Volume 144 Number 2 January 21, 2011
www.cell.com
Autophagosome Biogenesis Cued by Nutrient Sensing 2011 Global Funding Outlook
Your bacteria's favorite! Eppendorf Eporator ® The Eppendorf Eporator: the solution for efficiently transforming bacteria and yeast! The Eppendorf Eporator is the perfect fit for you and your lab. Ask the specialists—your bacteria and yeast—they will confirm that the product features and performance of the Eppendorf Eporator make it very easy for you to carry out your experiments and obtain excellent results.
Unique features of the Eppendorf Eporator Easy one-button start l Intuitive use guided by color display l Fast and freely programmable function keys l Integrated electroporation chamber for safety l Compact, space-saving design l USB port for data export l
For more information visit www.eppendorfna.com/eporator
www.eppendorf.com Þ Email:
[email protected] In the U.S.: Eppendorf North America, Inc. 800-645-3050 Þ In Canada: Eppendorf Canada Ltd. 800-263-8715
015.A1.0103.A © 2011 Eppendorf AG
NEW!
. W E N
Intelligent design
inCuSaFe™ copper enriched stainless steel interior
+
+
Single-beam, dual capture infrared CO2 sensor
+
SafeCell UV protection in situ
+
1
Hydrogen peroxide vapor sterilization in situ
Good laboratory technique
Sterisonic™ GxP Performance and Productivity Delivers Best Efficiency Value: 2 Hours SANYO Sterisonic™
14 Hours Brand X
H2O2 sterilization vs. high heat sterilization = Uptime (Hours) = Downtime (Hours)
7KHUDSLGin situ+2VHTXHQFHUHWXUQVWKH IXOO\VWHULOL]HG6WHULVRQLF*[3WRQRUPDOXVHTXLFNHU WKDQDQ\FRPSHWLWLYHLQFXEDWRULQWKHZRUOG
The industry’s first in situ H2O2 sterilization with the fastest turnaround. For maximum productivity in clinical, general purpose or the most highly compliant GMP applications, the new SANYO Sterisonic™ GxP CO2 incubator offers an impressive return on investment. With multiple contamination control safeguards, exclusive on-board H2O2 sterilization, FDA-21CFR data capture and graphical LCD display, the Sterisonic™ GxP rewards good laboratory technique with performance you can trust. Learn more, visit www.sterisonic.com or call 800-858-8442. pictured above: Sterisonic™ GxP MCO-19AICUVH with rapid H2O2 vapor sterilization system.
FREEER OFF [
D E TA IL
S
!
O N L IN E
]
FREE! FREE!
H2O2 sterilization system acessory, plus BD labware consumables kit. Limited time upgrade offer with purchase of the Sterisonic™ GxP MCO 19AICUVH. ($900 Value) BD stem cell starter kit with Sterisonic™ GxP quote. No purchase necessary! Act now. Supplies are limited. ($150 Value)
www.sterisonic.com ©2009 Sanyo Biomedical OWS 1015 05/09
Announcing an innovative new textbook from Academic Cell Primer to The Immune Response, Academic Cell Update Edition By Tak W. Mak and Mary Saunders
Primer to The Immune Response, Academic Cell Update Edition, is an invaluable resource for students who need a concise but complete and understandable introduction to immunology. Academic Cell textbooks contain premium journal content from Cell Press and are part of a new cutting-edge textbook/journal collaboration designed to help today’s instructors teach students to “think like a scientist.”
academiccell.com
Primer to The Immune Response Academic Cell Update Edition Tak W. Mak The Campbell Family Institute for Breast Cancer Research, Ontario, Canada Mary Saunders The Campbell Family Institute for Breast Cancer Research, Ontario, Canada Paperback/456 pages ISBN: 9780123847430 $79.95/£54.99/€64.95
Academic Cell is a dynamic textbook publishing partnership between Academic Press and Cell Press, two market-leading publishers bringing scientific advances from the world of life science research into the classroom. Order online now from: elsevierdirect.com/9780123847430 Request and examination copy from textbooks.elsevier.com
Twitter.com/academiccell
Facebook.com/academiccell
Editor Emilie Marcus Senior Deputy Editor Elena Porro Deputy Editor Robert Kruger Scientific Editors Karen Carniol Kara Cerveny Michaeleen Doucleff Fabiola Rivas Niki Scaplehorn Lara Szewczak Senior Managing Editor Meredith Adinolfi Deputy Managing Editor Andy Smith Lead Illustrator Andrew A. Tang Illustrators Yvonne Blanco Kate Mahan Production Staff Reyna Clancy Editorial Assistant Mary Beth O’Leary
Editorial Board Abul Abbas C. David Allis Genevie`ve Almouzni Uri Alon Angelika Amon Johan Auwerx Richard Axel Cori Bargmann Konrad Basler Bonnie Bassler David Baulcombe Jeffrey Benovic Carolyn Bertozzi Wendy Bickmore Elizabeth Blackburn Joan Brugge Lewis Cantley Joanne Chory David Clapham Andrew Clark Hans Clevers Stephen Cohen Pascale Cossart Jeff Dangl Ted Dawson Pier Paolo di Fiore Marileen Dogterom Julian Downward Bruce Edgar Steve Elledge Anne Ephrussi Ronald Evans Witold Filipowicz Marco Foiani Elaine Fuchs Yukiko Goda Stephen Goff Joe Goldstein
Douglas Green Leonard Guarente Taekjip Ha Daniel Haber Ulrike Heberlein Nobutaka Hirokawa Mark Hochstrasser Arthur Horwich Tony Hunter James Hurley Richard Hynes Thomas Jessell Narry Kim Mary-Claire King David Kingsley Frank Kirchhoff Richard Kolodner John Kuriyan Robert Lamb Mark Lemmon Beth Levine Wendell Lim Jennifer Lippincott-Schwartz Dan Littman Richard Losick Scott Lowe Tom Maniatis Matthias Mann Kelsey Martin Joan Massague´ Iain Mattaj Satyajit Mayor Ruslan Medzhitov Craig Mello Tom Misteli Tim Mitchison Alex Mogilner Paul Nurse Roy Parker
Dana Pe’er Kathrin Plath Carol Prives Klaus Rajewsky Venki Ramakrishnan Rama Ranganathan Anne Ridley James Roberts Alexander Rudensky Helen Saibil Joshua Sanes Randy Schekman Ueli Schibler Joseph Schlessinger Hans Scho¨ler Trina Schroer Geraldine Seydoux Kevan Shokat Pamela Sklar Nahum Sonenberg James Spudich Paul Sternberg Bruce Stillman Azim Surani Keiji Tanaka Craig Thompson Robert Tjian Ju¨rg Tschopp Ulrich von Andrian Gerhard Wagner Detlef Weigel Alan Weiner Jonathan Weissman Matthew Welch Tian Xu Shinya Yamanaka Marino Zerial Xiaowei Zhuang Huda Zoghbi
Cell Office Cell, Cell Press, 600 Technology Square, 5th Floor, Cambridge, Massachusetts 02139 Phone: (+1) 617 661 7057, Fax: (+1) 617 661 7061, E-mail:
[email protected] Online Publication: http://www.cell.com Cell (ISSN 0092-8674) is published biweekly by Cell Press, 600 Technology Square, 5th Floor, Cambridge, Massachusetts 02139. The institutional subscription rate for 2011 is $1,605 (US and Canada) or $1,847 (elsewhere). The individual subscription rate is $320 (US and Canada) or $363 (elsewhere). The individual copy price is $50. Periodicals postage paid at Boston, Massachusetts and additional mailing offices. Postmaster: send address changes to Elsevier Customer Service Americas, Cell Press Journals, 11830 Westline Industrial Drive, St. Louis, MO 63146, USA. The paper used in this publication meets the requirments of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed by Dartmouth Printing Company, Hanover, NH.
3TRUGGLINGåTOåKEEPåUPåWITHåå THEåLATESTåLIFEåSCIENCEåNEWS
#ELLå$AILY¬.EWS¬!GGREGATORåHASåTHEåSOLUTION 3UBSCRIBEåTOå&2%%å$AILYå.EWSå!LERTSåATåNEWSCELLCOMåANDå GETåTHEåLATESTåLIFEåSCIENCEåHEADLINESåDELIVEREDåTOåYOURåINåBOX
NEWSCELLCOM
Cell Press President & CEO Lynne Herndon Editor in Chief, Vice President of Content Development Emilie Marcus Vice President of Business Development Joanne Tracy Vice President of Web Development and Operations Keith Wollman Senior Product Manager Mark Van Hussen Director of Marketing Jonathan Atkinson Production Manager Meredith Adinolfi
Display Advertising Northeast/Mid-Atlantic: Victoria Macomber, ph: 508 928 1255; fax: 508 928 1256; e-mail:
[email protected] Midwest/Southeast/Eastern Canada: Inez Herrero-Redman, ph: 585 678 4395; fax: 585 678 4722; e-mail: i.herrero@elsevier. com Northwest/Southwest/Western Canada: Lori Young, ph: 646 370 6312; fax: 212 462 1915; e-mail:
[email protected] California: Elizabeth Loennborn, ph: 714 655 1877; fax: 214 452 9627; e-mail:
[email protected] UK/Europe: James Kenney, ph: +44 20 7424 4216; fax: +44 18 6585 3136; e-mail:
[email protected] Asia: Wendy Xie, ph: +86 10 8520 8827; e-mail: w.xie@ elsevier.com Classified Advertising United States and Canada: Gordon Sheffield, Key Account Manager, ph: 617 386 2189; fax: 617 397 2805; e-mail: g.sheffi
[email protected]
Press Officer Cathleen Genova
UK, Europe, and Asia: Sabrina Dodge, Key Account Manager, ph: +44 20 7424 4997; fax: +44 18 6585 3136; e-mail:
[email protected]
ª2011 Elsevier Inc. All rights reserved. This journal and the individual contributions contained in it are protected under copyright by Elsevier Inc., and the following terms and conditions apply to their use:
advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made. Although all advertising material is expected to conform to ethical (medical) standards, inclusion in this publication does not constitute a guarantee or endorsement of the quality or value of such product or of the claims made of it by its manufacturer.
Photocopying: Single photocopies of single articles may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee are required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for nonprofit educational classroom use. For information on how to seek permission, visit www.elsevier. com/permissions or call (+44) 1865 843830 (UK) / (+1) 215 239 3804 (US). Permissions: For information on how to seek permission, visit www.elsevier.com/ permissions or call (+44) 1865 843830 (UK) / (+1) 215 239 3804 (US). Derivative Works: Subscribers may reproduce tables of contents or prepare lists of articles including summaries for internal circulation within their institutions. Permission of the Publisher is required for resale or distribution outside the institution. Permission of the Publisher is required for all other derivative works, including compilations and translations (please consult www.elsevier.com/permissions). Electronic Storage or Usage: Permission of the Publisher is required to store or use electronically any material contained in this journal, including any article or part of an article (please consult www.elsevier.com/permissions). Except as outlined above, no part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without prior written permission of the Publisher. Notice: No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence, or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. Because of rapid
Reprints: Article reprints are available through Cell’s reprint service; for information, contact Nicholas Pavlow (e-mail:
[email protected]; ph: (+1) 212 633 3960). Subscription Orders and Inquiries: Mail, fax, or e-mail address changes to Elsevier Customer Service Americas, allowing 4–6 weeks for processing. Lost or damaged issues will be replaced, subject to availability, if Cell Press is notified within the claim period (US and airmail delivery: 3 months from issue date; surface delivery: 4 months from issue date). Periodical delivery in the US can take up to 3 weeks. Airmail delivery can take 2–4 weeks. The price of a single copy of Cell is $50 (excluding special issues). All orders must be prepaid and in writing. Please include the volume and issue number, payment (check or credit card, MasterCard, Visa, or American Express only), and a delivery address. Allow 4–6 weeks for delivery. Mailing address: Elsevier Customer Service Americas, Cell Press Journals, 11830 Westline Industrial Drive, St. Louis, MO 63146, USA. Toll-free phone within USA/Canada: 866 314 2355; phone for outside US/Canada: (+1) 314 453 7038; fax: (+1) 314 523 5170; e-mail:
[email protected]; internet: www.cellpress.com or <www.cell.com>. Funding Body Agreements and Policies: Elsevier has established agreements and developed policies to allow authors whose articles appear in journals published by Elsevier to comply with potential manuscript archiving requirements as specified as conditions of their grant awards. To learn more about existing agreements and policies, visit http://www.cell.com/cellpress/FundingBodyAgreements. Guide for Authors: For a full and complete guide for authors, please go to www.cell.com/ authors.
UNDERSTANDING CHANGE
New tools to advance epigenetics research For over 35 years, New England Biolabs has been committed to understanding the mechanisms of restriction and methylation of DNA. This expertise in enzymology has recently led to the development of a suite of validated products for epigenetics research. These unique solutions to study DNA methylation are designed to address some of the challenges of the current methods. EpiMark™ validated reagents simplify epigenetics research and expand the potential for biomarker discovery. Simplify DNA methylation analysis with MspJI
–
Hela
+
Plant (Maize) – +
–
Yeast
EpiMark™ validated products include: s Newly discovered methylation-dependent restriction enzymes
+
MspJI
s A novel kit for 5-hmC and 5-mC analysis and quantitation s Methyltransferases s Histones s Genomic DNAs
32 bp
To learn how these products can help you to better understand epigenetic changes, visit neb.com/epigenetics.
MspJI recognizes methylated and hydroxymethylated DNA and cleaves out 32 bp fragments for downstream sequencing analysis. Overnight digestion of 1 μg of genomic DNA from various sources with or without MspJI is shown. Note: Yeast DNA does not contain methylated DNA, therefore no 32-mer is detected.
CLONING & MAPPING
DNA AMPLIFICATION & PCR
RNA ANALYSIS
PROTEIN EXPRESSION & ANALYSIS
GENE EXPRESSION & CELLULAR ANALYSIS
www.neb.com
Leading Edge Cell Volume 144 Number 2, January 21, 2011 IN THIS ISSUE SELECT 163
The Many Faces of Cancer
ANALYSIS 167
Funding in 2011: East Heats Up as West Cools Down
C. Macilwain
PREVIEWS 170
Are Polycomb Group Bodies Gene Silencing Factories?
J.W. Hodgson and H.W. Brock
172
Rallying the Exocyst as an Autophagy Scaffold
J.-C. Farre and S. Subramani
PRIMER 175
High-Resolution Genome-wide Mapping of the Primary Structure of Chromatin
Z. Zhang and B.F. Pugh
SNAPSHOT 310
Chromatin Remodeling: SWI/SNF
M.M. Kasten, C.R. Clapier, and B.R. Cairns
Antibodies and Related Reagents for Signal Transduction Research
XP
™
Announcing
Monoclonal Antibodies,
eXceptional Monoclonal Technology™
exclusively from Cell Signaling Technology ®
development and production of antibodies with eXceptional Performance™ (XP™). This revolutionary and proprietary technology provides us with access to a broad range of antibody-producing B cells unattainable with traditional monoclonal technologies, allowing more comprehensive screening and the identification of XP monoclonal antibodies.
eXceptional specificity™ + eXceptional sensitivity™ + eXceptional stability and reproducibility™ = eXceptional Performance™ XMT coupled with our extensive antibody validation and stringent quality control delivers XP monoclonal antibodies with eXceptional Performance in the widest range of applications. untreated
EGF-treated (100 ng/ml, 15 min.) kDa
A43 1 Bx PC 3 He La
XMT™, an eXceptional Monoclonal Technology™ from Cell Signaling Technology®, enables the
Events
200
EGFR Receptor
140 100 80 60
EGF Receptor
human lung carcinoma
Jurkat cells (red) and Kyse70 cells (blue)
A-549 cells
EGF Receptor (D38B1) XP™ Rabbit mAb #4267 demonstrates eXceptional Performance in a wide range of applications. See www.cellsignal.com for full validation details.
for more information on XMT™ visit...
www.cellsignal.com China www.cst-c.com.cn
| Europe www.cellsignal.eu | Japan www.cstj.co.jp
Tel: 978-867-2300
| E-mail:
[email protected]
© 2010 Cell Signaling Technology, Inc. XMT™, XP™ , eXceptional Monoclonal Technology™, eXceptional specificity™, eXceptional sensitivity™, eXceptional stability and reproducibility™, eXceptional Performance™ and Cell Signaling Technology® are trademarks of Cell Signaling Technology, Inc. / DRAQ5® is a registered trademark of Biostatus Limited
Above: Confocal IF analysis of rat colon using Phospho-β-Catenin (Ser675) (D2F1) XP™ Rabbit mAb #4176. Red = DY-554 phalloidin. Blue = DRAQ5® #4084.
Articles Cell Volume 144 Number 2, January 21, 2011 187
Nuclear PTEN Regulates the APC-CDH1 Tumor-Suppressive Complex in a Phosphatase-Independent Manner
M.S. Song, A. Carracedo, L. Salmena, S.J. Song, A. Egia, M. Malumbres, and P.P. Pandolfi
200
Global Regulation of H2A.Z Localization by the INO80 Chromatin-Remodeling Enzyme Is Essential for Genome Integrity
M. Papamichos-Chronakis, S. Watanabe, O.J. Rando, and C.L. Peterson
214
Polycomb-Dependent Regulatory Contacts between Distant Hox Loci in Drosophila
F. Bantignies, V. Roure, I. Comet, B. Leblanc, B. Schuettengruber, J. Bonnet, V. Tixier, A. Mas, and G. Cavalli
227
Regulation of Mitochondrial Protein Import by Cytosolic Kinases
O. Schmidt, A.B. Harbauer, S. Rao, B. Eyrich, R.P. Zahedi, D. Stojanovski, B. Scho¨nfisch, B. Guiard, A. Sickmann, N. Pfanner, and C. Meisinger
240
Dual Action of ATP Hydrolysis Couples Lid Closure to Substrate Release into the Group II Chaperonin Chamber
N.R. Douglas, S. Reissmann, J. Zhang, B. Chen, J. Jakana, R. Kumar, W. Chiu, and J. Frydman
253
RalB and the Exocyst Mediate the Cellular Starvation Response by Direct Activation of Autophagosome Assembly
B.O. Bodemann, A. Orvedahl, T. Cheng, R.R. Ram, Y.-H. Ou, E. Formstecher, M. Maiti, C.C. Hazelett, E.M. Wauson, M. Balakireva, J.H. Camonis, C. Yeaman, B. Levine, and M.A. White
268
Delay in Feedback Repression by Cryptochrome 1 Is Required for Circadian Clock Function
M. Ukai-Tadenuma, R.G. Yamada, H. Xu, J.A. Ripperger, A.C. Liu, and H.R. Ueda
(continued)
Transfect with confidence. Building upon your voice and our vision,
Created to meet the needs of researchers who require high efficiency, ease of use and reproducibility,
Roche scientists have developed new reagents that position reliability and comprehensive efficiency across
X-tremeGENE reagents provide more viable cells with more physiologically relevant results. Our newest transfection reagents deliver results-driven performance, allowing you to focus your efforts on discovery, and helping you to shape the future. With Roche, you can
common and hard-to-transfect cell lines.
have complete confidence in the technology behind you.
Introducing X-tremeGENE DNA Transfection Reagents
Request your free sample today at x-tremegene.roche.com For life science research only. Not for use in diagnostic procedures. X-TREMEGENE is a trademark of Roche. © 2011 Roche Diagnostics. All rights reserved.
Roche Diagnostics Corporation Roche Applied Science Indianapolis, Indiana
282
RIM Proteins Tether Ca2+ Channels to Presynaptic Active Zones via a Direct PDZ-Domain Interaction
P.S. Kaeser, L. Deng, Y. Wang, I. Dulubova, X. Liu, J. Rizo, and T.C. Su¨dhof
RESOURCE 296
Densely Interconnected Transcriptional Circuits Control Cell States in Human Hematopoiesis
N. Novershtern, A. Subramanian, L.N. Lawton, R.H. Mak, W.N. Haining, M.E. McConkey, N. Habib, N. Yosef, C.Y. Chang, T. Shay, G.M. Frampton, A.C.B. Drake, I. Leskov, B. Nilsson, F. Preffer, D. Dombkowski, J.W. Evans, T. Liefeld, J.S. Smutko, J. Chen, N. Friedman, R.A. Young, T.R. Golub, A. Regev, and B.L. Ebert
POSITIONS AVAILABLE
On the cover: Cellular adaptation to nutrient-replete versus nutrient-constrained environments is driven by the mutually antagonistic actions of mTOR and ULK1, which specify cell growth versus autophagy. Here, Bodemann et al. (pp. 253–267) characterize distinct molecular platforms that control ULK1 versus mTOR activation and describe how these platforms are selectively assembled in response to nutrient availability. The image depicts the direct reciprocal inactivation relationship between the mTOR complex 1 and ULK1 kinases. This seemingly futile biochemical cycle is tamed through the nutrient-responsive orchestration of macromolecular protein complexes that couple ULK1 activation to autophagosome biogenesis. Art by Angela Diehl (UT Southwestern Medical Center).
Lasers are replacing conventional broadband light sources for fluorescence imaging applications due to desirable laser properties like high brightness, stability, long lifetime, and narrow spectral bandwidth. These features enable higher sensitivity, better image fidelity, and superior axial resolution in a variety of imaging applications using laser-scanning and spinning-disk confocal microscopes and total-internalreflection fluorescence (TIRF) microscopes. The narrow beam divergence, high spatial and temporal coherence, and well-defined polarization properties of lasers have enabled new fluorescence imaging techniques, such as super-resolution. The use of lasers as fluorescence light sources imposes new constraints on imaging systems and their components. For example, all optical filter wavelengths should be precisely keyed to the important laser lines and the spectra should offer steep transitions from the laser wavelength to fluorescence transmission. And exceptionally high transmission is critical to maximize system throughput, thus reducing acquisition time. Excitation filters act as “clean-up” filters, minimizing noise background resulting from the light away from the laser line, including spontaneous emission observed in solid-state lasers and the plasma lines of gas lasers. They should be hard-coated to withstand the high intensity of the laser beam. Emission filters should have deep blocking (optical density > 6) at all possible laser wavelengths to eliminate the intense stray laser light at the detector and very high transmission (> 97%) especially for low-light-level applications like single-molecule imaging. Emitters should also have low autofluorescence glass and excellent wavefront performance for minimal beam deviation and aberrations.
QP QP
7UDQVPLVVLRQ
Laser-based Fluorescence Imaging: What you need to know
7\SLFDO6SHFWUD ([FLWHU 'LFKURLF (PLWWHU /DVHU%DQGV
:DYHOHQJWKQP Long-pass filter set for laser microscopy - LF405/LP-A
Dichroic beamsplitters for laser applications must be anti-reflection (AR) coated to maximize transmission and eliminate coherent interference artifacts. They should also have high optical damage ratings like the exciters and low autofluorescence glass like the emitters. And it is critical for laser dichroics to exhibit sufficient flatness to eliminate axial focal shift and transverse aberrations associated with reflected laser light. Download our free white paper on this subject at: www.semrock.com Advances in thin-film filter technology pioneered by Semrock and embodied in all BrightLine® fluorescence filters permit the highest-performance fluorescence imaging, while resolving the longevity and handling issues that plague older, soft-coating technology.
We invite you to experience the difference at no risk. All Semrock filters have a 30 day, no questions asked return policy. We’ll even help you make the evaluation. Give us a call at 866-736-7625 for details.
A Unit of
Corporation
Visit www.semrock.com for a complete list of exclusively hard-coated filters.
Leading Edge
In This Issue A Long-Distance Relationship Goes Silent PAGE 214
How are distant genes coregulated? In this issue, Bantignies et al. show that colocalization of two Hox loci located 10 Mb from each other in Polycomb bodies contributes to their silencing. Their findings illustrate how positional organization of genes within the nucleus can regulate gene expression.
The APCs of PTEN’s Nuclear Function PAGE 187
PTEN is known as a cytoplasmic tumor-suppressing phosphatase. Song et al. now report that PTEN can inhibit cell proliferation in a rather different way in the nucleus—without using its enzymatic activity—by binding and activating the APC/C-CDH1 complex. This interaction may have important implications for cancer therapy, as suppression of both growth and cellular senescence appear to depend on PTEN regulating this complex.
Shuffling H2A.Z in Genome Stability PAGE 200
Chromatin-remodeling complexes regulate access to the genome for transcription. Papamichos-Chronakis et al. now shed light on how a chromatin-remodeling enzyme, INO80, functions as a key component of genomic stability in yeast. By removing histone H2A.Z from nucleosomes, INO80 enforces the proper genome-wide distribution of this histone variant, whose accurate positioning is required for completion of DNA replication.
Import Tax on TOM PAGE 227
Proteins entering mitochondria from the cytosol generally must pass through the translocase of the outer-membrane complex (TOM). Schmidt et al. now report that traffic through this gateway can be controlled by at least two distinct mechanisms. Phosphorylation of TOM components by casein kinase 2 stimulates channel construction, whereas phosphorylation by protein kinase A inhibits transfer across the membrane. Mitochondrial protein import can therefore be increased or decreased by different cytosolic kinases.
Keeping a Lid on Protein Folding PAGE 240
The protein-folding chamber of eukaryotic and archaeal chaperonins is sealed by a built-in lid. Douglas et al. demonstrate how ATP hydrolysis coordinates the closing of this lid with release of the unfolded substrate into the central chamber. Closing the lid may prevent premature escape of unfolded substrates, as both closing and release are required for successful folding. Their results help to explain how many proteins are directed down specific folding pathways. Cell 144, January 21, 2011 ª2011 Elsevier Inc. 159
60
years of leadership in human genetics research, education and service. 1948–2008 www.ashg.org
RalB Rallies Autophagy PAGE 253
Although many of the protein complexes that drive starvation-induced autophagy have been described, we understand very little about the signals that put these complexes into action. Bodemann et al. now show that, in response to nutrient deprivation, RalB triggers autophagy by remodeling the exocyst complex, which is better known for its role in docking vesicles with the plasma membrane during exocytosis. By exchanging components of this complex, it creates an autophagy-specific signaling platform on which the core vesicle nucleation machinery can be recruited for authophagosome biosynthesis.
A Cry in the Dark PAGE 268
Cryptochrome1 (Cry1) is an essential component of the mammalian circadian clock due to its role as a repressor of ‘‘morning-time’’ transcription. Ukai-Tadenuma et al. now uncover a combination of ‘‘day-time’’ and ‘‘night-time’’ cis-regulatory elements that establish Cry1’s ‘‘evening-time’’ peak in expression. They then use these findings to systematically engineer Cry1 constructs with a range of desired circadian phases and demonstrate that the phase of Cry1 expression affects the duration of circadian oscillation.
Calcium at the Synaptic RIM PAGE 282
Rapid communication between neurons depends on the ability of neurotransmitter-containing vesicles to quickly respond to presynaptic Ca2+ influx. Kaeser et al. show that RIM, an active zone protein, mediates this high-speed connection by acting as a molecular tether linking Ca2+ channels and synaptic vesicles. By situating calcium influx close to the vesicle pool, RIM ensures fast and synchronous neurotransmission.
A Blueprint for Hematopoiesis PAGE 296
Generating the diverse array of human blood cell types from a single multipotent progenitor requires a complex sequence of differentiation steps. Novershtern et al. untangle the transcriptional architecture underlying these lineages by comparing patterns of gene expression in 38 distinct purified populations of human hematopoietic cells. By coupling computational approaches with experimental validation, they identify specific regulators of hematopoiesis and decipher how the regulatory circuitry driving blood cell diversity is organized.
Cell 144, January 21, 2011 ª2011 Elsevier Inc. 161
BE THE FIRST
to read the latest issue of any Cell Press journal.
Register for Cell Press Email Alerts and get the complete table of contents as soon as the issue publishes online — FREE! Cell Press Email Alerts deliver the news, research, and commentaries featured in each journal’s latest issue, including the full title of every article, direct links to the articles, and the complete author list. Plus, to save you time, each research article has a brief summary highlighting its significant findings. You don’t have to be a subscriber to sign up for Cell Press Email Alerts. While subscribers have instant access to the full text of all articles listed in the Email Alerts, non-subscribers can read the abstracts of all articles as well as the full text of the issue’s Featured Article.
www.cellpress.com
Leading Edge
Select: The Many Faces of Cancer Although cancer is typically viewed as a consequence of unregulated cell proliferation, it has become increasingly clear that it also reflects a crisis of cellular identity. In this Select, we highlight recent papers that may help us to unravel the multiple personalities of tumor cells and, in doing so, bring us closer to conquering cancer in the clinic.
DIY Plumbing for Glioblastoma
Human tumor cells generate human blood vessels when transplanted into a mouse (human CD144 depicted in red). Image courtesy of M. Todaro and R. De Maria.
Stimulating the formation of new blood vessels confers a selective advantage on tumor cells and is therefore a valuable therapeutic target. It’s currently thought that in general, cancer cells ‘‘contract out’’ angiogenesis to noncancerous endothelial cells, which are recruited by tumor-derived proteins such as vascular endothelial growth factor (VEGF). However, two recent studies show that glioblastoma takes a radically different approach to building new blood vessels. Wang et al. (2010) and Ricci-Vitiani et al. (2010) show that the unusual vascular tissue of these aggressive brain tumors contains a subpopulation of cells derived directly from the tumor itself, not the surrounding endothelium, and that these cells share the same mutational profile as the parent tumor. The teams also show that a fraction of human glioblastoma cells can differentiate into endothelial cells in vitro and, importantly, form tumors with human vasculature when transplanted into mice. This human-derived vascular tissue appears to support tumor growth, as selective ablation of these cells in mice results in a significant reduction in tumor size. Understanding the tumor origins of glioblastoma blood vessels might explain why current angiogenesis inhibitors often fail to make an impact on tumor growth and will likely lead to more effective means of targeting angiogenesis in the future. Ricci-Vitiani, L., et al. (2010). Nature 468, 824–828. Wang, R., et al. (2010). Nature 468, 829–833.
Back to the Germline Gene expression profiling is a powerful tool to tease out the many different changes in cellular identity that shape the development of a tumor. Janic et al. (2010) now use a Drosophila brain tumor model to identify changes in gene expression driving the acquisition of cancer cell identity. Flies with a temperature-sensitive mutation in the lethal (3) malignant brain tumor (l(3)mbt) gene develop very large malignant growths in the larval brain. The gene encodes a transcriptional repressor that is thought to act with retinoblastoma (RB) and HP1 to trigger nucleosome compaction and gene silencing. In order to determine which genes are regulated by this silencing factor, Janic et al. compare the gene expression profile of l(3)mbt larval brains with those of a panel of controls. This panel contains not only brains without tumors but also brains from a different Drosophila brain tumor line (brat). By comparing the genes that are differentially expressed in each tumor, they can remove a large number of genes that are required for cell proliferation and focus on the remaining candidates that may define tumor cell identity. Remarkably, a quarter of these genes encode proteins that are normally expressed specifically in germline cells. Upregulation of some of these genes in larval brain tumors appears to be important for tumor growth, as flies that carry mutations in both l(3)mbt and several germline-specific genes fail to develop brain overgrowth. Inactivation of these germline genes does not inhibit tumor growth in a different Drosophila brain tumor model, suggesting that this soma-to-germline transition is not likely to be a global signature of tumor cells. However, a number of human tumors, including melanoma and several carcinomas, are known to upregulate germline-specific genes. This study suggests that germline genes may play a more important role in tumor growth than previously thought. Janic, A., et al. (2010). Science 330, 1824–1827.
A transplant of larval brain tissue from l(3) mbt mutant flies grows dramatically in the adult fly and depends on the expression of germline-specific genes. Image courtesy of C. Gonzales.
Cell 144, January 21, 2011 ª2011 Elsevier Inc. 163
.%7
Reach Your Ideal Candidate! så0OSTåONEåORåMULTIPLEåå JOBåOPENINGSå så0ROMOTEåANåUPCOMINGåå CONFERENCEåORåEVENTå så!NNOUNCEåAåGRANTAWARDå så%LEVATEåYOURåå ORGANIZATIONSåPROlLE
careers.cell.com
A Tale of Two Tumors Around 25% of medulloblastomas originate from cerebellar granule neuron precursors and are characterized by aberrant activation of the Sonic Hedgehog pathway. However, very little is known about the molecular underpinnings of the remaining cases. Gibson et al. (2010) now show that some of the variation between different medulloblastomas is likely to be due to their having very distinct developmental origins. By taking advantage of the Brain Explorer 2 Gene Atlas, a three-dimensional map of gene expression patterns in the mouse brain, Gibson et al. Two subtypes of medulloblastoma have distinct searched for brain regions that express the signature genes of a different developmental origins. Image courtesy of R. Gilbertson. form of medulloblastoma, known as the WNT-subtype. This form is characterized by activating mutations in catenin b1 (CTNNB1) and tends to arise in older children compared with the SHH-subtype. Unexpectedly, WNT-subtype signature genes tend to be expressed not in the cerebellum but in the dorsal brainstem. Moreover, mice with a conditional activating mutation in Ctnnb1 fail to generate tumors in the cerebellum but develop hyperplastic growths in the dorsal brainstem. If activation of Ctnnb1 is combined with inactivation of p53, fully developed medulloblastomas form. These mouse tumors are astonishingly similar to their human WNT-subtype counterparts, in terms of histology, gene expression profile, and even specific chromosomal deletions. Identifying the cell type of origin of this tumor subtype has therefore led directly to the development of an accurate mouse model of the disease. In general, this kind of approach is likely to become increasingly widely used to develop more specific treatments, allowing us to attack cancer at its bewilderingly diverse roots. Gibson, P., et al. (2010). Nature 468, 1095–1099.
Silencing Antitumor Immunity Attempts to stimulate antitumor immunity have so far failed to make a significant impact. Even though these treatments are often successful in raising a systemic immune reaction, they tend to have a limited effect on tumor growth, which suggests that tumors have means of evading the immune response. Now, Kraman et al. (2010) show that a population of stromal cells within tumors is responsible for defending cancer from attack by the immune system. Cells expressing fibroblast activation protein-alpha (FAP-a) are thought to play a role in wound healing by preventing an inappropriate immune response at the site of injury. Kraman et al. identify a small population of FAP-a-positive cells within lung carcinomas and pancreatic adenocarcinomas. Despite being relatively sparse, these cells appear to exert a powerful effect on the tumor microenvironment. Specific deletion of the FAP-a subpopulation is associated with a dramatic increase in the efficacy of therapeutic immunization against a tumor antigen, resulting in hypoxic necrosis of the tumor. This response occurs rapidly and locally and depends on the immune system, as it fails to occur in immunodeficient Rag2 mutant mice. The exact basis of this reaction, however, remains mysterious. Although antibodies against TNF-a and IFN-g are effective in blocking the antitumor response, there are no significant changes in the level of these and other inflammatory cytokines following deletion of FAP-a-positive cells. This suggests that suppression of antitumor immunity involves attenuation of the surrounding cells’ sensitivity to cytokines, rather than inhibition of their expression or secretion. Armed with this knowledge, it may now be possible to enhance the effectiveness of therapeutic vaccination strategies for cancer. Kraman, M., et al. (2010). Science 330, 827–830. Niki Scaplehorn
Cell 144, January 21, 2011 ª2011 Elsevier Inc. 165
Supplying Reagent, Pre-clinical and cGMP Grade Proteins to the Biopharmaceutical and Vaccine Development Community
Recombinant Vaccine Components ! ! P. aeruginosa rEPA ! ! E. coli • C. difficile COMING SOON!
• Circumsporozoite Protein Reagent Proteins is now your single source for reagent, pre-clinical and cGMP grade proteins. See our new website for a complete listing of products and special discounts of up to 30% on recombinant human proteins.
Win an iPad ! Make a purchase by Feb 28, 2011 for a chance to win an iPad!
Leading Edge
Analysis Funding in 2011: East Heats Up as West Cools Down The great slowdown in the global economy since 2008 is putting the brakes on government spending for health research funding worldwide. Could this be an opportunity for research centers in Asia to shift the power balance? For the past decade, global investment in biomedical research, both basic and clinical, has surged forward. The United States doubled the budget of the National Institutes of Health (NIH) between 1998 and 2003, and many other governments have eagerly supported the life sciences, sometimes at the expense of other areas of research. But in October 2008, the banking system collapsed and knocked the stuffing out of government balance sheets. Since then, eagerness for funding biomedical research
has stalled. And, according to analysts and senior researchers in the fifteen largest scientific powers, funding will go into reverse during 2011. With stories of deep cuts circulating, the atmosphere in many labs is one of alarm. However, a closer look at budget plans shows that governments aren’t axing spending, but rather retrenching. In fact, most nations are protecting science from steeper cuts applied to other areas of public spending. The United States, which utterly dominates global spending on
biomedical research, and Britain, which has energetically sought leadership in some areas of it, will each try to maintain spending at 2010 levels this year, even in the face of huge budgetary pressures. That said, some mid-sized players, including Italy, Spain, Canada, and Australia, have already cut back their spending since 2008. And Singapore is sharply reorientating its program by diverting 30% of its planned biomedical research spending to partnerships with industry. Other nations are pressing their researchers to generate visible results, fast. But the frigid funding outlook isn’t global. Scientific powers with strong export industries have been protected from the spending slump, as China, Korea, India, and Germany are planning substantial increases in the spending on biomedical research for 2011. Apples and Oranges International comparisons research spending are
of health notoriously
Cell 144, January 21, 2011 ª2011 Elsevier Inc. 167
difficult. Most nations lack a dominant agency akin to the US’s NIH, supporting both laboratory and clinical research. Many nations pay researchers’ salaries out of nonresearch budgets, and their basic research efforts aren’t easily allocated to specific missions, such as health. Thus, the world’s main repositories of research statistics—the OECD (Organisation for European Economic Cooperation) in Paris, Eurostat in Luxembourg, and the US National Science Foundation in Arlington, Virginia—struggle to agree on what constitutes health research spending. Nevertheless, spending on biomedical research is heavily skewed toward wealthy countries in general and the US in particular. According to an exhaustive 2007 study by the European Medical Research Councils, the US spends $40 billion annually. This is 2–3 times as much as the European Union, which in turn spends significantly more than the rest of the world combined. But could this balance of power be shifting? A rapid-fire global tour of the world’s largest spenders reveals the beginnings of a swing in government spending on biomedical research, with investments in Asia surging forward, although admittedly from a low base. 2012—NIH’s Year of Reckoning The outlook for the United States typifies the global picture in developed countries. Although the current financial year for the US began in October 2010, Congress isn’t expected to complete the 2011 budget until this March. Until then, spending is frozen at last year’s level of $31 billion, and any adjustment is likely to be downwards. The newly elected House of Representatives, now under the Republican’s control, will try to cut the budget, perhaps by $2 billion, and the biomedical research lobby will be looking to its allies in the Senate to protect it. This means that, in the most optimistic scenario, the actual value of the NIH budget will fall by 3% in 2011, once inflation is taken into account. Obama will propose his budget for 2012 this February, before the 2011 budget is even complete, and Congress is again expected to try and reduce whatever number he proposes. Every nation’s research budget has its quirks, and the biggest one for the US
right now is the continuing impact of Mr. Obama’s stimulus package, which injected an additional $10 billion into NIH in February 2009. Most of the money will be spent in 2010 and 2011, according to the Federation of American Societies for Experimental Biology. It will supplement core funding in each year by over $4 billion. ‘‘Really the key pressure point for NIH is the 2012 financial year,’’ says Jennifer Zeitzer, legislative director at the FASEB. The strategy in Canada appears similar: hold funding steady, at best. A decade ago, Canada tried hard to assert itself in the biomedical sphere by establishing the Canadian Institutes of Health Research. But after rapid growth initially, its budget has plateaued at about CN $930 million (US $940 million) in 2007 and 2008. Last year, Canada’s budget rose by 1.5%, and according to Mark Henderson, the editor of Canadian newsletter Research Money, it is ‘‘highly unlikely’’ to do better in 2011. Funding in Australia also grew quickly over the last decade. However, the budget for the National Health and Medical Research Council in Australia peaked at AU $714 million in 2009, and it has been frozen ever since. ‘‘The outlook for next year looks bleak,’’ says Julie Campbell, a biologist at the University of Queensland and president of the Association of Australian Medical Research Institutes. ‘‘There certainly is no hope for an increase, with a real threat that it may be decreased.’’ Asia Surges Forward In Asia, the picture is quite different. Korea’s spending in areas related to biotechnology powered ahead from less than US $100 million in 1998 to $850 million in 2008 and $1.25 billion in 2010. This money comes from three separate government departments and goes to both basic and clinical research. According to researchers and government officials, Korea’s funding will probably keep growing at a compound rate of around 20%. India has experienced a massive boom in private-sector research and development since 2005, when it transformed its large pharmaceutical industry by reforming its patent laws. Moreover, the Indian government is preparing to increase the
168 Cell 144, January 21, 2011 ª2011 Elsevier Inc.
combined budget for its biotechnology and science departments (US $700 million) by more than 20% this year. China’s budgets are notoriously opaque, and details of its research spending don’t feature in the OECD’s international comparisons. Nonetheless, China’s biomedical spending has been growing relentlessly, probably close to the annual compound rate that the Chinese report for research and development—an impressive 22.6% per year. Yet China’s efforts have to be kept in perspective. From 2006–2008, for example, China spent just 11.9 billion Yuan (US $1.8 billion) on research aimed at health and agriculture combined (see UNESCO report). Assuming China invests the same again on other basic research related to health, this puts China’s investment in biomedical research just above Korea’s and still considerably below that of Japan. Early last month, Reuters published an unconfirmed report that China might invest an astounding $1.5 trillion dollars in seven high-tech industry sectors— including biotechnology—under its five year plan for 2011–2015. Such leaks are used habitually in China to signal intent rather than to plan budgets. And, in the past, China has put agriculture ahead of health as a biotechnology priority. Nonetheless, the announcement offers a taste of what could lie ahead if China’s economy continues to expand. Perhaps the world’s most audacious experiment in building a biomedical research hub from scratch has taken place in Singapore. In 2000, Singapore established the BioMedical Research Council (BMRC), which offered large, unfettered grants for basic research. This strategy has developed a significant research capability, drawing significant global attention and attracting a large number of talented scientists in areas such as stem cell research. However, in September 2010, the government announced that its approach will change. The amount of money available for biomedical research will grow slightly in 2011–2015, to SI $3.7 billion (US $2.8 billion), but 30% of that will be available only for collaborations with drug companies. ‘‘The money will be distributed in a different way,’’ says George Radda, chair of the BMRC. ‘‘The government is
expecting researchers to show some returns for its investment.’’ The priority shift came after a governmental study found that investments in engineering research provided stronger economic returns for Singapore. ‘‘The principles are not bad,’’ comments one leading biologist in Singapore. ‘‘But the execution will be a challenge.’’ Japanese public spending on biomedical research, which is split between several agencies, has grown slowly over the past decade. Now it totals about 340 billion yen (US $4 billion). Its 2011 budget hadn’t been agreed at the time of writing, but it was expected to be flat or slightly down, according to analysts in Japan. Deutsche Boom, Euro Gloom Of the major European nations, only Germany is planning a significant expansion for biomedical funding in 2011. The science ministry will increase funding by 7%, and an additional V300 million is allocated to translational work. So the total biomedical budget could grow by as much as 11% in 2011.
France has also backed the life sciences over the last few years, though less emphatically than Germany. French politicians traded barbs last October over whether the total 2011 research budget across all agencies was actually up or down. But, nevertheless, the main biomedical agency in France, Inserm, will receive a 4.3% funding increase this year. After taking a swinging 14% cut last year, Spanish researchers are breathing a small sigh of relief in 2011—their science and innovation ministry is being protected from an austerity budget now pushing through the legislature. But the atmosphere in universities in southern Europe is still downbeat. Like colleagues in Italy, who experienced double-digit cuts across all disciplines last year, Spanish researchers are worried that their prospects of catching up with neighbors in northern Europe are receding, perhaps for good. ‘‘The situation in Italy is dismal,’’ says Ramon Marimon, a former Spanish science secretary currently working as an economist in Florence, ‘‘and it has been for some time.’’
In conclusion, nations whose economies have held up since the 2008 financial crisis may be rewarded in 2011 with an opportunity to make inroads into the established biomedical order. If Asia’s economic resurgence and scientific renaissance continue, a real change in the global balance of power in biomedical research could indeed transpire. But these future scientific powerhouses still have a lot of ground to cover to approach the funding levels of the United States. Web Resources
UNESCO Science Report 2010: http:// www.unesco.org/science/psd/ publications/science_report2010.shtml Eurostat: http://ec.europa.eu/ Organisation for Economic Co-operation and Development: http://www.oecd. org/ European Medical Research Councils 2007 White Paper: http://www.esf.org/ publications.html
Colin Macilwain Edinburgh, UK DOI 10.1016/j.cell.2011.01.002
Cell 144, January 21, 2011 ª2011 Elsevier Inc. 169
Leading Edge
Previews Are Polycomb Group Bodies Gene Silencing Factories? Jacob W. Hodgson1 and Hugh W. Brock1,* 1Department of Zoology, University of British Columbia, Molecular Epigenetics Group, Life Sciences Centre, 2350 Health Sciences Mall, Vancouver BC V6T 1Z3, Canada *Correspondence:
[email protected] DOI 10.1016/j.cell.2011.01.006
Polycomb group (PcG) proteins mediate long-range associations between Hox genes, which correlate with gene repression in vivo. Bantignies et al. (2011) identify a physiological role for the nuclear localization of Hox genes in PcG-mediated gene silencing, strengthening the evidence that nuclear positioning regulates gene expression. During the development of multicellular organisms, gene expression is controlled by both the spatial organization of the genome in the nucleus and the nuclear architecture. Segments of chromatin adopt highly organized structures in confined subregions of the nucleus called chromosome territories (Marshall et al., 1997; Okamoto and Heard, 2009), localized to specific subnuclear domains. These domains include ‘‘transcription factories,’’ which encompass clusters of actively transcribing genes, and Polycomb group (PcG) bodies, which colocalize with stably repressed Hox genes in Drosophila. Silencing of Hox genes requires longrange chromosomal interactions mediated by the PcG repressive complexes, which bind PcG response elements (PREs) of target genes (Vazquez et al., 2006; Mu¨ller and Verrijzer, 2009). Furthermore, the colocalization of PcG target genes within nuclear PcG bodies is regulated by the cell cycle (Buchenau et al., 1998). However, in contrast to transcription factories, which clearly mediate gene expression, it is not known whether PcG bodies contribute to gene silencing directly through their component PcGrepressive complexes or indirectly by positioning their target genes in nuclear domains. Now, Bantignies et al. (2011) show that compartmentalization of longrange interactions of PREs in PcG bodies contributes to epigenetic silencing of Hox genes by PcG proteins. Throughout the development of higher eukaryotes, the PcG genes maintain the regional identity of segments along the
anterior-posterior (i.e., head-to-tail) axis of the embryo by repressing Hox genes in specific regions. The Hox genes are organized into two clusters in Drosophila. The Antennapedia genes control the formation of a portion of the head and anterior thorax (termed parasegments [PS] 1–4), whereas the Bithorax genes regulate differentiation of the posterior thorax and abdominal segments (PS5– 14) of the fly (Figure 1A). Silencing of the Bithorax gene cluster in anterior parasegments is thought to occur by two major types of chromosomal interactions. First, PREs are present within regulatory regions of Bithorax genes, and these PREs can silence genes by interacting in cis or in trans through the pairing of homologous chromosomes (Lewis 1954; Pirrotta 1999). Second, long-range cis interactions between the PREs and their target promoters form higher-order three-dimensional (3D) chromatin structures, which also result in gene silencing in the anterior of the embryo (Lanzuolo et al., 2007). The higher-order structure of the Antennapedia gene cluster is not as well characterized. Nevertheless, expression of the Antennapedia (Antp) gene is silenced in the head regions of flies and in the posterior regions of the embryo by the Abdominal-B (AbdB) gene products at all developmental stages. Bantignies et al. now take advantage of the spatial patterns of Hox gene silencing in the fly embryo to determine whether higher-order chromatin assembly of the Antennapedia gene cluster contributes to silencing by PcG. Using a combination
170 Cell 144, January 21, 2011 ª2011 Elsevier Inc.
of high-resolution RNA fluorescence in situ hybridization (FISH) and immunostaining to Polycomb, the authors demonstrate that PcG bodies extensively colocalize with the Antp and the bithorax complex gene Abd-B. This colocalization occurs in cells only in anterior regions of the embryo or larval head tissues, where both genes are silenced. In addition, Antp and Ultrabithorax (Ubx), but not Abd-B, associate with PcG bodies in posterior regions of the embryo where Antp and Ubx, but not Abd-B, are silenced (Figures 1B and 1C). These long-range interactions between genes spaced 10 Mb apart require the PcG genes. These data together show that position-dependent association of PcG targets in PcG bodies correlates with silencing of the target. To determine which regulatory elements of Abd-B mediate the longrange contacts to Antp, the authors use Chromosome Conformation Capture on Chip (4C). This variation of chromosome conformation capture allows for the unbiased determination of interacting sequences over large genomic regions. They find that two PREs, Fab-7 and Mcp of the Abd-B gene, interact with Antp. These PREs are highly enriched for the PcG-dependent repressive epigenetic mark, H3K27me3 (i.e., trimethylation of the lysine 27 on histone H3). Bantignies and colleagues then show that mutating Fab-7 partially derepresses expression of the genes pb, Dfd, Scr, and Antp in larval head tissues. Finally, the authors demonstrate that the long-range chromosomal contacts between Antp and Fab-7
Figure 1. Mechanisms of Spatial Silencing of Hox Genes in Drosophila (A) In the genome of Drosophila melanogaster, the Hox genes are organized into two clusters, the Antennapedia (ANT-C) genes and the Bithorax (BX-C) genes, which are separated by 10 Mb. The ladybird early lbe gene is located 4.5 Mb upstream of the BX-C genes. (B) Polycomb group (PcG) bodies are subdomains of the nucleus, which contain long-range chromosomal interactions that correlate with the repression of Hox genes in vivo. For example, in the head region of the developing fly embryo (or larval eye disc region), the Antennapedia genes Antp and Abd-B are silenced, and they associate with each other inside of PcG bodies (Bantignies et al., 2011). Hox genes are colored as in (A). (C) In contrast, in the posterior parasegment 13 (PS13) of the fly embryo, the Antennapedia genes Antp and Ubx are both silenced and colocalize in PcG bodies (Bantignies et al., 2011).
are physiologically relevant. When they cross Fab-7 mutant flies (or Mcp1 mutants) with flies containing a sensitized allele of Antp (AntpNs), they observe an enhancement of the homeotic phenotypes of AntpNs, which is characteristic of PcG gene interactions. Bantignies and colleagues argue that long-range intrachromosomal interactions of Antp and Abd-B represent a higher order of chromatin gene regulatory mechanism. However, the in vivo effect that the long-range interactions have on Antp repression can be detected only in sensitized genetic backgrounds. Therefore, it remains possible that these long-range interactions represent a transient association that acts independently of short-range repression mediated by PREs. Alternatively long-range interactions may play a secondary role, such as stabilizing other short-range cis or trans interactions between the two gene clusters. The authors tested the physiological relevance of only a limited number of the
interactions detected between the Antennapedia and Bithorax gene clusters with 4C. Therefore, it remains to be seen whether the other long-range interactions exist between PREs of other genes and, if so, whether they are transient or perform compensatory roles in mutant chromosomes. Another outstanding question is what mechanism targets the PREs to the PcG bodies? Most researchers assumed that association of PRE with PcG bodies was constitutive, yet Bantignies and colleagues observe long-range interactions of PREs at PcG bodies only in regions of embryos where Hox genes are silenced. Furthermore, most studies show that binding of PcG proteins to PREs is also constitutive, suggesting that PcG proteins are necessary, but not sufficient, for targeting PREs to PcG bodies. Therefore, one attractive hypothesis is that the default chromatin structure of Hox gene clusters is association of PREs with PcG bodies, and then gene activation somehow prevents this
colocalization. Alternatively, non-PcG proteins or modified histones might recruit silenced genes to PcG bodies. This model predicts that individual PcG bodies could have different protein compositions, depending on which targets are present. This question could be addressed by determining whether PcG bodies in nuclei at different regions of the anterior-posterior axis are heterogeneous or whether they have uniform subunit composition or histone modifications. Indeed, the complex structure and regulation of Hox clusters may require specialized PcG bodies. If so, one would predict that Hox PREs would be recruited to a subset of PcG bodies within a nucleus and that other PREs would be recruited to a different subset. In that case, Hox-specific subsets of PcG bodies within a nucleus will have different subunit composition, histone modifications, or kinetics of association or dissociation compared to other PcG bodies. Answers to these questions and those raised by the authors will shed more light on gene silencing by the nuclear architecture and the 3D structure of chromatin. The findings by Bantignies and colleagues suggest the exciting possibility that the selective association of PREs with silencing factories has a role in the epigenetic gene silencing by the PcG. REFERENCES Bantignies, F., Roure, V., Comet, I., Leblanc, B., Schuettengruber, B., Bonnet, J., Tixier, V., Mas, A., and Cavalli, G. (2011). Cell 144, this issue, 214–226. Buchenau, P., Hodgson, J., Strutt, H., and ArndtJovin, D.J. (1998). J. Cell Biol. 141, 469–481. Lanzuolo, C., Roure, V., Dekker, J., Bantignies, F., and Orlando, V. (2007). Nat. Cell Biol. 9, 1167– 1174. Lewis, E.B. (1954). Am. Nat. 88, 225–239. Marshall, W.F., Straight, A., Marko, J.F., Swedlow, J., Dernburg, A., Belmont, A., Murray, A.W., Agard, D.A., and Sedat, J.W. (1997). Curr. Biol. 7, 930–939. Mu¨ller, J., and Verrijzer, P. (2009). Curr. Opin. Genet. Dev. 19, 150–158. Okamoto, I., and Heard, E. (2009). Chromosome Res. 17, 659–669. Pirrotta, V. (1999). Biochim. Biophys. Acta 1424, M1–M8. Vazquez, J., Mu¨ller, M., Pirrotta, V., and Sedat, J.W. (2006). Mol. Biol. Cell 17, 2158–2165.
Cell 144, January 21, 2011 ª2011 Elsevier Inc. 171
Leading Edge
Previews Rallying the Exocyst as an Autophagy Scaffold Jean-Claude Farre´1 and Suresh Subramani1,* 1Section of Molecular Biology, Division of Biological Sciences, University of California, San Diego, La Jolla, CA 92093-0322, USA *Correspondence:
[email protected] DOI 10.1016/j.cell.2011.01.005
Protein scaffolds coordinate the assembly of many multicomponent signaling complexes. Bodemann et al. (2011) now show that the exocyst, a protein complex involved in tethering transport vesicles to the plasma membrane, provides an assembly and activation platform for components of the autophagy machinery via a process requiring the GTPase RalB. Macroautophagy (henceforth referred to as autophagy) is a cellular degradation pathway for the clearance of damaged or superfluous proteins and organelles (Yang and Klionsky, 2010). Despite its importance in cellular homeostasis and in immunity against pathogens and despite accumulating knowledge regarding the autophagy machinery itself, the early regulatory signals that activate autophagy are unknown. Furthermore, several gaps remain in our understanding of how distinct autophagy complexes assemble and collaborate (Mehrpour et al., 2010). Bodemann et al. (2011) now identify some of the early regulatory and assembly steps, revealing a role for the exocyst, a complex that regulates postGolgi protein traffic. The authors demonstrate that nutrient deprivation, a condition that promotes autophagy, activates the Ras-like small GTPase RalB, which then engages the effector protein and exocyst component Exo84. This interaction promotes the assembly and activation of the autophagy complex using the exocyst as an assembly scaffold. During metabolic stress, including starvation, autophagy promotes the degradation of cytoplasmic components by the lysosome, and the recycling of their constituents promotes cell survival (Mehrpour et al., 2010; Yang and Klionsky, 2010). Autophagy involves formation of an isolation membrane, which elongates and fuses to form a double-membrane vesicle called an autophagosome. The autophagosome encloses cytoplasmic cargoes for delivery by fusion to the endosome or lysosome, eventually forming an autolysosome.
Autophagosome formation consists of three steps: nucleation, expansion, and fusion of the isolation membrane. Each step involves a specific set of protein complexes. The ULK (Unc-51-like kinase) and PI3K (phosphatidylinositol-3 kinase) complexes are most important for nucleation, whereas the ubiquitin-like (Ubl) conjugation system and the mAtg9 (mammalian autophagy-related gene 9) cycling complex, which is involved in transit of mAtg9 to and from the isolation membrane, facilitate expansion and closure of the isolation membrane (Mehrpour et al., 2010). The exocyst, a hetero-octameric complex containing the proteins Sec3, Sec5, Sec6, Sec8, Sec10, Sec15, Exo70, and Exo84 (recently renamed EXOC1-EXOC8) is involved in the postGolgi trafficking and tethering of vesicles to the plasma membrane (He and Guo, 2009; Munson and Novick, 2006). New evidence implicating a role for the exocyst complex in signaling during pathogen infection (Chien et al., 2006) led the authors to screen for proteins interacting with the exocyst subunit, Sec3. Using a high-throughput yeast two-hybrid screen, the authors found that both negative and positive regulators of autophagy interact with Sec3. The interactors include Rubicon (RUN domain and cysteine-rich domain containing), an inhibitor of autophagy, as well as Atg14L, a component of the PI3K complex, and FIP200, part of the ULK complex. Indeed, the authors find that several exocyst subunits (Sec3, Sec5, and Exo84) coimmunoprecipitate with Rubicon and Atg14L. Additionally, the core exocyst subunit, Sec8, associ-
172 Cell 144, January 21, 2011 ª2011 Elsevier Inc.
ates with Atg5 and Atg12, autophagy proteins involved in the ubiquitin-like system, cementing the association between the autophagy machinery and the exocyst. Given the association between autophagy and exocyst components and the fact that the small GTPases RalA and RalB mobilize exocyst assembly (Moskalenko et al., 2002, 2003), the authors next inquire whether RalA and RalB also play a role in autophagy. Indeed, Bodemann et al. show convincingly that activation of RalB, but not RalA, in cervical cancer and epithelial cell lines is necessary for autophagy. They find that competitive inhibitors of RalB inhibit the induction of autophagy during starvation, whereas constitutively activated forms of RalB stimulate autophagy even under nutrient-rich conditions. Thus, RalB is both necessary and sufficient for activation of autophagy. RalB and its related partner RalA cooperate in mitogen-induced signaling during oncogenic transformation by Ras. RalA is required to bypass normal restraints on cell proliferation, whereas RalB bypasses normal restraints on cell survival (Chien et al., 2006). Tumor cells have higher levels of RalB, and cells depleted of RalB exhibit survival defects (Bodemann and White, 2008). These observations may be explained, in part, by the finding that RalB promotes cell survival during starvation by inducing autophagy. Interestingly, the authors characterize two complexes containing both exocyst and autophagy components: an autophagy-active and autophagy-inactive complex. The RalB-Exo84 complex contains functional ULK and PI3K complexes,
suggesting that this complex both general autophagy and is active during starvationthe selective autophagic induced autophagy. In degradation of bacteria. contrast, upon inhibition of Together, the data are consisRalB signaling, Rubicon, an tent with the idea that the exoinhibitor of autophagy, assocyst functions as a scaffold for ciates with Exo84. The Sec5the core autophagy machinery ULK-PI3K complex correlates in mammalian cells. with the inactive autophagy Earlier work showed that state and is more abundant RalB competes with phosunder nutrient-rich condiphatidylinositol 3,4,5-tritions. These observations led sphosphate (PIP3) for binding the authors to propose a to the pleckstrin homology model for the activation of (PH) domain of Exo84 (Mosautophagy (Figure 1). They kalenko et al., 2003). It is suggest that assembly of the therefore exciting to hypotheULK and PI3K complexes on size that the binding of RalB Exo84 triggers autophagy by to Exo84 may trigger the generating an autophagymovement of the complex active complex. In contrast, from a PIP3-enriched environment, such as the plasma interaction between these membrane or recycling endocomplexes and Sec5 creates somes, to the autophagoan autophagy-inactive comsome assembly site. plex that is either a preinitiation The precise mechanism complex unable but poised to linking the activities of the trigger autophagy or is a signal ULK and PI3K complexes to termination complex for the the elongation and completion process. Consistent with this Figure 1. A Model for Exocyst Function in Autophagy of the isolation membrane by model, endogenous mammaBodemann et al. (2011) provide evidence that the exocyst, a protein complex the Ubl and mAtg9 cycling lian target of rapamycin involved in post-Golgi protein traffic, may function as a scaffold for the assembly of autophagy complexes. The authors suggest the following model complexes has been unclear complex 1 (mTORC1), which for activation of autophagy. Under nutrient-rich conditions, an exocyst sub(Mehrpour et al., 2010). The inhibits autophagy through complex containing the Sec5 protein associates with the Unc-51 like kinase direct interaction of these inactivation of the ULK com(ULK) and phosphatidylinositol-3 kinase (PI3K) complexes at the perinuclear complexes with the exocyst plex (Mehrpour et al., 2010), region forming an autophagy-inactive complex. Induction of autophagy (e.g., in response to starvation) leads to the activation of the Ras-like small GTPase, complex may provide this is present only in the autoRalB. The activated RalB interacts with the exocyst, promoting the replacemissing link. The work of Bodphagy-inactive complex. ment of Sec5 by another exocyst component, Exo84, and formation of an emann and colleagues shows The cellular localization of active autophagy complex that includes the ubiquitin-like (Ubl) conjugation system and the ubiquitin-like molecule LC3. This autophagy-active complex that all of these complexes complex components under localizes in cytosolic dots that could correspond to the isolation membrane. assemble on a common scafdifferent conditions suggests The Exo84 exocyst subcomplex may bring together complexes of the core fold, a recurring theme exthat the transition from an autophagic machinery or facilitate their concerted action. The exact subunit compositions of the autophagy-inactive and autophagy-active exocyst subploited by other signaling autophagy-inactive to an complexes remain unknown. PI3P, phosphatidylinositol 3-phosphate; PI, systems (Shaw and Filbert, autophagy-active complex phosphatidylinositol. WIPI-1 is a WD40 repeat autophagy protein that inter2009), suggesting that the exomay involve a change in localacts with phosphoinositides such as PI3P. cyst may coordinate molecular ization. Bodemann et al. find events in autophagy. that, under nutrient-rich conThe primary known function ditions, RalB associates with an exocyst subcomplex containing PI3K complex). These results suggest of the exocyst is the targeting and tethering Sec5, and components of the PI3K au- that activation of RalB and recruitment of post-Golgi vesicles to plasma memtophagy complex colocalize with RalB in of Exo84 may trigger translocation of the brane domains. By analogy, the exocyst perinuclear regions of the cell. Upon star- autophagy complex to sites of autopha- may also provide a targeting site for the autophagy machinery, perhaps tethering this vation and activation of RalB, Exo84 gosome membrane formation. replaces Sec5. Under these conditions Epistasis analysis also supports an early machinery to the isolation membrane for (starvation), the authors observe the role for RalB in triggering autophagy autophagosome expansion. Exo84 exocyst subcomplex at cytosolic upstream of the initial nucleation step dots, along with the PI3K autophagy mediated by the ULK complex. FurtherACKNOWLEDGMENTS complex, components of the Ubl com- more, depletion of several components of plex, and phosphatidylinositol 3-phos- the active exocyst complex (Sec3, Sec8, Our autophagy expertise is made possible by NIH phate (PI3P, the product of an active Exo70, Exo84, but not Sec5) suppresses grant GM069373 to S.S. Cell 144, January 21, 2011 ª2011 Elsevier Inc. 173
REFERENCES Bodemann, B.O., Orvedahl, A., Cheng, T., Ram, R.R., Ou, Y.-H., Formstecher, E., Maiti, M., Hazelett, C.C., Wauson, E.M., Balakireva, M., et al. (2011). Cell 144, this issue, 253–267. Bodemann, B.O., and White, M.A. (2008). Nat. Rev. Cancer 8, 133–140. Chien, Y., Kim, S., Bumeister, R., Loo, Y.M., Kwon, S.W., Johnson, C.L., Balakireva, M.G., Romeo, Y.,
Kopelovich, L., Gale, M., Jr., et al. (2006). Cell 127, 157–170. He, B., and Guo, W. (2009). Curr. Opin. Cell Biol. 21, 537–542. Mehrpour, M., Esclatine, A., Beau, I., and Codogno, P. (2010). Cell Res. 20, 748–762. Moskalenko, S., Henry, D.O., Rosse, C., Mirey, G., Camonis, J.H., and White, M.A. (2002). Nat. Cell Biol. 4, 66–72.
174 Cell 144, January 21, 2011 ª2011 Elsevier Inc.
Moskalenko, S., Tong, C., Rosse, C., Mirey, G., Formstecher, E., Daviet, L., Camonis, J., and White, M.A. (2003). J. Biol. Chem. 278, 51743– 51748. Munson, M., and Novick, P. (2006). Nat. Struct. Mol. Biol. 13, 577–581. Shaw, A.S., and Filbert, E.L. (2009). Nat. Rev. Immunol. 9, 47–56. Yang, Z., and Klionsky, D.J. (2010). Curr. Opin. Cell Biol. 22, 124–131.
Leading Edge
Primer High-Resolution Genome-wide Mapping of the Primary Structure of Chromatin Zhenhai Zhang1 and B. Franklin Pugh1,* 1Center for Comparative Genomics and Bioinformatics, Center for Eukaryotic Gene Regulation, Department of Biochemistry and Molecular Biology, The Pennsylvania State University, University Park, PA 16802, USA *Correspondence:
[email protected] DOI 10.1016/j.cell.2011.01.003
The genomic organization of chromatin is increasingly recognized as a key regulator of cell behavior, but deciphering its regulation mechanisms requires detailed knowledge of chromatin’s primary structure—the assembly of nucleosomes throughout the genome. This Primer explains the principles for mapping and analyzing the primary organization of chromatin on a genomic scale. After introducing chromatin organization and its impact on gene regulation and human health, we then describe methods that detect nucleosome positioning and occupancy levels using chromatin immunoprecipitation in combination with deep sequencing (ChIP-Seq), a strategy that is now straightforward and cost efficient. We then explore current strategies for converting the sequence information into knowledge about chromatin, an exciting challenge for biologists and bioinformaticians. Chromatin regulates remarkably diverse processes in eukaryotic organisms, from development and disease progression to cognition and aging. Not surprisingly then, deciphering how chromatin directs gene expression continues to be a major research priority. Chromatin is genomic DNA compacted into chromosomes by histone proteins, but RNA and other proteins are also important constituents. The fundamental repeating unit and building block of chromatin is the nucleosome; each nucleosome contains 147 base pairs of DNA wrapped approximately twice around a protein core and consists of two copies each of the four histones H2A, H2B, H3, and H4. Nucleosomes may contain histone variants, such as H2A.Z and H3.3, which are found at active genes (Malik and Henikoff, 2003). In addition, nucleosomes are often decorated with posttranslational modifications at specific amino acids of their histones. These modifications include acetylation, methylation, phosphorylation, ubiquitination, citrullination, SUMOylation, and ADP ribosylation (Kouzarides, 2007). The histone variants and modifications impart functionality to nucleosomes, such as the regulation of gene expression and the compaction of chromatin into higher-ordered structures. Indeed, a histone code may exist in which specific combinations of histone variants and modifications provide landmarks for gene regulatory proteins. These landmarks designate not only the start and end of genes but also the transcriptional status of a gene (Jenuwein and Allis, 2001). For example, trimethylation at lysine 4 on histone H3 (H3K4me3) marks the 50 region of active genes, whereas trimethylation of lysine 36 (H3K36me3) marks the middle-30 region of these genes. These histone modifications and variants may provide a ‘‘global positioning system’’ for assembly of proteins that regulate chromatin and the transcription machinery. The primary structure of chromatin consists of nucleosomes organized in and around genes (Figure 1) with an array of
uniformly spaced nucleosomes beginning at a fixed distance immediately downstream of most transcriptional start sites (Jiang and Pugh, 2009; Yuan et al., 2005). However, chromatin is more than this simple ‘‘beads-on-a-string’’ model. Nucleosomes dynamically interconvert to more compacted units, which further fold into higher-order structures. Histone modifications, in particular phosphorylation, likely direct chromatin folding and compaction, leaving genomic regions to reside in specific domains of the nucleus. In this Primer, we introduce experimental and analytical strategies for genome-wide characterization of chromatin at its primary organizational level—nucleosome positioning, occupancy, variant composition, and modifications. (For genomewide methods that map long-range interactions in higher-order chromatin, we refer readers to Lieberman-Aiden et al., 2009.) First, we describe examples of chromatin research that have or may benefit from nucleosome mapping studies, such as stem cell reprogramming and cancer therapeutics. Then we describe experimental considerations for mapping nucleosomes, and we conclude with a discussion of the computational strategies used to analyze large nucleosomal datasets. Chromatin Mapping Impacts Diverse Research Areas Studies dedicated to characterizing the primary structure of chromatin on a genomic scale aim to understand (1) how nucleosomes become organized across a genome; (2) how this organization influences evolution; and (3) how nucleosome organization regulates genes and other chromosomal elements, ultimately in relation to their impact on human health. Basic Organization of Nucleosomes We know that general features of DNA sequences either favor or disfavor nucleosome formation. For example, sequences with high GC content or with AA or TT dinucleotides in periodic 10 base-pair intervals favor nucleosome formation, whereas Cell 144, January 21, 2011 ª2011 Elsevier Inc. 175
Figure 1. Chromatin Architecture The primary structure of chromatin can be thought of as ‘‘beads-on-a-string’’ with uniformly spaced arrays of nucleosomes at a fixed distance downstream of transcriptional start sites. With the exception of specific regulatory situations, intact nucleosomes generally avoid the core promoter region, where the transcription machinery assembles. These nucleosome-free regions provide an opportunity to regulate gene expression at steps beyond simple promoter access, for example through elongation control of RNA polymerase II (Core and Lis, 2008). The protein core of nucleosomes is composed of histones, which often contain posttranslational modifications on specific amino acids and can be replaced by transcription-linked histone variants (dark blue and purple). As depicted, chromatin also folds into more compact structures aided by certain histone modifications.
sequences with tracts of deoxyadenosine nucleotides (poly (dA:dT)) disfavor nucleosome deposition (Hughes and Rando, 2009; Jiang and Pugh, 2009; Segal and Widom, 2009). However, beyond these general principles, the details for how DNA sequence and cellular factors influence the positioning, occupancy, and other properties of nucleosomes are still unknown (Figure 2A). Many sequence-specific DNA-binding proteins determine their genomic locations by making precise molecular interactions with as few as 6–8 base pairs of a DNA. An equivalent level of specificity applied to nucleosomal DNA but dispersed over its 147 base pairs would be difficult to discern. Moreover, unlike sequence-specific DNA-binding proteins, nucleosomes do not adopt a single position. In a population of molecules, such positions can be quite variable or ‘‘fuzzy’’ (Figure 2A). As nonoverlapping ‘‘beads-on-a-string,’’ the position of one nucleosome restricts the possible positions of adjacent nucleosomes. Consequently, the determinants of a nucleosome position may have distant origins, being propagated through adjacent nucleosomes. Furthermore, chromatin-remodeling complexes, such as SWR1, ISW2, and SWI/SNF, directly regulate nucleosome composition, positions, and occupancy levels, respectively. We know that these protein machines use the energy of ATP hydrolysis to drive nucleosomes to override intrinsic preferences for DNA sequence, but we do not know how they contribute to the organization that predominates in and around genes. Experiments aimed at understanding this question often delete or deplete remodeling complexes in vivo and then assess how the absence of these factors affects nucleosome organization. 176 Cell 144, January 21, 2011 ª2011 Elsevier Inc.
Nucleosome Organization Influences Evolution Nucleosome organization in vivo is not random, and it is now clear that DNA accessibility imparted by nucleosome positions alters DNA susceptibility to mutations, insertions, and deletions. This differential susceptibility shapes the genomic landscape, with mutations tending to be on nucleosomal DNA and insertions and deletions tending to be in linker regions (Sasaki et al., 2009). Nucleosome organization might impact human diversity, whereby single-nucleotide polymorphisms tend to be enriched in nucleosome-free promoter regions near nucleosomal edges (Schuster et al., 2010). Further research in this area may be directed at understanding how nucleosome organization shapes the evolution of promoter and enhancer elements, including sequence conservation of cis-regulatory elements. Nucleosome Organization Regulates Gene Expression Repositioning a nucleosome by as little as a few base pairs may be sufficient to change the accessibility of a DNA regulatory element. If the element is located on linker DNA between nucleosomes, then it may be accessible (Figure 2B). If the element resides on nucleosomal DNA, then it may be inaccessible, particularly if the helical nature of DNA faces the site inward toward the histone core (Jiang and Pugh, 2009; Segal and Widom, 2009). An alternative means for enhancing the accessibility of regulatory elements and coding sequences is through a complete or partial dismantling of nucleosomes. Such remodeling has been characterized by perturbing the cellular environment and mapping the resulting nucleosome reorganization (Schones et al., 2008). For example, genes that are induced by heat shock tend to lose nucleosomes in their promoter regions (Shivaswamy
Figure 2. Nucleosomal Properties Measured by High-Resolution Mapping Studies (A) Numerous properties of individual nucleosomes can be extracted from histone mapping studies, including the spacing between nucleosomes, the presence of histone variants and posttranslational modifications, nucleosome fuzziness, occupancy, and position relative to a genomic feature. Fuzziness is the degree to which a nucleosome deviates from its consensus position in a population measurement. Occupancy is a measure of nucleosome density. (B) Even a small change in a nucleosome’s location can alter a sequence’s accessibility to regulatory proteins. This schematic illustrates the rotational accessibility and inaccessibility of the DNA major groove on the surface of the core histone complex.
et al., 2008). Experiments aimed at understanding the underlying mechanism of remodeling involve depleting or deleting remodeling factors and then examining how nucleosome positions and occupancy levels change. One general rule emerging from these experiments is that robust transcriptional activity involves nucleosome depletion whereas transcriptional regulation may involve nucleosome repositioning. Histone modifications and variants have emerged as a fascinating yet still enigmatic means by which nucleosomes regulate gene expression. Histone variants, such as H2A.Z and H3.3, as well as certain acetylation sites, such as H3K9, 14 and H4K5, 8, 12, 16, create nucleosomes that may facilitate the eviction and/or repositioning of nucleosomes during transcription. Modifications, such as H3K4me3 and H3K36me3, bind proteins involved in the transcription cycle, whereas other marks, such as H3K9me3 and H3K27me3, bind proteins that create inaccessible repressive chromatin (Kouzarides, 2007). Furthermore, other modifications mark the locations of transcriptional enhancers (Heintzman et al., 2009). With more than 50 different histone modifications available, the potential for combinatorial control is bewildering, and deciphering this code will certainly be a focus of future research for many years. Nucleosomes at different positions in the genome serve unique functions. The first nucleosome downstream of a transcriptional start site may regulate the accessibility of the start site and/or the ability of RNA polymerase II to progress into a productive elongation state (although the evidence for such
linkages has been correlative rather than causative). In contrast, the first nucleosome upstream of the transcriptional start site may regulate the accessibility of cis-regulatory elements that bind sequence-specific transcription factors. Nucleosomes in the middle of genes may prevent spurious transcription initiation that might otherwise generate truncated gene products. Histone variants and modifications are selective to specific nucleosome positions (Figure 1) and thus are likely to endow nucleosomes with position-relevant functions. Because we do not currently know all the mechanistic steps of a transcription cycle (or the order of the steps), we have yet to learn how the combinatorial configuration of histone variants and modifications participates in transcriptional regulation. The primary organization of nucleosomes across a genome may be largely invariant from cell type to cell type. However, highly targeted changes in positioning, occupancy, histone composition, and modifications probably help define cell types. Therefore, current emphasis is being placed on generating genome-wide maps of histone modification states as cells or organisms undergo developmental programs (Shi, 2007). In this regard, the ENCODE (ENCyclopedia Of DNA Elements) project has provided a major boost for the field, as it reports on a large number of modification states across model cell lines (Birney et al., 2007). From this and other publicly generated data, a multitude of questions can be addressed. What modification states are associated with tissue differentiation, cell identity, and epigenetic inheritance? How are these modifications ‘‘read’’ to elicit such programs? For example, one study reports that cells already committed to a lineage have more promoter regions with repressive histone modification marks than embryonic stem cells (Hawkins et al., 2010). Are these marks reducing options for pluripotency by locking down promoters? Nucleosome Organization’s Influence on Human Health We are only beginning to understand the critical roles that chromatin plays in human health and disease. For example, induced pluripotent stem (iPS) cells have the potential to regenerate damaged tissue, but it is becoming clear that such cells, which originate from adult tissue, are not entirely equivalent to embryonic stem (ES) cells. Differences in these cells’ chromatin appear to be paramount. For example, at least in one case, regional states of repressive chromatin differ between ES and iPS cells, but reactivation of such regions allows the iPS cells to behave as ES cells when they are used to regenerate mice (Stadtfeld et al., 2010). A key aspect of keeping ES cells pluripotent is the maintenance of ‘‘open’’ chromatin states, which are generally depleted of repressive chromatin marks and rendered more dynamic by ATP-dependent chromatin remodelers, such as the chromodomain-helicase-DNA-binding protein 1 (CHD1) (Gaspar-Maia et al., 2009). Open regions may be thought of as providing ES cells with many transcriptional options for differentiation that would otherwise be eliminated by a closed chromatin state. Numerous studies are currently devoted to defining the architecture of these open states, including the identification of nucleosome positions, depletion levels, and modification states. Analogous to maintaining stem cells, chromatin of cancerous cells is also reprogrammed, and many studies are focused on mapping where chromatin regions change between ‘‘open’’ and ‘‘closed’’ states in cancer cells. Such maps may better Cell 144, January 21, 2011 ª2011 Elsevier Inc. 177
define distinct cancer subtypes to facilitate clinical treatments. For example, lower levels of certain histone modification states, such as H3K4me2, H3K9me2, and H3K18ac, have been strong prognostic indicators of the treatment outcomes for patients with pancreatic cancer (Manuyakorn et al., 2010). Cancer cells can become resistant to chemotherapies, but interestingly, such drugs become more effective when used in combination with inhibitors of chromatin modifiers, such as histone deacetylation (Sharma et al., 2010). In addition to its role in cancer, histone acetylation is also a key component of memory and behavior. For example, H4K12ac in the hypothalamus has been associated with memory formation (Peleg et al., 2010). Loss of this mark correlates with cognitive decline in mice, which can be restored with inhibitors of histone deacetylation. Similarly, an inability to methylate H3K9 has been linked to impaired learning (Schaefer et al., 2009). Histone deacetylase inhibitors have also been used to treat numerous neurological disorders, including anxiety and depression (Gundersen and Blendy, 2009). Future experiments will likely map histone modification states in relevant portions of the brain when mice are subjected to memory and behavioral tests. From yeast to mammals, life span has been linked to specific states of histone acetylation and methylation. Sirtuins, a class of histone deacetylases, promote gene silencing and longevity (Guarente, 2000). More recently, loss of H3K4 methylation and maintenance of H4K16 acetylation of histones have been attributed to increased life span (Dang et al., 2009; Greer et al., 2010). Genome-wide nucleosome mapping of histone modification states and nucleosome organization in old versus young cells and in cells with altered abilities to add or remove relevant acetylation and methylation marks may help to locate key genomic changes that alter life span. Experimental Considerations: From Sample to Sequence Tags Embarking on a genome-wide mapping project of chromatin is easier than one might expect, but finishing the mapping may be harder than anticipated. A basic molecular biology laboratory with access to either commercial or in-house whole-genome sequencers can prepare nucleosomes for mapping and determine their locations in a genome at single-nucleosome accuracy. The difficult but most rewarding part is turning these maps into knowledge about chromatin. Here we discuss strategies for conducting genome-wide mapping of nucleosomes, with a focus on using micrococcal nuclease (MNase) to generate mononucleosomes and then deep sequencing to identify their locations. We believe that MNase ChIP-Seq is probably the most effective means of mapping the primary structure of chromatin. However, first we briefly describe a few other genome-wide mapping strategies. Overview of Available Strategies Data from MNase ChIP-Seq provide population averages from a large number of cells. To map nucleosome configurations on individual DNA molecules, ectopic expression of DNA methyltransferases may be used in vivo or added to nuclei. Nucleosomal DNA is identified because its sequence is eventually altered, whereas linker DNA is not. For example, the M.CviPI methyltransferase methylates cytosine in 50 -GC-30 dinucleotides 178 Cell 144, January 21, 2011 ª2011 Elsevier Inc.
when it is present in linker DNA (Pardo et al., 2010). This methylcytosine, unlike cytosine, is protected against bisulfite conversion to uracil (and ultimately thymine) in vitro. After traditional Sanger sequencing, the configuration of a nucleosomal array on the original DNA molecule is inferred from the sequence. GC dinucleotides are inferred to be nucleosome-free, whereas 150 base-pair spans of GTs that are GCs in the reference (i.e., untreated) genome are interpreted as nucleosomal. The value of mapping an array of nucleosomes on a single DNA molecule is that adjacent nucleosome positions may appear to overlap in a population average but are actually mutually exclusive when examined on a single-molecule basis. Currently, this strategy has not been applied on a genomic scale, as it optimally requires high-throughput long-read (>1000 nucleotides) sequencing. Regions depleted of nucleosomes are candidates for regulatory regions. Therefore, if the primary purpose of chromatin mapping is to screen for nucleosome-depleted regions in many cell types or under various conditions, then FAIRE (formaldehyde-assisted isolation of regulatory elements) may be the appropriate method (Giresi and Lieb, 2009). FAIRE simply depends upon the differential partitioning of nucleosomal and nucleosome-free DNA in phenol-chloroform and aqueous phases. Thus, the major advantages of FAIRE are its simplicity and cost efficiency. However, its resolution is low compared to mapping the location of individual nucleosomes. DNase I hypersensitivity has been a classical means of mapping regions of accessible chromatin. Like FAIRE, it is a strategy used on a genomic scale in the ENCODE project (Hesselberth et al., 2009). However, due to frequent cleavages within nucleosomal DNA, nucleosome positions may be more difficult to discern when DNase I is used instead of MNase. Moreover, DNase I involves many sample handling steps and is complicated by technical variation in DNA digestion. Initial Preparation of Mononucleosomes From an operational perspective, there are two types of starting material for mapping nucleosomes: tissue excised from a multicellular eukaryotic organism and minimally aggregated cells, including those drawn from blood, grown in tissue culture, or cultured as free-living microorganisms. Excised tissue may contain a heterogeneous mixture of cells, which may obscure chromatin patterns specific to a cell type. Cellular heterogeneity may be minimized by highly selective and precise tissue excision, which may necessitate acquisition of less material. Although the minimum amount of excised material required to generate nucleosome maps is not known, a lower limit of 10,000 cells drawn from blood may provide a guide (Adli et al., 2010). More commonly, large numbers of cells are easily collected from tissue culture or microorganisms, such as yeast for which 107–108 cells are used. More sophisticated methods may be used to isolate tissue-specific nuclei (Deal and Henikoff, 2010). The production of genome-wide nucleosome maps has variously used or avoided formaldehyde crosslinking (Figure 3, step 1). Formaldehyde essentially ‘‘freezes’’ existing proteinprotein and protein-nucleic acid interactions in place, thereby preserving the in vivo status of interactions, without adverse effects on nucleosomes (Fragoso and Hager, 1997). Without crosslinking, nucleosomes may reorganize during cell
Figure 3. Flow Chart for Nucleosome Preparation, Mapping, and Analysis Images exemplify or illustrate the type of material or data at each stage. Steps 5 and 6 may be performed in either order. Left side, bottom image: Reprinted by permission from Macmillan Publishers Ltd., Nat. Rev. Genet., Jiang and Pugh, 2009. Right side, second image from the bottom: Reprinted by permission from Macmillan Publishers Ltd., Nat. Genet., Lee et al., 2007. Right side, bottom image: Reprinted by permission from Macmillan Publishers Ltd., Nat. Methods, Adli et al., 2010.
harvesting, chromatin preparation, and chromatin fragmentation. However, for most genes in yeast, we and other laboratories have found that nucleosome organization is largely the same in the presence or absence of formaldehyde when MNase is used for chromatin fragmentation (Kaplan et al., 2009). Nevertheless, we have also found genomic regions where nucleosome organization varies in the absence of formaldehyde, and thus, we recommend a simple formaldehyde crosslinking step.
Yeast and plants have cell walls, which require disruption through mechanical breakage (e.g., vigorous vortexing with glass beads) or enzymatic digestion (Albert et al., 2007; Rando, 2010) (Figure 3, step 2). Tissue or small whole animals, such as worms, may be disrupted by grinding of frozen material (Kolasinska-Zwierz et al., 2009). Tissue culture cells in sufficient quantities may be disrupted by douncing cells in a hypotonic buffer. If the amount of material is low, then it may be more practical to lyse with an ionic detergent, such as SDS, in combination with a freeze-thaw cycle (Adli et al., 2010). However, because SDS disruption is not compatible with subsequent MNase digestion, the chromatin must be fragmented by low-resolution sonication. Chromatin Fragmentation The method of chromatin fragmentation is critical to producing nucleosome maps of a desired resolution (Figure 3, step 3). Sonication produces DNA fragments ranging from 200 to 700 base pairs. The heterogeneity of fragment size and cleavage sites makes sonication suitable for characterizing chromatin states over wide regions encompassing many nucleosomes, but it is not optimal for mapping individual nucleosomes. MNase digestion, on the other hand, produces DNA fragments with ends that correspond to the ends of nucleosomes and, thus, produces maps with very high resolution. One potential limitation of MNase digestion is its bias toward cleaving at A or T more frequently than at G or C. However, extensive MNase digestion that predominantly produces mononucleosomes largely, but not entirely, overcomes this bias because even unfavorable cleavage sites become cleaved. Furthermore, residual bias can be computationally compensated (Albert et al., 2007). A limitation of extensive MNase digestion is the production of subnucleosomal-sized DNA fragments, particularly at highly transcribed genes where the DNA on the surface of remodeled or partially disassembled nucleosomes may be more exposed (Weiner et al., 2010). The lack of nucleosome-sized DNA fragments in such regions may be interpreted as being entirely nucleosome free, as opposed to the presence of remodeled or partial nucleosomes that escape detection. Different chromatin samples and preparations of MNase (i.e., commercial lots) may yield different degrees of MNase digestion. Therefore, it is prudent to titrate the MNase to achieve 80% of the DNA as mononucleosomal, which is detected by electrophoresis as a band at 150 base pairs (Figure 3, step 3) (Rando, 2010). In addition, pooling chromatin that has been fragmented to various extents from an MNase titration may help avoid biased isolation of mononucleosome subpopulations that differ in accessibility. Fragmentation by sonication releases insoluble chromatin fragments from the pellet to the supernatant. MNase treatment solubilizes mononucleosomes in yeast but is often less efficient in fly and mammalian systems. Therefore, a brief sonication in these latter two systems improves solubilization, without creating additional fragmentation. Alternatively, salt extractions of increasing strength can be used to selectively solubilize ‘‘active’’ chromatin (Henikoff et al., 2009). Gel analysis of histones and DNA released to the supernatant versus that retained in the pellet can be conducted to confirm full extraction. Cell 144, January 21, 2011 ª2011 Elsevier Inc. 179
Chromatin Immunoprecipitation Perhaps the most frequent use of nucleosome mapping is to characterize the distribution of histone modification states or histone variants. In these cases, immobilized antibodies against the particular modification or variant are necessary to immunoprecipitate (or ‘‘ChIP’’) chromatin fragments possessing the specific modification or variant (Figure 3, step 4) (Liu et al., 2005). Because only a small percentage of DNA becomes crosslinked to histones by formaldehyde, immunoprecipitation should be conducted in the presence of detergent (e.g., 0.05% SDS) to eliminate uncrosslinked DNA. Many antibodies, such as those against H3K4me3 and H2A.Z, are commercially available, providing a level of standardization and quality control of antibody specificity. However, one limitation of any antibody targeted against a modification is its potential to cross-react with the same or a similar modification located at other sites. Alternatively, an antibody may not recognize its epitope if a nearby amino acid is also modified, and such interfering modification might be present in only a subpopulation of the nucleosomes. Synthetic peptides harboring the modification or potentially confounding secondary modifications can be used to verify antibody specificity. Detection Historically, genome-wide detection of chromatin began with the use of low-resolution DNA microarrays in yeast. PCR probes of each intergenic and genic region were arrayed onto glass slides upon which fluorescently labeled ChIP material was hybridized (reviewed in Jiang and Pugh, 2009). Higher resolution was achieved with microarrays containing overlapping 50-nucleotide probes tiled every 20 base pairs across a small region of the yeast genome. Next high-density microarrays that spanned entire genomes were developed. These arrays, which remain in use today probably for a limited time, can generate maps of individual nucleosomes but with lower resolution compared to deep sequencing. Deep sequencing has the additional advantages of less background, better coverage, and a larger dynamic range compared to microarrays. That said, the fuzziness of nucleosome positions over a population (Figure 2A) precludes full realization of deep sequencing’s intrinsic high resolution. Regardless of the fragmentation method or whether ChIP is used, the resulting DNA should be gel purified in the 120–170 base-pair range to remove nonspecific, subnucleosomal, and polynucleosomal DNA fragments (Figure 3, step 5). Currently, deep sequencing of nucleosomal DNA requires library preparation, which essentially involves ligating DNA adapters to the ends of gel-purified mononucleosomal DNA (Figure 3, step 6). This allows for PCR amplification of the sample and creates a template by which sequencing initiates. By this stage, users typically have given their samples to a sequencing facility, which will construct the libraries for sequencing using kits provided by the manufactures of the sequencing instrument (Figure 3, step 7). Research laboratories that produce large numbers of libraries may develop their own library preparation protocols, which enhance cost efficiency. Adaptor sequences are available at company websites, and their ligation involves standard molecular biology manipulations. In this case, greater DNA yields may be obtained by gel purifying after library preparation and PCR amplification. 180 Cell 144, January 21, 2011 ª2011 Elsevier Inc.
Currently, the Illumina Genome Analyzer and the Applied Biosystems SOLiD sequencers are the most widely used deep sequencers for this type of work. Although a variety of deep sequencers will likely be available in the near future, the key instrument parameter for nucleosome mapping (and for ChIP-Seq in general) is not the read length but rather the tag count, which is the number of different DNA molecules that can be sequenced and mapped to the reference genome. In general, a technology platform should meet these minimum specifications: minimal steps for library construction, a sequencing read or tag length of 35 nucleotides, read accuracy of >99%, turnaround time of less than a few days, and a cost of under US $10,000 per run. In principle, biases during ligation, post-construction PCR amplification, gel purification, and sequencing can result in biased tag production, which may influence the apparent occupancy level and position of nucleosomes (Stein et al., 2010). Nevertheless, such biases may be compensated computationally because they manifest as anomalously high tag counts at specific genomic coordinates. For example, setting an upper limit on the normalized tag counts at a particular coordinate may correct such statistical outliers (Kaplan et al., 2009). In practice, sequencing bias may be rather innocuous because data are often aggregated in a way that eliminates outliers. Sequencing Tags Sample processing that includes MNase digestion, immunoprecipitation, and gel purification of mononucleosomes eliminates nonspecific background contamination of genomic DNA, which would otherwise degrade the quality of the maps. As such, each sequencing tag represents a measured nucleosome position, generally without the need for background correction. A general rule of thumb is that the number of sequencing tags needed to uniquely identify >90% of all nucleosomes is minimally ten times the number of estimated nucleosomes. An estimated number of total nucleosomes is the genome size divided by 200 (i.e., the average base-pair distance covered by a nucleosome core particle plus linker). Thus, complete yeast nucleosome maps require at least 600,000 tags, whereas human nucleosome maps require at least 150 million tags. However, more or fewer tags may be needed depending upon the goal of the experiment. If the goal is to measure occupancy levels, then 3–5 times more tags may be required to provide robust quantitative numbers of tags per nucleosome position or per genomic coordinate. If data are to be aggregated, for example by averaging the distribution of tags around a collection of genes, then substantially fewer tags may be sufficient. Indeed, not every nucleosome would need to be detected. Such minimal coverage is cost efficient when many experiments are conducted simultaneously, such as screening samples or titrating conditions. Because only a small portion of the library is sequenced, more coverage can be achieved by sequencing more of the library as needed. Similarly, histone modification states typically occur at only a fraction of all nucleosomes and, thus, in principle, require fewer tags. The number of needed sequencing tags for each sample must dovetail with the minimal sequencing ‘‘bandwidth’’ (Figure 4). Each channel of the sequencer flow cell (the current Illumina sequencer has 8 channels and the current SOLiD sequencer has 1–8 channels) represents the minimal bandwidth of the
Figure 4. How Multiplexing Can Influence Tag Production Left: The standard practice in ChIP-Seq is to index or ‘‘barcode‘‘ each genomic sample of nucleosomal DNA with a unique DNA sequence of 5–10 nucleotides. The barcode is ultimately sequenced and used to associate each tag with the sample from which it came, when many different samples are pooled together. One option is to PCR amplify each DNA sample, then pool equal mass proportions to generate equal numbers of tags for each nucleosomal sample. The problem with this approach is that a sample that might be expected to have a very low tag count, such as a negative control lacking an antibody or epitope, will yield approximately the same number of tags as real test samples. This could give the erroneous impression of high background in the test samples. Right: An alternative strategy is to pool samples prior to PCR amplification, based upon mixing equivalent numbers of cells (or some other metric of equivalency between samples). Then, the proportionality of tags between samples will, to a first approximation, remain constant. The problem with this approach is that any loss of sample or excess DNA contamination in a sample at any stage prior to pooling (including cell harvesting, chromatin immunoprecipitation, or library construction) will carry through to the end. As a result, tags from different test samples, which might be expected to have similar tag counts, could vary widely. Thus, the risk is that some samples may not yield enough tag counts to conduct the appropriate analysis.
sequencer. If a sequencer delivers, for example, 40 million mappable tags as its minimal bandwidth per channel, and the user requires 10 million tags per sample, then 4 multiplexed samples can be placed into each channel. Sample multiplexing, which is also called indexing or barcoding, is achieved by using commercially designed adapters. These adapters contain a unique predefined 5–10 nucleotide DNA sequences used to identify the sample. Current commercial systems allow up to 96 barcodes. Once indexed, samples can then be pooled in any desired ratio to achieve the requisite number of tags deliverable by the channel (but subject to the caveats illustrated in Figure 4). The combination of sample pooling and judicious apportioning of tags could drive sequencing costs below US $50 per sample. Technical improvements in deep sequencing will continue to increase the number and length of sequenced tags. For standard mapping, sequencing beyond the minimal length that is required to uniquely identify a tag in the genome offers little advantage and in fact has a number of disadvantages. The main drawback is cost. Once a tag is uniquely mapped in the genome, additional sequencing cycles add cost without adding more tags to the dataset. Other unnecessary disadvantages of longer reads include slower instrument turnaround and greater data storage needs. Sequencing error rates, in general, are not a significant issue because tags need only to be uniquely identified in the reference genome and thus can have multiple errors without impacting its uniqueness. This contrasts with detection of single-nucleotide polymorphisms or de novo sequencing of genomes in which accuracy is critical. The point of diminishing returns on sequence length is approximately 25–27 nucleotides. However, sequencing kits typically
produce 35 nucleotide tags, which represent a good compromise between the need for unique identification and the drawbacks of longer reads. One type of sequencing run, called ‘‘fragment’’ by an Applied Biosystems term and ‘‘single read’’ by Illumina, identifies only one end of the nucleosomal DNA molecule for each tag. In contrast, ‘‘paired-end’’ sequencing simultaneously identifies both ends in each tag. In principle, paired-end sequencing provides more accurate maps. However, the added accuracy may not be worth the roughly 2-fold increase in sequencing costs if it is not needed to address the questions at hand. For nucleosome mapping, both ends of a consensus nucleosome are already measured separately in a population of molecules detected by a fragment or single-read run. Moreover, consensus nucleosomes over a population are not at fixed positions but are rather ‘‘fuzzy’’ (Figure 2A), and thus the added accuracy of pairend sequencing may be moot. Paired-end and longer-read sequencing is advantageous when mapping nucleosomes in repetitive or low-complexity genomic regions, where the additional information provides a greater probability of uniquely identifying location. Bioinformatics: Turning Sequences into Nucleosomes Mapping DNA Sequence Tags Current Applied Biosystems SOLiD and Illumina GA/HiSeq platforms produce raw photographic image files of fluorescence intensities. The fluorescence resides on a two-dimensional surface that represents a detected nucleotide (Illumina) or dinucleotide (Applied Biosystems) incorporated into a group or cluster of identical clonally amplified DNA molecules. An entire flow cell houses hundreds of millions of clusters that undergo Cell 144, January 21, 2011 ª2011 Elsevier Inc. 181
Figure 5. Turning Tags into Nucleosomes Solid red and blue lines represent nucleosomal DNA ready for sequencing. The nucleosomal DNA is produced as MNase-resistant DNA fragments. In a population of molecules, the DNA fragments will have heterogeneous ends due to biases in digestion efficiency at different sequences, as well as the nucleosome not residing at a single position (i.e., ‘‘fuzzy’’ positioning). The asterisk, representing a unique sequence, provides a frame of reference. (A) In ‘‘fragment’’ or ‘‘single-read sequencing,’’ the DNA library is sequenced from only one of the adapters (green) (except in reading the barcode) and in the direction indicated by the blue or red arrows. Consequently, each nucleosome border is measured independently as a population. Tags can then be extended to 147 nucleotides or their 50 ends shifted by 73 nucleotides, as indicated to the right. Either way, the resulting frequency distributions, although looking different, have exactly the same uncertainty. (B) Paired-end sequencing allows both ends of the same DNA molecule to be sequenced. The midpoint of the pair defines the consensus nucleosome midpoint, which can be extended 73 nucleotides in both directions (right side).
20–150 cycles of sequencing (although 35 is the target), ultimately culminating in terabytes of image data. For many sequencing operations, long-term storage of these image files is cost prohibitive, and thus they are kept only short-term. In general, image files become obsolete once they are converted to FASTQ (Illumina) or CSFASTA (Applied Biosystems) files, which delineate the nucleotide sequence (i.e., ‘‘base calls’’). Physical DNA libraries may be kept indefinitely, should resequencing be necessary. Both sequencing platforms have on-board software to split off barcodes and map raw sequence data to a user-selected reference genome. ELAND software, packaged with the Illumina GA, aligns a sequence to a reference genome by first seeding an alignment with the first 32 nucleotides of the tag, then extends the alignment to the total tag length. Applied Biosystems provides the SOLiD System Analysis Pipeline Tool (Corona Lite). Other aligners are also available, including Maq, RMAP, Cloudburst, SOAP, SHRiMP, Bowtie, and BWA (Li and Homer, 2010). SHRiMP and Bowtie are particularly popular. 182 Cell 144, January 21, 2011 ª2011 Elsevier Inc.
From Mapped Tags to Nucleosomes The relevant part of a mapped tag is the genomic coordinate of its 50 end (lowest coordinate on the forward strand, highest coordinate on the reverse strand). It corresponds to a single measured nucleosome border, if the sequenced library originates from mononucleosomal DNA. The sequence specificity bias that is inherent to MNase digestion and incomplete protection of borders by histones collude to create imprecision in a mapped border, which in practice may be largely moot because nucleosome positions are intrinsically imprecise. Nonetheless, the resulting tag can be used to represent a nucleosome in multiple ways (Figure 5A): (1) as a strandspecific nucleosome border (unshifted tag) (Barski et al., 2007); (2) as an entire nucleosome by extending the tag in the 30 direction to a length of 147 base pairs (extended tag) (Kaplan et al., 2009); or (3) as a single coordinate representing a presumed nucleosome midpoint by shifting the 50 end of the tag 73 nucleotides toward the 30 direction (shifted tag) (Albert et al., 2007). For paired-end sequencing, the midpoint between the highest and
lowest coordinates of the two tags defines the nucleosome midpoint (Figure 5B). The midpoint can also be extended 73 nucleotides in both directions to define the presumed nucleosome length. When sonication is used to fragment chromatin, the resulting libraries largely lack single nucleosome precision. Nevertheless, their sequenced 50 ends can be shifted in the 30 direction by the average fragmentation size to estimate the midpoint of the nucleosome. Clusters of tags can be aggregated to define a consensus nucleosome position, which then represents a population average (Figure 3, step 8). For this, our laboratory uses GeneTrack software, which was the first peak calling software developed for mapping nucleosomes or any other data by ChIP-Seq (Albert et al., 2007). GeneTrack converts tag counts at each coordinate into a smoothed Gaussian distribution across multiple coordinates, implementing a user-defined standard deviation. GeneTrack then sums all instances of the distribution to create a smoothed continuous landscape across the genome. Local peaks are then identified, starting with the highest peak. A user-defined exclusion zone (e.g., 147 nucleotides) is centered over the peak to represent the steric exclusion of a nucleosome and prevent the calling of secondary peaks within the exclusion zone. Peak calling can be performed with unshifted tags on each DNA strand separately. The consensus midpoint for the nucleosome is then the midpoint distance between a peak on one strand and the next downstream (30 ) peak but located on the opposite strand. Alternatively, shifted tags from each strand can be first combined, then applied to GeneTrack. The former may be more accurate as it involves position-specific correction factors. The latter involves only a single correction for an entire dataset and may be more appropriate for raw data display in a browser. A number of other ‘‘peak-calling’’ algorithms have been used to define consensus nucleosome positions. Hidden Markov modeling has been applied to microarray data (Yuan et al., 2005) and can infer which DNA segments are occupied by nucleosomes after the algorithm trains on a dataset. On the other hand, template filtering aims to classify peak patterns (Weiner et al., 2010), then shifts peak pairs on opposite strands individually by whatever distances maximize area overlap between the two peaks. In contrast, model-based analysis of ChIP-Seq (MACS) uses empirical modeling of the length of protein-DNA interaction sites in combination with local biases in the genome based on a Poisson distribution (Zhang et al., 2008). Other peakcalling software, which may be applicable to nucleosome mapping, includes PeakFinder, FindPeaks, SISSRs, QuEST, CisGenome, PeakSeq, and Hpeak. Indeed, Laajala et al. (2009) compare these programs for use with ChIP-Seq data. Data Analysis Pipeline Data analysis can be divided into three stages (Figure 3, steps 8–10): primary, secondary, and tertiary analyses. Primary analysis starts with raw base calls, maps the tags to a reference genome, and then identifies peaks (i.e., ‘‘peak calls’’). This processing removes unmappable tags, aggregates the data, and provides some quality assessment of the dataset. Primary analysis, in principle, requires no biological knowledge and can be handled by trained computational staff.
During secondary analysis, nucleosome parameters are extracted from the dataset. These parameters include internucleosomal spacing, the ‘‘fuzziness’’ or variation of individual locations of nucleosomes, distances from a particular sequence element, and the extent to which nucleosomes occupy a region of genomic DNA (Figure 2A). Here knowledge of bioinformatics and genomics is required, particularly an understanding of genome annotation (e.g., how strands, coordinates, and features are defined), organization (e.g., how features, such as genes and regulatory elements are placed), and structure (i.e., how bendability varies across DNA sequences to how chromatin folds into higher-order structures). During tertiary analysis, a dataset is compared to many other experimental datasets (Figure 3, step 10). Examples of these analyses include the extent to which two histone modification states, such H3K27me3 and H3K4me3, colocalize throughout the genome and the distribution of nucleosomes around measured genomic features, such as transcriptional start sites. Tertiary analysis requires extensive biological knowledge to focus on key questions and avoid a seemingly endless number of less informative comparisons. Analysis pipelines can be developed in-house or assisted by online applications such as Galaxy (Goecks et al., 2010). Analysis Strategies: From Coordinates to Nucleosome Organization Multiple metrics of nucleosome organization provide insights into chromatin regulation, including nucleosome occupancy levels, nucleosome ‘‘fuzziness,’’ spacing distance between adjacent nucleosomes, and their distribution around genomic features (Figure 2A). The simplest display of nucleosome organization consists of a browser shot displaying the distribution of sequencing tags for a representative genomic region (Figure 3). A browser shot is attainable by the University of California, Santa Cruz (UCSC) genome browsers (Rosenbloom et al., 2010) and GeneTrack (Albert et al., 2008). The value of a browser shot is that it gives the most intuitive and unfiltered assessment of the data; however, it does represent only an anecdotal and potentially ‘‘cherry-picked’’ example. Nucleosome Occupancy Nucleosome occupancy can be assessed for a consensus nucleosome location or on a per base-pair basis (Figure 5). The former measures the number of shifted tags residing within ± 73 base pairs of a consensus nucleosome midpoint (Mavrich et al., 2008). By this measure, a single occupancy value is attributed to a consensus nucleosome position defined by a single coordinate. Preferential MNase digestion sites and fuzziness of the nucleosome may influence the consensus position but will have little effect on its occupancy level. In contrast, occupancy measured on a per base-pair basis counts the number of extended tags that cover each genomic coordinate without defining any consensus position (Kaplan et al., 2009). By this measure, occupancy levels at coordinates will be influenced by preferential MNase digestion sites and nucleosome fuzziness. Comparing occupancy levels across datasets requires normalization because the number of tags delivered by a sequencer is, to a first approximation, defined by the user Cell 144, January 21, 2011 ª2011 Elsevier Inc. 183
Figure 6. Data Normalization This schematic depicts an immunoblot measuring the bulk levels of chromatin-associated histone H3 and a particular histone modification. The corresponding immunoblot signals in a reference sample are set to 100. Assuming total levels of H3 are constant between samples, the relative level of each modification can be assessed. Nucleosomal tags generated from MNase ChIP-Seq can then be proportionally adjusted to reflect the relative modification state. This does not preclude further normalization on a locus-by-locus basis in which modification densities are calculated for a given amount of core histone (typically, H3) present in the sample.
rather than reflective of any biological property. Normalization simply involves setting the total number of uniquely mappable tags between samples equal. This assumes that the number of nucleosomes between the compared samples is indeed equal in the biological setting. Because different extents of MNase digestion may influence occupancy levels measured on a per base-pair basis (Weiner et al., 2010), digestion uniformity (i.e., 80% mononucleosomal) is a critical parameter. Although the amount of histones present across samples may be approximately constant, a particular histone modification may vary across samples. In such situations, if the total level of a histone modification across the whole genome can be measured independently, for example by immunoblotting, then total sample tag counts can be scaled to reflect this measured level (Figure 6). Measured levels of a particular histone modification (i.e., tag counts) at specific genomic locations have two contributing biological factors: nucleosome occupancy level and the amount of modification per nucleosome. The latter is the desired metric and can be derived by dividing the measured modification level for a consensus nucleosome or genomic interval by the corresponding level of nucleosome occupancy (Figure 6). Normalized levels of nucleosome occupancy or normalized modification densities can be compared across datasets on a per nucleosome basis or over defined intervals, such as every 500 base pairs. Datasets in which a large proportion of the occupancy levels are distributed over a rather narrow range of values will yield poor correlations between datasets. By analogy, any portion of a calm ocean will look like any other portion of a calm ocean. Data can be filtered in order to compare only nucleosomes or intervals that have particularly high or low occupancy levels (Kaplan et al., 2009). However, one caveat to this filtering is that any resulting correlation is applicable only to those regions. Nucleosome Positioning Most nucleosomes are not randomly placed in the genome, but they also do not associate with a particular DNA sequence in the same manner as sequence-specific transcription factors. Nevertheless, many nucleosomes, but not all, reside at preferred locations in the genome. The degree to which a cluster of nucleosomal tags deviates from its consensus position is a measure of its positioning or ‘‘fuzziness’’ (Albert et al., 2007) (Figure 2A). The 184 Cell 144, January 21, 2011 ª2011 Elsevier Inc.
fuzzier a position is, the less meaningful is its assigned consensus position. Multiple methods of measuring nucleosome fuzziness have been used (Albert et al., 2007; Kaplan et al., 2009; Weiner et al., 2010; Zhang et al., 2009). Nucleosome positions relative to a set of genomic sequences or features can be evaluated by plotting a frequency distribution of tag distances from those features (Figure 3, composite plot and cluster plot). Genomic features that are commonly examined include transcriptional start sites, specific cis-regulatory elements, and bound locations of specific proteins. Frequency distribution plots can be displayed as line graphs, which represent the collective tag distributions around a given set or subset of features, such as the transcriptional start sites for the most highly transcribed genes. These ‘‘composite’’ or ‘‘averaged’’ plots provide a simple and intuitive quantitative assessment of nucleosome occupancy, spacing, and positions in a single graph, but these plots do not reflect the variance in the system. For example, two dissimilar patterns may be, to a large extent, self-canceling when presented as a single composite plot. Plotting frequency distribution data as ‘‘cluster’’ plots make pattern variation more evident, but these plots are visually less quantitative. Cluster plots are essentially like line graphs for each region of interest, where the graph is collapsed to a onedimensional row running along the x axis (representing distance from a feature), and frequency bin counts (y axis) are represented by a color scale. Indeed, thousands of these rows can be aligned in the second dimension. These rows can then be sorted or organized, typically by K-means or hierarchical clustering. Perspective Historically, progress in the biological sciences was predicated on each individual experiment being relatively inexpensive, with the sum total of all experiments producing a body of knowledge. However, with the advent of genome-wide mapping by sequencing, each experiment has been comparatively expensive but has produced enough data to generate a substantial body of knowledge. Therefore, experiments have been carefully chosen to maximize the benefit to cost ratio. However, as the ‘‘low hanging fruit’’ of genome sequencing shrinks, researchers must penetrate deeper into the chromatin problem. We are only beginning to investigate the complexity of chromatin and its interplay with gene regulatory proteins on
a genome-wide scale. Nevertheless, research on genomic chromatin organization is expanding rapidly and encompassing a broader spectrum of biology, from the fundamental biophysical properties of chromosomes to human behavior. We still do not understand how DNA sequences work together with chromatin-remodeling proteins, such as RSC, CHD1, and SWI/ SNF, and other chromatin proteins to define the highly organized state of nucleosomes. We do not understand how histone modification states promote neural memory or cellular identity. Characterizing chromatin architecture and deciphering its code returns us back to the original state of biology in which many experiments sum together to comprise a body of research. Sequencing costs have dropped to the point where a large number of mapping experiments can now be completed within the scope of a typical grant. Thus, genome-wide experiments, such as high-resolution nucleosome mapping, are now accessible to a large number of researchers. As more investigators enter the field, its pace of discovery and impact across biology will only increase during the next decade. ACKNOWLEDGMENTS We thank Mike Kladde, Bryan J. Venters, Shinichiro Wachi, Kuangyu Yen, Liye Zhang, Sujana Ghosh, Megha Wal, Kiran Batta, Jing Hu, and Christine H. Walsh for helpful discussions. This work was supported by NIH grant HG004160.
REFERENCES Adli, M., Zhu, J., and Bernstein, B.E. (2010). Genome-wide chromatin maps derived from limited numbers of hematopoietic progenitors. Nat. Methods 7, 615–618. Albert, I., Mavrich, T.N., Tomsho, L.P., Qi, J., Zanton, S.J., Schuster, S.C., and Pugh, B.F. (2007). Translational and rotational settings of H2A.Z nucleosomes across the Saccharomyces cerevisiae genome. Nature 446, 572–576. Albert, I., Wachi, S., Jiang, C., and Pugh, B.F. (2008). GeneTrack-a genomic data processing and visualization framework. Bioinformatics 24, 1305–1306.
Giresi, P.G., and Lieb, J.D. (2009). Isolation of active regulatory elements from eukaryotic chromatin using FAIRE (Formaldehyde Assisted Isolation of Regulatory Elements). Methods 48, 233–239. Goecks, J., Nekrutenko, A., and Taylor, J. (2010). Galaxy: a comprehensive approach for supporting accessible, reproducible, and transparent computational research in the life sciences. Genome Biol. 11, R86. Greer, E.L., Maures, T.J., Hauswirth, A.G., Green, E.M., Leeman, D.S., Maro, G.S., Han, S., Banko, M.R., Gozani, O., and Brunet, A. (2010). Members of the H3K4 trimethylation complex regulate lifespan in a germline-dependent manner in C. elegans. Nature 466, 383–387. Guarente, L. (2000). Sir2 links chromatin silencing, metabolism, and aging. Genes Dev. 14, 1021–1026. Gundersen, B.B., and Blendy, J.A. (2009). Effects of the histone deacetylase inhibitor sodium butyrate in models of depression and anxiety. Neuropharmacology 57, 67–74. Hawkins, R.D., Hon, G.C., Lee, L.K., Ngo, Q., Lister, R., Pelizzola, M., Edsall, L.E., Kuan, S., Luu, Y., Klugman, S., et al. (2010). Distinct epigenomic landscapes of pluripotent and lineage-committed human cells. Cell Stem Cell 6, 479–491. Heintzman, N.D., Hon, G.C., Hawkins, R.D., Kheradpour, P., Stark, A., Harp, L.F., Ye, Z., Lee, L.K., Stuart, R.K., Ching, C.W., et al. (2009). Histone modifications at human enhancers reflect global cell-type-specific gene expression. Nature 459, 108–112. Henikoff, S., Henikoff, J.G., Sakai, A., Loeb, G.B., and Ahmad, K. (2009). Genome-wide profiling of salt fractions maps physical properties of chromatin. Genome Res. 19, 460–469. Hesselberth, J.R., Chen, X., Zhang, Z., Sabo, P.J., Sandstrom, R., Reynolds, A.P., Thurman, R.E., Neph, S., Kuehn, M.S., Noble, W.S., et al. (2009). Global mapping of protein-DNA interactions in vivo by digital genomic footprinting. Nat. Methods 6, 283–289. Hughes, A., and Rando, O.J. (2009). Chromatin ‘programming’ by sequence is there more to the nucleosome code than %GC? J. Biol. 8, 96. Jenuwein, T., and Allis, C.D. (2001). Translating the histone code. Science 293, 1074–1080. Jiang, C., and Pugh, B.F. (2009). Nucleosome positioning and gene regulation: advances through genomics. Nat. Rev. Genet. 10, 161–172. Kaplan, N., Moore, I.K., Fondufe-Mittendorf, Y., Gossett, A.J., Tillo, D., Field, Y., LeProust, E.M., Hughes, T.R., Lieb, J.D., Widom, J., et al. (2009). The DNA-encoded nucleosome organization of a eukaryotic genome. Nature 458, 362–366.
Barski, A., Cuddapah, S., Cui, K., Roh, T.Y., Schones, D.E., Wang, Z., Wei, G., Chepelev, I., and Zhao, K. (2007). High-resolution profiling of histone methylations in the human genome. Cell 129, 823–837.
Kolasinska-Zwierz, P., Down, T., Latorre, I., Liu, T., Liu, X.S., and Ahringer, J. (2009). Differential chromatin marking of introns and expressed exons by H3K36me3. Nat. Genet. 41, 376–381.
Birney, E., Stamatoyannopoulos, J.A., Dutta, A., Guigo, R., Gingeras, T.R., Margulies, E.H., Weng, Z., Snyder, M., Dermitzakis, E.T., Thurman, R.E., et al. (2007). Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature 447, 799–816.
Kouzarides, T. (2007). Chromatin modifications and their function. Cell 128, 693–705.
Core, L.J., and Lis, J.T. (2008). Transcription regulation through promoterproximal pausing of RNA polymerase II. Science 319, 1791–1792. Dang, W., Steffen, K.K., Perry, R., Dorsey, J.A., Johnson, F.B., Shilatifard, A., Kaeberlein, M., Kennedy, B.K., and Berger, S.L. (2009). Histone H4 lysine 16 acetylation regulates cellular lifespan. Nature 459, 802–807. Deal, R.B., and Henikoff, S. (2010). A simple method for gene expression and chromatin profiling of individual cell types within a tissue. Dev. Cell 18, 1030– 1040. Fragoso, G., and Hager, G.L. (1997). Analysis of in vivo nucleosome positions by determination of nucleosome-linker boundaries in crosslinked chromatin. Methods 11, 246–252. Gaspar-Maia, A., Alajem, A., Polesso, F., Sridharan, R., Mason, M.J., Heidersbach, A., Ramalho-Santos, J., McManus, M.T., Plath, K., Meshorer, E., et al. (2009). Chd1 regulates open chromatin and pluripotency of embryonic stem cells. Nature 460, 863–868.
Laajala, T.D., Raghav, S., Tuomela, S., Lahesmaa, R., Aittokallio, T., and Elo, L.L. (2009). A practical comparison of methods for detecting transcription factor binding sites in ChIP-seq experiments. BMC Genomics 10, 618. Lee, W., Tillo, D., Bray, N., Morse, R.H., Davis, R.W., Hughes, T.R., and Nislow, C. (2007). A high-resolution atlas of nucleosome occupancy in yeast. Nat. Genet. 39, 1235–1244. Li, H., and Homer, N. (2010). A survey of sequence alignment algorithms for next-generation sequencing. Brief Bioinform. 11, 473–483. Lieberman-Aiden, E., van Berkum, N.L., Williams, L., Imakaev, M., Ragoczy, T., Telling, A., Amit, I., Lajoie, B.R., Sabo, P.J., Dorschner, M.O., et al. (2009). Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science 326, 289–293. Liu, C.L., Kaplan, T., Kim, M., Buratowski, S., Schreiber, S.L., Friedman, N., and Rando, O.J. (2005). Single-nucleosome mapping of histone modifications in S. cerevisiae. PLoS Biol. 3, e328. Malik, H.S., and Henikoff, S. (2003). Phylogenomics of the nucleosome. Nat. Struct. Biol. 10, 882–891.
Cell 144, January 21, 2011 ª2011 Elsevier Inc. 185
Manuyakorn, A., Paulus, R., Farrell, J., Dawson, N.A., Tze, S., Cheung-Lau, G., Hines, O.J., Reber, H., Seligson, D.B., Horvath, S., et al. (2010). Cellular histone modification patterns predict prognosis and treatment response in resectable pancreatic adenocarcinoma: results from RTOG 9704. J. Clin. Oncol. 28, 1358–1365.
Schuster, S.C., Miller, W., Ratan, A., Tomsho, L.P., Giardine, B., Kasson, L.R., Harris, R.S., Petersen, D.C., Zhao, F., Qi, J., et al. (2010). Complete Khoisan and Bantu genomes from southern Africa. Nature 463, 943–947.
Mavrich, T.N., Ioshikhes, I.P., Venters, B.J., Jiang, C., Tomsho, L.P., Qi, J., Schuster, S.C., Albert, I., and Pugh, B.F. (2008). A barrier nucleosome model for statistical positioning of nucleosomes throughout the yeast genome. Genome Res. 18, 1073–1083.
Sharma, S.V., Lee, D.Y., Li, B., Quinlan, M.P., Takahashi, F., Maheswaran, S., McDermott, U., Azizian, N., Zou, L., Fischbach, M.A., et al. (2010). A chromatin-mediated reversible drug-tolerant state in cancer cell subpopulations. Cell 141, 69–80.
Pardo, C.E., Carr, I.M., Hoffman, C.J., Darst, R.P., Markham, A.F., Bonthron, D.T., and Kladde, M.P. (2010). MethylViewer: computational analysis and editing for bisulfite sequencing and methyltransferase accessibility protocol for individual templates (MAPit) projects. Nucleic Acids Res. Published online October 19, 2010. 10.1093/nar/gkq716. Peleg, S., Sananbenesi, F., Zovoilis, A., Burkhardt, S., Bahari-Javan, S., AgisBalboa, R.C., Cota, P., Wittnam, J.L., Gogol-Doering, A., Opitz, L., et al. (2010). Altered histone acetylation is associated with age-dependent memory impairment in mice. Science 328, 753–756. Rando, O.J. (2010). Genome-wide mapping of nucleosomes in yeast. Methods Enzymol. 470, 105–118. Rosenbloom, K.R., Dreszer, T.R., Pheasant, M., Barber, G.P., Meyer, L.R., Pohl, A., Raney, B.J., Wang, T., Hinrichs, A.S., Zweig, A.S., et al. (2010). ENCODE whole-genome data in the UCSC Genome Browser. Nucleic Acids Res. 38, D620–D625. Sasaki, S., Mello, C.C., Shimada, A., Nakatani, Y., Hashimoto, S., Ogawa, M., Matsushima, K., Gu, S.G., Kasahara, M., Ahsan, B., et al. (2009). Chromatinassociated periodicity in genetic variation downstream of transcriptional start sites. Science 323, 401–404. Schaefer, A., Sampath, S.C., Intrator, A., Min, A., Gertler, T.S., Surmeier, D.J., Tarakhovsky, A., and Greengard, P. (2009). Control of cognition and adaptive behavior by the GLP/G9a epigenetic suppressor complex. Neuron 64, 678–691. Schones, D.E., Cui, K., Cuddapah, S., Roh, T.Y., Barski, A., Wang, Z., Wei, G., and Zhao, K. (2008). Dynamic regulation of nucleosome positioning in the human genome. Cell 132, 887–898.
186 Cell 144, January 21, 2011 ª2011 Elsevier Inc.
Segal, E., and Widom, J. (2009). What controls nucleosome positions? Trends Genet. 25, 335–343.
Shi, Y. (2007). Histone lysine demethylases: emerging roles in development, physiology and disease. Nat. Rev. Genet. 8, 829–833. Shivaswamy, S., Bhinge, A., Zhao, Y., Jones, S., Hirst, M., and Iyer, V.R. (2008). Dynamic remodeling of individual nucleosomes across a eukaryotic genome in response to transcriptional perturbation. PLoS Biol. 6, e65. Stadtfeld, M., Apostolou, E., Akutsu, H., Fukuda, A., Follett, P., Natesan, S., Kono, T., Shioda, T., and Hochedlinger, K. (2010). Aberrant silencing of imprinted genes on chromosome 12qF1 in mouse induced pluripotent stem cells. Nature 465, 175–181. Stein, A., Takasuka, T.E., and Collings, C.K. (2010). Are nucleosome positions in vivo primarily determined by histone-DNA sequence preferences? Nucleic Acids Res. 38, 709–719. Weiner, A., Hughes, A., Yassour, M., Rando, O.J., and Friedman, N. (2010). High-resolution nucleosome mapping reveals transcription-dependent promoter packaging. Genome Res. 20, 90–100. Yuan, G.C., Liu, Y.J., Dion, M.F., Slack, M.D., Wu, L.F., Altschuler, S.J., and Rando, O.J. (2005). Genome-scale identification of nucleosome positions in S. cerevisiae. Science 309, 626–630. Zhang, Y., Liu, T., Meyer, C.A., Eeckhoute, J., Johnson, D.S., Bernstein, B.E., Nusbaum, C., Myers, R.M., Brown, M., Li, W., et al. (2008). Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137. Zhang, Y., Moqtaderi, Z., Rattner, B.P., Euskirchen, G., Snyder, M., Kadonaga, J.T., Liu, X.S., and Struhl, K. (2009). Intrinsic histone-DNA interactions are not the major determinant of nucleosome positions in vivo. Nat. Struct. Mol. Biol. 16, 847–852.
Nuclear PTEN Regulates the APC-CDH1 Tumor-Suppressive Complex in a Phosphatase-Independent Manner Min Sup Song,1 Arkaitz Carracedo,1,3 Leonardo Salmena,1 Su Jung Song,1 Ainara Egia,1 Marcos Malumbres,2 and Pier Paolo Pandolfi1,* 1Cancer Genetics Program, Beth Israel Deaconess Cancer Center, Department of Medicine and Pathology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA 02215, USA 2Cell Division and Cancer Group, Molecular Oncology Programme, Centro Nacional de Investigaciones Oncolo ´ gicas (CNIO), 28029 Madrid, Spain 3Present address: CIC bioGUNE, Technology Park of Bizkaia, 48160 Derio, Bizkaia, Spain *Correspondence:
[email protected] DOI 10.1016/j.cell.2010.12.020
SUMMARY
PTEN is a frequently mutated tumor suppressor gene that opposes the PI3K/AKT pathway through dephosphorylation of phosphoinositide-3,4,5-triphosphate. Recently, nuclear compartmentalization of PTEN was found as a key component of its tumor-suppressive activity; however its nuclear function remains poorly defined. Here we show that nuclear PTEN interacts with APC/C, promotes APC/C association with CDH1, and thereby enhances the tumor-suppressive activity of the APC-CDH1 complex. We find that nuclear exclusion but not phosphatase inactivation of PTEN impairs APC-CDH1. This nuclear function of PTEN provides a straightforward mechanistic explanation for the fail-safe cellular senescence response elicited by acute PTEN loss and the tumor-suppressive activity of catalytically inactive PTEN. Importantly, we demonstrate that PTEN mutant and PTEN null states are not synonymous as they are differentially sensitive to pharmacological inhibition of APC-CDH1 targets such as PLK1 and Aurora kinases. This finding identifies a strategy for cancer patient stratification and, thus, optimization of targeted therapies. INTRODUCTION PTEN (phosphatase and tensin homolog) is among the most frequently lost or mutated tumor suppressors, with a frequency of monoallelic mutations estimated at 50%–80% in endometrial carcinoma, glioblastoma, and prostate cancer and at 30%–50% in breast, colon, and lung tumors (Cairns et al., 1997; Feilotter et al., 1998; Gray et al., 1998; Li and Sun, 1997; Steck et al., 1997). Complete loss of PTEN is observed at highest frequencies in endometrial cancer and glioblastoma and is generally associated with metastatic cancers (Ali et al., 1999; Salmena et al.,
2008). Moreover, germline mutations of PTEN have been identified in cancer-susceptibility syndromes such as Cowden syndrome (Di Cristofano et al., 1998; Eng, 2003). PTEN can dephosphorylate phosphoinositide-3,4,5-triphosphate (PIP3), a potent activator of AKT (Maehama and Dixon, 1998). Loss of PTEN function leads to derepression of the PI3K/AKT pathway, which stimulates cell growth and survival (Stambolic et al., 1998; Sun et al., 1999). However, emerging evidence suggests that PTEN also has PI3K/AKT-independent functions (Salmena et al., 2008). Furthermore, cells harboring phosphatase-inactive PTEN mutants retain residual tumorsuppressive activity, leading to the hypothesis that PTEN exerts functions that are independent of its phosphatase activity (Blanco-Aparicio et al., 2007; Georgescu et al., 2000; Gildea et al., 2004; Koul et al., 2002; Maier et al., 1999). Early studies proposed that PTEN was exclusively cytoplasmic. However, recent reports clearly demonstrate that nuclear PTEN has important tumor-suppressive function (Fridberg et al., 2007; Perren et al., 2000; Whiteman et al., 2002; Zhou et al., 2002). Mechanistically, we have reported that ubiquitination of PTEN regulates its nuclear compartmentalization (Song et al., 2008; Trotman et al., 2007). However, the tumorsuppressive functions of PTEN within the nucleus still remain poorly defined. Cell-cycle progression is controlled by ubiquitination-mediated proteolysis of cell-cycle machinery. The two major E3 ubiquitin ligases controlling this process are SCF (Skp1/Cullin/F-box protein complex) and APC/C (anaphase-promoting complex/ cyclosome). SCF mainly controls target protein levels during S phase, whereas APC/C is thought to be active from mitosis to late G1 (Cardozo and Pagano, 2004; Peters, 2006). APC/C contains at least 11 different structural subunits, and its activity is controlled through the binding of CDC20 and CDH1, which recognize and recruit specific substrates. CDC20 is active in early mitosis whereas CDH1 activity is restricted to late mitosis and G1 (Pines, 2006; Sullivan and Morgan, 2007). Specific APC/C substrates include mitotic cyclins (Cyclins A and B), mitotic kinases (Aurora kinases, PLK1, Nek2A), proteins involved in chromosome segregation (Securin, Sgo1), DNA Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc. 187
A
Figure 1. Nuclear PTEN Interacts with APC/C
B
(A) Nuclear extracts from PTEN-deficient PC3 cells transfected with Myc-PTEN were immunoprecipitated with an anti-Myc antibody followed by mass-spectrometric peptide sequencing. APC/C components including APC3, APC4, APC5, and APC7 were identified. (B) DU145 cell lysates were immunoprecipitated (IP) without (Mock) or with anti-PTEN (left) or antiAPC3 (right) antibody followed by immunoblotting. Asterisk indicates heavy chain of IgG. (C) Recombinant GST-PTEN (aa 1–403), GSTPTEN-DC (aa 1–185), or GST-PTEN-DN (aa 185– 403) proteins, as shown in left diagram, were incubated with immunopurified APC/C from PC3 cells released 3 hr from nocodazole synchronization. * and # indicate heavy chain of IgG and nonspecific band, respectively. (D) Lysates from 12 hr after tetracyclin induction as in (E) were immunoprecipitated (IP) with anti-PTEN antibody and subjected to in vitro ubiquitination assay using 35S-labeled Cyclin B (WT and destruction box mutant [DBM]) in the presence of GST-CDH1 protein (50 ng) and immunoblotting with anti-polyubiquitinated proteins antibody. (E) PC3 cells were cotransfected with pcDNA4/TO/ Myc-His-PTEN and pcDNA6/TR to induce PTEN expression after the addition of tetracycline (Teton) for the indicated times or treated with 10 mM MG132 after 8 hr Tet induction for additional 4 hr, then subjected to immunoblotting (top) and flow cytometric analysis (bottom). The quantification of the relative immunoreactivity of each protein normalized to Actin is represented as the mean and the standard error of the mean (SEM) from three different experiments. (F) Accumulation of APC-CDH1 targets in PtenloxP/loxP;Probasin-Cre4 (Ptenpc/) mouse prostate epithelium. Lysates from anterior prostates in wild-type (WT-Cre) and Ptenpc/ mice at 11 weeks age were subjected to immunoblotting. See also Figures S1A–S1F.
C
D
E
replication proteins (Geminin, Cdc6), an F-box protein (SKP2), and transcription factors (Ets2, FoxM1) (Manchado et al., 2010; Wasch et al., 2010). Of importance, APC-CDH1 substrates, such as Cyclin A, PLK1, Aurora A, CDC20, or SKP2, are overexpressed in human tumors and are associated with chromosomal instability and poor prognosis (Carter et al., 2006). In mice, Cdh1 heterozygosity results in the development of epithelial tumors, suggesting that CDH1 may be a haploinsufficient tumor suppressor (Garcia-Higuera et al., 2008). Downregulation of CDH1 has been reported in many cancers, including those of prostate, ovary, liver, and brain, and during the malignant progression of a B-lymphoma cell line (Bassermann et al., 2008; Wang et al., 2000). Therefore, inactivation of APC-CDH1 in cancer may lead to unchecked accumulation of its targets with profound consequences for cell-cycle and genomic stability. 188 Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc.
F
In this study, we demonstrate that nuclear PTEN directly enhances the activity of APC/C by promoting its association with CDH1. Conversely, PTEN loss impairs the activity of the APC-CDH1 tumor-suppressive complex. Critically, PTEN activates APC-CDH1 in a phosphatase-independent manner, an observation that has important implications for cancer therapy. RESULTS Nuclear PTEN Interacts with APC/C In order to identify novel tumor-suppressive pathways regulated by PTEN, we immunoprecipitated exogenous Myctagged PTEN from the nuclear extracts of PTEN-deficient PC3 cells and identified interacting proteins by mass spectrometry (Figure 1A and Figure S1A available online). Remarkably, four APC/C constituents including APC3/CDC27, APC4,
APC5,and APC7 were identified as nuclear PTEN-associated proteins. We validated the binding between endogenous PTEN and the four APC/C components in DU145 cells by performing reciprocal immunoprecipitations (Figure 1B) and observed that the binding was restricted to the nucleus (Figures S1B and S1C). Notably, an in vitro binding assay revealed that the interaction of PTEN with APC3 is restricted to the COOH terminus of PTEN, indicating that the N-terminal phosphatase domain is dispensable for this interaction (Figure 1C). Critically, PTEN immunoprecipitation complexes displayed ubiquitinating ability toward the well-established APC/C target Cyclin B (Figure 1D). We next assessed the functional relevance of this physical interaction. To test whether PTEN directly affects the levels of APC/C targets, we developed a model system in which we could dissociate the consequences of acute PTEN induction from the ensuing cell-cycle effects. For this we used a Tetracyclin-inducible (Tet-on)-PTEN cellular system in asynchronously growing PC3 cells (Rubin et al., 1991). In this cellular system, acute PTEN induction precedes any sizable effects on cell-cycle distribution for at least 12 hr (Figure 1E). By contrast, in this timeframe PTEN induction led to a marked and rapid proteasome-dependent downregulation of APC/C targets. This was reverted by MG132 treatment, at time points much earlier than any detectable effect on cell-cycle distribution (Figure 1E). Moreover, we found that APC/CDH1 target levels were increased in vivo in Pten null prostates at time points where the incidence of prostate intraepithelial neoplasia (PIN) and the increase of cell proliferation are low (Figure 1F, Figure S1D, and data not shown). Similarly, PTEN silencing in DU145 cells led to the upregulation of the APC/C targets, Cyclin A2, Geminin, PLK1, Aurora A, and CDC20, whereas overexpression of PTEN in PTEN-deficient LNCaP and PC3 cells reduced APC/C target levels concomitant with G1 cell-cycle arrest (Figures S1E and S1F). Nuclear PTEN Enhances the Activity of APC-CDH1 Our data show that PTEN interacts with components of the APC/C complex and PTEN status alters the level of APC/C complex target proteins. Thus, we hypothesized that PTEN can regulate the ubiquitin ligase activity of APC/C. Indeed, in vitro APC/C ubiquitin ligase assays revealed that Cyclin B ubiquitination by APC/C was increased in a dose-dependent manner by recombinant PTEN protein in the presence of CDH1 (Figure 2A). By contrast, the ubiquitin ligase activity of APC/C was reduced in Pten/ mouse embryonic fibroblasts (MEFs) compared to wild-type cells (Figure 2B). Remarkably, nuclear but not cytoplasmic PTEN immunoprecipitates were capable of ubiquitinating Cyclin B (Figure 2C). Next, we sought to ascertain the mechanism by which PTEN regulates APC/C function. APC/C activation occurs upon the association with two different adaptor proteins, CDC20 and CDH1 (Cardozo and Pagano, 2004; Peters, 2006). Whereas CDC20 activates APC/C in early mitosis, APC-CDH1 is active in late mitosis and during G1 (Pines, 2006; Sullivan and Morgan, 2007). In vitro Cyclin B ubiquitination assays revealed that the ubiquitin ligase activity of PTEN immunoprecipitates in interphase was higher than in mitosis (Figure S1G). We there-
fore examined whether PTEN regulates the formation of the APC-CDH1 complex. In vivo, reintroduction of PTEN into PC3 cells increased the association between APC3 and CDH1 (Figure 2D, top panel), whereas PTEN knockdown in DU145 dramatically reduced the coimmunoprecipitation of APC3 and CDH1 (Figure 2D, bottom panel). In vitro binding assay revealed that PTEN favors the assembly between APC3 and CDH1 in a dose-dependent manner (Figure 2E). Taken together, these results suggest that PTEN promotes the association between APC and CDH1, thereby enhancing the activity of APC-CDH1. Nuclear Exclusion of PTEN Impairs Activation of APC-CDH1 Despite the increasing number of studies emphasizing the importance of nuclear PTEN as a tumor suppressor (Planchon et al., 2008; Salmena et al., 2008), how nuclear PTEN exerts its tumor-suppressive activity remains unclear. We have demonstrated that PTEN monoubiquitination at lysines 13 and 289 is essential for its nuclear localization and tumor-suppressive function, and conversely, deubiquitination of PTEN by HAUSP renders PTEN predominantly cytoplasmic (Song et al., 2008; Trotman et al., 2007). Therefore, we tested whether a nuclearexcluded PTENK13,289E mutant (Figure 3A) could still regulate APC-CDH1. In vivo ubiquitination assay showed that, unlike wild-type PTEN, nuclear-excluded PTENK13,289E was unable to promote the Cyclin B ubiquitination and to reduce the level of APC-CDH1 targets despite antagonizing AKT activation (Figure 3B and Figure S2A). Importantly, coimmunoprecipitation analysis revealed that PTENK13,289E mutant failed to interact with APC3 (Figure 3C). Because both PTEN and APC-CDH1 are known negative regulators of cell proliferation (Salmena et al., 2008; Wasch et al., 2010), we next examined the effects of nuclear exclusion of PTEN on cell growth. BrdU incorporation analysis revealed that nuclear-excluded PTENK13,289E mutant was ineffective in suppressing S phase entry after release from nocodazole block, whereas PTENWT significantly delayed the G1-S transition accompanied with a reduction in the level of APC-CDH1 targets (Figure 3D and Figure S2B). This observation is of critical relevance as it allowed us to uncouple the effect of PTEN on Akt and APC-CDH1 toward cell-cycle regulation. In line with this notion, PTENWT but not PTENK13,289E overexpression led to a downregulation of CDK2 activity due to reduction in the level of Cyclin A2 rather than Cyclin E at the G1-S transition, suggesting that nuclear PTEN might regulate the G1-S transition through reduction in the level of Cyclin A2, an APC-CDH1 target (Figure 3E). We also ascertained whether the delay in S phase entry by PTEN could be due to failure to be released from the nocodazole block or to exit from mitosis. PTENK13,289E as well as PTENWT overexpression had little impact on mitotic exit, as measured by the positivity of phospho-histone H3 (Figures S2C and S2D). In addition, PTENWT- or PTENK13,289E-expressing PC3 cells exhibited similar proportions of cells in mitosis upon nocodazole block, suggesting that PTEN might not be involved in spindle checkpoint (SAC) as well. Taken together, these results suggest that nuclear localization of PTEN is necessary for activation of APC-CDH1, thereby regulating cell-cycle progression. Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc. 189
A
Figure 2. Nuclear PTEN Regulates the Activity of APC-CDH1
B Cyclin B Cyclin B(DBM)
IP: IgG APC3 Pten
- + - + - - + + + - - + + +
Pten: Cyc B-Ub (D D.U.)
APC assay
GST-PTEN: IVT-CDH1:
* 35S-Cyclin
+/+ -/- +/+ -/- +/+ -/-
APC assay
Cyclin B-Ub conjugates
20 16 12 8 4 0
B
Cyclin B-Ub conjugates
35S-Cyclin
Input
IB: Poly-Ub
C Input
GST-CDH1: - - - + + + + -CNC NCN Cyc B-Ub (D.U.)
IVT-CDH1
Cyclin B Cyclin B(DBM) -
+ -
+
+
50 40 30 20 10 0
APC a assay
Cyc B-Ub (D.U.)
GST-PTEN
GST PTEN: GST-PTEN: IVT-CDH1:
GST-CDH1 Pten
IP: APC3 IP: PTEN
APC3
70 60 50 40 30 20 10 0
B
IB APC3 IB:APC3
+
Cyclin B-Ub conjugates j g
35S-Cyclin
D Input
APC3
B
GST-CDH1 APC3 PTEN Hsp90 Lamin B
PTEN CDH1
APC3 PTEN CDH1
His6-PTEN: Binding assay
E
Input
Lysate IP:CDH1
Lysate IP:CDH1
Growth-Suppressive Activity of PTEN Requires APC-CDH1 Next, we examined the requirement of APC-CDH1 function in PTEN-mediated cell growth suppression. The effects of PTEN on cell-cycle distribution as well as APC-CDH1 target modulation were completely abrogated by depletion of APC3 or CDH1 (Figures 4A and 4B). We then evaluated the contribution of APC-CDH1 to the growth-suppressive function of PTEN by using wild-type and Cdh1/ MEFs (among which the phosphatase activity of endogenous Pten was indistinguishable; Figure 4C). Strikingly, whereas PTEN overexpression induced a growth inhibition in immortalized Cdh1+/+ cells (or in Cdh1/ MEFs complemented with human CDH1), Cdh1/ cells were refractory to this effect of PTEN overexpression (Figure 4D, top panel). Consistent with these data, overexpression of PTEN was accompanied by the downregulation of Cyclin A2-associated CDK2 activity as 190 Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc.
+
++ +++ GST-CDH1 His6-PTEN APC3
(A) Immunopurified APC/C from PC3 cells, released 3 hr post-nocodazole synchronization, were subjected to in vitro ubiquitination assay using 35S-labeled in vitro-translated (IVT) wild-type or DBM Cyclin B in the presence of IVT-CDH1 (1 ml) and different amounts of GST-PTEN proteins (0, 50, 100 ng) and immunoblotting with antipolyubiquitinated protein antibody. Bottom panel illustrates the density unit (D.U.) of Cyclin B-Ub conjugates analyzed by the ImageJ 1.383 software (NIH). Asterisk indicates nonspecific band. (B) Lysates from wild-type and Pten/ MEFs subjected to 48 hr of serum depletion were immunoprecipitated (IP) with anti-APC3 or antiPten antibody and subjected to in vitro ubiquitination assay using 35S-labeled Cyclin B in the presence of GST-CDH1 protein (50 ng). Top panel shows the density unit (D.U.) of Cyclin B-Ub conjugates analyzed by the ImageJ 1.383 software. (C) Cytoplasmic (C) and nuclear (N) extracts from DU145 cells, released 3 hr from nocodazole synchronization, were IP with anti-APC3 or antiPTEN antibody and subjected to in vitro ubiquitination assay using 35S-labeled Cyclin B. Top panel shows the density unit (D.U.) of Cyclin B-Ub conjugates analyzed by the ImageJ 1.383 software. (D) Lysates from PTEN-deficient PC3 cells complemented with empty vector or PTEN (top) or PTEN-proficient DU145 cells transfected with siRNAs for Renilla luciferase (siControl) or PTEN (siPTEN-1) (bottom) were IP with anti-CDH1 antibody and then subjected to immunoblotting. (E) Immunopurified APC/C from PC3 cells, released 3 hr from nocodazole synchronization, were incubated with GST-CDH1 (100 ng) and different amounts of His6-PTEN proteins (0, 50, 100, 200 ng) for 1 hr followed by immunoblotting. See also Figure S1G.
GST-CDH1 His6-PTEN
well as other APC-CDH1 targets in Cdh1+/+ (or in Cdh1/ MEFs complemented with CDH1) but not in Cdh1/ cells (Figure 4D, bottom panel and Figure S3A). Conversely, the increase in proliferation and elevation of the level of APC-CDH1 targets induced by Pten silencing was abrogated in the absence of Cdh1 (Figure 4E and Figure S3B). Taken together, these results suggest that the growth-suppressive activity of PTEN is, at least in part, mediated by APC-CDH1. PTEN-Loss-Induced Cellular Senescence Involves APC-CDH1 Reduction of cellular levels of PTEN or CDH1 results in the decrease of their respective tumor-suppressive activity (Di Cristofano et al., 1998; Garcia-Higuera et al., 2008). However, acute and complete loss of both Pten and Cdh1 leads to a fail-safe cellular senescence response; this phenomenon has been
Figure 3. Nuclear Exclusion of Impairs Activation of APC-CDH1
B PTENWT
GFP
PTENK13,289E
Cyclin B Cyclin B(DBM) HA-Ub: GFP: WT: K13,289E: Mr(K)
-
+ + + + - - + - - +
-
+ + + + - - + - - + Cyclin B (U Ub)n
A
250 150
IP: Myc 100 IB: HA-Ub
IP: GFP P
C
#
50
*
IP: Myc 75 IB: Myc 50
*
APC3
Lysates
GFP-PTEN
GFP-PTEN
E
GFP WT K13,289E
GFP
Myc-Cyclin B GFP-PTEN
WT
K13,289E 32P-H1
CycA2 C
100
Cyclin B
Lysates
APC3
D
1 0.9 1.2 0.3 0.6 0.5 0.6 0.9 1.0 (fold)
Histone H1 * Cyclin A 2
70
40
*
10 10
11
12
Time (h)
13
32P-H1
CDK2 CycE1
BrdU-PE+/GFP+ c cells (%)
75
PTEN
(A) Immunofluorescence analysis of GFP, GFPtagged wild-type PTEN (PTENWT), and nuclearexcluded PTEN mutant (PTENK13,289E). Arrows indicate the nucleus in PC3 cells. (B) PC3 cells were cotransfected with a combination of wild-type or DBM Myc-Cyclin B, HA-ubiquitin (Ub), GFP, PTENWT, and PTENK13,289E and treated with the proteasome inhibitor MG132 (10 mM) for 4 hr before harvesting. Cell lysates were then immunoprecipitated (IP) with anti-Myc antibody and subjected to immunoblotting (IB). * and # indicate heavy chain of IgG and nonspecific band, respectively. (C) Lysates from PC3 cells transfected with GFP, PTENWT, or PTENK13,289E were IP with anti-GFP antibody and subjected to immunoblotting. (D) PC3 cells transfected with GFP, PTENWT, or PTENK13,289E were synchronized by growth in nocodazole (400 nM) for 24 hr, released for the indicated times, pulsed with BrdU for 30 min, and subjected to flow cytometric analysis. Data are the means from three different experiments and error bars represent the SEM. (E) Lysates from (D) were subjected to in vitro kinase assay with CDK2, Cyclin E1, or Cyclin A2. The relative kinase activities were normalized with histone H1 inputs. Asterisks indicate heavy chain of IgG. See also Figure S2. p value was determined by Student’s t test (*p < 0.01).
Cyclin E1 32P-H1
CDK2
vation of the p53 pathway (Chen et al., 2005). Interestingly, complete loss of Cdh1 and Pten led to a dramatic accumulation of p16, whereas Arf or p53/p21 was not affected compared to Pten loss alone. Critically, silencing of p16 prevented Pten-loss-induced senescence without affecting the p53 axis (Figures S4C–S4E). In line with previous reports from our lab, p19 silencing also reduced cellular senescence in Pten null MEFs without affecting p53 (Figures S4C–S4E) (Chen et al., 2009). As Ets2 has been shown to selectively regulate p16, but not Arf transcription (Huot et al., 2002; Ohtani et al., 2001), we analyzed Ets2 levels in Pten null cells to determine whether an increase in Ets2 could account for the enhanced expression of p16. Indeed, Ets2 protein but not mRNA level was markedly upregulated by acute loss of Pten (Figures 5C and 5G and Figure S4F). Of interest, consistent with the notion that APC-CDH1 regulates the ubiquitination-mediated degradation of Ets2 (Figure S4G) (Li et al., 2008), PTEN promoted the ubiquitination of wild-type but not destruction box mutated (DBM) Ets2 (Figure 5H). To determine the direct role of the Ets2/p16 pathway on active senescence in Pten null cells, we overexpressed or silenced Ets2 upon acute Pten loss. Overexpression of DBM Ets2 significantly promoted the senescence response by Pten loss even in the presence of CDH1 (Figure 5I), whereas silencing of Ets2 dramatically reduced senescence and the transcriptional induction of p16 (Figures S4H and S4I). Collectively, these data provide a compelling mechanism for
Time (h): 10 11 12 10 11 12 10 11 12
shown to profoundly suppress tumor progression upon Pten loss (Chen et al., 2005; Li et al., 2008). We therefore hypothesized that Pten-loss-induced senescence could be in part mediated by a reduction of Cdh1 activity toward mediators of cellular senescence. In order to test this hypothesis, we overexpressed CDH1 concomitantly with the acute inactivation of Pten in Ptenlox/lox MEFs. Strikingly, overexpression of CDH1 counteracted the senescence response (Figures 5A and 5B). Importantly, overexpression of CDH1 specifically prevented the increase of p16 mRNA and protein levels, but not those of Arf or p53 (Figures 5C and 5D and Figure S4A). Conversely, complete concomitant inactivation of Cdh1 and Pten in MEFs dramatically increased the distinctive morphology of senescent cells (flattened large cells; data not shown) and the positivity for senescence-associated b-galactosidase (SA-b-Gal), a hallmark of senescent cells (Figure 5E and Figure S4B). Consistent with these data, complete loss of Cdh1 upon Pten loss led to reduced proliferation compared to Pten loss alone, as measured by growth curves and BrdU incorporation analysis (Figure 5F). Of note, the senescence response elicited upon acute loss of Pten was accompanied by elevation of p19ARF(Arf)/p53/p21 as well as p16INK4A(p16) (Figures 5C and 5G). We have previously reported that senescence elicited by acute loss of Pten is accompanied by the acti-
Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc. 191
A
B
Figure 4. Growth Suppression by PTEN Relies on APC-CDH1 Function
C siCont. siCDH1-1 siCDH1-2
siCont. siAPC3-1 siAPC3-2
R2=0.998
Cyclin A2
Cyclin A2 1 0.48 1.59 1.64 2.11 1.94 (fold)
1 0.50 1.64 1.53 1.40 1.40 (fold)
Geminin
Geminin
1 0.60 1.46 1.53 1.67 1.67
1 0.51 1.67 1.42 1.39 1.30
Aurora A
0.150 0.100 0.050 0.000
Aurora A
1 0.41 1.69 1.58 2.04 1.95
0
1 0.40 1.35 1.24 1.22 1.19
PLK1
50 100 150 200
pmol phosphate
PLK1
1 0.71 2.01 2.11 2.59 2.55
1 0.45 1.41 1.34 1.31 1.26
CDC20 APC3
CDH1
PTEN
PTEN Hsp90
Hsp90 % Ce ells
% Ce ells
G1 S G2/M
100 80 60 40 20 0
100 80 60 40 20 0 / Cdh1-// Cont Cdh1+/+ Cont.
IP:PTEN
E 18 16 14 12 10 8 6 4 2 0
Cdh1+/+/Vector Cdh1+/+/PTEN Cdh1-/-/Vector Cdh1-/-/PTEN Cdh1-/-/CDH1/Vector Cdh1-/-/CDH1/PTEN
0
1 2 Time (d)
3
Cell numberr (x104)
D
G1 S G2/M
pmol phosphate
1 0.70 2.05 2.09 2.13 2.08
100 80 60 40 20 0
120
CDC20
1 0.66 1.43 1.53 1.91 1.64
Cell numberr (x104)
A620nm
0.200
16 14 12 10 8 6 4 2 0
Cdh1+/+/shGFP Cdh1+/+/shPten Cdh1-/-/shGFP Cdh1-/-/shPten Cdh1-/-/CDH1/shGFP Cdh1-/-/CDH1/shPten
0
Cdh1+/+ Cdh1-/- Cdh1-/-/CDH1
1 2 Time (d)
Cdh1+/+ Cdh1-/- Cdh1-/-/CDH1
32P-H1
Histone H1
CycA2
CycA2
32P-H1
1 0.41 2.40 2.57 1.79 0.88 (fold)
1 2.61 2.72 2.76 1.19 2.29 (fold)
Cyclin E1 32P-H1
CDK2
Histone H1 * IP:Cyclin A 2
CDK2 CycE1
CDK2 CycE1
* Cyclin A 2 32P-H1
3
(A and B) PC3 cells cotransfected with empty vector or HA-PTEN and Renilla luciferase siRNA (siControl), two independent APC3 siRNAs (siAPC3-1 and -2) (A), or CDH1 siRNAs (siCDH1-1 and -2) (B) were subjected to immunoblotting (top) and flow cytometric analysis to measure cell-cycle distribution (bottom). The relative immunoreactivity of each protein was quantified by normalizing with Hsp90. (C) diC8-PtdIns(3,4,5)P3 (40 mM) was incubated with and without (control) the immunopurified Pten (1 mg) at 37 C for 30 min. The amount of free phosphate was measured as Absorbance 620 nm (A620nm) and calculated using the standard curve line-fit data in top panel. Error bars represent SEM from three different experiments. (D) Growth curves (top) and in vitro kinase assay (bottom) of SV40-LT-immortalized wild-type and Cdh1/ MEFs, infected with a retroviral combination of human CDH1 and PTEN (with selection) as indicated and followed over a 3 day period. Error bars represent the SEM from three different experiments. Asterisks indicate heavy chain of IgG. (E) Growth curves (top) and in vitro kinase assay (bottom) of immortalized wild-type and Cdh1/ MEFs, infected with a retroviral combination of human CDH1 and Pten shRNA (with selection) as indicated and followed over a 3 day period. Error bars represent SEM from three different experiments. The relative kinase activities were normalized with histone H1 inputs. Asterisks indicate heavy chain of IgG. See also Figure S3.
the senescence response elicited by Pten loss through parallel CDH1/Ets2/p16 and Arf or p53/p21. PTEN Regulates APC-CDH1 Independently of Its Phosphatase Activity PTEN opposes the PI3K/AKT signaling pathway by catalyzing the dephosphorylation of PIP3 (Maehama and Dixon, 1998). Loss of PTEN leads to the activation of the PI3K/AKT cascade and stimulates cell growth and survival (Stambolic et al., 1998; Sun et al., 1999). Nevertheless, cells harboring phosphataseinactive PTEN mutants are phenotypically distinct from cells lacking PTEN protein, suggesting that PTEN exerts functions that are independent of its phosphatase activity (Blanco-Aparicio et al., 2007; Georgescu et al., 2000; Gildea et al., 2004; Maier et al., 1999; Okumura et al., 2005). In order to examine the contribution of the phosphatase activity of PTEN in the regulation of APC-CDH1, we first complemented PTEN null PC3 cells, with either wild-type or phosphatase-inactive PTEN (C124S or G129E) (Figure 6A). The levels of several canonical APC-CDH1 192 Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc.
32P-H1
Cyclin E1
targets were reduced upon expression of PTEN(C124S) or PTEN(G129E) phosCDK2 phatase-inactive mutants in this system (Figure S5A). Consistently, expression of phosphatase-inactive PTEN(C124S) in PC3 cells resulted in a delayed entry into S phase, a phenotype similar to that in wild-type PTEN-expressing PC3 cells or PTENproficient DU145 cells, implying that PTEN regulates the G1-S transition independent of its phosphatase activity (Figure 6B and Figures S5B and S5C). Importantly, expression of wildtype or PTEN(C124S) led to reduced Cyclin A2-associated CDK2 activity in PC3 cells, suggesting that PTEN(C124S) as well as wild-type PTEN suppresses the G1-S transition, at least in part, through APC-CDH1. Consistent with these data, in vivo xenograft generation with these variants of PC3 cells revealed a significant growth-suppressive activity of PTEN(C124S), in line with the modulation of APC-CDH1 targets (Figure 6C). Although expression of PTEN(C124S) led to a significant reduction in the level of late mitotic substrates of APC-CDH1, such as Aurora A, PLK1, and CDC20, the length of mitosis was not significantly shortened, compared to parental PC3 cells (Figures S5D and S5E). Importantly, phosphatase-inactive PTEN(C124S) mutants exhibited comparable activity to wild-type PTEN, enhancing the ubiquitin ligase activity of APC-CDH1 in vitro (as 32P-H1
A
B
D
E
G
C
F
H
I
Figure 5. Pten-Loss-Induced Cellular Senescence Involves APC-CDH1 (A–C) Cellular senescence assay (A), growth curves and BrdU incorporation assay (B), and immunoblot (C) of primary Ptenlox/lox MEFs, infected with retroviral Cre and human CDH1 selected for 4 days. Error bars represent SEM from three different experiments. p value was determined by Student’s t test. (D) Real-time RT-PCR analysis of murine p16 expression was quantified. Error bars represent SEM. (E–G) Cellular senescence assay (E), growth curves and BrdU incorporation assay (F), and immunoblotting (G) of primary conditional Ptenlox/lox;Cdh1+/+, Ptenlox/lox;Cdh1lox/+, Ptenlox/lox;Cdh1lox/lox MEFs, infected with retroviral Cre recombinase at 4 days after selection. Error bars represent SEM from three different experiments. Asterisk indicates nonspecific band. (H) PC3 cells were cotransfected with wild-type or DBM Flag-Ets2, His-ubiquitin (Ub), and HA-PTEN and treated with the proteasome inhibitor MG132 (10 mM) for 4 hr before harvesting. His-Ub-conjugated Ets2 was purified from cell lysates using Ni2+-NTA spin column under denaturing conditions. (I) Cellular senescence assay of primary Ptenlox/lox MEFs, infected with a retroviral combination of wild-type or Ets2(DBM), CDH1, and Cre as indicated, at 4 days after selection. Error bars represent SEM from three different experiments. See also Figure S4.
Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc. 193
B
120 100 80 60 40 20 0
100 80 60 40
PTEN
20
P-AKT AKT Actin
0 10
11
12
13
Time (h)
Tu umor volume (mm3)
140
pmol phosphate
C BrdU-positive cells (% %)
A
400 350 300 250 200 150 100 50 0
Vector WT C124S
#
* 4
6 8 10 12 14
Time (d)
D
Vector
WT
WT
C124S Cyclin A2 Geminin Aurora A PLK1 P-AKT AKT PTEN Actin
C124S
GST-PTEN: + IVT-CDH1: IVT CDH1: - - + + + + + + + +
Cyclin B-Ub conjugates
35S-Cyclin
B
Cyyclin B-Ub (D.U.)
APC assay
WT
IVT-CDH1
Input
C124S
45 40 35 30 25 20 15 10 5 0
GST-PTEN
GST-PTEN Cdh1+/+ Cdh1-/- Cdh1-/-/CDH1
32P-H1
Cdh1+/+/Vector Cdh1+/+/C124S Cdh1-/-/Vector Cdh1-/-/C124S Cdh1-/-/CDH1/Vector Cdh1-/-/CDH1/C124S
0
1 2 Time (d)
3
CycA2
18 16 14 12 10 8 6 4 2 0
1 0.30 1.88 2.10 1.46 0.46 (fold)
Histone H1 * Cyclin A 2
CDK2 CyycE1
Celll number
(x104)
E
32P-H1
Cyclin E1 32P-H1
CDK2
Figure 6. PTEN Regulates APC-CDH1 Independently of Its Phosphatase Activity (A) diC8-PtdIns(3,4,5)P3 (40 mM) was incubated without (control) and with PTEN (WT), PTEN(C124S), or PTEN(G129E) immunoprecipitates (1 mg) at 37 C for 30 min and then the free phosphate was measured using the Green Reagent. Error bars represent the SEM from three different experiments. (B) PC3 cells complemented with wild-type or phosphatase-inactive PTEN(C124S) were nocodazole synchronized, released for the indicated times, and pulsed with BrdU 30 min before harvesting. The proportion of BrdU+ cells was measured (left) and cell lysates were subjected to western blot (right). Error bars represent SEM from three different experiments. (C) PC3 cells complemented with wild-type or phosphatase-inactive PTEN(C124S) were injected subcutaneously into nude mice. Tumor volume was monitored (top) and tissue lysates at 2 weeks after injection were subjected to immunoblotting (bottom). Error bars represent SEM (n = 6 mice/group). (D) Immunopurified APC/C from PC3 cells, released 3 hr post-nocodazole synchronization, were subjected to in vitro ubiquitination assay using 35S-labeled Cyclin B in the presence of in vitro-translated (IVT)-CDH1 (2 ml) and different amounts of wild-type or PTEN(C124S) proteins (0, 50, 100, 200 ng). Right panel illustrates the density unit (D.U.) of Cyclin B-Ub conjugates analyzed by the ImageJ 1.383 software. (E) Growth curves (left) and in vitro kinase assay (right) of immortalized wild-type and Cdh1/ MEFs, infected with a retroviral combination of human CDH1 and PTEN(C124S) (with selection) as indicated and followed over a 3 day period. Error bars represent SEM from three different experiments. The relative kinase activities were normalized with histone H1 inputs. Asterisk indicates heavy chain of IgG. See also Figure S5 and Figure S6. p value was determined by Student’s t test (*p < 0.01; #p < 0.05).
194 Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc.
measured by Cyclin B ubiquitination) and in vivo (as measured by ubiquitination of Cyclin B and Ets2) (Figure 6D and Figures S6A and S6B). Finally, we found that, like wild-type PTEN, overexpression of phosphatase-inactive PTEN(C124S) resulted in a significant growth suppression accompanied with lower Cyclin A2-associated CDK2 activity in immortalized Cdh1+/+ (or Cdh1/ MEFs complemented with human CDH1) but not Cdh1/ cells (Figure 6E and Figure S6C). We then investigated the effects of pharmacological inhibition of PTEN phosphatase activity on APC-CDH1 function. Di-potassium bisperoxo (picolinato) oxovanadate [bpV(Hopic)] inhibits PTEN phosphatase activity (Lai et al., 2007; Rosivatz et al., 2006). Coimmunoprecipitation analysis revealed that the treatment of DU145 cells with 1 mM bpV(Hopic) had no effects on the formation of APC-CDH1 despite strong activation of AKT (Figure S6D). Additionally, the inhibition of PI3-kinase by treatment with 1 mM LY294002 failed to alter the interaction between APC3 and CDH1 or the ubiquitin ligase activity of APC-CDH1 (Figure S6E). Taken together, these results demonstrate that PTEN regulates APC-CDH1 in a phosphatase-independent manner. PTEN Loss but Not Inactivation of Its Phosphatase Activity Results in Hypersensitivity to Pharmacological Inhibition of APC-CDH1 Targets Inhibition of APC-CDH1 targets, such as PLK1 or Aurora kinases, has been pursued as a therapeutic modality to treat human cancers due to the fact that APC-CDH1 activity is reduced in tumors (Bassermann et al., 2008; Taylor and Peters, 2008). Indeed, PLK1 inhibitors as well as Aurora A inhibitors are currently being evaluated as anticancer agents (Harrington et al., 2004; Strebhardt and Ullrich, 2006). In our study we have demonstrated that levels of PLK1 and Aurora A are elevated upon PTEN loss in both asynchronous and synchronized conditions, suggesting that PTEN-deficient tumors might exhibit addiction to these kinases and hence hypersensitivity to their pharmacological inhibition. On the basis of the dispensability of the phosphatase activity of PTEN, we hypothesized that PTEN-deficient tumor cells would be more sensitive to such therapeutic approaches compared to phosphatase-inactive PTEN mutant cells. To test this hypothesis, we first pharmacologically inhibited PLK1 in wild-type and PTEN null cells. Both human and murine Pten null cells were highly sensitive to PLK1 pharmacological inhibition by BI 2536, a PLK1 inhibitor (Figures 7A and 7B and Figures S7A and S7B; PLK1 inhibition was directly measured by spindle assembly; Figure S7C). Importantly, the growth inhibitory effect of BI 2536 in PTEN null cells was associated with a profound mitotic arrest followed by apoptosis, as measured by cleavage of Parp and caspase-3 (Cpp32) (Figures 7A, right panel and Figure 7C). Critically, reconstitution of PTEN null cells with phosphatase-inactive PTEN(C124S) significantly restored the resistance to pharmacological inhibition of PLK1 (Figures 7B and 7C and Figure S7D). Similarly, PTEN null cells were hypersensitive to Aurora A inhibition by VX680 (MK0457) (Figures 7D and 7E and Figures S7E and S7F; Aurora A inhibition was directly measured by histone H3 phosphorylation on Ser10; Figure S7G). Of note, Aurora A inhibition by VX680 in PTEN null cells induced a robust accumulation of >4 N DNA content—as
measured by flow cytometry—and subsequent apoptosis (Figure 7D, right panel and Figure 7F). Importantly, the sensitivity of phosphatase-inactive PTEN(C124S)-expressing cells to VX680 was similar to that of wild-type PTEN-expressing cells (Figures 7E and 7F and Figure S7H). Taken together, these data suggest that PTEN loss but not phosphatase inactivation results in hypersensitivity to pharmacological inhibition of APCCDH1 targets PLK1 and Aurora A. DISCUSSION Our findings allow us to reach a number of relevant conclusions. First, we have identified a novel mechanism by which PTEN exerts its tumor-suppressive function within the nucleus by regulating the assembly and activity of APC-CDH1. Whereas high concentrations of nuclear PTEN are associated with G0/G1 phase and differentiated cells, lower concentrations of nuclear PTEN level are observed during S phase and in highly proliferating advanced tumors (Gimm et al., 2000; Ginn-Pease and Eng, 2003; Perren et al., 2000). Furthermore, Pten and Cdh1 are haploinsufficient tumor suppressors in the mouse and their heterozygous loss leads to diverse cancers, including cancers of the prostate and breast (Garcia-Higuera et al., 2008; Salmena et al., 2008). By contrast, complete acute loss of either tumor suppressor triggers a cellular senescence response (Chen et al., 2005; Li et al., 2008). In this study, we have demonstrated that Pten-loss-induced senescence is dependent on the Cdh1/Ets2/p16 pathway. On the basis of these data, we propose that PTEN loss elicits a potent senescence response through both phosphatase-dependent (superactivation of a PI3K/mTOR/p53 pathway; Alimonti et al., 2010; Chen et al., 2005) and phosphatase-independent (loss of APC-CDH1 function, unrelated to Arf or p53/p21 pathways) activities, in turn highlighting the complexity of the dose-dependent tumorpromoting and fail-safe cellular responses evoked by PTEN loss (Figure 7G). Second, we find that the novel function of nuclear PTEN presented herein is independent of its phosphatase activity. Cytoplasmic but not nuclear pools of PIP3 are sensitive to catalysis by PTEN, implying a potential role for nuclear PTEN beyond its phosphatase activity (Lindsay et al., 2006). Indeed, growth-inhibitory effects of nuclear PTEN are likely not mediated directly through the PI3K/AKT pathway (Blanco-Aparicio et al., 2007; Liu et al., 2005). Given that APC3 is highly phosphorylated on multiple residues in mitosis (Kraft et al., 2003), one could suspect that PTEN may be a potential phosphatase for APC3. However, in mitotic cells, we did not observe dephosphorylation of APC3 by PTEN neither in vivo nor in vitro (M.S.S. and P.P.P, unpublished data), and in addition, a catalytically inactive form of PTEN is fully functional toward APC. Overall, the phosphataseindependent activity of nuclear PTEN toward APC-CDH1 provides a straightforward explanation for the remnant tumorsuppressive activity associated with the phosphatase-inactive PTEN(C124S) mutation. Finally, we have demonstrated that PTEN-deficient and PTEN mutant cancer cells are differentially sensitive to pharmacological inhibition of PLK1 and Aurora A. Thus, our study indicates that patients with cancers harboring complete PTEN loss may Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc. 195
60 50 40 30 20 10 0
Growth inhibition (%)
B Growth inhibition (%)
A Cleaved Parp Cleaved Cpp32 Pt Pten Actin Pten+/+ Pten-/-
C
25
0
Vector
50
WT
50 40 30 20
10 0
(BI2536,nM)
Vector WT C124S
C124S
Vector WT C124S
Mitotic cells (%)
20
control
60
BI 2536
P-H3(pS10) DAPI
15
#
10
**
5 0
0
24
48
Time (h)
70 50
Cleaved Parp Cleaved Cpp32
40
Pten
*
60
30
Actin
20
Pten+/+ Pten-/-
0
control
VX680
Vector
Vec-dmso.001
400
600 FL2-A WT-dmso.002
WT
200
800
400
600 FL2-A C124S-dmso.003
C124S
200
800
1000
0
200
400
600 FL2-A WT-VX-1day.005
200
400
600 FL2-A
G1 G2/M
800
800
1000
31.1%
1000
0
200
400
600 800 FL2-A C124S-VX-1day.006
1000
Vector WT C124S
100
0
200
400
600 FL2-A
800
PTENWT
#
60
PTENC124S CDH1
APC
**
40 20 0
0 1000
CDH1
80
36.6%
15.2%
0
APC
56.4%
10.5% 0
G
Vec-VX-1day.004
12.2%
70 60 50 40 30 20 10 0
Vector WT C124S
500 (VX680,nM)
100
>4N cells (%)
F
0
Growth inhibition (%)
E Growth inhibition (%)
D
1000
G1 G2/M
24 Time (h)
48
Proliferation Cyclin A2 Geminin Aurora A PLK1
Senescence Ets2/p16
Figure 7. PTEN Loss but Not Phosphatase Inactivation Results in Hypersensitivity to Pharmacological Inhibition of APC-CDH1 Targets
(A) Growth inhibition of immortalized wild-type and Pten/ MEFs treated with the PLK1 inhibitor BI 2536 (50 nM) for 48 hr was measured (left). Western blots of cell lysates after BI 2536 treatment for 48 hr are shown on the right. Error bars represent SEM from three different experiments. (B) Growth inhibition of PC3 cells complemented with wild-type or PTEN(C124S) and treated with the PLK1 inhibitor BI 2536 (5 nM) was measured after the 48 hr treatment. Error bars represent SEM from three different experiments.
196 Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc.
benefit from pharmacological targeting of the APC-CDH1 pathway, whereas we predict mutant PTEN tumors to be much less sensitive. The APC-CDH1 targets PLK1 and Aurora kinases are overexpressed in human tumors, and this has prognostic and therapeutic potential in cancers (Meraldi et al., 2004; Strebhardt and Ullrich, 2006). Clinical trials in cancer patients are currently underway to test the effects of various PLK1 and Aurora A inhibitors as monotherapy or in combination with conventional chemotherapy. For example, phase II trials of BI 2536 in advanced small cell lung carcinoma (SCLC), non-small cell lung carcinoma (NSCLC), and acute myeloid leukemia (AML) and phase II trials of VX680 (MK0457) in chronic myeloid leukemia (CML) are currently ongoing (weblink: http://www. cancer.gov). It has been recently reported that oncogenic Ras activation engages the cells to be hypersensitive (synthetic lethal) to various mitotic inhibitors including BI 2536 (Luo et al., 2009). By contrast, and in full agreement with our findings, oncogenic PI3K activation does not respond to PLK1 inhibition, suggesting that PTEN-loss-driven hypersensitivity to inhibition of PLK1 and Aurora A is unlikely due to the PI3K/AKT pathway (Luo et al., 2009). Hence, our findings provide a rationale for cancer patient stratification based on PTEN loss versus PTEN mutation toward the optimization of targeted therapies. This principle is of great importance given the widespread incidence of these two states of PTEN gene disruption in human cancer. Furthermore, because the functions of PTEN toward PI3K/AKT signaling and the APC-CDH1 pathway are uncoupled, it is tempting to speculate that combinatorial therapy with PLK1/ Aurora kinases and PI3K/mTOR inhibitors may be an effective approach in PTEN null cancers. EXPERIMENTAL PROCEDURES Reagents and Antibodies Chemical reagents are from Sigma unless otherwise described. G418, puromycin, and hygromycin, which are used for establishment of stable cell lines, are purchased from Invitrogen. BrdU and anti-BrdU antibody were from BD Biosciences. The PLK1 inhibitor BI 2536 and the Aurora A inhibitor VX680 were from Chemie Tek. All the sources of antibodies used are listed in the Extended Experimental Procedures. Mass Spectrometry PTEN-deficient PC3 cells transfected with pRK5-Myc-PTEN were subjected to cellular fractionation. The details of nuclear fractionation method are described in the Extended Experimental Procedures. The nuclear extracts were immunoprecipitated with anti-Myc antibody and the PTEN-associated proteins were eluted with Myc peptide. The eluates were resolved by SDSPAGE on 4%–12% gradient gel (Invitrogen) for silver staining (Pierce). Specific
bands were cut out from the gel and subjected to mass-spectrometric peptide sequencing. Immunoprecipitation and In Vitro Binding Assay The details of immunoprecipitation and in vitro binding assay are described in the Extended Experimental Procedures. In Vitro Ubiquitination Assay The APC/C or PTEN was immunopurified with anti-APC3 (Sigma) or anti-PTEN (Cell Signaling) antibody, respectively, from PC3, DU145, wild-type, and Pten/ MEFs subjected to nocodazole synchronization and release for 2 or 3 hr. The ubiquitin ligase activity of APC-CDH1 complex was performed as described in detail in the Extended Experimental Procedures. In Vivo Ubiquitination Assay PC3 cells were cotransfected with pCS2-6Myc-Cyclin B (wild-type or DBM), provided by H. Yu, GFP-PTENWT or -PTENK13,289E, and HA-ubiquitin (Ub). Alternatively, cells were transfected with a combination of His-Ub and wildtype or DBM Ets2, provided by P. Zhang (Li et al., 2008). The details of in vivo ubiquitination assay are described in the Extended Experimental Procedures. Cell Culture and Flow Cytometry Primary MEFs and PC3 cells complemented with wild-type or phosphataseinactive PTEN(C124S) were established and maintained as described in the Extended Experimental Procedures. The details of cell-cycle analysis by flow cytometry are also described in the Extended Experimental Procedures. siRNA and shRNA All the sources of siRNA duplexes and shRNA constructs used are described in the Extended Experimental Procedures. PtdIns(3,4,5)P3 Phosphatase Assay We purchased diC8-PtdIns(3,4,5)P3 and the Green Reagent from Echelon and Biomol, respectively. Measurement of phosphate released from the substrate was performed according to the manufacturer’s instructions and is described in detail in the Extended Experimental Procedures. In Vitro Kinase Assay The immunoprecipitates with Cyclin A2, Cyclin E1, or CDK2 prepared from cell lysates were incubated with histone H1 (Roche) and [g-32P]ATP (PerkinElmer) for 30 min at room temperature. Reaction mixture was resolved by SDS-PAGE, and then phosphorylated histone H1 was analyzed by SDS-PAGE and autoradiography. Cell Proliferation and Senescence Assay Typically, cell proliferation and senescence assays were performed as described (Chen et al., 2005), and the details of methods are described in the Extended Experimental Procedures. For cell proliferation assays with the PLK1 inhibitor BI 2536 (Lenart et al., 2007; Steegmaier et al., 2007) or the Aurora A inhibitor VX680 (Harrington et al., 2004), SV40-LT-immortalized wild-type and Pten/ MEFs or PC3 cells complemented with wild-type or PTEN(C124S) were incubated in the presence of the indicated concentrations
(C) The cell population in mitosis of PC3 cells complemented with wild-type or PTEN(C124S) and treated with 5 nM BI 2536 was analyzed by immunofluorescence with anti-phospo-histone H3 (P-H3(pS10)) (left) and quantified for a 48 hr period (right). Error bars represent SEM from three different experiments. (D) Pten null cells are hypersensitive to Aurora A inhibition. Growth inhibition of immortalized wild-type and Pten/ MEFs treated with the Aurora A inhibitor VX680 (500 nM) was measured after the 48 hr treatment (left). Western blots of cell lysates after the treatment with different concentrations of VX680 for 48 hr are shown on the right. Error bars represent SEM from three different experiments. (E) Growth inhibition of PC3 cells complemented with wild-type or PTEN(C124S) and treated with the Aurora A inhibitor VX680 (500 nM) was measured after the 48 hr treatment. Error bars represent SEM from three different experiments. (F) The cell population harboring >4 N DNA content of PC3 cells complemented with wild-type or PTEN(C124S) and treated with VX680 (500 nM) was analyzed by flow cytometry (left) and quantified for a 48 hr period (right). Error bars represent SEM from three different experiments. (G) A proposed model for phosphatase-independent role of nuclear PTEN toward APC-CDH1 to regulate proliferation and cellular senescence. See also Figure S7. p values were determined by Student’s t test (*p < 0.01; #p > 0.05).
Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc. 197
of BI 2536 or VX680 (Chemie Tek) at a 48 hr period, and then cell growth was measured. Mouse Xenograft Tumor Model PC3 cells complemented with wild-type or PTEN(C124S) (1 3 107 cells per site) in suspension were mixed with equal volumes of matrigel (BD Bioscience) and injected subcutaneously into 6-week-old male nude mice (Charles River laboratory). Measurement of tumor size was performed twice a week, and tumor volume was estimated using the following formula: volume = length 3 width2 3 0.526. RNA Isolation and Quantitative Real-Time RT-PCR Total RNA was isolated with Trizol reagent (Invitrogen) and reverse-transcribed with the Superscript III reverse transcriptase (Invitrogen). Expression of specific mRNAs was determined with the LightCycler (Roche) using the SYBR Green PCR Master Mix (QIAGEN). All the sources of quantitative realtime RT-PCR primers used are listed in the Extended Experimental Procedures. SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures and seven figures and can be found with this article online at doi:10.1016/j.cell. 2010.12.020. ACKNOWLEDGMENTS We are grateful to former and present members of the Pandolfi lab for experimental support, advice, or discussion. We are thankful to Drs. Manuel Serrano, Pumin Zhang, Wenyi Wei, and Hongtao Yu for providing reagents. We thank Dr. John Asara for mass-spectrometric analysis. This work was supported by NIH grants to P.P.P. A.C. is supported by a Long-Term Fellowship Award from the European Molecular Biology Organization, and L.S. is supported by a Fellowship from the Canadian Institutes of Health Research. Received: May 20, 2010 Revised: September 20, 2010 Accepted: December 15, 2010 Published: January 20, 2011 REFERENCES Ali, I.U., Schriml, L.M., and Dean, M. (1999). Mutational spectra of PTEN/ MMAC1 gene: a tumor suppressor with lipid phosphatase activity. J. Natl. Cancer Inst. 91, 1922–1932. Alimonti, A., Nardella, C., Chen, Z., Clohessy, J.G., Carracedo, A., Trotman, L.C., Cheng, K., Varmeh, S., Kozma, S.C., Thomas, G., et al. (2010). A novel type of cellular senescence that can be enhanced in mouse models and human tumor xenografts to suppress prostate tumorigenesis. J. Clin. Invest. 120, 681–693. Bassermann, F., Frescas, D., Guardavaccaro, D., Busino, L., Peschiaroli, A., and Pagano, M. (2008). The Cdc14B-Cdh1-Plk1 axis controls the G2 DNAdamage-response checkpoint. Cell 134, 256–267. Blanco-Aparicio, C., Renner, O., Leal, J.F., and Carnero, A. (2007). PTEN, more than the AKT pathway. Carcinogenesis 28, 1379–1386. Cairns, P., Okami, K., Halachmi, S., Halachmi, N., Esteller, M., Herman, J.G., Jen, J., Isaacs, W.B., Bova, G.S., and Sidransky, D. (1997). Frequent inactivation of PTEN/MMAC1 in primary prostate cancer. Cancer Res. 57, 4997–5000. Cardozo, T., and Pagano, M. (2004). The SCF ubiquitin ligase: insights into a molecular machine. Nat. Rev. Mol. Cell Biol. 5, 739–751. Carter, S.L., Eklund, A.C., Kohane, I.S., Harris, L.N., and Szallasi, Z. (2006). A signature of chromosomal instability inferred from gene expression profiles predicts clinical outcome in multiple human cancers. Nat. Genet. 38, 1043–1048.
198 Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc.
Chen, Z., Trotman, L.C., Shaffer, D., Lin, H.K., Dotan, Z.A., Niki, M., Koutcher, J.A., Scher, H.I., Ludwig, T., Gerald, W., et al. (2005). Crucial role of p53dependent cellular senescence in suppression of Pten-deficient tumorigenesis. Nature 436, 725–730. Chen, Z., Carracedo, A., Lin, H.K., Koutcher, J.A., Behrendt, N., Egia, A., Alimonti, A., Carver, B.S., Gerald, W., Teruya-Feldstein, J., et al. (2009). Differential p53-independent outcomes of p19(Arf) loss in oncogenesis. Sci. Signal. 2, ra44. Di Cristofano, A., Pesce, B., Cordon-Cardo, C., and Pandolfi, P.P. (1998). Pten is essential for embryonic development and tumour suppression. Nat. Genet. 19, 348–355. Eng, C. (2003). PTEN: one gene, many syndromes. Hum. Mutat. 22, 183–198. Feilotter, H.E., Nagai, M.A., Boag, A.H., Eng, C., and Mulligan, L.M. (1998). Analysis of PTEN and the 10q23 region in primary prostate carcinomas. Oncogene 16, 1743–1748. Fridberg, M., Servin, A., Anagnostaki, L., Linderoth, J., Berglund, M., Soderberg, O., Enblad, G., Rosen, A., Mustelin, T., Jerkeman, M., et al. (2007). Protein expression and cellular localization in two prognostic subgroups of diffuse large B-cell lymphoma: higher expression of ZAP70 and PKC-beta II in the non-germinal center group and poor survival in patients deficient in nuclear PTEN. Leuk. Lymphoma 48, 2221–2232. Garcia-Higuera, I., Manchado, E., Dubus, P., Canamero, M., Mendez, J., Moreno, S., and Malumbres, M. (2008). Genomic stability and tumour suppression by the APC/C cofactor Cdh1. Nat. Cell Biol. 10, 802–811. Georgescu, M.M., Kirsch, K.H., Kaloudis, P., Yang, H., Pavletich, N.P., and Hanafusa, H. (2000). Stabilization and productive positioning roles of the C2 domain of PTEN tumor suppressor. Cancer Res. 60, 7033–7038. Gildea, J.J., Herlevsen, M., Harding, M.A., Gulding, K.M., Moskaluk, C.A., Frierson, H.F., and Theodorescu, D. (2004). PTEN can inhibit in vitro organotypic and in vivo orthotopic invasion of human bladder cancer cells even in the absence of its lipid phosphatase activity. Oncogene 23, 6788–6797. Gimm, O., Perren, A., Weng, L.P., Marsh, D.J., Yeh, J.J., Ziebold, U., Gil, E., Hinze, R., Delbridge, L., Lees, J.A., et al. (2000). Differential nuclear and cytoplasmic expression of PTEN in normal thyroid tissue, and benign and malignant epithelial thyroid tumors. Am. J. Pathol. 156, 1693–1700. Ginn-Pease, M.E., and Eng, C. (2003). Increased nuclear phosphatase and tensin homologue deleted on chromosome 10 is associated with G0-G1 in MCF-7 cells. Cancer Res. 63, 282–286. Gray, I.C., Stewart, L.M., Phillips, S.M., Hamilton, J.A., Gray, N.E., Watson, G.J., Spurr, N.K., and Snary, D. (1998). Mutation and expression analysis of the putative prostate tumour-suppressor gene PTEN. Br. J. Cancer 78, 1296–1300. Harrington, E.A., Bebbington, D., Moore, J., Rasmussen, R.K., Ajose-Adeogun, A.O., Nakayama, T., Graham, J.A., Demur, C., Hercend, T., Diu-Hercend, A., et al. (2004). VX-680, a potent and selective small-molecule inhibitor of the Aurora kinases, suppresses tumor growth in vivo. Nat. Med. 10, 262–267. Huot, T.J., Rowe, J., Harland, M., Drayton, S., Brookes, S., Gooptu, C., Purkis, P., Fried, M., Bataille, V., Hara, E., et al. (2002). Biallelic mutations in p16 (INK4a) confer resistance to Ras- and Ets-induced senescence in human diploid fibroblasts. Mol. Cell. Biol. 22, 8135–8143. Koul, D., Jasser, S.A., Lu, Y., Davies, M.A., Shen, R., Shi, Y., Mills, G.B., and Yung, W.K. (2002). Motif analysis of the tumor suppressor gene MMAC/ PTEN identifies tyrosines critical for tumor suppression and lipid phosphatase activity. Oncogene 21, 2357–2364. Kraft, C., Herzog, F., Gieffers, C., Mechtler, K., Hagting, A., Pines, J., and Peters, J.M. (2003). Mitotic regulation of the human anaphase-promoting complex by phosphorylation. EMBO J. 22, 6598–6609. Lai, J.P., Dalton, J.T., and Knoell, D.L. (2007). Phosphatase and tensin homologue deleted on chromosome ten (PTEN) as a molecular target in lung epithelial wound repair. Br. J. Pharmacol. 152, 1172–1184. Lenart, P., Petronczki, M., Steegmaier, M., Di Fiore, B., Lipp, J.J., Hoffmann, M., Rettig, W.J., Kraut, N., and Peters, J.M. (2007). The small-molecule
inhibitor BI 2536 reveals novel insights into mitotic roles of polo-like kinase 1. Curr. Biol. 17, 304–315. Li, D.M., and Sun, H. (1997). TEP1, encoded by a candidate tumor suppressor locus, is a novel protein tyrosine phosphatase regulated by transforming growth factor beta. Cancer Res. 57, 2124–2129. Li, M., Shin, Y.H., Hou, L., Huang, X., Wei, Z., Klann, E., and Zhang, P. (2008). The adaptor protein of the anaphase promoting complex Cdh1 is essential in maintaining replicative lifespan and in learning and memory. Nat. Cell Biol. 10, 1083–1089. Lindsay, Y., McCoull, D., Davidson, L., Leslie, N.R., Fairservice, A., Gray, A., Lucocq, J., and Downes, C.P. (2006). Localization of agonist-sensitive PtdIns(3,4,5)P3 reveals a nuclear pool that is insensitive to PTEN expression. J. Cell Sci. 119, 5160–5168. Liu, J.L., Sheng, X., Hortobagyi, Z.K., Mao, Z., Gallick, G.E., and Yung, W.K. (2005). Nuclear PTEN-mediated growth suppression is independent of Akt down-regulation. Mol. Cell. Biol. 25, 6211–6224. Luo, J., Emanuele, M.J., Li, D., Creighton, C.J., Schlabach, M.R., Westbrook, T.F., Wong, K.K., and Elledge, S.J. (2009). A genome-wide RNAi screen identifies multiple synthetic lethal interactions with the Ras oncogene. Cell 137, 835–848. Maehama, T., and Dixon, J.E. (1998). The tumor suppressor, PTEN/MMAC1, dephosphorylates the lipid second messenger, phosphatidylinositol 3,4,5-trisphosphate. J. Biol. Chem. 273, 13375–13378. Maier, D., Jones, G., Li, X., Schonthal, A.H., Gratzl, O., Van Meir, E.G., and Merlo, A. (1999). The PTEN lipid phosphatase domain is not required to inhibit invasion of glioma cells. Cancer Res. 59, 5479–5482. Manchado, E., Eguren, M., and Malumbres, M. (2010). The anaphasepromoting complex/cyclosome (APC/C): cell-cycle-dependent and -independent functions. Biochem. Soc. Trans. 38, 65–71. Meraldi, P., Honda, R., and Nigg, E.A. (2004). Aurora kinases link chromosome segregation and cell division to cancer susceptibility. Curr. Opin. Genet. Dev. 14, 29–36. Ohtani, N., Zebedee, Z., Huot, T.J., Stinson, J.A., Sugimoto, M., Ohashi, Y., Sharrocks, A.D., Peters, G., and Hara, E. (2001). Opposing effects of Ets and Id proteins on p16INK4a expression during cellular senescence. Nature 409, 1067–1070. Okumura, K., Zhao, M., DePinho, R.A., Furnari, F.B., and Cavenee, W.K. (2005). PTEN: a novel anti-oncogenic function independent of phosphatase activity. Cell Cycle 4, 540–542. Perren, A., Komminoth, P., Saremaslani, P., Matter, C., Feurer, S., Lees, J.A., Heitz, P.U., and Eng, C. (2000). Mutation and expression analyses reveal differential subcellular compartmentalization of PTEN in endocrine pancreatic tumors compared to normal islet cells. Am. J. Pathol. 157, 1097–1103. Peters, J.M. (2006). The anaphase promoting complex/cyclosome: a machine designed to destroy. Nat. Rev. Mol. Cell Biol. 7, 644–656. Pines, J. (2006). Mitosis: a matter of getting rid of the right protein at the right time. Trends Cell Biol. 16, 55–63.
Rubin, S.J., Hallahan, D.E., Ashman, C.R., Brachman, D.G., Beckett, M.A., Virudachalam, S., Yandell, D.W., and Weichselbaum, R.R. (1991). Two prostate carcinoma cell lines demonstrate abnormalities in tumor suppressor genes. J. Surg. Oncol. 46, 31–36. Salmena, L., Carracedo, A., and Pandolfi, P.P. (2008). Tenets of PTEN tumor suppression. Cell 133, 403–414. Song, M.S., Salmena, L., Carracedo, A., Egia, A., Lo-Coco, F., Teruya-Feldstein, J., and Pandolfi, P.P. (2008). The deubiquitinylation and localization of PTEN are regulated by a HAUSP-PML network. Nature 455, 813–817. Stambolic, V., Suzuki, A., de la Pompa, J.L., Brothers, G.M., Mirtsos, C., Sasaki, T., Ruland, J., Penninger, J.M., Siderovski, D.P., and Mak, T.W. (1998). Negative regulation of PKB/Akt-dependent cell survival by the tumor suppressor PTEN. Cell 95, 29–39. Steck, P.A., Pershouse, M.A., Jasser, S.A., Yung, W.K., Lin, H., Ligon, A.H., Langford, L.A., Baumgard, M.L., Hattier, T., Davis, T., et al. (1997). Identification of a candidate tumour suppressor gene, MMAC1, at chromosome 10q23.3 that is mutated in multiple advanced cancers. Nat. Genet. 15, 356–362. Steegmaier, M., Hoffmann, M., Baum, A., Lenart, P., Petronczki, M., Krssak, M., Gurtler, U., Garin-Chesa, P., Lieb, S., Quant, J., et al. (2007). BI 2536, a potent and selective inhibitor of polo-like kinase 1, inhibits tumor growth in vivo. Curr. Biol. 17, 316–322. Strebhardt, K., and Ullrich, A. (2006). Targeting polo-like kinase 1 for cancer therapy. Nat. Rev. Cancer 6, 321–330. Sullivan, M., and Morgan, D.O. (2007). Finishing mitosis, one step at a time. Nat. Rev. Mol. Cell Biol. 8, 894–903. Sun, H., Lesche, R., Li, D.M., Liliental, J., Zhang, H., Gao, J., Gavrilova, N., Mueller, B., Liu, X., and Wu, H. (1999). PTEN modulates cell cycle progression and cell survival by regulating phosphatidylinositol 3,4,5,-trisphosphate and Akt/protein kinase B signaling pathway. Proc. Natl. Acad. Sci. USA 96, 6199–6204. Taylor, S., and Peters, J.M. (2008). Polo and Aurora kinases: lessons derived from chemical biology. Curr. Opin. Cell Biol. 20, 77–84. Trotman, L.C., Wang, X., Alimonti, A., Chen, Z., Teruya-Feldstein, J., Yang, H., Pavletich, N.P., Carver, B.S., Cordon-Cardo, C., Erdjument-Bromage, H., et al. (2007). Ubiquitination regulates PTEN nuclear import and tumor suppression. Cell 128, 141–156. Wang, C.X., Fisk, B.C., Wadehra, M., Su, H., and Braun, J. (2000). Overexpression of murine fizzy-related (fzr) increases natural killer cell-mediated cell death and suppresses tumor growth. Blood 96, 259–263. Wasch, R., Robbins, J.A., and Cross, F.R. (2010). The emerging role of APC/ CCdh1 in controlling differentiation, genomic stability and tumor suppression. Oncogene 29, 1–10.
Planchon, S.M., Waite, K.A., and Eng, C. (2008). The nuclear affairs of PTEN. J. Cell Sci. 121, 249–253.
Whiteman, D.C., Zhou, X.P., Cummings, M.C., Pavey, S., Hayward, N.K., and Eng, C. (2002). Nuclear PTEN expression and clinicopathologic features in a population-based series of primary cutaneous melanoma. Int. J. Cancer 99, 63–67.
Rosivatz, E., Matthews, J.G., McDonald, N.Q., Mulet, X., Ho, K.K., Lossi, N., Schmid, A.C., Mirabelli, M., Pomeranz, K.M., Erneux, C., et al. (2006). A small molecule inhibitor for phosphatase and tensin homologue deleted on chromosome 10 (PTEN). ACS Chem. Biol. 1, 780–790.
Zhou, X.P., Loukola, A., Salovaara, R., Nystrom-Lahti, M., Peltomaki, P., de la Chapelle, A., Aaltonen, L.A., and Eng, C. (2002). PTEN mutational spectra, expression levels, and subcellular localization in microsatellite stable and unstable colorectal cancers. Am. J. Pathol. 161, 439–447.
Cell 144, 187–199, January 21, 2011 ª2011 Elsevier Inc. 199
Global Regulation of H2A.Z Localization by the INO80 Chromatin-Remodeling Enzyme Is Essential for Genome Integrity Manolis Papamichos-Chronakis,1,3 Shinya Watanabe,1 Oliver J. Rando,2 and Craig L. Peterson1,* 1Program
in Molecular Medicine of Biochemistry and Molecular Pharmacology University of Massachusetts Medical School, Worcester, MA 01655, USA 3Present Address: Institut Curie, UMR218 CNRS, 26 rue d’Ulm, 75248 Paris Cedex 5, France, INSERM, ATIP-Avenir team, 75248 Paris Cedex 5, France *Correspondence:
[email protected] DOI 10.1016/j.cell.2010.12.021 2Department
SUMMARY
INO80 is an evolutionarily conserved, ATP-dependent chromatin-remodeling enzyme that plays roles in transcription, DNA repair, and replication. Here, we show that yeast INO80 facilitates these diverse processes at least in part by controlling genome-wide distribution of the histone variant H2A.Z. In the absence of INO80, H2A.Z nucleosomes are mislocalized, and H2A.Z levels at promoters show reduced responsiveness to transcriptional changes, suggesting that INO80 controls H2A.Z dynamics. Additionally, we demonstrate that INO80 has a histone-exchange activity in which the enzyme can replace nucleosomal H2A.Z/H2B with free H2A/H2B dimers. Genetic interactions between ino80 and htz1 support a model in which INO80 catalyzes the removal of unacetylated H2A.Z from chromatin as a mechanism to promote genome stability. INTRODUCTION DNA damage and aberrant chromosome replication can jeopardize genome integrity with serious effects to an organism’s health and survival. Several mechanisms have evolved in eukaryotic cells to cope with damaged DNA and to promote proper duplication of the genome. During recent years, it has become apparent that chromatin structure plays an essential role in maintaining genomic integrity (Groth et al., 2007; Peterson and Cote, 2004). Specialized chromatin structures are formed during the DNA damage response or within S phase, promoting DNA repair and stabilizing replication forks. However, our understanding of how chromatin contributes to genome stability remains limited. In addition to posttranslational modifications of histones, the building blocks of chromatin, incorporation of variant histones within chromatin regions provides an additional regulatory mechanism (Talbert and Henikoff, 2010). Histone variants such as H3.3 and H2A.Z are expressed throughout the cell cycle, 200 Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc.
and they can be incorporated into chromatin in the absence of DNA replication. Incorporation of the H2A-like H2A.Z into nucleosomal arrays alters their biophysical properties (Fan et al., 2002, 2004), potentially creating distinct chromatin structures that may regulate diverse metabolic processes. H2A.Z is highly conserved from yeast to human, and likewise the H2A.Z variant is enriched within nucleosomes at the proximal promoter regions of euchromatic genes of all eukaryotes (Mavrich et al., 2008; Raisner et al., 2005; Zhang et al., 2005). H2A.Z is also highly dynamic, being lost from promoters upon transcriptional activation at a rate that exceeds that of the core H3/H4 tetramer (Hardy et al., 2009; Zhang et al., 2005). The SWI2/SNF2 family of ATP-dependent chromatin-remodeling enzymes use the energy of ATP hydrolysis to alter histone-DNA interactions, leading to movements of nucleosomes in cis (sliding), loss of some or all histones, or the exchange of H2A/H2B dimers (Clapier and Cairns, 2009). The Ino80 and Swr1 ATPases belong to the INO80 subfamily of the SWI2/SNF2 group of remodeling enzymes (Morrison and Shen, 2009). Both Swr1 and Ino80 are subunits of highly conserved, multisubunit complexes, SWR-C and INO80, that share several common subunits (e.g., Rvb1,2) (Kobor et al., 2004; Krogan et al., 2003; Mizuguchi et al., 2004; Shen et al., 2000). INO80 can catalyze nucleosome sliding in cis (Shen et al., 2000), whereas SWR-C, or its metazoan ortholog SRCAP (Kobor et al., 2004; Krogan et al., 2003; Mizuguchi et al., 2004; Ruhl et al., 2006), directs incorporation of H2A.Z into nucleosomes by a dimer-exchange reaction (Mizuguchi et al., 2004). In addition to a role in transcription, genetic studies indicate that INO80 regulates the DNA damage checkpoint response (Morrison et al., 2007; Papamichos-Chronakis et al., 2006) and stabilizes stalled replication forks (Papamichos-Chronakis and Peterson, 2008). Even though the importance of INO80 in genome stability is apparent, it is still unclear how INO80 contributes to these processes. Here, we investigate the molecular mechanism of INO80 function in budding yeast. We present evidence indicating that INO80 regulates the genome-wide distribution of H2A.Z and that it promotes the eviction of H2A.Z from promoters during transcriptional induction. We also demonstrate that purified INO80
complex can incorporate H2A into an H2A.Z nucleosome in vitro, indicating that it has a histone-exchange activity that replaces nucleosomal H2A.Z/H2B with free H2A/H2B dimers. Notably, glutamine substitutions of the four N-terminal acetylatable lysine residues of H2A.Z alleviate the sensitivity of ino80 mutants to both replication stress and DNA damage-inducing agents. Our data suggest that removal and replacement of unacetylated H2A.Z from chromatin by INO80 is an essential mechanism for maintaining genome integrity. RESULTS Aberrant Genome-wide Localization of H2A.Z in the Absence of INO80 Previously, we reported that a partial deletion of the INO80 gene (ino80D300) led to increased levels of H2A.Z within chromatin that surrounded a single, unrepaired DNA double-strand break (DSB) (Papamichos-Chronakis et al., 2006). When H2A.Z distribution was analyzed by chromatin immunoprecipitation in a strain harboring a complete deletion of INO80 (ino80D), altered levels of H2A.Z were detected at many genomic locations even in the absence of a DSB (data not shown). In order to understand how H2A.Z localization may be altered upon inactivation of the INO80 complex, we analyzed the distribution of H2A.Z across 4% of the genome (chromosome III and 230 additional promoter regions) at single nucleosome resolution using a combination of mononucleosome-resolution ChIP with dense tiling microarrays (ChIP-chip). Briefly, cells from wild-type (WT) and ino80D strains were arrested in the G1 phase of the cell cycle, cells were fixed with formaldehyde, cross-linked chromatin was digested to mononucleosomes, and chromatin immunoprecipitation was conducted with a polyclonal antibody directed against H2A.Z. Total and immunoprecipitated DNA from the two strains were subsequently amplified, labeled with Cy3 and Cy5, and hybridized to a custom, high-resolution tiling microarray (Liu et al., 2005; Yuan et al., 2005). In agreement with published results, the mononucleosomal H2A.Z map produced from the WT cells confirmed the distinct enrichment pattern of H2A.Z (Figures 1C and 1F). H2A.Z enrichment spanned a wide dynamic range, with the 2.5% most-enriched nucleosomes exhibiting 16-fold enrichment of H2A.Z relative to the 2.5% most H2A.Z-depleted nucleosomes (Figure 1A). However, the dynamic range of nucleosomal H2A.Z levels in the ino80D mutant was significantly compressed, with 95% of nucleosomes captured within a 6-fold dynamic range (Figure 1A). This is visualized by plotting nucleosome enrichments for H2A.Z in the ino80D cells against those of the WT cells, showing that the slope of WT versus ino80D enrichments is well below one (y = 0.547x) (Figure 1B). Because of standard microarray normalization, these data are equally consistent with global overincorporation of H2A.Z, global underincorporation, or mixed gain and loss at specific loci. However, equal levels of bulk, chromatin-associated H2A.Z were detected in the WT and ino80D strains (Figure 1D). These results indicate that INO80 does not impact the total amount of H2A.Z that is incorporated into chromatin, but rather there is an extensive reorganization of nucleosomal H2A.Z across the genome in the absence of Ino80. This can be seen at many loci as spurious incorporation of H2A.Z in
the ino80D mutant strain, together with a drop in H2A.Z levels at H2A.Z-rich domains (Figure 1C and Figure S1 available online). To gain further insight into the regulation of H2A.Z localization by the INO80 complex, we focused on RNAPII-transcribed genes. The typical open reading frame is characterized by high levels of H2A.Z at the first nucleosome (+1) at the transcription start site (TSS), with variable H2A.Z at the upstream (1) nucleosome and low levels of H2A.Z downstream of the +1 nucleosome (Albert et al., 2007; Raisner et al., 2005; Zhang et al., 2005). As shown in Figure 1E, H2A.Z mislocalization was especially pronounced at promoter nucleosomes (y = 0.487x). Interestingly, the decrease of H2A.Z at the +1 nucleosome was associated with concomitant gain of H2A.Z at nucleosomes inside the coding sequences [mid- and 30 coding sequence (CDS)], (Figure 1F). Together, these data demonstrate that H2A.Z becomes globally mislocalized in the ino80D mutant, and they support a role for the INO80 complex in regulating proper genome-wide H2A.Z localization. The INO80 Complex Promotes Transcription-Associated H2A.Z Dynamics In budding yeast, H2A.Z occupancy negatively correlates with transcription rates, with H2A.Z being highly enriched in most gene promoters but depleted upstream of very highly transcribed genes (Zhang et al., 2005). As an initial test to investigate whether INO80 plays a role in this process, a mononucleosomal ChIP-chip assay for H2A.Z was conducted in G1- and G2/M-arrested WT and ino80D cells. Scatter plot analysis of H2A.Z nucleosome occupancy demonstrated that H2A.Z genomic occupancy is altered between the two cell-cycle phases in the WT strain (y = 0.635x; Figure 2A). However, in the ino80D mutant, the H2A.Z nucleosomal pattern remains largely unchanged, consistent with INO80 regulating the dynamics of H2A.Z-containing nucleosomes (y = 0.953x; Figure 2B). Expression of KAR4 is highly induced when cells are arrested in G1 by mating pheromone, and it is repressed in G2 phase (Kurihara et al., 1996). As shown in the heat maps of the KAR4 locus in the WT strain (Figure 2C, left), H2A.Z is enriched at the repressed KAR4 promoter in G2/M and becomes, as expected, depleted during transcriptional activation in G1 cells. As expected, the enrichment of H2A.Z at other, non-cell-cycle-regulated genes remains unchanged between G1 and G2/M samples (Figure S2C). In contrast, H2A.Z levels in the ino80D mutant remain high and similar to the repressed level in both G1 and G2/M phases (Figure 2C, right). Notably, induction of KAR4 expression is not affected by inactivation of Ino80 (Figure S2A), and thus transcription levels do not explain the altered H2A.Z dynamics. These results also indicate that the failure to deplete H2A.Z during transcriptional induction has little effect on KAR4 expression. These results suggest that INO80 controls either the eviction of the H2A.Z/H2B dimers or the loss of H2A.Z-containing nucleosomes that occurs during transcriptional induction. INO80 Regulates Transcription-Dependent H2A.Z Eviction To investigate how INO80 controls H2A.Z dynamics, we used nucleosome-scanning analysis at the KAR4 locus. Forty-five overlapping primer pairs were used to monitor the translational Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc. 201
Figure 1. Global Alterations in the H2A.Z Nucleosomal Incorporation upon Disruption of the INO80 Complex (A–C) Increased misincorporation of H2A.Z and compression of the H2A.Z-enrichment range in the absence of INO80 as measured by mononucleosomal ChIP-chip analysis. (A) Distribution of the log2 median ratios of H2A.Z enrichment in WT and ino80 nucleosome populations. Value 0 on the x axis reflects the average H2A.Z enrichment in each strain. (B) Scatter plot analysis comparing H2A.Z enrichment in WT versus ino80 mutant. Blue dots represent H2A.Z nucleosomes. The trendline equation indicates the global change of the distribution of H2A.Z in the ino80 strain compared to WT. (C) Heatmap from a mononucleosomal ChIP-chip experiment for H2A.Z conducted in WT and ino80 cells. Each box represents 140 nucleotides. Squares in the same column represent the same nucleosomes. Green signal indicates H2A.Z enrichment below average and red signal indicates positive enrichment. Black indicates average enrichment. Scheme represents the respective scanned region of chromosome III. (D) Equal levels of chromatin-bound H2A.Z in WT and ino80 strains. Chromatin fractions were isolated from WT and ino80 cells and immunoblot analysis was performed for the indicated proteins. Similar results were found when nucleosomal histones were released from the chromatin pellet by MNase digestion prior to the SDS-PAGE analysis (data not shown). (E and F) Pronounced mislocalization of H2A.Z at RNA PolII genes. (E) Scatter plot analysis comparing H2A.Z enrichment at promoter nucleosomes in the indicated strains. Analysis conducted as in (B). (F) Distribution analysis of H2A.Z enrichment at the first nucleosome downstream of TSS (+1 nucleosomes) and in the coding sequence (mid- and 30 CDS) in WT and ino80 strains. Analysis conducted as in (A). See also Figure S1.
position and dynamics of H2A.Z nucleosomes at the KAR4 promoter in the presence or absence of INO80. When samples were analyzed from G2/M-arrested WT or ino80 cells, this analysis yielded four peaks, indicating four positioned nucleosomes flank the KAR4 promoter in the repressed state (Figure 2D). Notably, the positioning of these promoter proximal nucleosomes was identical in the presence or absence of Ino80 (Figure 2D). Strikingly, KAR4 promoter sequences were also severely depleted from mononucleosome samples of both G1-arrested WT and ino80 cells, indicating that nucleosomes are depleted when KAR4 is expressed (Figures 2D and 2E). To measure H2A.Z occupancy at KAR4 promoter nucleosomes, we conducted chromatin immunoprecipitation of H2A.Z. In the repressed state (G2/M), H2A.Z is enriched at each of the four promoter nucleosomes, with the highest levels seen for nucleosome +1 (Figure 2F). At this promoter, similar 202 Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc.
levels are observed in the ino80D mutant, as expected from Figure 2C. Upon activation of KAR4 (G1 cells), the amount of H2A.Z per nucleosome is lower at several KAR4 nucleosomes in the WT strain, indicating that H2A.Z is evicted (Figure 2G and Figure S2B). In contrast, the amount of H2A.Z per nucleosome does not decrease in the ino80D mutant, with levels remaining at the repressed, G2/M level or even higher (Figure 2G and Figure S2B). Taken together, these results suggest two independent and complementary pathways for H2A.Z eviction—the first pathway is driven by complete nucleosome loss, and the second, H2A.Z-specific eviction, requires INO80. INO80 Exchanges Nucleosomal H2A.Z/H2B with Free H2A/H2B Dimers In Vitro These data indicate that INO80 regulates the genome-wide localization of H2A.Z as well as the eviction of H2A.Z during
transcriptional induction. One simple possibility is that INO80 might preferentially slide H2A.Z nucleosomes or evict H2A.Z octamers during transcriptional induction. However, INO80 shows no detectable octamer eviction activity with either H2A or H2A.Z mononucleosomes, and we find that INO80 mobilizes H2A or H2A.Z nucleosomes with equal efficiency (Figures S3A and S3B). Since Ino80 and Swr1 belong to the same subfamily of Snf2-like ATPases, we tested whether Ino80 might catalyze an ATP-dependent H2A.Z/H2B dimer-exchange event that removes H2A.Z and incorporates H2A. Histone-exchange assays were performed with mononucleosomes reconstituted with recombinant yeast histones. Initially, mononucleosomes were assembled with H2A/H2B dimers, and these substrates were incubated with remodeling enzyme and free HA-tagged H2A.Z/H2B dimers. Following incubation, the reactions were electrophoresed on native PAGE to separate bona fide mononucleosome products from free histones or other types of nucleosomal products. Histone exchange was evaluated by western blotting, probing for loss of H2A and incorporation of HA-tagged H2A.Z into the mononucleosome (Figure 3A). In all experiments, mononucleosome integrity was analyzed by both western analysis of histone H2B and by visualizing DNA with ethidium bromide (Figure 3 and data not shown). As expected, the SWR-C complex showed robust, ATP-dependent incorporation of HA-H2A.Z and significant loss of H2A. In contrast, and consistent with previous studies, Ino80 showed little activity in this H2A.Z incorporation assay (Figure 3B and Figure S3E). Importantly, these same preparations of Ino80 showed robust ATPase and nucleosomal sliding activities (Figures S3B–S3D). Next, mononucleosomes were assembled with H2A.Z/H2B dimers and incubated with remodeling enzyme and FLAGtagged H2A/H2B dimers. Strikingly, Ino80 catalyzed ATP-dependent incorporation of FLAG-H2A into a mononucleosome product, whereas SWR-C, SWI/SNF, and RSC were inactive on this substrate (Figure 3C and data not shown). The INO80-dependent incorporation of FLAG-H2A was concentration dependent and increased with time of incubation (Figure 3D and data not shown), and titration of the mononucleosome substrate indicates that efficient exchange activity for 5 nM INO80 requires > 50 nM nucleosomes (data not shown). Strikingly, INO80 action catalyzed removal of 35% of the H2A.Z from the initial mononucleosome substrate, while the levels of H2B remained constant (Figure 3E). Notably, INO80 does not exhibit promiscuous dimer eviction activity, as INO80 does not catalyze H2A eviction from an H2A-containing mononucleosome (Figure S3E). Thus, these data indicate that INO80 catalyzes a dimer-exchange reaction in which nucleosomal H2A.Z/H2B is replaced with an H2A/H2B dimer. Previous analyses of ATP-dependent dimer-exchange activities have used biotinylated chromatin substrates immobilized on streptavidin magnetic beads (Mizuguchi et al., 2004). In these assays, the immobilized substrate is incubated with remodeling enzyme and free histones, and exchange events are monitored by western blot following magnetic bead capture of the chromatin substrate. We performed this strategy with biotinylated mononucleosomes reconstituted with H2A.Z/H2B dimers, and we found that Ino80 catalyzed FLAG-H2A incorporation in this assay as well (Figure 3F). Furthermore, no detect-
able incorporation of FLAG-H2A was observed when INO80 was incubated with a mononucleosome reconstituted with an H2A/H2B dimer (Figure 3G). Together, both types of assays indicate that INO80 can specifically replace nucleosomal H2A.Z with H2A. Decreased H2A.Z Expression Rescues Replication Defects of an ino80 Mutant INO80 plays roles in many nuclear events, including gene transcription, DNA replication, DNA repair, and sister chromatid cohesion (Conaway and Conaway, 2009). One possibility is that INO80 regulates these diverse events by its action on H2A.Z and, consequently, the defects observed in an ino80 mutant may be due to the mislocalization and aberrant chromatin dynamics of H2A.Z. One simple prediction of this model is that H2A.Z depletion might rescue the defects of an ino80 mutant. Unfortunately, htz1D ino80D and swr1D ino80D double mutants are inviable, suggesting that H2A.Z and Ino80 may play additional, redundant role(s) in an essential function (Figure S4A and data not shown). To overcome this problem, we created isogenic WT and ino80D strains in which HTZ1 is expressed from a chromosomal, truncated promoter at 10% WT levels (HTZ1CP) (Figure 4A). This reduced expression of HTZ1 leads to a 4-fold decrease in bulk H2A.Z chromatin association, and a 2-fold decrease at the positioned nucleosomes of the KAR4 locus (Figure 4B and Figure S4B). The HTZ1 CP allele fully complements the growth defect and thiobendazol sensitivity of an htz1D strain (data not shown), indicating that this level of H2A.Z is sufficient to perform its known functions. Previously, we showed that ino80 cells are incapable of completing DNA replication when exposed to replication stress conditions (Papamichos-Chronakis and Peterson, 2008). We investigated whether decreased expression of H2A.Z can rescue this replication defect. WT and ino80D cells that expressed either normal (HTZ1) or low levels of H2A.Z (HTZ1CP) were arrested in G1 and then released into media containing 40 mM hydroxyurea (HU), and their progression through S phase was followed by fluorescence-activated cell sorting (FACS). Both HTZ1 and HTZ1CP WT strains progressed normally through S phase (Figure 4C). As we showed previously, ino80D cells that express normal levels of H2A.Z are rapidly blocked in S phase (Figure 4C, left). Interestingly, lowering the expression of H2A.Z restored a normal rate of S phase progression in HU media in the absence of Ino80 (Figure 4C, right). In contrast, both the ino80 and HTZ1CP ino80 cells failed to grow in media lacking inositol (data not shown), indicating that lowered expression of HTZ1 cannot support transcription of the INO1 gene in the absence of INO80. Thus, these results indicate a close functional relationship between INO80 and H2A.Z and suggest that aberrant H2A.Z incorporation may have a negative impact on DNA replication fork stability. Functional Interactions between INO80 Complex and H2A.Z Acetylation The N-terminal domain of yeast H2A.Z is acetylated in vitro at lysines 3, 8, 10, and 14 by the NuA4 HAT complex (Babiarz et al., 2006; Keogh et al., 2006; Millar et al., 2006), and Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc. 203
Figure 2. H2A.Z Is Not Lost from Nucleosomes in the Absence of INO80 (A and B) Scatter plot analyses comparing H2A.Z enrichment in G2/M- versus G1-arrested WT (A) and ino80 mutant (B) cells. (C) Indicative heatmaps of H2A.Z nucleosomal occupancy at the KAR4 locus under repressed (G2/M) and activated (G1) conditions in the indicated strains. (D and E) Nucleosome positioning and nucleosome loss at the KAR4 promoter. (D) Representative graph demonstrating nucleosome positioning in WT and ino80 cells at the indicated conditions as measured by amplification of genomic mononucleosomal DNA by qPCR. Values reflect the ratio of the amplified tested DNA over the total DNA purified from mononucleosomes. Scheme represents the promoter and coding region of KAR4. Numbers of the nucleosomes are relative to the TSS. (E) Nucleosome loss was measured from (D) as the fold decrease of the DNA amplified in repressed overinduced conditions after correction of the ratios of amplification achieved with total MNased DNA. (F and G) H2A.Z is not evicted from the KAR4 promoter during transcriptional induction in the ino80 strain. (F) Mononucleosomal-ChIP assay for H2A.Z was conducted in the WT and ino80 strains in the indicated conditions. Values reflect the average absolute amplification of the tested DNA from three independent
204 Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc.
acetylation occurs at promoter nucleosomes in vivo after incorporation of H2A.Z into chromatin by SWR-C (Keogh et al., 2006). Given that H2A.Z was mislocalized in the absence of Ino80, we tested whether H2A.Z acetylation levels might be altered in the ino80D mutant. Strikingly, H2A.Z-K14 acetylation levels were much lower in the ino80D strain compared to WT (Figure 5A). We entertained the possibility that this defect in H2A.Z acetylation contributes to the genome instability phenotypes of the ino80D mutant. However, a strain that harbors a derivative of H2A.Z that cannot be acetylated, H2A.Z-K3,8,10,14R, does not show sensitivity to DNA damage or replication stress agents (Millar et al., 2006), indicating that the lack of H2A.Z acetylation is insufficient to cause genome instability phenotypes. Interestingly, H2A.Z-K3,8,10,14R shows synthetic sensitivity to replication stress and DNA damage agents when expressed in an ino80D strain (Figure 5B). These results reveal a role for Htz1 acetylation in DDR and replication stress survival and suggest a functional connection between H2A.Z acetylation and INO80. HDA1 encodes a histone deacetylase that regulates H2A.Z acetylation (Lin et al., 2008). As shown in Figure 5A, inactivation of Hda1 led to a large increase in H2A.Z-K14 acetylation in both the WT and ino80 strains. These data support a simple model in which H2A.Z can be acetylated in the absence of INO80, but it is deacetylated by Hda1, possibly due to its mislocalization. This data raise the interesting possibility that the accumulation of deacetylated H2A.Z in the ino80 mutant might be detrimental to genome stability. Deletion of HDA1 renders cells sensitive to DNA damage-inducing agents like methylmethanesulfonate (MMS) and zeocin but not to replication stress induced by hydroxyurea (Begley et al., 2002 and Figure S5). In our initial studies, we found that an ino80D hda1D double mutant has a severe slow-growth phenotype that made growth assays problematic. To circumvent this issue, we monitored the phenotype of an arp8D hda1D double mutant that did not show this synthetic phenotype. The Arp8 subunit is essential for the chromatin-remodeling activities of INO80, and an arp8D mutant shows sensitivity to replication stress (HU treatment) and DNA damaging agents (zeocin). Strikingly, deletion of HDA1 suppresses the HU sensitivity of an arp8D mutant (Figure 5C). These results suggest that the replication defects caused by inactivation of the INO80 complex can be rescued by removing the Hda1 HDAC. The H2A.Z-K(3,8,10,14)Q Acetylation Mimic Suppresses Genomic Instability Caused by Disruption of the INO80 Complex To further test whether constitutive H2A.Z acetylation can alleviate ino80D phenotypes, we created a putative H2A.Z acetyl mimic (HTZ1 K3,8,10,14Q). Initially, we tested whether expression of H2A.Z-K3,8,10,14Q could rescue the replication defects of an arp8 mutant during replication stress conditions. WT and arp8 cells that express either H2A.Z or H2A.Z-K3,8,10,14Q
were arrested in G1 and subsequently released into 40 mM HU, and their progression through S phase was followed by FACS. As shown in Figure 6A, WT cells progressed through S phase and completed DNA replication in approximately 100 min. In contrast, the arp8 cells proceeded through S phase slowly, unable to fully replicate their genome even after almost 6 hr in HU. However, expression of the H2A.Z panacetyl mimic in the arp8 strain enabled cells to duplicate their genome, albeit slowly (Figure 6A). In addition, expression of the H2A.Z panacetyl mimic appears to alleviate the fork collapse phenotype of an arp8 mutant, as WT levels of DNA pola are recovered at a stalled replication fork in the absence of Arp8 (Figure S6A). Expression of H2A.Z-K3,8,10,14Q also alleviated the growth sensitivity of arp8, arp5, and ino80 mutants to HU, as well as to the DNA damage-inducing agents MMS and zeocin (Figure 6B). Importantly, the htz1-4KQ strain has no apparent phenotype in the presence of INO80 (Figure S6B). In contrast, expression of H2A.Z-K3,8,10,14Q did not alleviate the MMS or HU sensitivity of an mre11D mutant (Figure S6C), indicating that suppression is specific to mutations that disrupt the INO80 complex. Interestingly, expression of H2A.Z-K3,8,10,14Q did not suppress the inositol auxotrophy of an arp8 mutant, and ARP8-dependent transcription of the INO1 gene was not alleviated by H2A.Z-K3,8,10,14Q (Figures 6C and 6D). Importantly, suppression of arp8D genome stability phenotypes by the panacetyl mimic are eliminated after reintroduction of a WT copy of HTZ1 (Figure 6E). These data indicate that H2A.Z-K3,8,10,14Q is a potent suppressor of the genomic instability phenotypes of strains that lack the INO80 complex. One simple explanation for why the H2A.Z-K3,8,10,14Q might suppress ino80 phenotypes posits that this H2A.Z derivative is not properly expressed or that it restores the WT chromatin distribution and dynamics of H2A.Z in the absence of Ino80. We find, however, that H2A.Z-K3,8,10,14Q is expressed and incorporated into chromatin at levels similar to WT as measured by ChIP and nucleosome-scanning assays (Figures S6B, S6D, and S6E). Moreover, mapping of H2A.Z-K3,8,10,14Q at KAR4 nucleosomes demonstrated that both H2A.Z-K3,8,10,14Q and WT H2A.Z were incorporated in high amounts in the arp8 mutant compared to the WT strain, and neither H2A.Z-K3,8,10,14Q nor H2A.Z were lost from KAR4 promoter nucleosomes upon transcriptional induction in the absence of INO80 (Figures 6F and 6G). These results suggest that the activity of INO80 is not sensitive to the acetylation status of H2A.Z and that both WT and the H2A.Z-K3,8,10,14Q derivative require INO80 action for proper localization. Consistent with this view, the in vitro histone-exchange activity of INO80 is not affected by substitution of H2A.Z N-terminal lysines with either arginine or glutamine residues (Figure S6F and data not shown). Collectively, these results are consistent with a model in which the mislocalization of unacetylated H2A.Z in the absence of INO80 is detrimental to genome integrity but constitutive H2A.Z acetylation counteracts these inhibitory effects.
experiments. (G) Mononucleosomal-ChIP assay for H2A.Z was conducted in the WT and ino80 strains in the indicated conditions. Values reflect the average enrichment of H2A.Z normalized to the respective input DNA from three independent experiments. Each bar represents the average enrichment of the tested DNA from six primer pairs inside the corresponding nucleosomal region. See also Figure S2.
Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc. 205
B
A
SWR1
INO80
Mononucleosomes
ATP Incubate with remodelers and dimers
-
+
-
+
1.01
1.00
α-HA (H2A.Z)
Native PAGE
α-H2B Western blot H2B amount 1.00
1.07
HA-H2A.Z dimer / H2A Nuc
C INO80 -
ATP
SWR1
+
-
SWI/SNF
+
-
+
D INO80
α-FLAG (H2A) -
+
1.00
0.94
H2A.Z/H2B ratio 1.00
0.65
ATP
α-H2B H2B amount
α-FLAG (H2A) 1.00 0.96 1.00
0.90 1.00 0.91
α-HA (H2A.Z) α-H2B
H2A-FLAG dimer / H2A.Z Nuc
H2B amount
E
H2A-FLAG dimer / HA-H2A.Z Nuc
INO80
-
α-FLAG (H2A)
F INO80
α-H2B
ATP
H2A-FLAG dimer / H2A.Z nuc
-
+
α-FLAG (H2A)
α-H2B H2B amount
1.00
1.05 +/- 0.18
G SWR1 ATP
-
+
INO80 -
H2A-FLAG dimer / H2A.Z Nuc
+
α-FLAG (H2A)
α-H2B H2B amount
1.00
1.15
1.00
0.96
H2A-FLAG dimer / H2A Nuc
Figure 3. INO80 Has ATP-Dependent Histone-Exchange Activity (A) Scheme of in vitro dimer-exchange assay with recombinant yeast mononucleosomes. Mononucleosomes were incubated with remodeling enzymes and free dimers. Incorporation of dimers into the nucleosomes was analyzed by native PAGE and western blotting. (B) SWR-C incorporates H2A.Z/H2B dimers into H2A-containing nucleosomes. H2A-containing mononucleosomes (100 nM) were incubated with the indicated remodeling enzymes (5 nM) and free, HA-tagged H2A.Z/H2B dimers (50 nM) in the presence or absence of ATP. (C) INO80 incorporates H2A/H2B dimers into H2A.Z-containing nucleosomes. H2A.Z-containing mononucleosomes (100 nM) were incubated with the indicated remodeling enzymes (5 nM) and free FLAG-tagged H2A/H2B dimers (50 nM) in the presence or absence of ATP. (D) Concentration dependence of INO80 on H2A/H2B dimer incorporation activity. Increasing amounts (3, 6, and 12 nM) of INO80 was used as in (C). (E) INO80 catalyzes a histone-exchange event. HA-tagged, H2A.Z-containing mononucleosomes (100 nM) were incubated with INO80 (5 nM) and free FLAGtagged H2A/H2B dimers (50 nM) in the presence or absence of ATP.
206 Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc.
A
B
C
Figure 4. Reduced H2A.Z Expression Suppresses the Replication Defects of the ino80 Mutant Strain (A) Left: Schematic representation of the HTZ1 locus carrying the truncated HTZ1 promoter. Right: Reduced H2A.Zp levels from the crippled promoter. Acid-extracted proteins from logarithmically grown cells were separated by SDS-PAGE and immunoblot analysis was performed for H2A.Z. Equal gel loading was confirmed by western blotting against histone H3. (B) Reduced incorporation of HTZ1cp in chromatin. Chromatin fractionation assay in WT, HTZ1cp and HTZ1cp, and ino80 strains and total and chromatin fraction of proteins were separated by SDS-PAGE and analyzed by immunoblotting for H2A.Z. The asterisk (*) indicates an apparent cytoplasmic protein that cross-reacts with the anti-H2A.Z antibody and serves as a fractionation control. (C) Expression of HTZ1cp promotes replication during replication stress conditions in the absence of INO80. Cells from the indicated strains were synchronized in G1 phase with alpha factor (aF) and subsequently released into nocodazole-containing YPD media with 40 mM HU. Cell samples were collected at the indicated times and analyzed for DNA content by flow cytometry analysis. See also Figure S4.
a negative impact on DNA DSB repair and DNA replication fork stability.
DISCUSSION Whereas previous studies have focused on the key role of the yeast SWR-C- and mammalian SRCAP-remodeling enzymes in directing the ATP-dependent deposition of the H2A.Z histone variant, here we have shown that the related INO80 enzyme catalyzes the replacement of nucleosomal H2A.Z for H2A within coding regions and the eviction of H2A.Z during transcriptional activation. Interestingly, this role for INO80 appears essential for the maintenance of genome stability, as decreased expression of H2A.Z or expression of a H2A.Z panacetyl mimic alleviates the sensitivity of ino80, arp5, or arp8 mutants to DNA-damaging or replication stress agents. Thus, our genetic interactions suggest that aberrant accumulation of unacetylated H2A.Z has
ATP-Dependent Histone Exchange by the INO80 Subfamily of Enzymes Although members of the SWI/SNF subfamily of remodeling enzymes are able to evict histone dimers or entire octamers from nucleosomal substrates, only members of the INO80 subfamily exhibit histone dimer deposition and/or exchange activity. Both the yeast SWR-C and mammalian SRCAP members can replace nucleosomal H2A with H2A.Z, and in this study we report that yeast INO80 can perform the opposite reaction, converting an H2A.Z mononucleosome into one that contains H2A. Why hasn’t the dimerexchange activity of INO80 been detected previously? We find that optimal dimer-exchange activity requires nucleosome concentrations >50 nM (S. W., unpublished data), whereas most previously published dimer-exchange assays have used much lower nucleosome concentrations (<10 nM; Mizuguchi et al., 2004). The Ino80 and Swr1 ATPases are the only members of the Snf2 family of ATPases that contain very large, 300–500 amino acid insertions between Helicase/ATPase
(F) INO80 incorporates H2A/H2B dimers into H2A.Z-containing nucleosomes. Streptavidin bead-immobilized H2A.Z-containing, biotinylated mononucleosomes (100 nM) were incubated with INO80 (5 nM) and free FLAG-tagged H2A /H2B dimers (50 nM) in the presence or absence of ATP. After washing, incorporation of dimers into the nucleosomes was analyzed by SDS-PAGE and western blotting. H2B values reported in this experiment reflect the standard deviation from three independent experiments. (G) INO80 does not incorporate H2A/H2B dimers into H2A-containing nucleosomes. H2A-containing biotinylated mononucleosomes (100 nM) were incubated with remodeling enzymes (5 nM) and free FLAG-tagged H2A/H2B dimers (50 nM) in the presence or absence of ATP. See also Figure S3.
Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc. 207
A
WT ino80 hda1 ino80,hda1
H2A.Z K14ac H2A.Z Eaf3
B
YPD
15mM HU
3μg/ml Zeocin
WT HTZ1 4KR ino80 HTZ1 4KR / ino80
C YPD
20mM HU
WT hda1 arp8 arp8,hda1 Figure 5. Functional Interactions between INO80 and H2A.Z Acetylation (A) Decrease of H2A.Z-K14 acetylation in the ino80 mutant is dependent on Hda1. Acid-extracted proteins from WT, ino80, hda1, and ino80 hda1 doublemutant cells were separated by SDS-PAGE and assayed for total and K14 acetylated H2A.Z by western blotting. Equal gel loading was confirmed by immunoblot analysis against Eaf3. (B) H2A.Z acetylation is essential for the viability of the ino80 mutant cells in DNA damage and replication stress conditions. WT, HTZ1-K(3,8,10,14)R (HTZ1 4KR), ino80, and ino80 mutant cells expressing the HTZ1-K(3,8,10,14)R derivative (HTZ1 4KR,ino80) were plated in 10-fold serial dilutions on YPD plates containing the indicated concentration of HU or zeocin to induce DNA replication stress or DNA DSBs, respectively. Pictures of the plates were taken after 2–5 days of incubation at 30 C. (C) Deletion of HDA1 suppresses the replication defects of the arp8 strain during replication stress conditions. Log-phase cells from the indicated strains were plated in 10-fold serial dilutions on YPD plates containing 20 mM HU. Pictures of the plates were taken after 2–4 days of incubation at 30 C. See also Figure S5.
motifs III and IV, and it seems likely that this insertion influences the outcome of the remodeling reaction (Clapier and Cairns, 2009). In addition, each enzyme has a unique complement of histone-binding subunits that may determine the specificity of the deposition reaction. For instance, the SWR-C contains the SWC2 subunit, key for H2A.Z recognition (Wu et al., 2005) as well as the Yaf9 subunit, which harbors a YEATS domain that binds H3/H4 (Wang et al., 2009). Furthermore, the SWR-C and INO80 complexes each harbor the Arp4 and Arp5 subunits that interact with H2A/H2B dimers (Shen et al., 2003), and previous mass spectrometry data indicate that both H2A and H2A.Z are associated with the purified INO80 complex even in the absence of DNA damage (Mizuguchi et al., 2004). 208 Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc.
Regulation of H2A.Z Dynamics by the INO80 Complex Why is H2A.Z mislocalized in the absence of INO80 and how is INO80 action targeted to create the WT pattern? The SWR-C enzyme is localized predominantly to the +1 and/or 1 nucleosomes proximal to many RNAPII promoters, consistent with the deposition of H2A.Z at these locations (Venters and Pugh, 2009; Shimada et al., 2008). One possibility is that SWR-C typically incorporates H2A.Z within a large number of nucleosomes that encompass and flank a target promoter (Figure 7). In this case, we envision that INO80 confers boundary function, removing the H2A.Z from coding region nucleosomes, reinforcing the targeted, SWR-C-dependent deposition at promoter nucleosomes. In this model, both SWR-C and INO80 may be recruited to the promoter or coding region of a target gene, or, alternatively, INO80 may function in a more general fashion, much like that proposed for the global action of histone deacetylases. Two studies have provided evidence that INO80 may be targeted to the coding regions of many genes transcribed by RNAPII (Klopf et al., 2009; Venters and Pugh, 2009), perhaps through interactions with the transcription elongation complex. A second model is based on the fact that INO80 is associated with stalled and elongating replication forks (PapamichosChronakis and Peterson, 2008; Shimada et al., 2008; Vincent et al., 2008), and in this capacity INO80 controls elongation rate and fork stability. One attractive possibility is that INO80 at the replication fork may remove or replace H2A.Z that might be mislocalized during the chromatin assembly process that occurs following fork passage. In this case, H2A.Z may be deposited ectopically by fork-associated histone chaperones or deposited aberrantly by SWR-C. In either case, removal by Ino80 might then facilitate the reincorporation of H2A.Z at the proper locations by SWR-C. INO80 and H2A.Z Acetylation H2A.Z contains several lysine residues that are subject to reversible acetylation in all systems where it has been investigated. In Tetrahymena, these H2A.Z lysines are essential for cell viability (Ren and Gorovsky, 2001), whereas in budding and fission yeast the substitution of H2A.Z N-terminal lysines by arginine results in sensitivity to drugs that impact chromosome segregation but no other obvious phenotypes (Keogh et al., 2006; Kim et al., 2009). Acetylated H2A.Z is enriched at transcriptionally active promoters where H2A.Z is preferentially evicted, and it has been suggested that H2A.Z acetylation may facilitate reassembly of H2A.Z nucleosomes during gene repression (Millar et al., 2006). However, substitution alleles that remove H2A.Z lysines do not have a major impact on gene expression profiles (Millar et al., 2006). Unfortunately, commercial antibodies that recognize acetylated H2A.Z do not function in ChIP assays, so detailed analysis of the distribution and dynamics of acetylated H2A.Z has not been possible (Keogh et al., 2006; M.P.-C. and C.L.P, unpublished data). We were surprised to find that INO80 has a large impact on the steady state levels of H2A.Z acetylation. Indeed, the analysis of bulk H2A.Z acetylation suggests that most, if not all, of the mislocalized H2A.Z is likely to be unacetylated in an ino80 mutant. Strikingly, the genetics indicate that it is the
mislocalization of unacetylated H2A.Z that has a major impact on genome stability, not mislocalization of H2A.Z per se. The combination of mutations that disrupt the INO80 complex and the H2A.Z 4K-Q version suppresses the sensitivity of ino80, arp5, or arp8 mutants to DNA damage and replication stress agents. In contrast, expression of the H2A.Z 4K - > R version causes an enhanced sensitivity to these same agents. These data indicate that mislocalization of unacetylated H2A.Z is an inhibitor of genome stability that must either be acetylated or be removed by INO80. Why Is Mislocalized H2A.Z Detrimental for Genome Integrity? The distinctive enrichment of the H2A.Z histone variant at promoter proximal nucleosomes has led to the pervasive view that H2A.Z is a key regulator of transcription that creates a more permissive environment for transcriptional activation. In yeast, loss of H2A.Z has a relatively minor effect on gene expression profiles, typically affecting only the transcriptional kinetics of a subset of inducible genes (Meneghini et al., 2003). However, our studies indicate that mislocalized H2A.Z exerts a general, repressive effect on processes that prevent genomic instability. Thus, although the promoter localization of H2A.Z provides a sensitive readout for proper deposition, perhaps the prevention of H2A.Z mislocalization by INO80 is more important than actual promoter proximal positioning. One intriguing possibility is that promoter localization places H2A.Z in a location that enhances its removal, thereby limiting its inhibitory effects on genome stability and allowing it to be used as a mechanism of transcriptional regulation. Why does mislocalized, unacetylated H2A.Z impact DSB repair and replisome function? One possibility is that nucleosomal arrays that contain large amounts of H2A.Z assume more compact, folded states that block access of repair enzymes or destabilize stalled forks. Indeed, in vitro studies indicate that H2A.Z incorporation facilitates formation of condensed 30-nm-like fibers (Fan et al., 2002, 2004). Alternatively, perhaps H2A.Z nucleosomes are inherently more dynamic, and genome stability is impacted by the inappropriate localization of dynamic nucleosome hotspots. And finally, the acetylation state of H2A.Z may regulate interactions with protein(s) that promote or hinder genome stability. In any of these cases, the H2A.Z panacetyl mimic suppresses defects in both DNA damage repair and replication stress pathways because of loss of INO80, suggesting a common function for acetylated H2A.Z. Elucidating the mechanisms that protect genome stability is an essential step toward understanding and fighting devastating diseases like cancer (Halazonetis et al., 2008). Our work has uncovered a chromatin-mediated pathway essential for the maintenance of genome integrity that implicates the function of the INO80 chromatin-remodeling enzyme on H2A.Z-containing chromatin. Recently, two groups provided evidence that the human INO80 complex also participates in DNA damage repair and in DNA replication, promoting genome stability (Hur et al., 2010; Wu et al., 2007). Additionally, studies in cancer patients have reported overexpression of H2A.Z in several major types of malignancies (Dunican et al., 2002; Rhodes et al., 2004; Svotelis et al., 2010; Zucchi et al., 2004). Given that the INO80
complex is highly conserved throughout evolution, both structurally and functionally (Conaway and Conaway, 2009), it would be particularly interesting to test whether the metazoan INO80 complex, similar to its yeast counterpart, regulates the localization and dynamics of the H2A.Z histone variant in higher eukaryotes. EXPERIMENTAL PROCEDURES Chromatin Immunoprecipitation ChIPs were performed as described (Liu et al., 2005; Papamichos-Chronakis and Peterson, 2008) with commercially available polyclonal antibodies raised against H2A.Z (Millipore and Abcam antibodies were used for microarray analyses; Millipore and Active Motif antibodies were used for nucleosome-scanning assays). Antibody specificity was confirmed by both ChIP and western analyses (Figure S7). Mononucleosomes were prepared as described (Liu et al., 2005). The recovered DNA was subjected to quantitative real-time PCR. All ChIPs were performed at least twice and the variation between experiments was 10%–25%. Primers used in the PCR reactions are available upon request. Microarray hybridization and analysis were conducted as described (Liu et al., 2005). Chromatin Fractionation and Protein Analysis Chromatin fractionation was conducted as described (Liang and Stillman, 1997; Wang et al., 2009). For MNase release of nucleosome-associated proteins from the chromatin pellet, pellets were resuspended in 200 ml Lysis 1% Triton X buffer containing 1 mM CaCl2 and 15 units of MNase. Samples were incubated at 37 C for 20 min and reaction was stopped by the addition of 1 mM EGTA and 1 mM EDTA. Samples were subsequently centrifuged at 14,000 rpm for 5 min at 4 C, and the supernatant was recovered for protein and DNA analysis. Equal MNase digestion was confirmed by agarose gel visualization of the released DNA. Cell-Cycle Arrest and Flow Cytometry Analysis Cell-cycle arrest and FACS were performed as described (PapamichosChronakis and Peterson, 2008). Protein Purifications INO80-TAP and SWR1-TAP were purified as described (Sinha et al., 2009). ATPase assay and remodeling assays were performed as described (Logie and Peterson, 1999). Recombinant yeast histones were expressed and purified from Escherichia coli, and octamers were reconstituted as described (Luger et al., 1999a, 1999b). In Vitro Histone-Exchange Assay Mononucleosomes were reconstituted by salt dialysis onto a 200 bp DNA fragment containing the 601 nucleosome-positioning sequence. Mononucleosomes were incubated with remodeling enzymes, free histone dimers, and 2 mM ATP in exchange buffer (70 mM NaCl, 10 mM Tris-HCl [pH8.0], 5 mM MgCl2, 0.1 mg/ml BSA, and 1 mM DTT at 30 C for 60 min). To reconstitute biotinylated mononucleosomes, 200 bp 601 DNA fragment was generated by PCR with biotinylated DNA primers. The biotinylated mononuclesomes were immobilized onto Dynabeads M-280 (Invitrogen). After washing to remove unbound mononucleosomes, the immobilized mononucleosomes were incubated with remodeling enzymes, free histone dimers, and 2 mM ATP in exchange buffer at room temperature for 60 min. The immobilized mononucleosomes were washed three times with exchange buffer and subjected to SDS-PAGE and western blotting.
ACCESSION NUMBERS The Gene Expression Omnibus accession number for the microarray data reported in this paper is GSE25722.
Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc. 209
A
B
C
E
F D
G
Figure 6. The Genomic Instability Phenotypes Caused by Disruption of the INO80 Complex Are Rescued by the H2A.Z-K(3,8,10,14)Q Acetylation Mimic Mutant (A) WT, arp8, and arp8 mutant cells expressing the HTZ1-K(3,8,10,14)Q allele (HTZ1 4KQ,arp8) were synchronized in G1 phase with alpha factor (aF) and subsequently released into nocodazole-containing YPD media with 40 mM HU. Cell samples were collected at the indicated times and analyzed for DNA content by flow cytometry analysis.
210 Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc.
Figure 7. Proposed Model for the Role of the INO80 Chromatin-Remodeling Complex in Establishing the Proper Chromatin Localization of H2A.Z In this model, INO80 performs a boundary function by removing H2A.Z that is deposited distal to the targeted SWR-C enzyme. See text for details.
SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures and seven figures and can be found with this article online at doi:10.1016/ j.cell.2010.12.021.
Received: May 26, 2010 Revised: October 18, 2010 Accepted: December 15, 2010 Published: January 20, 2011 REFERENCES
ACKNOWLEDGMENTS This work was supported by grants from the National Institutes of Health to C.L.P. (GM54096). O.J.R. is supported in part by a Career Award in the Biomedical Sciences from the Burroughs Wellcome Fund and grants from the National Institute of General Medical Sciences and Human Frontier Science Program. M.P.-C. is supported by the Avenir Program from Inserm. We thank John Holik [University of Massachusetts Medical School (UMMS)] for assistance with the tiling array studies, Jerry Workman (Stowers Institute) for yeast histone expression vectors, John Lescyz (UMMS) for mass spectrometry analysis, Nicholas Adkins (UMMS) for the western blot with Millipore a-Htz1 sera (Figure S7), Erica Hong (Harvard Medical School) for help with Figure 7, members of the Peterson lab for helpful discussions throughout the course of this work, and Genevieve Almouzni (Curie Institute), Angela Taddei (Curie Institute), and Valerie Borde (Curie Institute)for critical reading of the manuscript.
Albert, I., Mavrich, T.N., Tomsho, L.P., Qi, J., Zanton, S.J., Schuster, S.C., and Pugh, B.F. (2007). Translational and rotational settings of H2A.Z nucleosomes across the Saccharomyces cerevisiae genome. Nature 446, 572–576. Babiarz, J.E., Halley, J.E., and Rine, J. (2006). Telomeric heterochromatin boundaries require NuA4-dependent acetylation of histone variant H2A.Z in Saccharomyces cerevisiae. Genes Dev. 20, 700–710. Begley, T.J., Rosenbach, A.S., Ideker, T., and Samson, L.D. (2002). Damage recovery pathways in Saccharomyces cerevisiae revealed by genomic phenotyping and interactome mapping. Mol. Cancer Res. 1, 103–112. Clapier, C.R., and Cairns, B.R. (2009). The biology of chromatin remodeling complexes. Annu. Rev. Biochem. 78, 273–304. Conaway, R.C., and Conaway, J.W. (2009). The INO80 chromatin remodeling complex in transcription, replication and repair. Trends Biochem. Sci. 34, 71–77.
(B) WT and arp8, ino80, or arp5 single mutant cells ectopically expressing from the URA3 locus either the WT (HTZ1) or the HTZ1-K(3,8,10,14)Q (HTZ1 4KQ,arp8) allele were plated in 10-fold serial dilutions on YPD plates containing the indicated concentration of HU, methylmethanesulfonate (MMS) or zeocin to induce DNA replication stress, DNA damage during replication, or DNA DSBs, respectively. Pictures of the plates were taken after 2–5 days incubation at 30 C. (C) RT-qPCR analysis of INO1 mRNA isolated from the indicated strains grown either in the presence of inositol [yeast extract, peptone, dextrose (YPD)] or for 2 hours in synthetic media lacking inositol (inositol). The values of INO1 mRNA were normalized to the respective ACT1 transcripts and normalized INO1 transcript in WT cells grown in the absence of inositol was arbitrarily set as 100. Error bars represent variation between two separate experiments. (D) Cells from the indicated strains were plated in 10-fold serial dilutions on YPD plates or plates containing synthetic media lacking inositol. Pictures of the plates were taken after 2–4 days incubation at 30 C. (E) Resistance of the HTZ1-K(3,8,10,14)Q arp8 cells to genomic instability-inducing agents is due to the HTZ1-K(3,8,10,14)Q locus. pHTZ1/arp8, htz1D and pHTZ1/arp8, htz1-4KQD strains were plated in 10-fold serial dilutions on YPD plates containing the indicated concentration of HU and zeocin. Pictures were taken after 2–4 days days incubation at 30 C. (F) Nucleosome and H2A.Z loss during transcriptional induction are similar in the HTZ1 arp8 (arp8) and htz1-K(3,8,10,14)Q arp8 (htz1-4KQ arp8) strains. Representative graphs demonstrating nucleosome and H2A.Z loss at the KAR4 promoter in the indicated strains. The fold decrease rates were measured as in Figure 2E. (G) Similar high enrichment of H2A.Z and H2A.Z-K(3,8,10,14)Q at the KAR4 locus in the absence of arp8. Mononucleosomal-ChIP for H2A.Z was conducted in chromatin from the indicated strains arrested in G1 by alpha factor. Analysis was conducted as in Figure 2G. See also Figure S6.
Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc. 211
Dunican, D.S., McWilliam, P., Tighe, O., Parle-McDermott, A., and Croke, D.T. (2002). Gene expression differences between the microsatellite instability (MIN) and chromosomal instability (CIN) phenotypes in colorectal cancer revealed by high-density cDNA array hybridization. Oncogene 21, 3253–3257. Fan, J.Y., Gordon, F., Luger, K., Hansen, J.C., and Tremethick, D.J. (2002). The essential histone variant H2A.Z regulates the equilibrium between different chromatin conformational states. Nat. Struct. Biol. 9, 172–176. Fan, J.Y., Rangasamy, D., Luger, K., and Tremethick, D.J. (2004). H2A.Z alters the nucleosome surface to promote HP1alpha-mediated chromatin fiber folding. Mol. Cell 16, 655–661. Groth, A., Rocha, W., Verreault, A., and Almouzni, G. (2007). Chromatin challenges during DNA replication and repair. Cell 128, 721–733. Halazonetis, T.D., Gorgoulis, V.G., and Bartek, J. (2008). An oncogeneinduced DNA damage model for cancer development. Science 319, 1352– 1355. Hardy, S., Jacques, P.E., Gevry, N., Forest, A., Fortin, M.E., Laflamme, L., Gaudreau, L., and Robert, F. (2009). The euchromatic and heterochromatic landscapes are shaped by antagonizing effects of transcription on H2A.Z deposition. PLoS Genet. 5, e1000687. Hur, S.K., Park, E.J., Han, J.E., Kim, Y.A., Kim, J.D., Kang, D., and Kwon, J. (2010). Roles of human INO80 chromatin remodeling enzyme in DNA replication and chromosome segregation suppress genome instability (. Cell. Mol. Life Sci. 67, 2283–2296. Keogh, M.C., Mennella, T.A., Sawa, C., Berthelet, S., Krogan, N.J., Wolek, A., Podolny, V., Carpenter, L.R., Greenblatt, J.F., Baetz, K., et al. (2006). The Saccharomyces cerevisiae histone H2A variant Htz1 is acetylated by NuA4. Genes Dev. 20, 660–665. Kim, H.S., Vanoosthuyse, V., Fillingham, J., Roguev, A., Watt, S., Kislinger, T., Treyer, A., Carpenter, L.R., Bennett, C.S., Emili, A., et al. (2009). An acetylated form of histone H2A.Z regulates chromosome architecture in Schizosaccharomyces pombe. Nat. Struct. Mol. Biol. 16, 1286–1293. Klopf, E., Paskova, L., Sole, C., Mas, G., Petryshyn, A., Posas, F., Wintersberger, U., Ammerer, G., and Schuller, C. (2009). Cooperation between the INO80 complex and histone chaperones determines adaptation of stress gene transcription in the yeast Saccharomyces cerevisiae. Mol. Cell. Biol. 29, 4994–5007. Kobor, M.S., Venkatasubrahmanyam, S., Meneghini, M.D., Gin, J.W., Jennings, J.L., Link, A.J., Madhani, H.D., and Rine, J. (2004). A protein complex containing the conserved Swi2/Snf2-related ATPase Swr1p deposits histone variant H2A.Z into euchromatin. PLoS Biol. 2, E131. Krogan, N.J., Keogh, M.C., Datta, N., Sawa, C., Ryan, O.W., Ding, H., Haw, R.A., Pootoolal, J., Tong, A., Canadien, V., et al. (2003). A Snf2 family ATPase complex required for recruitment of the histone H2A variant Htz1. Mol. Cell 12, 1565–1576. Kurihara, L.J., Stewart, B.G., Gammie, A.E., and Rose, M.D. (1996). Kar4p, a karyogamy-specific component of the yeast pheromone response pathway. Mol. Cell. Biol. 16, 3990–4002. Liang, C., and Stillman, B. (1997). Persistent initiation of DNA replication and chromatin-bound MCM proteins during the cell cycle in cdc6 mutants. Genes Dev. 11, 3375–3386. Lin, Y.Y., Qi, Y., Lu, J.Y., Pan, X., Yuan, D.S., Zhao, Y., Bader, J.S., and Boeke, J.D. (2008). A comprehensive synthetic genetic interaction network governing yeast histone acetylation and deacetylation. Genes Dev. 22, 2062–2074. Liu, C.L., Kaplan, T., Kim, M., Buratowski, S., Schreiber, S.L., Friedman, N., and Rando, O.J. (2005). Single-nucleosome mapping of histone modifications in S. cerevisiae. PLoS Biol. 3, e328. Logie, C., and Peterson, C.L. (1999). Purification and biochemical properties of yeast SWI/SNF complex. Methods Enzymol. 304, 726–741. Luger, K., Rechsteiner, T.J., and Richmond, T.J. (1999a). Expression and purification of recombinant histones and nucleosome reconstitution. Methods Mol. Biol. 119, 1–16. Luger, K., Rechsteiner, T.J., and Richmond, T.J. (1999b). Preparation of nucleosome core particle from recombinant histones. Methods Enzymol. 304, 3–19.
212 Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc.
Mavrich, T.N., Jiang, C., Ioshikhes, I.P., Li, X., Venters, B.J., Zanton, S.J., Tomsho, L.P., Qi, J., Glaser, R.L., Schuster, S.C., et al. (2008). Nucleosome organization in the Drosophila genome. Nature 453, 358–362. Meneghini, M.D., Wu, M., and Madhani, H.D. (2003). Conserved histone variant H2A.Z protects euchromatin from the ectopic spread of silent heterochromatin. Cell 112, 725–736. Millar, C.B., Xu, F., Zhang, K., and Grunstein, M. (2006). Acetylation of H2AZ Lys 14 is associated with genome-wide gene activity in yeast. Genes Dev. 20, 711–722. Mizuguchi, G., Shen, X., Landry, J., Wu, W.H., Sen, S., and Wu, C. (2004). ATPdriven exchange of histone H2AZ variant catalyzed by SWR1 chromatin remodeling complex. Science 303, 343–348. Morrison, A.J., and Shen, X. (2009). Chromatin remodelling beyond transcription: the INO80 and SWR1 complexes. Nat. Rev. Mol. Cell Biol. 10, 373–384. Morrison, A.J., Kim, J.A., Person, M.D., Highland, J., Xiao, J., Wehr, T.S., Hensley, S., Bao, Y., Shen, J., Collins, S.R., et al. (2007). Mec1/Tel1 phosphorylation of the INO80 chromatin remodeling complex influences DNA damage checkpoint responses. Cell 130, 499–511. Papamichos-Chronakis, M., and Peterson, C.L. (2008). The Ino80 chromatinremodeling enzyme regulates replisome function and stability. Nat. Struct. Mol. Biol. 15, 338–345. Papamichos-Chronakis, M., Krebs, J.E., and Peterson, C.L. (2006). Interplay between Ino80 and Swr1 chromatin remodeling enzymes regulates cell cycle checkpoint adaptation in response to DNA damage. Genes Dev. 20, 2437–2449. Peterson, C.L., and Cote, J. (2004). Cellular machineries for chromosomal DNA repair. Genes Dev. 18, 602–616. Raisner, R.M., Hartley, P.D., Meneghini, M.D., Bao, M.Z., Liu, C.L., Schreiber, S.L., Rando, O.J., and Madhani, H.D. (2005). Histone variant H2A.Z marks the 50 ends of both active and inactive genes in euchromatin. Cell 123, 233–248. Ren, Q., and Gorovsky, M.A. (2001). Histone H2A.Z acetylation modulates an essential charge patch. Mol. Cell 7, 1329–1335. Rhodes, D.R., Yu, J., Shanker, K., Deshpande, N., Varambally, R., Ghosh, D., Barrette, T., Pandey, A., and Chinnaiyan, A.M. (2004). Large-scale meta-analysis of cancer microarray data identifies common transcriptional profiles of neoplastic transformation and progression. Proc. Natl. Acad. Sci. USA 101, 9309–9314. Ruhl, D.D., Jin, J., Cai, Y., Swanson, S., Florens, L., Washburn, M.P., Conaway, R.C., Conaway, J.W., and Chrivia, J.C. (2006). Purification of a human SRCAP complex that remodels chromatin by incorporating the histone variant H2A.Z into nucleosomes. Biochemistry 45, 5671–5677. Shen, X., Mizuguchi, G., Hamiche, A., and Wu, C. (2000). A chromatin remodelling complex involved in transcription and DNA processing. Nature 406, 541–544. Shen, X., Ranallo, R., Choi, E., and Wu, C. (2003). Involvement of actin-related proteins in ATP-dependent chromatin remodeling. Mol. Cell 12, 147–155. Shimada, K., Oma, Y., Schleker, T., Kugou, K., Ohta, K., Harata, M., and Gasser, S.M. (2008). Ino80 chromatin remodeling complex promotes recovery of stalled replication forks. Curr. Biol. 18, 566–575. Sinha, M., Watanabe, S., Johnson, A., Moazed, D., and Peterson, C.L. (2009). Recombinational repair within heterochromatin requires ATP-dependent chromatin remodeling. Cell 138, 1109–1121. Svotelis, A., Gevry, N., Grondin, G., and Gaudreau, L. (2010). H2A.Z overexpression promotes cellular proliferation of breast cancer cells. Cell Cycle 9, 364–370. Talbert, P.B., and Henikoff, S. (2010). Histone variants–ancient wrap artists of the epigenome. Nat. Rev. Mol. Cell Biol. 11, 264–275. Venters, B.J., and Pugh, B.F. (2009). A canonical promoter organization of the transcription machinery and its regulators in the Saccharomyces genome. Genome Res. 19, 360–371. Vincent, J.A., Kwong, T.J., and Tsukiyama, T. (2008). ATP-dependent chromatin remodeling shapes the DNA replication landscape. Nat. Struct. Mol. Biol. 15, 477–484.
Wang, A.Y., Schulze, J.M., Skordalakes, E., Gin, J.W., Berger, J.M., Rine, J., and Kobor, M.S. (2009). Asf1-like structure of the conserved Yaf9 YEATS domain and role in H2A.Z deposition and acetylation. Proc. Natl. Acad. Sci. USA 106, 21573–21578. Wu, W.H., Alami, S., Luk, E., Wu, C.H., Sen, S., Mizuguchi, G., Wei, D., and Wu, C. (2005). Swc2 is a widely conserved H2AZ-binding module essential for ATP-dependent histone exchange. Nat. Struct. Mol. Biol. 12, 1064–1071. Wu, S., Shi, Y., Mulligan, P., Gay, F., Landry, J., Liu, H., Lu, J., Qi, H.H., Wang, W., Nickoloff, J.A., et al. (2007). A YY1-INO80 complex regulates genomic stability through homologous recombination-based repair. Nat. Struct. Mol. Biol. 14, 1165–1172.
Yuan, G.C., Liu, Y.J., Dion, M.F., Slack, M.D., Wu, L.F., Altschuler, S.J., and Rando, O.J. (2005). Genome-scale identification of nucleosome positions in S. cerevisiae. Science 309, 626–630. Zhang, H., Roberts, D.N., and Cairns, B.R. (2005). Genome-wide dynamics of Htz1, a histone H2A variant that poises repressed/basal promoters for activation through histone loss. Cell 123, 219–231. Zucchi, I., Mento, E., Kuznetsov, V.A., Scotti, M., Valsecchi, V., Simionati, B., Vicinanza, E., Valle, G., Pilotti, S., Reinbold, R., et al. (2004). Gene expression profiles of epithelial cells microscopically isolated from a breast-invasive ductal carcinoma and a nodal metastasis. Proc. Natl. Acad. Sci. USA 101, 18147–18152.
Cell 144, 200–213, January 21, 2011 ª2011 Elsevier Inc. 213
Polycomb-Dependent Regulatory Contacts between Distant Hox Loci in Drosophila Fre´de´ric Bantignies,1,3,* Virginie Roure,1,3,4 Itys Comet,1,5 Benjamin Leblanc,1 Bernd Schuettengruber,1 Je´roˆme Bonnet,1,6 Vanessa Tixier,1,7 Andre´ Mas,2 and Giacomo Cavalli1,* 1Institut
de Ge´ne´tique Humaine, CNRS UPR 1142, 141, rue de la Cardonille, 34396 Montpellier Cedex 5, France de Mode´lisation Mathe´matique de Montpellier, CNRS UMR 5149, Universite´ Montpellier 2, Place Euge`ne Bataillon, 34095 Montpellier Cedex 5, France 3These authors contributed equally to this work 4Present address: Max-Planck Institute of Immunobiology, Stu ¨ beweg 51, D-79108 Freiburg, Germany 5Present address: Biotech Research & Innovation Centre, University of Copenhagen, Ole Maaløes Vej 5, DK-2200 Copenhagen N, Denmark 6Present address: Department of Bioengineering, Stanford University, Stanford, CA 93405, USA 7Present address: Unite ´ de Ge´ne´tique, Reproduction et De´veloppement, INSERM-CNRS UMR 6247, Universite´ de Clermont-Ferrand, 28 Place Henri Dunant, 63000 Clermont-Ferrand, France *Correspondence:
[email protected] (F.B.),
[email protected] (G.C.) DOI 10.1016/j.cell.2010.12.026 2Institut
SUMMARY
In Drosophila melanogaster, Hox genes are organized in an anterior and a posterior cluster, called Antennapedia complex and bithorax complex, located on the same chromosome arm and separated by 10 Mb of DNA. Both clusters are repressed by Polycomb group (PcG) proteins. Here, we show that genes of the two Hox complexes can interact within nuclear PcG bodies in tissues where they are corepressed. This colocalization increases during development and depends on PcG proteins. Hox gene contacts are conserved in the distantly related Drosophila virilis species and they are part of a large gene interaction network that includes other PcG target genes. Importantly, mutations on one of the loci weaken silencing of genes in the other locus, resulting in the exacerbation of homeotic phenotypes in sensitized genetic backgrounds. Thus, the three-dimensional organization of Polycomb target genes in the cell nucleus stabilizes the maintenance of epigenetic gene silencing. INTRODUCTION The organization of chromosomal domains in the cell nucleus plays an important role in the regulation of gene expression during cellular differentiation and development (Fraser and Bickmore, 2007; Williams et al., 2010). During interphase, eukaryotic chromosomes are organized as distinct domains called chromosome territories, which adopt specific structure and position in the cell nucleus. These territories are not strictly delimited in the nuclear space, allowing for some intermingling with other chromosomes (Branco and Pombo, 2006). It has been reported that chromosomal elements separated by large genomic distances are sometimes able to interact in the nuclear space. 214 Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc.
The phenomenon of long-range contacts between different chromosomal loci has been called ‘‘chromosome kissing’’ (Cavalli, 2007). Chromosome kissing has been reported for X chromosome inactivation, where the two copies come transiently in close proximity at the onset of the inactivation process (Okamoto and Heard, 2009). The same phenomenon has been observed in few other situations. However, it is not clear how these contacts may contribute to gene regulation in living organisms (Sexton et al., 2009; Williams et al., 2010). Long-range gene regulation may involve epigenetic components including proteins of the Polycomb group (PcG) (Grimaud et al., 2006; Sexton et al., 2007). PcG proteins are organized into nuclear foci called PcG bodies (Alkema et al., 1997; Buchenau et al., 1998; Messmer et al., 1992; Ren et al., 2008; Saurin et al., 1998). These bodies contain silenced PcG target chromatin (Grimaud et al., 2006; Lanzuolo et al., 2007), which is made of multimeric PcG complexes bound to cis-regulatory elements named PcG response elements (PREs) (Muller and Verrijzer, 2009; Schuettengruber et al., 2007). PcG protein binding to PREs silences genes involved in developmental patterning and cell proliferation (Merdes and Paro, 2009). In Drosophila melanogaster, Hox genes are the most prominent PcG targets. They are organized in two complexes: the Antennapedia complex (ANT-C) spans approximately 400 kb and comprises five Hox genes (lab, pb, Dfd, Scr, and Antp) that specify parts of the head and the anterior thorax (Kaufman et al., 1990), while the bithorax complex (BX-C) spans approximately 350 kb and contains three Hox genes (Ubx, abd-A, and Abd-B) involved in the development of the posterior thorax and the abdomen (Duncan, 1987; Lewis et al., 2003). The nuclear organization of the BX-C has been studied by Fluorescent In Situ Hybridization (FISH) and Chromosome Conformation Capture (3C) approaches (Lanzuolo et al., 2007). This study revealed the existence of long-distance interactions among the major elements bound by PcG proteins, including PREs and core promoters. Importantly, upon activation of the Abd-B gene, its interactions with the other genes of the complex are lost. Two particular PRE-containing elements called Fab-7 and
A
Antennapedia complex (ANT-C) Prox. pb
Dfd
B
Scr
Antp
Abd-B
Antp
Merge
Head
Abd-B
abd-A
20 15 10
PS4/5
Percentage of pairing
Embryos: Antp - Abd-B
Ubx
Dapi
C 25
Dist.
10 Mb lab
Chr. 3R
Figure 1. Kissing of Repressed Hox Genes
bithorax complex (BX-C)
* *
5
Head 1 (N= 276)
PS4/5 2 (N= 241)
Dapi
Posterior 3 (N= 326)
Abd-B
Antp
Polycomb
Merge
Posterior
Head
D
Posterior
0
E Percentage of pairing
Embryos:
F
25
25
20
20
15
15
Embryo Heads: Antp - Abd-B
G 30 25 20 15
10
10
Eye discs: Antp - Abd-B
*
*
(A) Schematic drawing illustrating the anterior and the posterior Hox gene clusters in D. melanogaster. The colored lines represent the approximate localization of the FISH probes for Antp (red) and Abd-B (green). (B) FISH in wild-type (WT) stage 10–11 embryos. Percentage colocalization between Antp and Abd-B. A maximum distance of 350 nm was used to define pairing between the two loci. The p values of the pairwise comparison are 2.045e-06 for Head versus PS4/5, 1.1e-04 for Head versus Posterior, 0.137 for PS4/5 versus Posterior. (C) Characteristic examples of individual nuclei. (D) FISH-I of Abd-B, Antp and PC. The regions chosen for image acquisition are indicated to the left. Figure images correspond to deconvolved single slices from 3D stacks. The scale bars represent 1 mm. (E) Percentage colocalization between beat-Vc and CG17622, two non PcG-targets, which are located 10 Mb away on Chromosome 3R. The p value is 0.707. (F) Percentage of colocalization between Antp and Abd-B in heads of stage 13–14 embryos from WT, PcXL5 and Pcl10 homozygous mutants. The p values are 1.066e-06 for WT versus PcXL5, 1.936e-09 for WT versus Pcl10, 0.259 for PcXL5 versus Pcl10. (G) Percentage of colocalization between Antp and Abd-B in third instar larval imaginal eye discs from WT, PcXL5 and Pcl10 heterozygous mutants. The p values are 2.3e-03 for WT versus PcXL5, 2.372e-04 for WT versus Pcl10, 0.623 for PcXL5 versus Pcl10. N indicates the total number of nuclei analyzed in 3–5 embryos or tissues. Asterisks indicate that the pairwise difference between samples corresponding to the left column and the other samples is significant. See also Figure S1 and Table S1.
*
10
approximately 10 Mb of euchromatic sequences that contain more than 1,200 0 0 0 annotated genes and represent more wt PcXL5 Pcl10 Pcl10 Head Posterior wt PcXL5 than one-third of the euchromatic fraction (N=379) (N=226) (N=251) (N= 208) (N= 231) (N=319) (N=384) (N= 391) of 3R (Figure 1A). In this work, we provide evidence that the two distant Hox clusters can be corepressed by PcG proteins Mcp, both involved in the regulation of Abd-B, have been directly via association in three-dimensional nuclear space between implicated in chromosome kissing events when extra copies the Antennapedia (Antp) gene from the ANT-C and the Abdomwere introduced in the fly genome as transgenes. In both cases, inal-B (Abd-B) and Ultrabithorax (Ubx) genes from the BX-C. contacts strengthen PcG-dependent gene silencing and kissing We define this phenomenon as ‘‘Hox gene kissing.’’ Furtherevents occur specifically at PcG bodies (Bantignies et al., 2003; more, 3C on chip (4C) confirmed Hox contacts and revealed Grimaud et al., 2006; Muller et al., 1999; Vazquez et al., 2006), additional Abd-B partner loci. Importantly, Hox gene kissing is although they are clearly tissue-specific and do not occur in conserved in D. virilis, a species evolutionarily separated from several larval tissues containing polytene chromosomes (Fedor- D. melanogaster by around 60 Myr. Moreover, we demonstrate ova et al., 2008). These data raise the question of whether PcG that the BX-C element Fab-7 participates in Hox gene kissing, bodies might be involved in the functional compartmentalization and that removal of this element weakens silencing of distant genes in the ANT-C locus, indicating that PcG-dependent of the Hox clusters or other Polycomb target genes. In D. melanogaster, the ANT-C and BX-C complexes are chromatin contacts have a functional role in stabilizing gene located on the right arm of chromosome 3 (3R), separated by silencing. 5
5
*
5
Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc. 215
A Fold change Log2 (probe number)
Chr. 3R 15 10 5
20 10
0
5 ANT-C
10 E5-ems
pros grn
15
Fab-7 BX-C
hth
Mb
25
20 NK-C
Drop
hh-pnt
ss srp-pnr 4C score
10
50
500
5000
Log2 (probe number)
10
5
srp pnr
Fold change
1
BX-C region
B
Fab-7 ss
Ubx
abd-A
Abd-B
30 20 10
12
11.5
Log2 (probe number)
C
ANT-C
Fold change
Mb
13.5
NK-C
hh and pnt
10
5
tin bap C15
Dfd
lbl lbe slou
lab pb Scr Antp
hh cnc pnt
30 20 10
12.5
D
Log-ratio
13
12.5
17.5
17
13
Pc
19
18.5
H3K27me3
19.5
Mb
H3K4me3
4
4C hits
4C hits
4C hits
3
Random sites
Random sites
Random sites
2 1 0 -1 -100kb
-50kb
0
50kb
Distance to 4C hits (bp)
100kb
-100kb
-50kb
0
50kb
Distance to 4C hits (bp)
100kb
-100kb
-50kb
0
50kb
100kb
Distance to 4C hits (bp)
Figure 2. Extensive Interactions of the Fab-7 Element and Other PcG Target Genes along the 3R Chromosome Arm (A) Whole 3R chromosome domainogram representation of WT 4C profile (top) and Pc ChIP enrichments as fold change in WT 4–12 hr embryos (bottom, black). The x axis represents chromosome 3R coordinates in Mb and the y axis of the domainogram represents domain sizes as the Log2 of the number of contiguous probes involved in the calculation of statistical scores (see Figure S2H for genomic length conversion). Purple arrowheads indicate the Fab-7 bait and the BX-C. Red arrowheads indicate strong hits within 2 Mb of the anchor region, Black arrowheads indicate the strongest long-range hits, Grey arrowheads indicate other significant long-range target regions.
216 Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc.
RESULTS The Nuclear Organization of the Antennapedia and Bithorax Complexes We first analyzed the relative positioning of the two Hox complexes in the nucleus, starting with the most distal genes within each complex: Abd-B from the BX-C, and Antp from the ANT-C. We used two-color FISH in whole mount embryos and larval tissues, followed by three-dimensional image analysis (Bantignies et al., 2003). We compared embryonic and larval nuclei with different Hox gene expression profiles (CastelliGair, 1998; Kosman et al., 2004; Morata et al., 1994). We first focused on developmental stage 10–11, when PcG-mediated regulation of homeotic genes has already initiated (Pirrotta et al., 1995). 3D image acquisitions were carried out in ectodermal interphase diploid nuclei in three different regions along the anteroposterior axis of the embryo: the head, where both genes are repressed, a thoracic region including parasegments (PS) 4 and 5, where Antp is expressed and Abd-B repressed, and the posterior tip of the embryo (PS 13 and 14), where the reciprocal situation is observed, i.e., Abd-B is activated and Antp repressed. Abd-B and the Antp gene rarely colocalized in the thoracic PS and in the posterior PS, with colocalization rates of 4% and 7.6%, respectively. In contrast, the association between the loci was significantly stronger in the head where both loci are repressed (18%, Figure 1B,C). This was reflected in a global three-dimensional distance distribution skewed toward shorter distances in the head (Table S1, available online). Later during development, the association between Antp and Abd-B was reinforced in anterior larval tissues compared to more posterior tissues such as leg and wing discs, where only Antp is active (Figure S1A and Table S2). Antp also colocalized with the repressed Ubx gene in the head compartment of embryonic nuclei, and this interaction was lost in the trunk (PS4/5), where Antp is active. In the posterior PS, however, where both Ubx and Antp are repressed, they colocalized significantly more than in PS4/5 (Figures S1B and S1C and Table S2). This indicates that Hox gene kissing correlates with the repression status of both Hox genes. Hox Gene Kissing Occurs at PcG Bodies and Depends on the Function of PcG Proteins Next we combined FISH with immunostaining using a Polycomb antibody (FISH-I) to analyze the nuclear localization of Antp, Abd-B, and Ubx relative to PcG bodies. Repressed genes are associated with large PcG bodies in 80% to 85% of the cases. In embryo heads, where both Antp and Abd-B are colocalized, the two genes were found in the same PcG body (Figure 1D). In posterior segments, Abd-B is clearly outside PcG bodies in 80% of the cases (Figure 1D), although the two Hox complexes can still associate via kissing of Antp and Ubx within PcG bodies
(Figure S1C). In the thoracic PS4/5, in which the Antp gene is activated, Antp is outside PcG bodies in 85% of the cases, while Abd-B and Ubx are inside (data not shown), correlating with the lack of colocalization between Antp and the other genes. We then analyzed whether PcG proteins are required for Hox gene kissing. First, two non-PcG target genes, beat-Vc and CG17622, which are located on chromosome 3R and separated by 10 Mb like the Hox complexes, colocalize in less than 5% of the nuclei (Figure 1E and Table S1). Second, in the head of Polycomb and Polycomb-like mutant embryos, the Antp and Abd-B genes colocalize in about 5% of the nuclei, i.e., much less than in WT and close to the two control genes beat-Vc and CG17622 (Figure 1F and Table S1). This indicates that PcG proteins are required for pairing of the two loci. Even heterozygous mutations in the Pc and Pcl genes reduced the frequency of gene kissing in anterior larval tissues (Figure 1G and Table S1). These data show that Hox gene kissing occurs within PcG bodies in a PcG-dependent manner. The BX-C Interacts Preferentially with Polycomb Enriched Regions In order to analyze whether Hox loci only contact each other or whether they have other interacting partners in the nucleus, we developed a modified 4C protocol (see Experimental Procedures, Figure S2, and Figure S3). The Fab-7 element from the BX-C regulates the Abd-B expression and plays a role in long distance interactions inside the BX-C as well as at greater genomic distances (Bantignies et al., 2003; Lanzuolo et al., 2007). Therefore, we used Fab-7 as the 4C bait fragment in order to analyze its interaction with other loci along the 3R chromosome arm in larval brain and anterior larval discs. We found an extensive series of interactions along the 3R chromosome arm (Figure 2A). The strongest interactions occur within the BX-C. Other strong interactors are two PcG target regions, the ss and srp-pnr loci, located at a distance of 0.5 and 1 Mb, respectively (see Figure 2B). Significant interaction events are also observed at long distances. Four are particularly strong, corresponding to the NK-C, E5/ems, prospero and the ANT-C loci, and four others are weaker, corresponding to the grn, hth, pnt, and Drop loci (Figures 2A and 2C). Strikingly, all these major 4C hits are Polycomb bound regions (Schuettengruber et al., 2009). Moreover, a global analysis indicates that the 4C hits are highly enriched in Polycomb and its associated H3K27me3 repressive chromatin mark (Figure 2D). Thus, the 4C analysis confirmed Hox gene kissing and revealed additional interactions. In order to validate the 4C results, we verified one of the strongest long-range contacts by FISH. The lbl/lbe genes from the NK homeobox gene complex (NKC; Garcia-Fernandez, 2005) are located approximately 4.5 Mb distally from the BX-C (Figure 3A). They are organized in tandem and expressed in a few specialized cells in the embryonic parasegments and in the head (Jagla et al., 1997a, 1997b, 1998). The
(B) Close-up view of the 4C domainogram for the BX-C region. (C) Close-up views for the ANT-C, NK-C and hh-pnt regions. Gene annotations and Pc ChIP enrichments are represented under these profiles. (D) Distribution of Pc, H3K27me3 and H3K4me3 ChIP enrichment at non BX-C 4C hits (colored lines) and at randomly selected non-BX-C sites (gray lines). Dashed lines represent the median enrichment level, and solid lines indicate enrichment level corresponding to median ±25% of the population. See also Figure S2 and Figure S3.
Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc. 217
cen
A
beat-Vc
ANT-C
Figure 3. Specific Association between Abd-B and Other Target Loci
tel
NK-C hh
BX-C
Chr. 3 4.5 Mb
10 Mb
6.2 Mb
Percentage of pairing
B
C
Abd-B - lbl/lbe
25 20 15
*
10 5 0
D
25 20
20
15
15
10
10
5
5
0 Head 1 (N=253)
E
PS4/5 2 (N=307)
Percentage of pairing
PS4/5 2 (N=300)
F
* Head 1 (N=269)
Posterior 3 (N=255)
25
20
20
15
15
10
10
5
5
0
PS4/5 2 (N=260)
Abd-B - ph
0 Head 1 (N=210)
Dapi
PS4/5 2 (N=210)
Posterior 3 (N=210)
Abd-B
Head 1 (N=210)
lbl/lbe
PS4/5 2 (N=210)
Polycomb
Posterior 3 (N=210)
Merge
PS13 3 (N=296)
Head
G
Antp - lbl/lbe
0 Head 1 (N=307)
PS13 3 (N=374)
Abd-B - hh
25
25
(A) Schematic drawing illustrating the chromosomal positions of the ANT-C, BX-C, NK-C, and hh loci on Chromosome 3R. (B–F) Percentage colocalization between Abd-B and lbl/lbe (B), Abd-B and beat-Vc (C), Antp and lbl/lbe (D), Abd-B and hh (E), Abd-B and ph, located on Chromosome X (F). The p values of the pairwise comparison are 0.847 for Head versus PS4/5, 7.081e-03 for Head versus PS13, 2.457e03 for PS4/5 versus PS13 (B), 0.552 for Head versus PS4/5, 0.917 for Head versus Posterior, 0.504 for PS4/5 versus Posterior (C), 0.045 for Head versus PS4/5, 0.208 for Head versus PS13, 1.411e-03 for PS4/5 versus PS13 (D), 0.189 for Head versus PS4/5, 0.481 for Head versus Posterior, 0.536 for PS4/5 versus Posterior (E), 1.0 for Head versus PS4/5, 0.736 for Head versus Posterior, 0.736 for PS4/5 versus Posterior (F). N indicates the total number of nuclei analyzed in 3–4 embryos. Asterisks indicate that the pairwise difference between samples corresponding to the left column and the other samples is significant. (G and H) FISH-I in the embryonic heads of Abd-B, lbl/lbe and PC (G), Abd-B, hh and PC ([H], top), Abd-B, ph and PC ([H], bottom). Figure images correspond to individual nuclei of deconvolved single slices from 3D stacks. The scale bars represent 1 mm. See also Figure S4 and Table S3.
Head
Head
to the lbl/lbe locus. In all tissues, the frequencies of colocalization were similar Dapi Abd-B hh Polycomb Merge to those observed for Abd-B and lbl/lbe in H a divergently expressing tissue, while they were lower than in tissues corepressing Abd-B and lbl/lbe (Figure 3C and Table S3). This suggests that PcGmediated silencing specifically induces gene kissing. Gene contacts between the more remote Antp and lbl/lbe loci, which are separated by approximately 14.5 Mb, were less frequent. Nonetheless, the Abd-B ph frequency of colocalization was significantly higher in tissues where both loci are silenced (Figure 3D and Table S3). BX-C and NK-C loci colocalized in approximately 20% of the These results confirm the observation that gene contacts nuclei in the head and in PS4/5 where both genes are repressed between PcG target genes are more frequent when the genes in most of the cells. This interaction decreased significantly in are corepressed. Accordingly, as in the case of the Hox gene PS13 where Abd-B is active and lbl mostly repressed (Figure 3B kissing, the contact between BX-C and NK-C occurred excluand Table S3). ‘‘Me´nage a` trois’’ type of contacts among Antp- sively in a PcG body (Figure 3G). Abd-B-lbl are rarely observed (approximately 1% of the nuclei In order to understand whether all PcG target genes associate in embryonic heads, Figure S4), indicating that gene contacts at PcG bodies in a random manner or whether PcG-mediated are not all occurring simultaneously. To appreciate whether the gene kissing involves a subset of all target genes, we further colocalization frequency of divergently expressed PcG target studied by 3D-FISH the Hox gene Abd-B and the PcG targets genes may reflect background behavior, we then analyzed by polyhomeotic (ph), which is located on chromosome X, and FISH the distance between Abd-B and beat-Vc, a non-PcG hedgehog (hh), which is located 6.2 Mb distal to Abd-B on chrotarget gene that lies 4.2 Mb away from Abd-B, a distance similar mosome 3R (Figure 3A). The colocalization frequency was low in 218 Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc.
Figure 4. Conservation of Hox Gene Kissing in D. virilis
A BX-C 26D
cen Chr. 2 D. virilis
B
ANT-C 24E > 6 Mb
C Antp - Abd-B
Dapi
Abd-B
Antp
Head
35 30
Merge
25 20
*
15 10 5 0 Head 1 (N= 192 )
Posterior
Percentage of pairing
40
(A) Schematic drawing illustrating the chromosomal position of ANT-C and BX-C and their relative spacing on D. virilis Chromosome 2. (B) Percentage colocalization between Antp and Abd-B in stage 11-12 D. virilis embryos. N indicates the total number of nuclei analyzed in three different embryos. (C) Examples of Antp and Abd-B nuclear positions in D. virilis. The scale bar represents 1 mm. See also Figure S5.
Posterior 2 (N= 211)
P = 2.583e-07
all embryonic tissues examined (Figures 3E and 3F and Table S3). In agreement with FISH analysis, no significant interaction between hh and Fab-7 was detected by 4C analysis (Figure 2C). Of note, hh and ph were generally localized in smaller PC bodies than those containing the Hox genes (Figure 3H), indicating that different classes of PcG bodies exist in the nucleus. Together, these data demonstrate the existence of specific long-range associations among Hox genes and other PcG target genes, which occur within PcG nuclear bodies and correlate with their transcriptional status along the anteroposterior body axis. Hox Gene Kissing Is Evolutionarily Conserved in Drosophila Species We reasoned that, if Hox gene kissing is functionally significant, it might be conserved through evolution. Therefore, we analyzed D.virilis, a species separated from the D. melanogaster lineage 40 to 60 Myr ago. In D. virilis, the two Hox clusters are split between Ubx and abd-A (Von Allmen et al., 1996) instead of between Antp and Ubx. They are located on chromosome 2 and are contained within two large sequence scaffolds (13047 and 12855, see http://flybase.org/cgi-bin/gbrowse/dvir), separated by more than 6 Mb (Figure 4A). A large-scale comparison of the chromosomal organization around the Hox genes in D. melanogaster and D. virilis shows that syntenic regions of limited sizes are conserved between the two species, but the global linear organization of the chromosome is highly divergent (Figure S5A). The D. melanogaster counterparts of the D. virilis genes map to independent regions that are scattered along the entire arm of chromosome 3R. In embryo head, where both genes are repressed, Antp colocalized with Abd-B in 31% of the nuclei. In posterior segments, where Abd-B is active and Antp is repressed, the colocalization frequency was strongly reduced (Figures 4B, 4C, and Figure S5B). We conclude that, despite a radical change in linear chromosome organization, silencing-dependent Hox gene kissing is evolutionarily conserved. This suggests that this phenomenon reflects specific
molecular interactions rather than a three-dimensional folding coincidence of the D. melanogaster 3R chromome arm. Hox Gene Contacts Depend on Regulatory Elements in the BX-C In transgenic systems, Fab-7 and Mcp have been shown to be involved in intraas well as interchromosomal interactions in diploid tissues (Bantignies et al., 2003, Vazquez et al., 2006). We thus analyzed Hox gene kissing in Drosophila lines in which Fab-7 is deleted. The Fab-712 line corresponds to homozygous deletion of the Fab-7 region containing the PRE and the chromatin boundary portion (Mihaly et al., 1997). We performed 4C in the Fab-712 line and compared it to the WT profile. While most of the profile is very similar (Figure 5A), some distinct differences can be seen (Figure 5B and Figure S3). In particular, the interaction with the ANT-C is significantly reduced, while the interaction with the NK-C is increased in the Fab-712 line (Figure 5C). We confirmed these data by FISH experiments in WT versus Fab-712 eyeantennal discs, showing that the colocalization frequencies are lower for Antp-Abd-B and higher for Abd-B -lbl/lbe (Figures 5D and 5E). We confirmed these effects by analyzing three more deletion lines. The Fab-71 and the Mcp1 lines correspond to homozygous deletion of Fab-7 and Mcp, respectively, and the McpH27Fab-71 stock corresponds to the homozygous deletion of both elements (Karch et al., 1994). In each of these lines, reduced levels of colocalization were observed between Antp and Abd-B (Figure S6). We therefore conclude that the colocalization between the two Hox genes involves specific sequences within the Abd-B locus, even though these sequences do not share extensive homology with sequences from Antp. The finding that the deletion of Fab-7 and Mcp reduces but does not abolish Hox gene kissing suggests that multiple DNA regions of the BX-C contribute to this phenomenon. This notion is corroborated by the colocalization between Ubx and Antp in posterior embryonic tissues where the Fab-7/ Mcp region is located outside PcG bodies (Figures S1B and S1C). Perturbation of Hox Gene Kissing Affects PcG-Dependent Silencing of the ANT-C Genes What is the functional significance of the colocalization of silenced genes of the BX-C and the ANT-C? We have previously shown that long range interaction of homologous PRE/boundary Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc. 219
Figure 5. Hox Gene Kissing Is Partially Reduced upon Deletion of the Fab-7 Element
A wt
Log2 (probe number)
15 10 5
Fab-712 15 10 5
0
ANT-C
5
10
BX-C
15
25
20
NK-C
Score
1
10
50
500
Mb
5000
B wt
Log2 (probe number)
15 10 5
Fab-712
15 10 5
0
ANT-C
5
10
C
* Percentage of pairing
30 20
*
20
E
30
*
25 20 15 10 5 0
0
P=
NK-C
35
40
10
15
Antp - Abd-B in eye-antennal discs
D
50
Maximal 4C score
BX-C
ANT-C
NK-C
hh hh
0.029
4.0e-08
0.987
wt
Fab-712
wt 1 (N=1071)
12 Fab7 2 (N=1170)
P = 3.306e-03
elements reinforces PcG-mediated silencing (Bantignies et al., 2003). Therefore, we hypothesized that kissing between Hox genes might stabilize PcG-dependent gene silencing. In this case, we predicted that the perturbation of gene kissing in the Fab-712 line would weaken silencing of ANT-C genes, while it should increase silencing at the NK-C locus. We measured, by reverse transcription followed by quantitative PCR (RT-qPCR) in eye-antennal discs, the level of Abd-B, from the BX-C, Antp, Scr, Dfd, and pb from the ANT-C, lbe from the NK-C and hh as a negative control of a gene where the frequency of contacts was not perturbed (Figures 6A and 6B). Abd-B was derepressed in the mutant, indicating that Fab-7 is involved in its repression in anterior tissues, in addition to the abdominal segments of the body plan (Galloni et al., 1993; Gyurkovics et al., 1990). Strik220 Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc.
25
(A) Whole 3R chromosome domainogram representation of wild-type and Fab-712 4C profiles (the 4C bait is located immediately downstream to the distal border of the deletion, allowing to directly compare WT with Fab-712). ANT-C, BX-C, and NKC regions are indicated by arrowheads. (B) Binarized domainogram representation indicating most statistically significant hits (black) outside the BX-C region. BX-C probes (gray) were excluded for the calculation. (C) Maximal scores for WT and Fab-712 4C interactions within ANT-C, NK-C and hh regions. P values correspond to the statistical difference between WT and Fab-712. (D and E) FISH in third instar larval eye-antennal discs comparing the percentage of colocalization between Antp and Abd-B (D), Abd-B and lbl/lbe (E) in WT versus Fab-712. N indicates the total number of nuclei analyzed in four antennal regions and four eye regions of the eye-antennal discs. Approximately the same numbers of nuclei were counted in both regions and the data were pooled for homogeneity and statistical analysis. P values correspond to the difference between the two populations. See also Figure S3 and Figure S6.
Mb
ingly, Antp, Scr, Dfd and pb were also derepressed, showing that the partial decrease in long-distance contacts 35 * induced by the deletion of Fab-7 is suffi30 cient to induce a decrease of silencing 25 at the ANT-C. In contrast, the lbe gene, 20 which contacts the BX-C with increased 15 frequency, is significantly more repressed 10 upon deletion of Fab-7. Finally, no effects 5 were seen at the control hh gene. 0 12 Can the perturbation of gene kissing wt Fab7 1 2 (N=1037) (N=1041) induce phenotypic changes in flies? Homeotic antenna to leg (A > L) transforP = 6.510e-03 mation is not observed in flies carrying Fab-7 or Mcp deletions. However, we reasoned that the transcriptional effect may be too subtle to induce a phenotype, and that phenotypic effects might be easier to detect in a sensitized genetic background. We thus analyzed the Antp Nasobemia (AntpNs) mutation (Talbert and Garber, 1994), in which the Antp P2 promoter is duplicated, inducing ectopic expression of the Antp protein in the antennal territory of larval imaginal discs and an incomplete antenna-to-leg (A > L) transformation (Figure 7A) that is sensitive to trxG and PcG functions (Vazquez et al., 1999 and data not shown). This increased expression is much lower than the WT expression of Antp in the wing disc, and does not correlate with altered Hox chromosome kissing (Figures S7A and S7B), suggesting that, on its own, it depends on cis-regulatory changes of the Antp locus rather than on perturbation of chromosome architecture. Abd-B - lbl/lbe in eye-antennal discs
A
cen
ANT-C
BX-C
pb Dfd Scr Antp
Abd-B
Figure 6. Transcriptional Effects at Long Distance upon Deletion of the Fab-7 Element
NK-C hh
Chr. 3 Genes tested:
lbe
hh
B
Normalized expression
Abd-B 1.5
2
1.0
Dfd 200
50 40
150
30 100 20 0.5
1
50
10 0
0
0.0 wt 1
Fab-7 2 12
wt 1
P = 0.009
0
Fab-7 2 12
Fab-7 2 12
wt 1
P = 0.077
pb
P = 2.2e-06
lbe
3
Normalized expression
Scr
Antp
3
hh
20
200
15
150
10
100
5
50
2
1
0
0 1 wt
2 12 Fab-7
P = 0.121
0 1 wt
2 12 Fab-7
P = 0.004
(A) Schematic drawing illustrating the chromosomal positions of ANT-C, BX-C, NK-C, and the hh loci on chromosome 3R, with the genes from each genomic region analyzed by RT-qPCR. (B) RT-qPCR analysis in WT and Fab-712 lines. The results are the mean of 22 independent experiments. The statistical significances of differences between the two lines were examined by paired student tests and the p value is indicated under each graph. Note that, for genes of the ANT-C, all genes except pb have a p value below 0.1. The hh negative control is not significantly affected by the Fab-7 deletion.
1 wt
2 12 Fab-7
P = 0.644
We generated different lines in which the AntpNs chromosome was recombined with the Fab-712, Fab-71, Mcp1, and the McpH27Fab-71 chromosomes. As controls, the AntpNs chromosome was recombined with a WT chromosome from a w1118 line and with a mutant chromosome containing an ebony (e1) recessive marker. Three AntpNs, seven AntpNs, Fab-712, seven AntpNs, Fab-71, four AntpNs, Mcp1, three AntpNs, McpH27Fab71 and four AntpNs, e1 recombinant lines were obtained and raised as balanced stocks at 21 C. No obvious differences were observed between these lines right after their establishment. However, clear differences emerged in subsequent generations, reaching a plateau after 3 to 5 generations: the Ns A > L phenotype was accentuated in the BX-C deletion lines (Figure 7A and Figures S7C and S7D). Transgenerational inheritance has been previously linked to PcG and trxG proteins and to the 3D organization of BX-C elements like Fab-7 (Bantignies et al., 2003; Sollars et al., 2003). Therefore, deletions in the BX-C might enhance the A > L transformations of AntpNs via loss of gene contacts and progressive decrease in chromatin silencing efficiency through subsequent generations.
In addition to A > L transformations, we also observed the emergence of outgrowth P = 0.015 phenotypes in the eye of AntpNs combined with Fab-712, Fab-71 or McpH27Fab-71 deletions (Figure S7F). The eye phenotypes are variable, ranging from holes in the eye to large outgrowths, often emerging from ommatidia. Electron microscopy (EM) analysis of these outgrowths indicates that they resemble proximal leg or thorax-like structures, which may indicate a derepression of Antp in the eye imaginal discs. Eye phenotypes were observed in approximately 10% heterozygous AntpNs, McpH27Fab-71 and in 2% to 10% AntpNs, Fab-712 and AntpNs, Fab-71 adult flies for each line, but were totally absent in AntpNs, Mcp1 and in the recombinant control lines. Moreover, the eye outgrowths are strongly enhanced in the rare homozygous escapers, with frequencies of 60% to 80% in AntpNs, McpH27Fab-71, and 25% to 60% in AntpNs, Fab-712 and AntpNs, Fab-71, whereas they are observed in less than 10% in the other lines. A reduction in eye size and fused ommatidia were also observed in the AntpNs lines combined with BX-C deletions (Figure S7F and data not shown). Similar results were observed upon production of recombinant lines using an AntpNs stock, obtained from the Bloomington Drosophila Center. Again, these phenotypes did not appear in the two recombinant control lines obtained. Since all these phenotypes are consistent with Antp derepression, we tested Antp expression by RT-qPCR on eye-antennal imaginal discs. Significant Antp derepression was observed in the AntpNs, Fab-712 (Figure 7B), AntpNs, Fab-71 and AntpNs, McpH27Fab-71 lines (Figure S7E) compared to the AntpNs line. In contrast, the levels of hh did not change significantly in the different lines. Finally, we tested whether Fab-7 deletion may affect phenotypes of other Hox gene mutations. We thus analyzed the heterozygous Scr4 Scrw allele combination. This chromosome carries a null mutation (Scr4) combined with a second mutation (Scrw) wt
Fab-712
Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc. 221
A
B Normalized expression
Antp 200
150
150
100
100
50
50
0
0 AntpNs
AntpNs, Fab-712
AntpNs
AntpNs, Fab-712
P = 0.646
P = 0.003
D
Scr4/+ vs Scr4Scrw/+ Percentage of T1 legs
Figure 7. Deletion of the Fab-7 Element Weakens Silencing of ANT-C Genes, Resulting in the Exacerbation of Homeotic Phenotypes in Sensitized Genetic Backgrounds (A) Electron microscopy images showing examples of the antenna-to-leg (A > L) transformations (white arrows) in AntpNs and in AntpNs, Fab-712 heterozygous flies. (B) RT-qPCR analysis of Antp and hh. Error bars represent the standard deviation of three independent experiments. (C and D) Distribution of the number of sex comb teeth per T1 legs in Scr4/+ versus Scr4ScrW/+ (C) and in Scr4ScrW/+ versus Scr4ScrW/Fab-712 flies (D). The mean value and the N number of legs analyzed in two independent experiments are indicated. The statistical significances of differences between the distributions were examined by chi-square tests and the p value is indicated under each graph. (E) Representative leg pictures for each genotype. See also Figure S7.
hh
200
C
AntpNs, Fab-712
AntpNs
wt
Scr4Scrw/+ vs Scr4Scrw/Fab-712
70
70
60
60
50
50
40
40
30
30
20
20
10
10
0
the sex comb teeth (Figures 7D and 7E), showing that Scr is partially derepressed upon loss of the Fab-7 element. Together, our data show that the removal of BX-C regulatory elements weakens silencing at the ANT-C, demonstrating that Hox gene kissing contributes to stabilize their corepression.
0 31
53 7 24 46 5 Number of sex comb teeth
68
346 3 42 1 35 Number of sex comb teeth
Scr4/+ (mean=6.2, N=178)
Scr4Scrw/+ (mean=4.5, N=190)
Scr4Scrw/+ (mean=4.5, N=190)
Scr4Scrw/Fab-712 (mean=4.9, N=236)
DISCUSSION
T1 legs
The Nature of Hox Gene Contacts 4C and FISH measurements reveal clear differences in the degree of interactions P = 3.0e-08 P = 2.2e-16 within the BX-C as opposed to those E Scr4/+ Scr4Scrw/+ Scr4Scrw/Fab-712 between BX-C and ANT-C genes. Earlier FISH experiments within the BX-C indicate contacts (as defined by FISH distances % 350 nm) between the Fab-7/Abd-B and the bxd/Ubx regions in the range of 85% (Lanzuolo et al., 2007). This is much higher than the 15 to 20% observed here between Fab-7/Abd-B and Antp in the same tissues, i.e., head of the embryo (Figure 1B). These FISH data are concordant with 4C, where the in the upstream regulatory region (Southworth and Kennison, signals within the BX-C are far stronger than those between 2002). This second mutation leads to Polycomb-dependent Fab-7 and the ANT-C. Therefore, the higher-order interactions repression of the Scr gene in the first (T1) leg (Pattatucci et al., between elements of the BX-C are tighter than those between 1991 and data not shown). Whereas the Scr4/+ males show an heterologous loci such as the BX-C and the ANT-C. We propose average of 6.2 sex comb teeth in the T1 leg, the additional pres- that this reflects a hierarchy of higher-order structures in the ence of Scrw causes a reduction to an average of 4.5, showing nucleus, where chromatin elements are much more likely to that the Scrw allele represses the WT Scr copy in trans (Figures interact with neighboring partners in cis before engaging in 7C and 7E). We then tested whether the deletion of Fab-7 may interactions with remote partners. A 3D loop-structure model attenuate this repressive effect. Crossing the Scr4 Scrw chromo- has been proposed for the BX-C (Lanzuolo et al., 2007). These some with the Fab-712 mutation induced a significant increase in structures may be highly dynamic entities enabling contacts 222 Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc.
between epigenetic elements from distant loci when they come into three-dimensional proximity in the same nuclear compartment. The Discovery of a 3D Polycomb Interactome Gene Network Our 4C analysis clearly reveals that the gene kissing is not limited to Hox genes, since we found that the BX-C is engaged in additional contacts with other partners on the same chromosome arm. The fact that all the main interaction partners are PcG-bound chromatin domains, suggests that PcG proteins contribute to the establishment of long-range contacts among their target genes in the 3D nuclear space. However, most PcG target genes are cobound by many other chromatin factors, including insulator proteins like CTCF and Su(Hw) (Bushey et al., 2009; Negre et al., 2010), and these proteins may contribute to drive long-distance contacts via chromatin boundary or insulator elements present at PcG target loci. The 4C analysis measured chromosome contact frequencies from a large population of fixed cells, however it cannot distinguish whether these interactions are simultaneous. Two-color FISH experiments validated 4C contacts between BX-C and ANT-C and between BX-C and NK-C. Three-color FISH experiments, where all three complexes are silenced, revealed that the contacts are not simultaneous, at least during embryogenesis (Figure S4). We thus propose that each of the Hox loci is in dynamic contact with the other one, as well as with other PcG target loci. The existence of multiple gene contacts as we discovered in 4C and FISH can thus explain why the two Hox loci do not interact in all nuclei, but only in a significant minority of them. Interestingly, the deletion of Fab-7 reduced Antp-Abd-B contacts while it increased Abd-B-lbl contacts. This might suggest that, once free from one interacting partner, Abd-B chromatin might be available for increased interactions with some of the other partners. In the future, further 4C studies and high-throughput FISH analysis with probes directed against the interacting 4C loci and the intervening regions should extend these observations to gain a systematic understanding of Polycomb dependent gene kissing. In this context, a study which combined 3C and ChIP (called 6C) using antibodies against the human PcG protein EZH2 revealed intra- and inter-chromosomal interactions between PcG-bound elements (Tiwari et al., 2008). Although that study analyzed a small number of interactions, it raises the possibility that a hard-wired Polycomb target gene network may also exist in vertebrates. Evolutionary Conservation and Functional Significance of Hox Gene Kissing The evolutionary conservation of Hox gene kissing between D. melanogaster and D. virilis strongly suggests that selective pressure maintains the spatial proximity of Hox complexes. This fact reinforces the idea that gene kissing might stabilize PcG-mediated silencing of Hox genes, contributing to the specification of body structures along the anteroposterior axis. Fab-7 and Mcp might represent examples of bifunctional regulatory elements, as they appear to have two distinct roles: on the one hand, they regulate the expression of their flanking genes in cis and, on the other, they mediate long-distance regulatory
interactions with Hox genes in the ANT-C. We propose that one of the roles of Fab-7 and Mcp was to promote locus-wide chromatin condensation of the ancestral Hox gene cluster by interacting with more anterior genes. The maintenance of these interactions after the physical split of the cluster may have originated the long-distance pairing that is still observed today. The analysis of deletions of BX-C elements reveals important features of long-range regulation. Within the BX-C, we noticed that the removal of Fab-7 reduces PcG-dependent silencing of the abd-A and Ubx genes (2- to 3-fold derepression, data not shown). This result indicates that multiple regulatory interactions take place for the strong and faithful repression of the BX-C in anterior segments. In this system, we exclude secondary effects in trans of Abd-B on Ubx and abd-A because the products of posterior Hox genes always repress, not activate, more anterior genes (a phenomenon called posterior dominance, Kuziora, 1993; Struhl and White, 1985). Indeed, this feature might attenuate the transcriptional effects of loss of long-distance Hox contacts. A second level of higher-order chromatin association involves larger scales, such as between the BX-C and ANT-C loci. These contacts are probably not indispensable for repression. For instance, Antp is silenced in every anterior cell, even if it contacts the BX-C in a minority of the nuclei. Cis regulatory elements are probably able to maintain silencing to a sufficient degree. However, small but clear transcriptional derepression of multiple ANT-C genes was observed upon mutations in the BX-C. In this context, one should note that the BX-C contacts many other loci in addition to the ANT-C. Likewise, the ANT-C may contact several other PcG target regions. These other contacts could functionally complement a loss of pairing upon BX-C mutations. ANT-C may maintain spatial association with other regions within the BX-C or with other domains that participate in the same spatial network of PcG target genes, thus remaining in an appropriate regulatory environment. However, this compensation cannot be complete, as illustrated by the increase in homeotic phenotypes seen in the sensitized Antp and Scr backgrounds. To our knowledge, this is the first report that links the function of a chromatin element involved in 3D chromosome contacts to specific phenotypes. Hox Contacts and Chromosomal Rearrangements Three different splitting events of the ancestral Hox gene complex have been detected to date in the Drosophila genus. Two of them, mentioned in this study, are represented by the separation of Antp and Ubx in D. melanogaster and of Ubx and abd-A in D. virilis. A third split has been described in D. buzzatii (Negre et al., 2003). This species, a member of the D. repleta group, shows a split between Ubx and abd-A as with D. virilis (Ranz et al., 1997), and an additional split, whereby the most anterior gene, lab, has been relocated flanking the two posterior Hox genes abd-A and Abd-B. In light of our data, the most likely scenario that might explain this third split involves translocation of the entire lab locus to the border of the posterior cluster due to PcG-dependent spatial proximity between the anterior and posterior Hox clusters. Related to this point, in mammalian cells the proximity of chromosome or chromosomal loci has been suggested to induce chromosomal rearrangements between Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc. 223
them (Branco and Pombo, 2006; Lin et al., 2009; Nikiforova et al., 2000; Roix et al., 2003). Moreover, chromosome kissing events dependent on colocalization in transcription factories were also shown to be correlated with a high rate of translocation (Osborne et al., 2007). It is interesting to note that Hox gene clusters have been submitted to considerable rearrangements during evolution of the animal kingdom (Garcia-Fernandez, 2005). We propose that split Hox clusters might have contributed to evolution of chromosomes bearing them. In conclusion, the data described here show that the specific nuclear organization imposed by the PcG proteins in Drosophila diploid tissues influences the maintenance of epigenetic states and might contribute to genome evolution. EXPERIMENTAL PROCEDURES Fly Stocks and Handling Flies were raised in standard cornmeal yeast extract media at 25 C. The Oregon-R w1118 line was used as wild-type (WT) D. melanogaster. A w stock (#15010-1051.17, from the Tucson species stock center) was used as WT D. virilis. The Pcl10/ KrGFP-CyO stock was used for selection of homozygous Pcl10 mutants (Bantignies et al., 2003). A PcXL5/ KrGFP-TM3,Sb stock was used for the selection of homozygous PcXL5 mutants. The Fab-712, Fab-71, Mcp1, and McpH27Fab-71 deletion lines were described in (Karch et al., 1994; Mihaly et al., 1997). The AntpNs stock used in this study was provided by W. Gehring. AntpNs stock from Bloomington was also used, although this stock shows lower penetrance of the A > L phenotype. All the AntpNs recombinant lines were maintained over the KrGFP-TM3,Sb balancer (from stocks BL#5195 of the Bloomington Drosophila Stock Center). The ebony (e1), Scr4 and Scr4ScrW stocks were from Bloomington (BL#2558, BL#2188 and BL#809, respectively). Staged eggs were collected on agar plates with standard vinegar/fresh yeast medium. Antp and Scr mutant Flies were grown at 21 C, and sex comb teeth were counted under a Nikon SMZ1000 binocular at 803 magnification. RT-qPCR Third-instar larval imaginal eye-antennal discs were dissected in Schneider’s Drosophila Medium (GIBCO) and 30–40 discs were taken for RNA isolation using TRIzol reagent (Invitrogen). 300–400 ngs of total RNA were used for the RT reaction. RT was performed using the Superscript III First Strand Synthesis Kit from Invitrogen following the manufacturer’s instructions and using hexamer primers. cDNA quantifications were performed by real-time PCR, using a Roche Light Cycler and the Light Cycler FastStart DNA Master SYBR green I kit. NdeI digested genomic DNA served for the standard curve. Expression levels were normalized to Rp49 and multiplied by 1.104. Primer sequences are listed in Table S4. Two-Color 3D-FISH and FISH-I These procedures are described in the Extended Experimental Procedures. Microscopy and Image Analysis Microscopy and 3D image analysis were as previously described (Bantignies et al., 2003). Minor modifications and EM are described in the Extended Experimental Procedures. Chromosome Conformation Capture on Chip (4C) Flies were grown at 25 C. Third-instar larval brain and anterior discs from 200 larvae were dissected in Schneider’s Drosophila Medium and used for the 3C. The 3C was performed as previously described (Hagege et al., 2007; Miele and Dekker, 2009) with the main differences being the use of DpnII (New England Biolabs), a 4 bp cutter restriction enzyme, and a fixation in 3% para-formaldehyde for 30 min, maximizing sensitivity and resolution of contact detection. The 4C method includes an ‘‘anchor biotinylated primer extension’’ procedure that
224 Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc.
is described in detail in the Extended Experimental Procedures. Microarray analyses are described in the same section and Figure S2 and Figure S3. SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, seven figures, and six tables and can be found with this article online at doi:10.1016/j.cell.2010.12.026. ACKNOWLEDGMENTS We would like to thank W. Gehring, F. Karch, H. Gyurkovics, J.M. Dura, and Alain Pelisson for fly lines and plasmids. We also thank T. Sexton and D. Cribbs for stimulating discussions; B. Pfeiffer and S. Celniker for sharing the D. virilis BX-C sequence before publication; Chantal Cazevieille for electron microscopy at the Centre de Ressources en Imagerie Cellulaire of Montpellier. F.B. is supported by the CNRS; V.R. and I.C. were supported by the Ministe`re de l’Enseignement Supe´rieur, the Association pour la Recherche sur le Cancer, and the Agence Nationale de la Recherche. B.L. was supported by the Ministe`re de l’Enseignement Supe´rieur. B.S. was supported by the Fondation de la Recherche Me´dicale. G.C. was supported by grants of the CNRS, the Human Frontier Science Program Organization, the European Union FP6 (Network of Excellence, The Epigenome, and STREP 3D Genome), the European Research Council (ERC-2008-AdG No 232947), by the Indo-French Centre for Promotion of Advanced Research, by the Agence Nationale de la Recherche, and by the Ministe`re de l’Enseignement Supe´rieur, ACI BCMS. Received: June 22, 2009 Revised: September 22, 2010 Accepted: December 17, 2010 Published: January 20, 2011 REFERENCES Alkema, M.J., Bronk, M., Verhoeven, E., Otte, A., van ’t Veer, L.J., Berns, A., and van Lohuizen, M. (1997). Identification of Bmi1-interacting proteins as constituents of a multimeric mammalian polycomb complex. Genes Dev. 11, 226–240. Bantignies, F., Grimaud, C., Lavrov, S., Gabut, M., and Cavalli, G. (2003). Inheritance of Polycomb-dependent chromosomal interactions in Drosophila. Genes Dev. 17, 2406–2420. Branco, M.R., and Pombo, A. (2006). Intermingling of chromosome territories in interphase suggests role in translocations and transcription-dependent associations. PLoS Biol. 4, e138. Buchenau, P., Hodgson, J., Strutt, H., and Arndt-Jovin, D.J. (1998). The distribution of polycomb-group proteins during cell division and development in Drosophila embryos: impact on models for silencing. J. Cell Biol. 141, 469–481. Bushey, A.M., Ramos, E., and Corces, V.G. (2009). Three subclasses of a Drosophila insulator show distinct and cell type-specific genomic distributions. Genes Dev. 23, 1338–1350. Castelli-Gair, J. (1998). Implications of the spatial and temporal regulation of Hox genes on development and evolution. Int. J. Dev. Biol. 42, 437–444. Cavalli, G. (2007). Chromosome kissing. Curr. Opin. Genet. Dev. 17, 443–450. Duncan, I. (1987). The bithorax complex. Annu. Rev. Genet. 21, 285–319. Fedorova, E., Sadoni, N., Dahlsveen, I.K., Koch, J., Kremmer, E., Eick, D., Paro, R., and Zink, D. (2008). The nuclear organization of Polycomb/Trithorax group response elements in larval tissues of Drosophila melanogaster. Chromosome Res. 16, 649–673. Fraser, P., and Bickmore, W. (2007). Nuclear organization of the genome and the potential for gene regulation. Nature 447, 413–417. Galloni, M., Gyurkovics, H., Schedl, P., and Karch, F. (1993). The bluetail transposon: evidence for independent cis-regulatory domains and domain boundaries in the bithorax complex. EMBO J. 12, 1087–1097.
Garcia-Fernandez, J. (2005). The genesis and evolution of homeobox gene clusters. Nat. Rev. Genet. 6, 881–892. Grimaud, C., Bantignies, F., Pal-Bhadra, M., Ghana, P., Bhadra, U., and Cavalli, G. (2006). RNAi Components Are Required for Nuclear Clustering of Polycomb Group Response Elements. Cell 124, 957–971. Gyurkovics, H., Gausz, J., Kummer, J., and Karch, F. (1990). A new homeotic mutation in the Drosophila bithorax complex removes a boundary separating two domains of regulation. EMBO J. 9, 2579–2585. Hagege, H., Klous, P., Braem, C., Splinter, E., Dekker, J., Cathala, G., de Laat, W., and Forne, T. (2007). Quantitative analysis of chromosome conformation capture assays (3C-qPCR). Nat. Protoc. 2, 1722–1733. Jagla, K., Frasch, M., Jagla, T., Dretzen, G., Bellard, F., and Bellard, M. (1997a). ladybird, a new component of the cardiogenic pathway in Drosophila required for diversification of heart precursors. Development 124, 3471– 3479. Jagla, K., Jagla, T., Heitzler, P., Dretzen, G., Bellard, F., and Bellard, M. (1997b). ladybird, a tandem of homeobox genes that maintain late wingless expression in terminal and dorsal epidermis of the Drosophila embryo. Development 124, 91–100. Jagla, T., Bellard, F., Lutz, Y., Dretzen, G., Bellard, M., and Jagla, K. (1998). ladybird determines cell fate decisions during diversification of Drosophila somatic muscles. Development 125, 3699–3708. Karch, F., Galloni, M., Sipos, L., Gausz, J., Gyurkovics, H., and Schedl, P. (1994). Mcp and Fab-7: molecular analysis of putative boundaries of cis-regulatory domains in the bithorax complex of Drosophila melanogaster. Nucleic Acids Res. 22, 3138–3146. Kaufman, T.C., Seeger, M.A., and Olsen, G. (1990). Molecular and genetic organization of the antennapedia gene complex of Drosophila melanogaster. Adv. Genet. 27, 309–362. Kosman, D., Mizutani, C.M., Lemons, D., Cox, W.G., McGinnis, W., and Bier, E. (2004). Multiplex detection of RNA expression in Drosophila embryos. Science 305, 846. Kuziora, M.A. (1993). Abdominal-B protein isoforms exhibit distinct cuticular transformations and regulatory activities when ectopically expressed in Drosophila embryos. Mech. Dev. 42, 125–137. Lanzuolo, C., Roure, V., Dekker, J., Bantignies, F., and Orlando, V. (2007). Polycomb response elements mediate the formation of chromosome higherorder structures in the bithorax complex. Nat. Cell Biol. 9, 1167–1174.
Muller, M., Hagstrom, K., Gyurkovics, H., Pirrotta, V., and Schedl, P. (1999). The mcp element from the Drosophila melanogaster bithorax complex mediates long-distance regulatory interactions. Genetics 153, 1333–1356. Negre, B., Ranz, J.M., Casals, F., Caceres, M., and Ruiz, A. (2003). A new split of the Hox gene complex in Drosophila: relocation and evolution of the gene labial. Mol. Biol. Evol. 20, 2042–2054. Negre, N., Brown, C.D., Shah, P.K., Kheradpour, P., Morrison, C.A., Henikoff, J.G., Feng, X., Ahmad, K., Russell, S., White, R.A., et al. (2010). A comprehensive map of insulator elements for the Drosophila genome. PLoS Genet. 6, e1000814. Nikiforova, M.N., Stringer, J.R., Blough, R., Medvedovic, M., Fagin, J.A., and Nikiforov, Y.E. (2000). Proximity of chromosomal loci that participate in radiation-induced rearrangements in human cells. Science 290, 138–141. Okamoto, I., and Heard, E. (2009). Lessons from comparative analysis of X-chromosome inactivation in mammals. Chromosome Res. 17, 659–669. Osborne, C.S., Chakalova, L., Mitchell, J.A., Horton, A., Wood, A.L., Bolland, D.J., Corcoran, A.E., and Fraser, P. (2007). Myc dynamically and preferentially relocates to a transcription factory occupied by Igh. PLoS Biol. 5, e192. Pattatucci, A.M., Otteson, D.C., and Kaufman, T.C. (1991). A functional and structural analysis of the Sex combs reduced locus of Drosophila melanogaster. Genetics 129, 423–441. Pirrotta, V., Chan, C.S., McCabe, D., and Qian, S. (1995). Distinct parasegmental and imaginal enhancers and the establishment of the expression pattern of the Ubx gene. Genetics 141, 1439–1450. Ranz, J.M., Segarra, C., and Ruiz, A. (1997). Chromosomal homology and molecular organization of Muller’s elements D and E in the Drosophila repleta species group. Genetics 145, 281–295. Ren, X., Vincenz, C., and Kerppola, T.K. (2008). Changes in the distributions and dynamics of polycomb repressive complexes during embryonic stem cell differentiation. Mol. Cell. Biol. 28, 2884–2895. Roix, J.J., McQueen, P.G., Munson, P.J., Parada, L.A., and Misteli, T. (2003). Spatial proximity of translocation-prone gene loci in human lymphomas. Nat. Genet. 34, 287–291. Saurin, A.J., Shiels, C., Williamson, J., Satijn, D.P., Otte, A.P., Sheer, D., and Freemont, P.S. (1998). The human polycomb group complex associates with pericentromeric heterochromatin to form a novel nuclear domain. J. Cell Biol. 142, 887–898.
Lewis, E.B., Pfeiffer, B.D., Mathog, D.R., and Celniker, S.E. (2003). Evolution of the homeobox complex in the Diptera. Curr. Biol. 13, R587–R588.
Schuettengruber, B., Chourrout, D., Vervoort, M., Leblanc, B., and Cavalli, G. (2007). Genome Regulation by Polycomb and Trithorax Proteins. Cell 128, 735–745.
Lin, C., Yang, L., Tanasa, B., Hutt, K., Ju, B.G., Ohgi, K., Zhang, J., Rose, D.W., Fu, X.D., Glass, C.K., et al. (2009). Nuclear receptor-induced chromosomal proximity and DNA breaks underlie specific translocations in cancer. Cell 139, 1069–1083.
Schuettengruber, B., Ganapathi, M., Leblanc, B., Portoso, M., Jaschek, R., Tolhuis, B., van Lohuizen, M., Tanay, A., and Cavalli, G. (2009). Functional Anatomy of Polycomb and Trithorax Chromatin Landscapes in Drosophila Embryos. PLoS Biol. 7, e13.
Merdes, G., and Paro, R. (2009). About combs, notches, and tumors: epigenetics meets signaling. Dev. Cell 17, 440–442.
Sexton, T., Bantignies, F., and Cavalli, G. (2009). Genomic interactions: chromatin loops and gene meeting points in transcriptional regulation. Semin. Cell Dev. Biol. 20, 849–855.
Messmer, S., Franke, A., and Paro, R. (1992). Analysis of the functional role of the Polycomb chromo domain in Drosophila melanogaster. Genes Dev. 6, 1241–1254. Miele, A., and Dekker, J. (2009). Mapping Cis- and Trans- Chromatin Interaction Networks Using Chromosome Conformation Capture (3C). Methods Mol. Biol. 464, 105–121. Mihaly, J., Hogga, I., Gausz, J., Gyurkovics, H., and Karch, F. (1997). In situ dissection of the Fab-7 region of the bithorax complex into a chromatin domain boundary and a Polycomb-response element. Development 124, 1809–1820. Morata, P., Jimenez-Mesa, J., Nunez de Castro, I., and Sanchez-Jimenez, F.M. (1994). Vinca alkaloids enhance the half-life of tumour ornithine decarboxylase. Cancer Lett. 81, 209–213. Muller, J., and Verrijzer, P. (2009). Biochemical mechanisms of gene regulation by polycomb group protein complexes. Curr. Opin. Genet. Dev. 19, 150–158.
Sexton, T., Umlauf, D., Kurukuti, S., and Fraser, P. (2007). The role of transcription factories in large-scale structure and dynamics of interphase chromatin. Semin. Cell Dev. Biol. 18, 691–697. Sollars, V., Lu, X., Xiao, L., Wang, X., Garfinkel, M.D., and Ruden, D.M. (2003). Evidence for an epigenetic mechanism by which Hsp90 acts as a capacitor for morphological evolution. Nat. Genet. 33, 70–74. Southworth, J.W., and Kennison, J.A. (2002). Transvection and silencing of the Scr homeotic gene of Drosophila melanogaster. Genetics 161, 733–746. Struhl, G., and White, R.A. (1985). Regulation of the Ultrabithorax gene of Drosophila by other bithorax complex genes. Cell 43, 507–519. Talbert, P.B., and Garber, R.L. (1994). The Drosophila homeotic mutation Nasobemia (AntpNs) and its revertants: an analysis of mutational reversion. Genetics 138, 709–720.
Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc. 225
Tiwari, V.K., Cope, L., McGarvey, K.M., Ohm, J.E., and Baylin, S.B. (2008). A novel 6C assay uncovers Polycomb-mediated higher order chromatin conformations. Genome Res. 18, 1171–1179. Vazquez, J., Muller, M., Pirrotta, V., and Sedat, J.W. (2006). The Mcp element mediates stable long-range chromosome-chromosome interactions in Drosophila. Mol. Biol. Cell 17, 2158–2165. Vazquez, M., Moore, L., and Kennison, J.A. (1999). The trithorax group gene osa encodes an ARID-domain protein that genetically interacts with the
226 Cell 144, 214–226, January 21, 2011 ª2011 Elsevier Inc.
brahma chromatin-remodeling factor to regulate transcription. Development 126, 733–742. Von Allmen, G., Hogga, I., Spierer, A., Karch, F., Bender, W., Gyurkovics, H., and Lewis, E. (1996). Splits in fruitfly Hox gene complexes. Nature 380, 116. Williams, A., Spilianakis, C.G., and Flavell, R.A. (2010). Interchromosomal association and gene regulation in trans. Trends Genet. 26, 188–197.
Regulation of Mitochondrial Protein Import by Cytosolic Kinases Oliver Schmidt,1,2,3,10 Angelika B. Harbauer,1,2,3,10 Sanjana Rao,1,3,4 Beate Eyrich,5 Rene´ P. Zahedi,5 Diana Stojanovski,1,6 Birgit Scho¨nfisch,1,2 Bernard Guiard,7 Albert Sickmann,5,8 Nikolaus Pfanner,1,2,* and Chris Meisinger1,2,9,* 1Institut
fu¨r Biochemie und Molekularbiologie, ZBMZ, Universita¨t Freiburg, 79104 Freiburg, Germany Centre for Biological Signalling Studies 3Fakulta ¨ t fu¨r Biologie 4Spemann Graduate School of Biology and Medicine Universita¨t Freiburg, 79104 Freiburg, Germany 5Leibniz-Institut fu ¨ r Analytische Wissenschaften-ISAS-e.V., 44139 Dortmund, Germany 6Department of Biochemistry, La Trobe University, 3086 Melbourne, Australia 7Centre de Ge ´ ne´tique Mole´culaire, CNRS, 91190 Gif-sur-Yvette, France 8Medizinisches Proteom-Center, Ruhr-Universita ¨ t Bochum, 44801 Bochum, Germany 9Department of Chemistry and Biochemistry, University of Bern, 3012 Bern, Switzerland 10These authors contributed equally to this work *Correspondence:
[email protected] (N.P.),
[email protected] (C.M.) DOI 10.1016/j.cell.2010.12.015 2BIOSS
SUMMARY
Mitochondria import a large number of nuclear-encoded proteins via membrane-bound transport machineries; however, little is known about regulation of the preprotein translocases. We report that the main protein entry gate of mitochondria, the translocase of the outer membrane (TOM complex), is phosphorylated by cytosolic kinases—in particular, casein kinase 2 (CK2) and protein kinase A (PKA). CK2 promotes biogenesis of the TOM complex by phosphorylation of two key components, the receptor Tom22 and the import protein Mim1, which in turn are required for import of further Tom proteins. Inactivation of CK2 decreases the levels of the TOM complex and thus mitochondrial protein import. PKA phosphorylates Tom70 under nonrespiring conditions, thereby inhibiting its receptor activity and the import of mitochondrial metabolite carriers. We conclude that cytosolic kinases exert stimulatory and inhibitory effects on biogenesis and function of the TOM complex and thus regulate protein import into mitochondria. INTRODUCTION Mitochondria play crucial roles in cellular energy conversion, numerous metabolic pathways, maintenance of ion concentrations, and regulation of apoptosis. Proteomic studies indicate that mitochondria contain 1000 (yeast) to 1500 (human) different proteins, 99% of which are being encoded by nuclear genes and synthesized as precursors on cytosolic ribosomes (Mootha et al., 2003; Neupert and Herrmann, 2007; Pagliarini et al., 2008; Chacinska et al., 2009). The central entry gate for
virtually all nuclear-encoded mitochondrial proteins is the preprotein translocase of the outer membrane (TOM complex). The receptors Tom20 and Tom70 initially recognize the precursor proteins: Tom20 preferentially preproteins with N-terminal presequences, and Tom70 hydrophobic precursors with internal targeting signals (Young et al., 2003; Dolezal et al., 2006; Neupert and Herrmann, 2007). Upon interaction with the initial receptors, the precursors are transferred to the central receptor Tom22 and from here to the import channel Tom40 (Neupert and Herrmann, 2007; Chacinska et al., 2009). Three small Tom proteins modulate the assembly and stability of the TOM complex. After passing through the TOM complex, the precursor proteins use different machineries to reach their functional destination in the four mitochondrial subcompartments: outer membrane, intermembrane space, inner membrane, and matrix (Dolezal et al., 2006; Neupert and Herrmann, 2007; Chacinska et al., 2009). Little is known about regulation of the translocases that mediate preprotein import into mitochondria. Reversible phosphorylation of proteins is a major means of regulation of cellular processes. Studies in recent years indicated that the number of mitochondrial phosphoproteins is larger than expected (Chi et al., 2007; Li et al., 2007; Reinders et al., 2007; Albuquerque et al., 2008; Gnad et al., 2009; Holt et al., 2009). The functional consequences of phosphorylation, however, have only been investigated for a limited set of mitochondrial proteins, including regulation of apoptotic processes, mitochondrial morphology, pyruvate dehydrogenase, and respiratory complexes; phosphorylation of a few preproteins was shown to modulate their interaction with Hsp70 chaperones and transfer into mitochondria (Desagher et al., 2001; Robin et al., 2003; Pagliarini and Dixon, 2006; de Rasmo et al., 2008; Soubannier and McBride, 2009). Large-scale studies of phosphoproteins in baker’s yeast indicated the existence of phosphorylation sites on the translocases of the mitochondrial membranes (Chi et al., 2007; Li et al., 2007; Albuquerque et al., 2008; Gnad et al., 2009; Holt et al., 2009); however, it has not been reported so far whether Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc. 227
C
Figure 1. Phosphorylation of TOM Receptors by CK2 and PKA
D
Tom20cd
[γ-33P]ATP Tom70cd
T T6 6A S1 74 T2 A 28 A
[γ-33P]ATP
W
CK 2 CK 1 PK A Ca M M KII AP CD K K CD 1-C K yc GS 2-C lin K3 ycl B in A
[γ-33P]ATP
W T S8 7A T9 2A S1 72 A
A
Tom22cd + CK2
+ PKA
+ CK1
+ CK1
+ PKA
+ GSK3
+ CamKII
+ CamKII
Tom20cd Tom70cd GSK3
Tom40 CK2α Tom6N CK2β 1
2
3
4
5
6
7
1
8
2
3
4
1
2
3
(A) Phosphorylation of purified TOM subunits with recombinant kinases, analyzed by SDS-PAGE and autoradiography. Kinase autophosphorylation bands are indicated (CK2a, CK2b, and GSK3). (B) Phosphorylation of Tom22cd phosphosite mutants. In samples 5–16, the applied kinase activity represented 50% for CK1 and 10% for CK2 of the one applied in (A) and samples 1–4. WT, wild-type. (C) In vitro phosphorylation of Tom20cd phosphosite mutants. (D) In vitro phosphorylation of Tom70cd phosphosite mutants. See also Table S1 and Figure S1.
4
B [γ-33P]ATP
min 0 60 120 0 60 120 Tom22pS46
+ PKA
Tom22pS46 (long exposure) Tom22
+ CamKII 5 1
2
3
6
7
8
9 10
4
any of these phosphosites are of functional relevance. Thus, despite our detailed knowledge about composition and function of the preprotein translocases, posttranslational regulatory mechanisms acting at the membrane translocases of mitochondria have not been identified. Here, we report that biogenesis and function of the TOM complex are regulated by protein kinases. Yeast mutant cells of casein kinase 2 (protein kinase CK2) show severe defects in the levels of the TOM complex. CK2 phosphorylates the receptor Tom22 and the mitochondrial import protein Mim1 that are critical for the biogenesis of further TOM subunits. Whereas CK2 plays a stimulatory role for mitochondrial biogenesis, protein kinase A (PKA) plays an inhibitory role. PKA phosphorylates the receptor Tom70 and thus impairs its activity for the import of metabolite carriers of the inner-mitochondrial membrane. Our study reveals that the main protein entry gate of mitochondria is not a static complex but is regulated by cytosolic kinases. RESULTS Phosphorylation of Tom Proteins by Cytosolic Kinases In Vitro To map phosphorylation sites on the TOM complex, we purified Saccharomyces cerevisiae mitochondria, outer-membrane vesi228 Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc.
A
46
cd S4 4,
22
To m
cd W T
22 To m
To m + CK1
22
A 46 cd S4 4,
cd W T
22
+ CK2
cles, and TOM complex (Figure S1A available online). Phosphopeptides were enriched and analyzed by tandem mass spectrometry (MS/MS). We identified 30 phosphosites in Tom proteins. Ten sites agreed with previously determined sites 0 60 120 0 60 120 (Chi et al., 2007; Li et al., 2007; Albuquerque et al., 2008; Gnad et al., 2009; Holt et al., 2009), and we identify 20 additional sites, leading to a total of 31 TOM phos11 12 13 14 15 16 phosites (Table S1 and Figure S1B). All Tom proteins were phosphorylated (including Tom71, a low-abundant isoform of Tom70). Most phosphosites were located on the cytosolic side of the outer membrane, particularly in the cytosolic domains of the receptors Tom20, Tom22, Tom70, and Tom71 (Figure S1B). Prediction of potential kinases for the TOM phosphosites yielded several cytosolic kinases—in particular, CK2 and PKA (Figure S1B). We performed an in vitro screen using recombinantly expressed TOM subunits, [g-33P]ATP, and the purified kinases CK2, CK1, PKA, calmodulin-dependent kinase II (CamKII), p42 mitogen-activated protein kinase (MAPK), cyclindependent kinase 1 (CDK1)-cyclin B complex, CDK2-cyclin A complex, and glycogen synthase kinase 3 (GSK3) (Figure 1A). Purified TOM subunits included the cytosolic domains of the receptors Tom22cd, Tom20cd, and Tom70cd; Tom40 renatured from inclusion bodies; and the cytosolic domain of Tom6 fused to glutathione S-transferase (GST). Tom22cd was phosphorylated by the acidophilic kinases CK1 and CK2 (Figure 1A and Figure S1C). Ser44 and Ser46 of Tom22 match the CK2 consensus sequence (Meggio and Pinna, 2003). Replacement of Ser44 and Ser46 by alanines strongly reduced phosphorylation of Tom22cd by CK2 (Figure 1B). We generated an antiserum that selectively recognized phospho-Ser46 of Tom22 (Figure S1D). Comparison of the phosphorylation by CK1 and CK2 using lower kinase concentrations revealed that CK2 functions as major kinase for Tom22 (Figure 1B, right). In addition, Tom22cd was weakly phosphorylated by PKA at Thr76 (Figure 1B). CK1
To m
CK2
W
Tom22cd
T S4 4, S4 46A 4, S4 46A 6A , T
76
A
Western blot
12 h 37°C
μg
ck2-ts
WT
24 h 37°C
ck2-ts
WT
24°C
ck2-ts
WT
10 20 40 10 20 40 10 20 40 10 20 40 10 20 40 10 20 40
μg
Tom22
12 h 37°C ck2-ts
WT
24 h 37°C
24 h 37°C
ck2-ts
WT
ck2-ts
10 20 40 10 20 40 10 20 40 10 20 40 10 20 40 10 20 40
OM45
Tom20
24°C
pa re ck nta 2- l t W s T
24°C WT
pa re ck nta 2- l t W s T
A
Pgk1 Cytosol
Tim23 Ssa1 Cox4
Tom40
Kar2 mtHsp70
Tom70
ER Sec61
F1β
Porin
37 38 39 10 11 12 13 14 15 16 17 18
TOM complex
440
440
ts T W
2-
2-
ck
T
ts
24 h 37°C
ck
Mito.
W
kDa
kDa
V2
III2/IV2 III2/IV
232
24°C
ts T W
ts
Mito.
2-
ts
TOM complex
kDa
αSu b
24 h 37°C
ck
T
Mito.
W
kDa 669
αCox4 24°C
24 h 37°C ts
24°C
ts T W
ts
2ck
T
2ck
W
Mito.
C
αTom22
24 h 37°C
T
αTom40 24°C
40 41 42
19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
T
9
2-
8
W
7
ck
6
W
5
2-
4
ck
B
3
2-
2
ck
1
669
669
V
232
140 140 66 1
2
3
4
440 5
6
7
8
1
2
3
4
440 5
6
7
8
Figure 2. Inactivation of CK2 in Yeast Causes Depletion of Mitochondrial TOM Complexes (A) WT and ck2-ts yeast were grown at 24 C and shifted to 37 C for 12–24 hr. Mitochondria or whole-cell extracts were analyzed by SDS-PAGE and immunoblotting. Parental, parental strain (YPH250) to ck2-ts and WT; mg, mitochondrial protein. (B) Blue native electrophoresis and immunoblot analysis of the TOM complex from WT and ck2-ts mitochondria. (C) Blue native electrophoresis and immunoblot analysis of the respiratory chain complexes III, IV, and V.
Tom20cd was mainly phosphorylated by CK2 and only weakly by several further kinases (Figure 1A and Figure S1C). Ser172 was the critical residue for the phosphorylation by CK2 (Figure 1C). Western blot analysis with an antiserum recognizing phosphoSer172 confirmed phosphorylation of Tom20 by CK2 (Figure S1E). Tom70cd was preferentially phosphorylated by PKA (Figure 1A). Mutant analysis revealed that Ser174 was the target residue for PKA, whereas the weak phosphorylation by CK1, GSK3, or CamKII was not altered in the mutant proteins analyzed (Figure 1D). Tom40 was phosphorylated by several kinases in vitro, including CK1, PKA (Ser54), and MAPK (Figure 1A and Figure S1F). Tom6 was phosphorylated on Ser16 by the cyclin-dependent kinase complexes, CDK1-cyclin B and CDK2-cyclin A (Figure S1G). These results indicate that purified Tom proteins can be phosphorylated by cytosolic kinases in vitro. The three receptors were found to be major targets for CK2 (Tom22 and Tom20) and PKA (Tom70), respectively. Inactivation of CK2 Leads to Defects of the Mitochondrial TOM Complex In Vivo To ask whether CK2 exerted an effect on the TOM complex in vivo, we used ck2 mutant yeast cells. Yeast CK2 consists of two catalytic a subunits, Cka1 and Cka2, and two regulatory b subunits, Ckb1 and Ckb2 (Poole et al., 2005). Because the activity of CK2 is essential for yeast viability, we employed the temperature-sensitive mutant strain ck2-ts, in which CKA1 and CKA2 have been deleted and a conditional allele of CKA2 is expressed from a plasmid (Hanna et al., 1995). When ck2-ts cells were grown at
permissive temperature (24 C), the levels of mitochondrial proteins were in the wild-type (WT) range (Figure 2A). Upon shift to nonpermissive conditions (37 C), the levels of TOM subunits decreased. After a 12 hr shift, in particular, the levels of Tom22 and Tom20 were reduced, and after a 24 hr shift, the levels of all TOM subunits analyzed were severely reduced (Figure 2A). The levels of further mitochondrial proteins tested were not or were only partially reduced after the 24 hr shift. Proteins of the cytosol and endoplasmic reticulum (ER) analyzed were not reduced in ck2-ts cells compared to wild-type cells after the 24 hr shift, including phosphoglycerate kinase (Pgk1), the chaperones Ssa1 and Kar2, and the translocase Sec61 (Figure 2A). The TOM complex was analyzed by blue native electrophoresis upon lysis of mitochondria with digitonin. The levels of the TOM complex were strongly reduced after the 24 hr shift of ck2-ts cells to nonpermissive conditions (Figure 2B). Innermembrane complexes such as supercomplexes of the respiratory chain (complexes III and IV) and the ATP synthase (complex V) were not or were only moderately affected (Figure 2C). We conclude that inactivation of CK2 causes a reduction of the amount of TOM complexes. Tom22 Is Quantitatively Phosphorylated by CK2 in Yeast To analyze the phosphorylation of Tom22 in a homologous system, we incubated isolated yeast mitochondria with yeast cytosolic extract in the presence of [g-33P]ATP. Phosphorylated Tom22 was only observed in the autoradiography when mitochondria had been dephosphorylated by alkaline phosphatase Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc. 229
A min
Figure 3. Tom22 Is Quantitatively Phosphorylated by CK2 in Yeast
C
Cytosolic extract with [γ-33P]ATP Tom22WT
Mito. AP
Tom22S44,46A
Cytosolic extract
cka2Δ
WT
AP
30 60 120 30 60 120 30 60 120 30 60 120
min Tom22
0
0
30
1
2
3
60 120
0
30
6
7
60 120
Tom22pS46 1
2
3
4
5
6
7
9 10 11 12
Tom22
22 W
T
6A
Porin
To m
To m
To m
22 W
22 S 4
T
4,
T
22 S 4
22 W
To m
To m
T
22 W
To m
To m
Mito.
22 S 44
A
46
A
B
8
Tom40
AP 4
5
8
Phostag
Tom22 1
2
3
D
4
5
6
7
8
9 10 11 12 13
E
Tom22cdWT Cytosolic extract
WT
cka1Δ
cka2Δ
WT
min 0 30 60120 0 30 60120 0 30 60120 0 30 60120
9
(A) Yeast mitochondria were treated with alkaline phosphatase (AP) as indicated and were subjected to phosphorylation by yeast cytosolic extract, followed by SDS-PAGE and autoradiography. (B) Mitochondria of Tom22 phosphomutant strains were treated with AP, lysed, and analyzed by Phos-tag SDS-PAGE and immunoblotting. (C) Mitochondria were treated with AP, incubated with cytosolic extract, and analyzed by SDS-PAGE and immunoblotting. (D) Phosphorylation of Tom22cdWT with cytosolic extract. (E) Phosphorylation of Tom20cdWT with cytosolic extract. See also Figure S2.
Tom20cdWT Cytosolic extract
cka1Δ
WT
cka2Δ
min 0 60 120 0 60 120 0 60 120
Tom22pS46
Tom20pS172
Tom22
Tom20
A cytosolic extract prepared from a yeast strain lacking Cka2, the a0 catalytic Cka2 Cka2 subunit of CK2, was strongly impaired in Pgk1 the phosphorylation of Ser46 on both Pgk1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 1 2 3 4 5 6 7 8 9 AP-treated mitochondria (Figure 3C) and purified Tom22cd (Figure 3D). A cytosolic Cytosolic WT fus3Δ snf1Δ psk2Δ extract Cytosolic WT fus3Δ snf1Δ psk2Δ extract prepared from a yeast strain that extract min 0 60 120 0 60 120 0 60 120 0 60 120 min 0 30 60120 0 30 60120 0 30 60120 0 30 60120 lacked the other catalytic subunit, Cka1, Tom20pS172 Tom22pS46 was able to phosphorylate Tom22cd (FigTom20 Tom22 ure 3D). For comparison, the phosphoryCka2 Cka2 lation of Ser172 of Tom20 was strongly Pgk1 Pgk1 impaired with cka1D as well as cka2D cytosolic extracts (Figure 3E), indicating 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 10 11 12 13 14 15 16 17 18 19 20 21 a differential dependence of Tom22 and Tom20 phosphorylation on the catalytic subunits of CK2. As control, cytosolic (AP) before the incubation with cytosolic extract (Figure 3A, extracts prepared from yeast strains lacking other cytosolic lanes 4–6), suggesting that wild-type mitochondria contained kinases like the MAPK Fus3, Snf1 (the yeast homolog of AMPTom22 in the phosphorylated state. Phosphorylation of Tom22 dependent protein kinase), or Psk2 (a PAS domain-containing was inhibited on mitochondria that were isolated from a yeast kinase) were competent in phosphorylation of Tom22 and strain expressing the Tom22S44,46A mutant form (Figure 3A), Tom20 (Figures 3D and 3E). To probe for an interaction of Tom22 with CK2, we performed demonstrating that Ser44/46 is the major site of Tom22 phosphorylation in organello. For an independent analysis, we used a pull-down assay with tagged Tom22cd. Cka1 and Cka2 were phosphate-affinity (Phos-tag) SDS-PAGE that retards the gel found in the eluate when wild-type yeast cytosol was used (Figmobility of the phosphorylated forms of proteins (Kinoshita ure S2A). However, both catalytic subunits were lacking in the et al., 2006). Treatment of wild-type mitochondria with AP led eluate when cka2D cytosolic extract was used, although the to a quantitative mobility shift of Tom22 detected with a holo- extract contained wild-type levels of Cka1 (Figure S2A), indiTom22 antibody (Figure 3B). In contrast, no shift was observed cating that the CK2 complex interacts with Tom22 via Cka2. with Tom22S44,46A mutant mitochondria; the mutant Tom22 The yield of Cka2 pull-down by Tom22 was influenced by the migrated at the position of dephosphorylated Tom22. The ability of Tom22 to be phosphorylated; the mutant form single-mutant forms Tom22S44A and Tom22S46A migrated Tom22cdS44,46A bound Cka2 with lower efficiency, and similarly, between the phosphorylated and dephosphorylated wild-type addition of ATP to the binding buffer to complete the phosphorspecies and were shifted to the faster-migrating form by AP ylation reaction decreased the interaction of Tom22 with Cka2 treatment (Figure 3B). Together with the observation that the (Figure S2B). Ser44/46-containing peptides of Tom22 were consistently observed in the doubly phosphorylated form by MS/MS Phosphorylation of Tom22 Promotes Its Import (Table S1 and Extended Experimental Procedures), these and Association with Tom20 results demonstrate that mitochondrial Tom22 is quantitatively Yeast mutant cells expressing Tom22S44,46A instead of wild-type Tom22 from the endogenous promoter contained reduced phosphorylated at Ser44 and Ser46. 230 Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc.
A
C Tom22S44,46A
Strain
Pre- 1 h culture
3
7
D
Tom22WT
Pre- 1 14 culture
3
Mito.
WT
[35S]precursor [35S]precursor
7 14
Tom22
Tom22WT
Tom22S44,46A
tom mutant
Tom22WT Tom22S44,46A Tom22WT Tom22S44,46A
min
5
10 20
5
10 20
5
10 20
5
10 20
AP tom20Δ
F1β
Phostag
Tom22
AAC
tom22Δ 1
2
3
4
5
6
Cox4 5
6
7
8
tom70Δ
9 10 1
B + CHX Tom22S44,46A
Tom22WT
Tom22S44,46A
0 2 4 6 8
0 2 4 6 8
0 2 4 6 8
0 2 4 6 8
E
3
4
5
6
7
8
9
[35S]Tom20
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Tom22S44,46A
Tom22 level (% of starting value t=0)
120
10
20
5
10
20
46
To m
22 S 4
4,
46 4,
22 S 4
T
22 W 5
AP
kDa 669
TOM complex
TOM complex
440 232 140
60
Tom22WT + CHX
40 Tom22S44,46A
20 0
min
Mito.
– CHX
Tom22WT
80
Mito.
Mito.
To m
Pgk1
To m
Porin
100
10 11 12
[35S]Tom20 A
Tom22
Tom22WT
h
Tom22WT
Tom22S44,46A
Strain
2
- CHX
T
4
22 W
3
A
2
To m
1
0
1
2
αTom40
%
90 3
105 100 4 5
24 6
28 7
39 8
TOM complex
32 9
30 100 28 10 11 12
66 %
2 4 6 8 Time after CHX addition (h)
Figure 4. Phosphorylation of Tom22 by CK2 Promotes Import and Assembly (A) Western blot of whole-yeast extract from Tom22WT and Tom22S44,46A strains grown from a stationary preculture to early/mid exponential growth phase at 30 C. (B) Western blot analysis of whole-yeast extract from Tom22WT and Tom22S44,46A strains grown in the presence or absence of cycloheximide (CHX). Quantification: the Tom22 level at CHX addition (t = 0 hr) was set to 100% for each growth condition. Data are represented as mean ± SEM (n = 3). (C) Tom22WT and Tom22S44,46A precursors were synthesized in reticulocyte lysate, treated with AP, and analyzed by Phos-tag SDS-PAGE and autoradiography. (D) Transport of Tom22WT and Tom22S44,46A to mitochondria from WT, tom20D, tom22D, and tom70D yeast strains. Analysis by SDS-PAGE and autoradiography. (E) (Left) Blue native electrophoresis and immunoblot analysis of Tom22WT and Tom22S44,46A mitochondria. (Middle and right) Assembly of [35S]Tom20 into the TOM complex of Tom22WT and Tom22S44,46A mitochondria, followed by treatment with AP as indicated. Analysis by blue native electrophoresis and digital autoradiography (quantification: control WT values set to 100%). See also Figure S3.
steady-state levels of the receptor (Figure 4A) (the antibody used recognized wild-type and mutant Tom22 with the same efficiency; Figure S3A). Possible explanations for reduced levels of Tom22 may be an impaired biogenesis or a faster turnover of the mutant protein. To test the second possibility, protein synthesis in the yeast cells was blocked by cycloheximide, and the half-life of Tom22WT and Tom22S44,46A was compared. Despite the lower protein level of Tom22S44,46A, the rate of degradation was indistinguishable between mutant and wildtype protein (Figure 4B). Tom22 is synthesized on cytosolic ribosomes and posttranslationally imported into mitochondria. We synthesized and 35S labeled the precursor in reticulocyte lysate (lacking mitochondria). Treatment with AP quantitatively shifted Tom22WT on Phos-tag SDS-PAGE, whereas the mobility of Tom22S44,46A was not altered by AP yet the mutant protein migrated like AP-treated Tom22WT (Figure 4C). We conclude that the precursor of Tom22 is quantita-
tively phosphorylated at Ser44/46 in the cytosol. We incubated radiolabeled Tom22 precursor with isolated mitochondria and observed that the association of Tom22S44,46A with mitochondria was of lower efficiency compared to Tom22WT (Figure 4D). To test which Tom receptor was critical for the import of phosphorylated Tom22 precursor into mitochondria, we used mitochondria from yeast strains lacking Tom20, Tom22, or Tom70. tom20D mitochondria were impaired in import of phosphorylated Tom22WT, and the difference in import efficiency of Tom22WT and Tom22S44,46A became much smaller than with wild-type mitochondria (Figure 4D, top). In contrast, tom22D and tom70D mitochondria showed the same difference in the efficiency of importing Tom22WT and Tom22S44,46A as wild-type mitochondria (Figure 4D), indicating that these mutant mitochondria discriminated between phosphorylated and nonphosphorylated Tom22 precursors. We conclude that the receptor Tom20 is required for recognition and efficient import of phosphorylated Tom22. Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc. 231
Thus, phosphorylation of the Tom22 precursor stimulates its targeting to mitochondria. As the levels of Tom20 were reduced in ck2-ts mutants (Figure 2A), we asked whether the phosphorylation of Tom20 was required for its biogenesis. Phos-tag gel analysis of yeast mitochondria, however, revealed that only a very minor fraction of Tom20 molecules was phosphorylated in contrast to Tom22 (Figure S3B, lane 1). Only upon incubation with CK2 was a larger fraction of mitochondrial Tom20 phosphorylated at residue Ser172 (Figure S3B). We generated yeast mutant strains that expressed Tom20S172A or Tom20S172E (phosphomimetic glutamate) instead of wild-type Tom20 yet did not observe any difference in the protein composition of the resulting mitochondria (Figure S3C). Import of the precursors of Tom20S172A and Tom20S172E into mitochondria was indistinguishable from that of wild-type Tom20, as analyzed by binding to mitochondria and insertion into the membranes (treatment with Na2CO3 at pH 11.5) (Figure S3D) and assembly into the TOM complex (Figure S3E) (in lanes 4–6 of Figure S3E, the wild-type precursor of Tom20 was treated with CK2, yet the assembly of Tom20 was not affected). Mutant mitochondria carrying Tom20S172A or Tom20S172E imported precursor proteins with wild-type efficiency both via the presequence pathway and via the carrier pathway (Figures S3F and S3G). Thus, mitochondrial Tom20 is phosphorylated by CK2 only to a low level, and replacement of the phosphorylated residue neither affects biogenesis of Tom20 nor import of precursor proteins. Because the phosphorylation status of Tom20 itself was not responsible for the reduced levels of Tom20 in ck2-ts cells, we asked whether the phosphorylation status of Tom22 affected the biogenesis of Tom20. The interaction between Tom22 and Tom20 is involved at two different stages of biogenesis of the TOM complex: (1) mature Tom20 functions as receptor for the precursor of Tom22, and (2) though targeting of the precursor of Tom20 to mitochondria does not require surface receptors, the subsequent assembly of Tom20 into the TOM complex depends on the interaction with Tom22 (Meisinger et al., 2001). Blue native analysis of the TOM complex from a yeast strain containing Tom22S44,46A revealed a double band (Figure 4E, lane 1), the lower band migrating like a TOM complex that lacked Tom20 (Meisinger et al., 2001). We synthesized the precursor of Tom20 and imported it into mitochondria. Mitochondria from the Tom22S44,46A yeast strain were indeed impaired in the assembly of Tom20 into the TOM complex (Figure 4E). Remarkably, when wild-type mitochondria with assembled Tom20 were treated with AP after the import reaction, the association of Tom20 with the TOM complex was strongly reduced (Figure 4E, lane 12). In Tom22S44,46A mitochondria, the amount of Tom20 found on the TOM complex was already at the low level that was not further affected by dephosphorylation (Figure 4E). Taken together, these results show that phosphorylation of Tom22 at Ser44/46 plays a dual role: it not only stimulates targeting of Tom22 to mitochondria, but also the association of Tom20 with the TOM complex. Phosphorylation of Mim1 by CK2 Is Required for the Biogenesis of Tom Proteins Import and assembly of the third import receptor, Tom70, was not affected in Tom22S44,46A mitochondria (Figure 5A), although the 232 Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc.
levels of Tom70 were reduced in ck2-ts yeast mutants (Figure 2A). Mitochondria isolated from ck2-ts yeast were impaired in the assembly of the Tom70 precursor (Figure 5B, lanes 7–9) (Tom70 is only loosely associated with the TOM complex and migrates as dimer on blue native gels; Becker et al., 2008). Import of the precursor of Tom70 does not require TOM receptors but depends on the outer-membrane protein Mim1 (Becker et al., 2008). We thus asked whether the levels of Mim1 were affected in ck2-ts yeast cells. Upon a short 8 hr shift of ck2-ts cells to nonpermissive temperature, the levels of Mim1 were partially reduced, similar to the levels of Tom22 (Figure 5C). After a 16 hr shift, the levels of Mim1 were strongly reduced (Figure 5C). Two high-confidence phosphorylation sites were predicted in Mim1, Ser12 and Ser16, both closely matching the CK2 consensus sequence (Meggio and Pinna, 2003). On Phos-tag gels, mitochondrial Mim1 migrated in two distinct bands that shifted to a single faster-migrating band upon treatment with AP (Figure 5D). We generated a yeast strain in which both Ser12 and Ser16 were replaced by alanine; the resulting Mim1S12,16A migrated as a single band and was not further affected by AP. Replacement of either Ser12 or Ser16 by alanine led to bands with intermediate mobility that were sensitive to treatment with AP (Figure 5D). We conclude that mitochondrial Mim1 is phosphorylated at Ser12 and Ser16. We tested the in vitro phosphorylation of the N-terminal domain of Mim1 fused to GST using the purified kinases CK1, CK2, PKA and CamKII. CK2 efficiently phosphorylated the Mim1 fusion protein in a site-specific manner (Ser12/Ser16) (Figure 5E). In a homologous yeast system, wild-type cytosolic extract phosphorylated the Mim1 fusion protein on Ser12/16 (Figure 5F). A cytosolic extract lacking Cka2 did not phosphorylate Mim1, whereas cka1D cytosol and mutants of further kinases phosphorylated Mim1 with wild-type efficiency (Figure 5F). Phos-tag gel analysis revealed that the precursor of Mim1 synthesized in reticulocyte lysate was partially phosphorylated at Ser12/16 (Figure 5G), indicating that full phosphorylation of Mim1 occurred at mitochondria. Taken together, we conclude that Mim1 is phosphorylated by CK2, mainly by the catalytic subunit Cka2. We analyzed the biogenesis of Tom receptors in Mim1S12,16A mutant mitochondria. The assembly of the precursor of Tom70 was impaired (Figure 5H). Assembly of Tom20 into the TOM complex was also reduced in Mim1S12,16A mitochondria (Figure 5H). In contrast, the assembly of Tom22 into the TOM complex was not affected in the Mim1S12,16A mitochondria (Figure 5H), in agreement with the Mim1-independent biogenesis of Tom22 (Becker et al., 2008). Thus, replacement of the CK2 target sites Ser12/16 of Mim1 impairs biogenesis of the receptors Tom70 and Tom20. Inactivation of CK2 Causes Defects of the Main Mitochondrial Protein Import Pathways Do the defects of the TOM complex in ck2-ts mutant cells lead to an impairment of the main mitochondrial protein import pathways? To directly determine the capacity of mitochondria for importing precursor proteins, we used the in organello import assay with radiolabeled precursor proteins (Stojanovski et al., 2007). Precursor proteins destined for the two major mitochondrial import pathways were analyzed: (1) presequence pathway into the
A Mito.
Tom22
min
5
WT
Tom22
10 20
5
24°C
C
[35S]Tom70
WT
S44,46A
WT
16 h 37°C
ck2-ts
WT
ck2-ts
μg 10 20 40 10 20 40 10 20 40 10 20 40 10 20 40 10 20 40
kDa
10 20
8 h 37°C ck2-ts
670
Mim1
440
Tom22
232 Tom70 dimer
140
Tom70
66
Tom20
1
2
3
4
5
F1β
6
ck2-ts
WT
min
1
3
1
9
3
9
1
9
1
3
F
24 9
19 10
25 104 11 12
%
[γ-33P]ATP S12A S16A
T
1W
1 S1
2A
im
im
1 S1
M
Mim1N
Autoradiography
Mim1N
Coomassie 1
Mim1N
M
[γ- P]ATP
E
T S1 2, 1 W 6A T S1 2, 1 W 6A T S1 2, 1 W 6A T S1 2, 16 A
13 8
33
W
5 7
9
8
7
6
5
4
3
2
1
Tom70
59 105 5 6
Phostag
Mim1
66
57 100 32 2 3 4
M
AP kDa 670 440
9
Tom70 dimer
25 1
im
T
3
232 140
Tom70 dimer
1 S1 2
1W WT
im
Mito.
ck2-ts
im
24 h 37°C
M
24°C Mito.
,1 6A
D
[35S]Tom70
M
B
6A
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
WT
2
3
CamKII
4
5
CK2
6
7
CK1
8
PKA
T
[35S]precursor
W
Δ
k2 Δ
ps
Δ
f1
s3 fu
Sn
W
T
ck a1 Δ ck a2 Δ W T
T W
W
T
G Cytosolic extract
Mim1WT
CK2 AP
Mim1N 1
2
3
4
5
6
7
8
9
10
Phostag
Mim1 1
H
[35S]Tom70 Mim1WT
Mito. min
Mim1S12,16A Mim1WT
1
3
1
3
9
Mim1WT
Mito. kDa 440 232
Tom70 dimer
140
Tom70
66 2
3
4
4
5
min
1
3
9
1
5
6
3
9
TOM complex
8
9
10
11
Mim1WT
Mito. kDa 670 440
7
6
7
8
9
[35S]Tom22
Mim1S12,16A
232
1
3
[35S]Tom20
Mim1S12,16A 9
2
min
5
15
45
Mim1S12,16A 5
15
45
TOM complex
kDa 670 440
140
232 140
66
66
12
13
14
15
16
17
18
Figure 5. Phosphorylation of Mim1 by CK2 Promotes Import of Tom70 and Tom20 (A) Assembly of [35S]Tom70 into Tom22WT and Tom22S44,46A mitochondria analyzed by blue native electrophoresis. (B) Import and assembly of [35S]Tom70 into WT and ck2-ts mitochondria. Quantification by digital autoradiography (control WT value [lane 3] set to 100%). (C) Levels of mitochondrial proteins of WT and ck2-ts yeast cells grown at 24 C or shifted to 37 C for 8–16 hr. Analysis by SDS-PAGE and immunoblotting. (D) Mitochondria were treated with AP and analyzed by Phos-tag SDS-PAGE and immunoblotting. (E) Phosphorylation of Mim1NWT-GST and Mim1NS12,16A-GST with recombinant kinases. (F) Phosphorylation of Mim1NWT-GST and Mim1NS12,16A-GST with yeast cytosolic extracts. (G) [35S]Mim1WT and Mim1S12,16A were synthesized in reticulocyte lysate in the presence or absence of CK2 (10 U/ml), treated with AP, and analyzed by Phos-tag SDS-PAGE. (H) Assembly of [35S]Tom70, Tom20, and Tom22 into Mim1WT and Mim1S12,16A mitochondria analyzed by blue native electrophoresis and autoradiography.
Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc. 233
24°C
[35S]F1β
min Δψ
24 h 37°C WT
ck2-ts
120
100 Mature F1β (% of control)
24°C WT
Mito.
37°C
120
ck2-ts
2 6 18 18 2 6 18 18 2 6 18 18 2 6 18 18
m
100
ck2-ts
Mature F1β (% of control)
A
80 60
WT
40 20 0 6 12 Time (min)
[35S]Su9-DHFR
WT
Mito. min Δψ
6 12 Time (min)
24°C
WT
ck2-ts
20 0
ck2-ts
2 6 18 18 2 6 18 18 2 6 18 18 2 6 18 18
p m 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
18
37°C
120
24 h 37°C Mature Su9-DHFR (% of control)
24°C
ck2-ts
40
18
120
100 ck2-ts
80 60
WT
40 20
Mature Su9-DHFR (% of control)
B
60
0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
WT
80
0
100 WT
80 60 40
ck2-ts
20 0
0
6 12 Time (min)
18
0
6 12 Time (min)
18
[35S]AAC
C
24 h 37°C WT
min 3 10 25 25 3 10 25 25 3 10 25 25 3 10 25 25 Δψ
24°C kDa 440 232 140
Assembled AAC
66
WT
80 60
ck2-ts
40 20 0
AAC precursor 2
3
4
5
6
7
8
9 10 11 12 13 14 15 16
80
WT
60 ck2-ts
40 20 0
0 1
37°C 100
100
Assembled AAC (% of control)
ck2-ts
ck2-ts
Assembled AAC (% of control)
24°C WT
Mito.
5
15 Time (min)
25
0
5
15 Time (min)
25
Figure 6. Inactivation of CK2 Causes Defects in Mitochondrial Protein Import (A) WT and ck2-ts yeast cells were grown at 24 C and shifted to 37 C for 24 hr. [35S]F1b was imported into mitochondria (Dc, membrane potential). Mitochondria were treated with proteinase K and analyzed by SDS-PAGE. p, precursor; m, mature. Quantifications are shown as mean ± SEM (n = 3). Import into WT mitochondria after the longest import time was set to 100% (control). (B) Import of [35S]Su9-DHFR was performed as described in (A). (C) Import of [35S]AAC was analyzed by blue native electrophoresis; quantification as in (A).
matrix: F1-ATPase subunit b and the model precursor Su9-DHFR, consisting of a mitochondrial presequence and the passenger protein dihydrofolate reductase (Stojanovski et al., 2007) and (2) carrier pathway into the inner membrane with the ADP/ATP carrier (AAC) as major substrate (Wiedemann et al., 2001; Dolezal et al., 2006; Neupert and Herrmann, 2007). All three precursor proteins were efficiently imported in a membrane potential (Dc)-dependent manner in ck2-ts mitochondria under permissive conditions (Figures 6A–6C). Upon shift of the mutant cells to nonpermissive temperature for 24 hr, the resulting mitochondria were strongly impaired in import of the different precursor proteins (Figures 6A–6C). We conclude that the inactivation of CK2 leads to major defects in the biogenesis of mitochondrial proteins, both for the presequence pathway and the carrier pathway. PKA Phosphorylates Mitochondrial Tom70 under Fermentable Conditions To study the phosphorylation of Tom70 in vivo, we generated an antiserum that selectively recognized phospho-Ser174 and 234 Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc.
observed a specific signal of mitochondrial Tom70 from wildtype yeast, but not from mutant yeast in which Ser174 was replaced by alanine (Figure 7A, lanes 1 and 2). PKA is a heterotetramer composed of two catalytic subunits and two regulatory subunits. In yeast, the closely related genes TPK1, TPK2, and TPK3 encode the catalytic subunits, and BCY1 encodes the regulatory (inhibitory) subunit (Zaman et al., 2008). Binding of cAMP to Bcy1 releases the active catalytic subunits. Yeast cytosolic extract phosphorylated the cytosolic receptor domain of yeast Tom70 at Ser174; H89 that inhibits the catalytic subunits of PKA blocked the phosphorylation (Figure 7B, top). When cytosolic extract with a low phosphorylating activity was used, the PKA activator 8Br-cAMP stimulated the phosphorylation of Tom70cd (Figure 7B, middle). Cytosolic extract from a yeast mutant in which the catalytic subunits were inactivated (mutant form of TPK3 and deletion of TPK1 and TPK2) was strongly impaired in phosphorylation of Tom70 (Figure 7B, bottom). Yeast PKA is activated on fermentable medium, whereas the PKA activity is low under nonfermentable
YPG
Kemptide-GST Tom70cd Kemptide-GST
Tom70 Pgk1 1
Tom40
10
YPG + Glucose
20 30
10 20 30
0
Tom70pS174
Western Coomassie Blot
PKA Tom70pS174
0
min
Tom70cd
Pgk1 3
4
5
6
7
8
9
F
10
[35S]AAC, WT mito. Tom70cdWT 1
0
0
PKA
0.25 0.5
1
μg kDa 669
100
232 140
Assembled AAC
2
120
440
D 1
2
3
Tom70S174A
Mito.
4
min Δψ
2
18
6
Tom70WT 18
2
6
18
min 5
H89
H89
10 20 5
10 20
18
2
6
18
4
5
6
7
kDa 440
140
0
Load
AAC precursor
66
Elution
PKA
Elution Ssa1
4
5
24 1
107 122 2 3
6
0 4
65 5
81 6
Tom70S174A
Mito. 2
min Δψ
100 7
0 8
22 9
23 10
8 11
0 12
20 40
6
18
Tom70WT
%
Tom70
100
18
2
6
18
Tom70pS174
Tom70S174E 18
2
6
18
18
5
232
20 40 Assembled PiC PiC precursor
Pgk1
Assembled PiC 8
9
10 11 12
50
82 186
0
48
70 100
0
33
36
40
0
13
14
16
17
18
20
21
22
23
24
15
5
tpk1Δ tpk2Δ tpk3 ts
10 20 5
10 20
Tom70pS174
19
H
%
Tom70S174A 5
Tom70WT
10 20 20
5
Tom70S174E
10 20 20
5
3
4
5
tom70Δ
10 20 20 20 20
bcy1Δ
13 14 15 16 17 18 1
2
3
4
5
6
7
8
9
10
11
12
13
7
8
9
μg
5
10 20
10
bcy1Δ
WT rho
10 20 40
5
10 20
WT rho 5
10 20
Tom5
Tom40 Mim1 Tom22
440
Receptorbound AAC
25
WT rho
10 20 40 10 20 40
kDa 669
Pgk1
50
6
Tom70
[35S]AAC, ATP depleted
min Prot. K
2
WT rho μg
Mito.
75
0 1
66
E WT
Bmh1
140
Tom70pS174
7
Pgk1
kDa 440
8Br-cAMP
8Br-cAMP min 5
Tom70
PKA
[ S]PiC 3
0.25 0.5 0.75 1 Recombinant Tom70cd (μg)
Tom70cd
35
2
0
8
Ssa1
Pgk1 1
40
G
18
Assembled AAC
Assembled AAC
Tom70pS174
3
– PKA 60
Tom70S174E
232
B
2
Elution (% of control)
Porin
1
80
20
66 [35S]AAC
min
0.5 0.25
+ PKA
Tom70cdWT
PKA
Assembled AAC (% of control)
YPG
Tom70cd
Western Coomassie Western Blot Blot
C
[γ−33P]ATP
74
To m 70 S 1
T
4A
To m 70 W
To m 70 S 17
Mito.
A
YPD
A
Hsp60
232 140
Tom20
66
AAC
14
PiC 1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18
Figure 7. Phosphorylation of Tom70 by Protein Kinase A Impairs Carrier Import (A) WT and mutant yeast strains were grown on fermentable medium, and mitochondria were isolated and treated with PKA as indicated. Analysis by SDS-PAGE and western blotting. (B) Phosphorylation of Tom70cd with yeast cytosolic extract, analyzed by immunoblotting. (Lanes 1–6) Cytosolic extract of WT yeast (grown on YPD) treated with the PKA inhibitor H89 as indicated. (Lanes 7–12) Cytosolic extract of WT yeast (grown on YPG) supplemented with 8Br-cAMP as indicated. (Lanes 13–18) Cytosolic extracts of WT and PKA mutant yeast strains (grown on YPD), supplemented with 8Br-cAMP. (C) (Left) Cytosolic extract of WT yeast grown on YPD or YPG was incubated with Tom70cd or Kemptide-GST in presence of [g-33P]ATP. (Right) WT yeast cells grown on YPG were incubated with glucose for 15 min as indicated before preparation of cytosolic extract, which was used for phosphorylation of Tom70cd. (D) [35S]AAC and PiC were incubated with Tom70WT, Tom70S174A, and Tom70S174E mitochondria. Mitochondria were treated with proteinase K, lysed with digitonin, and analyzed by blue native electrophoresis (quantification: control WT values [lanes 7 and 19] set to 100%). (E) [35S]AAC was incubated with isolated mitochondria under ATP-depleted conditions, followed by treatment with proteinase K (Prot. K) as indicated. The samples were analyzed by blue native electrophoresis. (F) Assembly of [35S]AAC into WT mitochondria in the presence of soluble Tom70cdWT (mg) that was phosphorylated by PKA prior to the import reaction as indicated. Data are represented as mean ± SEM (n = 4). Assembly in the absence of Tom70cd was set to 100%. (G) Pull-down from cytosolic extract of WT yeast cells with His10-tagged Tom70cdWT that was pretreated with PKA as indicated. Ssa1, Pgk1, and Bmh1: 5% of load, 100% of ATP-elution; Tom70 and Tom70pS174: 5% of load, 5% of imidazole-elution. Data are represented as mean ± SEM (n = 5); values without PKA were set to 100%. (H) WT (rho+), bcy1D, and WT rho yeast strains were grown on fermentable medium, and mitochondria were analyzed by SDS-PAGE and immunoblotting. mg, mitochondrial protein loaded. See also Figure S4.
conditions (Zaman et al., 2008), shown in Figure 7C for the phosphorylation of Tom70cd and the PKA model substrate Kemptide. MS analyses detected phospho-Ser174 containing peptides of mitochondrial Tom70 only from cells grown on fermentable medium, but not from cells grown on nonfermentable medium (Chi et al., 2007; Albuquerque et al., 2008; Extended Experimental Procedures).
The phospho-Ser174 signal observed after growth of yeast on glucose (Figure 7A, lane 2) was strongly increased when the isolated mitochondria were treated with PKA (Figure 7A, lane 3), indicating that only a fraction of the Tom70 molecules had been phosphorylated at Ser174 in vivo. The total mitochondrial levels of Tom70, detected with Tom70-holo antiserum, were not altered by its phosphorylation status (Figure 7A). Thus, also Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc. 235
on a fermentable medium, only a fraction of Tom70 molecules were phosphorylated. Phosphorylation of Tom70 Impairs Import of Mitochondrial Metabolite Carriers For a functional analysis, we first asked whether phosphorylation of Tom70 influenced its targeting to mitochondria. We synthesized the Tom70 precursor in reticulocyte lysate and analyzed its targeting and membrane integration (resistance to extraction at pH 11.5). When Ser174 was replaced by alanine or the phosphomimetic residue glutamate, targeting and membrane insertion of Tom70 were not affected (Figure S4A); the blue native mobility of the Tom70 dimer and the TOM complex were also not altered (Figure S4B). These results indicate that neither targeting of Tom70 to mitochondria nor the oligomeric state of the TOM machinery was altered by the replacement of Ser174. We thus asked whether the replacement of Tom70-Ser174 influenced the import of precursor proteins into mitochondria. Tom70 functions as a main receptor for noncleavable hydrophobic precursors, including the AAC, phosphate carrier (PiC), and dicarboxylate carrier (DIC) of the inner membrane (Neupert and Herrmann, 2007; Chacinska et al., 2009). Import and assembly of the 35S-labeled carrier proteins to the mature dimers can be monitored by blue native electrophoresis (Wiedemann et al., 2001). Mutant mitochondria containing a glutamate instead of Tom70-Ser174 were impaired in import of the carrier proteins, whereas the replacement of Ser174 by alanine enhanced their import (Figure 7D and Figure S4C). To exclude indirect effects of the Tom70 mutants on mitochondrial protein biogenesis, we used two precursor proteins that are not imported by the Tom70-carrier pathway (Chacinska et al., 2009): the presequence-carrying model preprotein b2-DHFR that is transported to the inner membrane in a Dc-dependent manner and the precursor of Tim9 that is transported into the intermembrane space. Both precursor proteins were efficiently imported into the Tom70-Ser174 mutant mitochondria (Figures S4D and S4E). These results indicate that replacement of Tom70-Ser174 by a phosphomimetic residue diminishes the efficiency of protein import via the carrier pathway, whereas replacement by a nonphosphorylatable residue enhances the import efficiency. In wild-type mitochondria, a fraction of Tom70 molecules is phosphorylated, leading to an intermediate import efficiency. The in organello binding of carrier precursors to Tom70 can be visualized by blue native electrophoresis when [35S]AAC is accumulated at the outer membrane of mitochondria upon depletion of ATP (Wiedemann et al., 2001). The precursor proteins are observed in high molecular mass complexes in a Tom70-dependent manner and are accessible to externally added protease (Figure 7E). Replacement of Tom70-Ser174 by alanine enhanced the binding, whereas the replacement by glutamate diminished the binding (Figure 7E). To obtain further evidence that the phosphorylation status of Tom70 is important for precursor binding, we used a competition assay with the soluble cytosolic receptor domain. Tom70cd added to in organello import assays has been shown to bind AAC precursors and thus to compete with their import into mitochondria (Brix et al., 2000). The wild-type form of recombinant Tom70cd, which is not phosphorylated upon synthesis in E. coli, competed with AAC import into mitochon236 Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc.
dria, as did recombinant Tom70cdS174A (Figure S4F). The phosphomimetic mutant form Tom70cdS174E did not affect the import of AAC (Figure S4F). To obtain direct evidence that phosphorylation of Tom70 by PKA impairs precursor binding, we phosphorylated Tom70cd by PKA before the competition assay. Indeed, PKA-treated Tom70cd did not compete for the import of AAC, in contrast to nonphosphorylated Tom70cd (Figure 7F). Carrier proteins are delivered to Tom70 in a complex with the cytosolic chaperone Hsp70 that specifically binds to Tom70 (Young et al., 2003). Ser174 is located in close proximity to the Hsp70-binding pocket of Tom70 (Li et al., 2009). As the negatively charged C-terminal EEVD motif of Hsp70 is involved in binding to Tom70 (Li et al., 2009), we hypothesized that introduction of negative charge by phosphorylation might interfere with Hsp70 binding. We incubated yeast cytosol with tagged Tom70cd and observed binding of Hsp70 (Ssa1) to the receptor, but not of control proteins like Pgk1 or Bmh1 (Figure 7G). Phosphorylation of Tom70cd by PKA diminished the binding of Ssa1 to Tom70 (Figure 7G). We conclude that PKA-mediated phosphorylation of Tom70 at Ser174 impairs the activity of the receptor for binding of Hsp70 and import of carrier precursors. In yeast lacking the regulatory subunit Bcy1, the activity of PKA is not controlled anymore by cAMP levels. bcy1D cells are unable to grow on nonfermentable medium (Zaman et al., 2008). We compared the levels of mitochondrial proteins from bcy1D cells to that of wild-type cells. The levels of TOM proteins, as well as of Mim1 and the control matrix protein Hsp60, were not or were only moderately affected in bcy1D cells (Figure 7H). However, the levels of AAC and PiC were considerably decreased in bcy1D cells (Figure 7H). To exclude that decreased levels of carrier proteins were a general phenotype of strains that cannot grow on nonfermentable medium, we also analyzed mitochondrial protein levels of a rho strain that lacks functional mitochondrial DNA (but contains the subunits of PKA). The levels of the carrier proteins in the rho cells were close to that of the rho+ cells (Figure 7H). In agreement with the in organello studies, the in vivo results indicate that activation of PKA exerts a negative effect on the levels of mitochondrial metabolite carriers. DISCUSSION The TOM complex is essential for the import of most mitochondrial proteins from the cytosol. Though numerous studies have investigated the cytosolic-mitochondrial network in apoptosis and information on the regulation of mitochondrial morphology is increasing (Desagher et al., 2001; Pagliarini and Dixon, 2006; Soubannier and McBride, 2009), little has been known about the regulation of mitochondrial protein import, and posttranslational regulatory mechanisms acting at the preprotein translocases of the mitochondrial membranes have been unknown. We report that the TOM complex is targeted by cytosolic kinases. CK2 and PKA differentially regulate biogenesis and function of the TOM complex and thus control mitochondrial protein homeostasis. CK2 plays important roles in the regulation of numerous cellular processes, including gene expression, cell-cycle progression, cell polarity, and ion homeostasis; roles of CK2 in protection from apoptosis and phosphorylation of enzymes
involved in mitochondrial lipid metabolism have also been reported (Hanna et al., 1995; Desagher et al., 2001; Meggio and Pinna, 2003; Onorato et al., 2005; Poole et al., 2005; Tripodi et al., 2010). We found that CK2 phosphorylates two key components required for the biogenesis of the TOM complex, Tom22 and Mim1, and thus ck2 mutant yeast cells contain strongly reduced levels of the TOM complex. (1) The precursor of Tom22 is quantitatively phosphorylated by CK2 in the cytosol. Phosphorylation of Ser44/46 stimulates targeting of the precursor to mitochondria. The mechanistic basis is that the interaction of Tom22 with its import receptor Tom20 is enhanced by the phosphorylation. The phosphorylation-stimulated interaction of Tom22 and Tom20 serves a dual function in TOM biogenesis: it is not only required for importing the precursor of Tom22, but also for the assembly of imported Tom20 into the TOM complex. (2) The outer-membrane protein Mim1 is not a structural part of the TOM complex but transiently interacts with the complex and is required for the import of several Tom proteins, including the precursors of Tom70, Tom20, and small Tom proteins, but not for the import of Tom22 (Chacinska et al., 2009). CK2 efficiently phosphorylates Mim1, and the levels of Mim1 are reduced in ck2 mutant yeast. Mitochondria containing a mutant form of Mim1 that cannot be phosphorylated by CK2 are impaired in the import of Tom70 and Tom20. Taken together, CK2 controls the biogenesis of all TOM receptors: Tom22 by direct phosphorylation, Tom70 via phosphorylation of Mim1, and Tom20 via Mim1 and Tom22. The strong reduction in the levels of all three TOM receptors, in turn, leads to further defects, of particular importance being the biogenesis of the import channel Tom40 that involves all three receptors, Tom20, Tom22, and Tom70 (Chacinska et al., 2009). The low TOM levels then cause major defects in mitochondrial biogenesis, including the main protein import pathways, presequence pathway and carrier pathway, that strictly depend on the TOM complex. CK2 is constitutively active, though its activity for specific substrates may be modulated by a regulation of copy number, subcellular localization, conformational changes, and interaction partners (Poole et al., 2005). It has been observed that the activity of CK2 is higher in rapidly growing cells, both in mammals and yeast, i.e., under conditions in which more mitochondria are needed (Meggio and Pinna, 2003; Tripodi et al., 2010). By phosphorylation of Tom22 and Mim1, CK2 plays an important stimulatory role in the biogenesis of the TOM complex. CK2 thus controls the levels of the main mitochondrial entry gate, suggesting a role of CK2 in promoting biogenesis of mitochondria. In contrast to the stimulatory role of CK2 on mitochondrial biogenesis, PKA impairs the receptor activity of Tom70. Addition of glucose to yeast cells leads to an activation of this cAMPdependent protein kinase (Zaman et al., 2008). PKA has been found in association with different cell organelles and has been found to affect mitochondrial morphology, oxidative metabolism, and apoptosis; cAMP-dependent phosphorylation of a few nuclear-encoded proteins was shown to modulate their interaction with Hsp70 chaperones and transfer into mitochondria (Robin et al., 2003; Pagliarini and Dixon, 2006; de Rasmo et al., 2008; Soubannier and McBride, 2009). The identification of Tom70 as a target of PKA reveals that PKA not only affects
mitochondrial import via phosphorylation of a small set of precursor proteins, but also exerts a direct regulatory effect on the TOM machinery. Phosphorylation of Tom70 by PKA impairs its receptor activity by diminishing the interaction with cytosolic Hsp70 and the import of hydrophobic precursors such as the inner-membrane metabolite carriers. On nonfermentable medium, cells depend on respiration, and thus a high activity of mitochondria is required, including an increased need for exchange of metabolites between mitochondria and the rest of the cell. Under these conditions, Tom70 is not phosphorylated by PKA and can exert its full receptor activity. When shifting to growth on glucose (fermentable medium), metabolite exchange between mitochondria and the rest of the cell is diminished, and fewer metabolite carrier proteins are needed; a fraction of Tom70 molecules becomes phosphorylated and is impaired in the import of metabolite carriers. The phosphorylation by PKA thus serves as a regulatory means to adjust the import rate of metabolite carriers to the metabolic state of the cell. In summary, the protein import machinery of the mitochondrial outer membrane is regulated by cytosolic kinases at several levels, TOM biogenesis and receptor activity, and under different growth conditions. The regulatory effects of CK2 and PKA on the TOM receptors demonstrate that the main mitochondrial entry gate for preproteins is not functioning autonomously but is embedded into cellular signaling networks. These findings reveal a posttranslational mechanism for controlling mitochondrial preprotein translocases. EXPERIMENTAL PROCEDURES Analysis of Yeast Strains S. cerevisiae cells were grown on YPD (2% glucose), YPG (3% glycerol), or synthetic medium (Trp; 3% glycerol) at 23 C –37 C. Mitochondria were isolated by differential and gradient centrifugation. Yeast strains expressing Tom22WT or Tom22S44,46A from the authentic promoter were inoculated from an overnight preculture to an OD600nm of 0.1 in YPG and were grown at 30 C. For analysis of protein turnover, cycloheximide (50 mg/ml) was added 3 hr after inoculation. YPH250 (parental strain), YDH6 (WT, cka1::HIS3 cka2::TRP1 + p(CEN6/ARSH4 LEU2 CKA2)), and YDH13 (ck2-ts, cka1::HIS3 cka2::TRP1 + p(CEN6/ARSH4 LEU2 cka2–13)) (Hanna et al., 1995) were inoculated to an OD600nm of 0.2 in YPG and were grown for 2 hr at 24 C, followed by an incubation at 24 C or 37 C for 24 hr. Samples were taken, and whole-cell extracts were generated by precipitation with trichloroacetic acid or postalkaline extraction and analyzed by SDS-PAGE and western blotting. See Extended Experimental Procedures. In Vitro and In Organello Phosphorylation Substrate proteins were expressed in E. coli and were purified. In vitro phosphorylation with recombinant protein kinases (CK2, CK1, PKA, CamKII, p42 MAPK, CDK1-cyclinB, CDK2-cyclinA, and GSK3; New England Biolabs) was done in the presence of [g-33P]ATP (Perkin Elmer). For generation of cytosolic extract, yeast cells were treated with zymolyase 20T and lysed with glass beads; cell debris was removed by centrifugation, and the protein content was adjusted by absorbance at 280 nm. The cytosolic extract was supplemented with ATP and mixed with recombinant protein or isolated mitochondria. Phosphorylation was detected by digital autoradiography or western blotting with phosphosite-specific antibodies. Alkaline Phosphatase Treatment and Phosphate-Affinity SDS-PAGE Mitochondria were resuspended in ST buffer (10 mM Tris/HCl [pH 7.2], 250 mM sucrose, 2 mM MgCl2, and 2 mM PMSF) to a protein concentration of 1 mg/ml, and 1 U AP per 5 mg mitochondrial protein was added. Samples
Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc. 237
were incubated for 30 min at 25 C and mild shaking. Mitochondria were washed, lysed, and subjected to SDS-PAGE or Phos-tag electrophoresis (Kinoshita et al., 2006) on 12.5%–15% acrylamide gels containing 50–100 mM Phos-tag and 100–200 mM MnCl2. Protein Import Assays [35S]Methionine-labeled precursor proteins were synthesized in rabbit reticulocyte lysate, and levels were adjusted as necessary. Precursors were incubated with isolated mitochondria in import buffer (250 mM sucrose, 80 mM KCl, 5 mM MgCl2, 2 mM KH2PO4, 5 mM methionine, 3% (w/v) fatty acid-free bovine serum albumin, and 10 mM MOPS/KOH [pH 7.2]) with 2 mM ATP and 2–3 mM NADH (Stojanovski et al., 2007). Mitochondria were washed, treated with proteinase K or Na2CO3, and analyzed by SDS-PAGE or blue native electrophoresis and digital autoradiography. For competition assays (Brix et al., 2000), recombinant Tom70cd was phosphorylated with PKA. SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, four figures, and one table and can be found with this article online at doi:10. 1016/j.cell.2010.12.015. ACKNOWLEDGMENTS We thank Drs. C.V. Glover, A.P. Bidwai, K.M. Dombek, and C. Mann for strains and antisera; Drs. N. Wiedemann, K. Wagner, T. Becker, N. Gebert, M. Bohnert, M. Gebert, and D. Stroud for discussion; and C. Prinz for expert technical assistance. This work was supported by the Deutsche Forschungsgemeinschaft, Excellence Initiative of the German Federal and State Governments (EXC 294 BIOSS; GSC-4 Spemann Graduate School), Bundesministerium fu¨r Bildung und Forschung (Dynamo), Ministerium fu¨r Innovation, Wissenschaft, Forschung und Technologie des Landes Nordrhein-Westfalen, Sonderforschungsbereich 746, Gottfried Wilhelm Leibniz Program, and Landesforschungspreis Baden-Wu¨rttemberg. This paper is dedicated to Klaus Paal (1955–2009), who is deeply missed. Received: March 2, 2010 Revised: September 4, 2010 Accepted: December 7, 2010 Published online: January 6, 2011 REFERENCES Albuquerque, C.P., Smolka, M.B., Payne, S.H., Bafna, V., Eng, J., and Zhou, H. (2008). A multidimensional chromatography technology for in-depth phosphoproteome analysis. Mol. Cell. Proteomics 7, 1389–1396. Becker, T., Pfannschmidt, S., Guiard, B., Stojanovski, D., Milenkovic, D., Kutik, S., Pfanner, N., Meisinger, C., and Wiedemann, N. (2008). Biogenesis of the mitochondrial TOM complex: Mim1 promotes insertion and assembly of signal-anchored receptors. J. Biol. Chem. 283, 120–127. Brix, J., Ziegler, G.A., Dietmeier, K., Schneider-Mergener, J., Schulz, G.E., and Pfanner, N. (2000). The mitochondrial import receptor Tom70: identification of a 25 kDa core domain with a specific binding site for preproteins. J. Mol. Biol. 303, 479–488. Chacinska, A., Koehler, C.M., Milenkovic, D., Lithgow, T., and Pfanner, N. (2009). Importing mitochondrial proteins: machineries and mechanisms. Cell 138, 628–644. Chi, A., Huttenhower, C., Geer, L.Y., Coon, J.J., Syka, J.E., Bai, D.L., Shabanowitz, J., Burke, D.J., Troyanskaya, O.G., and Hunt, D.F. (2007). Analysis of phosphorylation sites on proteins from Saccharomyces cerevisiae by electron transfer dissociation (ETD) mass spectrometry. Proc. Natl. Acad. Sci. USA 104, 2193–2198. de Rasmo, D., Panelli, D., Sardanelli, A.M., and Papa, S. (2008). cAMP-dependent protein kinase regulates the mitochondrial import of the nuclear encoded NDUFS4 subunit of complex I. Cell. Signal. 20, 989–997.
238 Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc.
Desagher, S., Osen-Sand, A., Montessuit, S., Magnenat, E., Vilbois, F., Hochmann, A., Journot, L., Antonsson, B., and Martinou, J.C. (2001). Phosphorylation of Bid by casein kinases I and II regulates its cleavage by caspase 8. Mol. Cell 8, 601–611. Dolezal, P., Likic, V., Tachezy, J., and Lithgow, T. (2006). Evolution of the molecular machines for protein import into mitochondria. Science 313, 314–318. Gnad, F., de Godoy, L.M., Cox, J., Neuhauser, N., Ren, S., Olsen, J.V., and Mann, M. (2009). High-accuracy identification and bioinformatic analysis of in vivo protein phosphorylation sites in yeast. Proteomics 9, 4642–4652. Hanna, D.E., Rethinaswamy, A., and Glover, C.V. (1995). Casein kinase II is required for cell cycle progression during G1 and G2/M in Saccharomyces cerevisiae. J. Biol. Chem. 270, 25905–25914. Holt, L.J., Tuch, B.B., Villen, J., Johnson, A.D., Gygi, S.P., and Morgan, D.O. (2009). Global analysis of Cdk1 substrate phosphorylation sites provides insights into evolution. Science 325, 1682–1686. Kinoshita, E., Kinoshita-Kikuta, E., Takiyama, K., and Koike, T. (2006). Phosphate-binding tag, a new tool to visualize phosphorylated proteins. Mol. Cell. Proteomics 5, 749–757. Li, X., Gerber, S.A., Rudner, A.D., Beausoleil, S.A., Haas, W., Villen, J., Elias, J.E., and Gygi, S.P. (2007). Large-scale phosphorylation analysis of a-factorarrested Saccharomyces cerevisiae. J. Proteome Res. 6, 1190–1197. Li, J., Qian, X., Hu, J., and Sha, B. (2009). Molecular chaperone Hsp70/Hsp90 prepares the mitochondrial outer membrane translocon receptor Tom71 for preprotein loading. J. Biol. Chem. 284, 23852–23859. Meggio, F., and Pinna, L.A. (2003). One-thousand-and-one substrates of protein kinase CK2? FASEB J. 17, 349–368. Meisinger, C., Ryan, M.T., Hill, K., Model, K., Lim, J.H., Sickmann, A., Mu¨ller, H., Meyer, H.E., Wagner, R., and Pfanner, N. (2001). Protein import channel of the outer mitochondrial membrane: a highly stable Tom40-Tom22 core structure differentially interacts with preproteins, small Tom proteins, and import receptors. Mol. Cell. Biol. 21, 2337–2348. Mootha, V.K., Bunkenborg, J., Olsen, J.V., Hjerrild, M., Wisniewski, J.R., Stahl, E., Bolouri, M.S., Ray, H.N., Sihag, S., Kamal, M., et al. (2003). Integrated analysis of protein composition, tissue diversity, and gene regulation in mouse mitochondria. Cell 115, 629–640. Neupert, W., and Herrmann, J.M. (2007). Translocation of proteins into mitochondria. Annu. Rev. Biochem. 76, 723–749. Onorato, T.M., Chakraborty, S., and Haldar, D. (2005). Phosphorylation of rat liver mitochondrial glycerol-3-phosphate acyltransferase by casein kinase 2. J. Biol. Chem. 280, 19527–19534. Pagliarini, D.J., and Dixon, J.E. (2006). Mitochondrial modulation: reversible phosphorylation takes center stage? Trends Biochem. Sci. 31, 26–34. Pagliarini, D.J., Calvo, S.E., Chang, B., Sheth, S.A., Vafai, S.B., Ong, S.E., Walford, G.A., Sugiana, C., Boneh, A., Chen, W.K., et al. (2008). A mitochondrial protein compendium elucidates complex I disease biology. Cell 134, 112–123. Poole, A., Poore, T., Bandhakavi, S., McCann, R.O., Hanna, D.E., and Glover, C.V. (2005). A global view of CK2 function and regulation. Mol. Cell. Biochem. 274, 163–170. Reinders, J., Wagner, K., Zahedi, R.P., Stojanovski, D., Eyrich, B., van der Laan, M., Rehling, P., Sickmann, A., Pfanner, N., and Meisinger, C. (2007). Profiling phosphoproteins of yeast mitochondria reveals a role of phosphorylation in assembly of the ATP synthase. Mol. Cell. Proteomics 6, 1896–1906. Robin, M.A., Prabu, S.K., Raza, H., Anandatheerthavarada, H.K., and Avadhani, N.G. (2003). Phosphorylation enhances mitochondrial targeting of GSTA4-4 through increased affinity for binding to cytoplasmic Hsp70. J. Biol. Chem. 278, 18960–18970. Soubannier, V., and McBride, H.M. (2009). Positioning mitochondrial plasticity within cellular signaling cascades. Biochim. Biophys. Acta 1793, 154–170. Stojanovski, D., Pfanner, N., and Wiedemann, N. (2007). Import of proteins into mitochondria. Methods Cell Biol. 80, 783–806.
Tripodi, F., Cirulli, C., Reghellin, V., Marin, O., Brambilla, L., Schiappelli, M.P., Porro, D., Vanoni, M., Alberghina, L., and Coccetti, P. (2010). CK2 activity is modulated by growth rate in Saccharomyces cerevisiae. Biochem. Biophys. Res. Commun. 398, 44–50. Wiedemann, N., Pfanner, N., and Ryan, M.T. (2001). The three modules of ADP/ATP carrier cooperate in receptor recruitment and translocation into mitochondria. EMBO J. 20, 951–960.
Young, J.C., Hoogenraad, N.J., and Hartl, F.U. (2003). Molecular chaperones Hsp90 and Hsp70 deliver preproteins to the mitochondrial import receptor Tom70. Cell 112, 41–50.
Zaman, S., Lippman, S.I., Zhao, X., and Broach, J.R. (2008). How Saccharomyces responds to nutrients. Annu. Rev. Genet. 42, 27–81.
Cell 144, 227–239, January 21, 2011 ª2011 Elsevier Inc. 239
Dual Action of ATP Hydrolysis Couples Lid Closure to Substrate Release into the Group II Chaperonin Chamber Nicholai R. Douglas,1 Stefanie Reissmann,1,4 Junjie Zhang,3 Bo Chen,2 Joanita Jakana,3 Ramya Kumar,1 Wah Chiu,2,3 and Judith Frydman1,* 1Department
of Biology and BioX Program, Stanford University, Stanford, CA 94305-5020, USA in Structural and Computational Biology and Molecular Biophysics 3National Center for Macromolecular Imaging Baylor College of Medicine, Houston, TX 77030, USA 4Present address: MPI for Terrestrial Microbiology, Marburg D-35043, Germany *Correspondence:
[email protected] DOI 10.1016/j.cell.2010.12.017 2Program
SUMMARY
Group II chaperonins are ATP-dependent ringshaped complexes that bind nonnative polypeptides and facilitate protein folding in archaea and eukaryotes. A built-in lid encapsulates substrate proteins within the central chaperonin chamber. Here, we describe the fate of the substrate during the nucleotide cycle of group II chaperonins. The chaperonin substrate-binding sites are exposed, and the lid is open in both the ATP-free and ATP-bound prehydrolysis states. ATP hydrolysis has a dual function in the folding cycle, triggering both lid closure and substrate release into the central chamber. Notably, substrate release can occur in the absence of a lid, and lid closure can occur without substrate release. However, productive folding requires both events, so that the polypeptide is released into the confined space of the closed chamber where it folds. Our results show that ATP hydrolysis coordinates the structural and functional determinants that trigger productive folding. INTRODUCTION Achieving correct protein folding is critical for cellular health and viability. Failure to fold and maintain protein homeostasis is associated with a growing number of diseases (Hartl and Hayer-Hartl, 2009; Powers et al., 2009). Accordingly, cell viability is dependent on a class of proteins called molecular chaperones, which bind nonnative proteins and facilitate their folding (Bigotti and Clarke, 2008; Frydman, 2001; Hartl and Hayer-Hartl, 2009; Spiess et al., 2004). Among these, the group II chaperonins found in eukaryotic cells and archaea have a unique ring-shaped structure that determines their functional characteristics (Bigotti and Clarke, 2008; Go´mez-Puertas et al., 2004; Spiess et al., 2004). For instance the eukaryotic chaperonin TRiC/CCT assists 240 Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc.
the folding of 10% of newly translated proteins, including essential cytoskeletal proteins, cell-cycle regulators, and tumor suppressors (Thulasiraman et al., 1999; Yam et al., 2008). Intriguingly, many of its substrates, such as actin, cannot be folded by other chaperone systems (Spiess et al., 2004), suggesting that TRiC possesses unique mechanistic features absent from other chaperones. Group II chaperonins are large complexes consisting of two stacked rings of eight (or less frequently nine) subunits each (Bigotti and Clarke, 2008; Go´mez-Puertas et al., 2004; Spiess et al., 2004). Individual subunits are generally different, ranging from one to four in archaea, to eight different subunits for TRiC/ CCT. The general subunit architecture is conserved across group II chaperonins. Each subunit consists of an equatorial, ATP-binding domain, an intermediate hinge domain, and an apical domain, which contains the substrate-binding sites; a flexible protrusion extends from the apical domain and acts as a built-in lid. ATP binding and hydrolysis drives group II chaperonins through a conformational cycle that is not well understood. In the absence of nucleotide, the lid-containing segments are open, and the complex binds substrate. The open-state structures of TRiC/CCT and an archaeal chaperonin from Methanococcus maripaludis are remarkably similar (Booth et al., 2008; Pereira et al., 2010; Zhang et al., 2010). Incubation with hydrolyzable ATP induces a compact conformation, where the lid segments of each subunit form a beta-stranded iris that closes over the central cavity of the complex. The structure of this closed state is also virtually the same in eukaryotic and archaeal chaperonins (Booth et al., 2008; Cong et al., 2010; Ditzel et al., 1998; Pereira et al., 2010; Zhang et al., 2010). The presence of an intact lid is dispensable for substrate binding and ATP hydrolysis in both eukaryotic and archaeal chaperonins. However, the lid confers allosteric coupling of subunits within the complex and is essential for substrate folding (Kanzaki et al., 2008; Meyer et al., 2003; Reissmann et al., 2007). Although the fully open and fully closed states are emerging in some structural detail, little is known about the trajectory of the chaperonin through the conformation cycle or how substrate folding is achieved (Bigotti and Clarke, 2008).
A number of studies using archaeal and eukaryotic chaperonins have suggested that ATP binding suffices to close the built-in lid and trigger substrate folding (Iizuka et al., 2003; Llorca et al., 2001; Villebeck et al., 2007; Stuart et al., 2011). Subsequent ATP hydrolysis would serve to reopen the lid and release the folded protein. In contrast, other studies reported that ATP binding alone is unable to close the lid or promote substrate folding (Bigotti et al., 2006; Meyer et al., 2003; Reissmann et al., 2007). Instead, these studies identified the transition state of ATP hydrolysis as the critical step in the ATPase cycle that promotes the closed conformation (Meyer et al., 2003; Reissmann et al., 2007). A fundamental question for group II chaperonins concerns the fate of the substrate during the ATPase cycle. The current model proposes that group II chaperonins do not release the substrate during folding (Go´mez-Puertas et al., 2004; Stuart et al., 2011). Instead, ATP binding would cause the apical domains with their bound substrate to move, and this movement mechanically forces substrate folding. In this view, substrate liberation occurs after nucleotide hydrolysis, perhaps after nucleotide release and the subsequent return of the chaperonin to the open state. Some experimental results are not reconciled easily with the ‘‘mechanical force’’ model. The substrate-binding sites of group II chaperonins have been mapped to the vicinity of helix 11 (Spiess et al., 2006), which is unavailable to the central cavity in the ATP-induced closed state. The mechanical model of group II chaperonin action suggests that the cavity is not necessarily a folding chamber per se, rather it is used as a mechanical scaffold for active remodeling. This led to the suggestion that the lids primarily assist in the conformational cycle of the chaperonin (Kanzaki et al., 2008). However, ATP incubation of a group II chaperonin lacking a lid (Cpn-Dlid) produces an identical conformation to that of wild-type but is unable to promote substrate folding (Reissmann et al., 2007; Zhang et al., 2010). Thus, the movement of the apical domains does not require the presence of the lid; however, their movement alone is insufficient to promote folding. Here, we use the group II chaperonin from mesophilic archaea Methanococcus maripaludis, herein Cpn, to define the fate of the polypeptide substrate during the conformational cycle of group II chaperonins. The allosteric regulation and structure of this Cpn are similar to those of TRiC/CCT (Pereira et al., 2010; Reissmann et al., 2007; Zhang et al., 2010). We find that ATP hydrolysis has a dual role in group II chaperonin function, promoting both lid closure and release of the substrate into the cavity. Importantly, both events must occur for successful substrate folding. We suggest an alternate model for group II chaperonin function, whereby folding relies on the release of the substrate into a unique chemical environment within the closed chamber.
principle, folding of a polypeptide with a strict chaperonin requirement, i.e., a stringent substrate, could require several cycles of Cpn binding and release (Figure 1A). Alternatively, the substrate could fold in a single ATPase-cycle event, without requiring multiple rounds of binding and release. To test these possibilities we employed rhodanese, a stringent Cpn substrate (Martin et al., 1991). 35S-rhodanese binds to nucleotide-free Cpn in an unstructured, proteinase K (herein PK)-sensitive state (Figure 1A, left arrow, Figure 1B, lane 2 bottom panel, and Figure 1C for native gel analysis). Addition of ATP induces lid closure and encapsulates the substrate within the closed chamber (Meyer et al., 2003; Reissmann et al., 2007) (Figure 1A). Upon closure, the Cpn lid segments and the encapsulated 35S-rhodanese are protected from proteolytic digestion (Figure 1B, lane 3). Importantly, ATP addition causes the time-dependent folding of rhodanese (Figure 1D, red symbols). Comparing the kinetics of rhodanese folding (t1/2 12 min) with the estimated kinetics of a single round of ATP hydrolysis (Bigotti et al., 2006; Reissmann et al., 2007) indicates that completion of rhodanese folding involves several cycles of ATP binding and release. Similar results are observed for malate dehydrogenase (MDH) (see Figure S1B available online; data not shown). Importantly, addition of protease at any time following ATP addition interrupted the folding reaction (Figure 1D, PK, shown for t = 0 and t = 13 min). Because PK can only degrade the substrate if the lid is open, this result suggests that the Cpn-substrate complex undergoes repeated cycles of ATP-driven opening and closing during the folding reaction. We next examined whether such iterative cycling is required to achieve folding by exploiting the observation that addition of AlFx together with ATP locks group II chaperonins in a symmetrically closed state that fully encapsulates the substrate (Meyer et al., 2003) (Figure 1A, right arrow). The ATPdAlFx-induced state of Cpn-rhodanese was locked closed, leading to full proteolytic protection of both Cpn and substrate (Figure 1B, lane 4) and a characteristic electrophoretic migration shift on native gels (Figure 1C). Under these conditions, ATPase cycling is interrupted (Figure S1A), and the substrate undergoes a single round of binding and encapsulation, allowing us to evaluate whether iterative cycling is required for group II chaperonin folding (Figure 1E). Strikingly, the rate and yield of rhodanese folding under these noncycling conditions were identical to those observed for the actively cycling chaperonin (Figure 1E). Addition of PK to the ATPdAlFx reaction did not interrupt folding, confirming that there was no reopening of the Cpn and no release of the nonnative substrate under these conditions. We conclude that the closed chamber of group II chaperonins is the folding-active compartment. Furthermore, a single round of encapsulation in this chamber can achieve maximum rhodanese folding, with similar kinetics and yield as observed under cycling conditions. Thus, although iterative cycling does occur, it is not strictly required for Cpn-dependent folding.
RESULTS A Single Round of Encapsulation within the Closed Chamber Suffices for Substrate Folding We initially examined whether the folding reaction is completed within the closed central chamber of group II chaperonins. In
The Closed, Folding-Active, State of Group II Chaperonins Requires ATP Hydrolysis To examine whether ATP binding suffices to promote the foldingactive state of group II chaperonins, we specifically impaired the ATPase-active site by targeting Asp386, which is essential to Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc. 241
ATP
A PK
AlFx
Figure 1. Role of Substrate Encapsulation and Iterative Cycling in Group II Chaperonin Action
38
0
80
Rhodanese Folding (% of Native)
Rhodanese Folding (% of Native)
10
0
(A) Cpn cycle between an open, substrate-accepting state, and an ATP-induced closed state. In each cycle the substrate (in blue) is released in either the folded or unfolded state. Unfolded or substrate rebinds Cpn for iterative rounds of folding. Incubation with ATP and AlFx interrupts 35 Cpn• S-Rho C B iterative cycling by locking Cpn in a closed state that encapsulates the substrate. In the absence of + PK ATP + ATP, PK (scissors) digestion interrupts iterative AlFx ATP + + cycling by specifically digesting the substrate (B) + AlFx and the open lid segments in Cpn. (B) PK sensiCoom. Blue tivity of open and closed Cpn states. PK leads to Coom. full digestion of the open Cpn lids (Coomassie Blue stain, top panel) and the bound substrate, 35 S-rhodanese (35S-Rho, bottom panel; lane 2). 1 2 3 4 35 S-Rho ATP-induced cycling to the closed state protects 35 S-Rho both the Cpn lids and the substrate (lane 3). Incubation with ATPdAlFx locks the complex % Rho protected closed leading to complete PK protection of both lids and encapsulated 35S-rhodanese (lane 4). A purified complex of Cpndrhodanese at 0.25 mM was incubated in the presence or absence of 1 mM ATP and/or 1 mM AlFx for 10 min at 37 C Cycling (ATP) E No Cycling (ATP•AlFx) D and digested with 20 mg/ml PK for 5 min at 25 C. ATP PK ATP PK (C) Native gel analysis of Cpn-substrate complexes. Incubation with ATPdAlFx shifts the f(t) f(t) AlFx 80 mobility of Cpn (top panel Coomassie blue stain), 80 No PK which carries the encapsulated substrate (bottom 70 70 panel for autoradiography of 35S-rhodanese). 60 60 Nonnative rhodanese aggregates cannot migrate 50 ATP into the native gel (data not shown). PK 50 ATP•AlFx 40 t=13 min (D) Folding under cycling conditions. ATP (5 mM) 40 ATP•AlFx 30 was added to initiate Cpn-mediated folding of 30 PK t=0 min 20 rhodanese, measured at the indicated time points. PK AlFx 20 Addition of PK at the times indicated immediately 10 t=0 min 10 interrupts the folding reaction, indicating that the Cpn is cycling between open and closed states 0 20 40 60 80 100 120 0 20 40 60 80 100 120 during folding. Time (min) Time (min) (E) Folding under noncycling conditions. Cpn mediated folding as in (D), except that folding was initiated either by addition of ATP (cycling allowed), ATPdAlFx (no cycling allowed), AlFx (control), or ATPdAlFx and PK (no cycling allowed, no rebinding of released Rho). The folding yields and rates were identical for all conditions, indicating that cycling is not required for group II chaperonin-mediated folding. See also Figure S1.
coordinate the water molecule that participates as a nucleophile during the hydrolysis of the phosphate-anhydride bond (CpnD386A) (Figure 2A). Cpn-D386A cannot hydrolyze ATP but retains efficient ATP binding (data not shown; Reissmann et al. [2007]). Importantly, unlike Cpn-WT, Cpn-D386A is unable to fold the stringent Cpn substrates rhodanese (Figure 2B) and malate dehydrogenase (data not shown). This demonstrates that ATP binding is insufficient to induce the fully folding-active state observed upon ATP hydrolysis. We next assessed the proposal that ATP binding leads to partial (Clare et al., 2008) or full (Iizuka et al., 2003; Llorca et al., 2001) lid closure. To this end the structure of ATP-bound Cpn-D386A was derived to 15 A˚ resolution by single-particle cryo-EM (Figure 2C, blue). Comparison of these structures with the ATP-free and ATP-bound states of Cpn-WT, derived to 10 and 6 A˚, respectively, revealed the conformational changes induced by ATP binding, distinguishing them from those induced 242 Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc.
by ATP hydrolysis (Figures 2D; Figure S2). ATP incubation with Cpn-WT induces lid closure, yielding a symmetrically closed structure similar to that previously obtained for Cpn-WT with ATPdAlFx (Figure 2C, cyan; see also Figure S5) (Pereira et al., 2010; Zhang et al., 2010). In contrast, ATP binding to CpnD386A yielded an open structure that resembled the nucleotide-free state (Figure 2D for overlay,; Figures S2A and S2B). Further addition of AlFx did not result in closure (data not shown). Despite leaving the lid open, ATP binding induced a 20 A˚ constriction in the chaperonin opening (Figures 2D; Figure S2B, 110 A˚ span versus 130 A˚ in the ATP-free state). Closer analysis of the conformational changes in a single subunit indicated that ATP binding induces an en masse rigid body tilt of the entire intermediate and apical domains toward the ATP-binding equatorial domain (Figure S2C). We conclude that ATP binding is insufficient to close the lid but triggers domain movements that lead, upon hydrolysis, to the closed state. These results are
A
B Rhodanese Folding (% of WT)
O
F
O
Mg 2+
100
H Asp386
O ADP Al H O- O O F F OH OH
Rhodanese
80 60 40 20
ATP
-
+ WT
C
WT - ATP
D386A + ATP
WT + ATP
-
+
D386A
No Cpn
D
Side View
ATP-free vs. ATP-bound Top View
ATP-binding
ATP-free
E
WT ATP AlFx
-
+ -
+ +
-
+ -
+ +
WT ATP AlFx Tot
D386A
PK - + - +
Coom. Blue
1 35
S-Rho•Cpn Autoradiography
(A) ATP-binding pocket of group II Cpn from T. acidophilum (pdb ID 1A6E) highlighting Asp386, essential for ATP hydrolysis. (B) Rhodanese folding for Cpn-WT and CpnD386A. ATP hydrolysis is required to support rhodanese folding; data are represented as mean ± SEM (n = 3). (C) Single-particle cryo-EM reconstructions of Cpn-WT and Cpn-D386A. Shown are side and top views of Cpn-WT without (left, gold) and with ATP (right, cyan) and Cpn-D386A with ATP (middle, blue). (D) Overlay of EM density maps for Cpn-WT ATP and Cpn-D386A +ATP highlights the changes induced by ATP binding. (E) Role of ATP and ATPdAlFx on the 35S-rhodanese interaction with Cpn-WT and Cpn-D386A. Cpn complexes were analyzed on 4% native gels, and the Rho-containing Cpn was visualized by autoradiography. (F) PK digestion of Cpn-WT and Cpn-D386A complexes with 35S-rhodanese, analyzed by SDSPAGE followed by Coomassie blue staining (top), and autoradiography (bottom). Cpn-D386A is incapable of closing (compare lane 3 for WT with lane 6, lane 7 for D386A). See also Figure S2.
ATP-hydrolysis
F
D386A
Figure 2. ATP Hydrolysis Is Required for the Closed Folding-Active State
35
2
3
S-Rho
consistent with fluorescence experiments on the thermosome from Thermoplasma acidophilum, indicating a rapid rearrangement attributed to ATP binding, followed by a slower rearrangement attributed to ATP hydrolysis and lid closure (Bigotti and Clarke, 2005; Reissmann et al., 2007). The effects of ATP binding on the conformation of both the substrate and the lid were further examined using biochemical assays (Figures 2E and 2F). As described above, addition of ATPdAlFx to Cpn-WT stabilizes the closed state, locking the encapsulated substrate inside the chamber and leading to proteolytic protection of both the lids and the substrate (Figures 1B and 1C and Figures 2E and 2F, lane 3; top panel for Cpn, bottom panel for 35S-Rho for 35S-Rho-Cpn-WT complex). Both ATP and ATPdAlFx induce a structurally similar closed state in Cpn-WT (e.g., Figure 2C, right panel), but the ATPdAlFx state displays a characteristic faster electrophoretic migration on native gels (Figure 2E) (a similar effect is observed for TRiC/
CCT [Meyer et al., 2003]). In contrast to Cpn-WT, incubation of Cpn-D386A with either ATP or ATPdAlFx failed to produce Tot the signature mobility shift (Figure 2E). Furthermore, both the lid and the substrate remained in a largely unstructured, protease-sensitive state upon ATP binding (Figure 2F, lanes 5–7), 4 5 6 7 consistent with the result that ATP binding leaves Cpn in an open state (Figures 2C and 2D). Importantly, the lid also remains open under conditions where only one ring binds nucleotide (0.2 mM; Reissmann et al. [2007]) or if Cpn-WT is incubated with the nonhydrolyzable ATP analogs AMPPNP or ATPgS (at either 0.2 or 1 mM; data not shown), further supporting the conclusion that ATP binding to either one ring or both does not suffice to close the lid. PK + + - +
ATP Hydrolysis Triggers Substrate Release from the Chaperonin-Binding Site Lid closure and substrate encapsulation are essential for folding substrates such as actin for TRiC (Meyer et al., 2003) and rhodanese (Reissmann et al., 2007) and MDH for Cpn (Figures S3A and S3B). We next examined whether lid closure modulates the interaction of the substrate with the chamber. The ‘‘mechanical force’’ model proposes that the chaperonin does not release the substrate proteins into the closed cavity; in this scenario the chaperonin-substrate interaction persists in the closed state leading to the mechanical remodeling of the substrate Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc. 243
A Cpn-WT
without lid
Cpn- lid
Trap
Figure 3. ATP Hydrolysis Triggers Substrate Release from Group II Chaperonins
(A) Proposed models for how closure affects substrate interactions with the central Cpn chamber. (i) The substrate remains bound in the closed state, or (ii) the substrate is released into (i) (ii) (i) (ii) the central cavity. The closed Cpn-WT retains substrate in either model (left). Removal of the lid, - / + ATP - / + ATP B Cpn Cpn C Native Gel Native Gel 35 35 yielding Cpn-Dlid, allows testing of these models. S-Rho S-Rho no Trap + Trap Cpn-Dlid will lose the substrate if closure weakens lid lid the interaction with the chaperonin (model (ii) WT lid WT lid D386A D386A right). A decrease in substrate affinity might also p ATP + + + + + ATP + a Tr be revealed using a GroEL-derived trap (Frydman and Hartl, 1996). Cpn substrate-binding sites Cpn Cpn 35 S-Rho shown as pink lines. 35 S-Rho (B) Effect of ATP binding and hydrolysis on the Trap Cpn-Dlid-substrate interaction. The indicated 35 S-Rho 100 91 100 18 100 102 Cpn-35S-rhodanese complexes, incubated with or Cpn-bd 90 46 90 3 88 86 % Rho bound without 1 mM ATP for 10 min at 37 C, were 97 12 14 Trap-bd 10 54 10 analyzed by native gel electrophoresis followed by % Rho bound autoradiography. The amount of 35S-rhodanese that remains Cpn bound in each condition is indiATP F D cated. lid WT (C) Effect of ATP binding and hydrolysis on AlFx ATP + + release of nonnative substrate from Cpn com+ + AlFx plexes. Autoradiography of native gel for reactions carried out as in (B), but in the presence of 35 Trap S-Rho equimolar GroEL-Trap, which functions as a scavenger for released nonnative proteins. A lid WT E reaction where denatured rhodanese is added Coom. ATP/AlFx + + directly to the Trap is included as a control. The Blue PK + + + + amount of Cpn-bound and Trap-bound rhoda35 nese was calculated for each reaction from the S-Rho native gel analysis. (D) Transition state mimic ATPdAlFx locks the Cpn-Dlid in the symmetrically closed state + ATP + ATP-AlFx H G I immediately halting the ATPase cycle (see Fig3.6 4.0 3.6 ure S1; Zhang et al. [2010]). lid-NR-Rho 3.4 (E) PK digestion of 35S-rhodanese complexes with 3.0 3.4 -ATP Cpn-WT or Cpn-Dlid in the presence or absence 3.2 of ATPdAlFx. The 35S-rhodanese is completely 3.2 2.0 digested in the closed Cpn-Dlid. 3.0 NR-Rho 3.0 (F) Native gel analysis of 35S-rhodanese-chaperin buffer 1.0 2.8 onin complexes incubated as in (E). Cpn-Dlid + 2.8 ATPdAlFx fully releases its substrate (top panel, 35 S-Rho) even though both Cpns undergo the 200 400 600 200 400 600 550 600 650 700 750 800 same conformational change with ATPdAlFx Time (s) Time (s) Wavelength (bottom panel; Coom. Blue). (G) Fluorescence emission spectra of NR-Rho in the presence and absence of Cpn-Dlid (Kim et al., 2005). Binding to the Cpn causes an increase in fluorescence intensity at 630 nm. (H) Time-dependent changes in the fluorescence intensity of NR-Rho emission at 630 nm. Red trace indicates NR-Rho-Cpn-Dlid complex in the absence of ATP. Addition of ATP (arrow) causes a decrease in fluorescence (blue trace). (I) Time-dependent changes in the fluorescence intensity of NR-Rho emission at 630 nm as in (H); arrow indicates addition of ATPdAlFx, which causes a qualitatively similar decrease in fluorescence intensity (cyan trace). See also Figure S3. ATP
ATP
or
conformation (Figure 3Ai, left) (Llorca et al., 2001). Alternatively, ATP hydrolysis could promote substrate release into the closed chamber (Figure 3Aii, left). Because monitoring the substratechaperonin interaction inside the closed chamber is complicated by the presence of the lid, we exploited the previously characterized Cpn-Dlid variant that lacks the entire lid-forming segments (Pereira et al., 2010; Reissmann et al., 2007; Zhang et al., 2010). Importantly, Cpn-Dlid achieves the same ATP-induced ‘‘closed’’ conformation as Cpn-WT (Zhang et al., 2010), and its 244 Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc.
NR-Rho Fluorescence (x 104)
NR-Rho Fluorescence (x 104)
NR-Rho Fluorescence (x 104)
or
ATPase activity and substrate-binding ability are unaffected (Reissmann et al., 2007). These features of Cpn-Dlid allowed us to distinguish between the above models (Figure 3A, right panels). Thus, the model that proposes that the polypeptide remains associated with the chaperonin throughout the ATPase cycle predicts that the substrate will remain bound to Cpn-Dlid upon addition of ATP or ATPdAlFx (Figure 3Ai, ‘‘Dlid’’ right). In contrast if ATP weakens the chaperonin-substrate interaction, the absence of the lid will allow the polypeptide to diffuse
away from the chaperonin (Figure 3Aii, ‘‘Dlid’’ right). Of note, Cpn-Dlid cannot promote folding of substrates such as rhodanese and MDH (Figures S3A and S3B; Reissmann et al. [2007]); thus, substrate release from Cpn-Dlid cannot be ascribed to completion of folding. Purified 35S-rhodanesedCpn complexes were incubated in the presence or absence of ATP for 10 min and analyzed using native gels followed by autoradiography (Figure 3B). Cpn-WT comigrates with the substrate under both conditions (Figure 3B, WT), as expected given that 35S-rhodanese is encapsulated in the closed complex (Figures 1B and 1C). The small ATP-induced reduction in bound substrate is presumably due to loss through ATPase cycling and/or folding (see below, Figure 3C). Strikingly, incubation of Cpn-Dlid with ATP led to a dramatic reduction in the amount of Cpn-bound rhodanese (Figure 3B, Dlid). This ATP-dependent loss of rhodanese required ATP hydrolysis because it was not observed when the Cpn-Dlid also carried the D386A mutation (Figure 3B, Dlid/D386A). Similar results were obtained for other Cpn-bound polypeptides, including MDH (data not shown) and actin (see below; Figure 5). The ATP-induced reduction in Cpn-substrate affinity was further evinced through the use of a ‘‘Trap,’’ a modified GroEL that scavenges nonnative polypeptides (Figure 3C) (Frydman and Hartl, 1996). Trap will not bind to folded rhodanese but will bind to nonnative polypeptides once they are released from the Cpn (Frydman and Hartl, 1996) (Figure 3C, see Trap lane). For all Cpn variants tested, little or no 35S-rhodanese was captured by the Trap in the absence of ATP, suggesting that rhodanese binds stably to all nucleotide-free Cpn variants and cannot be displaced by the Trap (Figure 3C, ATP). Addition of ATP to Cpn-WT allowed a fraction of rhodanese to bind to the more rapidly migrating Trap (Figure 3C, WT+ATP). Comparing the WT incubations in the presence and absence of Trap (i.e., Figures 3B and 3C) suggests that during normal ATP cycling a fraction of the substrate is released in a nonnative form that rebinds to the chaperonin for another round of folding. This nonnative polypeptide is captured by the Trap, which thus prevents Cpn rebinding and interrupts the cycle. Importantly, addition of ATP to Cpn-Dlid-35S-rhodanese caused a nearcomplete transfer of the bound polypeptide to the Trap (Figure 3C, Dlid), indicating that ATP induces substrate release from the chaperonin. Furthermore, no increase in substrate transfer to the Trap was observed upon ATP addition to CpnDlid D386A (Figure 3C, Dlid/D386A), indicating that substrate dissociation from Cpn requires ATP hydrolysis. The experiments above show that ATP hydrolysis has a function that is completely lid independent, namely, to release the substrate from the chaperonin-binding sites. We next employed ATPdAlFx, which mimics the trigonal-bipyramidal transition state of ATP hydrolysis (Meyer et al., 2003) (Figure 3D). As with CpnWT (Figure S1A), the addition of AlFx to Cpn-Dlid immediately arrests its ATPase activity, suggesting that inhibition of ATP hydrolysis and trapping of the closed state occurs after a single cycle (Figure S3C). Whereas incubation of Cpn-WT-35S-rhodanese with ATPdAlFx closes the chamber and encapsulates the substrate (Figures 3E and 3F), the substrate remains protease sensitive following incubation of Cpn-Dlid-35S-rhodanese with ATPdAlFx (Figure 3E). Native gel analysis showed that Cpn-
Dlid with ATPdAlFx undergoes the same signature shift as Cpn-WT, consistent with structural analyses showing that both Cpns adopt the same closed conformation upon incubation with ATPdAlFx (Pereira et al., 2010; Zhang et al., 2010). ATPdAlFx induced a complete release of a broad panel of polypeptides (Figure 3F for 35S-rhodanese; Figures S3D–S3G for other substrates; Figure 5 for Actin), indicating that ATP hydrolysis blocks general access to the substrate-binding sites. The same conclusion was reached using size exclusion chromatography of purified Cpn-35S-rhodanese complexes incubated in the presence or absence of ATPdAlFx and analyzed on a Bio-Sil SEC-400-5 column (Figure S3H). This experiment also indicated that ATPdAlFx induces full substrate release from Cpn-Dlid. The effect of nucleotide hydrolysis on Cpn-substrate interactions was further examined using rhodanese carrying the environmentally sensitive fluorescent moiety Nile Red (Kim et al., 2005) (herein NR-Rho; Figures 3G–3I). In free solution, NR-Rho exhibits a low fluorescence emission spectrum characteristic of an aqueous, polar environment, with a maximum at 650 nm (Figure 3G, gray trace). However, binding to Cpn caused a fluorescence intensity increase as well as a blue shift of the maximal intensity to 630 nm (Figure 3G, red trace for Cpn-Dlid; similar results obtained for Cpn-WT; data not shown). This change in fluorescence upon Cpn binding is a diagnostic for rhodanese occupying a more hydrophobic environment (Kim et al., 2005). We used the maximal fluorescence at 630 nm to monitor the effect of nucleotides on the substrate-chaperonin interaction. The CpnDlid-NR-Rho fluorescence signal remained stable in the absence of nucleotide (Figures 3H and 3I, red traces). Addition of ATP produced a rapid decay in fluorescence intensity (Figure 3H, ‘‘+ATP,’’ blue trace). This supports our previous conclusion that ATP cycling by Cpn leads to substrate release. Addition of ATPdAlFx yielded similar results (Figure 3I, ‘‘+ATPdAlFx,’’ cyan trace), supporting the idea that the ATP hydrolysis-transition state induces substrate release. We conclude that ATP hydrolysis has a dual function within the chaperonin cycle; it promotes lid closure (Figure 2) and also triggers substrate release from the chaperoninbinding sites (Figure 3). Strikingly, the latter function is not dependent on the presence of a lid. The Chaperonin Substrate-Binding Sites Are Unavailable in the Closed State A simple model explaining our results is that the ATP-induced Cpn conformation no longer exposes the substrate-binding sites. We tested this model using an order of addition experiment (Figure 4). In the control condition (Figure 4, Ctrl), substrate was added to the open, apo-Cpn, which exposes the substrate-binding sites. The second condition added the substrate first, prior to incubation with ATPdAlFx (Figure 4, S/A); this condition measured the extent of ATPdAlFx-induced substrate release. In the third condition we incubated with ATPdAlFx first and then added substrate to the chaperonin (Figure 4, A/S); this measured the ability of an ATPdAlFx-preincubated closed complex to bind substrate (Figure 4A). If the binding sites are still available in the closed state, we might expect some substrate binding for closed Cpn-Dlid in the A/S condition, which still retains a large opening allowing access to the central cavity (Pereira et al., 2010; Zhang et al., 2010). Because the pore size may restrict polypeptide entry to Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc. 245
A
B
Condition
Ctrl
S A
Order of Addition
Cpn + Substrate
1: Substrate 2: ATP•AlFx
A
S
1: ATP•AlFx 2: Substrate
ii. PepB (12 aa)
i. Rhodanese (293 aa) WT
Δlid
WT
Δlid
Ctrl S A A S
Ctrl S A A S
Ctrl S A A S
Ctrl S A A S
C
S A
Ctrl
WT
Δlid
WT
A
Δlid
WT
the cavity and may sterically interfere with substrate binding, we used both rhodanese (Figure 4Bi) and a small 12-mer peptide substrate (herein PepB) (Figures S3G; Figure 4Bii). The small peptide substrate should be able to freely diffuse inside the closed chamber in the Cpn-Dlid. In the absence of nucleotide, both substrates bound to Cpn-WT and Cpn-Dlid (Ctrl; Figures 4B and 4C; Figure S3G). As expected, addition of ATPdAlFx to the Cpn-substrate complex (Figure 4, S/ A) promoted substrate encapsulation for Cpn-WT (WT S/A) (Figures 4B and 4C) and substrate release for Cpn-Dlid (Dlid S/A) (Figures 4B and 4C). In the case of A/S, closing the Cpn-WT chamber with ATPdAlFx precluded substrate binding; thus, the closed lid blocks access to the central cavity (WT A/ S) (Figure 4B and scheme in Figure 4C). For Cpn-Dlid, substrate should bind the chaperonin in the A/S condition provided that the binding sites are still available in the closed conformation. This was not the case; instead the ATPdAlFx preincubated CpnDlid was unable to bind either 35S-rhodanese or the small PepB (Figure 4B) (Cpn-Dlid compare S/A and A/S). Thus, the ATPdAlFx state of Cpn-Dlid no longer exposes the substrate-binding sites. Given that the ATPdAlFx conformations of Cpn-WT and Cpn-Dlid are virtually identical (Pereira et al., 2010; Zhang et al., 2010), these experiments show that the substrate-binding sites are no longer available upon ATP hydrolysis. Mechanism of ATP-Induced Substrate Release What is the possible mechanism for substrate release in group II chaperonins? A structural analogy with the distantly related bacterial group I chaperonins, e.g., GroEL, is not possible, given that they use a detachable lid, GroES, which upon ATP binding, both caps the cavity and displaces the substrate. In contrast we 246 Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc.
S
Figure 4. The Substrate-Binding Sites Are Unavailable in the Closed Cpn State (A) Order of addition experiment to test availability of substrate-binding sites in the open and closed Cpn states. Without nucleotide, both Cpn-WT and Cpn-Dlid are open and bind substrate (Ctrl). Substrate addition prior to incubation with ATPdAlFx allows the substrate to bind first before closure (S/A); incubation with ATPdAlFx prior to substrate to addition examines if the closed state can bind substrate (A/S). (B) Native gel analysis of the above incubations. Two Cpn-binding substrates of different sizes were used: (i) rhodanese (293 aa), and (ii) PepB (12 aa). The smaller peptide should access more readily the substrate-binding sites. 35 S-Rhodanese was detected by autoradiography; Alexa 488PepB was detected by fluorescence scan. (C) With Ctrl, both substrates bind chaperonin in the open state. S/A shows that when the lid is present (Cpn-WT), closure of the Cpn-substrate complex retains the substrate in the chamber; when the lid is absent (Cpn-Dlid), the substrate escapes the cavity. A/S illustrates how the ATPdAlFx state blocks substrate binding to both Cpn-WT and Cpn-Dlid. Because Cpn-Dlid retains access to the inner chamber in the closed state, this result indicates that the substrate-binding sites are hidden in the closed state.
Δlid
show that substrate release in group II chaperonins is lid independent and requires ATP hydrolysis. Closer examination of Cpn structures in the open and closed states led to a hypothesis for how ATP hydrolysis induces substrate eviction (Figure 5A) (Pereira et al., 2010; Zhang et al., 2010). In the open state the substrate-binding region around helix 11 is well exposed (Figure 5A, pink in left panel) (Spiess et al., 2006), leaving ample space to accommodate the bound substrate. In contrast the closed state brings the apical domains from adjacent subunits into close proximity (Figure 5A). Closure causes helix 11 to form a tightly packed interface with a loop spanning residues 327–331 in its neighboring subunit (Figure 5A, cyan). Such lateral intra-ring contacts might displace the substrate from its binding site, causing the 327–331 region to act as a ‘‘release loop for the substrate’’ (herein rls loop). To disrupt this lateral interface, we made Ala substitutions in four loop residues making key contacts with helix 11 yielding the Cpn-rls variants (Figure 5A, T327A, N328A, K330A, and D331A). To better understand the role of the rls loop within the chaperonin cycle, we used cryo-EM to obtain a detailed structural characterization of the conformation of both Cpn-rls and Cpn-rls-Dlid in the presence or absence of ATP or ATPdAlFx (Figure S4A for Fourier shell correlation analysis of resolutions; Figure S4B for Cpn-rls-Dlid; and Figure S5 for Cpn-rls). The rls chaperonins achieve essentially the same closed state as the wild-type counterparts (Figures 5B and 5C; Figure S4B for Cpn-rls-Dlid; Figure S5 for Cpn-rls). Consistent with their ability to reach a closed state, the Cpn-rls mutants were competent for ATP binding and hydrolysis (data not shown). We initially focused on Cpn-rls-Dlid because the absence of a lid simplifies analysis of substrate release (Figures 5D–5F).
Open State
A
Closed State
Figure 5. Structural Basis of ATP-Induced Substrate Release in Group II Chaperonins
B Cpn-rls + ATP
C Cpn-rls- lid + ATP Cpn-rls T327A, N328A, K330A, D331A
D
E * Cpn-rls
*
ATP
* ? *
lid WT ATP -
+
lid -
rls- lid
+
-
*
*
Cpn-rls lid
lid
WT
+
ATP/AlFx -
+
* ? *
ATP/AlFx
-
rls- lid
+
-
+
Coom. Blue
Coom. Blue
Rhodanese
Rhodanese
Actin
1
2
3
4
5
Actin
6 1
i. Cpn- lid
3
4
5
ii. Cpn-rls- lid + ATP
3.6 3.4 3.2
NR-Rho Fluorescence (x 104)
NR-Rho Fluorescence (x 104)
F
2
+ ATP
+ AlFx
3.8
6
(A) Structures of group II chaperonins in the open and closed states highlighting helix 11, the locus of substrate binding (pink). ATP-induced closure brings together adjacent apical domains, creating a tight interface between helix 11 of one subunit and loop 327–331 of the neighboring subunit (green). The indicated tetra-alanine substitution in loop 327–331 (herein rls) was introduced in both Cpn and Cpn-Dlid (herein Cpn-rls and Cpn-rlsDlid). (B and C) ATP induces the compact closed state in Cpn-rls (B) and Cpn-rls-Dlid (C). Top view of structures of indicated Cpn states obtained by single-particle cryo-EM reconstructions to 4–6 A˚ (see also Figure 6, Figure S4, and Figure S5). (D) ATP fails to induce substrate release in Cpnrls-Dlid. The indicated Cpn-substrate complexes were incubated in the presence and absence of ATP; substrate release assessed using native gel electrophoresis followed by Coomassie staining to visualize the Cpn(s) (top panel) and fluorescence scans to view substrate(s) (middle and bottom panels). (E) ATPdAlFx triggers substrate release in Cpn-rlsDlid. Incubations and analysis as in (D), except that incubations were carried out in the presence and absence of ATPdAlFx. (F) Nucleotide-induced changes in the environment of NR-Rho bound to Cpn-Dlid (i) or Cpn-rlsDlid (ii). Experiments performed as in Figure 3H. Starting from a nucleotide-free NR-Rho-Cpn complex (red trace), ATP was added to reaction at time indicated by an arrow (blue trace) and incubation continued. For Cpn-rls-Dlid no drop in fluorescence was observed upon ATP addition; after 5 min, AlFx was added to the ATP reaction and incubation continued (cyan trace). See also Figure S4.
3.6 3.4
blue panel), consistent with the cryo-EM analysis. Surprisingly, unlike ATP, incu* * 2.8 3.0 bation with ATPdAlFx caused Cpn-rlsDlid to efficiently release all the 200 400 600 300 600 900 Time (s) Time (s) substrates tested (Figure 5E for rhodanese and actin). This observation was striking given the apparent similarity between the ATP and ATPdAlFx strucCpn-WT and Cpn-Dlid served as controls. In the absence of ATP, tures of Cpn-rls variants (Figure 5E; Figure S4B). Thus, it appears all chaperonins bound rhodanese and actin efficiently, as shown that, in the rls mutant, the conformation promoting substrate by native gel analysis (Figure 5D). Strikingly, Cpn-rls-Dlid was release cannot be stably populated by ATP alone, whereas incapable of releasing either substrate in the presence of ATP, ATPdAlFx can stabilize this state and evict the substrate. Fluorescence spectroscopy provided independent support for unlike Cpn-Dlid (Figure 5D, compare lane 6 to lane 4). This suggests that the lateral contacts between helix 11 and the rls the above conclusions. As for Cpn-Dlid, NR-Rho bound to loop 327–331 are indeed important for releasing the substrate Cpn-rls-Dlid had an emission spectrum characteristic of a hydrophobic environment (data not shown). In contrast to upon ATP hydrolysis. We next examined the effect of the transition state mimic Cpn-Dlid (Figure 5Fi, blue trace), ATP incubation did not cause ATPdAlFx (Figure 5E). Native gel analysis showed that any appreciable change in the fluorescence of NR-Rho bound Cpn-rls-Dlid adopts the same fast migrating conformation to Cpn-rls-Dlid (Figure 5Fii, blue trace), indicating that ATP alone observed for Cpn-Dlid and Cpn-WT (Figure 5E, Coomassie cannot release the bound substrate. However, when AlFx was 3.0
ATP
3.2
*
* ATP *
AlFx
*
Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc. 247
Figure 6. Substrate Release into the Central Chamber Is Required for Group II Chaperonin-Mediated Folding (A) Use of Cpn-rls to test the role of substrate release in group II chaperonin folding. Incubation with ATP should lead to lid closure without substrate release, whereas addition of ATPdAlFx should release the substrate into the closed cavity. (B) Side views of single-particle cryo-EM reconstructions of ATPdAlFx induced state of Cpn-rls and Cpn-WT highlight the similarity of both closed structures (see also below; Figure S5). (C–F) Comparative structural analysis of the ATP and ATPdAlFx states of Cpn-WT and Cpn-rls. i. Top views of overlays for the electron density maps. ii. Superimposition of apical domain region for a single subunit from the overlaid chaperonin models. Superimposition of structures obtained for Cpn-rls and Cpn-WT reveals that the ATP state of Cpn-WT (purple) is virtually identical to the ATPdAlFx states of both Cpn-WT (blue) and Cpnrls (cyan). In contrast, ATP induces a different closed state in Cpn-rls (yellow); comparison with ATPdAlFx states reveals major differences in the position of helix 11 (red arrow) and the rls loop (blue arrow, residues 327–331). (G) Cpn-rls binds rhodanese efficiently and encapsulates the substrate upon ATP or ATPdAlFx induced closure. i. Native gel analysis of 35 S-rhodanese bound to Cpn-rls in the presence or absence of ATP or ATPdAlFx. ii. PK digestion of incubations from (i). Cpn-rls produces a proteaseresistant lid in the presence of ATP or ATPdAlFx (top panel) that fully encapsulates the substrate (bottom panel for 35S-rhodanese). Note similarity with Cpn-WT in Figures 2E and 2F. (H) Rhodanese folding requires substrate release into the central chamber. Rhodanese complexes with Cpn-WT or Cpn-rls were incubated with the indicated nucleotides, and folding was assessed as in Figure 1; data are represented as mean ± SEM (n = 3). See also Figure S5.
added to an ongoing incubation of NR-RhodCpn-rls-Dlid with ATP, the fluorescence rapidly dropped, indicating substrate release from the chaperonin (Figure 5Fii, cyan trace). A similar reduction in fluorescence intensity was observed if ATP and AlFx were added together but was absent if only AlFx was added (data not shown). We conclude that weakening the lateral contacts between helix 11 and its neighboring subunit prevents substrate release, even though Cpn-rls-Dlid can hydrolyze ATP and achieve the closed state. However, stabilizing the posthydrolysis state by addition of AlFx populates the conformation that evicts the substrate. Structural Basis of Substrate Release and Encapsulation We next examined the effect of the rls mutations in the Cpn with the intact lid (herein Cpn-rls, Figure 6). Detailed structural 248 Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc.
analyses of the ATP and ATPdAlFx induced states in both CpnWT and Cpn-rls revealed interesting differences between these chaperonins (Figures 6B–6F; Figure S5). Single-particle cryoEM reconstructions were obtained to 4–6 A˚ for both chaperonins in the presence of either ATP or ATPdAlFx (Cpn-WT-ATP 6 A˚, Cpn-WT-ATPdAlFx 4.3 A˚, Cpn-rls-ATP 5 A˚, Cpn-rls-ATPdAlFx 6 A˚, Figure 6; Figure S5). Models of these structures were then built by flexible fitting into the density map with Rosetta (Figure 6; Figure S5A; see Figure S5B for goodness of fit between model and density map) (DiMaio et al., 2009). Cpn-rls achieved a closed state with either ATP or ATPdAlFx, similar to those obtained with Cpn-WT. Notably, superimposition of the structures of Cpn-WT and Cpn-rls in the different nucleotide states revealed variations in their structure, particularly in the region corresponding to the apical domains (Figures 6C–6F; i. top view of superimposed EM density
maps). These differences were also evident when comparing the apical domain regions in the respective chaperonin models (Figures 6C–6F; ii. detail of apical domain and lid for a subunit within the complex). The ATP (magenta) and ATPdAlFx (blue) states of Cpn-WT were essentially identical (Figure 6C). Thus, ATPdAlFx generates the same closed state observed under ATP-cycling conditions (e.g., Figure 1E). Importantly, we observed a shift in the apical domain regions between the closed Cpn-rls states induced by ATP (yellow) and ATPdAlFx (cyan) (Figure 6D). Cpn-rls-ATP also exhibited noticeable differences with both closed WT structures (e.g., Figure 6F). The apical domain protrusions in Cpn-rls-ATP are shifted clockwise, and the apical domains, including the lid, are tilted up compared to the ATPdAlFx state, exhibiting significant variations in helix 11 (ii. red arrow) and the rls loop (ii. blue arrow). In contrast the ATPdAlFx states of Cpn-WT and Cpn-rls were nearly identical (Figure 6E). These structural analyses demonstrate that even though Cpnrls can close with ATP, impairment of the helix 11/loop 327– 331 contacts results in aberrant intra-ring interactions between the apical domains. This is consistent with the inability of Cpnrls-Dlid to release the substrate in the presence of ATP (Figure 5Fii). Furthermore, ATPdAlFx induces a closed conformation in Cpn-rls that is indistinguishable from the closed state of CpnWT with either ATP or ATPdAlFx. This is consistent with, and explains, the finding that ATPdAlFx leads to substrate release in Cpn-rls-Dlid (Figure 5Fii). Substrate Release and Encapsulation Are Required for Productive Folding The identification of a mechanism that evicts the bound polypeptide upon closure allowed us to test the relevance of substrate release for the folding cycle. First, the ability of Cpn-rls to encapsulate a bound substrate was examined by native gel analysis (Figure 6Gi) and PK digestion (Figure 6Gii), as shown above for Cpn-WT. Incubation of the Cpn-rls with rhodanese yielded a binary complex that behaved exactly as that of Cpn-WT (Figure 6Gi). Protease digestion analysis indicated that, in the absence of nucleotide, the substrate binds in an unstructured conformation (Figure 6Gii, lane 2). Importantly, incubation with either ATP or ATPdAlFx led to proteolytic protection of both the chaperonin lid segments (Figure 6Gii, lanes 3 and 4, top panel) and the bound 35S-rhodanese (Figure 6Gii, lanes 3 and 4, bottom panel). Thus, both ATP and ATPdAlFx induce stable lid closure and fully encapsulate the substrate within the central chamber of Cpn-rls. Rhodanese-chaperonin complexes were prepared for Cpn-rls and Cpn-WT, which served as a control (Figure 6H). As expected, addition of ATP or ATPdAlFx to the Cpn-WT complex induced rhodanese folding (Figure 6H, black traces). Strikingly, addition of ATP to the Cpn-rls complex failed to promote rhodanese folding, even though the substrate was encapsulated within the closed chamber (Figure 6H, green trace). We hypothesized that failure to fold stems from the failure to release the bound substrate into the central chamber. Therefore, we tested the effect of ATPdAlFx, which should promote substrate release (Figure 5). Addition of ATPdAlFx to the Cpn-rls reaction caused efficient rhodanese folding (Figure 6H). These experiments indicate that lid closure and substrate encapsulation are, by them-
selves, unable to promote substrate folding. Importantly, they demonstrate that substrate release into the central closed chamber is essential for productive folding by group II chaperonins. DISCUSSION Our study defines how the ATPase cycle of group II chaperonins modulates the interaction with substrates (Figure 7). We find that ATP hydrolysis triggers substrate release from the chaperonin through a hitherto unanticipated mechanism involving lateral intra-ring contacts between adjacent apical domains. Given the high degree of structural and mechanistic similarity among all group II chaperonins, our findings have broad implications to understand cellular folding in eukaryotes and archaea. Role of ATP Binding in the Chaperonin-Conformational Cycle To resolve the role of ATP binding in group II chaperonin action, we specifically impaired hydrolysis by targeting D386 (Ditzel et al., 1998). We find that ATP binding alone does not support substrate folding or lid closure, similar to previous findings for TRiC/CCT (Meyer et al., 2003). ATP binding does induce a conformational change that constricts the Cpn chamber entrance from 130 to 110 A˚ (Figure S2; Figures 7A and 7B). The movement results from an en bloc counterclockwise rotation of the intermediate and apical domains with respect to the equatorial, ATP-binding domain (Figure S2). Notably, a similar concerted movement of intermediate and apical domains has previously been observed during lid closure for TRiC/CCT (Booth et al., 2008). Our results indicate that ATP hydrolysis is generally required for lid closure and folding in group II chaperonins, underscoring the general conservation of architecture and mechanism between archaeal and eukaryotic chaperonins. The Closed Group II Chaperonin Chamber Is a ‘‘Folding-Active’’ Compartment ATP hydrolysis has a dual role within the group II chaperonin cycle: it both triggers lid closure and releases the substrate from the apical domains into the cavity (Figure 7). Importantly, both events are required for productive folding. Lid closure in the absence of substrate release is also insufficient to achieve folding (Figure 6H). This contrasts with the previously proposed mechanical force model, which suggests that folding occurs through movement of the apical domains without releasing the substrate. The observation that we can generate a chaperonin state that can close the lid without releasing the substrate raises the possibility that lid closure and substrate encapsulation precede release (Figures 7A and 7B, shown in brackets). Such a mechanism would ensure that substrates are confined inside the chamber prior to their release, thereby avoiding the premature escape of nonnative aggregation-prone species into the cytosol. The released substrate folds while encapsulated in the central cavity (Figure 1E). No folding was observed when the substrate was released into the bulk solution (Cpn-Dlid) (Figures S3A and S3B), indicating that the chemical and physical characteristics of the closed central chamber create a folding-active compartment. The nature of this compartment will depend on the side Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc. 249
A
ATP
ATP
Substrate
Substrate
Binding
Hydrolysis
Release
Folding
Folding Active
Folding Inactive
B
ATP
ATP
Binding
Hydrolysis
C
Figure 7. Model for Group II Chaperonin-Folding Mechanism (A) ATP regulation of the Cpn substrate cycle. In the absence of ATP, chaperonins are open, exposing substrate-binding sites (pink). Upon ATP binding, the lid remains open and the substrate bound, but a subtle conformational change is observed. ATP hydrolysis has a dual function: close the lid, and release the substrate by hiding the substrate-binding sites. We hypothesize that lid closure may precede substrate release, transiently generating a closed but foldinginactive state (in brackets). Substrate release into the closed chamber is required for folding, which occurs within the central chamber. (B) Top view of the chaperonin-substrate cycle in (A), highlighting the mechanism of polypeptide release upon ATP hydrolysis. The open, ATP-free and ATP-bound, states expose the substrate-binding region (pink). ATP hydrolysis creates a lateral contact with the rls loop (green) that displaces the substrate into the central cavity. (C) Top views of open and closed crystal structures from Cpn (Pereira et al., 2010) highlighting the substrate-binding region (pink) and rls loop (green).
chains exposed in the closed state as well as the effect of crowding on the solvent properties of the chamber (Tang et al., 2006). The hetero-oligomeric nature of most group II chaperonins may lead to a diversification of the chamber properties (Cong et al., 2010), which may contribute to the folding of specific substrates. Although a single encapsulation step suffices for optimal folding in vitro, it is important to consider that in the cellular context, 250 Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc.
cycling on and off the chaperonin likely fulfills an important homeostatic function. Thus, each cycle may expose the substrate polypeptide to additional folding cofactors as well as quality control components, thereby preventing folding-incompetent proteins from clogging the chaperonin. How the balance between processivity and clearance is achieved in vivo is an important question for future research.
ATP Hydrolysis Triggers Substrate Release through a Unique Interdomain Displacement Mechanism ATP hydrolysis releases the bound substrate from its chaperonin-binding sites through a hitherto unanticipated mechanism; namely, a conformational change that brings together vicinal apical domains. This creates a lateral interface between helix 11 of one domain and loop 327–331 in the adjacent subunit (Figures 7B and 7C, pink and green, respectively). The crystal structure suggests that formation of this lateral H-bonded network is incompatible with substrate binding. We hypothesize that these lateral intersubunit contacts displace the substrate from its binding site (pink in Figure 7). The precise mechanism of release will require further investigation. One possibility is that the intersubunit interaction sterically interferes with substrate rebinding during thermal breathing of the chaperonin-substrate interaction. Alternatively, the helix 11-rls loop interaction could create an entropic zipper that displaces the substrate. Yet another model is that the rls interaction helps stabilize a conformation that cannot bind substrate. The presence of ATPdAlFx may compensate energetically for the loss of the H-bonded network between substrate-binding region and rls loop, and by itself induce the subtle conformational change required to release the substrate. The unique nature of substrate release in group II chaperonins may have important implications for hetero-oligomeric chaperonins, particularly in light of recent findings that different subunits recognize distinct motifs in the substrate (Spiess et al., 2006). Because the mechanism for substrate release depends on the nature of a specific intersubunit interface, rather than a general GroES-binding interface as observed in GroEL, the local kinetics of substrate release could vary for a specific apical domain (e.g., shading in Figure 7B). The order of release of different regions of a substrate polypeptide from their respective subunits may be influenced by the strength of this interaction vis-a`-vis the timing of conformational change and formation of the lateral interface. The ensuing sequential mechanism of substrate release from the chaperonin could provide exquisite control of the folding pathway of the substrate, which in turn contributes to the unique ability of these chaperonins to fold specific proteins. One could envision that subunit-specific substrate remodeling and/or ordered release directs substrates of group II chaperonins along specific folding trajectories. Exploring these exciting possibilities may have profound implications for our understanding and ability to control cellular folding pathways. EXPERIMENTAL PROCEDURES Biochemical Approaches All Cpn variants were produced by site-directed mutagenesis; purification and functional analyses were performed as described (Reissmann et al., 2007). MDH refolding was performed as in Hayer-Hartl (2000). Fluorescent proteins were generated as in Kim et al. (2005), and fluorescence was measured on a FluoroLog-3 Fluorometer (HORIBA Jobin Yvon). Cryo-EM Analyses Samples were embedded in vitreous ice on 400-mesh R1.2/1.3 Quantifoil grids (Quantifoil Micro Tools GmbH, Jena Germany) and imaged on a JEM3200FSC electron cryo-microscope and JEM2010F electron cryo-microscope (JEOL Ltd., Tokyo, Japan) with field emission guns. Details about the image acquisition parameters are in Table S1 of the Extended Experimental Procedures. The
image processing steps followed those described in (Baker et al., 2010). The figures were prepared using MacPyMOL (http://www.pymol.org) and UCSF Chimera (Pettersen et al., 2004).
ACCESSION NUMBERS Coordinates have been deposited in the Electron Microscopy Databank and Protein Data Bank under ID codes EMD-5244, PDB:3IZH; EMD-5245, PDB:3IZI; EMD-5246, PDB:3IZJ; EMD-5247, PDB:3IZK; EMD-5248, PDB:3IZL; EMD-5249, PDB:3IZM; and EMD-5250, PDB:3IZN.
SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, five figures, and one table and can be with this article online at doi:10.1016/j.cell. 2010.12.017. ACKNOWLEDGMENTS We thank Dr. W.E. Moerner for providing Nile Red maleimide and advice on fluorescence microscopy. We also thank members of the J.F. laboratory and Raul Andino for critical reading of the manuscript; Frank DiMaio and David Baker for advice on modeling; Erik Miller for useful discussions; and Jeremy England and Erik Miller for providing PepB. This work was supported by NIH Grants GM74074 (to J.F.), 5PN2EY016525 (to J.F. and W.C.), and P41RR002250 (to W.C.). N.R.D. was a recipient of predoctoral ARCS fellowships and was supported by an NIH training grant (T32-007276). J.Z. was a recipient of NIH training grants on nanobiology (R90 DK071054 and T90 DA022885). Received: July 21, 2010 Revised: October 26, 2010 Accepted: December 10, 2010 Published: January 20, 2011 REFERENCES Baker, M.L., Zhang, J., Ludtke, S.J., and Chiu, W. (2010). Cryo-EM of macromolecular assemblies at near-atomic resolution. Nat. Protoc. 5, 1697–1708. Bigotti, M.G., and Clarke, A.R. (2005). Cooperativity in the thermosome. J. Mol. Biol. 348, 13–26. Bigotti, M.G., and Clarke, A.R. (2008). Chaperonins: the hunt for the group II mechanism. Arch. Biochem. Biophys. 474, 331–339. Bigotti, M.G., Bellamy, S.R., and Clarke, A.R. (2006). The asymmetric ATPase cycle of the thermosome: elucidation of the binding, hydrolysis and productrelease steps. J. Mol. Biol. 362, 835–843. Booth, C.R., Meyer, A.S., Cong, Y., Topf, M., Sali, A., Ludtke, S.J., Chiu, W., and Frydman, J. (2008). Mechanism of lid closure in the eukaryotic chaperonin TRiC/CCT. Nat. Struct. Mol. Biol. 15, 746–753. Clare, D.K., Stagg, S., Quispe, J., Farr, G.W., Horwich, A.L., and Saibil, H.R. (2008). Multiple states of a nucleotide-bound group 2 chaperonin. Structure 16, 528–534. Cong, Y., Baker, M.L., Jakana, J., Woolford, D., Miller, E.J., Reissmann, S., Kumar, R.N., Redding-Johanson, A.M., Batth, T.S., Mukhopadhyay, A., et al. (2010). 4.0-A resolution cryo-EM structure of the mammalian chaperonin TRiC/CCT reveals its unique subunit arrangement. Proc. Natl. Acad. Sci. USA 107, 4967–4972. DiMaio, F., Tyka, M.D., Baker, M.L., Chiu, W., and Baker, D. (2009). Refinement of protein structures into low-resolution density maps using rosetta. J. Mol. Biol. 392, 181–190. Ditzel, L., Lo¨we, J., Stock, D., Stetter, K., Huber, H., Huber, R., and Steinbacher, S. (1998). Crystal structure of the thermosome, the archaeal chaperonin and homolog of CCT. Cell 93, 125–138.
Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc. 251
Frydman, J. (2001). Folding of newly translated proteins in vivo: the role of molecular chaperones. Annu. Rev. Biochem. 70, 603–647.
chaperonin reveal the open and closed states associated with the protein folding cycle. J. Biol. Chem. 285, 27958–27966.
Frydman, J., and Hartl, F.U. (1996). Principles of chaperone-assisted protein folding: differences between in vitro and in vivo mechanisms. Science 272, 1497–1502.
Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt, D.M., Meng, E.C., and Ferrin, T.E. (2004). UCSF Chimera—a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612.
Go´mez-Puertas, P., Martı´n-Benito, J., Carrascosa, J.L., Willison, K.R., and Valpuesta, J.M. (2004). The substrate recognition mechanisms in chaperonins. J. Mol. Recognit. 17, 85–94.
Powers, E.T., Morimoto, R.I., Dillin, A., Kelly, J.W., and Balch, W.E. (2009). Biological and chemical approaches to diseases of proteostasis deficiency. Annu. Rev. Biochem. 78, 959–991.
Hartl, F.U., and Hayer-Hartl, M. (2009). Converging concepts of protein folding in vitro and in vivo. Nat. Struct. Mol. Biol. 16, 574–581.
Reissmann, S., Parnot, C., Booth, C.R., Chiu, W., and Frydman, J. (2007). Essential function of the built-in lid in the allosteric regulation of eukaryotic and archaeal chaperonins. Nat. Struct. Mol. Biol. 14, 432–440.
Hayer-Hartl, M. (2000). Assay of malate dehydrogenase. A substrate for the E. coli chaperonins GroEL and GroES. Methods Mol. Biol. 140, 127–132. Iizuka, R., Yoshida, T., Shomura, Y., Miki, K., Maruyama, T., Odaka, M., and Yohda, M. (2003). ATP binding is critical for the conformational change from an open to closed state in archaeal group II chaperonin. J. Biol. Chem. 278, 44959–44965. Kanzaki, T., Iizuka, R., Takahashi, K., Maki, K., Masuda, R., Sahlan, M., Yebenes, H., Valpuesta, J.M., Oka, T., Furutani, M., et al. (2008). Sequential action of ATP-dependent subunit conformational change and interaction between helical protrusions in the closure of the built-in lid of group II chaperonins. J. Biol. Chem. 283, 34773–34784. Kim, S., Semyonov, A., Twieg, R., Horwich, A., Frydman, J., and Moerner, W. (2005). Probing the sequence of conformationally induced polarity changes in the molecular chaperonin GroEL with fluorescence spectroscopy. J. Phys. Chem. B 109, 24517–24525. Llorca, O., Martı´n-Benito, J., Grantham, J., Ritco-Vonsovici, M., Willison, K., Carrascosa, J., and Valpuesta, J. (2001). The ‘sequential allosteric ring’ mechanism in the eukaryotic chaperonin-assisted folding of actin and tubulin. EMBO J. 20, 4065–4075. Martin, J., Langer, T., Boteva, R., Schramel, A., Horwich, A., and Hartl, F. (1991). Chaperonin-mediated protein folding at the surface of groEL through a ‘molten globule’-like intermediate. Nature 352, 36–42. Meyer, A.S., Gillespie, J.R., Walther, D., Millet, I.S., Doniach, S., and Frydman, J. (2003). Closing the folding chamber of the eukaryotic chaperonin requires the transition state of ATP hydrolysis. Cell 113, 369–381. Pereira, J.H., Ralston, C.Y., Douglas, N., Meyer, D., Knee, K.M., Goulet, D.R., King, J.A., Frydman, J., and Adams, P.D. (2010). Crystal structures of a group II
252 Cell 144, 240–252, January 21, 2011 ª2011 Elsevier Inc.
Spiess, C., Meyer, A., Reissmann, S., and Frydman, J. (2004). Mechanism of the eukaryotic chaperonin: protein folding in the chamber of secrets. Trends Cell Biol. 14, 598–604. Spiess, C., Miller, E., McClellan, A., and Frydman, J. (2006). Identification of the TRiC/CCT substrate binding sites uncovers the function of subunit diversity in eukaryotic chaperonins. Mol. Cell 24, 25–37. Stuart, S.F., Leatherbarrow, R.J., and Willison, K.R. (2011). A two-step mechanism for the folding of actin by the yeast cytosolic chaperonin. J. Biol. Chem. 286, 178–184. Tang, Y.C., Chang, H.C., Roeben, A., Wischnewski, D., Wischnewski, N., Kerner, M.J., Hartl, F.U., and Hayer-Hartl, M. (2006). Structural features of the GroEL-GroES nano-cage required for rapid folding of encapsulated protein. Cell 125, 903–914. Thulasiraman, V., Yang, C., and Frydman, J. (1999). In vivo newly translated polypeptides are sequestered in a protected folding environment. EMBO J. 18, 85–95. Villebeck, L., Persson, M., Luan, S.L., Hammarstrom, P., Lindgren, M., and Jonsson, B.H. (2007). Conformational rearrangements of tail-less complex polypeptide 1 (TCP-1) ring complex (TRiC)-bound actin. Biochemistry 46, 5083–5093. Yam, A.Y., Xia, Y., Lin, H.T., Burlingame, A., Gerstein, M., and Frydman, J. (2008). Defining the TRiC/CCT interactome links chaperonin function to stabilization of newly made proteins with complex topologies. Nat. Struct. Mol. Biol. 15, 1255–1262. Zhang, J., Baker, M.L., Schroder, G.F., Douglas, N.R., Reissmann, S., Jakana, J., Dougherty, M., Fu, C.J., Levitt, M., Ludtke, S.J., et al. (2010). Mechanism of folding chamber closure in a group II chaperonin. Nature 463, 379–383.
RalB and the Exocyst Mediate the Cellular Starvation Response by Direct Activation of Autophagosome Assembly Brian O. Bodemann,1 Anthony Orvedahl,2 Tzuling Cheng,1 Rosalyn R. Ram,1 Yi-Hung Ou,1 Etienne Formstecher,4 Mekhala Maiti,1 C. Clayton Hazelett,5 Eric M. Wauson,3 Maria Balakireva,6 Jacques H. Camonis,6 Charles Yeaman,5 Beth Levine,2 and Michael A. White1,* 1Department
of Cell Biology of Internal Medicine and Microbiology and Howard Hughes Medical Institute 3Department of Pharmacology UT Southwestern Medical Center, Dallas, TX 75390-9039, USA 4Hybrigenics, Inc., 75014 Paris, France 5Department of Anatomy and Cell Biology, University of Iowa, Iowa City, IA 52242, USA 6Institut Curie, Inserm U-548, 75248 Paris, France *Correspondence:
[email protected] DOI 10.1016/j.cell.2010.12.018 2Department
SUMMARY
The study of macroautophagy in mammalian cells has described induction, vesicle nucleation, and membrane elongation complexes as key signaling intermediates driving autophagosome biogenesis. How these components are recruited to nascent autophagosomes is poorly understood, and although much is known about signaling mechanisms that restrain autophagy, the nature of positive inductive signals that can promote autophagy remain cryptic. We find that the Ras-like small G protein, RalB, is localized to nascent autophagosomes and is activated on nutrient deprivation. RalB and its effector Exo84 are required for nutrient starvation-induced autophagocytosis, and RalB activation is sufficient to promote autophagosome formation. Through direct binding to Exo84, RalB induces the assembly of catalytically active ULK1 and Beclin1-VPS34 complexes on the exocyst, which are required for isolation membrane formation and maturation. Thus, RalB signaling is a primary adaptive response to nutrient limitation that directly engages autophagocytosis through mobilization of the core vesicle nucleation machinery. INTRODUCTION The critical role of macroautophagy (herein referred to as autophagy) in tissue homeostasis, cellular adaptation to nutrient restriction, and in clearance of pathogens and dysfunctional organelles suggests de novo generation of the doublemembrane autophagosome requires responsiveness to induc-
tive signals that specify location, contents, and duration (Kissova´ et al., 2004; Noda et al., 1995; Yang et al., 2006). A number of key signaling events have been identified that specify autophagosome biogenesis. Among the earliest is the dephosphorylation of inhibitory mTOR-dependent sites on the ULK1-Atg13FIP200 induction complex (Hosokawa et al., 2009a, 2009b). This presumably releases ULK1 activity to facilitate auto-phosphorylation of the ULK1-ATG13-FIP200 complex and assembly with the vertebrate-specific autophagy protein ATG101 (Hosokawa et al., 2009a, 2009b; Jung et al., 2009; Mercer et al., 2009). Through currently undescribed mechanisms, this leads to the activation of an autophagy specific class III PI(3)K complex, the Beclin1-ATG14L-VPS34-VPS15 complex. This activity coats a cup-shaped isolation membrane with phosphatidylinositol-3-phosphate, PI(3)P, which serves as a recruitment signal for the ATG16-ATG5/ATG12 component of the isolation membrane elongation machinery (Suzuki et al., 2001). Two ubiquitin-like molecules, ATG12 and LC3, undergo conjugation to ATG5 and phosphatidylethanolamine respectively to promote autophagosome formation. ATG12 is activated by ATG7 (E1), transferred to ATG10 (E2), followed by covalent linkage to an internal lysine on ATG5 (Mizushima et al., 1998a, 1998b). In the second conjugation system, LC3 is first cleaved by the cysteine protease, ATG4, which exposes a C-terminal glycine residue. ATG7 (E1) activates LC3 and transfers it to ATG3 (E2) (Amar et al., 2006; Tanida et al., 2002, 2004). LC3 is then conjugated to phosphatidylethanolamine with assistance of ATG5/12 conjugates (Fujita et al., 2008; Hanada et al., 2007; Kabeya et al., 2000). The lipidated LC3, LC3-II, coats the inner and outer surfaces of the autophagosome, and along with ATG5, serves as a discrete marker of autophagosomes and autophagosome precursors, respectively (George et al., 2000; Kabeya et al., 2000, 2004; Mizushima et al., 2001). These key signaling events are coordinated with dynamic membrane events to culminate in the formation of a double-membrane autophagosome. The autophagosome ultimately fuses with a lysosome that facilitates the Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc. 253
turnover of engulfed material by lysosomal/vacuolar acid hydrolases. How signaling intermediates are coordinated with the dynamic membrane events during the autophagosome biogenesis is currently unknown. RalA and RalB are close relatives to the founding members of the Ras GTPase superfamily. They are engaged in response to mitogenic, trophic, and hormonal signals by a diverse group of guanyl nucleotide exchange factors that fall into two major groups: those that are directly Ras-responsive via a carboxy terminal Ras binding domain and those that are apparently mobilized by phosphoinositide second messengers via a carboxy terminal Pleckstrin homology domain (Bodemann and White, 2008; Feig, 2003). Although a number of RalGTP effector proteins have been identified that couple RalA/B activation to dynamic cell biological processes, an overarching occupation of the Ral GTPases is the direct regulation of the Sec6/8, or exocyst, complex (Bodemann and White, 2008; Feig, 2003). Two members of the heterooctomeric exocyst complex, Sec5 (EXOC2) and Exo84 (EXOC8), are bona fide effector molecules that mediate RalA/B regulation of dynamic secretory vesicle targeting and tethering processes (Bodemann and White, 2008; Moskalenko et al., 2002, 2003; Sugihara et al., 2002). RalA-dependent mobilization of exocyst holocomplex assembly is critical for maintenance of apical/basolateral membrane identity in polarized epithelial cells (Moskalenko et al., 2002, 2003) and for insulin-stimulated Glut4 delivery to the plasma membrane in adipocytes (Chen et al., 2007). Distinct from regulation of membrane trafficking, RalB has been demonstrated to mediate signal transduction cascades supporting the host defense response. On Toll-like receptor activation, RalB/Sec5 complex assembly directly participates in activation of the innate immune signaling kinase TBK1 to facilitate an interferon response (Chien et al., 2006). This combination of roles, vesicle trafficking/tethering and signal cascade assembly/activation, suggests that Ral/exocyst effector complexes may coordinate dynamic membrane trafficking events with stimulus-dependent signaling events. Here we show that the small G protein, RalB, and an Exo84dependent subcomplex of the exocyst are critical for nutrient starvation and pathogen-induced autophagosome formation. Native RalB proteins localize to sites of nascent autophagosome formation and accumulate in the ‘‘active’’ GTP-bound state under nutrient limited conditions. RalB, but not its close homolog RalA, is required for autophagosome biogenesis and is sufficient to activate autophagy in human epithelial cells. The mechanism of action is through direct triggering of vesicle nucleation by assembly of an active ULK1-Beclin1-VPS34 initiation complex on the RalB effector protein Exo84. Thus, the RalB-Exo84 effector complex defines a key proximal regulatory component of the cellular response to nutrient deprivation. RESULTS Association of the Exocyst with Autophagosome Assembly Machinery Accumulating observations indicate direct participation of the heterooctomeric exocyst (aka Sec6/8) complex in adaptive responses to pathogen challenge (Bhuvanakantham et al., 254 Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc.
2010; Chien et al., 2006; Ishikawa and Barber, 2008; Ishikawa et al., 2009). Most strikingly, core innate immune signaling through TBK1 and STING is supported by the Sec5 subunit of the exocyst (Chien et al., 2006; Ishikawa and Barber, 2008; Ishikawa et al., 2009). To help generate molecular leads that may account for the participation of exocyst components in host defense signaling, we used high throughput yeast twohybrid screening to isolate a cohort of proteins that can associate with exocyst subunits (Formstecher et al., 2005). Among this cohort, both negative (RUBICON) and positive (FIP200, ATG14L) modulators of autophagy were isolated in the firstdegree interaction neighborhood of Sec3 (see Experimental Procedures). Given the functional convergence of Ral/exocyst signaling and autophagy in pathogen recognition and clearance, we examined the association of exocyst components and autophagy proteins in human epithelial cell cultures. The interaction of Sec3 with RUBICON and ATG14L was validated by expression co-IP (Figures 1A and 1B). In addition Exo84 and Sec5 could interact with RUBICON and ATG14L, as would be expected if autophagosome machinery/exocyst interactions occur in the context of multisubunit exocyst complexes (Figures 1C–1F). Immunoprecipitation of the core exocyst subunit, Sec8, recovers all characterized components of the exocyst complex (Grindstaff et al., 1998). Therefore, to examine if the exocyst may be associated with the LC3-modification machinery that drives elongation of isolation membranes, we probed Sec8 complexes for the presence of ATG5/ATG12 conjugates. As shown, Sec8-ATG5/ATG12 complexes were recovered from both overexpression co-IPs (Figure 1G) and by coimmunoprecipitation of endogenous proteins (Figure 1H), indicating a physical integration of the exocyst and autophagosome assembly machinery. RalB Signaling Is Required and Sufficient for Induction of Autophagosome Formation Mobilization of exocyst assembly in response to regulatory inputs is a major occupation of the Ras-like GTPases RalA and RalB (Balakireva et al., 2006; Cascone et al., 2008; Chen et al., 2007; Chien et al., 2006; Frische et al., 2007; Hase et al., 2009; Jin et al., 2005; Lalli and Hall, 2005; Moskalenko et al., 2002, 2003; Rosse´ et al., 2006; Spiczka and Yeaman, 2008; Sugihara et al., 2002). To examine the potential participation of Ral GTPase signaling in the regulation of autophagy, we first tested the consequence of blocking Ral-GTP/effector interactions on amino acid starvation-induced autophagosome accumulation and on isolation membrane encapsulation of bacterial pathogens. Expression of the minimal Ral-binding domain of the Ral effector RalBP1/RLIP76 (RLIP[RBD]) is dominant inhibitory to the action of endogenous RalA and RalB proteins through direct competition with Ral effector molecules (Chien et al., 2006; Moskalenko et al., 2002). As previously demonstrated (Fass et al., 2006; Pattingre et al., 2005), serum and amino-acid starvation of HeLa cells with Earle’s basic salt solution (EBSS) induced relocalization of endogenous LC3 protein from a diffuse cytosolic distribution to a condensed punctate pattern, and significantly decreased the total LC3 signal; consistent with starvation-induced autophagosome formation and maturation. RLIP (RBD) expression blocked both LC3 punctae formation and LC3
turnover (Figure 1I). As a surrogate measure for autophagic flux, we quantitated the total endogenous LC3 signal of individual cells after amino acid starvation, and found that RLIP(RBD) expression inhibited LC3 protein turnover in a dose-dependent fashion (Figure 1K). To investigate the contribution of Ral signaling to pathogen-responsive LC3 modification of membranes, mRFP-LC3 expressing HeLa cells were infected with GFP-labeled Salmonellae typhimurium. As expected if Ral signaling supports this response, RLIP(RBD) expression blocked recruitment of LC3 to internalized Salmonellae (Figure 1J). In addition, we found that ectopic expression of RalB was sufficient to induce the accumulation of LC3 punctae in cervical cancer cells (Figures 1L and 1M) and in immortalized bronchial epithelial cells (Figured 1N and 1O) in the absence of amino acid starvation or pathogen exposure. Remarkably, RalB(G23V) expression in nutrient rich conditions was sufficient to induce an accumulation of LC3 punctae that was 4–5-fold higher than that induced by amino acid deprivation (Figure 1O). This accumulation is likely associated with increased autophagic flux as RalB(G23V)induced LC3 punctae were further increased by chloroquinemediated inhibition of autophagosome turnover (Figure 1O; p = 0.011, Student’s t test), and this correlated with accumulation of phosphatidylethanolamine-conjugated LC3 (Figure 1P). Thus Ral signaling appears to be necessary and sufficient to engage autophagy. Evaluation of interactions between Ral signaling and autophagy in animals was carried out in Drosophila dRal hypomorphs (Ral35d), which have a weak loss-of-bristle phenotype associated with post-mitotic cell-specific apoptosis (Balakireva et al., 2006). Depletion of ATG14L, ATG1 (ULK1), ATG8a (LC3), ATG6 (Beclin), or VPS34, by in vivo expression of corresponding dsRNA, significantly enhanced the Ral35d phenotype (see Table S1 available online). To directly investigate the individual contributions of human Ral GTPases and exocyst proteins to regulated autophagosome biogenesis, we next tested the consequence of siRNA-mediated RalA, RalB, and exocyst subunit depletion on nutrient starvationinduced autophagy. Depletion of RalA, in a stable GFP-LC3 expressing cell line, had no consequence on GFP-LC3 signal accumulation or punctae formation. In contrast, RalB depletion significantly impaired starvation-induced LC3 punctae formation and LC3 turnover (Figure 2A). The extent of autophagosome inhibition observed on RalB depletion, as monitored by quantitation of GFP-LC3 punctae and total GFP-LC3 signal intensity, was equivalent to that seen on depletion of the known components of autophagosome formation, ATG5 or Beclin1 (Figures 2B and 2C). An equivalent analysis of the exocyst subunits Sec8, Sec5, and Exo84 indicated selective contributions of exocyst components to presumed autophagosome formation. Depletion of Sec8, a central exocyst subunit, had equivalent consequences as depletion of RalB, ATG5, or Beclin1. In contrast, among the two Ral effectors in the exocyst, Sec5 and Exo84, only Exo84 depletion impaired starvation-induced LC3 punctae formation and increased LC3 accumulation (Figures 2B and 2C). Evaluation of each of the eight exocyst subunits suggested that in addition to Sec8 and Exo84, Sec3 and Exo70 are limiting for support of autophagocytosis (Figure 2D and Figure S1A). The selective requirement for Exo84 versus Sec5 indicates that RalB regulation of autophagy is likely independent of the previ-
ously characterized RalB/Sec5/TBK1 signaling pathway (Chien et al., 2006). Additional observations supporting RalB-selective support of autophagosome formation were made through examination of starvation-induced changes in endogenous LC3 localization and LC3 posttranslational modification (Figures 2E–2G). These combined observations indicate that RalB and discrete components of the exocyst are required for autophagosome formation in multiple biological contexts. Activation of endogenous Ral GTPases may also be sufficient to induce autophagy, as depletion of endogenous RalGAP in the absence of nutrient limitation was sufficient to activate RalB and induce autophagic flux (Figures S1B–S1E). RalB Is Recruited to Sites of Nascent Autophagosome Formation To investigate the physical proximity of RalB to autophagosome formation, we examined the subcellular localization of endogenous RalB and components of the autophagosome initiation and elongation machinery. In telomerase and CDK4-immortalized normal human airway epithelial cells (HBEC30-KT), we noticed conspicuous colocalization of endogenous Beclin1 and endogenous RalB in perinuclear structures. On amino acid starvation (EBSS for 90 min), we observed redistribution of both RalB and Beclin1 to vesicular structures throughout the cell body (Figure 3A). The majority of these RalB positive structures co-labeled with a GFP-2X-FYVE reporter that localizes to sites of PI-(3)-P enrichment (Gillooly et al., 2000), the product of the Beclin1-associated class III PI3K VPS34 (Figure 3B). In addition, we found marked colocalization of endogenous RalB with GFP-ATG5 after a 90 min incubation in starvation media (Figure 3C). By 4 hr, GFP-LC3 punctae had accumulated, many of which were RalB positive (Figure 3D). To investigate the recruitment of RalB to a discrete membrane site undergoing LC3-modification, we utilized GFP-expressing S. typhimurium as a detectable, proximal signal for LC3-modification of the vacuole. Three hours after postinfection antibiotic selection to remove extracellular Salmonellae, we found that endogenous ATG5 was present along the surface of internalized GFP-Salmonellae, which colocalized with RalB (Figure 3E). Finally, an autophagic response of HBEC cells to Sendai virus exposure induced a redistribution of RalB but not RalA to cytosolic vesicular structures and promoted accumulation of endogenous RalBATG5/ATG12 protein complexes (Figures 3F and 3G). Nutrient Starvation and RalB Drive Assembly of Exo84-Beclin1 Complexes Given that Beclin1, a central regulatory node engaged to initiate autophagic responses to diverse stimuli, colocalized with RalB, we examined the relationship between Beclin1 and exocyst subunits. We found that nutrient starvation induced a dramatic assembly of Exo84/Beclin1 complexes in HEK293 cells (Figure 4A). In stark contrast, abundant Sec5/Beclin1 complexes present under nutrient-rich growth conditions were disassembled within 90 min of nutrient deprivation (Figure 4B), which could be reversed by addition of nonessential amino acids (Figure 4C). Sec8/Beclin1 complexes, on the other hand, were present under both nutrient-rich and nutrient-poor growth conditions (Figure 4D). Analysis of Beclin1 deletion constructs Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc. 255
B + -
GFP-Sec3: Flag-ATG14L:
+ +
+ -
Flag-ATG14L IP Flag-RUBICON
GFP-Sec3
GFP-Sec3
GFP-Sec3
GFP-Sec3
Flag-RUBICON IP WCL
C
+ +
Flag-ATG14L
WCL
+
+
Flag-ATG14L:
Myc-Exo84:
-
+
HA-Sec5:
+ -
+ Flag-ATG14L
HA-Sec5 IP
Vector Only
EBSS 4 hours
Rlip (RBD)
WCL
HA-Sec5
Sec8 IP
Myc IP ATG5/ATG12 Exo70 Sec5
Myc-Sec8
WCL
GFP-ATG5
Salmonella (GFP)
L
GFP
mRFP-LC3
EBSS
1.5 1010
GFP-ATG5 R2 = 0.7722
5.0 1009 0
p < 0.0001
80 60 40 20 0
1.5 1010
on
GFP-RalB
m
RF
P-
LC
3
Total Rlip(RBD) signal / cell
al
1.0 1010
Merge
M
mRFP-LC3
mRFP-LC3 punctae / cell
Total LC3 signal / cell
K
5.0 1009
HA-Sec5 Flag-RUBICON
GFP-ATG5
anti-LC3
DMEM 4 hours
0
Flag-RUBICON
Myc-Sec8
J Rlip(RBD)
1.0 1010
+ +
HA-Sec5
I
2.0
+ -
WCL
Myc-Exo84
+ +
Myc-Sec8 IP
Flag-ATG14L
WCL
Myc-Exo84
1010
Myc-Exo84
Flag-RUBICON: HA-Sec5: HA-Sec5 IP
H + -
HA-Sec5
Flag-ATG14L
WCL
Flag-RUBICON
WCL
GFP-ATG5: Myc-Sec8:
+
Flag-ATG14L Myc-Exo84
+ +
G
F Flag-ATG14L: Myc-Exo84 IP
+ -
Flag-RUBICON
Flag-ATG14L
Flag-RUBICON
E
D Flag-RUBICON: Myc-Exo84: Myc-Exo84 IP
GF P- e A FP tg5 -R al B
GFP-Sec3: Flag-RUBICON:
G
A
O
N
anti-LC3
50μM CQ
P 80
RalB(23V)
LC3 punctae / cell
vo
50μM CQ: Flag-RalB(23V):
Vehicle 50 M CQ
+ -
+ +
60
LC3(I) LC3(II)
40
Flag-RalB(23V) β-actin
20
0
Growth Media: EBSS: RalB(23V): 50μM CQ:
+ -
+ -
+ + -
+ +
+ +
+ + +
Figure 1. Physical and Functional Interaction of the Ral-Exocyst Complex with Autophagy Machinery (A–G) Exocyst subunits interact with autophagy proteins. The indicated proteins were overexpressed in HEK293 cells, then immunoprecipitated with an antibody directed to the specified tag. Immunoprecipitates were analyzed for coprecipitation with (A and B) GFP-Sec3; (C and D) Flag-RUBICON; (E and F) Flag-ATG14L; and (G) GFP-ATG5 as indicated. Whole-cell lysate: WCL, immunoprecipitation: IP. (H) Endogenous Sec8 complexes contain ATG5-12 conjugates. The endogenous exocyst complex was immunoprecipitated from HEK293 cells with anti-Sec8 antibody and analyzed for coprecipitation of ATG5/ATG12 conjugates (Sec8 IP) using anti-ATG5 antibody. Anti-Myc immunoprecipitates served as a negative
256 Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc.
indicated that both Exo84 and Sec5 required the amino-terminal BCL2-interacting domain for interaction with Beclin1 (88–150), whereas the evolutionarily conserved domain (244–337) was dispensable (Figure S2A). However, Exo84 and Sec5 likely have distinct binding determinants within the BCL2-interacting domain, as Beclin1(F123A), which fails to bind BCL2, interferes with Sec5 but not Exo84 association (Figures 4E and 4F). Importantly, we found that immunoprecipitation of endogenous Beclin1 from nutrient-deprived versus nutrient replete cells resulted in selective coprecipitation of endogenous Exo84 under starvation conditions (Figure 4G). These observations suggest that Beclin1 is recruited to distinct exocyst subcomplexes in response to nutrient availability. Previous observations from our group indicated that discrete macromolecular Exo84 and Sec5 complexes can be detected by density gradient centrifugation in pheochromocytoma cells (Moskalenko et al., 2003). Accordingly, we found that endogenous Exo84 and Sec5 display very distinct localization patterns in epithelial cells (Figure 4H). Likewise, ectopic expression of Exo84 or Sec5 was sufficient to differentially recruit Beclin1 to these distinct subcellular compartments (Figured 4I and 4J). The Exo84 compartment and the Sec5 compartment were reminiscent of the staining patterns observed with endogenous Beclin1 in the nutrient starved versus fed states respectively (Figure 3A). Size exclusion chromatography of cleared lysates, from proliferating cells, indicated the bulk of endogenous Exo84 and Sec5 eluted in separate fractions of 500 kDa and >700 kDa respectively, both of which partially cofractionated with Beclin1. Intriguingly, the ATG1 ortholog, ULK1, displayed a bimodal distribution presumably representative of distinct high and low molecular weight complexes (Figure S2B). Consistent with a sentinel role in the cellular response to nutrient deprivation, RalB was activated by nutrient deprivation as indicated by accumulation of the GTP-bound conformation. RalA, on the other hand, was unaffected (Figure 5A). Accordingly, expression of a constitutively active RalB variant (RalB
[23V]) was sufficient to induce Beclin1/Exo84 complex formation in the absence of nutrient deprivation (Figure 5B). To examine whether a direct RalB-Exo84 effector interaction was necessary for RalB to drive Exo84-Beclin1 association, we employed partial loss of function RalB variants selectively uncoupled from Exo84 versus Sec5 (Cascone et al., 2008; Jin et al., 2005). RalB (G23V,A48W) has a 43-fold higher affinity for Sec5 versus Exo84, and RalB(G23V,E38R) has a 104-fold higher affinity for Exo84 versus Sec5 (Jin et al., 2005). As shown, RalB (G23V,E38R) was considerably more effective at promoting Exo84/Beclin1 complex formation as compared to its Exo84binding defective counterpart, RalB(G23V,A48W) (Figure 5C). In contrast to Exo84, Beclin1/Sec5 complexes can be isolated under nutrient rich conditions. Interestingly, inhibition of Ral signaling by RLIP(RBD) expression eliminates the accumulation of Beclin1/Sec5 complexes suggesting that RalA/B signaling under nutrient rich conditions is required for this interaction (Figure 5D). A point mutation of the Ral-binding domain of Sec5 (T11A), which abolishes binding to Ral-GTP, also abolished the Sec5-Beclin1 interaction (Figure 5E). No interaction of Beclin1 with RLIP76, an exocyst-independent Ral effector, was observed (Figure 5F), indicating that Ral family modulation of Sec5/Beclin1 and Exo84/Beclin1 complexes is specific. The sufficiency of RalB interactions to drive both Sec5-Beclin1 and Exo84-Beclin1 complexes, coupled with our observations of the selective responsiveness of these complexes to nutrient status, suggested the possibility that nutrient availability results in distinct RalB-effector coupling. Indeed, endogenous RalB preferentially associated with Exo84 in nutrient poor conditions and Sec5 under nutrient rich conditions (Figures 5G and 5H). Importantly, endogenous ULK1, a key kinase that promotes initiation of autophagy (Chan et al., 2007; Hosokawa et al., 2009a; Jung et al., 2009), was selectively enriched in RalB immunoprecipitates on nutrient depletion (Figure 5G). These observations indicate that direct RalB/exocyst effector interactions differentially deliver Beclin1 to Sec5- or Exo84-containing
control (Myc IP). Two independent experiments are shown. Representation of the examined proteins in the input whole-cell nondenaturing lysates is shown (WCL). (I) Inhibition of Ral signaling blocks the LC3 response to amino-acid starvation. Forty-eight hours posttransfection with Myc-Rlip(RBD) HeLa cells were incubated in DMEM or EBSS for an additional 4 hr as indicated. Myc-Rlip(RBD) and LC3 were detected by immunofluorescence using anti-myc and anti-LC3 antibodies, respectively. Vector control cells were similar to untransfected cells. Scale bar represents 20 mm. (J) Inhibition of Ral signaling blocks the LC3 response to pathogen infection. HeLa cells were transfected with monomeric RFP-LC3 together with Myc-Rlip(RBD) or an empty vector control as indicated. Forty-eight hours posttransfection, cells were infected with Salmonella typhimurium-GFP for 1 hr followed by 3 hr of postinfection selection for intracellular Salmonella. Internalized Salmonella and LC3 were visualized using their respective fluorescent fusions. High magnification of the subcellular regions indicated by the boxes are shown in the panels on the right. Dashed lines indicated cell borders as visualized in a saturated exposure. Scale bar represents 10 mm. (K) Total fluorescence intensity corresponding to Myc-Rlip(RBD) (anti-myc) and endogenous LC3 (anti-LC3) at single-cell resolution for EBSS-treated cells as shown in (I) (n = 82, R2 = 0.7722). (L) RalB is sufficient to induce accumulation of LC3 punctae. HeLa cells expressing monomeric RFP-LC3 together with GFP-ATG5 or GFP-RalB are shown as indicated. Scale bar represents 10 mm. (M) mRFP-LC3 punctae in cells treated as in (L) were quantitated. The distribution of mRFP-LC3 punctae/cell is displayed as box-and-whisker plots. The three bands of the box illustrate the 25th (lower), 50th (middle), and 75th (upper) quartiles. The whiskers go 1.5 times the interquartile distance or to the highest or lowest point, whichever is shorter. The + designates the mean. P-values were calculated using the Student’s t test. (N) HBEC3-KT cells expressing RalB(23V) or transfected with vector control were incubated in growth medium containing 50 mM Chloroquine (CQ), to prevent LC3 turnover by autophagolysosomes, for 4 hr followed by detection of endogenous LC3 with anti-LC3 antibody. (O) RalB is sufficient to induce autophagic flux. HBEC3-KT cells treated as in (N) were incubated in growth media or amino-acid free Earle’s balanced salt solution (EBSS) for 4 hr with or without 50 mM Chloroquine (CQ), to prevent LC3 turnover in autophagolysosomes, as indicated. Immunofluorescence was performed with anti-LC3 antibody and LC3 punctae were quantitated. Data are represented as mean ± standard error of the mean (SEM). (P) Whole-cell lysates from HBEC3-KT cells transfected with Flag-RalB(G23V) or vector control were analyzed for the relative accumulation of LC3(I) and LC3(II) when incubated in growth medium containing 50 mM CQ for 4 hr. b-actin is shown as a loading control. See also Table S1.
Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc. 257
B Hela GFP-LC3
Figure 2. RalB and an Exo84-Containing Subcomplex of the Exocyst Are Necessary for Amino Acid Starvation-Induced Autophagy
ns
60
GFP-LC3 punctae / cell
A
ns
40 *
**
***
***
20
***
***
tro
C
siRalA + siRalB
Mean total GFP intensity
si
si
C
on
siRalA
siRalB
si AT l G Be 5 cl in si 1 R al si si A R R al al A+ B R si alB Ex o8 si 4 Se c si 5 Se c8
0
siControl
60 ***
40
***
***
***
***
*** ns
20
D
ns
on **
si
*
C
*** ns
si
60
E
40 ***
***
20
***
***
siC on tro l siA TG 5 siR alA siR alB
GFP-LC3 punctae / cell
t si r o l AT G Be 5 cl in si 1 R al si si A R R al al A+ B R si alB Ex o8 si 4 Se c si 5 Se c8
0
0
β-actin
si C on tr si ol Se si c3 Se si c5 Se si c6 Se si c8 Se c1 0 si Se c1 5 si Ex o7 0 si Ex o8 4
F
+ -
GFP-LC3(I) GFP-LC3(II)
Endogenous LC3
G 15
siControl
siATG5
LC3 punctae / cell
ns
10
5 ***
***
***
siRalA
si
si
C
on
tro AT l G si 5 R al A si siR R al alB A+ R al B
0
siRalB
subcomplexes in response to nutrient availability and that RalB/ Exo84 complexes may specify activation of autophagosome formation. Consistent with this, we found that inhibition of Ral signaling by Rlip(RBD), which inhibits autophagy, promoted the association of the autophagy inhibitor protein RUBICON with Exo84 (Figure 5I). The RalB/Exo84 Effector Pathway Mobilizes VPS34 Activity To further probe the relationship of Exo84 versus Sec5 complexes to mobilization of autophagosomes, we examined the consequence of nutrient depletion or Ral activation on recruitment of VPS34 to exocyst/Beclin1 complexes. Beclin1 258 Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc.
- Control siRNA + Target siRNA ATG5/ATG12 RalA RalB
(A) RalB depletion inhibits accumulation of GFPLC3 punctae. HeLa cells stably expressing GFPLC3 were depleted of the indicated proteins by siRNA transfection. Cells were imaged by GFP fluorescence 96 hr after transfection. Scale bar represents 10 mm. (B) Sec5 and Exo84 selectively participate in accumulation of GFP-LC3 punctae. GFP-LC3 punctae in cells treated as in (A) were quantitated. The mean distribution of GFP-LC3 punctae/cell is displayed as a bar graph, data are represented as mean ± SEM. P-values were calculated by oneway ANOVA followed by Dunnett’s multiple comparison test. (C) Inhibition of GFP-LC3 punctae correlates with accumulation of LC3 protein. The mean total intensity of GFP-LC3 in cells treated as in (A) was quantitated. The distribution of the mean total GFP intensity is displayed as a bar graph, data are represented as mean ± SEM. P-values were calculated by one-way ANOVA followed by Dunnett’s multiple comparison test. (D) A subset of exocyst subunits are limiting for accumulation of GFP-LC3 punctae. The indicated siRNAs were evaluated as in (B). (E) RalB depletion inhibits accumulation of LC3lipid conjugates. Whole-cell lysates from HBEC3KT cells stably expressing GFP-LC3 transfected with the indicated siRNAs were assayed for the relative accumulation of GFP-LC3(I) and GFP-LC3 (II). b-actin is shown as a loading control. siRNAmediated target depletion is shown 96 hr post transfection (right). (F) RalB participates in accumulation of endogenous LC3 punctae. HeLa cells were depleted of the indicated proteins by siRNA transfection. 96 hr after transfection, cells were incubated in amino acid free EBSS for 4 hr. Endogenous LC3 was imaged by anti-LC3 immunofluorescence. Scale bar represents 10 mm. (G) Endogenous LC3 punctae in cells treated as in (F) were quantitated. The mean distribution of LC3 punctae/cell is displayed as a bar graph, data are represented as mean ± SEM. P-values were calculated by one-way ANOVA followed by Dunnett’s Multiple Comparison Test. See also Figure S1.
has been heavily implicated as a positive regulatory cofactor of the VPS34 lipid kinase, which is thought to be a biochemical trigger for initiation of autophagosome isolation membrane assembly and elongation (Vergne et al., 2009; Zeng et al., 2006). Like the Exo84/Beclin1 relationship, we found that nutrient depletion resulted in accumulation of Exo84/VPS34 complexes (Figure 6A). Again, as we had seen with Beclin1, the opposite relationship was observed with Sec5 (Figure 6B). Expression of active RalB in the absence of nutrient depletion mirrored these observations. Namely, RalB(G23V) drove assembly of Exo84/VPS34 complexes (Figure 6C) and drove disassembly of Sec5/VPS34 complexes in a manner dependent on direct RalB/effector interactions (Figure 6D).
A
Beclin1
RalB
Merge
GFP-ATG5
RalB
Merge
B
GFP 2X-Fyve
RalB
Merge
LC3-GFP
RalB
Merge
Growth Medium
EBSS
C
D
F
E
G SeV (5hr):
ATG5
RalB
Salmonella
-
+
SeV (5hr): + Sec8 ATG5/12
RalA
RalB RalB IP
RalB
GFP + ATG5
GFP + RalB
Merge
Figure 3. Native RalB Colocalizes with Autophagy Machinery (A) Beclin1 and RalB colocalize in cells pre and post induction of autophagy. Endogenous immunofluorescence of Beclin1 (anti-Beclin1) and RalB (anti-RalB) in HBEC30-KT cells incubated for 90 min in fresh growth medium or EBSS as indicated. Dashed line indicates cell outline. Scale bar represents 10 mm. (B–D) RalB colocalizes with early and late markers of autophagosome biogenesis. HBEC30-KT cells were transfected with (B) GFP-2X-Fyve; (C) GFP-ATG5; and (D) GFP-LC3. Cells were incubated in EBSS for (B) 30 min; (C) 90 min; or (D) 3 hr. GFP fluorescence and endogenous RalB (anti-RalB) immunofluorescence is shown. High magnification of 10 mm 3 10 mm regions indicated by the boxes are shown in the bottom panels. Scale bar represents 10 mm. (E) ATG5 and RalB are recruited to sites of incipient isolation membrane formation. Endogenous immunofluorescence of ATG5 (anti-ATG5) and RalB (anti-RalB) in HBEC30-KT cells infected with Salmonella typhimurium-GFP. Cells were exposed to S. typhimurium-GFP for 1 hr followed by 3 hr of postinfection antibiotic selection against extracellular Salmonella. Scale bar represents 2 mm. (F) SenV infection selectively alters the subcellular distribution of RalB versus RalA. Endogenous immunofluorescence of RalA (anti-RalA) and RalB (anti-RalB) in HBEC3-KT cells mock infected or infected with Sendai virus for 5 hr. Scale bar represents 10 mm. (G) SenV infection induces accumulation of endogenous RalB/ATG5-12 complexes. Endogenous RalB complexes were immunoprecipitated from mock infected or Sendai virus infected HBEC3-KT cells with anti-RalB antibodies and analyzed for coprecipitation of ATG5/ATG12 conjugates.
Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc. 259
A
B Flag-Beclin1: + + Myc-Exo84: + + DMEM: 4h EBSS: - 90m
+ + 4h Myc-Exo84
Myc-Exo84 IP
Flag-Beclin1 Myc-Exo84
WCL
+ + 4h
Myc-Sec8 IP
HA-Sec5
EBSS: Goat IgG: anti-Beclin1:
HA-Sec5 Flag-Beclin1
WCL
Beclin1 Exo84
IP
WCL
Myc-Exo84
Flag-Beclin1
F + + +
WCL
Flag-Beclin1(F123A): + HA-Sec5: + DMEM: + EBSS: -
+ + +
Myc-Exo84 Flag-Beclin1(F123A)
HA-Sec5 IP
HA-Sec5 Flag-Beclin1(F123A)
Myc-Exo84 Flag-Beclin1(F123A)
WCL
HA-Sec5 Flag-Beclin1(F123A)
Sec5
Exo84
IP
+ + -
HA-Sec5
H
+ +
Beclin1 Exo84
DMEM: Goat IgG: anti-Beclin1:
Flag-Beclin1
WCL
Flag-Beclin1
+ + -
HA-Sec5
HA-Sec5 IP
Flag-Beclin1
WCL
Flag-Beclin1(F123A): + Myc-Exo84: + DMEM: + EBSS: Myc-Exo84 Myc-Sec8 IP Flag-Beclin1 Myc-Sec8
WCL
Flag-Beclin1: + + + HA-Sec5: + + + DMEM: 90m - EBSS: - 90m EBSS+NEAA: - - 90m
+ + 4h
E Flag-Beclin1: + + Myc-Sec8: + + DMEM: 4h EBSS: - 90m
I
+ + + + 4h - 90m
Flag-Beclin1
D
G
C Flag-Beclin1: HA-Sec5: DMEM: EBSS: HA-Sec5 IP
Merge
+ +
Beclin1 Exo84 Beclin1 Exo84
Flag-Beclin1
Merge
J
HA-Sec5
Flag-Beclin1
Merge
Figure 4. Nutrient Deprivation Drives Assembly of Exo84/Beclin1 Complexes (A–D) Nutrient limitation induces Beclin1/Exo84 interactions and inhibits Beclin1/Sec5 interactions. Forty-eight hours posttransfection with tagged Beclin1 and exocyst expression constructs, HEK293 cells were incubated in DMEM, EBSS, or EBSS with 13 nonessential amino acids for 90 min or 4 hr as shown. The indicated proteins were then immunoprecipitated with antibodies directed to the specified tag. Immunoprecipitates were analyzed for coprecipitation with Flag-Beclin1. (E and F) Beclin1(F123A) mutant interacts with Exo84 but not Sec5. Coexpression, co-IPs with the indicated proteins were performed as in (A–D). (G) Endogenous Beclin1/Exo84 complexes accumulate in response to nutrient deprivation. Endogenous Beclin1 was immunoprecipitated from HEK293 cells incubated in EBSS (top) or DMEM (bottom) for 90 min and analyzed for coprecipitation of Exo84 (IP). Host species-matched nonspecific IgG immunoprecipitates served as negative controls. Representation of the examined proteins in the input whole-cell nondenaturing lysates is shown (WCL).
260 Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc.
The presence of ATG14L in complex with Beclin1 and VPS34 is thought to specify the participation of this complex in autophagy as opposed to other cell processes where VPS34 activity has been implicated (Itakura et al., 2008; Matsunaga et al., 2009; Sun et al., 2008; Zhong et al., 2009). We found that dynamic interactions of Exo84 and Sec5 with ATG14L were also remarkably similar to those observed with Beclin1. Specifically, under nutrient rich growth conditions, Exo84 association with ATG14L was induced by RalB(G23V) expression (Figure 6E) whereas preexisting Sec5/ATG14L complexes were inhibited in the presence of the dominant inhibitory peptide Rlip(RBD) (Figure 6F). Furthermore, RalB(E38R) but not RalB(A48W) expression was sufficient to drive accumulation of PI-(3)-P positive punctae, the product of active VPS34, as visualized by accumulation of the GFP-2X-FYVE probe (Figures 6G and 6H). Similar results were observed for the accumulation of GFPLC3 punctae (Figure 6I). These observations suggest that induction of autophagy proceeds through assembly of Beclin1ATG14L-VPS34 complexes on Exo84, whereas the interaction of these components with Sec5 may represent organization of inactive components in a pre-initiation complex and/or a signal termination complex. Importantly, the increased accumulation of GFP-LC3 observed on RalB(G23V) expression was reversed by coexpression of kinase dead mutant ULK1(K46N), suggesting that RalB may act upstream of ULK1 to promote autophagy (Figure 6I). Active ULK1 Assembles on Exo84 upon Induction of Autophagy ULK1 activation is the most apical positive inductive signal, among Atg proteins, yet identified for initiation of autophagy. We have found that RalB is activated on nutrient starvation and that this correlates with the induction of RalB/ULK1 complexes (Figures 5A and 5G, respectively). Remarkably, RalB(G23V) expression was sufficient to promote assembly of ULK1/Beclin1 complexes, which have not been described previously but which may represent a mechanistic link between ULK1 activation and the VPS34 vesicle nucleation complex (Figure 7A). Furthermore, either nutrient depletion or RalB(G23V) expression was sufficient to induce assembly of ULK1/Exo84 complexes (Figures 7B and 7F). Depletion of Exo84 eliminated the capacity of RalB(G23V) to induce ULK1/Beclin1 complex formation, indicating that Exo84 is required for this assembly event (Figure 7C). In contrast to ULK1/Exo84 interactions, we observed increased Sec5/ULK1 complex assembly when Ral signaling was blocked by expression of Rlip(RBD) (Figure 7D). Analysis of Beclin1 deletion constructs indicated that, unlike Exo84 and Sec5, ULK1 requires the evolutionarily conserved domain (aa 244–337) for Beclin1 association (Figure S3A). Importantly, although ULK1 was present in both Exo84 and Sec5 complexes under nutrient poor conditions (Figures 7E and 7F), only Exo84-associated ULK1 displayed significant catalytic activity (Figures 7F and 7G).
ULK1 and mTORC1 have been reported to inhibit each other by reciprocal phosphorylation (Hosokawa et al., 2009a; Lee et al., 2007). Consistent with catalytically active Exo84/ULK1 complexes, expression of RalB(G23V) and Exo84 was sufficient to inhibit base-line mTORC1 activity as observed by reduced accumulation of phospho-threonine 389 on p70S6K (Figure 7B). In contrast, expression of RalB(G23V) and Sec5 resulted in increased accumulation of phospho-p70S6K (Figure 7D). Of note, endogenous mTORC1 was present in Sec5 but not Exo84 immunoprecipitates (Figure S3B). The assembly of ULK1 with Exo84 and disassembly of Beclin1 from Sec5 are responsive to mTOR inhibition, but only on chronic exposure to rapamycin, suggesting this is a consequence of the indirect effects that mimic nutrient starvation (Figures S3C and S3D). These combined observations indicate that the RalB/Exo84 effector relationship engages autophagy through direct modulation of a ULK1/Beclin1 initiation complex. DISCUSSION Our findings are consistent with a model in which the Ras-like G protein RalB acts as a regulatory switch to promote autophagosome biogenesis, in response to inductive signaling events, by mobilizing assembly of ULK1/Beclin1/VPS34 autophagy initiation complexes (Figure 7H). RalB is activated during the autophagic response, is localized to sites of incipient autophagosome formation, and is necessary and sufficient for induction of autophagic flux. In response to RalB activation, the direct RalB effector Exo84 is engaged as an essential assembly platform for catalytically active autophagy induction (ULK1/FIP200) and vesicle initiation (Beclin1/ATG14L,VPS34) complexes. In all cases examined, we found that dynamic assembly of active autophagosome biogenesis machinery on Exo84 was coordinated with disassembly of this same machinery from Sec5. Both Exo84 and Sec5 are Ral family G protein effectors and subunits of the exocyst, a heterooctomeric secretory vesicle trafficking complex (Lipschutz and Mostov, 2002). Previous work has shown that distinct Exo84 and Sec5 subcomplexes are directly engaged by Ral signaling to mobilize exocyst holocomplex formation in support of the dynamic vesicle targeting and tethering events required for stimulus-dependent exocytosis (Moskalenko et al., 2002, 2003). Remarkably, the Sec5/autophagy protein disassembly event and the Exo84/autophagy protein assembly event, described here, both require direct interaction with active RalB proteins. Thus induction of autophagy through RalB activation triggers dynamic autophagy protein reassembly events centered on two independent exocyst subunits. The tethering of autophagosome biogenesis machinery to distinct exocyst subcomplexes may provide appropriate spatial and temporal resolution of localized autophagic triggers. Consistent with this, we find that Sec5 and Exo84 accumulate in discrete cellular compartments that segregate with localization of Beclin1
(H) Exo84 and Sec5 are enriched in distinct subcellular compartments. Endogenous immunofluorescence of Sec5 (anti-Sec5) and Exo84 (anti-Exo84) in MDCK cells. Scale bar represents 10 mm. (I and J) Exo84 and Sec5 can recruit Beclin1 to distinct subcellular compartments. HEK293 cells were transfected with (F) Flag-Beclin1 and Myc-Exo84; (G) FlagBeclin1 and HA-Sec5. Immunofluorescence of the indicated fusion tags was performed. High magnification of 10 mm 3 10 mm regions indicated by the boxes are shown in the bottom panels. Scale bar represents 10 mm. See also Figure S2.
Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc. 261
A
B
EBSS:
-
1
5
10 15
Flag-Beclin1: Myc-Exo84: Flag-RalB(23V): Rlip(RBD):
30 60 90 240 minutes
+ + + + + + - + - - + Myc-Exo84
RalB(GTP)
Flag-Beclin1
Myc-Exo84 IP
RalB
Sec8
RalA(GTP)
Flag-RalB(23V)
RalA
Flag-Beclin1
Normalized Ral(GTP) / Total Ral
WCL
Sec8 Flag-RalB(23V)
C
2.4
Flag-Beclin1: Myc-Exo84: Flag-RalB(23V):
RalB RalA
2.0
+ + -
+ + +
+ +
+ +
38R 48W
Myc-Exo84
Myc-Exo84 IP
1.6
Flag-Beclin1
1.2
Myc-Exo84 0.8 0
5
10
15
60
WCL
120 180 240
Flag-Beclin1 Flag-RalB(23V)
Time (minutes)
D
E
Flag-Beclin1: HA-Sec5: Flag-RalB(23V): Rlip(RBD):
+ + + + - + - -
+ + +
Flag-Beclin1: Myc-Sec5(T11A): Flag-RalB(23V): HA-Sec5
+ + -
+ + +
Sec8
Flag-Beclin1
Flag-RalB(23V)
Myc-Sec5(T11A)
Myc-RLIP76
+ -
+ Exo84 ULK1
RalB IP
ATG14L
Flag-Beclin1 Flag-RalB(23V)
I Flag-RalB: HA-Sec5: 90m DMEM: 90m EBSS: Flag-RalB IP
+ + + -
HA-Sec5
+ + -
+ + +
Myc-Exo84 IP
Flag-RalB HA-Sec5
WCL
WCL
Flag-RalB
RalB Normailzed IP / input:
Flag-RUBICON: Myc-Exo84: Rlip(RBD):
+ + +
UVRAG Sec5
WCL
Flag-RalB(23V)
H 90m DMEM: 90m EBSS:
Flag-Beclin1
Sec8
Sec8
Flag-RalB(23V)
G
+ + +
Myc-RLIP76
Flag-Beclin1
WCL
Sec8
+ + -
Myc-RLIP76 IP
Flag-Beclin1
Flag-RalB(23V)
WCL
Flag-Beclin1: Myc-RLIP76: Flag-RalB(23V):
Myc-Sec5(T11A)
Myc-Sec5 (T11A) IP
Flag-Beclin1
HA-Sec5 IP
F
Myc-Exo84 Flag-Rubicon Myc-Exo84 Flag-Rubicon
1.0 0.28
Figure 5. RalB Drives Assembly of Exo84/Beclin1 Complexes through Direct RalB-Exo84 Effector Binding (A) Amino-acid depletion activates RalB. Endogenous GTP-bound RalA and RalB were collected by GST-Sec5-RBD mediated affinity purification from HEK293 cells incubated in EBSS for the indicated times and visualized with specific anti-RalA and anti-RalB antibodies. The normalized GTP-loaded index for RalA and RalB was calculated as Ral(GTP)/Total Ral to generate the scatterplot. (B–F) RalB regulates Beclin1/exocyst subcomplex interactions. The indicated proteins were expressed in HEK293 cells and immunoprecipitated with antibodies directed to the appropriate tag. Immunoprecipitates were analyzed for coprecipitation with Flag-Beclin1, Flag RalB(23V), and endogenous Sec8 as shown. (G) Nutrient status specifies distinct endogenous RalB/effector interactions. Endogenous RalB was immunoprecipitated with anti-RalB antibody from HEK293 cells incubated in DMEM or EBSS for 90 min as indicated and analyzed for coprecipitation of Exo84, Sec5, ATG14L, UVRAG, and ULK1. (H) Forty-eight hours posttransfection, HEK293 cells were incubated in DMEM or EBSS for 90 min as indicated. Flag-RalB immunoprecipitates were examined for coprecipitation of HA-Sec5. The indicated normalized IP/input ratio was calculated by dividing immunoprecipitated Sec5 by total expressed Sec5, then normalizing the calculated values to DMEM condition. (I) Ral-inhibition induces accumulation of Exo84/Rubicon interactions. Coexpression, co-IPs with the indicated proteins were performed as in (B–F).
pre- (Sec5) and post- (Exo84) induction of autophagy. It will be of great interest to determine if these locations represent the source location and assembly sites, respectively, of the membrane proteins required for autophagosome isolation membrane construction. Known exocyst subunit-autonomous mechanisms specifying subcellular localization patterns include interactions with organelle-specific proteins and membraneselective phosphoinositides (He and Guo, 2009). 262 Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc.
Recent observations indicate that detection of conserved pathogen-associated molecular patterns (PAMPs) by Toll-like receptors will mobilize autophagy together with activation of canonical innate-immune pathway activation (Delgado et al., 2008; Shi and Kehrl, 2010). Within this context of host defense surveillance and response systems, we have previously shown that RalB can engage Sec5 to activate the innate immunity signaling kinase TBK1 and the subsequent IRF3 transcription
A
Flag-VPS34: Myc-Exo84: 90m DMEM: 90m EBSS: Myc-Exo84 IP
+ + + -
B
+ + + Myc-Exo84 Flag-VPS34
+ + + -
C
+ + +
+ + -
Flag-VPS34: Myc-Exo84: RalB(23V): Rlip(RBD): Myc-Exo84 IP
HA-Sec5 Flag-VPS34
+ + + -
+ + + Myc-Exo84 Flag-VPS34
HA-Sec5
WCL
Myc-Exo84
WCL
Flag-VPS34: HA-Sec5: 90m DMEM: 90m EBSS: HA-Sec5 IP
Myc-Exo84
Flag-VPS34
WCL
Flag-VPS34
Flag-VPS34
Flag-RalB(23V)
D
E Flag-VPS34: HA-Sec5: RalB(23V): HA-Sec5 IP
+ + -
+ + + + + + + 38R 48W
HA-Sec5 Flag-VPS34
F Flag-ATG14L: Myc-Exo84: RalB(23V): Rlip(RBD): Myc-Exo84 IP
HA-Sec5 WCL
Flag-VPS34 Flag-RalB
+ + -
+ + + -
+ + +
+ + -
Flag-ATG14L: HA-Sec5: RalB(23V): Rlip(RBD): Myc-Exo84 Flag-ATG14L
+ + + -
+ + + HA-Sec5
HA-Sec5 IP
Flag-ATG14L
Myc-Exo84
WCL
HA-Sec5
WCL Flag-ATG14L
G
Flag-ATG14L
H
p < 0.0001
GFP-2xFyve Punctae / cell
60
GFP-2X-Fyve
FLAG-RalB
40
20
)
R
48 B(
al R
al R
W
)
t)
38
w
B(
B( al R
ra nt
Flag-RalB(A48W)
U
Flag-RalB(E38R)
ns
fe
ct
ed
0
p-value:
GFP-LC3 Punctae / cell
I
< 0.0001 0.0005
< 0.0001
60
40
20
ve R ct al or B( 23 R onl a V) lB y +U (2 LK 3V) R 1 al B( (46 N 2 3 R V ) al B( ,38 23 R V, ) 48 W )
0
Figure 6. RalB Expression Drives Assembly of Exo84/Vps34 and Exo84/ATG14L Complexes (A–F) VPS34 and ATG14L/exocyst subcomplexes are regulated by nutrient limitation and RalB activation. HEK293 cells expressing the indicated proteins were incubated in DMEM or EBSS for 90 min as indicated. Tagged exocyst subunits were immunoprecipitated and analyzed for coprecipitation with Flag-VPS34 and Flag-ATG14L where indicated. (G) RalB/Exo84 effector interactions mobilize VPS34 activity. HeLa cells expressing GFP-2X-Fyve together with RalB partial loss of function mutants Flag-RalB (E38R) or Flag-RalB(A48W) are shown as indicated. (H) GFP-2X-Fyve punctae in cells treated as in (G) were quantitated. The distribution of GFP-2X-Fyve punctae/cell is displayed as box-and-whisker plots. P-values were calculated using the Student’s t test. (I) RalB(G23V) and RalB(G23V,E38R) are sufficient to induce accumulation of GFP-LC3 punctae and kinase dead ULK1(K46N) blocks the increase observed with RalB(G23V) expression. HeLa cells stably expressing GFP-LC3 were transfected with the indicated constructs then visualized by immunofluorescence of the indicated tags. The distribution of GFP-LC3 punctae/cell is displayed as box-and-whisker plots. P-values were calculated using the Student’s t test.
Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc. 263
B
A Flag-Beclin1: HA-ULK1: RalB(23V): HA-ULK1 IP
+ + -
C
HA-ULK1: Myc-Exo84: RalB(23V): Rlip(RBD):
+ + + HA-ULK1
+ + -
+ + + -
+ + + Myc-Exo84
Myc-Exo84 IP
Flag-Beclin1
Myc-Exo84
Flag-Beclin1 Flag-RalB
+ + + + -
+ + +
+ + + + HA-ULK1
HA-ULK1 IP
HA-ULK1
HA-ULK1 WCL
Flag-Beclin1: HA-ULK1: RalB(23V): siControl: siExo84:
Flag-Beclin1
HA-ULK1
WCL
Flag-Beclin1
p-p70S6K(T389)
WCL
Exo84 Flag-RalB
Flag-RalB(23V)
E
D HA-ULK1: Myc-Sec5: RalB(23V): Rlip(RBD):
+ + -
+ + + -
HA-ULK1: Myc-Sec5: 90m DMEM: 90m EBSS:
+ + + Myc-Sec5
Myc-Sec5 IP
+ + + -
+ + + Myc-Sec5
Myc-Sec5 IP
HA-ULK1
HA-ULK1
Myc-Sec5 HA-ULK1
WCL
Myc-Sec5
WCL
HA-ULK1
p-p70S6K(T389) Flag-RalB(23V) + + + -
+ + +
+ + -
+ +
+ + + + -
+ + + + -
+ + + +
+ + + +
G
+ + + + HA-ULK1 Myc-Sec5 Myc-Exo84 P Exo84
32
P MBP
75 50 25 0
H
Ex
o8
4
32
100
Se IP c5 IP
Myc-Sec5: Myc-Exo84: HA-ULK1: Myc IP: 90m DMEM: 90m EBSS:
normalized activity index
F
Nutrient Deprivation, Pathogen Invasion
RalB
GTP
GDP
RalGEFs GDP
Pi
RalGAPα1/β
RalB
PI PI-(3)-P
GTP PI
P PI-(3)-
mTORC1
P
P P
P P P
ULK1
ULK1 Beclin1
Autophagy Suppression
Sec5 Exocyst subcomplex
Exo84 Exocyst subcomplex VPS34
RalB
VPS15
GTP
ATG14L
Autophagy Induction
PI
VPS34
PI PI-(3)-P
PI-(3)-P
PI PI-(3)-P
VPS15 PI-(3)-P
PI
Nutrients
Figure 7. Active ULK1 Associates with Exo84 (A) RalB induces ULK1/Beclin1 Complex formation. ULK1 immunoprecipitates were analyzed for coprecipitation with Flag-Beclin1 on RalB(23V) expression as indicated. (B) ULK1/Exo84 complexes are regulated by RalB. The indicated proteins were expressed in HEK293 cells. Myc-tagged Exo84 was immunoprecipitated and analyzed for coprecipitation with HA-ULK1. (C) RalB-induced ULK1/Beclin1 complexes require Exo84. HEK293 cells were first transfected with siControl or siExo84 siRNAs before the indicated proteins were expressed 24 hr later. ULK1 immunoprecipitates were analyzed for coprecipitation with Flag-Beclin1 on RalB(23V) expression as indicated. (D) ULK1/Sec5 complexes accumulate on Ral inhibition. Coexpression, co-IPs with the indicated proteins were performed as in (B). (E) ULK1/Sec5 complexes dissociate on nutrient deprivation. Coexpression, co-IPs with the indicated proteins were performed as in (B) with the addition of 90 min incubation in DMEM or EBSS as indicated. (F) Amino-acid starvation induces association of Exo84 with catalytically active ULK1. Exo84 and Sec5 complexes were assayed for coprecipitation with ULK1 and for associated protein kinase activity as indicated. (G) The normalized activity ratio for EBSS stimulated Exo84 and Sec5 precipitates was calculated by the division of MBP 32P incorporation by the HA-ULK1 signal coprecipitated from (F). (H) Working model of RalB/exocyst dependent mobilization of autophagy. See also Figure S3.
264 Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc.
factor-dependent interferon response (Chien et al., 2006). Our observations here indicate that RalB can separately engage Exo84 to facilitate activation of the autophagy kinase ULK1 and induction of autophagosome biogenesis. Together, this suggests that RalB and the exocyst represent a regulatory hub, through bifurcating activation of TBK1 and Beclin1/ VPS34, that helps engage concomitant activation of the gene expression and organelle biogenesis responses supporting systemic pathogen recognition and clearance. The required coordination of such time and location-specified responses may account for the adaptation of exocyst function to support of signal transduction cascades in metazoans. The RalA and RalB G-proteins are signal propagation molecules coupled to mitogenic, trophic, and cytokine signaling systems (Bodemann and White, 2008; Feig, 2003). This connectivity potentially provides appropriate functional coupling of autophagic responses to diverse cellular milieus. Ral activation occurs through engagement of one or more of a family of 5 Ral-specific guanyl nucleotide exchange factors that can selectively couple to RalA or RalB through mechanisms that remain to be determined (Bodemann and White, 2008). Of note, RalA has recently been shown to promote mTORC1 activation, potentially through PLD1 and phosphatidic acid-dependent mTORC1/2 assembly (Maehama et al., 2008; Toschi et al., 2009; Voss et al., 1999). This regulatory relationship could be directly antagonistic to autophagosome biosynthesis given the capacity of mTORC1 to restrain ULK1 activity through direct inhibitory phosphorylation events (Hosokawa et al., 2009a; Jung et al., 2009). Whether these are independent or interconnected regulatory arms remains to be determined. However, the RalB GDP/GTP cycle and its effector relationships comprise a regulatory mechanism that can directly control dynamic transition between metabolic states supporting cell growth versus cell maintenance. EXPERIMENTAL PROCEDURES Plasmids and Antibodies Detailed information on plasmid and antibody origins, immunofluorescence protocols, image capture procedures, and image quantitation are described in the Extended Experimental Procedures.
Yeast Two-Hybrid The coding sequence for full-length human SEC3 (GenBank gi:7023219) was cloned into pB27 as a C-terminal fusion to LexA and used as a bait to screen at saturation a high-complexity random-primed human placenta cDNA library as previously described (Fromont-Racine et al., 1997).
Immunoprecipitation and Kinase Assays Immunoisolation of tagged or native proteins was performed using standard procedures from nondenaturing cell extracts (20 mM Tris-HCl pH 7.4, 137 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 10 mM MgCl2, 2 mM ethylene glycol tetraacetic acid [EGTA]). Kinase assays were carried out in (25 mM MOPS pH 7.5, 1 mM EGTA, 0.1 mM sodium vanadate, 15 mM MgCl2, 5 mM b-glycerol phosphate). See Extended Experimental Procedures for extended details.
S. typhimurium Infection Exposure to GFP-expressing S. typhimurium (obtained from Mary O’Riordan, University of Michigan) was performed as described (Radtke et al., 2007).
SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, three figures, and one table and can be found with this article online at doi:10.1016/j. cell.2010.12.018. ACKNOWLEDGMENTS We thank Noboru Mishuzima for GFP-Atg5 and GFP-LC3, Zhenyu Yue for Flag-ATG14L and Flag-RUBICON, and Noriko Okazaki for HA-ULK1 and HA-ULK1(K46N). We thank members of the White and Levine labs for productive discussions. We thank Melanie Cobb for helpful discussion and assistance with kinase assays. This work was supported by grants from the National Institutes of Health (CA71443 and CA129451 to M.W. and CA84254 and CA109618 to B.L.), ARC4845 and GenHomme Network Grant 02490-6088 to J.C., and the Welch Foundation (I-1414 to M.W.). B.B. was supported by DOD Award Number W81XWH-06-1-0749. R.R. was supported by T32GM008203. Y.O. was supported by CPRIT RP101496. Received: July 12, 2010 Revised: October 21, 2010 Accepted: December 6, 2010 Published: January 20, 2011 REFERENCES Amar, N., Lustig, G., Ichimura, Y., Ohsumi, Y., and Elazar, Z. (2006). Two newly identified sites in the ubiquitin-like protein Atg8 are essential for autophagy. EMBO Rep. 7, 635–642. Balakireva, M., Rosse´, C., Langevin, J., Chien, Y.C., Gho, M., Gonzy-Treboul, G., Voegeling-Lemaire, S., Aresta, S., Lepesant, J.A., Bellaiche, Y., et al. (2006). The Ral/exocyst effector complex counters c-Jun N-terminal kinasedependent apoptosis in Drosophila melanogaster. Mol. Cell. Biol. 26, 8953– 8963. Bhuvanakantham, R., Li, J., Tan, T.T., and Ng, M.L. (2010). Human Sec3 protein is a novel transcriptional and translational repressor of flavivirus. Cell. Microbiol. 12, 453–472. Bodemann, B.O., and White, M.A. (2008). Ral GTPases and cancer: linchpin support of the tumorigenic platform. Nat. Rev. Cancer 8, 133–140. Cascone, I., Selimoglu, R., Ozdemir, C., Del Nery, E., Yeaman, C., White, M., and Camonis, J. (2008). Distinct roles of RalA and RalB in the progression of cytokinesis are supported by distinct RalGEFs. EMBO J. 27, 2375–2387. Chan, E.Y., Kir, S., and Tooze, S.A. (2007). siRNA screening of the kinome identifies ULK1 as a multidomain modulator of autophagy. J. Biol. Chem. 282, 25464–25474. Chen, X.W., Leto, D., Chiang, S.H., Wang, Q., and Saltiel, A.R. (2007). Activation of RalA is required for insulin-stimulated Glut4 trafficking to the plasma membrane via the exocyst and the motor protein Myo1c. Dev. Cell 13, 391–404. Chien, Y., Kim, S., Bumeister, R., Loo, Y.M., Kwon, S.W., Johnson, C.L., Balakireva, M.G., Romeo, Y., Kopelovich, L., Gale, M., Jr., et al. (2006). RalB GTPase-mediated activation of the IkappaB family kinase TBK1 couples innate immune signaling to tumor cell survival. Cell 127, 157–170. Delgado, M.A., Elmaoued, R.A., Davis, A.S., Kyei, G., and Deretic, V. (2008). Toll-like receptors control autophagy. EMBO J. 27, 1110–1121. Fass, E., Shvets, E., Degani, I., Hirschberg, K., and Elazar, Z. (2006). Microtubules support production of starvation-induced autophagosomes but not their targeting and fusion with lysosomes. J. Biol. Chem. 281, 36303–36316. Feig, L.A. (2003). Ral-GTPases: approaching their 15 minutes of fame. Trends Cell Biol. 13, 419–425. Formstecher, E., Aresta, S., Collura, V., Hamburger, A., Meil, A., Trehin, A., Reverdy, C., Betin, V., Maire, S., Brun, C., et al. (2005). Protein interaction mapping: a Drosophila case study. Genome Res. 15, 376–384.
Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc. 265
Frische, E.W., Pellis-van Berkel, W., van Haaften, G., Cuppen, E., Plasterk, R.H., Tijsterman, M., Bos, J.L., and Zwartkruis, F.J. (2007). RAP-1 and the RAL-1/exocyst pathway coordinate hypodermal cell organization in Caenorhabditis elegans. EMBO J. 26, 5083–5092. Fromont-Racine, M., Rain, J.C., and Legrain, P. (1997). Toward a functional analysis of the yeast genome through exhaustive two-hybrid screens. Nat. Genet. 16, 277–282. Fujita, N., Itoh, T., Omori, H., Fukuda, M., Noda, T., and Yoshimori, T. (2008). The Atg16L complex specifies the site of LC3 lipidation for membrane biogenesis in autophagy. Mol. Biol. Cell 19, 2092–2100. George, M.D., Baba, M., Scott, S.V., Mizushima, N., Garrison, B.S., Ohsumi, Y., and Klionsky, D.J. (2000). Apg5p functions in the sequestration step in the cytoplasm-to-vacuole targeting and macroautophagy pathways. Mol. Biol. Cell 11, 969–982. Gillooly, D.J., Morrow, I.C., Lindsay, M., Gould, R., Bryant, N.J., Gaullier, J.M., Parton, R.G., and Stenmark, H. (2000). Localization of phosphatidylinositol 3-phosphate in yeast and mammalian cells. EMBO J. 19, 4577–4588. Grindstaff, K.K., Yeaman, C., Anandasabapathy, N., Hsu, S.C., RodriguezBoulan, E., Scheller, R.H., and Nelson, W.J. (1998). Sec6/8 complex is recruited to cell-cell contacts and specifies transport vesicle delivery to the basal-lateral membrane in epithelial cells. Cell 93, 731–740. Hanada, T., Noda, N.N., Satomi, Y., Ichimura, Y., Fujioka, Y., Takao, T., Inagaki, F., and Ohsumi, Y. (2007). The Atg12-Atg5 conjugate has a novel E3-like activity for protein lipidation in autophagy. J. Biol. Chem. 282, 37298–37302. Hase, K., Kimura, S., Takatsu, H., Ohmae, M., Kawano, S., Kitamura, H., Ito, M., Watarai, H., Hazelett, C.C., Yeaman, C., and Ohno, H. (2009). M-Sec promotes membrane nanotube formation by interacting with Ral and the exocyst complex. Nat. Cell Biol. 11, 1427–1432. He, B., and Guo, W. (2009). The exocyst complex in polarized exocytosis. Curr. Opin. Cell Biol. 21, 537–542. Hosokawa, N., Hara, T., Kaizuka, T., Kishi, C., Takamura, A., Miura, Y., Iemura, S., Natsume, T., Takehana, K., Yamada, N., et al. (2009a). Nutrient-dependent mTORC1 association with the ULK1-Atg13-FIP200 complex required for autophagy. Mol. Biol. Cell 20, 1981–1991. Hosokawa, N., Sasaki, T., Iemura, S., Natsume, T., Hara, T., and Mizushima, N. (2009b). Atg101, a novel mammalian autophagy protein interacting with Atg13. Autophagy 5, 973–979.
Lalli, G., and Hall, A. (2005). Ral GTPases regulate neurite branching through GAP-43 and the exocyst complex. J. Cell Biol. 171, 857–869. Lee, S.B., Kim, S., Lee, J., Park, J., Lee, G., Kim, Y., Kim, J.M., and Chung, J. (2007). ATG1, an autophagy regulator, inhibits cell growth by negatively regulating S6 kinase. EMBO Rep. 8, 360–365. Lipschutz, J.H., and Mostov, K.E. (2002). Exocytosis: the many masters of the exocyst. Curr. Biol. 12, R212–R214. Maehama, T., Tanaka, M., Nishina, H., Murakami, M., Kanaho, Y., and Hanada, K. (2008). RalA functions as an indispensable signal mediator for the nutrientsensing system. J. Biol. Chem. 283, 35053–35059. Matsunaga, K., Saitoh, T., Tabata, K., Omori, H., Satoh, T., Kurotori, N., Maejima, I., Shirahama-Noda, K., Ichimura, T., Isobe, T., et al. (2009). Two Beclin 1-binding proteins, Atg14L and Rubicon, reciprocally regulate autophagy at different stages. Nat. Cell Biol. 11, 385–396. Mercer, C.A., Kaliappan, A., and Dennis, P.B. (2009). A novel, human Atg13 binding protein, Atg101, interacts with ULK1 and is essential for macroautophagy. Autophagy 5, 649–662. Mizushima, N., Noda, T., Yoshimori, T., Tanaka, Y., Ishii, T., George, M.D., Klionsky, D.J., Ohsumi, M., and Ohsumi, Y. (1998a). A protein conjugation system essential for autophagy. Nature 395, 395–398. Mizushima, N., Sugita, H., Yoshimori, T., and Ohsumi, Y. (1998b). A new protein conjugation system in human. The counterpart of the yeast Apg12p conjugation system essential for autophagy. J. Biol. Chem. 273, 33889–33892. Mizushima, N., Yamamoto, A., Hatano, M., Kobayashi, Y., Kabeya, Y., Suzuki, K., Tokuhisa, T., Ohsumi, Y., and Yoshimori, T. (2001). Dissection of autophagosome formation using Apg5-deficient mouse embryonic stem cells. J. Cell Biol. 152, 657–668. Moskalenko, S., Henry, D.O., Rosse, C., Mirey, G., Camonis, J.H., and White, M.A. (2002). The exocyst is a Ral effector complex. Nat. Cell Biol. 4, 66–72. Moskalenko, S., Tong, C., Rosse, C., Mirey, G., Formstecher, E., Daviet, L., Camonis, J., and White, M.A. (2003). Ral GTPases regulate exocyst assembly through dual subunit interactions. J. Biol. Chem. 278, 51743–51748. Noda, T., Matsuura, A., Wada, Y., and Ohsumi, Y. (1995). Novel system for monitoring autophagy in the yeast Saccharomyces cerevisiae. Biochem. Biophys. Res. Commun. 210, 126–132.
Ishikawa, H., and Barber, G.N. (2008). STING is an endoplasmic reticulum adaptor that facilitates innate immune signalling. Nature 455, 674–678.
Okazaki, N., Yan, J., Yuasa, S., Ueno, T., Kominami, E., Masuho, Y., Koga, H., and Muramatsu, M. (2000). Interaction of the Unc-51-like kinase and microtubule-associated protein light chain 3 related proteins in the brain: possible role of vesicular transport in axonal elongation. Brain Res. Mol. Brain Res. 85, 1–12.
Ishikawa, H., Ma, Z., and Barber, G.N. (2009). STING regulates intracellular DNA-mediated, type I interferon-dependent innate immunity. Nature 461, 788–792.
Pattingre, S., Tassa, A., Qu, X., Garuti, R., Liang, X.H., Mizushima, N., Packer, M., Schneider, M.D., and Levine, B. (2005). Bcl-2 antiapoptotic proteins inhibit Beclin 1-dependent autophagy. Cell 122, 927–939.
Itakura, E., Kishi, C., Inoue, K., and Mizushima, N. (2008). Beclin 1 forms two distinct phosphatidylinositol 3-kinase complexes with mammalian Atg14 and UVRAG. Mol. Biol. Cell 19, 5360–5372.
Radtke, A.L., Delbridge, L.M., Balachandran, S., Barber, G.N., and O’Riordan, M.X. (2007). TBK1 protects vacuolar integrity during intracellular bacterial infection. PLoS Pathog. 3, e29.
Jin, R., Junutula, J.R., Matern, H.T., Ervin, K.E., Scheller, R.H., and Brunger, A.T. (2005). Exo84 and Sec5 are competitive regulatory Sec6/8 effectors to the RalA GTPase. EMBO J. 24, 2064–2074.
Rosse´, C., Hatzoglou, A., Parrini, M.C., White, M.A., Chavrier, P., and Camonis, J. (2006). RalB mobilizes the exocyst to drive cell migration. Mol. Cell. Biol. 26, 727–734.
Jung, C.H., Jun, C.B., Ro, S.H., Kim, Y.M., Otto, N.M., Cao, J., Kundu, M., and Kim, D.H. (2009). ULK-Atg13-FIP200 complexes mediate mTOR signaling to the autophagy machinery. Mol. Biol. Cell 20, 1992–2003.
Shi, C.S., and Kehrl, J.H. (2010). TRAF6 and A20 regulate lysine 63-linked ubiquitination of Beclin-1 to control TLR4-induced autophagy. Sci. Signal. 3, ra42.
Kabeya, Y., Mizushima, N., Ueno, T., Yamamoto, A., Kirisako, T., Noda, T., Kominami, E., Ohsumi, Y., and Yoshimori, T. (2000). LC3, a mammalian homologue of yeast Apg8p, is localized in autophagosome membranes after processing. EMBO J. 19, 5720–5728.
Spiczka, K.S., and Yeaman, C. (2008). Ral-regulated interaction between Sec5 and paxillin targets Exocyst to focal complexes during cell migration. J. Cell Sci. 121, 2880–2891.
Kabeya, Y., Mizushima, N., Yamamoto, A., Oshitani-Okamoto, S., Ohsumi, Y., and Yoshimori, T. (2004). LC3, GABARAP and GATE16 localize to autophagosomal membrane depending on form-II formation. J. Cell Sci. 117, 2805–2812. Kissova´, I., Deffieu, M., Manon, S., and Camougrand, N. (2004). Uth1p is involved in the autophagic degradation of mitochondria. J. Biol. Chem. 279, 39068–39074.
266 Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc.
Sugihara, K., Asano, S., Tanaka, K., Iwamatsu, A., Okawa, K., and Ohta, Y. (2002). The exocyst complex binds the small GTPase RalA to mediate filopodia formation. Nat. Cell Biol. 4, 73–78. Sun, Q., Fan, W., Chen, K., Ding, X., Chen, S., and Zhong, Q. (2008). Identification of Barkor as a mammalian autophagy-specific factor for Beclin 1 and class III phosphatidylinositol 3-kinase. Proc. Natl. Acad. Sci. USA 105, 19211–19216.
Suzuki, K., Kirisako, T., Kamada, Y., Mizushima, N., Noda, T., and Ohsumi, Y. (2001). The pre-autophagosomal structure organized by concerted functions of APG genes is essential for autophagosome formation. EMBO J. 20, 5971– 5981.
Vieira, O.V., Botelho, R.J., Rameh, L., Brachmann, S.M., Matsuo, T., Davidson, H.W., Schreiber, A., Backer, J.M., Cantley, L.C., and Grinstein, S. (2001). Distinct roles of class I and class III phosphatidylinositol 3-kinases in phagosome formation and maturation. J. Cell Biol. 155, 19–25.
Tanida, I., Tanida-Miyake, E., Komatsu, M., Ueno, T., and Kominami, E. (2002). Human Apg3p/Aut1p homologue is an authentic E2 enzyme for multiple substrates, GATE-16, GABARAP, and MAP-LC3, and facilitates the conjugation of hApg12p to hApg5p. J. Biol. Chem. 277, 13739–13744.
Voss, M., Weernink, P.A., Haupenthal, S., Mo¨ller, U., Cool, R.H., Bauer, B., Camonis, J.H., Jakobs, K.H., and Schmidt, M. (1999). Phospholipase D stimulation by receptor tyrosine kinases mediated by protein kinase C and a Ras/ Ral signaling cascade. J. Biol. Chem. 274, 34691–34698.
Tanida, I., Sou, Y.S., Ezaki, J., Minematsu-Ikeguchi, N., Ueno, T., and Kominami, E. (2004). HsAtg4B/HsApg4B/autophagin-1 cleaves the carboxyl termini of three human Atg8 homologues and delipidates microtubule-associated protein light chain 3- and GABAA receptor-associated protein-phospholipid conjugates. J. Biol. Chem. 279, 36268–36276. Toschi, A., Lee, E., Xu, L., Garcia, A., Gadir, N., and Foster, D.A. (2009). Regulation of mTORC1 and mTORC2 complex assembly by phosphatidic acid: competition with rapamycin. Mol. Cell. Biol. 29, 1411–1420. Vergne, I., Roberts, E., Elmaoued, R.A., Tosch, V., Delgado, M.A., ProikasCezanne, T., Laporte, J., and Deretic, V. (2009). Control of autophagy initiation by phosphoinositide 3-phosphatase Jumpy. EMBO J. 28, 2244–2258.
Yang, Z., Huang, J., Geng, J., Nair, U., and Klionsky, D.J. (2006). Atg22 recycles amino acids to link the degradative and recycling functions of autophagy. Mol. Biol. Cell 17, 5094–5104. Zeng, X., Overmeyer, J.H., and Maltese, W.A. (2006). Functional specificity of the mammalian Beclin-Vps34 PI 3-kinase complex in macroautophagy versus endocytosis and lysosomal enzyme trafficking. J. Cell Sci. 119, 259–270. Zhong, Y., Wang, Q.J., Li, X., Yan, Y., Backer, J.M., Chait, B.T., Heintz, N., and Yue, Z. (2009). Distinct regulation of autophagic activity by Atg14L and Rubicon associated with Beclin 1-phosphatidylinositol-3-kinase complex. Nat. Cell Biol. 11, 468–476.
Cell 144, 253–267, January 21, 2011 ª2011 Elsevier Inc. 267
Delay in Feedback Repression by Cryptochrome 1 Is Required for Circadian Clock Function Maki Ukai-Tadenuma,1,8 Rikuhiro G. Yamada,1,8 Haiyan Xu,4 Ju¨rgen A. Ripperger,5 Andrew C. Liu,4 and Hiroki R. Ueda1,2,3,6,7,* 1Laboratory
for Systems Biology, RIKEN Center for Developmental Biology Genomics Unit, RIKEN Center for Developmental Biology 3Laboratory for Synthetic Biology, RIKEN Quantitative Biology Center 2-2-3 Minatojima-minamimachi, Chuo-ku, Kobe, Hyogo 650-0047, Japan 4Department of Biological Sciences, The University of Memphis, Memphis, TN 38152-0001, USA 5Department of Medicine/Unit of Biochemistry, University of Fribourg, 5, rue du Muse ´ e, CH-1700 Fribourg, Switzerland 6Graduate School of Science, Osaka University, 1-1 Machikaneyama, Toyonaka, Osaka 560-0043, Japan 7Department of Mathematics, Graduate School of Science, Kyoto University, Kitashirakawa Oiwake-cho, Sakyo-ku, Kyoto, Kyoto 606-8502, Japan 8These authors contributed equally to this work *Correspondence:
[email protected] DOI 10.1016/j.cell.2010.12.019 2Functional
SUMMARY
Direct evidence for the requirement of delay in feedback repression in the mammalian circadian clock has been elusive. Cryptochrome 1 (Cry1), an essential clock component, displays evening-time expression and serves as a strong repressor at morning-time elements (E box/E0 box). In this study, we reveal that a combination of day-time elements (D box) within the Cry1-proximal promoter and night-time elements (RREs) within its intronic enhancer gives rise to evening-time expression. A synthetic composite promoter produced eveningtime expression, which was further recapitulated by a simple phase-vector model. Of note, coordination of day-time with night-time elements can modulate the extent of phase delay. A genetic complementation assay in Cry1/:Cry2/ cells revealed that substantial delay of Cry1 expression is required to restore circadian rhythmicity, and its prolonged delay slows circadian oscillation. Taken together, our data suggest that phase delay in Cry1 transcription is required for mammalian clock function. INTRODUCTION Circadian clocks are thought to consist of autoregulatory loops with delayed transcriptional/translational feedback repression in which delayed expression of clock components is critical for maintaining circadian rhythmicity (Dunlap, 1999; Reppert and Weaver, 2002; Young and Kay, 2001). However, the underlying molecular mechanism giving rise to such delay remains 268 Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc.
unknown, hindering formal validation of its biological relevance. In mammalian clocks, circadian transcriptional program is mediated through at least three clock-controlled DNA elements, morning-time (E box/E0 box, or E/E0 box: CACGT[G/T]) (Gekakis, 1998; Hogenesch et al., 1997; Ueda et al., 2005; Yoo et al., 2005), day-time (D box: TTA[T/C]GTAA) (Falvey et al., 1996; Ueda et al., 2005), and night-time elements (Rev-Erb/ROR-binding element, or RRE: [A/T]A[A/T]NT[A/G]GGTCA) (Harding and Lazar, 1993; Preitner et al., 2002; Ueda et al., 2002, 2005). The E/E0 box-mediated transcriptional program has a critical role in the core autoregulatory loop of the mammalian circadian clock (Gekakis, 1998; Sato et al., 2006; Ueda et al., 2005). In this core loop, bHLH-PAS transcription activators such as BMAL1 and CLOCK form heterodimers that bind to E/E0 box cis-elements in the promoter regions of their target genes, including the Per and Cry genes; CRYs, in turn, form repressor complexes that physically associate with the BMAL1/CLOCK complex to inhibit E/E0 box-mediated transcription (Dunlap, 1999; Griffin et al., 1999; Kume et al., 1999; Reppert and Weaver, 2002; Young and Kay, 2001). Thus, the CRYs play an integral role in the circadian clock by ‘‘closing’’ the core negative feedback loop. Although Cry1/ mice and their SCN slices display circadian rhythms at the organismal and tissue levels, respectively, dissociated Cry1/ SCN neurons and fibroblasts are largely arrhythmic. The issue of cell autonomy has been carefully examined in several recent studies (Brown et al., 2005; DeBruyne et al., 2007; Liu et al., 2007). By contrast to Cry1-deficient cells, dissociated Cry2/ SCN neurons and fibroblasts exhibit robust rhythmicity, implying that CRY2 cannot substitute for Cry1 deficiency at the cellular level (Liu et al., 2007). Therefore, we focused on transcriptional regulation of Cry1 gene. CRY1 and its expression pattern play a pivotal role in the core autoregulatory loop. Either overexpression of CRY1 or interference of CRY1’s repressor activity on E/E0 box-mediated
transcription can abolish circadian transcriptional oscillations (Sato et al., 2006; Ueda et al., 2005). Remarkably, Cry1 displays delayed gene expression relative to other genes with E/E0 box elements (Ueda et al., 2002, 2005). Circadian expression of Cry1 peaks at evening phases in the SCN (CT12) (Ueda et al., 2002, 2005), which is much later than for typical morning-time E/E0 box-regulated genes such as Rev-Erba, and is intermediate between day-time D box- and night-time RRE-regulated genes such as Per3 and Bmal1, respectively (Ueda et al., 2002, 2005). Dual roles of Cry1 as a strong repressor for E/E0 box activity and a time delay mediator fit well with the current model of the circadian clock, i.e., feedback repression with delay may depend on the unique mode of transcriptional regulation of Cry1. Previous studies identified an E0 box and an E box in Cry1’s regulatory region (Ueda et al., 2005; Fustin et al., 2009) and two RREs in its first intron (Ueda et al., 2005). In this study, we also identified additional D boxes in the promoter region and confirmed their functionality in conferring day-time expression (delayed phase relative to E box). We further discovered that a combination of the promoter containing E/E0 boxes and D boxes with the first intron sequence of Cry1 containing RREs generated delayed-phase expression of Cry1, in which the strength of nighttime elements (RREs) can modulate the extent of phase delay. Of note, a simple phase vector model predicts that coordination between day-time and night-time elements can determine the extent of phase delay. Based on this model, we generated an array of Cry1 constructs that display different phases, and these constructs were used in a genetic complementation assay to restore circadian oscillation in Cry1/:Cry2/ cells. These experiments reveal that substantial delay of Cry1 expression is required to restore single-cell level rhythmicity and that prolonged delay of Cry1 expression can slow circadian oscillation. These results suggest that phase delay in transcriptional feedback repression is required for mammalian clock function. RESULTS Cry1 Promoter Confers Phase and Amplitude Intermediate between Those Conferred by E/E0 Box and D Box Circadian Elements To examine Cry1 promoter activity, we generated a reporter construct, P(Cry1)-Luc, in which a 1.5 kbp DNA fragment containing the Cry1 promoter was fused to the Luciferase (Luc) gene. Cry1 promoter-driven bioluminescence reached its peak at circadian time (CT) 9.60 ± 0.11 (n = 3, mean ± standard deviation), which was rather close to that of a D box-P(SV40)-Luc reporter harboring three tandem repeats of D boxes fused to an SV40 promoter (CT10.42 ± 0.16), and delayed > 6 hr relative to an E0 box-P(SV40)-Luc reporter harboring three tandem repeats of E0 boxes fused to an SV40 promoter (CT3.53 ± 0.04) (Figure 1A and Figure S1A and Table S1 available online). On the other hand, the Cry1 promoter produced a higher-amplitude rhythm than that of the D box-P(SV40)-Luc reporter (Figure 1B). The amplitude of E0 box-driven bioluminescence rhythms (Figure 1B, E0 box-P(SV40)-Luc) was even higher than those driven by the Cry1 promoter. These data place the Cry1 promoter intermediate between D box and E0 box in both phase and amplitude
of driven rhythms and suggest that the Cry1 promoter might contain both D box and E0 box elements. D Box in Cry1 Promoter Confers Phase Delay and Day-Time Expression We investigated the genomic sequences of the Cry1 promoter and found five highly conserved regions, of which two sequences (50 -TTCAGAAA-30 and 50 -AAACGTGA-30 ) most closely resemble a D box according to position weight matrix analysis. Interestingly, these sequences overlap with the conserved E0 box sequences in the promoter region (Figure 1C). We designated this region of the Cry1 promoter as a Cry1proD element and constructed a Cry1proD-P(SV40)-Luc reporter by fusing three tandem repeats of this element to an SV40 promoter. NIH 3T3 cells transiently transfected with this construct showed circadian oscillation of bioluminescence with a peak at day-time (CT12.37 ± 0.05, n = 3; Figure 1D, Figure S1B, and Table S1) and a relative amplitude between those of E0 box and D box constructs (Figure 1E). Because the E0 box and two putative D boxes in Cry1proD element overlap, it is not practical to isolate each CCE for analysis individually. Instead, we tested whether clock factors involved in E0 box- or D box-mediated transcription could activate or repress the Cry1proD element. Cotransfection of E0 box activators BMAL1/CLOCK strongly induced not only E0 box, but also Cry1proD activity in NIH 3T3 cells (Figure 1F). Cotransfection of a D box repressor E4BP4 inhibited BMAL1/CLOCK induction of Cry1proD activity in a dose-dependent manner (Figure 1F). Interestingly, the E0 box repressor CRY1 also inhibited this induction (Figure S1C), and the D box activators DBP, HLF, and TEF (Mitsui et al., 2001) also induced Cry1proD activity (Figure S1D). These results suggest that Cry1proD is regulated by classical transcriptional regulators of both D box and E/E0 box activities, consistent with the observation that bioluminescence rhythms driven by the Cry1 promoter display properties intermediate between those driven by D box and E0 box constructs. To confirm that Cry1 expression is delayed relative to E/E0 box activity, we measured temporal mRNA profiles of endogenous Per2, Bmal1, and Cry1 in NIH 3T3 cells expressing a P(Cry1)Luc or a P(Per2)-Luc reporter (Figure 1G, left). As shown, the phase of Luc mRNA driven by the Per2 promoter was almost the same as that of endogenous Per2 mRNA (Figure 1G, right). Furthermore, the phase of Luc mRNA driven by the Cry1 promoter was delayed relative to those of endogenous Per2 mRNA or Luc mRNA driven by the Per2 promoter, with a phase difference of 1–2 hr. These results further support the notion that functional D boxes in the Cry1 promoter contribute to phase delay of Cry1 expression. Cry1 Intron Acts as an Enhancer to Confer Phase Delay In addition to the phase delay caused by D boxes within the Cry1 promoter, we also observed further phase delay of the endogenous Cry1 mRNA by at least 2 hr relative to Luc mRNA driven by the Cry1 promoter (Figure 1G). The endogenous Cry1 mRNA displayed 3–4 hr phase delay relative to endogenous Per2 mRNA and 7–8 hr advance relative to Bmal1 mRNA. This observation is consistent with previous reports (Baggs et al., 2009; Etchegaray et al., 2003; Liu et al., 2008; Preitner Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc. 269
4
5 2.5
5
P(Cry1)
0 Normalized Bioluminescence Low High -1 0 1
E'-box-P(SV40)
3.41 10.38
D-box-P(SV40) 0
1
2
Day
3
4
5
6 3
Cry1proD -P(SV40)
Cry1proD-mut -P(SV40)
P(SV40)
Cry1 promoter
-93 ~ -70
+328
mouse : AAACACACTTCAGAAACGTGAGGT
0.4
************************
human : AAACACACTTCAGAAACGTGAGGT D-box consensus:
TTATGTAA TTACGTAA AACGTG
E'-box consensus: 0
1
2
3
4
Day
P(Cry1)-Luc / NIH3T3
4
10000000
3
7500000
2
5000000
1
2500000
0
0
1
Day
2
3
0
18
12
6
0
Per2
Bmal1 Cry1
Luc
endogenous RNA
P(Per2)-Luc / NIH3T3
3
6000000
2
4000000
1
2000000
0
0
0
1
Day
2
3
Bioluminescence (counts/min)
- + - +++ - + - +++ - + - +++ - + - +++ - + - +++ --+ --+ --+ --+ --+ D-P(SV40)
4
Cry1proD -P(SV40) E'-P(SV40) D-P(SV40)
mRNA/Gapdh (average=1)
Relative Luc Activity
12
E'-P(SV40)
3
Cry1proD element
G
15
0 BMAL1/CLOCK E4BP4
2
Bioluminescence (counts/min)
F
1
Day
5
Normalized Bioluminescence Low High -1 0 1
-1208 0
10
0
+328
mammal conservation
0.8 Phase (h)
12.37
Cry1 exon1
(TSS) -1
0.2
E
15
Phase (h)
Cry1proD-P(SV40)
Cry1proD -P(SV40)
D
0.4
chr10 -1208
Phase (h)
Day
3
D-P(SV40)
2
D-P(SV40)
1
E'-P(SV40)
0
Relative Amplitude
10.42
D-box-P(SV40)
7.5
Relative Amplitude
3.53
P(Cry1) E'-P(SV40) D-P(SV40)
mRNA/Gapdh (average=1)
E'-box-P(SV40)
C
0.6
10 Phase (h)
9.60
B
12.5
Phase (h) P(Cry1)
E'-P(SV40)
A
P(Cry1)-Luc P(Per2)-Luc
Cry1 mRNA Per2 mRNA Bmal1 mRNA Luc mRNA LUC activity
Figure 1. The Cry1 Promoter Contains Both E/E0 Box and D Boxes
(A) The phases of circadian transcriptional activities induced by P(Cry1), E0 box-P(SV40), and D box-P(SV40) promoters. Each promoter was fused with a Luciferase reporter gene (Luc) and transiently transfected into NIH 3T3 cells. Time series of bioluminescence expression were recorded in real time using a photomultiplier tube (PMT). Heatmaps represent average promoter activities from three independent samples. Raw data were detrended for baseline and amplitude and then scaled into a range of 1 to 1 (left). Phases were estimated from the time series data by fitting a cosine wave (right). (B) The relative amplitudes of circadian transcriptional activities induced by P(Cry1), E0 box-P(SV40), and D box-P(SV40) promoters. (C) E0 box and its overlapping D boxes in Cry1 promoter. Genomic positions relative to the transcription start site (TSS, designated also as ‘‘1’’) of Cry1 gene are indicated along with evolutionary conservation scores among mammalian species. Colored letters indicate nucleotides matching the consensus sequence of D box and E0 box. (D) The phases of circadian transcriptional activities induced by Cry1proD-P(SV40), E0 box-P(SV40), and D box-P(SV40) promoters. The experiment was performed as in (A). (E) The relative amplitudes of circadian transcriptional activities induced by Cry1proD-P(SV40), E0 box-P(SV40), and D box-P(SV40) promoters. (F) The Cry1proD element can be activated by BMAL1/CLOCK and repressed by E4BP4. Each of the indicated promoters was fused to a Luciferase reporter gene and transiently transfected into NIH 3T3 cells. Relative Luciferase activity for each promoter was scaled so that the activity without transcriptional regulation is normalized to 1. (A–F) Data are representative of two independent experiments. Error bars represent SD determined from three measurements for each sample (n = 3). (G) mRNA expression patterns of endogenous Cry1, Per2, and Bmal1 and exogenous Luciferase (Luc). NIH 3T3 cells were transfected with P(Cry1)-Luc or P(Per2)-Luc reporter. Relative mRNA levels of each gene were measured. In parallel, transcriptional activities of P(Cry1) and P(Per2) promoters were monitored by bioluminescence recording. Phases of measured rhythm are indicated on the right. (Right) Error bars represent SEM (n = 3). (Left) Error bars represent SD (n = 3). See also Figure S1 and Table S1.
et al., 2002; Sato et al., 2006; Ueda et al., 2005), which have speculated that a phase delay would be generated by two functional RREs present in the intron regions of Cry1 (Ueda et al., 2005). To provide experimental evidence for the mechanism of further phase delay of Cry1 expression, we focused on one of the highly conserved regions of the Cry1 gene—the first intron, 270 Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc.
which contains two RREs, designated here as R1 and R2 (Figure 2A). These RREs are highly conserved and aligned in a head-to-head arrangement, perfectly matched to the consensus RRE sequence ([A/T]A[A/T]NT[A/G]GGTCA). We cloned a 1.03 kbp fragment containing the conserved intronic RREs and inserted it into the P(Cry1)-Luc reporter plasmid
to generate a P(Cry1)-Cry1 intron 1.03k-Luc reporter. Cells expressing this reporter displayed a bioluminescence peak at CT14.62 ± 0.20, whereas absence of the 1.03 kbp intron sequence resulted in a peak at CT10.51 ± 0.30, a difference of 4 hr (Figure 2B, Figure S2A, and Table S1). We next focused on a highly conserved region of 336 bp within the 1.03 kbp intron sequence for further analysis. Whereas cells expressing a P(Cry1)-Cry1 intron D336-Luc reporter exhibited a peak at CT10.37 ± 0.24, those expressing P(Cry1)-Cry1 intron 336-Luc peaked at CT14.32 ± 0.23 (Figure 2B, Figure S2A, and Table S1), a 4 hr phase delay. Thus, Cry1 first intron sequences containing RREs likely underlie the delayed phase of Cry1 expression. In addition, their effects appear to be independent of locations (Figure 2C, Figure S2B, and Table S1), suggesting that this sequence functions as a transcriptional enhancer. Next, we analyzed the regulatory regions for the presence of their corresponding transcription factors in vivo using chromatin immunoprecipitation (ChIP) assays with time series samples from mouse liver. Chromatin from wild-type, Dbp/ (Lopez-Molina et al., 1997), or Rev-Erba/ (Preitner et al., 2002) mice was immunoprecipitated by anti-BMAL1, anti-DBP, or anti-REV-ERBa antibodies (Figure 2D). The levels of BMAL1 and DBP binding to Cry1proD displayed circadian oscillation in wild-type, Dbp/, or Rev-Erba/ mice, whereas DBP binding in Dbp/ mice was significantly reduced (p < 0.01 by two-way ANOVA) with residual signals potentially deriving from TEF and/or HLF binding to the same element. On the other hand, no significant reduction was observed for the binding of DBP to this region in Rev-Erba/ mice. The level of REV-ERBa binding to the Cry1 first intron region also displayed circadian oscillation in wild-type and Dbp/ mice, whereas it was significantly reduced in Rev-Erba/ mice (p < 0.01 by two-way ANOVA). The levels of BMAL1 binding to the Dbp promoter region displayed circadian oscillation, whereas there was only background binding of DBP and REV-ERBa to this region. This result confirmed that BMAL1 and DBP bind to the Cry1 promoter region, and REV-ERBa binds to the Cry1 first intron region. The peak binding time of each transfactor is consistent with previous reports of its in vivo binding or its nuclear accumulation (Lopez-Molina et al., 1997; Mitsui et al., 2001; Preitner et al., 2002; Ripperger and Schibler, 2006). In addition to the biochemical interaction between the Cry1 promoter and D box trans-regulators described above, we also examined the role of the D box using genetic approaches; we measured mRNA expression patterns from time course liver samples of triple-knockout mice of PAR bZip genes (Tef, Hlf, and Dbp) (Gachon et al., 2004). Although these mice displayed normal circadian behavior (possibly due to compensation rendered by posttranslational mechanisms intracellularly and/or intercellular coupling of clock cells in vivo) (Gachon et al., 2004; Lee et al., 2001; Liu et al., 2007), we found that Cry1’s circadian expression level was different from wild-type and its peak of expression delayed (Figure 2E and Figure S2C), whereas those of other measured clock genes (Bmal1, RevErba, and Per1) were not. Importantly, the observed peak delay was reproducible and significant in three independent experiments (p < 0.01 by two-way ANOVA). These results further confirm that PAR bZip genes are important for the proper phase of expression of Cry1.
Strength of Intronic RREs Correlates with Phase Delay To determine whether the strength of RREs in 336 bp of Cry1 first intron sequence correlates with the phase delay, we generated an array of intron sequences harboring mutant RREs, including deletion, mutation, and inversion of the two RREs (Figure 3A). We inserted these mutant intron sequences into P(SV40)-Luc vector to generate an array of P(SV40)-Cry1 intron 336-Luc reporter constructs. As one measurement for the strength of intronic RREs, we first examined transcriptional activation of these constructs by RORa, an activator of RRE, in a reporter assay. We found that induced Luciferase activities varied significantly among constructs, ranging from strong induction by wild-type RRE to almost no induction by double-mutant or deleted RREs (R1 and R2) (Figure 3B). These results indicate that the RREs within the intron sequence are functionally responsive to RORa and that intronic RREs of various strengths can be obtained from different RRE mutations. As an independent measurement for the strength of intronic RREs, we next examined the amplitude of circadian oscillations expressed by these constructs in reporter rhythm assays (Figure S3A and Table S1). Rhythm amplitude was low when an intron sequence of low RORa responsiveness was used to drive reporter expression and high when an intron sequence of high RORa responsiveness was used (Figures 3B and 3C). Overall, there was a significant positive correlation between the two measurements for the strength of intronic RREs: RORa responsiveness and rhythm amplitude among the intronic RRE mutants (r2 = 0.95, p < 0.01; Figure 3D). These mutant intron sequences allowed us to analyze quantitatively the role of intronic RREs in the phase delay mechanism. Specifically, we examined how intronic RRE mutation affects phase delay using a reporter rhythm assay (Figure 3E, Figure S3B, and Table S1). We found that the observed phase delay significantly correlated with the first measurement for the strength of the intronic RRE mutants, i.e., RORa responsiveness (r2 = 0.82, p < 0.01; Figure 3F, top). Similarly, phase delay also correlated well with the second measurement for the strength of the intronic RRE mutants, i.e., the rhythm amplitude (r2 = 0.90, p < 0.01; Figure 3F, bottom). Taken together, these data suggest that the strength of RREs correlates with the phase delay, further corroborating our finding that the RREs in the Cry1 intron act as an enhancer to further delay the phase conferred by Cry1 promoter. Combination of Day- and Night-Time Elements Produces Evening Phase Control Given that the delayed expression of Cry1 is a combined effect of its promoter and intron, we sought to understand whether this combinatorial effect is a general design principle in the circadian transcriptional network or a mechanism unique to the transcriptional regulation of Cry1. We first asked whether the phase of endogenous Cry1 expression could be synthesized using an artificial promoter in clock cells. We constructed three sets of reporters, with each harboring one of the three CCEs (i.e., E/E0 box, D box, and RRE) in the presence or absence of the RRE-containing intron sequence from the Cry1 gene (Figures 4A and 4B). Real-time bioluminescence recording of transfected NIH 3T3 cells showed that the RRE-containing Cry1 intron sequence, as expected, did not dramatically alter the phase of Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc. 271
-1 (TSS)
A
23.8kb
chr10
Cry1 gene region mammal conservation
isolated 1.03kb region
isolated 336 base region
R1 R2
Flag Cry1 intron
P(Cry1)
Luc
polyA
Phase (h)
B P(Cry1)
10.51
P(Cry1)-Cry1 intron 1.03k
14.62
336
P(Cry1)-Cry1 intron 336
14.32
336
P(Cry1)-Cry1 intron Δ336
10.37 0
1
2
3
4
Normalized Bioluminescence Low High
5
-1
Day
0
1
Phase (h)
C no Cry1 intron
10.04
P(Cry1)
Flag
upstream of P(Cry1)
12.22
downstream of P(Cry1)
14.94
P(Cry1)
336
inside of the coding sequence
13.91
P(Cry1)
Flag
12.61
P(Cry1)
downstream of the coding sequence 0
1
2
3
4
336
P(Cry1)
Luc
polyA
Flag
Luc
polyA
Flag
Luc
polyA
Luc
polyA
336
Flag
Luc
336
polyA
5
Day
D
percent of input
2.5
anti-BMAL1
anti-DBP
Cry1 promoter
Cry1 promoter
0.4
2
0.15
0.3 0.2
0.1
0.1
0.05
1 0.5
0
0 0 4 8 12 16 20
0 4 8 12 16 20
Dbp-/-
wild type
0 4 8 12 16 20
Rev-Erbα-/-
0 4 8 12 16 20
Dbp-/-
wild type
Dbp promoter
0 4 8 12 16 20
0 4 8 12 16 20
Rev-Erbα-/-
wild type
2
0 4 8 12 16 20
0 4 8 12 16 20
Dbp-/-
Dbp promoter
2.5 percent of input
Cry1 1st intron
0.2
1.5
0 ZT(h) 0 4 8 12 16 20
Rev-Erbα-/-
Dbp promoter
0.4
0.2
0.3
0.15
1.5 0.2
0.1
0.1
0.05
1 0.5
0 ZT(h) 0 4 8 12 16 20
0 4 8 12 16 20
0 4 8 12 16 20
Dbp-/-
Rev-Erbα-/-
wild type
E
anti-REV-ERBα
Cry1 0.015
0
0 0 4 8 12 16 20
0 4 8 12 16 20
0 4 8 12 16 20
0 4 8 12 16 20
0 4 8 12 16 20
0 4 8 12 16 20
wild type
Dbp-/-
Rev-Erbα-/-
wild type
Dbp-/-
Rev-Erbα-/-
Bmal1
0.005
0.015
Rev-Er bα
Per 1
0.009
mRNA/Gapdh
0.004 0.01
0.003 0.002
0.005
0.01
0.006
0.005
0.003
0.001 0 ZT(h) 0 4 8 12 16 20
wild type
0 4 8 12 16 20
Dbp-/-:Tef-/-:Hlf-/-
0
0 4 8 12 16 20
0 4 8 12 16 20
wild type
Dbp-/-:Tef-/-:Hlf-/-
0
0 4 8 12 16 20
0 4 8 12 16 20
wild type
Dbp-/-:Tef-/-:Hlf-/-
0
0 4 8 12 16 20
0 4 8 12 16 20
wild type
Dbp-/-:Tef-/-:Hlf-/-
Figure 2. Cry1 Intron Acts as an Enhancer to Confer Phase Delay (A) The first intron of Cry1 contains RRE sequences. The marked 1.03 kbp and 336 bp of Cry1 first intron sequence, which are highly conserved in mammals, were cloned and examined in this study. Two RREs are indicated as R1 and R2, respectively. (B) Cry1’s first intron confers phase delay. The Cry1 promoter was combined with the Cry1 1.03 kbp intron, 336 bp intron, or Cry1 intron D336 deletion mutant to generate composite promoters. The experiment was performed as in Figure 1A. (C) Cry1’s first intron sequence confers phase delay independently of its location. The 336 bp of Cry1 intron sequence was inserted upstream or downstream of the Cry1 promoter and inside or downstream of the coding sequence. Data are representative of two independent experiments (B and C). (D) Binding of BMAL1 (an E/E0 box regulator) and DBP (a D box regulator) to the Cry1 promoter region and REV-ERBa (a RRE regulator) to the Cry1 first intron region in vivo. Chromatin from wild-type (gray), Dbp/ (orange), or Rev-Erba/ (purple) mice was prepared at 4 hr intervals from mice held in a 12 hr light/12 hr dark cycle (LD 12:12). The binding of each regulator to its regulatory region was analyzed by ChIP with the indicated antibodies. Note that DBP binding in Dbp/ mice was significantly reduced (p < 0.01 by two-way ANOVA), with residual signals potentially deriving from TEF and/or HLF binding. Specific TaqMan probes
272 Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc.
RRE-mediated reporter expression, albeit with an increase in amplitude (Figure S4). The intron sequence sometimes caused double peaks for the E0 box-driven rhythms (Figure 4A and Figure S4). When these rhythms were fitted to a circadian cosine curve, we observed a reduction of the relative amplitude and a slight but reproducible phase advance (Figure 4A, Figure S4, and Table S1). Importantly, the combination of D box in the promoter and RRE-containing Cry1 intron sequence conferred a substantial phase delay of > 5 hr (CT14.48 ± 0.21) when compared to the D box alone (CT9.31 ± 0.16) (Figure 4, Figure S4, and Table S1). It is important to note that our result indicates that E/E0 boxes are dispensable for the generation of delayedphase expression of Cry1. This is because the synthetic composite ‘‘D box + RRE’’ promoter (i.e., a combination of a synthetic D box-driven promoter and RRE-containing Cry1 intron sequence) lacks functional E boxes, unlike the Cry1 promoter. Thus, the D box and the RRE can combine to generate a distinct intermediate phase. We were able to recapitulate these experimental measurements in a simple model using ‘‘phase vectors.’’ A phase vector represents phase and amplitude of the oscillation as direction and length of the vector in polar coordinates. In this way, the combination of two oscillations can be represented by the vector sum of two corresponding phase vectors (Extended Experimental Procedures). We plotted measured oscillations (CCE without intron sequence and intron sequence without CCE) and obtained the summed phase vector of the CCE-intron sequence combinations (Figure 4B, left three circles). Interestingly, the summed phase vectors corresponded well with the measured oscillations (Figure 4B, rightmost). These results support the notion that combining two CCEs that otherwise function independently can be a general mechanism for generation of new phases and, more specifically, the combined phase may be predicted, to a first-order approximation, by a vector sum. Delayed Expression of Cry1 Restores Circadian Rhythmicity in Cry1/:Cry2/ Cells To address the functional importance of the RRE-mediated phase delay, we employed cell-based genetic complementation, testing for phenotypic rescue in arrhythmic Cry1/:Cry2/ cells. We hypothesized that, if phase delay is an important property of Cry1, its delayed expression, peaking at evening-time, should restore circadian oscillations in these cells. To test this hypothesis, we established mouse embryonic fibroblasts from Cry1/:Cry2/ double-knockout mice (van der Horst et al., 1999). Similar to negative control (Figure 5A, without Cry1), Cry1 expression driven only by the Cry1 promoter, P(Cry1), did not rescue circadian oscillations in these cells (Figure 5A, P(Cry1)). However, when Cry1 expression was regulated by
P(Cry1)-Cry1 intron 336, which contains the Cry1 promoter and the RRE-containing 336 bp of Cry1 intron sequence, its exogenous expression restored circadian rhythmicity in these cells, with a period length of 26.73 ± 0.19 hr (Figure 5A, Cry1 intron 336). The observed rescue capability was independent of the CRY1 protein level, vector type, or method of DNA delivery (Figures S5A and S5B). Taken together, these results demonstrate that delay of Cry1 expression, conferred by the Cry1 intron, is required for rescue of circadian rhythmicity. To further assess the contribution of delayed Cry1 expression to the rescued circadian oscillation, we tested the rescue capability of the intronic RREs mutants that possess different RRE strengths, as described above (Figure 3 and Figure 5A, nine panels on the right, and Table S2). The ability of the intronic RRE mutants to rescue rhythmicity, represented as amplitude of circadian oscillations, significantly correlated with the strength of intronic RREs, as measured by bioluminescence levels derived from P(SV40)-Cry1 intron 336-Luc (r2 = 0.87, p < 0.01; Figure 5B, left). Similarly, the rescue capability also correlated with another measurement of strength of intronic RREs, i.e., RORa responsiveness of the intronic RRE mutants (r2 = 0.97, p < 0.01; Figure 5B right). More directly, the rescue capability correlated with the phase delay conferred by the intronic RRE mutants that was measured in bioluminescence rhythms of P(Cry1)-Cry1 intron 336-Luc (r2 = 0.71, p < 0.01; Figure 5C). It should be noted that the rescue capability in these experiments does not correlate with either amplitude or basal bioluminescence levels of P(Cry1)-Cry1 intron 336-Luc (Figure S5C), suggesting that the rescue capability is most likely attributable to the delayed phase of Cry1 expression conferred by RREs. To directly confirm this, we demonstrated that the pure RREs, when combined with Cry1 promoter, rescued circadian rhythmicity in Cry1/:Cry2/ cells, whereas Cry1 promoter alone could not reliably rescue rhythms (Figure S5D and Figure 5). The Cry1-rescued Cry1/:Cry2/ cells (a Cry2 knockout, in essence) showed a rather long period length of 27 hr (Figure 5A), which is consistent with previous reports showing that Cry2/ single-knockout cells display long periods compared to wild-type cells (24–25 hr) (Liu et al., 2007). We confirmed that genetic complementation of Cry1 in Cry1/:Cry2/ cells recapitulates the circadian phenotype in Cry2 single-knockout cells, thus phenotypically validating the Cry1 rescue assay (Figure S5E and Table S3). Cry1 Phase Delay Modulates Circadian Period Length The genetic complementation assay expressing Cry1 of various phases revealed that delay of Cry1 expression is required to restore circadian rhythmicity, consistent with the proposed design principle for circadian clocks, i.e., transcriptional/translational
were used to detect the Cry1 promoter region, the Cry1 first intron region, or the Dbp promoter region (control region). ZT, Zeitgeber time. Mean and SD represent three ChIP experiments. (E) Circadian expression profiles of Cry1, Bmal1, Rev-Erba, and Per1 in vivo. RNA from liver of wild-type (gray) or Dbp/:Tef/:Hlf/ (orange) mice was prepared at 4 hr intervals from mice held in LD 12:12 cycle. Relative mRNA levels of each gene were measured. Note that Cry1 expression in Dbp/:Tef/:Hlf/ mice was different from wild-type and its peak delayed. All RNA samples were normalized to Gapdh mRNA accumulation. Mean and SEM from two pools of three mice each per time point. The representative data from three independent experiments was shown. The observed peak delay was reproducible and significant in three independent experiments (p < 0.01 by two-way ANOVA). See also Figure S2 and Table S1.
Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc. 273
P(SV40)
Flag Cry1 intron
B
polyA
Luc
20 Relative Bioluminescence
A
R1 R2
336
15 10 5 0
0.4
Relative amplitude of P(SV40)-Cry1 intron 336-Luc
inversion
mutation
C
no Cry1 intron wild type Cry1 intron 336 R1 deletion R2 deletion R1,2 deletion R1 mutation R2 mutation R1,2 mutation R1 inversion R2 inversion R1,2 inversion
0.3
0.2
0.1
0
1
2
3
mutation Cry1 intron 336
D
4
y = 0.018x + 0.052 r2= 0.95 0.3
0.2
0.1
0.0 0
5
Flag Cry1 intron
Luc
polyA
F Phase (h)
R1 R2
no Cry1 intron
10.63
wild type Cry1 intron 336
14.03
R1 deletion
11.70
R2 deletion
11.65
R1,2 deletion
10.36
R1 mutation
12.49
mutation
deletion
336
10
15
20
RORα-responsiveness 5
Phase-delay from P(Cry1)-Luc (h)
P(Cry1)
inversion
0.4
Day
y=0.26x-0.21 r2=0.82 2.5
0
0
5
10
15
20
-2.5
RORα-responsiveness
inversion
5
R2 mutation
11.18
R1,2 mutation
9.61
R1 inversion
13.43
R2 inversion
10.48
R1,2 inversion
10.96
Normalized Bioluminescence Low -1
0
1
2
3 Day
High 0
1
4
Phase-delay from P(Cry1)-Luc (h)
E
deletion
control element
Relative amplitude of P(SV40)-Cry1 intron 336-Luc
deletion
RORα - + - + - + - + - + - + - + - + - + - + - + - + - + - + E' D RRE (-) Wild R1 R2 R1,2 R1 R2 R1,2 R1 R2 R1 ,2
y = 14.86x - 1.01 r2= 0.90 2.5
0 0.0
0.1
0.2
0.3
0.4
5 -2.5
Relative amplitude of P(SV40)-Cry1 intron 336-Luc
Figure 3. The Strength of Intronic RREs Correlates with Phase Delay (A) Mutant reporter constructs derived from P(SV40)-Cry1 intron 336-Luc. P(SV40)-Cry1 intron 336-Luc contains the 336 bp of Cry1 intron sequence (wild-type); the RREs or R1/R2 within the Cry1 intron sequence were altered to generate three sets of Cry1 intron sequence mutants: deletions, mutations, and inversions. Red arrow indicates the direction of R1 and R2. Purple, light blue, and white rectangles represent wild-type, mutated, and deleted RREs, respectively. (B) RORa responsiveness of mutated intron sequences. Each reporter construct in (A) was transiently transfected into NIH 3T3 cells in the absence (–) or presence (+) of RORa. Luciferase activities were scaled so that basal activity without RORa was 1. (C) Relative amplitudes of circadian transcriptional activities induced by constructs presented in (A). (D) Correlation between two measurements for the strength of intronic RREs, the RORa responsiveness (B), and the relative rhythm amplitude (C) of mutated intron sequences. (E) Mutant reporter constructs derived from P(Cry1)-Cry1 intron 336-Luc and their phases. The SV40 promoter in P(SV40)-Cry1 intron 336-Luc was replaced with Cry1 promoter P(Cry1) to generate P(Cry1)-Cry1 intron 336-Luc. The RRE mutations in P(Cry1)-Cry1 intron 336-Luc are the same as in P(SV40)-Cry1 intron 336-Luc constructs in (A). (F) Phase delay correlates with two measurements for the strength of intronic RREs, the RORa responsiveness and the relative rhythm amplitude. The RORa responsiveness presented in (B) (top) and relative rhythm amplitudes presented in (C) (bottom) of mutated intron sequences are plotted against phase delay of P (Cry1)-Cry1 intron 336-Luc activity relative to P(Cry1)-Luc activity presented in (E). Data are representative of two independent experiments (B, C, and E). Error bars represent SD (n = 3) (B and C). Mean and SD (error bar) of two independent experiments are shown (each experiment contains three samples; n = 3 unless otherwise indicated in Table S1) (D and F). See also Figure S3 and Table S1.
feedback repression with delay. This design principle further predicts that Cry1 expression with a more prolonged delay can slow circadian oscillations. To test this prediction, we first attemp274 Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc.
ted to generate constructs expressing Cry1 with prolonged delays. According to the phase-vector model described above (Figure 4B), we should be able to generate evening-to-night
A 3× CCE
P(SV40)
Flag Cry1 intron
Luc
polyA
Phase (h) R1 R2
E'-box-P(SV40)
3.27
336
336
D-box-P(SV40)
9.31
D-box-P(SV40)-Cry1 intron 336
14.48
RRE-P(SV40)
17.04
RRE-P(SV40)-Cry1 intron 336
16.32 0
1
2
3
4
Normalized Bioluminescence
2.13 *
E'-box-P(SV40)-Cry1 intron 336
336
High 1
0
-1 Low
5
Day
Predicted
Measured
0
21
18
1.0
21
3 0.5
6
0
15
9 12 Phase (h)
18
1.0
21
3 0.5
6 18
0
15
9 12 Phase (h)
Amplitude
RRE + Cry1 intron 336
0
Amplitude
D-box + Cry1 intron 336
0
Amplitude
E'-box + Cry1 intron 336
0
1.0
21
3 0.5
6 18
0
15
9 12 Phase (h)
Amplitude
B
1.0
3 0.5
6
0
15
9 12 Phase (h)
Figure 4. Combination of Intronic RREs with Known Circadian cis-Elements Gives Rise to Emergent Phases that Can Be Predicted by Phase Vectors (A) Combination of the Cry1 intron sequence with known CCEs gives rise to emergent phases. A promoter was constructed by inserting 3 3 E0 box, 3 3 D box, or 3 3 RRE sequences in the upstream of P(SV40). In the reporter construct, Luciferase expression was under the control of the 3 3 CCE-P(SV40) promoter in the absence or presence of the 336 bp of Cry1 intron sequence. The experiment was performed as in Figure 1A. Phases were estimated by fitting a cosine wave with circadian period corresponding to maximum autocorrelation of the time series using detrended bioluminescence data. This method allowed phase estimation of even distorted wave form expressed by E0 box + Cry1 intron sequence (marked by asterisk). (B) A phase vector model recapitulates the emergent phases. The phase vector of each CCE (E0 box, green arrow; D box, orange arrow; RRE, purple arrow) and Cry1 intron sequence (black arrow) and the vector sum of the two phase vectors (center of colored ellipsoid) are plotted in the polar coordinate (left three circles). The ellipsoidal disk represents 95% confidence region. The phase vectors (colored arrows of black border) represent measured circadian transcriptional activities induced by the combined regulation of Cry1 intron sequence and each CCE (rightmost circle). Data are representative of two independent experiments. See also Figure S4 and Table S1.
expression with a more prolonged delay by weakening the daytime promoter but keeping a constant strength of the night-time enhancer of Cry1 intron sequence. Therefore, we generated an array of day-time promoters with various strengths of D boxes, containing 1, 2, or 3 tandem repeats of D boxes or Cry1proD elements; we confirmed that these day-time promoters displayed day-time phased bioluminescence rhythms of various relative amplitudes, as expected (Figure 6A, Figure S6A, and Table S1). We then generated another set of constructs by combining these day-time promoters with the Cry1 intron sequence (Figure 6B, Figure S6B, and Table S1). These constructs displayed evening-tonight phases of bioluminescence rhythms (Figure 6B, Figure S6B, and Table S1). Importantly, there was a significant correlation between the observed phases and the predicted phases from the simple phase-vector model (r2 = 0.77, p < 0.01; Figure 6C, rightmost panel).
Next, we asked whether evening-to-night Cry1 expression with prolonged delay could slow circadian oscillations (Figure 6D and Table S3). Interestingly, the periods of rescued circadian oscillations ranged from 27 to 31 hr. In particular, period length correlated with the delay prolonged by weakening the day-time promoter: the more the Cry1 phase was delayed, the longer the rescued period (r2 = 0.81, p < 0.01; Figure 6E and see also Figure S6C). We also confirmed that the period length did not significantly correlate with either amplitude or basal activity of Cry1 expression by using a different constitutive promoter (Figure S6D and Table S3). In addition, CRY1 protein level was not responsible for the changes in rescued period (Figure S6E). These results showed that Cry1 expression with a prolonged delay slows circadian oscillations, further supporting the proposed design principle of circadian clocks—transcriptional/ translational feedback repression with delay. Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc. 275
B
0
6
deletion 1
2
3
4
5
P(Cry1)
12000000 8000000 4000000 0
0
1
2
3
4
5
P(Cry1)-Cry1 intron 336
15000000 10000000 5000000 0
0
1
2
3
Day
4
5
6
R2
Relative amplitude of rescued oscillation
y = 1.87x + 0.14 r2= 0.87 0.6
0.3
0.0 0.0
15000000
10000000
10000000
10000000
5000000
5000000
5000000
0
0 0
1
2
3
4
5
6
0.2
0.4
0 0
1
2
3
4
15000000 10000000
10000000
5000000
5000000
5000000
0
0
1
2
3
4
5
0
6
0
15000000
15000000
10000000
10000000
5000000
5000000
1
2
3
4
5
0
6
0
1
2
3
4
5
6
1
2
3
4
5
6
1
2
3
4
5
6
10000000
5000000
0
0
1
2
3
4
5
6
0
0
1
Day
no Cry1 intron wild type Cry1 intron 336 R1 deletion R2 deletion R1,2 deletion R1 mutation R2 mutation R1,2 mutation R1 inversion R2 inversion R1,2 inversion
0.6
0.3
12
0
10000000
2
3
4
18
RORα-responsiveness
5
0
6
0
Day
C
6
6
15000000
y = 0.036x + 0.23 r2= 0.97
0
5
15000000
0.0
Relative amplitude of P(SV40)-Cry1 intron 336-Luc
R1,2
15000000
0.9
0.9
Relative amplitude of rescued oscillation
0
R1
15000000
Day 0.9
Relative amplitude of rescued oscillation
Cry1-/-:Cry2-/- double KO MEF
6
Bioluminescence (counts/min)
dLuc
Bioluminescence (counts/min)
P(Per2)
Bioluminescence (counts/min)
Plasmid DNA for reporter
10000000
mutation
Cry1
Bioluminescence (counts/min)
Flag Cry1 intron
inversion
Plasmid DNA for rescue P(Cry1)
without Cry1 20000000
Bioluminescence (counts/min)
Bioluminescence (counts/min)
A
-2.5
0.6
0.3 y = 0.11x + 0.29 r2= 0.71 0.0 0.0
2.5
5.0
Phase-delay of P(Cry1)-Cry1 intron 336-Luc (h)
Figure 5. Delayed Expression of Cry1 Restores Circadian Rhythmicity in Cry1/:Cry2/ Cells
(A) Genetic complementation of Cry1 rescues circadian oscillation in Cry1/:Cry2/ cells. A schematic diagram of Cry1 rescue constructs is shown on the left. The composite promoter contains P(Cry1) and the 336 bp Cry1 intron sequence of wild-type or a mutant (deletion, mutation, or inversion of the R1 and R2 sequences as in Figure 3), which controls Cry1 expression. Cry1 rescue constructs were each cotransfected with a destabilized Luciferase reporter construct, P (Per2)-dLuc, into Cry1/:Cry2/ mouse embryonic fibroblast cells (left), followed by bioluminescence recording. Whereas mock-transfected Cry1/:Cry2/ cells were completely arrhythmic and those expressing P(Cry1)-Cry1 were only transiently rhythmic during the first 2 days of recording, P(Cry1)-Cry1 intron 336Cry1 expression restored circadian oscillation with a period length of 26.73 ± 0.19 hr (bottom in the center column). Rescue effects varied among the intronic RRE mutants (right nine panels). Data are representative of two independent experiments. (B) Relative amplitude of rescued circadian oscillation correlates with the strength of intronic RREs. The relative amplitudes of rescued oscillation are plotted against two measurements for the strength of intronic RREs, the relative amplitudes of P(SV40)-Cry1 intron 336-Luc oscillation presented in Figure 3C, and the RORa responsiveness presented in Figure 3B. (C) Relative amplitude of rescued circadian oscillation correlates with phase delay. The relative amplitudes of rescued oscillations are plotted against the phase delay of various P(Cry1)-Cry1 intron 336-Luc activities relative to P(Cry1)-Luc activity presented in Figure 3E. Mean and SD (error bar) of two independent experiments are shown (each experiment contains three samples; n = 3 unless otherwise indicated in Table S1). See also Figure S5 and Table S2.
Single-Cell Analysis Confirms the Importance of Cry1 Phase Delay in Feedback Repression Arrhythmic phenotypes observed in population of cells might be due to rapid damping of individual cells or lack of synchronization among individual cells. To discriminate between these possibilities, we monitored bioluminescence levels in real time at the level of single-cell resolution (Sato et al., 2006; Ukai et al., 2007). As with whole-well assays, single-cell analysis showed that most individual cells expressing Cry1 with a normal delay, driven by the intron sequence containing wild-type RREs, were robustly rhythmic, with a circadian period of 26.77 ± 0.12 hr 276 Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc.
(Figures 7A and 7B and Table S3), whereas most cells expressing Cry1 without delay, driven by an intron sequence harboring mutated RREs, were arrhythmic (Figure 7B and Movie S1). Moreover, individual cells expressing Cry1 with a prolonged delay driven by the Cry1 intron sequence alone (i.e., in the absence of Cry1 promoter) displayed long circadian periods of up to 32.00 ± 0.58 hr (Figures 7A and 7B and Table S3). The circadian oscillations in Figure 7A with delayed Cry1 expression were statistically significant (p < 0.01 by autocorrelation) and reproducible in different series of experiments. Thus, singlecell analysis confirmed the circadian phenotypes observed in
A
P(SV40)
3× CCE
Flag
Luc
B
polyA
Flag Cry1 intron
P(SV40)
3× CCE
Luc
polyA
Relative Phase (h) amplitude
Phase (h)
3×E'-box-P(SV40)
3.37
0.40
P(Cry1)
14.61
3×RRE-P(SV40)
18.55
0.26
3×D-box-P(SV40)
15.46
P(Cry1)
9.61
0.18
2×D-box-P(SV40)
16.60
3×D-box-P(SV40)
9.84
0.14
1×D-box-P(SV40)
17.35
2×D-box-P(SV40)
11.48
0.08
3×Cry1proD-P(SV40)
16.64
1×D-box-P(SV40)
---
0.06
2×Cry1proD-P(SV40)
16.56
3×Cry1proD-P(SV40)
13.43
0.15
1×Cry1proD-P(SV40)
16.61
2×Cry1proD-P(SV40)
12.70
0.18
P(SV40)
1×Cry1proD-P(SV40)
12.00
0.11
---
0.06
P(SV40) 0
1
2
3
4
1
2
3
4
5
Day
Normalized Bioluminescence Low
5
High
-1
Day
C
17.39 0
0
1 18
18
Measured phase
15
9 0.4
12 Phase (h)
D
6
Bioluminescence (counts/min)
P(SV40)
6
D-box-P(SV40)
20000000 15000000 10000000 5000000
0
1
2
3
4
5
15000000 10000000 5000000 0 1
2
3 Day
4
5
Cry1proD-P(SV40)
P(Cry1) 20000000
0
15
9
3× element
12 Phase (h)
2× element 20000000
15000000
15000000
15000000
10000000
10000000
10000000
5000000
5000000
5000000
1
2
3
4
5
6
0
0
1
2
3
4
5
15
14
6
0
0
1
2
3
15000000
15000000
10000000
10000000
10000000
5000000
5000000
5000000
0
0
1
2
3 4 Day
5
6
0
0
1
2
3 4 Day
5
6
0
0
16
17
18
32
4
5
6
30
28 y = 0.94x + 27.53 r2= 0.81
-1 15000000
15
Predicted phase (h)
1× element
20000000
0
16
E
20000000
0
y = 1.22x - 3.54 r2= 0.77 17
0.4
+ Cry1 intron 336
Bioluminescence (counts/min)
Bioluminescence (counts/min)
Bioluminescence (counts/min)
+ Cry1 intron 336
0
0.2
P(Cry1)-Cry1 intron 336 3×D-box-P(SV40)-Cry1 intron 336 2×D-box-P(SV40)-Cry1 intron 336 1×D-box-P(SV40)-Cry1 intron 336 3×Cry1proD-P(SV40)-Cry1 intron 336 2×Cry1proD-P(SV40)-Cry1 intron 336 1×Cry1proD-P(SV40)-Cry1 intron 336
Amplitude
Amplitude
Predicted phase 0.2
6
Period of rescued oscillation (h)
6
Measured phase (h)
Measured phase
Predicted phase
18
26
0
1
2
3
4
Phase-delay of composite promoters (h)
1
2
3 4 Day
5
6
P(SV40) P(Cry1) 3×D-box-P(SV40) 2×D-box-P(SV40) 1×D-box-P(SV40) 3×Cry1proD-P(SV40) 2×Cry1proD-P(SV40) 1×Cry1proD-P(SV40)
Figure 6. Prolonged Delay of Cry1 Expression Slows Circadian Oscillations in Cry1/:Cry2/ Cells (A) Promoters harboring various CCEs display different circadian phases. The promoters contain one, two, or three tandem copies of D box or Cry1proD element, which were inserted into the P(SV40)-Luc vector to generate an array of reporter constructs. (B) The 336 bp of Cry1 intron sequence confers phase delay to D box and Cry1proD element. Reporter constructs were generated similarly as in (A) except that the 336 bp of Cry1 intron sequence was inserted. The experiment was performed as in Figure 1A (A and B). (C) The measured phases conferred by the composite promoters are consistent with those predicted by phase vectors. (Left) The phase vectors of oscillations driven by various promoters without the intron sequence (colored arrows) and those driven by the intron sequence (CT17.5) or P(Cry1) (CT10) (two black arrows) are plotted with summed phase vectors (center of colored ellipsoidal disks). The ellipsoidal disk represents 95% confidence region. (Right) The summed phase vectors in the left circle are plotted with phase vectors of measured oscillations driven by the composite promoters. (Rightmost) The predicted phases from the simple phase-vector model are plotted against the observed phases. Error bars represent SD (n = 3). (D) Cry1 rescue of circadian oscillation in Cry1/:Cry2/ cells using synthetic composite promoters. The composite promoters presented in (B) were used to drive Cry1 expression. Cry1/:Cry2/ cells were cotransfected with a Cry1 expression construct and a P(Per2)-dLuc reporter. (E) Prolonged phase delay of Cry1 expression correlates with period length of rescued oscillations. The period lengths of rescued oscillations are plotted against the phase delay of various composite promoters’ activity relative to P(Cry1)-Cry1 intron 336-Luc activity presented in (B). Mean and SD (error bar) of two independent experiments are shown (each experiment contains three samples; n = 3). Data are representative of two independent experiments (A, B, and D). See also Figure S6, Table S2, and Table S3.
Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc. 277
Bioluminescence (counts/min)
A
P(Cry1)-Cry1 intron 336 R1,2 mutation-Cry1
P(Cry1)-Cry1 intron 336-Cry1
P(SV40)-Cry1 intron 336 R1,2 mutation-Cry1
P(SV40)-Cry1 intron 336-Cry1
15000000
15000000
15000000
15000000
10000000
10000000
10000000
10000000
5000000
5000000
5000000
5000000
0
B
0
1
2
3 Day
4
5
0
6
1
2
3 Day
4
5
0
6
P(Cry1)-Cry1 intron 336 R1,2 mutation-Cry1
P(Cry1)-Cry1 intron 336-Cry1 Activity (% of maximum)
0
0
1
2
3 Day
4
5
100
100
100
80
80
80
80
60
60
60
60
40
40
40
40
20
20
20
20
0
1
2
3
4 Day
5
6
7
0
1
2
3
4 Day
5
6
0
7
0
10-8 -6
1
2
3
4 Day
5
6
20
10-4
10-8
40
10-6
40
60
10-4
60
60
80
10-2
80
80
100
100
0
1
2
3 4 Day
5
6
7
40
10-4
10-2
1
0
2
3 4 Day
5
6
7
2
3 Day
4
5
6
1
2
3
4 Day
5
6
7
10-4
20
10-2
40 60
10-2
0
1
2
3 4 Day
5
6
7
80 100
0
1
2
3 4 Day
5
6
7
Amplitude
100
10-6
20
1
0
10-8
10
20
0
7
0
0
P(SV40)-Cry1 intron 336 R1,2 mutation-Cry1
P(SV40)-Cry1 intron 336-Cry1
100
0 Ranked cell
0
6
C
Period Delay Cry1 promoter + Cry1 intron
Cry1 promoter
D
Cry1 intron
E E/E'box
D-box
E/E'box
D-box
E/E'box
D-box
Period Am pli
CRY1
tud
Evening
e
RRE
RRE
Figure 7. Single-Cell Analysis Confirms the Requirement of Cry1 Phase Delay
RRE
(A) P(Per2)-dLuc bioluminescence levels in transfected Cry1/:Cry2/ cells as recorded. The P(Per2)-dLuc reporter and a Cry1 expression construct as indicated were cotransfected into Cry1/:Cry2/ cells, and bioluminescence expression was recorded with a PMT. Data from three independent samples are shown. (B) P(Per2)-dLuc bioluminescence levels in transfected individual Cry1/:Cry2/ cells as recorded by a luminescence microscope (n = 100). Reporter activities from each cell were normalized so that the maximum and minimum bioluminescence values are 100% and 0%, respectively. The mean reporter activity for all of the analyzed single cells at each time point is indicated by a thick black line (top row). Time series of bioluminescence expression shown in the top row were redrawn as heatmaps (bottom row). Each row in the heatmap represents a time series of P(Per2)-dLuc reporter activities from a single cell. The corresponding p value of rhythmicity at the period of maximum autocorrelation was evaluated for each time series and is depicted on the right. One-hundred cells were randomly selected and individually analyzed. Data are representative of two independent experiments (A and B). (C) The roles of phase delay in Cry1 expression. Through regulation of Cry1 expression, the promoter and intron primarily affect the amplitude and period of the clock system, respectively.
278 Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc.
Cry1-rescued Cry1/:Cry2/ cells at the cell population level, which lends strong support for our finding that delay of Cry1 expression is required for circadian clock function. DISCUSSION Cry1 Phase Control Mechanism In this report, we provided experimental data, as well as model predictions, for a ‘‘combinatorial regulatory mechanism’’ to explain the delayed expression of Cry1. We newly identified D boxes, which overlap with the E/E0 box and confer phase delay over E/E0 box activity. In addition, we also demonstrated that the previously identified RREs in the first intron (Ueda et al., 2005) can confer the additional phase delay in Cry1 expression. Furthermore, we observed that the synthetic pure RREs, in combination with Cry1 promoter, rescued circadian rhythmicity in Cry1/:Cry2/ cells with statistical significance (p < 0.01; Figure S5D). Together with the observation that the combination of pure D box and Cry1 intron sequence also rescued rhythms (Figure 6D), we conclude that the RRE and D box elements can recapitulate the basic function of Cry1 intron and Cry1 promoter, respectively. It should be noted that the possible contribution from unknown elements in Cry1 intron 336 sequence could not be completely excluded because the relative amplitude of the pure RRE elements (combined with Cry1 promoter) is slightly lower than that of the wild-type Cry1 intron 336 (combined with Cry1 promoter, Figure S5D). General Design Principles for New Phases As revealed in this study, these multiple distinct regulatory sites (i.e., two RREs in first intron and the E/E0 box and D boxes in the promoter region) function in a coordinated fashion to generate substantial phase delay, leading to evening-time expression. Interestingly, in an effort to study design principles of the circadian clockwork, we employed a simple phase-vector model in which the new evening-time could be predicted by the combination of two component phase vectors. Although the phase-vector model was not used for phase prediction in our previous study (Ukai-Tadenuma et al., 2008), such a model is also applicable to this previous study when we take into account the time delay associated with transcription/translation of regulator proteins and the Luciferase reporter (r2 = 0.99, p < 0.01; Figures S7A and S7B and Table S4). These results show that a new phase can be generated through combinatorial synthesis of either two transcriptional regulators or two clock-controlled DNA elements and also can be described, at least to a first-order approximation, by a phase-vector model. Taken together, this combinatorial regulatory mechanism for the generation of new circadian phases of transcription represents a general design principle underpinning the complex system behavior. Although the phase-vector model predicts the phase of a synthesized oscillation, it is only a first-order approximation.
For example, there are some discrepancies between predicted and measured amplitudes. Also, the wave form generated by the combination of E0 box-driven promoter and the Cry1 intron sequence appeared like a ‘‘two-peak’’ wave form (Figure S4, upper-left), indicative of nonlinear effects or involvement of yet unknown factors. In detailed analysis, we found that the 11 hr period oscillation was actually a significant component in the two-peak wave form (Figure S4, upper-right; p < 0.01). This is the first demonstration of synthesized ultradian rhythms, which may lead to insights into mechanisms of ultradian gene expression with harmonic periods, as recently reported (Hughes et al., 2009). Delayed Cry1 Expression Contributes to Clock Robustness In this study, we focused on the level of transcriptional regulation and demonstrated the importance of delay in feedback repression at the intracellular level. Constitutive expression of Cry1 abolished circadian rhythmicity in wild-type cells (Ueda et al., 2005) and failed to restore circadian oscillation in arrhythmic Cry1/:Cry2/ cells (Figure S6D and Figure 7A), suggesting that rhythmic expression of Cry1 is important for clock function. We revealed that the transcriptional oscillation of Cry1 with a correct phase with substantial delay was sufficient and required to rescue circadian oscillation in arrhythmic Cry1/:Cry2/ cells (Figure 5) and that transcriptional oscillation of Cry1 with a prolonged delay slows circadian oscillation (Figure 6). Importantly, we also confirmed that the amount (baseline) of CRY1 protein was not responsible for the changes in amplitude (Figure S5A) and period (Figure S6E) of rescued oscillations. These results suggest that the phase of Cry1 expression is responsible for the changes in rescued amplitude and period rather than the amount (baseline) of CRY1 protein. Because we confirmed the significant linear correlation between transcriptional activities and protein levels (p < 0.01), when monitored by firefly Luciferase, and CRY1 protein amounts, when monitored by fusion Renilla Luciferase, irrespective of cell types (r2 = 0.93 in NIH3T3 and r2 = 0.90 in Cry1/:Cry2/ cells), we speculated that phase of CRY1 protein level would be responsible for the amplitude and period of rescued oscillations. It should be noted that the CRY1 protein expression levels in our experiments are within a certain range (Figure S5A and Figure S6E), and we do not exclude (and our current results are not in conflict with) the notion that CRY1 protein amounts may affect the parameters of clock function when CRY1 protein levels drastically differ from those in our experimental system, as previously reported (Baggs et al., 2009; Ueda et al., 2005). In addition, we do not exclude the possibility that other regulatory mechanisms such as posttranscriptional modifications (Lee et al., 2001; Liu et al., 2008) play important roles in attaining the robustness of the clock. For example, rhythmic expression of PER2 is recently reported to play a prominent role in CRY1 function (Chen et al.,
(D) A schematic diagram of a minimal circuit for the mammalian circadian transcriptional network. The network can be represented by a simple circuit, consisting of two transcriptional activations (green arrows) and four transcriptional repressions (red arrows) on three regulatory elements (three rectangles). (E) The minimal circuit envisaged as a composite of two distinct oscillatory network motifs: (1) A repressilator that is composed of three repressions (left) and (2) a delayed negative feedback loop, which is composed of two activations and one repression (right). See also Figure S7, Movie S1, and Table S3.
Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc. 279
2009). And PER2 is phosphorylated by CKId/3, which is also an essential mammalian clock component (Lee et al., 2009) and has been recently implicated in temperature compensation (Isojima et al., 2009). In addition, it has been reported that constant supply of membrane-permeable CRY1 and CRY2 proteins rescued circadian oscillation in Cry1/:Cry2/ cells (Fan et al., 2007). In line with this observation, our single-cell analysis indicated that a fraction of individual cells transfected with Cry1 driven by a constitutive promoter exhibited weak circadian oscillation even though the rhythms are rather transient (Figure 7B), implying that constant Cry1 expression might partially rescue circadian clock function. This qualitatively less-robust clock function is probably attributable to posttranscriptional and posttranslational mechanisms (Lee et al., 2001; Liu et al., 2008). Even in this context, it appears that phase delay in rhythmic Cry1 expression may contribute to the robustness of clock function by ensuring properly timed nuclear translocation of CRY proteins. This idea is strongly supported by our results presented in this study: delayed Cry1 expression via D box-mediated transcription (i.e., from Cry1 promoter) allowed partial rescue, and further delay via the RREs from the Cry1 intron restored circadian rhythmicity with amplitude and persistence comparable to wild-type cells. Design Principle for a Circadian Transcriptional Network Delayed feedback repression is one of the most prevailing but as yet unverified design principles for a circadian transcriptional network. This design principle predicts that decreased delay dampens circadian oscillations and that prolonged delay slows down circadian oscillations (Figure 7C, Figures S7C–S7F, and Extended Experimental Procedures) (Bernard et al., 2006; Lewis, 2003; Novak and Tyson, 2008). The results presented in this study are consistent with the two predictions from the delayed feedback repression, suggesting that it is an applicable design principle in the mammalian circadian transcriptional network. A Minimal Circuit for a Circadian Transcriptional Network In a previous effort to identify a minimal circuit of the complex autoregulatory transcriptional networks in the mammalian circadian clock, we showed that day-time promoter activity can be reconstructed by combining a morning-time activator and a night-time repressor and night-time promoter activity by combining a day-time activator and a morning-time repressor (Ukai-Tadenuma et al., 2008). In this study, we succeeded in synthesizing the evening-time phase control of transcription. Our previous and current results suggest that the complex mammalian transcription network can be reduced to a relatively simple diagram (Figure 7D) that would consist of three regulatory elements and six transcriptional regulations (two activations and four repressions). It is noteworthy that this diagram can be envisaged as a composite of two distinct oscillatory network motifs (Figure 7E). The first oscillatory network motif is composed of three repressions (i.e., E/E0 box to RRE, RRE to D box, and D box to E/E0 box), comprising a cyclic negative feedback loop—a repressilator (Elowitz and Leibler, 2000). The second oscillatory network motif is composed of two activations (i.e., 280 Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc.
E/E0 box to D box and D box to RRE) and one repression (i.e., RRE to E/E0 box), comprising a delayed negative feedback loop. It is interesting to note that oscillatory properties of both network motifs were experimentally suggested by synthetic approaches (Elowitz and Leibler, 2000; Stricker et al., 2008). Therefore, further experimental and theoretical analyses of the composite of these oscillatory network motifs lie ahead. EXPERIMENTAL PROCEDURES Preparation of Embryonic Fibroblasts from Cry1/:Cry2/ Double-Knockout Mice Cry1/:Cry2/ double-knockout mice (van der Horst et al., 1999) were carefully kept and handled according to the RIKEN Regulations for Animal Experiments. The dissociated cells (mouse embryonic fibroblasts [MEF] from Cry1/:Cry2/ double-knockout mice; Cry1/:Cry2/ cells) were suspended and cultured in DMEM (Invitrogen) supplemented with 10% FBS (JRH Biosciences) and antibiotics (see Extended Experimental Procedures for details). Real-Time Circadian Reporter Assay Using NIH 3T3 Cells and Cry1/:Cry2/ Cells Real-time circadian assays were performed as previously described (Sato et al., 2006; Ueda et al., 2005) with the following modifications. NIH 3T3 cells were transfected with the Luciferase reporter plasmids. Cry1/:Cry2/ cells were transfected with pGL3-P(Per2)-dLuc reporter plasmid (Sato et al., 2006) and each Cry1 gene expression vector. The cells were stimulated by 10 mM (NIH 3T3) or 30 mM (Cry1/:Cry2/ cells) forskolin (Fermentek), and the bioluminescence was measured at 30 C (see Extended Experimental Procedures for details).
SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, seven figures, four tables, and one movie and can be found with this article online at doi:10.1016/j.cell.2010.12.019. ACKNOWLEDGMENTS This research was supported by an intramural Grant-in-Aid from the RIKEN Center for Developmental Biology (CDB) (H.R.U.), Uehara Memorial Foundation (H.R.U.), The Mitsubishi Foundation (H.R.U.), the President’s Fund from RIKEN (H.R.U.), KAKENHI (Grant-in-Aid for Scientific Research) on Priority Areas ‘‘Systems Genomics’’ from the Ministry of Education, Culture, Sports, Science and Technology of Japan (H.R.U.), the National Science Foundation (IOS-0920417) (A.C.L.), and the SNF/FNS (J.A.R.). We thank Ueli Schibler, Urs Albrecht, and Fre´de´ric Gachon for their valuable reagents; Akira Yasui and Gijsbertus T.J. van der Horst for Cry1/:Cry2/ double-knockout mice; Hideki Ukai and Koh-hei Masumoto for technical support in establishment of Cry1/:Cry2/ cells; Yohei Koyama for critical comments on the theoretical analysis of delayed negative feedback loop; Hajime Tei for Cry1 and Cry2 expression vectors; and David K. Welsh for critical reading of the manuscript. H.R.U. and A.C.L. designed the research scheme. M.U.-T. constructed most of materials and performed real-time luminescence assays. R.G.Y. conducted bioinformatic, statistic, and theoretical analyses. H.X. constructed the virus vector and performed the corresponding real-time luminescence assays. J.A.R. performed ChIP analysis and mRNA accumulation analysis. All authors discussed the results and commented on the manuscript text. Received: May 21, 2010 Revised: August 21, 2010 Accepted: December 10, 2010 Published online: January 13, 2011
REFERENCES Baggs, J.E., Price, T.S., DiTacchio, L., Panda, S., FitzGerald, G.A., and Hogenesch, J.B. (2009). Network Features of the Mammalian Circadian Clock. PLoS Biol. 7, e1000052. Bernard, S., Cajavec, B., Pujo-Menjouet, L., Mackey, M.C., and Herzel, H. (2006). Modelling transcriptional feedback loops: the role of Gro/TLE1 in Hes1 oscillations. Philos. Transact. A Math. Phys. Eng. Sci. 364, 1155–1170. Brown, S.A., Fleury-Olela, F., Nagoshi, E., Hauser, C., Juge, C., Meier, C.A., Chicheportiche, R., Dayer, J.M., Albrecht, U., and Schibler, U. (2005). The period length of fibroblast circadian gene expression varies widely among human individuals. PLoS Biol. 3, e338. Chen, R., Schirmer, A., Lee, Y., Lee, H., Kumar, V., Yoo, S.-H., Takahashi, J.S., and Lee, C. (2009). Rhythmic PER Abundance Defines a Critical Nodal Point for Negative Feedback within the Circadian Clock Mechanism. Mol. Cell 36, 417–430. DeBruyne, J.P., Weaver, D.R., and Reppert, S.M. (2007). Peripheral circadian oscillators require CLOCK. Curr. Biol. 17, R538–R539. Dunlap, J.C. (1999). Molecular bases for circadian clocks. Cell 96, 271–290. Elowitz, M.B., and Leibler, S. (2000). A synthetic oscillatory network of transcriptional regulators. Nature 403, 335–338. Etchegaray, J.P., Lee, C., Wade, P.A., and Reppert, S.M. (2003). Rhythmic histone acetylation underlies transcription in the mammalian circadian clock. Nature 421, 177–182.
Lee, H., Chen, R., Lee, Y., Yoo, S., and Lee, C. (2009). Essential roles of CKIdelta and CKIepsilon in the mammalian circadian clock. Proc. Natl. Acad. Sci. USA 106, 21359–21364. Lewis, J. (2003). Autoinhibition with transcriptional delay: a simple mechanism for the zebrafish somitogenesis oscillator. Curr. Biol. 13, 1398–1408. Liu, A.C., Tran, H.G., Zhang, E.E., Priest, A.A., Welsh, D.K., and Kay, S.A. (2008). Redundant function of REV-ERBalpha and beta and non-essential role for Bmal1 cycling in transcriptional regulation of intracellular circadian rhythms. PLoS Genet. 4, e1000023. Liu, A.C., Welsh, D.K., Ko, C.H., Tran, H.G., Zhang, E.E., Priest, A.A., Buhr, E.D., Singer, O., Meeker, K., Verma, I.M., et al. (2007). Intercellular coupling confers robustness against mutations in the SCN circadian clock network. Cell 129, 605–616. Lopez-Molina, L., Conquet, F., Dubois-Dauphin, M., and Schibler, U. (1997). The DBP gene is expressed according to a circadian rhythm in the suprachiasmatic nucleus and influences circadian behavior. EMBO J. 16, 6762–6771. Mitsui, S., Yamaguchi, S., Matsuo, T., Ishida, Y., and Okamura, H. (2001). Antagonistic role of E4BP4 and PAR proteins in the circadian oscillatory mechanism. Genes Dev. 15, 995–1006. Novak, B., and Tyson, J.J. (2008). Design principles of biochemical oscillators. Nat. Rev. Mol. Cell Biol. 9, 981–991. Preitner, N., Damiola, F., Lopez-Molina, L., Zakany, J., Duboule, D., Albrecht, U., and Schibler, U. (2002). The orphan nuclear receptor REV-ERBalpha controls circadian transcription within the positive limb of the mammalian circadian oscillator. Cell 110, 251–260.
Falvey, E., Marcacci, L., and Schibler, U. (1996). DNA-binding specificity of PAR and C/EBP leucine zipper proteins: a single amino acid substitution in the C/EBP DNA-binding domain confers PAR-like specificity to C/EBP. Biol. Chem. 377, 797–809.
Reppert, S.M., and Weaver, D.R. (2002). Coordination of circadian timing in mammals. Nature 418, 935–941.
Fan, Y., Hida, A., Anderson, D.A., Izumo, M., and Johnson, C.H. (2007). Cycling of CRYPTOCHROME proteins is not necessary for circadian-clock function in mammalian fibroblasts. Curr. Biol. 17, 1091–1100.
Ripperger, J.A., and Schibler, U. (2006). Rhythmic CLOCK-BMAL1 binding to multiple E box motifs drives circadian Dbp transcription and chromatin transitions. Nat. Genet. 38, 369–374.
Fustin, J.M., O’Neill, J.S., Hastings, M.H., Hazlerigg, D.G., and Dardente, H. (2009). Cry1 circadian phase in vitro: wrapped up with an E box. J. Biol. Rhythms 24, 16–24.
Sato, T.K., Yamada, R.G., Ukai, H., Baggs, J.E., Miraglia, L.J., Kobayashi, T.J., Welsh, D.K., Kay, S.A., Ueda, H.R., and Hogenesch, J.B. (2006). Feedback repression is required for mammalian circadian clock function. Nat. Genet. 38, 312–319.
Gachon, F., Fonjallaz, P., Damiola, F., Gos, P., Kodama, T., Zakany, J., Duboule, D., Petit, B., Tafti, M., and Schibler, U. (2004). The loss of circadian PAR bZip transcription factors results in epilepsy. Genes Dev. 18, 1397–1412. Gekakis, N. (1998). Role of the CLOCK protein in the mammalian circadian mechanism. Science 280, 1564–1569. Griffin, E.A., Jr., Staknis, D., and Weitz, C.J. (1999). Light-independent role of CRY1 and CRY2 in the mammalian circadian clock. Science 286, 768–771. Harding, H.P., and Lazar, M.A. (1993). The orphan receptor Rev-ErbA alpha activates transcription via a novel response element. Mol. Cell. Biol. 13, 3113–3121. Hogenesch, J.B., Chan, W.K., Jackiw, V.H., Brown, R.C., Gu, Y.Z., Pray-Grant, M., Perdew, G.H., and Bradfield, C.A. (1997). Characterization of a subset of the basic-helix-loop-helix-PAS superfamily that interacts with components of the dioxin signaling pathway. J. Biol. Chem. 272, 8581–8593. Hughes, M.E., DiTacchio, L., Hayes, K.R., Vollmers, C., Pulivarthy, S., Baggs, J.E., Panda, S., and Hogenesch, J.B. (2009). Harmonics of circadian gene transcription in mammals. PLoS Genet. 5, e1000442. Isojima, Y., Nakajima, M., Ukai, H., Fujishima, H., Yamada, R.G., Masumoto, K.H., Kiuchi, R., Ishida, M., Ukai-Tadenuma, M., Minami, Y., et al. (2009). CKIepsilon/delta-dependent phosphorylation is a temperature-insensitive, period-determining process in the mammalian circadian clock. Proc. Natl. Acad. Sci. USA 106, 15744–15749. Kume, K., Zylka, M.J., Sriram, S., Shearman, L.P., Weaver, D.R., Jin, X., Maywood, E.S., Hastings, M.H., and Reppert, S.M. (1999). mCRY1 and mCRY2 are essential components of the negative limb of the circadian clock feedback loop. Cell 98, 193–205. Lee, C., Etchegaray, J., Cagampang, F.R., Loudon, A.S., and Reppert, S.M. (2001). Posttranslational mechanisms regulate the mammalian circadian clock. Cell 107, 855–867.
Stricker, J., Cookson, S., Bennett, M.R., Mather, W.H., Tsimring, L.S., and Hasty, J. (2008). A fast, robust and tunable synthetic gene oscillator. Nature 456, 516–519. Ueda, H.R., Chen, W., Adachi, A., Wakamatsu, H., Hayashi, S., Takasugi, T., Nagano, M., Nakahama, K., Suzuki, Y., Sugano, S., et al. (2002). A transcription factor response element for gene expression during circadian night. Nature 418, 534–539. Ueda, H.R., Hayashi, S., Chen, W., Sano, M., Machida, M., Shigeyoshi, Y., Iino, M., and Hashimoto, S. (2005). System-level identification of transcriptional circuits underlying mammalian circadian clocks. Nat. Genet. 37, 187–192. Ukai-Tadenuma, M., Kasukawa, T., and Ueda, H.R. (2008). Proof-by-synthesis of the transcriptional logic of mammalian circadian clocks. Nat. Cell Biol. 10, 1154–1163. Ukai, H., Kobayashi, T.J., Nagano, M., Masumoto, K.H., Sujino, M., Kondo, T., Yagita, K., Shigeyoshi, Y., and Ueda, H.R. (2007). Melanopsin-dependent photo-perturbation reveals desynchronization underlying the singularity of mammalian circadian clocks. Nat. Cell Biol. 9, 1327–1334. van der Horst, G.T., Muijtjens, M., Kobayashi, K., Takano, R., Kanno, S., Takao, M., de Wit, J., Verkerk, A., Eker, A.P., van Leenen, D., et al. (1999). Mammalian Cry1 and Cry2 are essential for maintenance of circadian rhythms. Nature 398, 627–630. Yoo, S.H., Ko, C.H., Lowrey, P.L., Buhr, E.D., Song, E.J., Chang, S., Yoo, O.J., Yamazaki, S., Lee, C., and Takahashi, J.S. (2005). A noncanonical E box enhancer drives mouse Period2 circadian oscillations in vivo. Proc. Natl. Acad. Sci. USA 102, 2608–2613. Young, M.W., and Kay, S.A. (2001). Time zones: a comparative genetics of circadian clocks. Nat. Rev. Genet. 2, 702–715.
Cell 144, 268–281, January 21, 2011 ª2011 Elsevier Inc. 281
RIM Proteins Tether Ca2+ Channels to Presynaptic Active Zones via a Direct PDZ-Domain Interaction Pascal S. Kaeser,1,3,6 Lunbin Deng,1,3,6 Yun Wang,3 Irina Dulubova,4,7 Xinran Liu,3 Josep Rizo,4 and Thomas C. Su¨dhof1,2,3,5,* 1Department
of Molecular and Cellular Physiology Hughes Medical Institute Stanford University, Lorry Lokey Building, 265 Campus Dr., Stanford 94305-5453, USA 3Department of Neuroscience 4Department of Biochemistry 5Howard Hughes Medical Institute UT Southwestern Medical Center, Dallas, TX 75390, USA 6These authors contributed equally to this work 7Present address: Reata Pharmaceuticals, 2801 Gateway Dr., Irving, TX 75063, USA *Correspondence:
[email protected] DOI 10.1016/j.cell.2010.12.029 2Howard
SUMMARY
At a synapse, fast synchronous neurotransmitter release requires localization of Ca2+ channels to presynaptic active zones. How Ca2+ channels are recruited to active zones, however, remains unknown. Using unbiased yeast two-hybrid screens, we here identify a direct interaction of the central PDZ domain of the active-zone protein RIM with the C termini of presynaptic N- and P/Q-type Ca2+ channels but not L-type Ca2+ channels. To test the physiological significance of this interaction, we generated conditional knockout mice lacking all multidomain RIM isoforms. Deletion of RIM proteins ablated most neurotransmitter release by simultaneously impairing the priming of synaptic vesicles and by decreasing the presynaptic localization of Ca2+ channels. Strikingly, rescue of the decreased Ca2+channel localization required the RIM PDZ domain, whereas rescue of vesicle priming required the RIM N terminus. We propose that RIMs tether N- and P/ Q-type Ca2+ channels to presynaptic active zones via a direct PDZ-domain-mediated interaction, thereby enabling fast, synchronous triggering of neurotransmitter release at a synapse. INTRODUCTION At a synapse, action potentials induce Ca2+ influx into a presynaptic terminal, which triggers rapid synchronous neurotransmitter release, thereby initiating synaptic transmission (Katz, 1969). Release is mediated by synaptic vesicle exocytosis at the active zone, a specialized region of the presynaptic plasma 282 Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc.
membrane that docks and primes vesicles for exocytosis (Wojcik and Brose, 2007). Fast synchronous release requires colocalization of Ca2+ channels with the release machinery at the active zone (Llinas et al., 1992; Meinrenken et al., 2002). Voltage-gated Ca2+ channels consist of a pore-forming a1 subunit and accessory b and a2d subunits (Catterall et al., 2005). Presynaptic neurotransmitter release almost exclusively depends on N- and P/Q-type Ca2+ channels containing CaV2.1 and CaV2.2 a1 subunits, respectively; R-type Ca2+ channels containing CaV2.3 a1 subunits may also contribute, whereas L-type and T-type Ca2+ channels containing CaV1 and CaV3 a1 subunits do not (Castillo et al., 1994; Dietrich et al., 2003; Luebke et al., 1993; Poncer et al., 1997; Regehr and Mintz, 1994; Takahashi and Momiyama, 1993; Wu et al., 1999). However, how N- and P/Q-type Ca2+ channels are specifically localized to active zones and coupled to the release machinery is unknown. Active zones are composed of evolutionarily conserved proteins, including Munc13s, RIMs, RIM-BPs (RIM-binding proteins), ELKS’s, and a-liprins (Wojcik and Brose, 2007). Of these proteins, RIMs are likely the central organizers because they directly or indirectly interact with all other known active-zone proteins and with synaptic vesicles (Mittelstaedt et al., 2010). RIM proteins are expressed in three principal isoforms (Kaeser et al., 2008; Wang et al., 1997, 2000): RIM1a and RIM2a that contain all RIM domains (i.e., N-terminal Rab3- and Munc13binding sequences, central PDZ domains, and C-terminal C2A and C2B domains with an intercalated PxxP sequence that binds to RIM-BPs); RIM1b and RIM2b that are identical to RIM1a and RIM2a but lack only the N-terminal Rab3-binding sequences (RIM1b) or both the N-terminal Rab3- and Munc13-binding sequences (RIM2b); and RIM2g, RIM3g, and RIM4g that are composed only of C2B domains and are not considered here further. Genetic experiments in C. elegans and mice revealed that RIMs are essential for synaptic vesicle docking and priming and for presynaptic plasticity (Castillo et al., 2002; Fourcaudot
et al., 2008; Gracheva et al., 2008; Kaeser et al., 2008; Koushika et al., 2001; Schoch et al., 2002, 2006), but their mechanism of action remains unclear. Several presynaptic proteins were shown to interact with Ca2+ channels. However, none of the reported interactions is selective for N- and P/Q-type Ca2+ channels. For example, (1) the RIM C2A and C2B domains bind to a1 subunits of L- and N-type Ca2+ channels (Coppola et al., 2001), (2) the RIM C2B domain interacts with the b4 Ca2+ channel subunit (Kiyonaka et al., 2007) that associates with all Ca2+ channels subtypes but is not required for neurotransmitter release (Qian and Noebels, 2000), and (3) the proline-rich sequences of RIMs bind to RIM-BPs (Wang et al., 2000) that in turn bind to L-, N-, and P/Q-type Ca2+ channels (Hibino et al., 2002). Thus, no plausible hypothesis at present suggests how N- and P/Q-type Ca2+ channels are selectively recruited to presynaptic active zones. More importantly, no presynaptic protein has been identified that is essential for recruiting N- and P/Q-type Ca2+ channels to presynaptic terminals. Only a-neurexins, which are presynaptic cell-adhesion molecules, were found to be required for presynaptic Ca2+-channel function (Missler et al., 2003). a-neurexins, however, are also essential for the organization of other components of the presynaptic release machinery, and no molecular mechanism is known that links a-neurexins to active zones or Ca2+ channels. Using an unbiased yeast two-hybrid screen, we here identify a direct interaction of P/Q- and N-type Ca2+ channels with RIM PDZ domains. To test whether RIMs act to localize Ca2+ channels to active zones, and whether this function requires the PDZ domains of RIMs, we generated conditional double knockout (KO) mice of all RIM isoforms that contain PDZ domains. Using these mice, we show by electrophysiological recordings, Ca2+ imaging, and quantitative immunostaining of Ca2+ channels that RIMs are essential for localizing Ca2+ channels to release sites. Moreover, we show that only two RIM sequences are required for localization of Ca2+ influx to active zones: their PDZ domains that bind to Ca2+ channels and their proline-rich sequences that bind to RIM-BPs, which in turn bind to Ca2+ channels. Thus, we propose that RIMs tether Ca2+ channels to active zones via two parallel essential interactions, direct binding of Ca2+ channels to the RIM PDZ domains that is specific for N- and P/Q-type Ca2+ channels and indirect binding of Ca2+ channels to RIMs via RIM-BPs that is shared among different types of Ca2+ channels.
RESULTS A Screen for Synaptic Proteins Binding to Ca2+ Channels We performed yeast-two hybrid screens for proteins that interact with the cytoplasmic C-terminal sequences of P/Qand N-type Ca2+ channels. We chose these Ca2+-channel baits because P/Q- and N-type Ca2+ channels mediate nearly all presynaptic Ca2+ influx, and their C termini have been implicated in targeting Ca2+ channels to the active zone (Catterall et al., 2005). The C termini of N- and P/Q-type Ca2+ channels contain three conserved sequence motifs (Figure 1A and Figure S1A available online): an SH3 domain-binding motif (RQLPGTP), a PNGY motif, and a C-terminal sequence motif
(DxWC). Among 84 and 134 independent prey clones obtained in P/Q- and N-type Ca2+-channel screens, 33 and 16 prey clones, respectively, represented RIM-BPs (Figure 1B), consistent with earlier studies (Hibino et al., 2002). In addition, 2 and 3 independent prey clones, respectively, contained RIM1 fragments, whose only overlapping sequence encoded the PDZ domain (Figure 1C). The direct interaction of the RIM1 PDZ domain with P/Q- and N-type Ca2+ channels was unexpected, prompting us to quantify it using liquid yeast two-hybrid assays. We found that the RIM1 PDZ domain strongly and specifically bound to P/Q- and N-type Ca2+ channels, whereas no other RIM1 domain tested exhibited Ca2+-channel binding activity (Figures 1D–1F). Mutations in the three conserved motifs of the cytoplasmic sequences of N- and P/Q-type Ca2+ channels showed that only the C-terminal sequence motif was essential for binding to RIM1 PDZ domains, as predicted for a PDZ-domain interaction (Figures 1E and 1F). To validate the interaction of RIM PDZ domains with Ca2+ channels by an independent method, we employed NMR spectroscopy. We produced a recombinant 15N-labeled RIM1 PDZ domain and acquired 1H-15N heteronuclear single quantum coherence (HSQC) spectra in the absence or presence of unlabeled peptide from the C terminus of the P/Q-type Ca2+ channel (residues RHDAYSESEDDWC). Previous studies of the 1H-15N HSQC spectrum of the RIM1 PDZ domain (Lu et al., 2005) allowed us to assign cross-peak shifts induced by the Ca2+channel peptide to specific residues in the RIM PDZ domain (see Extended Experimental Procedures, experimental rationale 1). We found that the C terminus of P/Q-type Ca2+ channels directly bound to the ligand-binding pocket of the RIM1 PDZ domain (Figures 1G and 1H and Figure S1B). Additional 1 H-15N HSQC experiments revealed analogous binding of the C terminus of N-type Ca2+ channels to the RIM1 PDZ domain, and of Ca2+ channels to the RIM2 PDZ domain (Figures S1C and S1D). An atomic model suggests that the Ca2+-channel sequence fits well into the PDZ-domain-binding pocket (Figures 1H and 1I). The binding envisioned in this model agrees with the shifts we observed in the HSQC spectra in cross-peaks from residues such as K651, K653, and K694, which contribute to the binding pocket of the RIM1 PDZ domain and move upon P/Q-type Ca2+channel peptide binding (Figure 1G). Furthermore, we confirmed these interactions by isothermal calorimetry with the RIM1 PDZ domain (Figures S1E and S1F), yielding dissociation constants similar to those of other PDZ-domain interactions (P/Q-type Ca2+ channel, 10.3 ± 0.6 mM; N-type Ca2+ channel, 23.4 ± 1.7 mM; Wiedemann et al., 2004). Thus, RIM PDZ domains stoichiometrically interact with the C-terminal sequences of N- and P/Q-type Ca2+ channels, which are not found in L- and T-type Ca2+ channels. Conditional KO Mice for Presynaptic RIM Isoforms The binding of RIM PDZ domains to Ca2+ channels was unexpected because RIM PDZ domains interact with the activezone protein ELKS (a.k.a., Rab6-binding protein, CAST, or ERC; Ohtsuka et al., 2002; Wang et al., 2002). To test whether the RIM PDZ domain physiologically binds to presynaptic Ca2+ channels, we generated conditional double KO mice in which Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc. 283
A
outside inside
PxxP I
II
III
PNGY DDWC (P/Q-type) DHWC (N-type)
IV
+
H3N
Ca2+-channel bait RIM-BP2
B Ca2+channel bait
P/Q-type
Ca2+channel bait
P/Q-type
PDZ
C 2A
PxxP
C2B
350 386
100 aa
48
56 65
N-type
1.0
0.5
n.d.
n.d.
0
RIM1 PDZ + P/Q-type Ca2+-channel
RIM1 PDZ
PDZ
1.6
0.8
n.d. 0
wild-type Ca2+-channel prey
ld-
RIM1 48 clone RIM1 56 clone RIM1 65 clone
RIM1 bait
PD C2 Z A C2 B C2 AB
pLexN
wi
n.d.
0
ΔCterm
PNGYM
interaction strength (a.u.)
2
F
N-type Ca2+-channel bait PxxPM
interaction strength (a.u.) RI M clo 1 pr ne ey s co ntr ol n.d. RI M clo 1 pr ne ey s co ntr ol n.d. RI M1 clo pr ne ey s co ntr ol RI M clo 1 pr ne ey s co ntr ol
interaction strength (a.u.) RI M1 clo pr ne ey s co ntr ol n.d. RI M clo 1 pr ne ey s co ntr ol
E
prey clones
typ e SH PN 3 M GY M ΔC t e co rm ntr ol
Zn
4
G
SH3 SH3
N-type
P/Q-type Ca2+-channel bait wild-type pLexN
D
FN3 FN3 FN3
prey clones
RIM1α
C
SH3
RIM1 PDZ
H
102
G662
N
G675 107
C G680 G646
S695
G667
D659
C αB
N βB
Q698 117
15
N (ppm)
112
T650
K651
122
N637
K653
I
E696
L629 A681 K694
L671 127
W673sc
K651 K633
I701 K653
G631
C
N674
K694
L691
132
N
K654 11
10
9
8
Y687
7
1
H (ppm)
Figure 1. Direct Interaction of P/Q- and N-Type Ca2+ Channels with RIM PDZ Domains (A) Structure of the a1 subunits of P/Q- and N-type Ca2+ channels. Following the 4 3 6 transmembrane regions (I–IV), P/Q-type and N-type Ca2+ channels contain a C-terminal cytoplasmic tail with conserved SH3-domain binding sequences (PxxP), PNGY motifs, and C-terminal sequence motifs (DxWC).
284 Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc.
all RIM isoforms containing PDZ domains (RIM1a, 1b, 2a, and 2b) can be deleted by cre-recombinase (Figures 2A–2C and Figure S2), thereby enabling us to avoid the lethality of RIM-deficient mice (Kaeser et al., 2008; Schoch et al., 2006). We cultured neurons from newborn conditional double KO mice and infected these neurons with lentiviruses expressing EGFP-tagged active or inactive cre-recombinases. Rescue experiments were performed by coexpressing various RIM proteins from the same lentiviruses via an IRES sequence (Kaeser et al., 2009). Immunoblotting demonstrated that after 10 days in vitro (DIV10), neurons expressing active cre-recombinase (referred to as cDKO neurons) lack RIM proteins, whereas neurons expressing inactive cre-recombinase (referred to as controls) retain RIM expression (Figure 2D). Despite lacking RIM proteins, cDKO neurons exhibited an overall normal morphology with unchanged synapse size and density (Figures 2E and 2F). Electron microscopy revealed that in cDKO neurons, the number of docked synaptic vesicles per active zone was decreased nearly 2-fold (Figures 2G and 2H), consistent with a role for RIM in vesicle docking (Gracheva et al., 2008; see also Han et al., 2011). All other measured parameters were unchanged, and removal of RIMs did not alter the structure of presynaptic dense projections visualized by phosphotungstic acid staining (Figure 2G). RIM Deletion Severely Impairs Neurotransmitter Release Electrophysiologically, deletion of RIM proteins caused a 3- to 4fold decrease in the frequency of spontaneous ‘‘minis,’’ in the amplitude of postsynaptic currents evoked by isolated action potentials, and in the size of the readily releasable pool measured by application of hypertonic sucrose (Figures 3A–3D and Deng et al., 2011). Moreover, deletion of RIMs significantly decelerated and desynchronized release, as evidenced by an increase in rise times and in rise-time variability (Figures 3E– 3G) but did not change the relative contributions of P/Q- and N-type Ca2+ channels to evoked synaptic responses (Figures S3A and S3B). Similarly, the RIM deletion massively decreased release induced by stimulus trains (30 stimuli at 10 Hz; Figures 3H and 3I), decelerated release as manifested by a relative increase in delayed release (Figures 3J and 3K), and again
strongly desynchronized release (Figures 3L and 3M). These data suggest that RIMs are essential not only for vesicle priming, a previously identified RIM function that partly accounts for the decrease in release (Koushika et al., 2001; Schoch et al., 2002), but also for the synchronous timing of fast release that is not accounted for by a priming deficit. Given that RIMs directly bind to Ca2+ channels (Figure 1), and that a loss of Ca2+ channels from presynaptic terminals would explain the impaired synchronous timing of release, we hypothesized that RIM binding to Nand P/Q-type Ca2+ channels may tether Ca2+ channels to the active zone, thereby increasing the efficiency, speed, and synchrony of release. RIM Deletion Alters the Ca2+ Dependence of Release To explore the hypothesis that RIMs localize Ca2+ channels to active zones, we first examined the speed with which addition of a membrane-permeable Ca2+ buffer (EGTA-AM) decreases Ca2+-triggered release, measured as the inhibitory postsynaptic current (IPSC) amplitude (see Extended Experimental Procedures, experimental rationale 2). EGTA-AM caused a significantly faster rate of decline in IPSC amplitude in RIM-deficient cDKO neurons than in control neurons (Figures 3N–3P). Thus, EGTAdependent chelation of Ca2+ inhibits release more effectively in RIM-deficient than in control neurons, consistent with a longer average distance between Ca2+ channels and release sites in RIM-deficient synapses. We next examined the dependence of release in RIM-deficient synapses on the extracellular Ca2+ concentration ([Ca2+]ex; Figure 4). If RIM-deficient presynaptic terminals contain fewer tethered Ca2+ channels, presynaptic Ca2+ influx should be decreased, and more [Ca2+]ex should be required for equivalent amounts of release—i.e., release should exhibit a higher [Ca2+]ex dependence without a change in apparent Ca2+ cooperativity (see Extended Experimental Procedures, experimental rationale 3). However, because Ca2+ influx through Ca2+ channels saturates at high [Ca2+]ex (Church and Stanley, 1996; Schneggenburger et al., 1999), [Ca2+]ex titrations underestimate the change in Ca2+ dependence of release in mutant synapses, and only relative changes are interpretable. RIM-deficient cDKO neurons exhibited a large reduction in neurotransmitter release at all [Ca2+]ex (Figures 4B and 4C) and
(B and C) Summary of the RIM-BP (B) and RIM prey clones (C) isolated in yeast two-hybrid screens with the C-terminal sequences of N- and P/Q-type Ca2+ channels. (D–F) Liquid yeast-two hybrid assays with baits containing wild-type C-terminal sequence of the P/Q-type Ca2+ channel and the three indicated RIM prey clones (D); baits containing the C-terminal sequence of the N-type Ca2+ channel with point mutations in the PxxP (PxxPM) or the PNGY sequence (PNGYM), or with a deletion of the 4 C-terminal residues (DCterm), and the indicated RIM prey clones (E); and baits containing the indicated RIM1 domains (PDZ, PDZ domain only; C2A or C2B, C2A or C2B domains only; C2AB, both C2 domains and the intercalated PxxP motif) and preys consisting of the wild-type C-terminal sequence of the N-type Ca2+ channel (F, left bars), or the indicated mutants of this Ca2+ channel (F, right bars). For all assays, pLexN served as a control; a.u. = arbitrary units; n.d. = not detectable (means ± standard error of the mean [SEM]). (G) Analysis of P/Q-type Ca2+ channel binding to the RIM1 PDZ domain by NMR spectroscopy. 1H-15N HSQC spectra of the 15N-labeled RIM1 PDZ domain (38 mM) were acquired in the absence (black contours) and presence (blue contours) of unlabeled P/Q peptide (0.1 mM). Selected cross-peak assignments from residues on the periphery of the binding site are indicated; cross-peaks from three lysine residues in the binding pocket that shift upon peptide binding are labeled in bold, underlined typeface (K651, K653, and K694). (H) Model of the RIM1 PDZ domain (blue ribbon diagram; orange indicates residues corresponding to shifted cross-peaks assigned in panel G) bound to the six C-terminal residues of the P/Q-type Ca2+ channel peptide represented as a stick model with color-coded atoms (carbon, yellow; oxygen, red; nitrogen, blue; sulfur, orange). Strand bB and helix aB, the two structural elements that line the peptide-binding site (Lu et al., 2005), are indicated. (I) Close-up view of the surface of the RIM1 PDZ domain peptide-binding pocket with the bound P/Q-type Ca2+ channel peptide (colors are identical to panel H). For additional 1H-15N HSQC spectra and affinity measurements by isothermal titration calorimetry, see Figure S1.
Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc. 285
A
RIM2αβγ floxed or KO 1’
2-4
5
1’’
2α
7-25
1’’’
2β
1’’’
B
6
2γ
100 kb 27 - 29
26
wild-type allele targeting construct original KI allele RIM2αβγ
26-30
N
C
N
C
DT
floxed
C
RIM2αβγ KO 2 kb
RIM1αβfloxed or KO
Zn
PxxP PDZ
1β
C2A
C2B
RIM2αβγ floxed or KO
Zn
PxxP
ol
14
ntr
cD
ntr
KO
ol
12
co
cD
ntr
ol
10
co
cD
ol
8
ntr
cD
co
ntr
ol
6
KO
DIV
KO
D
C 2B
2γ
KO
2β
C 2A
cD
PDZ
co
2α
200 amino acids
KO
1α
co
C
RIM1/2 VCP β-actin
0 .4 0 .2
2
ntr cD ol KO
co PSD length (μm)
3 2 1 0
20
0 0.4
0.2
0.0
286 Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc.
o cD l KO
0
40
ntr
PSD
2
o cD l KO
PSD
DP
bouton circumference (μm)
DP
** *
4
vesicles per bouton
H
ntr
cDKO
co
control
4
0
0 .0
docked vesicles per active zone
G
0 .6
co ntr c D ol KO
merge
Syn
puncta size (μm2)
F
co
cDKO
puncta per 10 μm dendrite
control
MAP-2
E
Figure 2. Conditional Deletion of RIM Proteins in Mice (A) Structure of the RIM2 gene (a.k.a., Rims2). Exons are shown as black boxes and numbered, positions of exons containing the initiator codons for RIM2a, RIM2b, and RIM2g are labeled 10 , 100 , and 10 00 , respectively. The first exon that is shared by all RIM2 isoforms (exon 26) was used for gene targeting in the conditional RIM2abg KO mice (shaded blue area). (B) RIM2abg targeting strategy. The diagram shows (from top to bottom) an expanded map of the RIM2 gene surrounding exon 26; the targeting vector (C = ECFP-tetracysteine tag in exon 26; blue triangles = loxP sites; N = neomycin resistance cassette; green circles = frt recombination sites; DT = diphtheria toxin gene cassette); the knockin allele (KI); the RIM2abgfloxed allele (neomycin resistance cassette was removed by flprecombination); and the KO allele (cre recombination deleted exon 26, creating a nontranslated, unstable mRNA). (C) Domain structures of RIM1a, 1b, 2a, 2b, and 2g that are deleted in the RIM1/RIM2 conditional double KO neurons. Coils surrounding the N-terminal Zn2+-finger domain (Zn) signify Rab3-binding sequences. (D) Representative immunoblots of RIM1 and RIM2 proteins in cultured hippocampal neurons from RIM1/RIM2 double conditional KO mice infected with lentiviruses expressing inactive (control) or active crerecombinase (cDKO). Neurons were infected on DIV3 and analyzed at the indicated times (DIV6–DIV14). (E) Representative images of cDKO and control neurons stained with antibodies to MAP2 (green) and synapsin (red). Scale bar = 5 mm, applies to all images. (F) Quantitations of the size and density of synapses analyzed as shown in (E) (control, n = 18 neurons/3 independent cultures; cDKO neurons, 17/3). (G) Electron micrographs of osmium tetroxide- (top) or phosphotungstic acid-stained (bottom) control and cDKO neurons (scale bars, 200 nm). (H) Quantitations of synaptic ultrastructure in electron micrographs. Docked vesicles are defined as vesicles touching the plasma membrane (DP, dense presynaptic projections; PSD, postsynaptic density). Data in (F) and (H) show means ± SEM. Statistical significance by Student’s t test: ***p < 0.001; for additional detailed information, see Figure S2 and Table S1.
B
cDKO
Sucrose
charge transfer (nC)
16
3
12
2
cDKO
Sucrose
co 0.6
2
1 0
co
cD
0.4
co
co
M
cDKO
control
SD of 20-80% rise time (ms)
5
#2-10
AP number
0 min 6 min 12 min control
cDKO 0.5 nA 100 ms
30
O
KO
15
P 1.0
control
0.8 0.6 0.4
cDKO
0.2 0.0
0
4
8
time (min)
12
6
*
4 2 0
KO
EGTA-AM
0 0
AP number normalized IPSC ampl.
N
1
10
cD
30
2
co
15
3
ol
10
20
4
ol
20
0.2 nA 10 ms
ntr
IPSC 20-80% rise time (ms)
IPSC 20-80% rise time (ms)
#1
0
**
#1
#2-10
1 nA 10 ms
0
ntr o cD l KO
0.0
10
ntr o cD l KO
ntr o cD l KO
0.0
** *
(A and B) Excitatory synaptic responses in cultured hippocampal control and cDKO neurons evoked by an action potential (A) or hypertonic sucrose application (B) (left, representative traces; right, summary graphs of amplitudes or charge transfers; A: control, n = 8 neurons/3 independent neuronal cultures; cDKO, n = 9/3, B: control, n = 9/3; cDKO, n = 10/3). (C and D) Inhibitory synaptic responses evoked by an action potential (C) or hypertonic sucrose application (D) (C: control, n = 21/4; cDKO, n = 18/4; D: control, n = 11/3; cDKO, n = 11/3). (E–G) Analysis of the kinetics of isolated IPSCs (E, superposed representative traces from control and cDKO neurons; F, 20%–80% IPSC rise times; and G, rise time variability as expressed by the standard deviation (SD) of the 20%– 80% rise time [control, n = 31/6; cDKO, n = 36/6]). (H–K) Synaptic responses elicited by a 10 Hz stimulus train in cDKO and control neurons (H, representative IPSCs; I–K, summary graphs of the synaptic charge transfer for the first IPSC [I] and for delayed release [J; release starting 100 ms after the last stimulus] and of the ratio of delayed release/first response) (K; control, n = 20/4; cDKO, n = 21/4). (L and M) Analysis of the kinetics of IPSCs during 10 Hz stimulus trains (L, superposed representative traces of the first 10 IPSCs during a 10 Hz stimulus train applied for 3 s [top, first response indicated as a thick line, later responses represented as thin lines], and 20%–80% rise times for three sample trains [bottom]; M, standard deviation (SD) of the 20%–80% rise times during the 10 Hz stimulus train as a measure of synchrony; control, n = 7/3; cDKO, n = 9/3). (N–P) Time course of the decrease in IPSCs induced by addition of the membrane-permeable Ca2+-chelator EGTA-AM (10 mM; N, sample traces; O, summary graphs; P, decay time constants; control, n = 8/3; cDKO, n = 9/3). Decay time constants t were calculated by fitting individual experiments to a single exponential function. All data are means ± SEMs; *p < 0.05, **p < 0.01, ***p < 0.001 as determined by Student’s t test. Numerical values of electrophysiology results are in Table S2, further analysis of synaptic responses elicited at 10 Hz in Figure S3.
**
20
co
** *
0.8
KO
ol ntr
ol KO
0.2
K 1.2
ntr
first peak charge (nC)
J 0.3
delayed release (nC)
I
1s
0.2 0.0
co
5 ms
0.5 nA
0.4
ntr
0.1 nA
0.1
***
G
***
3
SD of 20-80% rise time (ms)
F
20-80% rise time (ms)
co
5s
cDKO
cDKO
ntr
0.5 nA
0.5 nA
control
0
o cD l KO
0
** *
4
cD
1
8
cD
** *
control
L
ntr
control
Sucrose
5 ms
H
co
IPSCs
D
4
100 ms
E
o cD l KO
ntr o cD l KO
5s
ntr o cD l KO
0.5 nA
** *
2
0.25 nA
co
cDKO
IPSC amplitude (nA)
control
4
0
0.0
IPSCs
C
charge transfer (nC)
** *
6
co
0.1 nA 10 ms
Sucrose 0.4
0.2
Figure 3. RIM Deletion Decreases, Decelerates, and Desynchronizes Neurotransmitter Release
EPSCs control
delayed release/ 1st peak charge ratio
cDKO
EPSC amplitude (nA)
EPSCs control
decay time const. (min)
A
a major shift in the [Ca2+]ex dependence of neurotransmitter release (Figure 4D). Both phenotypes were equally observed in inhibitory and excitatory synapses and fully rescued by fulllength wild-type RIM1a (Figures 4A–4D and Figures S4A–S4N). Fitting the [Ca2+]ex-response curve of individual experiments to a Hill function showed that the RIM deletion increased the [Ca2+]ex requirement for release almost 2-fold, without changing the apparent Ca2+ cooperativity of release (Figure 4E and Figure S4H). Note that in the Hill function fits, the experimentally measured amplitudes suggest near saturation at 10 mM [Ca2+]ex, allowing direct comparison of the fitted parameters (Figure S4I). The [Ca2+]ex titration provided us with a facile assay to examine which RIM sequences determine vesicle priming and the [Ca2+]ex dependence of release. We first tested rescue of the RIM double KO phenotype by the N-terminal RIM-RZ fragment that contains the Rab3-binding (‘‘R’’) and Zn2+-finger domains (‘‘Z’’) of RIM1a and that had been previously implicated in vesicle priming (Dulubova et al., 2005), and the C-terminal RIM-PASB fragment that contains its PDZ (‘‘P’’), C2A (‘‘A’’), PxxP (‘‘S’’ for SH3 domain binding), and C2B domains (‘‘B’’) (Figure 4A). At 2 mM [Ca2+]ex, the RIM-RZ and RIM-PASB fragments each partially rescued the decrease in release and alleviated the previously described decrease in Munc13 protein levels in cDKO neurons (Figures 4F and 4I; Figure S4D; Schoch et al., 2002). However, whereas the RIM-PASB fragment completely reversed the impairment of the [Ca2+]ex dependence of release in cDKO neurons, the RIM-RZ fragment did not (Figures 4F–4K). RIMRZ, conversely, rescued the readily releasable pool, whereas Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc. 287
R Z
A RIM1α RIM-RZ
Zn Zn H
RIM-PASB
B
[Ca2+]ex (mM):
0.5
P
A
S
B
PDZ
C 2A
PxxP
C2 B
C 2A
PxxP H
PDZ
1
2
C2B
5
10
control cDKO cDKO + RIM1α cDKO + RIM-RZ cDKO + RIM-PASB
2 nA 250 ms
E
2 cDKO 0
0.6 control 0.4 cDKO + RIM1α
0.2 cDKO
0.0 10
1
2+
[Ca ]ex (mM)
cDKO + RIM-RZ cDKO
H 1.0 0.8
control
0.6 cDKO + RIM-RZ
0.4
cDKO
0.2 0.0
0 10
1
2+
[Ca ] ex (mM)
2 cDKO
K 0.8
cDKO + RIM-PASB control
0.6 0.4
*
3
*** ** **
2 1
co
ntr o cD l K cD O RIMKO 1α + co ntr o cD l cD KO RIMKO -R + co Z ntr cD ol cD KO RIM K -PAO + SB
rise time (ms)
L
1
[Ca2+] ex (mM)
M 0.8
***
***
3 2
0
10
*** **
**
0.4
0.0
RIM Deletion Impairs Presynaptic Ca2+ Influx The increased [Ca2+]ex dependence, decreased synchrony, and lowered speed of release in RIM-deficient neurons support the hypothesis that RIMs localize Ca2+ channels to the active zone. However, these measurements are indirect. Their results could be explained by other hypotheses, for example that RIM directly regulates Ca2+ triggering of release by binding to synaptotagmin (Coppola et al., 2001; Schoch et al., 2002). To address this issue with an independent approach, we monitored actionpotential-induced Ca2+ transients by Ca2+ imaging in presynaptic boutons and postsynaptic dendrites. We engineered new active and inactive EGFP-tagged crerecombinase proteins that exhibit a tight nuclear localization (Figures S5A–S5C), and expressed them in neurons from conditional KO mice. We then filled single neurons via a patch pipette with Alexa 594 and the Ca2+ indicator Fluo5F, identified presynaptic axonal boutons and second-order dendrites by imaging Alexa 594, elicited isolated action potentials by brief somatic current injections, and monitored the resulting Ca2+ transients in boutons and dendrites by imaging Fluo5F in line scans (at
1
ntr o cD l K cD O RIMKO 1α + co ntr o cD l cD KO RIMKO -R + co Z ntr cD ol cD KO RIM K -PAO + SB
10
SD of rise time (ms)
1
[Ca2+] ex (mM)
0
cDKO
0.2 0.0
0
0
10
EC50 of [Ca2+] ex (mM)
cDKO + RIM-PASB
4
normalized IPSC
IPSC amplitude (nA)
control
1
[Ca ] ex (mM)
1.0
6
2
2+
J
I
******
3
co
1
0
co
2
1
co
IPSC amplitude (nA)
control
normalized IPSC
G
2
co
[Ca ] ex (mM)
F 4
10 2+
EC50 of [Ca2+] ex (mM)
1
***
3
ntr cD ol K cD O RIMKO -R + Z
cDKO + RIM1α
0.8
ntr o cD l KO c RIMDKO -PA + SB
control 4
ntr o cD l KO cD RIMKO + 1α
1.0
EC50 of [Ca2+] ex (mM)
D normalized IPSC
IPSC amplitude (nA)
C
RIM-PASB did not (Deng et al., 2011). Note that different from synaptotagmin mutations (Shin et al., 2009), none of the RIM manipulations changed the apparent Ca2+ cooperativity of release (Figure S4H; see also Table S3). We next tested whether RIM-RZ or RIM-PASB rescued the speed and synchrony of release. Consistent with the [Ca2+]ex titrations, the RIM-PASB fragment fully reversed the deceleration and desynchronization of release in cDKO neurons, as assessed at 2 mM [Ca2+]ex, whereas the RIM-RZ fragment did not (Figures 4L and 4M and Figures S4O–S4Q). Together, these data indicate that the N-terminal RIM domains function in vesicle docking and priming (Betz et al., 2001; Dulubova et al., 2005; Gracheva et al., 2008; Koushika et al., 2001; Schoch et al., 2002), whereas the C-terminal RIM domains function in the Ca2+ dependence and synchrony of release.
Figure 4. Mutational Dissection of the RIM KO Phenotype (A) Diagram of RIM rescue proteins expressed in cDKO neurons via an IRES sequence from the same mRNA as cre-recombinase. The single-letter code above the RIM1a diagram identifies the various domains (R, Rab3-binding a-helical region; Z, Zn-finger region, P, PDZ domain; A, C2A domain; S, proline-rich SH3-binding PxxP motif; B, C2B domain); H marks the presence of a human influenza hemagglutinin (HA)-tag.
288 Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc.
(B) Representative traces of IPSCs evoked at the indicated extracellular Ca2+ concentrations [Ca2+]ex in control neurons, cDKO neurons without rescue, cDKO neurons with full-length RIM1a rescue, and cDKO neurons with rescue with the RIM-RZ or the RIM-PASB fragments. Each rescue experiment was performed with independent control groups. (C–K) Summary plots of absolute (C, F, and I) and normalized IPSC amplitudes (D, G, and J; normalized to the 10 mM [Ca2+]ex response) evoked at the indicated [Ca2+]ex, and summary graphs of the Ca2+ dependence of release (E, H, and K; expressed as the [Ca2+]ex producing a half-maximal IPSC amplitude [EC50], as determined by fitting in individual experiments the [Ca2+]ex dependence of the IPSC amplitude [C, F, and I] to a Hill function). Control neurons and cDKO neurons were analyzed in comparison with cDKO neurons rescued with RIM1a (C–E), RIM-RZ (F–H), or RIM-PASB (I–K). (C and D): n = 8 neurons/3 independent batches of culture in control, 6/3 in cDKO, 9/3 in cDKO + RIM1a; (E and F): n = 7/3 in control, 7/3 in cDKO, 7/3 in cDKO + RIM-RZ; (G and H): n = 6/3 in control, 5/3 in cDKO, 8/3 in cDKO + RIM-PASB. (L and M) Summary graphs of 20%–80% rise times (L) and rise time variability (M) for the indicated rescue experiments at 2 mM [Ca2+]ex (for sample traces, see Figure S4O; for number of neurons/independent cultures analyzed, see panels C–K). Data shown are means ± SEM, ***p < 0.001 by one-way ANOVA; detailed statistical analysis for all data points can be found in Table S3. For Ca2+ cooperativity and Imax, see Figure S4.
A
control
cDKO
cDKO + RIM-PASB
Figure 5. RIM Deletion Decreases Presynaptic Ca2+ Transients
ΔG/G0
cumulative probability
ΔG/G0 (% change)
***
co ntr ol cD KO
Fluo5F
GFP/Fluo5F/Alexa594
(A) Representative fluorescence images of control neurons, cDKO neurons, and cDKO neurons rescued with the C-terminal RIM-PASB fragment. Neurons were filled via a patch pipette with Fluo5F and Alexa 594 (red); nuclear EGFP-fluorescence (produced by the active and inactive cre-recombinase EGFP-fusion proteins; see Figures S5A– S5C) is shown in green; and coincident Alexa 594 and EGFP- or Fluo5F signals are shown in yellow. Insets (bottom right) show areas in dotted rectangles containing a sample axonal bouton (gray lines = positions B of the patch pipette; white lines = positions of line scans for 20 mV the Ca2+ transients shown in B). Scale bar (bottom left) = 50 ms -70 mV 20 mm. (B) Representative action potentials (top); line scans of Ca2+ transients in presynaptic boutons induced by these action potentials and monitored via Fluo5F fluorescence 50% ΔG/G0 (middle; colored white for better visibility); and quantita50 ms tions of Ca2+ transients (bottom; averaged across the bouton). (C and D) Summary plots of action potential-induced changes in Ca2+-indicator fluorescence monitored in bouton dendrite bouton C D E n.s. presynaptic boutons from control neurons, cDKO neurons, 80 202 60 120 and cDKO neurons rescued with the C-terminal RIM-PASB 1.0 RIM1/2 40 fragment (C, time course of the Ca2+-indicator fluorescence 100 20 CaV2.1-A 202 [inset: the same plot for dendrites]; D, the cumulative 0.8 0 0 200 400 600 80 probability of the peak Ca2+-indicator fluorescence, exCaV2.1-S 202 time (ms) 0.6 pressed as DG/G0). Data in (C) are means (line) ± SEM CaV2.2 60 control 202 (shaded area); ***p < 0.001 as assessed by two-way ANOVA cDKO CaVα2/δ1 0.4 cDKO + for peak amplitudes during the first 60 ms after action 40 116 RIM-PASB potential induction (C) or by Kolmogorov-Smirnov test (D); control CaVβ4 20 50 0.2 cDKO control, n = 45 boutons/10 neurons/4 independent cultures; Liprin-α3 cDKO + cDKO, n = 46/11/4; cDKO + RIM-PASB, n = 44/11/4. RIM-PASB 116 0 87 GDI 0.0 (E) Immunoblot analysis of Ca2+-channel subunit levels in 0.0 0.5 1.0 1.5 2.0 50 0 100 200 300 400 500 600 control and cDKO neurons. Blots were probed with antiβ-actin ΔG/G0 peak amplitude time (ms) bodies to the indicated Ca2+-channel proteins (P/Q-type [CaV2.1-A and CaV2.1-S] and N-type [CaV2.2] a subunits, and a2/d1 and b4 subunits) and control proteins (GDI, GDP-dissociation inhibitor); numbers indicate positions of molecular weight markers. For analysis of dendritic Ca2+ transients, statistical values, and quantitative assessment of mRNA levels, see Figure S5 and Table S4.
***
333 Hz, 100–150 mm from the cell body). To ensure that the observed Ca2+ transients were not due to passive depolarizations in response to the somatic current injections, we decreased in control experiments the injected current to a threshold level that only sometimes evoked an action potential. In these experiments, Ca2+ transients in boutons and dendrites strictly depended on the induction of action potentials, confirming that we were monitoring action-potential-induced Ca2+ transients (Figures S5D and S5E). In RIM-containing control boutons, isolated action potentials induced a brief, 100% increase in Fluo5F Ca2+-indicator fluorescence, whereas in RIM-deficient cDKO boutons, action potentials induced only an 50% increase (Figures 5A–5D). Similar to the impaired [Ca2+]ex dependence of release (Figure 4), the decreased Ca2+ influx in RIM-deficient cDKO neurons was fully rescued by the C-terminal RIM-PASB fragment (Figures 5A–5D). Deletion of RIM proteins did not alter dendritic Ca2+ transients (Figure 5C, inset and Figures S5F–S5H), suggesting that the RIM deletion did not generally impair Ca2+-channel function or Ca2+ buffering. Moreover, we detected no change in expression levels of various Ca2+-channel subunits in cDKO neurons (Figure 5E and Figure S5I). Thus, RIM deletions selectively
decreased presynaptic Ca2+ influx in hippocampal neurons. A parallel study extends this finding by direct measurements of Ca2+ currents in RIM-deficient calices of Held (Han et al., 2011). RIM PDZ Domain Is Required for Localizing Ca2+ Influx We next tested whether the RIM PDZ domain mediates the RIMdependent presynaptic localization of Ca2+ influx, using systematic rescue experiments combined with [Ca2+]ex titrations as an assay. Deletion of either the PDZ domain or the PxxP motif from the RIM-PASB fragment (Figure 6A, top) blocked rescue of the impaired [Ca2+]ex dependence in RIM-deficient cDKO neurons, whereas deletion of the C2A or the C2B domains had no effect (Figures 6A–6E and Figures S6A–S6C). Moreover, expression of a RIM fragment composed only of the PDZ domain and PxxP motif (‘‘RIM-PS’’) in cDKO neurons fully rescued the impaired [Ca2+]ex dependence of release (Figures 6F–6H and Figures S6D–S6F). In contrast, expression of a RIM fragment composed of its two C2 domains (‘‘RIM-AB’’) did not rescue the [Ca2+]ex dependence of release but partially reversed the decrease in IPSC size in RIM-deficient cDKO neurons, Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc. 289
RIM-ASB
B
PxxP
C 2B
C2A
PxxP H
PDZ PDZ
C2A
RIM-PAS
PDZ
C2A
RIM-PS
PDZ
C 2B
C2B
PxxP H
RIM-PSB RIM-PAB
H
C 2B
PxxP H PxxP H
C2A
RIM-AB
H
C 2B
control RIM-PSB RIM-PAB RIM-PAS
4
RIM-ASB
2
cDKO
0.8 control 0.6 0.4
RIM-PSB RIM-PAS RIM-ASB RIM-PAB cDKO
0.2 0.0
10
*** ***
1
10
[Ca2+]ex (mM)
[Ca ]ex (mM)
10
2
0 1
2+
***
3
co
1
5
2 nA 250 ms
E 1.0
0
2
cDKO cDKO + RIM-ASB cDKO + RIM-PSB cDKO + RIM-PAB cDKO + RIM-PAS cDKO + RIM-PS cDKO + RIM-AB
D
6
normalized IPSC
IPSC amplitude (nA)
control
S
C2A
1
RIM -A RIM SB -PS RIM B -P RIM AB -PA S
PDZ
[Ca2+] ex (mM): 0.5
ntr ol cD KO
R Z
Zn
C
B
RIM1α P A
EC50 of [Ca2+] ex (mM)
A
cDKO + rescue
0.6
control cDKO
0.4
RIM-AB
0.2 0.0
0 1
[Ca2+]ex (mM)
10
1
10
[Ca2+]ex (mM)
suggesting that the C2 domains are dispensable for localizing presynaptic Ca2+ influx but enhance the efficacy of release. To directly probe whether the RIM PDZ domain is actually essential for localizing presynaptic Ca2+ influx to active zones, we next performed rescue experiments with full-length wild-type RIM1a or RIM1a lacking the PDZ domain (RIM-DPDZ). Whereas the former rescued all RIM cDKO phenotypes (Figures 4C–4E), the latter was unable to reverse the impairment in [Ca2+]ex dependence of release in RIM-deficient neurons, although it completely rescued the decrease in evoked IPSC amplitude at high [Ca2+]ex concentrations (Figures 7A–7E and Figures S7A–S7C). Moreover, full-length but not PDZ-domain-deleted RIM1a rescued the deceleration and desynchronization of release in RIM-deficient neurons (Figures 4L and 4M and Figures 7F–7H). Possibly most importantly, RIM1a lacking the PDZ domain did not restore normal Ca2+ influx into presynaptic nerve terminals in RIM-deficient neurons as measured by Ca2+ imaging, whereas full-length RIM1a fully rescued Ca2+ influx (Figures 7I and 7J and Figures S7D and S7E). Thus, the RIM-PDZ domain is critical for localizing Ca2+ influx to presynaptic active zones, thereby promoting the normal Ca2+ dependence, speed, and precision of neurotransmitter release. 290 Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc.
***
*** 2
1
0
The RIM PDZ Domain Localizes P/Q-Type Ca2+ Channels to Presynaptic Terminals The decrease in presynaptic Ca2+ influx in RIM-deficient terminals could be due to a loss of presynaptic Ca2+ channels or cDKO + rescue to a decrease in their activity. To address this question, we measured by quantitative immunofluorescence presynaptic levels of P/Q-type Ca2+ channels, which mediate >80% of the synaptic responses in our preparation (Figures S3A–S3C). Strikingly, the RIM deletion reduced presynaptic P/Q-type Ca2+ channel levels 40% but had no effect on the active-zone protein bassoon (Figures 8A and 8B and Figures S8A and S8B). Finally, to test whether the presynaptic localization of P/Q-type Ca2+ channels depends on the RIM PDZ domain, we examined rescue of the Ca2+-channel localization deficit in RIM-deficient cDKO neurons with full-length wild-type RIM1a or RIM1a lacking the PDZ domain (Figures 8A and 8B). In agreement with the electrophysiological and Ca2+-imaging results described above (Figure 7), deletion of the PDZ domain rendered RIM1a unable to rescue the decrease in Ca2+-channel levels in presynaptic terminals (Figures 8A and 8B), suggesting that the RIM PDZ domain is essential for tethering of Ca2+ channels to presynaptic terminals. -P RIM S -AB
cDKO
0.8
(A) Domain structures of rescue proteins. (B) Sample traces of IPSCs in control neurons, cDKO neurons, and cDKO neurons rescued with the indicated proteins. (C–E) Systematic rescue analyses of the [Ca2+]ex dependence of release in RIM-deficient cDKO neurons with RIM fragments containing three of the four RIM domains present in the RIM-PASB fragment. Absolute IPSC amplitudes (C), IPSC amplitudes normalized to the response at 10 mM [Ca2+]ex (D), and apparent Ca2+ affinities (EC50 values; E) are indicated (control, n = 9 cells/3 independent batches of cultures; cDKO, n = 9/3; cDKO + RIM-ASB, n = 9/3; cDKO + RIM-PSB, n = 8/3; cDKO+RIM-PAB, n = 8/3; cDKO + RIM-PAS, n = 10/3). (F–H) Rescue analyses of the Ca2+ dependence of release with RIM fragments containing either only the PDZ domain and PxxP motif (RIM-PS), or only the C2A and C2B domains (RIM-AB) of RIM1 (control, n = 8/3; cDKO, n = 7/3; cDKO + RIM-PS, n = 9/3; cDKO + RIM-AB, n = 9/3). Data shown are means ± SEM; ***p < 0.001 by one-way ANOVA. Cooperative factor n and Imax can be found in Figure S6; all numerical data are in Table S5.
RIM
2
H
RIM-PS
ntr ol cD KO
RIM-AB RIM-PS
1.0
EC50 of [Ca2+] ex (mM)
4
normalized IPSC
IPSC amplitude (nA)
control
co
G
F
Figure 6. RIM PDZ Domain and PxxP Motif Restore Normal [Ca2+]ex Dependence to RIM-Deficient Synapses
DISCUSSION Based on protein/protein interaction studies (Figure 1), generation of conditional KO mice (Figure 2), electrophysiological
Zn RIM-ΔPDZ Zn
A
PDZ
RIM1α
B
[Ca2+]ex (mM): control
1
0.5
C2A
PxxP
C2B
C2A
PxxP
C2B
2
10
5
cDKO
cDKO + RIM-ΔPDZ
5 4
3 control 2 cDKO
1
0.8
E cDKO + RIM-ΔPDZ
0.6 control 0.4 0.2 cDKO
0
1
rise time (ms)
G 1 nA 5 ms
1 0
H
** **
2
2
10
[Ca2+] ex (mM)
control
*** ***
3
co
[Ca2+]ex (mM)
0.0
10
SD of rise time (ms)
1
F
1.0
ntr o cD l K RI cDK O M- O ΔP + DZ
D
6
normalized IPSC
IPSC amplitude (nA)
C
1 nA 0.1 s
EC50 of [Ca2+] ex (mM)
cDKO + RIM-ΔPDZ
*
0.6
*
recordings (Figure 3, Figure 4, Figure 6, and Figure 7), Ca2+ imaging (Figure 5 and Figure 7), and quantitative immunofluorescence (Figure 8), we here propose that the PDZ domains of RIM proteins stoichiometrically interact with N- and P/Q-type Ca2+ channels in vitro, that RIM proteins are essential for tethering Ca2+ channels to presynaptic terminals in vivo, and that the RIM PDZ domain is required for this function. In addition, RIMs indirectly interact with Ca2+ channels via their RIM-BP binding sequence (Hibino et al., 2002), which we show is also essential. Thus, RIMs perform two parallel interactions with Ca2+ channels: a direct interaction via their PDZ domains that is specific for N- and P/Q-type Ca2+ channels and an indirect interaction via RIM-BPs that is not specific for N- and P/Q-type Ca2+ channels. Our results suggest a physiologically validated mechanism by which Ca2+ influx is localized to the active zone, as required for fast, synchronous triggering of neurotransmitter release (Figure 8C) (also see Han et al., 2011).
Figure 7. RIM-Dependent Localization of Presynaptic Ca2+ Influx Requires the RIM PDZ Domain
The RIM PDZ-Domain/Ca2+-Channel Interaction We found that the C-terminal sequences of N- and P/Q-type Ca2+ channels specifically bind to RIM PDZ domains. R-type Ca2+ channels have a similar C-terminal sequence and may also interact, whereas L- and T-type Ca2+ channels do not (Figure 1 and Figure S1). The PDZ-domain/Ca2+ -channel interaction was surprising because the PDZ-domain proteins Mints and CASK were previously shown to bind to Ca2+ channels (Maximov et al., 1999), and because the RIM PDZ domain is known to bind to ELKS proteins (Ohtsuka et al., 2002; Wang et al., 2002). However, it remains unclear whether these previously described interactions are physiologically important; in fact, Mint- and CASK-deficient synapses exhibit multiple abnormalities that do not resemble a Ca2+-influx impairment (Atasoy et al., 2007; Ho et al., 2006), whereas ELKS2-deficient synapses display increased release at inhibitory synapses (Kaeser et al., 2009), and ELKS levels are not detectably changed in RIM cDKO neurons (data not shown). Thus, it seems unlikely that the effects we observe here are indirectly mediated via Mints, CASK, or ELKS. It is possible, however, that different PDZdomain-binding reactions compete with each other at the active zone. For example, ELKS binding to RIM PDZ domains may inhibit Ca2+-channel binding and thereby attenuate neurotransmitter release; this inhibitory role of ELKS binding could be regulated during plasticity, which might account for the central role of RIM in short- and long-term plasticity (Castillo et al., 2002; Fourcaudot et al., 2008; Kaeser et al., 2008; Schoch et al., 2002).
(A) Domain structures of rescue proteins. (B–E) Sample traces and quantitative analysis of [Ca2+]ex dependence of release of IPSCs in control neurons, cDKO neurons, and cDKO neurons rescued with the PDZ-domain-deficient RIM-DPDZ fragment. Absolute IPSC amplitudes (C), IPSC amplitudes normalized to the response at 10 mM [Ca2+]ex (D), and apparent Ca2+ affinities (EC50 values; E) are indicated (control, n = 10/ 3; cDKO, n = 9/3; cDKO + RIM-DPDZ, n = 10/3). (F–H) Speed and synchrony of neurotransmitter release in control neurons, cDKO neurons and cDKO neurons rescued with RIM-DPDZ (control, n = 7/3; cDKO, n = 10/3; cDKO + RIM-DPDZ, n = 12/3). (I and J) Sample line scans (I) and summary data (J) of action potential-evoked
Ca2+ transients in presynaptic boutons of control neurons, cDKO neurons, and cDKO neurons rescued with RIM1a or RIM-DPDZ. (Boutons: control, n = 40 boutons/6 neurons/5 independent cultures; cDKO, n = 52/7/5; cDKO + RIM1a, n = 51/7/5; cDKO + RIM-DPDZ, n = 52/7/5; dendrites: control, n = 22/6/5; cDKO, n = 22/7/5; cDKO + RIM1a, n = 19/7/5, cDKO + RIM-DPDZ, n = 22/7/5.) For cumulative peak amplitudes and statistical values, see Figure S7 and Table S6. Statistical analyses: *p < 0.05; **p < 0.01; ***p < 0.001; (E, G, and H) One-way ANOVA; (J) Two-way ANOVA for peak amplitudes during the first
control
0.2 0.0
co
ntr
0.5 nA 5 ms
I
0
ntr ol cD K RI cDK O M- O ΔP + DZ
cDKO + RIM-ΔPDZ
1
0.4
cDKO
co
0.5 nA 5 ms
o cD l K RI cDK O M- O ΔP + DZ
cDKO
cDKO + RIM1α
cDKO + RIM-ΔPDZ 20 mV 50 ms
Fluo5F
-70 mV
50% ΔG/G0 50 ms
***
J
ns
bouton ΔG/G0 % change
ΔG/G0 (% change)
120
80
dendrite
80 60 40 20 0 0
40
200
400
600
time (ms)
control cDKO cDKO + RIM1α cDKO + RIM-ΔPDZ
0 0
200
time (ms)
400
600
Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc. 291
A
cDKO + RIM-ΔPDZ
cDKO + RIM1α
cDKO
merge
Bsn
Syn
merge CaV2.1
Syn
control
B
1.0
0.8
*
**
0.6
normalized synaptic levels
normalized synaptic levels
1.0
0.4 0.2 0
0.8 0.6
cDKO + RIM-ΔPDZ
0.4 0.2 0
CaV2.1
C
control cDKO cDKO + RIM1α
bassoon zinc finger-domain
presynaptic neuron x Ca2+
x Ca2+
PDZ-domain PxxP motif non-Ca2+-binding C2-domain Ca2+-binding C2-domain MUN-domain SH3-domain FN3 repeat
synaptic cleft
postsynaptic neuron
synaptic vesicle Munc13 x
Rab3/27 GT
P
x
RIM
Ca2+
x
Syt1 Ca2+
x
Ca
2+
Ca
RIM-BP
2+
Ca2+ Ca2+
Ca2+
Ca2+
2+ Ca2+ Ca2+ Ca Ca2+ Ca2+ Ca2+ Ca2+ Ca2+
Ca
Ca2+
Ca2+
Ca2+ Ca2+ Ca2+ Ca2+ Ca2+
Ca2+ Ca2+ Ca2+ Ca2+
Ca2+ Ca
2+
Ca2+
Ca2+
Ca2+
Synaptobrevin SNAP-25
Complexin
2+
Ca2+-channel
Munc18
Ca2+
Ca2+
Syntaxin
Ca2+
Ca2+ Ca2+
292 Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc.
synaptic cleft
Figure 8. RIM-Dependent Ca2+-Channel Tethering Linked to Synaptic Vesicle Docking and Priming (A and B) Immunofluorescent stainings (A) and quantitative immunolocalization analyses (B) of P/Q-type Ca2+ channels (top panel in A) and presynaptic bassoon (bottom panel in A) in control and RIM-deficient cDKO neurons and in cDKO neurons rescued with RIM1a or RIM-DPDZ (means ± SEM, n=3 cultures per condition, *p < 0.05; **p < 0.01 by Student’s t test compared to control; a second, independent experiment is found in Figure S8 and Table S7). (C) Model of the presynaptic release machinery. The drawing illustrates the structures of major active-zone proteins (RIMs, Munc13s, and RIM-BPs), P/Q- or N-type Ca2+ channels, a partially assembled SNARE complex (composed of synaptobrevin/VAMP on synaptic vesicles and SNAP-25 and syntaxin-1 on the plasma membrane), Munc18-1, complexin, and key synaptic vesicle proteins (Rab3 and synaptotagmin-1 [Syt1]). Domain identification is provided on the top right. We propose that RIMs determine the specific localization of P/Qand N-type Ca2+ channels at the active zone via a direct Ca2+channel/PDZ-domain interaction and via indirect binding of Ca2+ channels to RIMs via RIM-BPs (Hibino et al., 2002). In addition, RIMs form an N-terminal priming complex with Rab3 and Munc13, in which Munc13 likely acts by binding to SNARE complexes (not depicted due to restrictions of the twodimensional presentation). Synaptotagmin-1 on the vesicles serves as the Ca2+ sensor for exocytosis. With this architecture, Ca2+ channels and Ca2+ sensors are in close proximity, accounting for the speed, synchrony, and extent of release.
The RIM PDZ Domain Is Essential for Localizing Ca2+ Channels to the Active Zone Using rescue experiments in RIM-deficient neurons, we found that the RIM PDZ domain was invariably required in various RIM rescue constructs to reverse the impairment in presynaptic Ca2+ influx in RIM-deficient neurons (Figure 4, Figure 5, Figure 6, and Figure 7) and for localizing P/Q-type Ca2+ channels to presynaptic boutons (Figure 8). In addition, loss of RIM-BP-binding sequences blocked rescue of Ca2+ influx (Figure 6). These experiments suggest that RIMs tether Ca2+ channels to the active zone via two parallel interactions: directly by binding to Ca2+ channels via their PDZ domains and indirectly by binding to RIM-BPs, which in turn bind to Ca2+ channels (see model in Figure 8C and reconstitution of the tripartite complex in Figure S8C). Our data also account for the specificity of N- and P/Q-type Ca2+ channels in release, as the RIM PDZ domain/Ca2+-channel interaction is specific for these Ca2+-channel types (Figure 1). Active-Zone Functions of RIMs By showing that, besides their role in vesicle docking and priming, RIMs are essential for tethering Ca2+ channels to active zones, our findings corroborate the notion that RIM proteins are central organizers of active zones (Betz et al., 2001; Gracheva et al., 2008; Kaeser et al., 2008; Koushika et al., 2001; Schoch et al., 2002, 2006) (Figure 3, Figure 4, Figure 5, Figure 6, Figure 7, and Figure 8) (see also Deng et al., 2011 and Han et al., 2011). Moreover, RIM proteins perform additional functions, as indicated by the fact that although the RIM C2 domains had no detectable role in Ca2+ influx, they boosted neurotransmitter release, possibly by binding to a-liprins, the b4 Ca2+ channel subunit, or other proteins (Kiyonaka et al., 2007; Schoch et al., 2002). Moreover, in a parallel study we found that RIM proteins activate priming by binding to Munc13 proteins, thereby disrupting their homodimerization and reversing the autoinhibitory effects of the Munc13 homodimerization (Deng et al., 2011). Thus, RIMs occupy the center of an interaction network in the molecular anatomy of the active zone and influence all aspects of neurotransmitter release (Figure 8C). However, our data also raise new questions. How do RIM C2 domains boost release without altering Ca2+ influx? Why does deletion of just one RIM isoform, RIM1a, which has only a partial effect on neurotransmitter release due to its redundancy with other RIM isoforms, block multiple forms of presynaptic long-term synaptic plasticity (Castillo et al., 2002; Fourcaudot et al., 2008; Kaeser et al., 2008)? Is the RIM/Ca2+-channel interaction a mechanism by which synaptic strength can be regulated? With the availability of double conditional KO mice described here, these questions can now be addressed. EXPERIMENTAL PROCEDURES In Vitro Protein-Binding Assays Two yeast two-hybrid screens of a rat brain cDNA library using bait vectors encoding the CaV2.2 N-type Ca2+ channel C terminus (residues 2163–2339) or CaV2.1 P/Q-type Ca2+ channel C terminus (residues 2213–2368) and liquid yeast two-hybrid assays were performed as described (Wang et al., 1997). Of the 134/84 isolates with the N-type/P/Q-type bait, 8/16 clones corresponded to RIM-BP1, 8/17 to RIM-BP2, and 3/2 to RIM1. For mapping of the interaction region and the relative strength, yeast strain L40 was cotrans-
formed with the various Ca2+ channel or RIM1 bait vectors and the Ca2+ channel or RIM1 prey vectors. HSQC spectroscopy was performed with rat RIM1 (residues 596–704, expressed as described; Lu et al., 2005), and nonlabeled P/Q-type Ca2+-channel peptides. 1H–15N HSQC spectra were acquired in a Varian Inova500 spectrometer at a 40–200 mM protein concentration. Generation of Double Conditional RIM KO Mice The RIM2abg targeting vector was constructed from a l-phage DNA clone isolated from a genomic library, and conditional RIM2abg KO mice were generated by homologous recombination in R1 embryonic stem cells. The recombined stem cells were used for blastocyst injections to obtain chimeric mice. After germline transmission of the mutant allele, the newly generated conditional RIM2abg KO mice were crossed to conditional RIM1ab KO mice (Kaeser et al., 2008). Electrophysiology Whole-cell patch-clamp recordings were performed in cultured hippocampal neurons at DIV13–15. Synaptic responses were elicited by a local stimulation electrode and were acquired with a multiclamp 700B amplifier. The extracellular solution contained (in mM) 140 NaCl, 4 KCl, 2 CaCl2, 2 MgCl2, 10 HEPES-NaOH (pH 7.3), and 10 glucose, with 315 mOsm, and either 50 mM picrotoxin (excitatory postsynaptic currents, EPSCs) or 10 mM CNQX and 50 mM D-APV (IPSCs). For all electrophysiological experiments, the experimenter was blind to the condition/genotype of the cultures analyzed. Ca2+ Imaging All experiments were performed in a Zeiss LSM 510 confocal microscope. Cultured hippocampal neurons were examined at DIV14–18 in whole-cell patch-clamp configuration after filling with Fluo5F and Alexa 594 dyes for 10 min. Action potentials were induced by short, somatic current injections through the patch pipette (typically 5 ms, 600 pA); Ca2+ transients were measured with line scans through presynaptic boutons and second-order dendrites at a frequency of 333 Hz, typically 100–150 mm away from the neuronal cell body. Fluorescent signals were quantified as mean region of interest and plotted as GG0/G0 (G = average green emission in a given line; G0 = average of 20 line scans before action potential induction). The experimenter was blind to the condition/genotype until all recordings and analyses were completed. Immunofluorescence Staining of Cultured Neurons Cultured neurons were fixed in 4% paraformaldehyde, permeabilized in 0.1% Triton X-100/3% bovine serum albumin, and incubated overnight with antiCaV2.1 rabbit polyclonal antibodies (Alomone labs, 1:100) or anti-bassoon rabbit polyclonal antibodies (Synaptic Systems, 1:250) and anti-synapsin mouse monoclonal antibodies (Synaptic Systems, 1:1000). Alexa Fluor 546 anti-mouse and Alexa Fluor 633 anti-rabbit secondary antibodies were used for detection with a confocal microscope. Single sections were acquired with identical settings applied to all samples in an experiment and were used to quantify levels of P/Q-type Ca2+ channels and bassoon in ImageJ (NIH); the data were normalized to synapsin staining and expressed relative to control cultures. The experimenter was blind to the condition/genotype in all experiments. Data Analysis All data are shown as means ± standard errors of the mean (SEM). Statistical significance was determined by one-way ANOVA (some electrophysiological recordings), two-way ANOVA (Ca2+-imaging peak amplitude), KolmogorovSmirnov test (cumulative distribution of peak amplitudes in Ca2+ imaging), c-test (mouse survival analysis), or Student’s t test (all other experiments). Ninety-five percent confidence intervals for [Ca2+]ex-titration data and fitting parameters were calculated based on the covariance matrix. All numerical and statistical values and the tests used can be found in Table S1, Table S2, Table S3, Table S4, Table S5, Table S6, and Table S7. Miscellaneous Mixed hippocampal cultures and lentiviruses generated in transfected HEK293T cells expressing EGFP-tagged active or inactive cre-recombinases
Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc. 293
followed by an IRES sequence for expression of rescue constructs were produced as described (Kaeser et al., 2009). SDS-PAGE gels, immunoblotting, and electronmicroscopic analyses were done according to standard methods described in the Extended Experimental Procedures. All animal experiments were performed according to institutional guidelines. A detailed methods section can be found in the Supplemental Information. SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, eight figures, and seven tables and can be found with this article online at doi:10. 1016/j.cell.2010.12.029. ACKNOWLEDGMENTS
Dulubova, I., Lou, X., Lu, J., Huryeva, I., Alam, A., Schneggenburger, R., Sudhof, T.C., and Rizo, J. (2005). A Munc13/RIM/Rab3 tripartite complex: from priming to plasticity? Embo J. 24, 2839–2850. Fourcaudot, E., Gambino, F., Humeau, Y., Casassus, G., Shaban, H., Poulain, B., and Luthi, A. (2008). cAMP/PKA signaling and RIM1alpha mediate presynaptic LTP in the lateral amygdala. Proc. Natl. Acad. Sci. USA 105, 15130– 15135. Gracheva, E.O., Hadwiger, G., Nonet, M.L., and Richmond, J.E. (2008). Direct interactions between C. elegans RAB-3 and Rim provide a mechanism to target vesicles to the presynaptic density. Neurosci. Lett. 444, 137–142. Han, Y., Kaeser, P.S., Su¨dhof, T.C., and Schneggenburger, R. (2011). RIM determines Ca2+ channel density and vesicle docking at the presynaptic active zone. Neuron 69, in press. Published online January 26, 2011. 10.1016/j.neuron.2010.12.014. Hibino, H., Pironkova, R., Onwumere, O., Vologodskaia, M., Hudspeth, A.J., and Lesage, F. (2002). RIM binding proteins (RBPs) couple Rab3-interacting molecules (RIMs) to voltage-gated Ca(2+) channels. Neuron 34, 411–423.
We thank E. Borowicz, I. Kornblum, L. Fan, J. Mitchell, H. Ly, and I. Huryeva for technical assistance, Dr. R.E. Hammer for blastocyst injections, Dr. N. Brose for Munc13-1 antibodies, Dr. C. Acuna-Goycolea for advice on Ca2+-imaging experiments, Dr. Z. Ma for assistance with yeast two-hybrid screening, Drs. Z. Pang, T. Bacaj, and C. Fo¨ldy for help with data analysis, and Dr. R. Schneggenburger for comments. This work was supported by grants from the NIH (NINDS 33564 to T.C.S., NS37200 to J.R., DA029044 to P.S.K.), a Swiss National Science Foundation Postdoctoral Fellowship (to P.S.K.), and a NARSAD Young Investigator Award (to P.S.K.).
Ho, A., Morishita, W., Atasoy, D., Liu, X., Tabuchi, K., Hammer, R.E., Malenka, R.C., and Sudhof, T.C. (2006). Genetic analysis of Mint/X11 proteins: essential presynaptic functions of a neuronal adaptor protein family. J. Neurosci. 26, 13089–13101.
Received: April 1, 2010 Revised: September 2, 2010 Accepted: November 15, 2010 Published: January 20, 2011
Kaeser, P.S., Kwon, H.B., Chiu, C.Q., Deng, L., Castillo, P.E., and Sudhof, T.C. (2008). RIM1alpha and RIM1beta are synthesized from distinct promoters of the RIM1 gene to mediate differential but overlapping synaptic functions. J. Neurosci. 28, 13435–13447.
REFERENCES
Katz, B. (1969). The Release of Neural Transmitter Substances (Liverpool: Liverpool Univ. Press).
Atasoy, D., Schoch, S., Ho, A., Nadasy, K.A., Liu, X., Zhang, W., Mukherjee, K., Nosyreva, E.D., Fernandez-Chacon, R., Missler, M., et al. (2007). Deletion of CASK in mice is lethal and impairs synaptic function. Proc. Natl. Acad. Sci. USA 104, 2525–2530. Betz, A., Thakur, P., Junge, H.J., Ashery, U., Rhee, J.S., Scheuss, V., Rosenmund, C., Rettig, J., and Brose, N. (2001). Functional interaction of the active zone proteins Munc13-1 and RIM1 in synaptic vesicle priming. Neuron 30, 183–196.
Kaeser, P.S., Deng, L., Chavez, A.E., Liu, X., Castillo, P.E., and Sudhof, T.C. (2009). ELKS2alpha/CAST deletion selectively increases neurotransmitter release at inhibitory synapses. Neuron 64, 227–239.
Kiyonaka, S., Wakamori, M., Miki, T., Uriu, Y., Nonaka, M., Bito, H., Beedle, A.M., Mori, E., Hara, Y., De Waard, M., et al. (2007). RIM1 confers sustained activity and neurotransmitter vesicle anchoring to presynaptic Ca2+ channels. Nat. Neurosci. 10, 691–701. Koushika, S.P., Richmond, J.E., Hadwiger, G., Weimer, R.M., Jorgensen, E.M., and Nonet, M.L. (2001). A post-docking role for active zone protein Rim. Nat. Neurosci. 4, 997–1005. Llinas, R., Sugimori, M., and Silver, R.B. (1992). Microdomains of high calcium concentration in a presynaptic terminal. Science 256, 677–679.
Castillo, P.E., Weisskopf, M.G., and Nicoll, R.A. (1994). The role of Ca2+ channels in hippocampal mossy fiber synaptic transmission and long-term potentiation. Neuron 12, 261–269.
Lu, J., Li, H., Wang, Y., Sudhof, T.C., and Rizo, J. (2005). Solution structure of the RIM1alpha PDZ domain in complex with an ELKS1b C-terminal peptide. J. Mol. Biol. 352, 455–466.
Castillo, P.E., Schoch, S., Schmitz, F., Sudhof, T.C., and Malenka, R.C. (2002). RIM1alpha is required for presynaptic long-term potentiation. Nature 415, 327–330.
Luebke, J.I., Dunlap, K., and Turner, T.J. (1993). Multiple calcium channel types control glutamatergic synaptic transmission in the hippocampus. Neuron 11, 895–902.
Catterall, W.A., Perez-Reyes, E., Snutch, T.P., and Striessnig, J. (2005). International Union of Pharmacology. XLVIII. Nomenclature and structure-function relationships of voltage-gated calcium channels. Pharmacol. Rev. 57, 411–425.
Maximov, A., Sudhof, T.C., and Bezprozvanny, I. (1999). Association of neuronal calcium channels with modular adaptor proteins. J. Biol. Chem. 274, 24453–24456.
Church, P.J., and Stanley, E.F. (1996). Single L-type calcium channel conductance with physiological levels of calcium in chick ciliary ganglion neurons. J. Physiol. 496, 59–68. Coppola, T., Magnin-Luthi, S., Perret-Menoud, V., Gattesco, S., Schiavo, G., and Regazzi, R. (2001). Direct interaction of the Rab3 effector RIM with Ca2+ channels, SNAP-25, and synaptotagmin. J. Biol. Chem. 276, 32756– 32762.
Meinrenken, C.J., Borst, J.G., and Sakmann, B. (2002). Calcium secretion coupling at calyx of held governed by nonuniform channel-vesicle topography. J. Neurosci. 22, 1648–1667. Missler, M., Zhang, W., Rohlmann, A., Kattenstroth, G., Hammer, R.E., Gottmann, K., and Sudhof, T.C. (2003). Alpha-neurexins couple Ca2+ channels to synaptic vesicle exocytosis. Nature 424, 939–948. Mittelstaedt, T., Alvarez-Baron, E., and Schoch, S. (2010). RIM proteins and their role in synapse function. Biol. Chem. 391, 599–606.
Deng, L., Kaeser, P.S., Xu, W., and Su¨dhof, T.C. (2011). RIM proteins activate vesicle priming by reversing autoinhibitory homodimerization of Munc13. Neuron 69, in press. Published online January 26, 2011. 10.1016/j.neuron. 2011.01.005.
Ohtsuka, T., Takao-Rikitsu, E., Inoue, E., Inoue, M., Takeuchi, M., Matsubara, K., Deguchi-Tawarada, M., Satoh, K., Morimoto, K., Nakanishi, H., et al. (2002). Cast: a novel protein of the cytomatrix at the active zone of synapses that forms a ternary complex with RIM1 and munc13-1. J. Cell Biol. 158, 577–590.
Dietrich, D., Kirschstein, T., Kukley, M., Pereverzev, A., von der Brelie, C., Schneider, T., and Beck, H. (2003). Functional specialization of presynaptic Cav2.3 Ca2+ channels. Neuron 39, 483–496.
Poncer, J.C., McKinney, R.A., Gahwiler, B.H., and Thompson, S.M. (1997). Either N- or P-type calcium channels mediate GABA release at distinct hippocampal inhibitory synapses. Neuron 18, 463–472.
294 Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc.
Qian, J., and Noebels, J.L. (2000). Presynaptic Ca(2+) influx at a mouse central synapse with Ca(2+) channel subunit mutations. J. Neurosci. 20, 163–170. Regehr, W.G., and Mintz, I.M. (1994). Participation of multiple calcium channel types in transmission at single climbing fiber to Purkinje cell synapses. Neuron 12, 605–613. Schneggenburger, R., Meyer, A.C., and Neher, E. (1999). Released fraction and total size of a pool of immediately available transmitter quanta at a calyx synapse. Neuron 23, 399–409. Schoch, S., Castillo, P.E., Jo, T., Mukherjee, K., Geppert, M., Wang, Y., Schmitz, F., Malenka, R.C., and Sudhof, T.C. (2002). RIM1alpha forms a protein scaffold for regulating neurotransmitter release at the active zone. Nature 415, 321–326. Schoch, S., Mittelstaedt, T., Kaeser, P.S., Padgett, D., Feldmann, N., Chevaleyre, V., Castillo, P.E., Hammer, R.E., Han, W., Schmitz, F., et al. (2006). Redundant functions of RIM1alpha and RIM2alpha in Ca(2+)-triggered neurotransmitter release. EMBO J. 25, 5852–5863. Shin, O.H., Xu, J., Rizo, J., and Sudhof, T.C. (2009). Differential but convergent functions of Ca2+ binding to synaptotagmin-1 C2 domains mediate neurotransmitter release. Proc. Natl. Acad. Sci. USA 106, 16469–16474. Takahashi, T., and Momiyama, A. (1993). Different types of calcium channels mediate central synaptic transmission. Nature 366, 156–158.
Wang, Y., Okamoto, M., Schmitz, F., Hofmann, K., and Sudhof, T.C. (1997). Rim is a putative Rab3 effector in regulating synaptic-vesicle fusion. Nature 388, 593–598. Wang, Y., Sugita, S., and Sudhof, T.C. (2000). The RIM/NIM family of neuronal C2 domain proteins. Interactions with Rab3 and a new class of Src homology 3 domain proteins. J. Biol. Chem. 275, 20033–20044. Wang, Y., Liu, X., Biederer, T., and Sudhof, T.C. (2002). A family of RIM-binding proteins regulated by alternative splicing: Implications for the genesis of synaptic active zones. Proc. Natl. Acad. Sci. USA 99, 14464–14469. Wiedemann, U., Boisguerin, P., Leben, R., Leitner, D., Krause, G., Moelling, K., Volkmer-Engert, R., and Oschkinat, H. (2004). Quantification of PDZ domain specificity, prediction of ligand affinity and rational design of super-binding peptides. J. Mol. Biol. 343, 703–718. Wojcik, S.M., and Brose, N. (2007). Regulation of membrane fusion in synaptic excitation-secretion coupling: Speed and accuracy matter. Neuron 55, 11–24. Wu, L.G., Westenbroek, R.E., Borst, J.G., Catterall, W.A., and Sakmann, B. (1999). Calcium channel types with distinct presynaptic localization couple differentially to transmitter release in single calyx-type synapses. J. Neurosci. 19, 726–736.
Cell 144, 282–295, January 21, 2011 ª2011 Elsevier Inc. 295
Resource
Densely Interconnected Transcriptional Circuits Control Cell States in Human Hematopoiesis Noa Novershtern,1,2,3,11 Aravind Subramanian,1,11 Lee N. Lawton,4 Raymond H. Mak,1 W. Nicholas Haining,5 Marie E. McConkey,6 Naomi Habib,3 Nir Yosef,1 Cindy Y. Chang,1,6 Tal Shay,1 Garrett M. Frampton,2,4 Adam C.B. Drake,2,7 Ilya Leskov,2,7 Bjorn Nilsson,1,6 Fred Preffer,8 David Dombkowski,8 John W. Evans,5 Ted Liefeld,1 John S. Smutko,9 Jianzhu Chen,2,7 Nir Friedman,3 Richard A. Young,2,4 Todd R. Golub,1,5,10 Aviv Regev,1,2,10,12,* and Benjamin L. Ebert1,5,6,12,* 1Broad
Institute, 7 Cambridge Center, Cambridge MA, 02142, USA of Biology, Massachusetts Institute of Technology, Cambridge MA, 02140, USA 3School of Computer Science, Hebrew University, Jerusalem 91904, Israel 4Whitehead Institute for Biomedical Research, 9 Cambridge Center, Cambridge, MA 02142, USA 5Dana-Farber Cancer Institute, Boston, MA 02115, USA 6Brigham and Women’s Hospital, Boston, MA 02115, USA 7Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, MA 02139 8Massachusetts General Hospital, Boston, MA 02114, USA 9Nugen Technologies, San Carlos, CA 94070, USA 10Howard Hughes Medical Institute, Chevy Chase, MD 20815-6789, USA 11These authors contributed equally to this work 12These authors contributed equally to this work *Correspondence:
[email protected] (A.R.),
[email protected] (B.L.E.) DOI 10.1016/j.cell.2011.01.004 2Department
SUMMARY
Though many individual transcription factors are known to regulate hematopoietic differentiation, major aspects of the global architecture of hematopoiesis remain unknown. Here, we profiled gene expression in 38 distinct purified populations of human hematopoietic cells and used probabilistic models of gene expression and analysis of cis-elements in gene promoters to decipher the general organization of their regulatory circuitry. We identified modules of highly coexpressed genes, some of which are restricted to a single lineage but most of which are expressed at variable levels across multiple lineages. We found densely interconnected cis-regulatory circuits and a large number of transcription factors that are differentially expressed across hematopoietic states. These findings suggest a more complex regulatory system for hematopoiesis than previously assumed. INTRODUCTION Hematopoiesis is an ideal model for the study of multilineage differentiation in humans. More than 2 3 1011 hematopoietic cells from at least 11 lineages are produced daily in humans from a small pool of self-renewing adult stem cells (Quesenberry and Colvin, 2005). Production of each cell type is highly regulated and responsive to environmental stimuli. Mutations or 296 Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc.
aberrant expression of regulatory proteins cause both benign and malignant hematologic disorders. The hematopoietic system is also well suited for an analysis of the global architecture of the molecular circuits controlling human cellular differentiation. Hematopoietic stem cells, progenitor cells, and terminally differentiated cells can be isolated using flow cytometry. Moreover, many aspects of hematopoietic differentiation can be recapitulated in vitro. Finally, high-speed multiparameter flow cytometry and cDNA amplification procedures allow us to purify and profile gene expression from rare subpopulations (Ebert and Golub, 2004). A dominant model of hematopoiesis posits that it is controlled by a hierarchy of a relatively small number of critical transcription factors (TFs) that are sequentially expressed, are largely restricted to a specific lineage, and can interact directly to mediate and reinforce cell fate decisions (Iwasaki and Akashi, 2007). Genetically engineered mice have been used to map the maturation stage at which key TFs are essential (Orkin and Zon, 2008). Recent genome-wide studies suggest a more complex architecture in regulatory circuits involving larger numbers of TFs that control different combinations of modules of coexpressed genes (Amit et al., 2009; Suzuki et al., 2009). Complex circuits with a larger number of TFs than previously assumed, each with a major regulatory effect, are emerging from studies in immune cell types (Amit et al., 2009; Suzuki et al., 2009), stem cell populations (Mu¨ller et al., 2008), and cell differentiation in invertebrates (Davidson, 2001). These two views leave open several key questions in understanding the regulatory architecture of human hematopoiesis. (1) Are distinct hematopoietic cell states characterized mostly
lin– CD133+
HSC1 CD34dim
lin– CD38–
HSC2 CD34+
CD34+ CD38+
CMP IL-3Rα lo + CD45RA–
CD34+ CD38+ IL-3Rα– CD45RA–
MEP
GMP
CD34+ CD38+ IL-3Rα lo + CD45RA+
CD34+
ERY1 CD71+ GlyA–
CD34–
ERY2 CD71+ GlyA–
CD16– CD34+ CD41+ CD61+ CD45–
CD34–
ERY4 CD71 lo GlyA+
ERY5
MEGA2
CD34– CD71– GlyA+
CD34– CD41+ CD61+ CD45–
Pre-BCELL2 CD10+ CD19+
GRAN1 CD11b–
GlyA+
MEGA1
CD34+
CD34– SSC hi CD45+
CD34–
ERY3 CD71+
ERY MEGA
CD34–
MONO1
Pre-BCELL3 CD10+ CD19+
CD34– CD33+ CD13+
CD34 SSC hi CD45+ GRAN2 CD11b+ CD16–
GRAN3 FSC hi SSC hi CD16+ CD11b+
CD19+
CD8+
BCELLa1 lgD+
TCELL2 CD62L+
CD27–
MONO2
EOS2
BASO1
FSC hi SSC lo CD14+ CD45dim
FSC hi SSC lo IL3Rα+ CD33dim+
FSC hi SSC lo CD22+ CD123+ CD33+/– CD45dim
GRAN/MONO
DENDa2 HLA DR+ CD3– CD14– CD16– CD19– CD56– CD123– CD11c+
CD45RA+
DENDa1 BCELLa2 BCELLa3 BCELLa4
NKa1
NKa2
NKa3
NKa4
HLA DR+ CD3– CD14– CD16– CD19– CD56– CD123+ CD11c–
CD56– CD16+ CD3–
CD56+ CD16+ CD3–
CD56– CD16– CD3–
CD14– CD19– CD3+ CD1d+
DC
CD19+ lgD+ CD27+
CD19+ lgD– CD27–
B-cells
CD19+ lgD– CD27+
NK-cells
TCELL1 CD8+ CD62L– CD45RA+
TCELL3 CD8+ CD62L– CD45RA–
TCELL4 CD8+ CD62L+ CD45RA–
CD4+
TCELL6 CD62L+
CD45RA+
TCELL7 CD4+ CD62L– CD45RA–
TCELL8 CD4+ CD62L+ CD45RA–
T-cells
Figure 1. Hematopoietic Differentiation The 38 hematopoietic cell populations purified by flow sorting and analyzed by gene expression profiling are illustrated in their respective positions in hematopoiesis. (Gray) Hematopoietic stem cell (HSC1,2), common myeloid progenitor (CMP), megakaryocyte/erythroid progenitor (MEP). (Orange) Erythroid cells (ERY1–5). (Red) CFU-MK (MEGA1) and megakaryocyte (MEGA2). (Purple) Granulocyte/monocyte progenitor (GMP), CFU-G (GRAN1), neutrophilic metamyelocyte (GRAN2), neutrophil (GRAN3), CFU-M (MONO1), monocytes (MONO2), eosinophil (EOS), and basophil (BASO). (Blue) Myeloid dendritic cell (DENDa2) and plasmacytoid dendritic cell (DENDa1). (Light green) Early B cell (Pre-BCELL2), pro-B cell (Pre-BCELL3), naive B cell (BCELLa1), mature B cell, class able to switch (BCELLa2), mature B cell (BCELLa3), and mature B cell, class switched (BCELLa4). (Dark green) Mature NK cell (NK1–4). (Turquoise) Naive CD8+ T cell (TCELL2), CD8+ effector memory RA (TCELL1), CD8+ effector memory (TCELL3), CD8+ central memory (TCELL4), naive CD4+ T cell (TCELL6), CD4+ effector memory (TCELL7), and CD4+ central memory (TCELL8). See Table S1 for markers information.
by induction of lineage-specific genes or by a unique combination of modules, wherein the distinct capacities of each cell type are largely determined through the reuse of modules? (2) Is hematopoiesis determined solely by a few master regulators, or does it involve a more complex network with a larger number of factors? (3) What are the regulatory mechanisms that maintain cell state in the hematopoietic system, and how do they change as cells differentiate? Here, we measured mRNA profiles in 38 prospectively purified cell populations, from hematopoietic stem cells, through multiple progenitor and intermediate maturation states, to 12 terminally differentiated cell types (Figure 1). We found distinct, tightly integrated, regulatory circuits in hematopoietic stem cells and
differentiated cells, implicated dozens of new regulators in hematopoiesis, and demonstrated a substantial reuse of gene modules and their regulatory programs in distinct lineages. We validated our findings by experimentally determining the binding sites of four TFs in hematopoietic stem cells, by examining the expression of a set of 33 TFs in erythroid and myelomonocytic differentiation in vitro, and by investigating the function of 17 of these TFs using RNA interference. Our data provide strong evidence for the role of complex interconnected circuits in hematopoiesis and for ‘‘anticipatory binding’’ to the promoters of their target genes in hematopoietic stem cells. Our data set and analyses will serve as a comprehensive resource for the study of gene regulation in hematopoiesis and differentiation. Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc. 297
HSC2
RESULTS TCELL CD4
CD8
NK
BCELL
MONO EOS BASO DEND2 DEND1 PBCELL
GRAN
MEGA
Early ERY
Late ERY
MEP
GMP
CMP
HSC1
A
HSC1 CMP
HSC2
MEP Early ERY Late ERY MEGA GMP GRAN MONO EOS BASO DEND2 DEND1 PBCELL BCELL NK
CD8 TCELL CD4 Pearson correlation –1
B
HSC/early ERY
Late ERY GRAN/MONO
B-cell
+1
T-cell HOXA9 N-MYC HMGA2 CD34 GATA2 CDK6 ANK1 HBQ1 MRC2 RHCE SPTB CD64 TLR2 FCN1 HNMT TREM1 VENTX CD40 SOX5 CD19 IL9R SWAP70 IGHA1 CD3E IL7R CD28 LAT RORA CD27
An Expression Map of Hematopoiesis Reveals Cell State-Specific Profiles We defined 38 distinct cell states based on cell surface marker expression, representing hematopoietic stem and progenitor cells, terminally differentiated cells, and intermediate states (Figure 1 and Table S1 available online). For each state, we purified samples separately from four to seven independent donors by multiparameter flow cytometry (Experimental Procedures), yielding 211 samples. Cells from all stem and progenitor populations were purified from umbilical cord blood. Terminally differentiated lymphocyte populations were purified from peripheral blood, as terminal differentiation is completed in these cells upon exposure to antigens after birth (Table S1). In all cases, cells were harvested fresh and were processed and sorted immediately. We isolated mRNA from each cell type and measured expression profiles using Affymetrix microarrays (Experimental Procedures). The global transcriptional profiles are consistent with the established topology of hematopoietic differentiation. Replicate samples from a single state but different donors and samples from multiple states within a lineage are highly correlated with each other, and profiles from related lineages are also similar (Figure 2A). Of note, hematopoietic stem cell (HSC) samples do not form a separate cluster but are highly similar to early progenitors in the megakaryocyte/erythrocytes lineage (MEGA/ERY), suggesting that their transcriptional state is largely maintained in some of the early progenitors. These findings are also apparent in a systematic unsupervised analysis using nonnegative matrix factorization (NMF) (Brunet et al., 2004) (Figure S1A) and hierarchical clustering (Figure S1B). We further validated our data set by confirming that previously published lineage-specific gene signatures are significantly enriched in the expected lineage compared to other lineages (FDR < 0.25; Figure S1C and Extended Experimental Procedures).
Log 2 scale –1
+1
Number of genes (n = 13,647)
C 6000 GNF2 Hemato Breast Lung Lymphoma
5000 4000 3000 2000 1000 0
z > 0.5
z>1
z>2
z>3
z>4
z>5
z > 10
Distance from mean
Figure 2. A Transcriptional Map of Hematopoietic Differentiation Identifies Lineage-Specific Transcription (A) Similarity in global expression profiles between proximate differentiation states. The heat map shows the pairwise Pearson correlation coefficients between all 211 samples ordered according to the differentiation tree (right and top). A positive correlation is portrayed in yellow and a negative correlation in purple. (B) Signature genes characterizing the five main lineages. Expression levels are shown for the top 50 marker genes (rows) for each of four major lineages plus hematopoietic stem and progenitor cells. High relative expression is
298 Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc.
Unique and Complex Gene Signatures Characterize Distinct Hematopoietic Lineages In a supervised analysis, we found that each of the five dominant states in our data set—HSPCs, differentiated erythroid cells, granulocytes/monocytes B cells, and T cells—is distinguished by a set of significantly differentially expressed genes specific to each lineage as compared to the others (Figure 2B and Table S2). Some of these genes are expressed more than 100-fold higher in one cell type (e.g., granzyme genes in certain T cell and NK cell populations, PROM1 [CD133 antigen], and HOXA9 in stem and progenitor cells). shown in red and low relative expression in blue; the expression of each gene is normalized to a mean expression of zero across all the samples; labels as in Figure 1. Genes were selected by high expression in one lineage compared to the others (t test). (C) The number of genes that are differentially expressed, according to an outlier statistic, was calculated for all hematopoietic cell states profiled (red); a compendium of 79 tissues in the GNF atlas (Su et al., 2004) (blue); and data sets of lymphomas (Monti et al., 2005) (turquoise), lung cancers (Bhattacharjee et al., 2001) (purple), and breast cancers (Chin et al., 2006) (green). See also Figure S1.
HSC2
HSC1
B
CD4
TCELL CD8
NK
BCELL
EOS BASO DEND2 DEND1
GRAN
MONO
PBCELL
GMP Late ERY
Early ERY
MEGA
MEP
Module size
250
CMP
A
685 607 985 835 1021 901 841 811 691 913 673 949 715 769 613 943 991 817 859 667 955 829 757 703 1003 595 931 589 1027 649 961 853 907 739 661 793 847 583 883 967 601 625 895 727 889 775 577 787 655 399 679 973 919 925 823 733 871 751 643 1009 565 805 631 865 709 697 745 1015 781 937 637 997 559 763 799 877 619 721 571 979
Carbohydrate metabolism;Growth Hormone Signaling Pathway;Lysosome Ribosome Ribosome RNA processing Ribosome RNA processing Purinergic nucleotide receptor
Cytoskeleton Antigen presentation;IFN alpha signaling pathway;MHC class II CTL mediated immune response ;T Cell Receptor Signaling Pathway;T Helper Cell Surface Molecules Cell communication;T Cell Receptor Signaling Pathway;Tyrosine kinase signaling
Antigen presentation;BCR Signaling Pathway;MHC class II
Monovalent inorganic cation transporter activity;Oxidative phosphorylation Oxidative phosphorylation Oxidative phosphorylation Oxidoreductase activity Protein amino acid glycosylation Blood group antigen;Organic cation transporter activity Cell cycle;Cell proliferation;Mitosis Cell cycle checkpoint Calcium ion binding activity Chromatin;DNA packaging Cell cycle;DNA replication ER;Energy pathway;Mitochondrion;Oxidative phosphorylation;Oxidoreductase activity Ribonucleotide metabolism Actin Organization and Cell Migration;Cell junction;ER;Hydrolase Hemoglobin complex Cell proliferation Serine-type endopeptidase activity Vision Voltage-gated ion channel activity
Morphogenesis Cell differentiation Cell communication;Granzyme A mediated Apoptosis Pathway;Interleukin receptor activity;Ligand-gated ion channel Non-membrane spanning protein tyrosine phosphatase activity Prostanoid receptor activity Ligand-gated ion channel activity Receptor activity Antibacterial peptide activity;Serine-type endopeptidase activity Immunoglobulin Inflammatory response
Log2-Ratio -1
0
1
Figure 3. Expression Pattern and Functional Enrichment of 80 Transcriptional Modules (A) Average expression levels of 80 gene modules. Shown is the average expression pattern of the gene members in each of the 80 modules (rows) across all 211 samples (columns). Colors and normalization as in Figure 2B. The samples are organized according to the differentiation tree topology (top) with abbreviations as in Figure 1. The number of genes in each module is shown in the bar graph (left). The expression profiles of a few example modules discussed in the text are highlighted by vertical yellow lines. The expression of individual genes in each module is shown in Figure S2. (B) Functional enrichment in gene modules. Functional categories with enriched representation (FDR < 5%) in at least one module are portrayed. Categories were selected for broad representation. The complete list appears in Table S3. See also Figure S2 and Figure S7.
The signature genes are enriched for molecular functions and biological processes consistent with the functional differences between lineages (Figure S1D and Table S2). Of note, a set of 16 genes comprised of the 50 partners of known translocations in leukemias (Mitelman et al., 2010) is enriched in the HSPC population (p < 0.013). This suggests that the 50 partners of leukemiacausing translocations, containing the promoters of the fusion genes, tend to be selectively expressed in stem and progenitor cell populations. The diversity of gene expression across hematopoietic lineages is comparable to the diversity in gene expression observed across a host of human tissue types. The number of genes that are differentially expressed throughout our hematopoiesis data set (outlier analysis) (Tibshirani and Hastie, 2007) (Extended
Experimental Procedures) is comparable to that determined for an atlas of 79 different human tissues (Su et al., 2004) and far higher than in lymphomas (Monti et al., 2005), lung cancers (Bhattacharjee et al., 2001), or breast cancers (Chin et al., 2006) (Figure 2C). Coherent Functional Modules of Coexpressed Genes Are Reused across Lineages To dissect the architecture of the gene expression program, we used the Module Networks (Segal et al., 2003) algorithm (Experimental Procedures) to find modules of strongly coexpressed genes and associate them with candidate regulatory programs that (computationally) predict their expression pattern. We identified 80 gene modules (Figure 3A; modules are numbered Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc. 299
arbitrarily by the algorithm) covering the 8968 genes that are expressed in the majority of the samples of at least one cell population. The genes in each of the modules are tightly coexpressed (Figure S2), and the 80 modules have largely distinct expression patterns (Figure 3A and Figure S2) and are enriched for genes with distinct biological functions (Figure 3B and Table S3). A small number of modules are expressed in very specific cell states and reflect the unique functional capacities of a single lineage. For example, module 889 is expressed in terminal erythroid differentiation and is enriched for genes encoding blood group antigens and organic cation transporters; module 691 is expressed in B lymphocytes and is enriched for genes encoding immunoglobulins and BCR-signaling pathway components; and module 721 is expressed in granulocytes and monocytes and includes genes encoding enzymes and cytokine receptors that are essential for inflammatory responses. Conversely, most modules are expressed at varying levels across multiple lineages, suggesting reuse of their genes in multiple hematopoietic contexts. These include modules expressed in both HSC and progenitor populations (e.g., numbers 865, 679, and 805), in both B and T cells (e.g., 673 and 703), in both granulocyte/monocyte populations and lymphocytes (e.g., 817, 799, and 649), and across all myeloid (e.g., 583) or all lymphoid cells (e.g., 931). Reuse of modules reflects the differential functional requirements for specific biochemical programs in the various cell states. For example, mitochondrial and oxidative phosphorylation modules (e.g., 847, 583, and 883) are induced in erythroid progenitors that produce high levels of heme and are affected most by mitochondrial mutations (Chen et al., 2009; Fontenay et al., 2006), as well as in granulocytes and monocytes, which are capable of a respiratory burst following phagocytosis. Module States Persist through Multiple Differentiation Steps To delineate the relation between gene expression and differentiation, we projected each module’s expression pattern onto the known topology of the differentiation tree (Figure 4 and Figure S4). For example, consider module 865 (Figure 4A and Figure S3), which is strongly induced in hematopoietic stem and progenitor cells and contains genes encoding key HSPC cell surface markers (CD34 and CD117) and transcriptional regulators (GATA2, HOXA9, HOXA10, MEIS1, and N-MYC). By projecting the module on the differentiation tree, we observe that its induced state in HSCs persists through several consecutive differentiation steps and is repressed at three main points (Figure 4A, arrowheads): (1) after the granulocyte/monocyte progenitor, (2) after erythroid progenitors, and (3) in the differentiation of HSCs toward the lymphocyte lineage. We identified a host of such differentiation-associated patterns in gene regulation. One major pattern (31 modules) is HSC-persistent states: such modules are active in the HSC state and persist in an active state in several progenitor populations on the erythroid/myeloid branch (Figures 4A and 4E), the lymphoid branch (Figure S4A), or both (Figures S4B and S4H). The HSC state changes gradually at different points in different modules. Indeed, only module 631 (Figure S4C) is primarily HSC specific and includes the known stem cell-specific TFs NANOG and 300 Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc.
SMAD1 (Xu et al., 2008). In other patterns, modules have low or inactive expression in HSCs but are activated in a single lineage (10 modules) on either the erythroid/myeloid branch (Figures 4B and 4C and Figure S4D) or the lymphoid branch (Figure 4D). In most cases (39 modules), modules are inactive in HSPCs but are activated in multiple independent lineages (Figure 4F and Figure S4F). A Sequence-Based Model of the Regulatory Code The high degree of coexpression of genes within modules suggests that they may be coregulated by common transcriptional circuits. We therefore examined each module for enrichment of known and candidate cis-regulatory elements in their promoters (Extended Experimental Procedures). We used six motif-finding methods and a motif-clustering pipeline to identify a nonredundant library of enriched elements. We scored each module for the enrichment of each of the candidate sites or of known elements or binding events (Sandelin et al., 2004; Subramanian et al., 2005) (Extended Experimental Procedures). We identified 156 sequence motifs and 28 binding profiles of 12 TFs (measured by ChIP) that were enriched in the promoters of at least one module (data available on http://www.broadinstitute. org/dmap/). Of these, 66 are previously unannotated motifs, and 118 are associated with 72 TFs (Table S4). Of these 72 TFs, 11 are known hematopoietic factors (Table S4), and their sites are often enriched in modules consistent with their known functions. For example, the site for the erythroid TF GATA1 (Pevny et al., 1991) is enriched in the late erythroid module 889, and sites for the lymphocyte regulators Helios and NFATC (Aramburu et al., 1995; Hahm et al., 1998) are enriched in the T and NK module 559. We also found significant enrichments for TFs with roles in other differentiation processes, which were not previously implicated in hematopoiesis, such as HNF4 a (in the HSPC Module 865) and HNF6 (in the lymphoid modules 859 and 961). Tightly Integrated cis-Regulatory Circuits Govern Differentiation States To explore how these cis-regulatory associations can give rise to stable cell states, we assembled the regulatory circuits connecting the 276 TFs whose binding sites were enriched in any gene set with each other (Figure 5). We connected an edge from each factor with a known motif to all of the factors that harbor this motif in their gene promoters (Extended Experimental Procedures) and focused only on those factors that were expressed in a given cell state. For example, the circuit of HSC-expressed TFs with known binding sites (Figure 5A) includes many major known regulators of the HSC state (Orkin and Zon, 2008), which are densely interconnected through autoregulatory (12 of 23 active factors), feedback (15 and 39 loops of size 2 and 3), and feedforward (206 loops of size 3) loops. Abnormal expression of many of the circuit’s TFs is known to cause hematologic malignancies (Look, 1997). This integrated circuitry can give rise to a robust transcriptional network in terminally differentiated cells and HSCs. Of note, because the sequence of the binding site for most TFs is unknown, including 66 of the putative enriched binding sites, the density of regulation is likely even greater than we observed.
A
B
HSC and progenitor module (#865)
Late erythroid module (#727)
PBX1, SOX4
NFE2*
PBX1
NFE2*
PBX1 NFE2
SOX4
SOX4
FOXO3A, GATA1, NFIX1, MYT1
GATA2, HOXA9, HOXA10, MEIS1, MYCN, DNMT3B, ZNF323, HMGA2
C
Granulocytes and monocytes module (#721)
D
B-cell module (#589)
MNDA, CEBPD
POU2AF1, HOXC4
MNDA
POU2AF1 CEBPD CEBPD HOXC4 POU2AF1
CEBPA, VDR, SPI1, ATF3, CREB5, PPARGC1A, VENTX, MYCL1
E
KLF8, E2F5, GABPA, BHLHB3, GCM1
F
HSC and erythroid module (#655)
Granulocyte, B- and T-cell “re-use” module (#817)
TAL1
PIAS
TAL1
TAL1
HHEX
PIAS
PIAS PIAS
NCOA4, Timeless, CSDA
TRIM22, ISGF3G, TRIM38, SP110, IRF1, JUNB, ARNTL, STAT1, NCOA3, NCOA1
Figure 4. Propagation and Transitions in Modules’ Expression along Hematopoiesis Shown are the mean expression levels of the module’s genes in each cell state (colored squares) and selected changes in the predicted regulators, as highlighted in the text (upward arrowhead, regulator induced; downward arrowhead, regulator repressed). Member genes (rather than regulators) in each module encoding TFs are noted below each module, as these may reflect alternative regulators at the same differentiation points. TFs that were validated as regulators of erythroid or granulocyte/monocyte differentiation in a functional assay (Figure 7) are highlighted in bold. The color bar at the bottom of each tree denotes the key lineages, as in Figure 1. (A) HSC and progenitor expression in module 865. (B) Lineage-specific induction in late erythrocytes in module 727. (C) Lineage-specific induction in granulocytes and monocytes in module 721. (D) Lineage-specific induction in B cells in module 589. (E) One-sided propagation of induced state from HSC to the erythroid lineage in module 655. (F) Reuse of module 817, which is inactive in HSCs and independently induced in both lymphoid cells and granulocytes. See also Figure S3 and Figure S4.
During the course of differentiation, the HSC circuit gradually disappears along multiple lineages due to loss of expression of the relevant TFs (Figure 5A and data available on http://www.
broadinstitute.org/dmap/). Conversely, in terminally differentiated cells, other dense circuits emerge through the induction of other TFs. For example, the 14 factors in the erythroid circuit Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc. 301
A HSC network
B Late erythrocyte network
HSC
MEP Correlated
TF
Uncorrelated
Active in phase
Early ERY
Late ERY
Inactive
Figure 5. Dynamic Organization of Tightly Integrated cis-Regulatory Circuits in HSCs and Erythroid Cells (A and B) Shown are cis-regulatory networks between TFs (nodes) that are enriched in at least one gene set and are expressed (fold change > 1.5) in (A) HSCs or (B) late erythroid cells. Nodes represent TFs that are expressed (purple) or not (gray) in each of the four phases of the erythroid lineage (HSC, MEP, early ERY, and late ERY). An edge from node a to node b indicates that the promoter of the gene in node b has a binding site for the TF encoded by the gene in node a. Edge colors indicate the Pearson correlation between the expression profiles of the TFs in the connected nodes: red, positive correlation (coefficient > 0.4); black, no correlation (absolute Pearson % 0.4); gray, nonactive edge (at least one of the two connected nodes was not expressed in that phase). See Table S4 for enriched motif information.
include many of the known major regulators of erythroid differentiation (Cantor and Orkin, 2002), including GATA1, LMO2, FOXO4, NFE2, and RXRA (Figure 5B). We find similarly distinct networks in the granulocyte lineage, T cells, and B cells. Hundreds of Transcription Factors Are Differentially Expressed across Lineages in Coherent Modules The dense regulatory circuits between TFs in our sequencebased model suggest that the expression of TF genes is likely to be highly regulated in hematopoiesis. Indeed, supervised analysis finds that many TF genes are strongly differentially expressed in each primary lineage (Figure 6A and Figure S5A) and that the diversity of TF gene expression is comparable between hematopoiesis and the tissue compendium (Su et al., 2004) (Figure S5B). Some TFs are expressed predominantly in a single lineage, including well-studied TFs that are known to be essential for differentiation in HSCs or a particular lineage (Figure S6). However, the expression of those factors often increases gradually along differentiation (Figures S6D, S6H, and S6I), similar to the gradations observed in gene modules (Figure 4 and Figure S4). Many other TFs are ‘‘reused’’ across lineages either through persistent expression from a single progenitor population or by independent activation in multiple lineages (Figure 4 and Figure S4). For example, module 793 (Figure S4F), which is 302 Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc.
induced in both B cells and late erythroid cells, includes several TFs and chromatin regulators. Among these, KLF3 has a reported role in erythroid cells (Funnell et al., 2007), whereas NFAT5 has a demonstrated function in B cells (Kino et al., 2009). Many TFs—not previously associated with these lineages—are expressed similarly to known factors and belong to the same modules, suggesting that the transcriptional circuit consists of a greater number of TFs than previously assumed. For example, the late erythroid module 727 (Figure 4B) contains four TFs: two are known erythroid TFs (GATA1 and FOXO3A) (Bakker et al., 2007), whereas the others (NFIX1, MYT1) were not previously linked to erythropoiesis. Similarly, the granulocytes/monocytes module 721 (Figure 4C) contains eight TFs, only two with known roles in the lineage (CEBPA and PU.1/SPI1). An Expression-Based Model of the Regulatory Code of Hematopoiesis Identifies Putative Regulators Controlling Changes in Differentiation To identify the potential regulatory role of differentially expressed TFs, we examined the combinations of TFs (regulatory program), which the Module Networks algorithm (Segal et al., 2003) used in order to ‘‘explain’’ the expression of each of the 80 modules (Experimental Procedures). For example, the algorithm associated module 865 (Figure S3, bottom) with five regulators, most prominently PBX1 (‘‘top regulator’’) and SOX4 (‘‘2nd level
Figure 6. Lineage-Specific Regulation of TF Expression Signature TF genes with lineage-specific expression in the five main lineages. Shown are the expression levels of the top 50 marker TF genes (rows) selected for each of four major lineages plus hematopoietic stem and progenitor cells (labels as in Figure 1). Genes were selected by high expression in one lineage compared to the others (t test). High expression is shown in red and low expression in blue; the expression of each gene is normalized to a mean expression of zero across all the samples. See also Figure S5 and Figure S6.
regulator’’) (Figure S3, top). It predicts that, when both PBX1 and SOX4 are induced (in HSCs, CMPs, MEPs, GMPs, early ERY, and early MEGA cells), the module’s genes are induced too. PBX1 is an established regulator of HSPCs, and SOX4 has recently been shown to be a direct target of HOXB4, a known HSC regulator (Lee et al., 2010), supporting the algorithm’s result. The regulators were chosen by their expression alone, and though the model chooses one combination of ‘‘representative’’ regulators, there may be several highly similar TFs that could fulfill the role. We next interpreted these regulatory connections within the context of the lineage tree. We associated each regulator with the tree positions (Figure 4 and Figure S4, arrowheads), in which a change in the regulator’s expression is associated with a change in the module’s expression. For example, there are four such positions for PBX1 and SOX4 in module 865 (Figure 4A, arrowheads), such as the association between the repression of PBX1 and the repression of the module in differentiation toward lymphoid lineages (Figure 4A, downward arrows, labeled PBX1). In this way, we predict the roles of distinct TFs at distinct
differentiation points, such as MNDA at the granulocyte/monocytes progenitor (Figure 4C and Figure S4G) or NCOA4 and KLF1 at late erythrocytes (Figure S4D). Overall, the algorithm associated 220 TFs (Table S3) with at least one regulatory program and 63 TFs as top regulators (e.g., Figure S3, top) of at least one module. These include 15 TFs previously associated with hematopoiesis (e.g., TAL1, KLF1, BCL11b, LMO2, and MYB) and 7 associated with differentiation in other systems (e.g., CREG1, MEF2A, and NHLH2). For example, we correctly found HOXA9 associated with HSPCs and early erythroid induction (module 679); NFE2, RXRA, KLF1, and FOXO3 associated with late erythroid induction (modules 727, 895, 889, and 739) (Figure 4B and Figure S5D); HIVEP2 and BCL11b associated with T cell induction (modules 859, 949, and 667); and HOXC4 and POU2AF1 associated with B cell induction (module 589) (Figure 4D). In addition, the algorithm predicted a regulatory role for proteins that were not previously associated with regulating hematopoietic differentiation (e.g., MNDA and NCOA4). The selected regulators are enriched for TFs that are known to participate as 30 partners in fusions in hematologic Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc. 303
Expression regulator Sequence regulator
A
Time
10d
ERY
10d Lineage GM
ERY PROG
ERY
GRAN
MONO
MNDA CEBPD CEBPA VDR ATF3 ELF4 TRIOBP SPI1 CREB5 BCL6 HDAC4 FOS EGR2 RNF13 ZFP106 MEFV CEBPG TFCP2 CEBPB MTF1 SP1 HOXA5 HIF3A KLF3 KLF1 AFF1 E2F1 CDK8 GATA1 COPS2 TAL1 YY1 FOXO3A
[87]
[64]
[64]
[94]
[79]
D
Control (Luc)
104
7.7%
FOXO3A
104
1.6%
18.1%
103
102
102
102
101
101
101
52.1% 102
103
104
HIF3A
104
15.5%
10
101
102
103
104
YY1
4
2.0%
26.0%
100 10
2.7%
102
102
102
104
2 SP
I1
sh
sh
sh 3A
3A
3.1%
22.0%
100 103
104
AFF1
8.8%
100
103
101
101
37.5% 102
102
10.2% 103
101
101
4
103
101
XO FO
60.1% 100
100
103
100
1.3%
9.1% 100
101
XO
1
2 sh
sh
F1
F1
1.1%
103
100
SPI1
104
1.6%
103
100
CD11b
AF
GM
[82]
FO
ERY
AF
luc
sh
AF Sequence reg. Expression reg.
[81]
2
[67]
sh
[94]
I1
* *
1
* *
1
*
* *
2
**
*
sh
*
* *
3A
** ** **
* *
HIF
*
* *
h2
*
* *
0 [Knockdown (%)] [80]
1
sh
**
*
2
1s
*
F1 KL F GA 1 T FO A1 XO 3A YY 1 HIF 3A TA L CO 1 PS 2 SP 1 MN DA EL F4 CE BP A E2 F1 CE BP B VD CE R BP D SP I1
** ** ** ** **
*
3
3A
0
4
YY
LUC
5
5
HIF
10
*
Cord blood
h1
15
Adult bone marrow 6
1s
GlyA/CD11b
20
7
SP
Normalized GlyA / CD11b ratio
Erythroid to myelomonocytic ratio
1
C
25
YY
B
100 100
101
102
103
104
100
101
102
103
104
GlyA
Figure 7. Experimental Validation of 33 TFs (A) The expression of 33 TFs was detected in primary human bone marrow CD34+ progenitor cells undergoing differentiation in vitro, harvested at 12 time points between days 3 to 10 of differentiation, and detected by a multiplexed assay using LMA followed by fluorescent bead-based detection (left heat map). In the heat map in the right panel, the expression of the same TFs in the original Affymetrix data set is illustrated. The labels at the far left indicate whether the TF was chosen as a regulator in the expression-based model or in the sequence-based model. (B) Differentiation following TF silencing with shRNA. Human bone marrow CD34+ cells expressing shRNAs targeting TFs were induced to differentiate in vitro for 10 days, and the ratio of erythroid (glycophorin A-positive) and myelomonocytic (CD11b-positive) cells was measured by flow cytometry. Each black dot represents an individual shRNA (mean of three replicates), and bars indicate their average. The effect of a control shRNA targeting the luciferase gene, which is not
304 Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc.
cancers (Mitelman et al., 2010) (25 of the regulators; p < 0.028), consistent with a regulatory role in hematopoiesis. Finally, we compared the predictions of the expression- and sequence-based models. The two models were different due to two reasons. First, 85% of the TFs chosen as regulators in the expression model (187 of 220) do not have a characterized binding motif in current databases and cannot be identified in the sequence model. Second, 29 of 41 TFs (70%) whose known sites are incorporated in the sequence model and appear in the expression model show little or no correlation in expression (absolute Pearson < 0.4) to the module with which they are associated in the sequence model (data available on http://www. broadinstitute.org/dmap/). Thus, the two models are likely complementary, each capturing a substantial but distinct number of known regulators in the relevant states. To gain confidence in their predictions, we next pursued experimental approaches. Direct Targets of MEIS1, TAL1, IKAROS, and PU.1 in HSPCs Reveal Dense Circuits and Anticipatory Binding To validate and further investigate the gene modules and ciscircuits, we examined the direct binding of TFs across the genome using chromatin immunoprecipitation followed by sequencing (ChIP-Seq) in HSPCs. We analyzed the binding of MEIS1, TAL1, PU.1/SPI1, and IKAROS/IKZF1, four key regulators of the specification, maintenance, or differentiation of HSCs (Argiropoulos et al., 2007; Le´cuyer and Hoang, 2004; Ng et al., 2007; Singh et al., 1999), in two replicates, often in independently expanded populations of primary human HSPCs (Extended Experimental Procedures). We scored each experiment for statistically significant binding (Extended Experimental Procedures and Table S5) and tested each of our expression modules for enrichment in binding events (Table S5). In modules whose genes are highly induced in terminal differentiation, we found enrichment of binding by corresponding lineage specific factors in HSPCs, suggesting anticipatory regulation. For example, module 727 (Figure 4B), expressed in terminally differentiated erythroid cells, was enriched with target genes bound in HSPCs by TAL1, an erythroid transcription factor (Table S5). Similarly, genes in the granulocyte/monocyte module 763 were enriched for targets bound by PU.1 in HSPCs (Table S5), and genes in the lymphoid module 949 were enriched for target genes bound by IKAROS in HSPCs (Table S5). In many (but not all) cases, expression of the target module is already moderate in HSCs and increases with differentiation. This strongly supports an anticipatory regulation in which relevant differentiation TFs are bound at target promoters in HSPCs, resulting in mild expression of targets that persists and further increases upon differentiation. Some of our expression-based model’s predictions for HSPCs are supported by the ChIP-Seq data. For example, the two
modules that are induced in HSPCs and are associated in our model with either MEIS1 (module 961) or its known binding partner PBX1 (module 865, Figure 4A) are enriched in target genes bound by MEIS1. MEIS1 and HOXA9 are members of module 865, consistent with MEIS1’s autoregulatory binding (Table S5). The ChIP-Seq data also support module reuse. For example, several of the modules enriched with PU.1 are reused in granulocytes and B lymphoid cells (e.g., modules 853, 649, 979, 769, and 817), consistent with an established role for PU.1 in both lineages. In other cases, module reuse may be mediated by combinatorial binding of two factors (e.g., by both PU.1 and IKAROS in module 607, which is expressed in granulocytes, monocytes, and some lymphoid cells). The individual binding events in our profiles also support the overall organization observed in the cis-circuits in the sequence model. First, three of the factors bind their own promoter (IKAROS and MEIS1) or enhancer (PU.1), forming autoregulatory loops, as observed for many known master regulators (Boyer et al., 2005) and in our sequence model. Second, PU.1, IKAROS, and MEIS1 are integrated in a feed-forward loop. Third, there is a significant overlap between the targets of any pair of factors (Table S5). Finally, in aggregate, the factors bind 13 of the 23 other TFs in our HSC circuit, further increasing its density. Differential Expression of Candidate Transcription Factors during In Vitro Differentiation We confirmed the lineage-specific expression of 33 TFs in primary human hematopoietic progenitor cells induced to differentiate in vitro. We focused on the erythroid and myelomonocytic lineages, as differentiation of primary human hematopoietic progenitor cells can be faithfully recapitulated and genetically manipulated along these lineages in vitro. We selected a set of 33 TFs identified in either the sequence or gene expression-based models as candidate regulators of these two lineages. We developed a quantitative, multiplexed assay to detect the expression of the signature genes in a single well using ligationmediated amplification (LMA) followed by amplicon detection on fluorescent beads (Peck et al., 2006). We cultured primary human CD34+ cells from adult bone marrow in vitro in cytokine conditions promoting either erythroid or myelomonocytic differentiation. We harvested cells at 12 time points between days 3 and 10 of erythroid and myelomonocytic differentiation and determined TF gene expression using the multiplexed beadbased assay. We confirmed that the 33 TFs are differentially expressed between the two lineages, providing a robust expression signature that can distinguish between the two states independent of profiling platform in cells derived from adult bone marrow or umbilical cord blood and in cells that differentiated in vivo or in vitro (Figure 7A).
expressed in human cells, is indicated with a dashed line. Below the shRNA labels, * or ** indicates p < 0.05 for one or both shRNAs, respectively. (Bottom) Classification of the TFs according to their roles in the expression-based and sequence-based models and to their induction pattern in the LMA profiling. (C) The effects of additional shRNAs targeting candidate TFs expressed in CD34+ cells derived from both umbilical cord blood and adult bone marrow and assayed as in (B) (*p < 0.01). (D) Representative flow cytometry scatter plots from shRNAs expressed in umbilical cord blood. See additional information in Table S5, Table S6, and Table S7.
Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc. 305
Changes in Expression Levels in Transcription Factor Circuits Functionally Modulate Differentiation In Vitro We next tested whether acute loss of expression of each TF using RNA interference can functionally affect erythroid and myelomonocytic differentiation. We used our multiplexed bead-based assay to identify short hairpin RNAs (shRNAs) that effectively knock down each TF and found 17 genes with at least two different effective shRNAs. Next, we infected primary human adult bone marrow CD34+ cells with the validated lentiviral shRNAs, cultured the cells in cytokine conditions supporting both erythroid and myelomonocytic differentiation, and assessed the number of erythroid (glycophorin A-positive) cells relative to myelomonocytic (CD11b-positive) cells by flow cytometry (Figure 7B). In most cases, the shRNA perturbation dramatically altered differentiation, with the ratio of erythroid to myeloid cells ranging from less than 1:10 to more than 10:1 with different shRNAs. The perturbations associated with the lowest fraction of erythroid cells in culture corresponded to the samples expressing shRNAs targeting nine TFs expressed at higher levels in the erythroid lineage (Table S6). Consistent with our models, six were regulators in either the expression or the sequence model, and the other three were members of erythrocyte-induced modules (Figure 7B, bottom). These include GATA-1 and KLF1, TFs with well-established roles in erythroid differentiation (Funnell et al., 2007; Pevny et al., 1991), and TAL1 and FOXO3A, which have been implicated in erythroid differentiation (Aplan et al., 1992; Bakker et al., 2007). The TF YY1 was identified in our sequence-based models, has higher expression in erythroid cells, and was functionally validated by our shRNA screen. A physical association between YY1 and GATA-1 was reported in the chicken a-globin enhancer (Rinco´n-Arano et al., 2005). Finally, we validated a new role for HIF3A and AFF1 (AF4) in erythroid differentiation based on module membership and perturbation. Of note, AFF1 is a common translocation partner with the MLL gene in leukemia (Li et al., 1998). Conversely, eight perturbations resulted in the lowest fraction of myelomonocytic cells and corresponded to samples expressing shRNAs targeting seven TFs induced in granulocyte/monocyte cells and one (E2F1) with higher expression in erythroid cells. Four TFs were predicted by the expression model to regulate modules induced in granulocytes/monocytes, and five were predicted in the sequence network (Figure 7B, bottom). These include the well-established granulocyte/monocyte TFs, PU.1/SPI1 and C/EBP family members (Hirai et al., 2006; Scott et al., 1994), and VDR, a gene that has been implicated in myeloid differentiation (Liu et al., 1996). We further validated three TFs that had not previously been associated with erythroid differentiation (AFF1, HIF3A, and YY1) alongside a known erythroid regulator (FOXO3A) and a known granulocyte regulator (PU.1/SPI1) (Figures 7C and 7D). We tested additional shRNAs for each gene by quantitative PCR and identified two shRNAs per gene that decrease expression of their target genes by 63% to 95% in human CD34+ cells derived from both adult bone marrow and umbilical cord blood. Using flow cytometry for lineage-specific markers following 10 days of differentiation, we validated our initial findings that AFF1, HIF3A, and YY1 decrease the relative production of erythroid 306 Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc.
lineage cells. These results are further supported by profiling mRNA levels following knockdown of these five TFs at 4 days following lentiviral infection. Compared to a control shRNA, gene expression in cells expressing AFF1, HIF3A, and FOXO3A were anticorrelated with erythroid profiles and positively correlated with granulocytes (Figure S5C), and knockdown of PU.1/SPI1 had the inverse pattern, as expected. Knockdown of YY1 caused a transcriptional profile more similar to HSCs, indicating a more substantial block in terminal differentiation. Taken together, our findings indicate that modulating the expression of TF genes can powerfully alter hematopoietic differentiation. Web-Based Portal as a Research Resource To facilitate interrogation of our hematopoietic gene expression database by the broader scientific community, we have created a Web-based portal (http://www.broadinstitute.org/dmap) to provide access to the primary data, sample information, processed results from both models, and a suite of analytic tools. DISCUSSION General Principles of Transcriptional Circuits in Differentiation The changes in gene expression over the course of hematopoietic differentiation are profound. The number of differentially expressed genes is similar within hematopoiesis and across human tissues, suggesting comparable complexity. Our findings reveal several major principles about the organization of this transcriptional program. Gene expression in hematopoiesis can be decomposed into modules of tightly coexpressed genes, some of which are restricted to specific lineages, whereas most are reused in multiple lineages. Furthermore, a module’s transcriptional state persists through multiple differentiation steps. For example, the transcriptional state of HSCs is not switched off immediately but instead persists with gradually decreasing expression in progenitor cells. Many of the TFs with known binding sites can be assembled into densely interconnected circuits. These can provide a mechanism for robust gene regulation in both terminally differentiated cells and HSCs. Because the binding sites for many factors remain unknown, we therefore expect that the circuit’s density and complexity is even higher. A large number of TFs are differentially expressed across hematopoiesis, often in tightly coregulated modules, and at comparable complexity to that of the other (nonregulatory) genes. Leveraging this correspondence, we associated TFs to the modules and differentiation states that they may regulate. We automatically rediscovered (without prior knowledge) many of the key known TFs and predict regulatory functions for numerous additional TFs. By monitoring the binding of four major TFs in HSPCs, we found that anticipatory regulation may be a major feature of these circuits. In such cases, TFs that direct lineage-specific differentiation bind a significant portion of their target genes in HSPCs. These target genes are often moderately expressed in the stem and progenitor cells, with substantial further induction as differentiation progresses in the relevant lineage. This is
consistent with the concept of ‘‘lineage priming’’ in HSCs (Akashi, 2005), providing flexibility in cell fate commitments. Discovering and Validating Transcriptional Regulators in Hematopoiesis Our examination of the global architecture of hematopoietic differentiation offers a complementary strategy to studies of individual genes in murine models. In this approach, gene expression and sequence-based analyses nominate a host of candidate regulators and point to groups of factors that may act together and hence introduce redundancies. The two computational approaches complement each other: the expression model may identify factors whose binding specificity is unknown, the sequence model may help detect those factors whose mRNA levels do not change or do not correspond to changes in targets (Lu et al., 2009). We used a perturbation-based approach to validate TFs derived from the sequence and expression models in an in vitro differentiation system. Modulating expression of candidate TFs with RNA interference altered differentiation of hematopoietic progenitor cells in vitro in the direction predicted by our models. We reconfirmed the role of several known factors and identified several new ones (e.g., YY1, AFF1, and HIF3A). In vitro manipulation can be more sensitive than genetic ablation experiments in vivo, wherein perturbations may be corrected by homeostatic mechanisms, such as cytokine or transcriptional feedback loops. A Transcriptional Roadmap for Hematologic Malignancies Balanced translocations involving TFs play a major role in the pathogenesis of human leukemias. Of 200 known translocations in AML (Mitelman et al., 2010), there are 53 in which at least one translocation partner is a TF, 16 as a 50 partner, and 43 as a 30 partner (6 in both). Twenty-five of these 43 known 30 partners are among the 220 regulators in the expression model (p < 0.028), and 5 of the known 30 partners are among the 72 known TFs in the sequence model. Furthermore, 50 partners are enriched in genes expressed in HSPCs. These results support the role of chosen regulators in differentiation and are consistent with a broader paradigm in which lineage-specific promoters can dysregulate key TFs to disrupt differentiation (Rosenbauer and Tenen, 2007; Tomlins et al., 2005). Impaired or blocked hematopoietic differentiation is a defining characteristic of leukemia, and the gene expression profiles of leukemias cluster strongly into subgroups that correspond to specific molecular subgroups (Bullinger et al., 2004; Tamayo et al., 2007; Valk et al., 2004). Gene signatures induced in various leukemias significantly overlap those induced in normal hematopoiesis (Figures S7A and S7B). In most cases, there is a coherent overlap between the leukemia subtype and the cell type from which it is known to arise. However, human leukemias often express more complex combinations of modules that are not observed in normal samples, including HSPC modules, as has been reported in murine models of leukemia (Krivtsov et al., 2006). Toward a Programming ‘‘Code’’ of Hematopoietic Differentiation A more complete understanding of hematopoietic differentiation will likely require an integration of gene expression data with
other genomic data, including epigenetic analyses, genomewide ChIP-Seq studies, proteomics, and systematic functional studies. Given the ability to produce high-quality measurements from small numbers of cells, gene expression data provides a first draft of the transcriptional program controlling hematopoiesis, opening the way to manipulate and reprogram these circuits, through perturbation and manipulation of each regulatory factor. These can highlight avenues for therapeutic intervention, including ‘‘reprogramming’’ of cells to more desired states. Though deriving mechanistic models from mammalian gene expression profiling data has been challenging, hematopoiesis provides a paradigm for the testing of more advanced algorithms (Kim et al., 2009). Our data set and analyses provide a resource for further inquiries into normal and pathologic hematopoietic differentiation in humans. EXPERIMENTAL PROCEDURES Further details for data analysis, chromatin immunoprecipitation, and functional validation experiments are described in the Extended Experimental Procedures.
Subjects and Samples Human umbilical cord blood was harvested from postpartum placentas at Brigham and Women’s Hospital under an Institution Review Board (IRB)approved protocol. Peripheral blood samples were obtained from healthy volunteers at the Dana-Farber Cancer Institute with informed consent under an IRB-approved protocol. The majority of cells were purified from umbilical cord blood, an enriched source of undifferentiated populations. However, terminally differentiated lymphocyte populations, including T cells (TCELL1-8), B cells (BCELL1-4), natural killer cells (NKa1, NKa2, NKa3, and NKT), and dendritic cells (DENDa1 and DENDa2), were purified from adult peripheral blood because terminal differentiation in these populations requires exposure to antigens after birth. For each cell population, we purified samples from four to seven distinct donors. All blood samples were harvested fresh and immediately processed for flow sorting.
Cell Sorting Strategy and Flow Cytometry First, mononuclear cells were isolated by Ficoll-Hypaque sedimentation. For relatively rare populations, including hematopoietic stem cell populations (HSC1 and HSC2), progenitor populations such as common myeloid progenitor (CMP), megakaryocyte/erythroid progenitor (MEP), granulocyte-monocyte progenitor (GMP), and the erythroid lineage populations (ERY1–5), lineage depletion was performed using antibodies against CD2, CD3, CD4, CD5, CD8, CD11b, CD14C, CD19, and CD56 with a magnetic column (Miltenyi Biotec, Auburn, CA). Positive selection was then performed using flow cytometry for labeled antibodies to the markers described in Table S1. For the more common or terminally differentiated populations, including neutrophil populations (GRAN1-3), basophils (BASO1), monocytes (MONO1–2), eosinophils (EOS2), megakaryocytes (MEGA1–2), B-lymphoid progenitor (PRE_BCELL1), pro and early B lymphocytes (PRE_BCELL2 and PRE_BCELL3), dendritic cells (DENDa1 and DENDa2), mature T cells (TCELL1–8), mature B cells (BCELL1–4), and natural killer cells (NKa1–3, NKT), cells were positively selected using flow scatter properties and antibodies based on the immunophenotypes described in Table S7. The gene expression profiles for a subset of the lymphoid populations has been analyzed previously (Haining et al., 2008). Sorting was performed with Vantage SE. Diva or FACSAria flow cytometers (Becton Dickinson, San Jose, CA). Cell populations of interest were collected into tubes containing PBS in a collection unit at 4 C. The > 95% purity of populations was confirmed by performing FACS analysis of the sorted cells. Sorted cells were spun down, immediately resuspended in TriZol (Invitrogen, San Diego, CA), and stored at 70 C.
Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc. 307
Microarray Data Acquisition Total RNA was isolated from TriZol. The concentration of RNA was quantified using the RiboGreen RNA Quantitation Kit (Invitrogen, San Diego, CA). Ten nanograms of total RNA were amplified using the Ovation Biotin RNA Amplification and Labeling System (NuGEN, San Carlos, CA). The cDNA was fragmented, labeled, and hybridized to Affymetrix HG_U133AAofAv2 microarrays (Affymetrix, Santa Clara, CA), which contain 22,944 probes. Expression-Based Module Networks Model The modules and their regulation programs were automatically learned using the Module Networks algorithm (Segal et al., 2003). This method detects modules of coexpressed genes and their shared regulation programs. The regulation program is a small set of genes whose expression is predictive of the expression level of the module genes using a decision (regression) tree structure. Given the expression values and a pool of candidate regulator genes, a set of modules and their associated regulation programs are automatically inferred by an iterative procedure. This procedure searches for the best gene partition into modules and for the regulation program of each module while optimizing a target function. The target function is the Bayesian score derived from the posterior probability of the model (see Segal et al., 2005 for a detailed description of the algorithm).
Argiropoulos, B., Yung, E., and Humphries, R.K. (2007). Unraveling the crucial roles of Meis1 in leukemogenesis and normal hematopoiesis. Genes Dev. 21, 2845–2849. Bakker, W.J., van Dijk, T.B., Parren-van Amelsvoort, M., Kolbus, A., Yamamoto, K., Steinlein, P., Verhaak, R.G., Mak, T.W., Beug, H., Lo¨wenberg, B., and von Lindern, M. (2007). Differential regulation of Foxo3a target genes in erythropoiesis. Mol. Cell. Biol. 27, 3839–3854. Bhattacharjee, A., Richards, W.G., Staunton, J., Li, C., Monti, S., Vasa, P., Ladd, C., Beheshti, J., Bueno, R., Gillette, M., et al. (2001). Classification of human lung carcinomas by mRNA expression profiling reveals distinct adenocarcinoma subclasses. Proc. Natl. Acad. Sci. USA 98, 13790–13795. Boyer, L.A., Lee, T.I., Cole, M.F., Johnstone, S.E., Levine, S.S., Zucker, J.P., Guenther, M.G., Kumar, R.M., Murray, H.L., Jenner, R.G., et al. (2005). Core transcriptional regulatory circuitry in human embryonic stem cells. Cell 122, 947–956. Brunet, J.-P., Tamayo, P., Golub, T.R., and Mesirov, J.P. (2004). Metagenes and molecular pattern discovery using matrix factorization. Proc. Natl. Acad. Sci. USA 101, 4164–4169.
ACCESSION NUMBERS
Bullinger, L., Do¨hner, K., Bair, E., Fro¨hling, S., Schlenk, R.F., Tibshirani, R., Do¨hner, H., and Pollack, J.R. (2004). Use of gene-expression profiling to identify prognostic subclasses in adult acute myeloid leukemia. N. Engl. J. Med. 350, 1605–1616.
Data set is available on http://www.ncbi.nlm.nih.gov/geo/, GSE24759.
Cantor, A.B., and Orkin, S.H. (2002). Transcriptional regulation of erythropoiesis: an affair involving multiple partners. Oncogene 21, 3368–3376.
SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, seven figures, and seven tables and can be found with this article online at doi:10. 1016/j.cell.2011.01.004. ACKNOWLEDGMENTS We thank E. Lander, I. Amit, and I. Gat-Viks for critical review of the manuscript; D. Scadden for helpful discussions; L. Gaffney and S. Hart for assistance with figure generation; and D. Peck, J. Lamb, R. Onofrio, and the Broad Genetic Analysis Platform for assistance with expression arrays. The work was funded by the NIH (grants R01 HL082945 and P01 CA108631 to B.L.E. and the PIONEER award to A.R.), the Burroughs-Wellcome Fund (CAMS to B.L.E. and CASI to A.R.), funds from Landon and Lavinia Clay (R.A.Y.) and HHMI (T.R.G. and A.R.). A.R. is an investigator of the Merkin Foundation for Stem Cell Research at the Broad Institute. J.S.S. is an employee of NuGEN Technologies, Inc. Received: June 19, 2010 Revised: October 18, 2010 Accepted: January 4, 2011 Published: January 20, 2011 REFERENCES Akashi, K. (2005). Lineage promiscuity and plasticity in hematopoietic development. Ann. N Y Acad. Sci. 1044, 125–131. Amit, I., Garber, M., Chevrier, N., Leite, A.P., Donner, Y., Eisenhaure, T., Guttman, M., Grenier, J.K., Li, W., Zuk, O., et al. (2009). Unbiased reconstruction of a mammalian transcriptional network mediating pathogen responses. Science 326, 257–263.
Chen, M.L., Logan, T.D., Hochberg, M.L., Shelat, S.G., Yu, X., Wilding, G.E., Tan, W., Kujoth, G.C., Prolla, T.A., Selak, M.A., et al. (2009). Erythroid dysplasia, megaloblastic anemia, and impaired lymphopoiesis arising from mitochondrial dysfunction. Blood 114, 4045–4053. Chin, K., DeVries, S., Fridlyand, J., Spellman, P.T., Roydasgupta, R., Kuo, W.L., Lapuk, A., Neve, R.M., Qian, Z., Ryder, T., et al. (2006). Genomic and transcriptional aberrations linked to breast cancer pathophysiologies. Cancer Cell 10, 529–541. Davidson, E.H. (2001). Genomic Regulatory Systems: In Development and Evolution (Burlington, MA: Academic Press). Ebert, B.L., and Golub, T.R. (2004). Genomic approaches to hematologic malignancies. Blood 104, 923–932. Fontenay, M., Cathelin, S., Amiot, M., Gyan, E., and Solary, E. (2006). Mitochondria in hematopoiesis and hematological diseases. Oncogene 25, 4757–4767. Funnell, A.P., Maloney, C.A., Thompson, L.J., Keys, J., Tallack, M., Perkins, A.C., and Crossley, M. (2007). Erythroid Kru¨ppel-like factor directly activates the basic Kru¨ppel-like factor gene in erythroid cells. Mol. Cell. Biol. 27, 2777–2790. Hahm, K., Cobb, B.S., McCarty, A.S., Brown, K.E., Klug, C.A., Lee, R., Akashi, K., Weissman, I.L., Fisher, A.G., and Smale, S.T. (1998). Helios, a T cellrestricted Ikaros family member that quantitatively associates with Ikaros at centromeric heterochromatin. Genes Dev. 12, 782–796. Haining, W.N., Ebert, B.L., Subrmanian, A., Wherry, E.J., Eichbaum, Q., Evans, J.W., Mak, R., Rivoli, S., Pretz, J., Angelosanto, J., et al. (2008). Identification of an evolutionarily conserved transcriptional signature of CD8 memory differentiation that is shared by T and B cells. J. Immunol. 181, 1859–1868. Hirai, H., Zhang, P., Dayaram, T., Hetherington, C.J., Mizuno, S., Imanishi, J., Akashi, K., and Tenen, D.G. (2006). C/EBPbeta is required for ‘emergency’ granulopoiesis. Nat. Immunol. 7, 732–739. Iwasaki, H., and Akashi, K. (2007). Hematopoietic developmental pathways: on cellular basis. Oncogene 26, 6687–6696.
Aplan, P.D., Nakahara, K., Orkin, S.H., and Kirsch, I.R. (1992). The SCL gene product: a positive regulator of erythroid differentiation. EMBO J. 11, 4073–4081.
Kim, H.D., Shay, T., O’Shea, E.K., and Regev, A. (2009). Transcriptional regulatory circuits: predicting numbers from alphabets. Science 325, 429–432.
Aramburu, J., Azzoni, L., Rao, A., and Perussia, B. (1995). Activation and expression of the nuclear factors of activated T cells, NFATp and NFATc, in human natural killer cells: regulation upon CD16 ligand binding. J. Exp. Med. 182, 801–810.
Kino, T., Takatori, H., Manoli, I., Wang, Y., Tiulpakov, A., Blackman, M.R., Su, Y.A., Chrousos, G.P., DeCherney, A.H., and Segars, J.H. (2009). Brx mediates the response of lymphocytes to osmotic stress through the activation of NFAT5. Sci. Signal. 2, ra5.
308 Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc.
Krivtsov, A.V., Twomey, D., Feng, Z., Stubbs, M.C., Wang, Y., Faber, J., Levine, J.E., Wang, J., Hahn, W.C., Gilliland, D.G., et al. (2006). Transformation from committed progenitor to leukaemia stem cell initiated by MLL-AF9. Nature 442, 818–822. Le´cuyer, E., and Hoang, T. (2004). SCL: from the origin of hematopoiesis to stem cells and leukemia. Exp. Hematol. 32, 11–24. Lee, H.M., Zhang, H., Schulz, V., Tuck, D.P., and Forget, B.G. (2010). Downstream targets of HOXB4 in a cell line model of primitive hematopoietic progenitor cells. Blood 116, 720–730. Li, Q., Frestedt, J.L., and Kersey, J.H. (1998). AF4 encodes a ubiquitous protein that in both native and MLL-AF4 fusion types localizes to subnuclear compartments. Blood 92, 3841–3847. Liu, M., Lee, M.H., Cohen, M., Bommakanti, M., and Freedman, L.P. (1996). Transcriptional activation of the Cdk inhibitor p21 by vitamin D3 leads to the induced differentiation of the myelomonocytic cell line U937. Genes Dev. 10, 142–153. Look, A.T. (1997). Oncogenic transcription factors in the human acute leukemias. Science 278, 1059–1064. Lu, R., Markowetz, F., Unwin, R.D., Leek, J.T., Airoldi, E.M., MacArthur, B.D., Lachmann, A., Rozov, R., Ma’ayan, A., Boyer, L.A., et al. (2009). Systems-level dynamic analyses of fate change in murine embryonic stem cells. Nature 462, 358–362. McLean, C.Y., Bristor, D., Hiller, M., Clarke, S.L., Schaar, B.T., Lowe, C.B., Wenger, A.M., and Bejerano, G. (2010). GREAT improves functional interpretation of cis-regulatory regions. Nat. Biotechnol. 28, 495–501. Mitelman, F., Johansson, B., and Mertens, F. (2010). Mitelman Database of Chromosome Aberrations and Gene Fusions in Cancer (http://cgap.nci.nih.gov/Chromosomes/Mitelman). Monti, S., Savage, K.J., Kutok, J.L., Feuerhake, F., Kurtin, P., Mihm, M., Wu, B., Pasqualucci, L., Neuberg, D., Aguiar, R.C., et al. (2005). Molecular profiling of diffuse large B-cell lymphoma identifies robust subtypes including one characterized by host inflammatory response. Blood 105, 1851–1861. Mu¨ller, F.J., Laurent, L.C., Kostka, D., Ulitsky, I., Williams, R., Lu, C., Park, I.H., Rao, M.S., Shamir, R., Schwartz, P.H., et al. (2008). Regulatory networks define phenotypic classes of human stem cell lines. Nature 455, 401–405. Ng, S.Y., Yoshida, T., and Georgopoulos, K. (2007). Ikaros and chromatin regulation in early hematopoiesis. Curr. Opin. Immunol. 19, 116–122. Orkin, S.H., and Zon, L.I. (2008). Hematopoiesis: an evolving paradigm for stem cell biology. Cell 132, 631–644. Peck, D., Crawford, E.D., Ross, K.N., Stegmaier, K., Golub, T.R., and Lamb, J. (2006). A method for high-throughput gene expression signature analysis. Genome Biol. 7, R61. Pevny, L., Simon, M.C., Robertson, E., Klein, W.H., Tsai, S.F., D’Agati, V., Orkin, S.H., and Costantini, F. (1991). Erythroid differentiation in chimaeric mice blocked by a targeted mutation in the gene for transcription factor GATA-1. Nature 349, 257–260. Quesenberry, P.J., and Colvin, G.A. (2005). Hematopoietic Stem Cells, Progenitor Cells, and Cytokines. In Williams Hematology, E. Beutler, M.A. Lichtman, B.S. Coller, T.J. Kipps, and U. Seligsohn, eds. (New York: McGraw-Hill), pp. 153–174.
Sandelin, A., Alkema, W., Engstro¨m, P., Wasserman, W.W., and Lenhard, B. (2004). JASPAR: an open-access database for eukaryotic transcription factor binding profiles. Nucleic Acids Res. 32(Database issue), D91–D94. Scott, E.W., Simon, M.C., Anastasi, J., and Singh, H. (1994). Requirement of transcription factor PU.1 in the development of multiple hematopoietic lineages. Science 265, 1573–1577. Segal, E., Shapira, M., Regev, A., Pe’er, D., Botstein, D., Koller, D., and Friedman, N. (2003). Module networks: identifying regulatory modules and their condition-specific regulators from gene expression data. Nat. Genet. 34, 166–176. Segal, E., Pe’er, D., Regev, A., Koller, D., and Friedman, N. (2005). Learning module networks. J. Mach. Learn. Res. 6, 557–588. Singh, H., DeKoter, R.P., and Walsh, J.C. (1999). PU.1, a shared transcriptional regulator of lymphoid and myeloid cell fates. Cold Spring Harb. Symp. Quant. Biol. 64, 13–20. Su, A.I., Wiltshire, T., Batalov, S., Lapp, H., Ching, K.A., Block, D., Zhang, J., Soden, R., Hayakawa, M., Kreiman, G., et al. (2004). A gene atlas of the mouse and human protein-encoding transcriptomes. Proc. Natl. Acad. Sci. USA 101, 6062–6067. Subramanian, A., Tamayo, P., Mootha, V.K., Mukherjee, S., Ebert, B.L., Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., and Mesirov, J.P. (2005). Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 102, 15545–15550. Suzuki, H., Forrest, A.R., van Nimwegen, E., Daub, C.O., Balwierz, P.J., Irvine, K.M., Lassmann, T., Ravasi, T., Hasegawa, Y., de Hoon, M.J., et al; FANTOM Consortium; Riken Omics Science Center. (2009). The transcriptional network that controls growth arrest and differentiation in a human myeloid leukemia cell line. Nat. Genet. 41, 553–562. Tamayo, P., Scanfeld, D., Ebert, B.L., Gillette, M.A., Roberts, C.W., and Mesirov, J.P. (2007). Metagene projection for cross-platform, cross-species characterization of global transcriptional states. Proc. Natl. Acad. Sci. USA 104, 5959–5964. Tibshirani, R., and Hastie, T. (2007). Outlier sums for differential gene expression analysis. Biostatistics 8, 2–8. Tomlins, S.A., Rhodes, D.R., Perner, S., Dhanasekaran, S.M., Mehra, R., Sun, X.W., Varambally, S., Cao, X., Tchinda, J., Kuefer, R., et al. (2005). Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science 310, 644–648. Valk, P.J., Verhaak, R.G., Beijen, M.A., Erpelinck, C.A., Barjesteh van Waalwijk van Doorn-Khosrovani, S., Boer, J.M., Beverloo, H.B., Moorhouse, M.J., van der Spek, P.J., Lo¨wenberg, B., and Delwel, R. (2004). Prognostically useful gene-expression profiles in acute myeloid leukemia. N. Engl. J. Med. 350, 1617–1628. Valouev, A., Johnson, D.S., Sundquist, A., Medina, C., Anton, E., Batzoglou, S., Myers, R.M., and Sidow, A. (2008). Genome-wide analysis of transcription factor binding sites based on ChIP-Seq data. Nat. Methods 5, 829–834.
Rinco´n-Arano, H., Valadez-Graham, V., Guerrero, G., Escamilla-Del-Arenal, M., and Recillas-Targa, F. (2005). YY1 and GATA-1 interaction modulate the chicken 30 -side alpha-globin enhancer activity. J. Mol. Biol. 349, 961–975.
Xu, R.H., Sampsell-Barron, T.L., Gu, F., Root, S., Peck, R.M., Pan, G., Yu, J., Antosiewicz-Bourget, J., Tian, S., Stewart, R., and Thomson, J.A. (2008). NANOG is a direct target of TGFbeta/activin-mediated SMAD signaling in human ESCs. Cell Stem Cell 3, 196–206.
Rosenbauer, F., and Tenen, D.G. (2007). Transcription factors in myeloid development: balancing differentiation with transformation. Nat. Rev. Immunol. 7, 105–117.
Zhang, Y., Liu, T., Meyer, C.A., Eeckhoute, J., Johnson, D.S., Bernstein, B.E., Nusbaum, C., Myers, R.M., Brown, M., Li, W., and Liu, X.S. (2008). Modelbased analysis of ChIP-Seq (MACS). Genome Biol. 9, R137.
Cell 144, 296–309, January 21, 2011 ª2011 Elsevier Inc. 309
Scientific Editor, Cell
Cell is seeking a scientist to join its editorial team. The minimum qualification for this position is a PhD in a relevant area of biomedical research, although additional postdoctoral or editorial experience is preferred. This is a superb opportunity for a talented individual to play a critical role in promoting science by helping researchers disseminate their findings to the wider community. As an editor, you would be responsible for assessing submitted research papers and overseeing the refereeing process, and you would commission and edit material for Cell's Leading Edge. You would also travel frequently to scientific conferences to follow developments in research and to establish and maintain close ties with the scientific community. You would have opportunities to pioneer and contribute to new trends in scientific publishing. The key qualities we look for are breadth of scientific interest and the ability to think critically about a wide range of scientific issues. The successful candidate will be highly motivated and creative and able to work independently as well as in a team. This is a full-time, in-house editorial position, based at the Cell Press office in Kendall Square near MIT in Cambridge, Massachusetts. Cell Press offers an attractive salary and benefits package and a stimulating working environment that encourages innovation.
To apply, please visit http://reedelsevier.taleo.net/careersection/51/jobdetail.ftl?lang=en&job=SCI0005W.
Scientific Editor, Molecular Cell
Molecular Cell is seeking a full-time scientific editor to join its editorial team. We will consider qualified candidates with scientific expertise in any area that the journal covers. The minimum qualification for this position is a PhD in a relevant area of biomedical research, although additional experience is preferred. This is a superb opportunity for a talented individual to play a critical role in the research community away from the bench. As a scientific editor, you would be responsible for assessing submitted research papers, overseeing the refereeing process, and choosing and commissioning review material. You would also travel frequently to scientific conferences to follow developments in research and establish and maintain close ties with the scientific community. The key qualities we look for are breadth of scientific interest and the ability to think critically about a wide range of scientific issues. The successful candidate will also be highly motivated and creative and able to work independently as well as in a team. This is a full-time in-house editorial position, based at the Cell Press office in Cambridge, Massachusetts. Cell Press offers an attractive salary and benefits package and a stimulating working environment. Applications will be held in the strictest of confidence and will be considered on an ongoing basis until the position is filled.
To apply Please submit a CV and cover letter describing your qualifications, research interests, and reasons for pursuing a career in scientific publishing, as soon as possible, to our online jobs site: http://www.elsevier.com/wps/find/job_search.careers. Click on “search for US jobs” and select “Massachusetts.” Or: http://reedelsevier.taleo.net/careersection/51/jobdetail.ftl?lang=en&job=SCI0005X.
No phone inquiries, please. Cell Press is an equal opportunity/affirmative action employer, M/F/D/V.
Business Project Editor Cell Press is seeking a Business Project Editor to plan, develop, and implement projects that have commercial or sponsorship potential. By drawing on existing content or developing new material, the Editor will work with Cell Press's commercial sales group to create collections of content in print or online that will be attractive to readers and sponsors. The Editor will also be responsible for leveraging new online opportunities for engaging the readers of Cell Press journals. The successful candidate will have a PhD in the biological sciences, broad scientific interests, a fascination with technology, good commercial instincts, and a true passion for both science and science communication. They should be highly organized and dedicated, with excellent written and oral communication skills, and should be willing to work to tight deadlines. The position is full time and based in Cambridge, MA. Cell Press offers an attractive salary and benefits package and a stimulating work environment. Applications will be considered on a rolling basis. For consideration, please apply online and include a cover letter and resume.
To apply Please visit the career page at http://www.elsevier.com and search on keywords “Business Project Editor.”
No phone inquiries. Elsevier-Cell Press is an Equal Opportunity Employer.
EDITOR-IN-CHIEF
SENIOR EDITORS
ASSOCIATE EDITORS
F.E. Bloom La Jolla, CA, USA
J.F. Baker Chicago, IL, USA P.R. Hof New York, NY, USA G.R. Mangun Davis, CA, USA J.I. Morgan Memphis, TN, USA F.R. Sharp Sacramento, CA, USA R.J.Smeyne Memphis, TN, USA A.F. Sved Pittsburgh, PA, USA
G. Aston-Jones Charleston, SC, USA J.S. Baizer Buffalo, NY, USA J.D. Cohen Princeton, NJ, USA B.M. Davis Pittsburgh, PA, USA J. De Felipe Madrid, Spain M.A. Dyer Memphis, TN, USA M.S. Gold Pittsburgh, PA, USA G.F. Koob La Jolla, CA, USA
T.A. Milner New York, NY, USA S.D. Moore Durham, NC, USA T.H. Moran Baltimore, MD, USA T.F. Münte Magdeburg, Germany K-C. Sonntag Belmont, MA, USA R.J. Valentino Philadelphia, PA, USA C.L. Williams Durham,NC, USA
1
23
Twenty-three to the Power of One.
One re-unified journal, nine specialist sections, 23 receiving Editors ← Authors receive first editorial decision within 30 days of submission ← “Young Investigator Awards” for innovative work by a new generation of researchers ←
Brain Research take another look www.elsevier.com/locate/brainres
Positions Available
OPPORTUNITY AT THE UNIVERSITY OF GENEVA THE FACULTY OF MEDICINE of GENEVA is seeking applications for 3 positions of: ASSISTANT PROFESSOR OR ASSOCIATE PROFESSOR in the Section of Fundamental Medicine These full-time positions at the level of assistant or associate professor involve pre- and postgraduate teaching and research in the fields of metabolism or cardiovascular disease or biology of cancer or host pathogen interactions or stem cells. Candidates should have demonstrated experience in one of these specific fields. They must be able to develop and lead research programs in a particular aspect of the field and to coordinate research projects in collaboration with other medical specialties. They must be willing to participate in interdisciplinary projects as well as assuming all pertinent administrative tasks. A Doctorate of Science or Medicine (MD) is required and some knowledge of French is expected at term. The starting date for the position is June 1st 2011 , or according to agreement. Applications must be sent before March 1st 2011, to: The Dean of the Faculty of Medicine Centre médical universitaire 1 rue Michel-Servet 1211 Genève 4 - Switzerland Information concerning applications and job description are available from
[email protected] Tel. +41 22 379 50 26 Women are encouraged to apply
Chaired Professor Position Children’s Hospital Boston and Harvard Medical School The Vascular Biology Program at Children’s Hospital Boston is seeking applications for a chaired faculty position at the level of Professor in a field related to the study of Vascular Biology. The successful candidate will be appointed as the James and Thea Stoneman Professor of Vascular Biology at Harvard Medical School. Candidates for this position will hold a Ph.D. and/or M.D. degree and will have an internationally recognized record of research excellence and a successful history of obtaining independent funding. Opportunities to collaborate with scientists and clinicians whose areas of expertise extend from fundamental cellular and molecular mechanisms to clinical and translational research in the field of vascular biology will be available. Applicants whose primary interest is in vascular biology and/or diseases that have a vascular component are encouraged to apply. Interested applicants should forward their curriculum vitae and inquiries to: Faculty Search Committee Vascular Biology Program Karp Family Research Building, Room 11.124 Children’s Hospital Boston 300 Longwood Avenue Boston, MA 02115 Children’s Hospital Boston is an Equal Opportunity Affirmative Action Employer. Applications from qualified women and underrepresented minorities are encouraged.
Expand your stem cell library and save today on the latest books on stem cells and regenerative medicine Stem Cells
Stem Cell Anthology
Scientific Facts and Fiction
From Stem Cell Biology, Tissue Engineering, Regenerative Medicine, Cloning and Stem Cell Methods
Christine Mummery, Ian Wilmut, Anja Van de Stolpe and Bernard Roelen November 2010 | 400 pages | Paperback | $79.95 | €57.95 | £48.99 | ISBN: 9780123815354
Principles of Regenerative Medicine, 2nd Edition
Bruce M. Carlson October 2009 | 450 pp. | Hardback | $150.00 | €100.00 | £95.00 |AU$222.00 | ISBN: 9780123756824
Essential Stem Cell Methods
Anthony Atala and Robert Lanza November 2010 | 1400 pages | Hardback | $199.95 | €143.00 | £125.00 | ISBN: 9780123814227
A Volume in the Reliable Lab Solutions Series
Heart Development and Regeneration, 2-Volume Set
Tissue Engineering
Robert Lanza and Irina Klimanskaya April 2009 | 628 pp. | Paperback | $75.00 | €50.95 | £45.99 |AU$111.00 | ISBN: 9780123750617
Nadia Rosenthal and Richard P. Harvey June 2010 | 1072 pp. | Hardback | $199.95 | €143.00 | £125.00 | AU$296.00 | ISBN: 9780123813329
Clemens van Blitterswijk, Peter Thomsen, Jeffrey Hubbell, Ranieri Cancedda, Anders Lindahl Sahlgrenska, Jerome Sohier and David F. Williams March 2008 | 760 pp. | Hardback | $115.00 | €76.95 | £69.99 |AU$170.00 | ISBN: 9780123708694
Essentials of Stem Cell Biology, 2nd Edition
Human Stem Cell Manual
Robert Lanza, Roger Pedersen, John Gearhart, E. Donnall Thomas, Brigid Hogan, James Thomson, Douglas Melton and Sir Ian Wilmut June 2009 | 600 pp. | Hardback | $199.95 | €134.00 | £125.00 | AU$302.00 | ISBN: 9780123747297
Jeanne F. Loring, Robin L. Wesselschmidt and Philip H. Schwartz June 2007 | 488 pp. | Spiral bound | $88.95 | €59.95 | £53.99 |AU$132.00 | ISBN: 9780123704658
Foundations of Regenerative Medicine Clinical and Therapeutic Applications Anthony Atala, Robert Lanza, James Thomson and Robert Nerem September 2009 | 750 pp. | Hardback | $99.95 | €66.95 | £60.99|AU$148.00 | ISBN: 9780123750853
A Laboratory Guide
Handbook of Stem Cells 2-Volume Set with CD-ROM Vol. 1–2 Vol. 1 – Embryonic Stem Cells Vol. 2 – Adult & Fetal Stem Cells Robert Lanza, Roger Pedersen, Helen Blau, E. Donnall Thomas, John Gearhart, James Thomson, Brigid Hogan, Catherine Verfaillie, Douglas Melton, Irving Weissman, Malcolm Moore and Michael West September 2004 | 1,760 pp. | Hardback | $566.00 | €380.00 | £345.00 | AU$817.00 | ISBN: 9780124366435
Cell Stem Cell subscribers save 25% on their book order Secure ordering online at elsevierdirect.com Enter promo code 28024 at check out Prices and publication dates subject to change without notice.
T H E N E W E D I T I O N O F F U N D A M E N TA L N E U R O S C I E N C E I S N O W AVA I L A B L E !
FUNDAMENTAL NEUROSCIENCE THIRD EDITION EDITED BY:
Larry R. Squire, VA Medical Center and University of California, San Diego, La Jolla, California, USA February 2008 Hardback
SENIOR EDITORS: Darwin Berg, Floyd E. Bloom, Sascha du Lac, Anirvan Ghosh, and Nicholas C. Spitzer
1,280 pp. 9780123740199 Print list price: $99.95
This comprehensive text reference presents the entire spectrum of modern neuroscience. Addressing feedback on the previous edition, Fundamental Neuroscience, Third Edition is now more concise and reader-friendly than ever before. Each chapter is heavily illustrated and provides clinical boxes describing experiments, disorders, and methodological approaches and concepts. A companion site contains an image bank of the figures for use in poster presentations, slides, and handouts. Capturing the promise and excitement of this fast-moving field, Fundamental Neuroscience, Third Edition is the book that you will reference throughout your neuroscience career! NEW TO THIS EDITION: • 30% new material including new chapters on Dendritic Development and Spine Morphogenesis; Chemical Senses; Cerebellum; Eye Movements; Circadian Timing; Sleep and Dreaming; and Consciousness • Companion Web site with all figures and Web links to additional material • Multiple model system coverage beyond rats, mice, and monkeys • Extensively expanded index for easier referencing WHY BUY THIS EDITION? • It presents the entire spectrum of modern neuroscience as a cohesive reference rather than just a collection of review articles • It is written at a level appropriate for graduate-level students and researchers with various science backgrounds who need a solid foundation in neuroscience principles • Clinical conditions are discussed in neuroscience to help readers establish the connection between the basic biology and the neurological symptoms and syndromes • It includes more than 650 illustrations, including color photographs, imaging, and micrographs, also available on the companion Web site
SECURE ONLINE ORDERING AT ELSEVIERDIRECT.COM
.%7
Reach Your Ideal Candidate! så0OSTåONEåORåMULTIPLEåå JOBåOPENINGSå så0ROMOTEåANåUPCOMINGåå CONFERENCEåORåEVENTå så!NNOUNCEåAåGRANTAWARDå så%LEVATEåYOURåå ORGANIZATIONSåPROlLE
careers.cell.com
http://www.eupa.org/
Editor in Chief: Juan J. Calvete, Valencia, Spain Executive Editors: Proteomics in Cell Biology Jean-Jacques Diaz, Lyon, France
EuPA now has its own journal!
Covered by PubMed
Proteomics in Microbiology Concha Gil, Madrid, Spain Proteomics in Plant Systems Jesus V. Jorrín, Córdoba, Spain Proteomics in Animal Models Dario Neri, Zürich, Switzerland Proteomics in Protein Science Jasna Peter-Katalinic, Münster, Germany Biomedical Applications of Proteomics and Congress Proceedings Jean-Charles Sanchez, Geneva, Switzerland Proteomics of Body Fluids and Proteomic Technologies Pier Giorgio Righetti, Milan, Italy Bioinformatics in Proteomics Peter Højrup, Odense, Denmark
Submitting Authors Manuscripts can be submitted to the Journal of Proteomics at http://ees.elsevier.com/jprot
For a complete listing of the editorial board, visit the journal’s homepage
To receive more information register at:
http://www.elsevier.com/locate/jprot
For the past 40 years FEBS Letters has published award-winning research papers crafted by distinguished experts in the field.
www.febsletters.com
Scopus is the largest abstract and citation database of peer-reviewed literature and quality web sources with smart tools to track, analyze and visualize research.
enrich your experience
www.scopus.com
SnapShot: Chromatin Remodeling: SWI/SNF Margaret M. Kasten, Cedric R. Clapier, and Bradley R. Cairns HHMI, Huntsman Cancer Institute, University of Utah, Salt Lake City, UT 84112, USA Composition and functions of the SWI/SNF family of remodelers Yeast SWI/SNF
Swp82
Snf6
Swi3 Swi3 Pol II activation Elongation DSB repair DNA replication
Swi1/ Adr6
Yeast RSC
Arp7
Swp73 Swi2/Snf2
Rtt102 Arp9
Snf5 Snf11 Taf14
Fly BAP/PBAP BAP111/ dalao
BAF155/ MOR
BAP60 BAF155/ Brahma MOR SNR1/ BAP45
BAP
Osa/ Eyelid
BAP55
Actin Pol II regulation Polybromo Elongation Cell-cycle/proliferation PBAP BAF170 Immune system function Development (Metamorphosis) SUBUNIT LEGEND Catalytic ATPase with bromodomain
Actin-like
Additional bromodomain subunits
Subfamily specific
Pol II and Pol III activation DSB repair Cell signaling Cell-cycle progression, spindle-assembly checkpoint Chromosome/plasmid segregation, cohesion
Rsc8/ Swh3 Rsc8/ Swh3 Rsc9
Rsc30 Rsc3 Rsc14/ Htl1 Ldb7 Arp7
Rsc7
Rsc6
Rtt102
Sth1
Arp9
Rsc4 Rsc10/ Rsc1 Sfh1 Rsc56 Rsc5 or 2
Elongation Human BAF/PBAF DSB repair, nucleotide excision BAF45a, BAF250a,b/ BAF155 repair b,c,d BAF57 BAF53 hOSA1 Signaling BAF60 a,b BAF a,b,c hBRM or Proliferation and BAF170 BRG1 differentiation E-actin Stem cell self-renewal/ hSNF5/ pluripotency BAF47/ BAF180 DNA replication INI1 PBAF Splicing BAF200 BRD7 Tumor Suppressor Development TEXT LEGEND Transcription DNA repair Cell-cycle and Differentiation Others Subunits with similar colors within a complex indicate functional modules, and identical colors between organisms denote related subunits.
Remodeler/nucleosome complex Histone H3 tail
DNA entry/exit points
Model of the RSC-nucleosome complex Sth1 conducts ATP-dependent DNA translocation DNA is drawn from one side of the nucleosome and pumped toward the other Disruption of histone-DNA contacts leads to remodeling outcomes (see below) Predicted translocase binding site
Dyad axis
The different outcomes of SWI/SNF chromatin remodeling SITE EXPOSURE
Repositioning DNA-binding protein
Octamer ejection
Nucleosome
Unwrapping
Remodeler
ATP
ADP
+ Dimer ejection
310
Cell 144, January 21, 2011 ©2011 Elsevier Inc.
DOI 10.1016/j.cell.2011.01.007
See online version for legend and references
¬,OOK¬!GAIN
$ISCOVER¬-ORE s¬!CCESS¬TO¬THE¬¬#ELL¬0RESS¬PRIMARY¬RESEARCH¬JOURNALS¬AND¬ ¬4RENDS¬REVIEWS¬TITLES ¬ALL¬ON¬THE¬SAME¬PLATFORM s¬)MPROVED ¬MORE¬ROBUST¬ARTICLE¬AND¬AUTHOR¬SEARCH s¬6IDEO ¬ANIMATIONS¬AND¬SOUND¬lLES s¬%ASY¬TO¬NAVIGATE¬HOME¬PAGE ¬ARTICLES¬PAGES¬AND¬ARCHIVE
WWWCELLCOM
the difference between discover and do over
Gibco®. Because every little thing matters. Fact: your research will only be as accurate, as efficient, as groundbreaking as the cell culture you’re working with. That’s why you’re settling for nothing less than Gibco® media. Considering it could be the difference between publishing your results and reading someone else’s, why experiment with your experiment?
invitrogen.com/discovergibco download the free mobile app at http://gettag.mobi