Some Articles Planned for FutureVolumes
The RNA World in Plant Mitochondria STEFAN BINDER, MICHAELAHOFFMANN,JOSEPH KUtTN, AND KLAUSDASCHNER
CTD Phosphatose: Role in RNA Polymerase II Cycling and the Regulation of Transcript Elongation MICHAEL E. DAHMUS, NICK MARSIIALL,AND PATRICKLIN
ATP Synthase: The Missing Link STANLEYD. DUNN, D. T. MCLACHLIN,AND M.
j.
REVINGTON
Functional Analysis of MUC1, a Carcinoma-Associated Mucin SANDRAJ. GENI)LER HIV-1 Nucleoprotein: Retroviral/Retrotransposon Nucleoproteins JEAN-LUG DABLIX
Biochemistry of Methiogenesis: Pathways, Genes, and Evolutionary Aspects UWE DEPPENMEIER
Manipulation of Aminoacylation Properties of tRNAs by Structure-Based and Combinational in Vitro Approaches t/ICHARI) GIEGE AND JOEM PUETZ
Shunting and Reinitiation: Viral Strategies to Control Initiation of Translation THOMAS HOHN
Functions of Alphavirus Nonstructural Proteins in RNA Replication LEEV] KAARIAINENAND TERO AHOLA
DNA-Protein Interactions Involved in the Initiation and Termination of Plasmid Rolling Circle Replication SALEEM A. KAHN, T.-L. CttANG, M.G. KRAMER,AND M. ESPINOSA
Specificity and Diversity in DNA Recognition by E. coil Cyclic AMP Receptor Protein JAMES C. LEE
Molecular Mechanisms of Error-Prone DNA Repair ZVI LIVNEH
Catalytic Properties of the Translation Factors Necessary for mRNA Activation and Binding to 40S Subunits WILLIAM C. MERRICK
x
SOME ARTICLES PLANNED FOR FUTURE VOLUMES
Initiation of Eukaryotic DNA Replication and Mechanisms HEINZ-PETER NASHEUER
FGF3: A Gene with a Finely Tuned Spatiotemporal Pattern of Expression during Development CHRISTIAN LAVIALLE
Molecular Basis of Fidelity of DNA Synthesis and Nucleotide Specificity of Retroviral Reverse Transcriptase LUIS MENENDEZ-ARIAS
Initiation of Eukaryotic DNA Replication and Mechanisms HEINZ-PETER NASHEUER, KLAUS WEISSHART, AND FRANK GROSSE
A Growing Family of Guanine Nucleotide Exchange Factors Is Responsible for the Activation of Ras-FamilyGTPases LAWRENCE A. QUILLIAM
Translational Factors That Affect 5'-3' mRNA Interaction NAttUM SONENBERG AND FRANCIS POULIN
HIV Transcriptional Regulation in the Context of Chromatin ERIc VERDIN
The Molecular Biologyof the Group VIA Ca2+-Independent PhospholipaseA2 Z H O N G M I N MA 1 AND JOHN TURK
Division of Endocrinolof~y, Diabetes, and Metabolism Department of Medicine Washington University School of Medicine St. Louis, Missouri 63110 I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Classification and Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii. Sequence and Structural Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Lipase Consensus Motif GXSXG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. ATP-Binding Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Ankyrin-Repeat Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Bipartite Nuclei Localization Signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Caspase-3 Cleavage Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . F. Proline-Rieh Region of Hnman Long Group VIA PLA2 lsoform . . . . . . G. Other Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Gene Structure, Alternative Splicing, and Chromosomal Localization . . . . . V. Tissue Distribution and Expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. Enzymology of Group VIA PLA2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Phospholipase A2 and Phospholipase AI Aeti~dties of Group VIA PLA2. B. Selectivity of Group VIA PLA2 for Phospholipids . . . . . . . . . . . . . . . . . . C. Lysophospholipase, PAF Aeetylhydrolase, and Transaeylase Activities of Group VIA PLA~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Potential Celhdar Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Signaling Function in Insulin-Secreting Cells . . . . . . . . . . . . . . . . . . . . . . B. Apoptosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2. Membrane Phospholipid Remodeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Membrane Homeostasis and Other Functions . . . . . . . . . . . . . . . . . . . . . VIII. Future Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 3 4 5 S 10 14 15 15 16 17 20 20 20 20 21 22 22 24 25 "26 28 29
The group VIA PLA2 is a member of the PLAg superfamily. This enzyme, which is cytosolie and Cag+-independent, has been designated iPLA2fl to distinguish it from another recently eloned Ca2+-independent PLA2. Features of iPLA2/3 moleeular strueture offer some insight into possible eellular funetions of the enzyme. At least two catalytically active iPLAzfi/isoforms and additional 1To whom eorrespondenee should he addressed.
Progress in Nucleic Acid Research and Molecular Biology,Vo]. 67
l
Copyright O 2001 by Academic Press. All rights of reproduction m any fonn reserved. 0079-6603/01 $35,00
2
ZHONGMIN MA AND JOHN TURK
splicing variants are derived from a single gene that consists of at least 17 exons located on human chromosome 22q13.1. Potential tumor suppressor genes also reside at or near this locus. Structural analyses reveal that iPLA2~ contains unique structural features that include a serine lipase consensus motif (GXSXG), a putative ATP-binding domain, an ankyrin-repeat domain, a caspase-3 cleavage motif DVTD138y/N, a bipartite nuclear localization signal sequence, and a proline-rich region in the human long isoform, iPLA2~ is widely expressed among mammalian tissues, with highest expression in testis and brain, iPLA2~ prefers to hydrolyze fatty acid at the sn-2 fatty acid substituent but also exhibits phospholipase AI, lysophospholipase, PAF acetylhydrolase, and transacylase activities, iPLA~/3 may participate in signaling, apoptosis, membrane phospholipid remodeling, membrane homeostasis, arachidonatc release, and exocytotic membrane fusion. Structural features and the existence of multiple splicing variants ofiPLAg/~ suggest that iPLAg/3 may be subject to complex regulatory mechanisms that differ among cell types. Further study of its regulation and interaction with other proteins may yield insight into how its structural features are related to its function.
© 2001 Academic Press.
I. Introduction In response to cellular stimulation, membrane phospholipids are often hydrolyzed to generate intraeellular and intercellular messengers. Phospholipase A-2(PLA2) enzymes catalyze hydrolysis ofsn-2 fatty acid substituents from glycerophospholipid substrates to yield a free fatty acid and a 2-1ysophospholipid (1). This group of enzymes has been intensively studied because they play crucial roles in diverse cellular responses, including phospholipid digestion and metabolism, host defense and signal transduction, and production of proinflammatory mediators, such as prostaglandins and leukotrienes, through the release of arachidonie acid (AA) from membrane phospholipids (2, 3). The lysophospholipid generated in PLA2 hydrolysis serves as a precursor for the proinflammatory molecule platelet-activating factor (PAF), and lysophosphatidie acid is a potent mitogen (4). PLA2 enzymes are a rapidly growing superfamily of diverse enzymes that have been classified into at least 11 groups (5). Recent advances in DNA and protein databases that permit BLAST analyses and EST searches have permitted cloning of new PLA2 species. This chapter summarizes the molecular biology of a recently cloned intracellular Ca2+-independent PLA2 that has been classified as group VIA PLA2 (5) and is designated iPLA2fl here to distinguish it from another recently cloned Ca2+-independent PLA2 (6). iPLA2/~ was first purified from the nmrine P338D1 macrophage-like cells as an 80-kDa protein on sodium doeeeyl sulfate-polyaerylamide gel electrophoresis (SDS-PAGE) (7). The enzyme was subsequently isolated from chinese hamster ovary (CHO) cells (8), which led to the cloning of its cDNAs from several sources (8-12). Analyses
GROUPVIACa2+-INDEPENDENTPHOSPHOLIPASEA2
3
of its primary sequence have revealed structural characteristics that may provide clues about the roles of the enzyme in cellular processes. Determination of the structure of the human iPLA2~ gene has yielded insight into the geneses of multiple iPLA,2~ splice variants (11, 12), and the gene has been found to reside in a chromosomal location that contains loci for genes associated with human diseases.
II. Classification and Nomenclature Based on their dependence on Ca 2+ for their enzymatic activity, PLA2 enzymes can be dMded into Ca'2+-dependent and Ca2+-independent classes. The former includes several groups of secretory PLA2s (sPLA2), which require millimolar Ca '2+ concentrations for catalytic activity, and group IV Ca2+-dependent cytosolie PLA,2isoenzymes (cPLA2~ and -fl), which require submieromolar Ca 2+ concentrations to associate with membrane substrates. The Ca'2+-independent PLA2s appear to represent a diverse group of enzymes that can be further subdivided into several categories: group VIA intraeellular Ca2+-independent PLA,2 (iPLA2fl) (8-12), membrane-associated Ca%-independent PLA.)(iPLA2F) (6), 61-kDa group IV cytosolic PLA2F (ePLA2F) (13, 14), and PAF ace@hydrolases (1.5, 16). A common feature of these Ca2+-independent PLA.2s is the presence of the lipase consensus motif GXSXG. These enzymes exhibit no other similarities. iPLAafi was initially identified and purified from murine P388DI maerophage-like cells (7, 17) and classified as group VI PLA2 (1, 18) and subsequently as group VIA PLA,2 (5). In the remainder of this chapter, the group VIA PLA,~will be designated iPLA.2fl for simplicity, unless otherwise indicated. The eDNA encoding this enzyme was first cloned from CHO cells (8) and subsequently from other sources (9-12). iPLA2fi has two recognized enzymatically active isoforms (12) and exhibits lysophospholipase activity in addition to PLA2 activity (7, 8, 19). Sequence analyses reveal that iPLA.~fl contains several interesting structural features that may be related to its functions ill cells. Analyses of a predicted 40-kDa protein identified by the human genome project and a TBLASTN database search of GenBank led to the cloning of a novel Ca2+-independent, membrane-associated PLA.2that has been designated iPLA.~F (6). The deduced amino acid sequence fi'om this transcript showed no homology to known Ca2+-independent PLA9 enzymes except the putative ATPbinding and GXSXG lipase consensus motifs that also occur in iPLA2fl. Both of these motifs also exist in a 40-kDa enzyme from potato with Ca'e+-independent phospholipase A2 activity (20, 21) that has been designated iPLA,)oe (6). The classification sehelne of Six and Dennis designates iPLA,~fl as group VIA and iPLA_gF as group VIB PLA2, respectively (5). iPLA2F also contains a C-terminal
4
ZHONGMIN MA AND JOHN TURK
peroxisomal targeting sequence (SKL) (22, 23). Because iPLA2F is tightly bound to membrane fractions in cell homogenates, it may be that the major subcellular location of iPLA2F is in the peroxisomal matrix enclosed within the peroxisomal membrane (6). By combined BLAST and EST database searches and 5'-RACE methods, a largely membrane-bound PLA2 with a calculated molecular mass of 60.9 kDa homologous to ePLA20t (group IVA PLA2) was cloned (13, 14). This protein, which exhibits Ca'2+-independent PLA2 activity, has been designated cPLA2F (13, 14). According to the scheme of Six and Dennis, this enzymes is classified as group VIC PLA2 (5). The deduced amino acid sequence indicates that the cPLA.2F protein lacks the C2 domain of cPLA2a, and accordingly has no dependence upon Ca ~+ for membrane association or catalytic activity. This enzyme is thus a Ca'2+-independent PLA2. cPLA2F protein contains a pren~lation motif(-CCLA) (24) at the C terminus (13). The isoprenoid precursor [ H]mevalonolactone is incorporated into the prenylation motif ofcPLA2F when expressed in COS cells, and the mutagenesis of CCLA to SSLA at the C terminus of cPLA_gF prevents the ['3H]mevalonolactone incorporation, suggesting that the consensus prenylation site is indeed utilized. This may account for the membrane localization of cPLA2F (13). Platelet-activating factor (PAF) aeetylhydrolases are also Ca2+-independent PLA2 (16). PAF acetylhydrolases are structurally diverse isoenzymes that catalyze hydrolysis of the sn-2 aeyl group of choline glycerolipids containing an sn-1 alkyl ether linkage and a short-chain or oxidized sn-2 substituent (16). The classification scheme of Six and Dennis (5) places these enzymes into two groups. The group VII enzymes have molecular masses of 40-45 kDa and include both secreted isozymes found in plasma (group VIIA) (25) and intracellular, myristoylated enzymes found in lung and kidney (group VIIB) (26). The group VIII enzymes are intraeeIlular, have molecular mass of 29-30 kDa, and are found in brain (16, 27).
III. Sequence and Structural Characteristics The iPLA2fl cDNAs have been cloned from several sources (8-12). Rodent iPLA2fi and the human short isoform of iPLA2fi eDNA species encode a single 752-amino acid protein with calculated molecular mass of about 85 kDa. The long isoform of human iPLA2fi eDNA encodes an 807-amino acid protein which has a 55-amino acid residue insertion at position 395 (Fig. 1) (11, 12). The iPLA2fl enzymes share no sequence similarity with other known PLA2 enzymes. Among the consensus structural features of sPLA2 enzymes are a Ca2+-binding -loop with the typical glyclne-neh sequence Y ' -G-C-X-C-G-X-G-G-X-X-X-P (the number of amino acid residues is based on Type I PLA2) and the residue Asp49, and an active site His48 (28). Asp49 is located adjacent to the catalytic •
•
•
25
37
GROUPVIACa2+-INDEPENDENTPHOSPHOLIPASEA2 Caspase-3 cleavage site
DVTDlS3y
R-iPLA2 I~ SH-iPLA2 I~
,5
Lipase consensus motif
GTS~'rG
LH-iPLA2 ~ [
•
DVTDIg3y
GTS619TG
Eight Ankyrin-repeats domain
[ ] Bipartite nuclear localization signal
ATPbindingdomain
[] Proline-richregion
FIG. 1. Schematicrepresentationof the structure of iPLA:lfi.The upper bar represents the rodent or human short isoformofiPLA2fl,and the lowerbar represents the humanlongisofnrmof iPLA2/~.The positionof the eight anlg,rin repeats, the putativeATP-bindingdomain,the bipartite nuclear localizationsignal,the proline-richregionof HL-iPLA2/~,the caspase-3cleavagesite, and the lipaseconsensusmotifare shown.
His 4s, forming the so-called His/Asp dyad. Mechanistic studies indicate that the sPLA.2 do not form a classic aeyl enzyme intermediate that is characteristic of serine esterases. Instead, they utilize the catalytic site His, assisted by Asp, to polarize a bound water molecule that then attacks the substrate earbonyl group. The Ca '2+ ion, bound in the conserved Ca2+-binding loop, stabilizes the transition state. Serine esterases such as iPLA,2 employ a mechanism for catalysis that is different from that of sPLA2 enzymes. The group IVA cytosolic Ca'2+-dependent PLA.2 (ePLA2ot) has a Ca2+-dependent lipid-binding (CaLB) domain at its N terminus that is responsible for transloeation of cPLA2 from cytosol to membranes in response to rises in cytosolic [Ca2+] induced by extracellular signals (29, 30). The CaLB domain of ePLA2~ exhibits significant homology with the C2 domains in proteins such as protein kinase C, GTPaseactivating protein (GAP), synaptotagmin, and phospholipase C. Such domains bind to phospholipid membranes in a Ca'2+-dependent manner (30, 31). It is not yet certain what t
A. LipaseConsensusMotif GXSXG The amino acid sequence of iPLA2f contains a lipase consensus motif GXS465XG (Ser 619 in long isoform of iPLA2/3) (Fig. 1) that is commonly found in neutral lipases and is essential for enzymatic activity (32, 33). Indeed, an iPLA2/3 mutant protein expressed in CHO cells with an iPLA2/3 eDNA
Z H O N G M I N MA AND JOHN TURK
containing a mutation a t Ser 465 to Ala completely abolished the catalytic activity of iPLA2/~, supporting the participation of Ser 465 in catalysis (8). In the case of cPLA2ot, mutation of Sel2es to Ala at GLS22SGS, which is similar to but slightly different from the classical serine lipase consensus, also completely eliminates both PLA2 and lysophospholipase activities of ePLA2~ without significantly altering its structure (34). Like cPLA2oe, iPLA.2/3 also can be inhibited by arachidonyl trifluoromethyl ketone (AATFMK, also referred to as AACOCF3) and methyl arachidonyl fluorophosphonate (MAFP) (17, 18), perhaps because both the cPLA2a and iPLA2/~ appear to use a central Ser in similar catalytic mechanisms. The activities of iPLA2/3 and cPLA2a are differentially affected by a bromoenol lactone (BEL) suicide substrate at a concentration as low as 1 #M. At such low concentrations, BEL inhibits iPLA2/~ activity but not sPLA2s or cPLA2a activities (35). BEL has been found to bind covalently to iPLA2/3 (17), although the site of attachment has not been determined. This covalent modification inactivates the enzyme irreversibly. Identification of the BEL binding site could be useful in relating iPLA2]3 structural features to its catalytic activities. A partial lipase consensus GPS'252GF also occurs in hamster iPLA2/~ (8). A similar sequence in cPLA2ot contains the active site serine, but site-directed mutagenesis Ser 2'52 to Ala in iPLA2/~ does not alter catalytic activity when the enzyme is expressed in CHO cells (8). The Ser 25'2 in hamster iPLA.2/~ is also not conserved in iPLA2]3 cloned from other species. cPLA2F (group IVC) contains a GvsS'2GS motif that also occurs in cPLA2~ (13, 14). Mutagenesis of Ser "22s,Arga°°, Asp549, or Arg566 also abolishes cPLA,)c~ activity (14, 36). These amino acid residues are conserved in cPLA2F as Arg54, Asp asS, and Arg4°'2 (13, 14). Mutagenesis of these residues to Ala greatly reduces cPLA2F activity, indicating that Arg54, Asp3s5, and Arg4°2 are required for catalysis in addition to the central serine in the GVSS2GSsequence (14). Analysis of the deduced amino acid sequence of iPLA2F (group VIB PLA2) revealed that this novel enzyme, like iPLAeJ3, also contains a Gxs4SaXG motif (6). This conserved sequence motif occurs in two translational variants with apparent molecular masses of 77 and 63 kDa. These proteins may arise from use of alternative translational initiation sites at Met 101 and Met 221 ' , respectively. It is likely that the GXSXG consensus motif contains the active serine required for catalysis. There is also a partial lipase consensus sequence (GDS2°3FY) in the deduced amino acid sequence of iPLA2F (6). This motif is probably included in the 77-kDa protein but not in the 63-kDa protein based on the proposed translation initiation sites. Further studies by site-directed mutagenesis will be required to determine whether the Gxs4S3XG or GDS2°3Fy sequences contain an active site Ser that is required for catalysis by iPLA2F. Group VII enzymes include a secreted plasma form (25) and an intracellular isoform II (26). Group VIII includes the/3 and F subunits of intracellular
GROUP VIA Ca2+-INDEPENDENT PHOSPHOLIPASE A,~
7'
isoform Ib (16, 37, 38). The cloning of a eDNA encoding the plasma form of PAF aeetylhydrolase was reported in 1995 (25), and the deduced amino acid sequence contained a GHS 273FG motif (Table I). Using site-directed mutagenesis, Set 273 was found to be essential for catalysis. Both Asp 296 and His 351 are also essential for catalytic activity (25). The orientation and spacing of these catalytic residues are consistent with an ot/fl hydrolase conformation that has been observed in other lipases (25). The deduced amino acid sequence ofisoform II of PAF acetylhydrolases (group VIIB) exhibits 41% identity with that of plasma PAF acetylhydrolase (group VIIA) and also has a GHSFG consensus motif that occurs in plasma PAF aeetylhydrolase (group VIIA). The group VIIB enzyme is inactivated by diisopropyl fluorophosphate consistent with involvement of a Ser residue in catalysis, and the substrate specificity of the group VIIB enzyme is similar to that of plasma PAF aeetylhydrolase (39). The I3 and y subunits of isoform Ib (group VIII) exhibit 63.2% identity in deduced amino acid sequence overall. There is 86% identity in the catalytic and PAF-receptor-homologous domains. There is no overall homology with other PAF aeetylhydrolases (38). This group of enzymes contains a GXSXV motif instead of GXSXG (Table I). Mutagenesis of Ser 48 of the GXS 48XV consensus motif of the y-subunit of the group VIII enzyme abolishes enzymatic activi~, indicating that Ser 4s is involved in catalysis (38). The consensus sequence of GXSX(G/V) is a common feature of all Ca2+-independent PLA.2 so far cloned, and the observations summarized above
TABLE I o+ CtIARACTERISTICS OF THE CLONED Ca- -INDEPENDENT PHOSPI1OLIPASEA2 Name
Location
Source
Size (kDa)
Active site
References
iPLA2fl (group VIA PLA2)
Intraeelhdar
85 and 88
GTSTG
8-12
Patatins and Patatin-like (iPLA2a) iPLA2F (group VIB PLA,2)
Intraeelhdar
Islet, CHO, P338D, B lymphoe~es Potato tubes
40
GTSTG
(5, 20, 21
Membrane -associated Iutracellular Secreted
Human heart and skeletal nmscle Brain EST Plasma
88.5, 77, and 63 61 45
GVSTG
6
GVSGS GttSFG
13, 14 2,5
Liver and kidney
40
GHSFG
26
Brain
30
GDSLV
16, 27
Brain
29
GDSMV
16, 27
ePLA2y (group IVC PLA,2) PAF aee~lhydrolase (group VIIA) PAF aeetylhydrolase Intraeellular Isoforms II (group VIIB) PAF acetylhydrolase Ib, Intracellular /J subunit (group VIIIA) PAF acetylhydrolase Ib, Intraeellular F subunit (group VIIIB)
8
ZHONGMINMAANDJOHNTURK
indicate that the central serine in this motif participates in catalysis in these enzymes (Table I).
B. ATP-Binding Domain When the cloned rat islet iPLA2fi eDNA is transiently expressed in COS-7 or CHO cells, its actMty is stimulated about 2-4-fold by adenosine triphosphate (ATP) (9). This is similar to the effect of ATP on the iPLA2fl activity isolated from P388D1 macrophage-like cells (7). Furthermore, iPLA2fl overexpressed in S. frugiperda (Sfg) cells adsorbs to both ATP-agarose and ealmodulin-agarose matrices, and this can be exploited in affinity chromatographic purification of the enzyme (40, 41). Recently, Yang et al. (42) reported the purification of an 80-kDa Ca2+-independent PLA2 from rat brain using the same strategy. Amino acid sequencing of several peptides from tryptie digests of the purified rat brain iPLA2 protein indicated that it corresponds to the iPLA2fl cloned from rat panereatie islets by Ma et al. (9). These observations suggest that an ATP-binding motif exists within the iPLA2fi molecule. Surveys of the protein kinase family reveal that their catalytic domains exhibit a highly conserved G-X-G-X-X-G sequence motif that appears to partieipate in ATP binding (43). The consensus motif the G-X-G-X-X-G is also found in many nueleotide binding proteins in addition to the protein kinases (43, 44). Alignment of the iPLA2fl sequence with the ATP-binding domain of protein kinases revealed that the sequence from amino acid residue 431 to 457 of iPLA2fl exhibits similarity to the region of protein kinases involved in ATP binding (Fig. 2). A model for the ATP-binding site of the protein kinase v-src, based on the three-dimensional structures from other nucleotide-binding proteins, shows that the G-X-G-X-X-G residues form an elbow around ATR with the first glyeine in contact with the ribose moiety and the second glyeine lying near the terminal pyrophosphate. Mutagenesis analyses of the ATP-binding site of cAMP-dependent protein kinase indicated that replacing the third Gly5'5 had minimal effects on steady-state kinetic parameters, whereas replacement of either Gly5° or Gly'52 had major effects on both K,,, and kc~t values consistent with the predicted importance of the tip of the glyeine-rieh loop for catalysis (45). A nearly invariant Val residue lies within subdomain I of v-src (Fig. g) and is located just two positions on the carboxyl-terminal side of the G-X-G-X-X-G consensus. This residue may contribute to the positioning of conserved glyeines (43). In addition, several amino acid residues in subdomain II of v-src, such as alanine'29s, lysine 3°°, and leucine3°2, are represented at analogous positions in the sequence ofiPLA2fi Lys3°°in v-src appears to be directly involved in the phospho-transfer reaction (46). Since iPLA2CJadsorbs to ATP-agarose and is desorbed by ATR it is possible that ATP directly binds to the ATP-binding domain of iPLA.2JL Sueh binding may modulate the function of iPLA~, although
GROUP VIA Ca2+-INDEPENDENTPHOSPHOLIPASEA2 Serine/threonineprotein kinase family cAPK-0~ 40 cAPK-~ SRA3 TPKI PKC-~ PKC-~ PKC-y CDC28
"r
9 "r'r
41 92 84 336 339 332 5
Tyrosine protein kinase 1 Src EGFR INS.R PDGFR
267 685 1020 565
iPLA2 ~ 420 Consensus
FIr'. 2. Alignment of the sequence of iPLA2fl with the ATP-bindingdomain of some of serine/threonine protein kinases and of tyrosine protein kinases. The sequence of iPLA2fl and the consensusof ATP-bindingdolnain are shown at the bottom of the figure. Hyphensrepresent gaps introducedfbr optim~dalignment. The kinase subdolnains are indicated by roman nmnerals at the top of the alignment Residues conservedin all or nearlydl protein kinases are enclosedin black boxes.
this possibility has not been verified experimentally. Interestingly, analyses of hmnan iPLA2fl gene strneture revealed that the ATP-binding domain and lipase eonsensus lnotif are encoded by the same exon (12), and this structural location suggests that these two domains eould interact to regulate the iPLA2fi fimetion. ATP could affect iPLA2fl function by at least three mechanisms. First, ATP could bind to the ATP-binding domain of iPLA2fi to stimulate its intrinsie enzymatic aetivity. Second, ATP could induce association of iPLA2fl monomers to form muhimerie aggregates. Third, ATP could stimulate transloeation ofiPLA2fl from eytosol to membranes, where its substrates are located. In many eases, iPLA2fi activity has been found to be stimulated by ATE as well as other di- and triphosphate nueleosides (• 9, 12). In some eases (8, i0, 19), ATP has been observed not to stimulate activity of purified or partially purified iPLA2fl. Interestingly, activity of the recombinant short isoforIn of human iPLA,2fi expressed in Sf9 cells is not affected by ATR but activity of the long isoform of human iPLA2fi increases about 3-fold in the presence of ATP (12). A study by Kio and Dennis (19) suggested that ATE stabilizes iPLA2fl and protects it from denaturation. In cells, iPLA2fi might interact with other proteins that negatively regulate its activity. ATP may induce dissociation of iPLA2fi from such negative modulators and result in the activation of iPLA2fi. Nevertheless, the faet that iPLA2fi can bind to an ATP-affinity column suggests that ATP might interact with iPLA.~fi in vivo; this possibility deserves further study, for example, by deletion and mutagenesis analyses.
10
ZHONGMIN MAAND JOHN TURK
iPLA2F, like iPLA2fl, contains an ATP-binding motif at amino acid sequence 449GGGTRGW457. All glycines in this motif are conserved between iPLA2/3 and iPLA2F, as is the Val residue (Ile in human iPLA.2fl) two positions after Gly4'55. In contrast to iPLA2fi, both the ATP-binding motif and the lipase consensus sequence are encoded by two adjacent exons in the iPLA2F gene (6).
C. Ankyrin-RepeatDomain Examination of the iPLA2fi deduced amino acid sequence by dot matrix analysis revealed that the domain between amino acid residues 150 and 414 (Fig. 1) is composed of eight strings of a repetitive sequence motif of approximately 33 amino acid residues (Fig. 3) (9). This repeating-sequence motif is highly homologous to that of an 89-kDa domain ofankyrin (8, 9). In 1987, Breeden and Nasmyth (47) reported an "~33-residue repeating motif in the sequence of two yeast cellcycle regulators, SWi6 and cdcl0, and in the Notch and LIN-12 developmental regulators from Drosophila melanogaster and Caenorhabditis elegans. Subsequently, the discovery of 24 copies of this sequence in the cytoskeletal protein ankyrin led to the designation of this motif as the ankyrin repeat (48). Ankyrinrepeat-containing proteins perform a wide variety of biological functions and have been detected in organisms ranging from viruses to humans. The motif has now been recognized in >400 proteins, and the number of repeats within any one protein is highly variable (49). Molecules that contain such repeats include ankyrin proteins that link integral membrane proteins to cytoskeletal elements
DNK ~ 3
NN Q
4
-
5
PRY
8
ND
L
GP
Q ~ D N]p~-IIr, Q QMG
K~Q_~M~RV
KFSQKGCIAIEM
---~N~aM
F
E
• . .G.TPLH.
K I S K[Q]- -ILIQ D A ......
E.V.
QY-CH Q DVT GKNAS G NQV -L-CN R NIM SMDSNQIHSKD KRG C - - DHD S T TYGANIG R-- V-FG~E~D T P PV S R A R K~A F I L L . . . . .
A . . . . .
Consents
i I FIG.3. Alignmentof the eightstringsof ankyrin-repeatsequencesofiPLA2fl.Identicalamino acid residues are enclosedin black boxes and conservativechanges are enclosedin open boxes. The ankyrin-repeatconsensusand definedstructure are shownat the bottom of the figure.Arrows represent the fl hairpins,and cylindersrepresentthe u helices.Ankyrinrepeatwas definedas a fl2c~2 motif. [ReprintedwithpermissionfromZ. Ma, S. Ramanadham,K. Kempe,X. S. Chi,J. Ladenson, andJ. Ture,J. Biol. Chem. 272, 11118-11127 (1997).]
11
GROUP VIA Ca2+-INDEPENDENT PHOSPHOLIPASE A2
(48, 50), developmental regulators in Drosophila and C. elegans, cell-cycle control proteins in yeast, transcriptional factors, toxins, and viral proteins (Table II) (49-5l). Recently, a novel family of postsynaptic-density proteins, Shank, has been reported to contain seven ankyrin repeats at the N terminus (52). These molecules are present in the nucleus, cytoplasm, and membranes, as well as the extracellular milieu (49). The role of ankyrin repeats in mediating protein-protein interactions is well documented, and their presence is often interpreted as an indicator of a sin> ilar filnction in otherwise uncharacterized systems. The presence of ankyrin repeats in iPLA2fl suggests that intra- or intermolecular protein-protein interactions may regulate its function. Ankyrin is a linker molecule between membrane and cytoskeletal proteins (50). Its C-terminal domain binds to cytoskeletal proteins such as spectrin and tubulin, while its N-terminal 89-kDa ankyrin-repeat domain binds to integral membrane proteins, such as ion channels and cell adhesion/signaling molecules (Table III) (50). The ion channels that ankyrinrepeat domains bind include the Na+,K+-ATPase of renal basolateral men> branes, a renal amiloride-sensitive Na + channel, the red cell anion exchanger, a cerebellar inositol triphosphate receptor, and voltage-dependent Na + channel in myelinated neurons (50). The binding of ankyrin repeats to integral membrane proteins raises two possible roles for ankyrin repeats in the iPLA2 protein. TABLE II PROTEINS WITII ANKYRIN REPEATS
Groups
Examples
Ankyrin proteins
Ankyrins
Developmental proteins
une-44 (C. elegans) in-12, glp-1, ibm-1 (C. elegans)
Viral host range proteins
Notch (D. melanogaster) TAN-1 (human, Notch-like) 53BP'2, c' lx Bs," NF-x B preeursor( 110 kDa), Lyt-10, CDK inhibitors," GABPo~//~,a ode10, SWI4, SWI6" (yeast) Shank proteins Latrotoxin/latroinsectotoxin (black widow spider) Vaeeina 32-1d (vaeeinia vires)
Other proteins
Cowpox HRP (cowpox virus) Fowlpox 47-kd (fbwlpox virus) Group VI PLA2 (iPLA,2fl)
Transcriptional factors and cell-cycle regulato D' proteins Synaptie proteins Toxins
Myotrophin a PYK-2" (Nonreceptor tyrosine kinase) 2-5A RNAse
Number of repeats 24 24
6 6 6 2-8
7 19 3
8 4 4 9
"Tile three-dimensionalstructuresof the an~,rin repeat have been determinedby X-ray and/or NMR.
12
ZHONGMIN MA AND JOHN TURK TABLE III INTEGRALMEMBRANEPROTEINSTHAT BIND TO ANKYRINREPEAT Integral membrane proteins Ion channels Bed cell anion exchanger (AE1) AE1 (kidney isofbrm) IP3 receptor (270 kDa) Voltage-sensitive Na + channel (homologous to c¢ subunits of voltagedependent Ca2+ channel) Na+,K+-ATPase (~ subunit) Amiloride-sensitive Na + channel Cell adhesion/signaling molecules Lymphocyte adhesion antigen CD44 (putative hyaluronic acid receptor) GP116 (CD44-1ike endothelial protein) Ankyrin-binding glycoprotein 205 (AB-GP205)
Ankyrins
Erythroid ankyrin (Ankl) Epithelial anlgwin(s) Brain ankyrin (Ank2) Brain ankyrin (Ank2)
Epithelial ankyrin(s) Epithelial ankyrin(s) Lymphocyte ankyrin Erythroid ankyrin (Ankl) Brain ankyrin (Ank2)
(1) Although cPLA2 has a CaLB domain that shares homology with the C2 domain in the conventional isoforms of PKC, phospholipase CF, synaptotamin, and so forth (29, 30), iPLA,213has no similar sequence that might mediate association with membrane phospholipids. The translocation iPLA213 from cytosol to membrane is likely to be important for its function in cells because its phospholipid substrates are in membranes. It is possible that iPLA213 is able to associate with membrane phospholipids through the binding of its ankyrin-repeat domain to integral membrane proteins. Recently, we demonstrated that iPLA213 can be induced to associate with a membrane fraction of INS-1 insulinoma cells upon cell stimulation (manuscript in preparation). The long isoform of human iPLA.213 was also found to be associated with membrane when overexpressed in COS-7 cells (53). Interestingly, Western blot analysis of iPLA213 in adult rat ventrieular myocytes revealed that full-length iPLA213 is detected only in the membrane fraction (54). (2) Regulation of ionic fluxes is critical to the function of pancreatic islet 13 cells, neurons, and muscle cells, and proteins containing ankyrin repeats associate with a number of ion transporters (Table III). Arachidonic acid (AA) and other polyunsaturated fatty acids affect many ion channels (55). The concentrations of free AA and other polyunsaturated acids within cells are very low. It is possible that, upon activation, iPLA213 might transloeate to associate with membrane ion channels via its ankyrin-repeat domain. Hydrolysis of AA and other polyunsaturated acids from phospholipids catalyzed by iPLA213 could then yield high regional concentrations of polyunsaturated acids that could affect ion channel functions. It was reported that Na + flux induced by angiotension II
GROUPVIACa2+-INDEPENDENTPHOSPHOLIPASEA2
13
required the activation of a BEL-sensitive iPLA2 in LLC-PK1 cells (56), consistent with the possibility that iPLAefl might interact with ion channels. The first three-dimensional structure of an ankyrin-repeat-containing molecule was determined by X-ray crystal structural analysis of 53BP2 bound to the p53 cell-cycle tumor suppressor (57). Subsequently, the ankyrin-repeat structures of several proteins, including cyclin-dependent kinase (CDK) inhibitors, GABPa/fl, myotrophin, IKBa-NF4cB, Swi6, and PYK2, were determined by X-ray and/or NMR methods (Table II) (58, 59). The ankyrin repeat consists of pairs of antiparallel a helices stacked side-by-side and connected by a series of intervening fl-hairpin motifs. The extended fl sheet projects away from the helical pairs almost at right angles to them, resulting in a characteristic L-shaped cross section. This assembled structure has been likened to a cupped hand: the fi hairpins form the fingers, and the concave, inner surface of the anlcyrin groove, which is made up of solvent-exposed residues from the a-helical bundle, Ibrms the palm (,58, 59). The crystal structure analyses revealed that ankyrin repeats play a critical role in forming fimctional complexes. It has been reported that iPLA2 exists as a multimerie complex of 270-350 kDa (8). Deletion of the N-terminal 150-amino acid residues plus the eight ankyrin repeats of iPLA,2fl results in loss of catalytic actMty (8). This could indicate that the ankyrin-repeat domain is important for formation of a multimeric complex of iPLA2fl and that this is the catalytic active ibrm. A recent report by Larsson et al. (11) demonstrated that cells cotransfeeted with full-length iPLA2fl and with a deletion mutant that contained the ankyrin-repeat domain but not the catalytic domain (ankyrin-iPLA2-1) exhibited decreased activity compared to cells transfected with full-length iPLA2fl alone. This suggests that multimerie complexes of iPLA,2/? represent the functional forms. The fact that there is residual iPLA2fi activity in the cotransfectants could reflect the existence of a subpopulation of homomultimeric complexes. Alternatively, heteromultimeric complexes might retain some activity. In either case, these results suggest that the ankyrin-repeat domain participates in formation of fimctional of iPLA2/? complexes. Inflamed tissue expresses a host of proteins that are not normally expressed. Many of the genes encoding such proteins are activated by NF-x B, a transcriptional factor that is normally in an inactive form in the cytoplasm. NFKB can be activated by a variety of proinflammatory and noxious stimuli (60, 61). Under resting conditions, NF-gB is tightly associated with IKBs, a class of specific inhibitory proteins that prevent nuclear transloeation and DNA binding of NF-K B. The structural hallmark of the various IicB proteins is an ankyrin-repeat domain containing six or seven closely adjacent repeats (6,2). Crystal structure analyses of the bc B/p65/p50 complex indicate a fundamental role of ankyrin repeats in the formation of inactive IKB-NF4cB complexes. The ankyrin-repeat domain
14
ZHONGMINMAANDJOHNTURK
of IKBa forms a slightly bent cylinder with five loops protruding from the packed arrangement of stacked a helices. The loops between the repeats contain residues that specifically recognize NF-KB. The appearance of IxB in the NF-xB complex is reminiscent of a backbone lying between two lungs, where each ankyrin repeat is a vertebra (62-64). Interestingly, ankyrin repeats 1 and 2 interact with sequences encompassing the nuclear localization signal (NLS) of p65 (62). Repeats 3 to 5 bind tightly over a large surface to the C-terminal Ig-fike domains of both p50 and p65 Rel homology domains (RHDs) (62). Structural analyses ofiPLA2fi reveal that the enzyme contains a bipartite nuclear localization signal sequence (NLS), as discussed below. This raises the possibility that the ankyrinrepeat domain might bind the NLS of iPLA.2/~ intramolecularly to regulate the translocation of iPLA2/3 from cytoplasm to nucleus. This would be analogous to the binding of the ankyrin repeats of IKBot to the NLS of NF-KB in IxB-NF-KB complexes.
D. Bipartite Nuclei Localization Signal Using the ExPaSy (Expert Protein Analysis System) profileScan to scan the iPLA2/3 amino acid sequence against protein profile databases (including PROSITE), only two domains in iPLA2J3 yielded significant matches with consensus domains in the databases. These are the ankyrin-repeat domain and a bipartite nuclear localization signal sequence (Fig. 1). The two best defined NLSs are that of SV40 large T antigen (SV40TAg), which has a simple basic NLS (PKKKRKV) sequence (65), and that of nncleoplasmin (66), which has a bipartite basic NLS sequence (KRPAATKKAGQAKKKK), in which two interdependent clusters of basic amino acids are separated by a flexible spacer (66, 67). Usually, the first two adjacent basic amino acids (Arg or Lys) in this sequence are followed by a spacer region of any ten residues and at least three basic residues (Arg or Lys) in the five positions after the spacer region (67). Proteins containing NLS are transported into the nucleus in a process that involves NLS binding to the nuclear import receptor importin/~ together with members of the importin a family (68). The sequence511KREFGEHTKMTDV KKPK52' of rodent iPLA2fl (or 565KREFGEHTKMTDV--ff-KPKTM of the long is~orm of human iPLA2/3) perfectly matches the bipartite nuclear localization signal in nucleoplasmin (66), suggesting that iPLA2fl might have the ability to translocate to the nucleus. Recently, we found that iPLA2fl protein can be identified immunochemicallyin nuclei isolated from iPLA2fl-overexpressing IN S-1 insulinoma cells (manuscript in preparation). We found that a small percentage of iPLA2fi was detected in nuclei compared with cytosol under resting conditions. We are exploring the possibility that the translocation ofiPLA2fl to nuclei can be stimulated under some conditions. As suggested by crystal structures of IKB-NF-xB complexes (63, 64), we propose that the ankyrin-repeat domain of iPLA2fi might
GROUP VIA Ca2+-INDEPENDENT PHOSPHOLIPASE A2
15
regulate nuclear translocation through its binding with the NLS of iPLA2fi intramolecularly. The ankyrin repeats might form a slightly bent cylinder like those of IxBce. The NLS might fold back to contact the ankyrin-repeat domain and thereby block the NLS of iPLA2fl and prevent recognition by nuclear import receptors. Upon stimulation, modulatory factors might interact with the ankyrinrepeat domain to release the NLS of iPLA0fi and lead to nuclear translocation.
E. Caspase-3 Cleavage Site Recently, Atsumi et al. (69, 70) reported that treatment of human promonoeytie U937 cells with apoptosis-indncing agents, such as anti-Fas antibody or TNFa/cyeloheximide (CHX), was accompanied by a time-dependent increase in [3H]araehidonie acid (AA) release from prelabeled cells. The time-dependent [3H]AA release paralleled the accumulation of apoptotie cells. By immunoblotting analyses of U937 cells with anti-iPLA2j~ antibody, this group observed that, in addition to an intact 85-kDa iPLA2J? protein, another immunoreaetive band with an estimated molecular mass of 70 kDa became visible 6-12 h after treatment with TNFot/CHX. During this time period, TNFo~/CHX-indueed AA release and easpase-3 activity increased significantly (70), suggesting that the 70-kDa immunoreactive band might be produced by caspase-3 action. Indeed, a potential caspase-3 cleavage site, DXXDIX (71), occurs in iPLAafl around Asp ls3', which is located near the N-terminal end of the first ankyrin repeat. Moreover; ifiPLA2fl is cleaved at this site (DVTD 183Y in humans, DVTD 183N in rodents), the predicted size of the resulting C-terminal fragment would be consistent with the size of the cleaved fragment observed in this study (70). Caspase-3, one of the key executioners of apoptosis, participates in proteolytic cleavage of many key proteins, such as the nuclear enzyme poly(ADP-ribose} polymerase (PARP), during apoptosis ( 72}. A survey of many protein substrates of easpase-3 indicates that each contains a common cleavage motif DXXDSX (71). The fact that iPLA2fl is a substrate for caspase-3 was further confirmed by eotransfection ofcaspase-3 and iPLA2fi. Cleavage at Asp is'3resulted in the activation ofiPLA2fi activity {70}. cPLA2ol also contains a caspase-3 cleavage motif DELD5225A, and cleavage of cPLA2a at Asp5~2leads to inactivation (69), snggesting that cPLA2a and iPLA2~ activities are differentially modulated in apoptosis.
F. Proline-Rich Region of Human Long Group VIA PLA2 Isoform Cloning of human islet iPLA2fl eDNA species from pancreatic islets and insulinoma cells (12) revealed two isoforms of different lengths (Fig. 1). The short iPLA2/3 isoform (SH-iPLA,2fi) corresponds to rodent iPLA.2]~, and the long iPLA2/3 isoform (LH-iPLA2fl) corresponds to that cloned from human B-lymphocyte-derived cell lines (11). The amino acid sequence for the long
16
ZHONGMIN MA AND JOHN TURK
E P
P
H N~N G ] H I L I Q ~ . P J M ~ P I P [H P GIH Y ~ V
DAF-3 LH-iPLA2~ H
~A
Smad4
p x ¢ . _G ¢ P . . Q n x ¢ ¢ . I:! !t P F_ x , . 1~ ¢ Q_p _p ¢ ¢ S . N . L . ¢ Q Consensus Fie,. 4. Alignment of proline-rich region of LH-iPLA2 with the proline-rich middle linker domains of the Smad proteins DAF-3 and Smad4. Identical residues are enclosed in black boxes and conservative changes are enclosed in open boxes. In the consensus sequence, amino acid residues that are identical between at least two of the three sequences are indicated by underlined capihdized letters. Conservative changes are based on functional groups of amino acids. Basic residues (H, K, and R) by/~; hydrophobie residues (A, F, I, L, M, P, V, and W) by qb; polar residues (C, G, N, Q, S, T, and Y) by n. Positions at which there is no similarity are denoted by dots. [Reprinted with permission from Z. Ma, X. Wang, W. Nowatzke, S. Ramanadham, and J. Turk, J. Biol. Chem. 274, 9607-9616 (1999).]
isoform differs from the short isoform by the presence of a 54-amino acid insert in the region of the eighth ankyrin repeat. This insert corresponds exactly to the amino acid sequence encoded by exon 8 of the human iPLA2fl gene (12, 53). This insert is proline-rich, and a BLAST search revealed similarities to the proline-rich middle linker domain of the DAF-3 Smad protein from C. elegans (73), which is most closely related to mammalian Smad4 (Fig. 4) (74). Smad4 is a Mad-related protein and has been identified as the product of the tumor suppressor gene dpc4. This gene is deleted or mutated in a proportion of human pancreatic (75), breast, ovarian (76), and co!orectal tumors (77). The tumor suppressor activity of Smad4 is probably attributable to its participation in the signaling pathway of a family of cytokines that includes TGF-fl (78). The proline-rich middle linker region of Smad4 shares a PXsPXsHHPX12NX4Q motif with the corresponding region of DAF-3 and the proline-rich region in the long human iPLA2fl isoform. The Smad4 middle linker domain mediates protein interactions with signaling partners (74), is located near the center of the protein, and separates an N-terminal MH1 domain with DNA binding activity from a C-terminal MH2 domain with transcriptional activity (79). The prolinerich region in the long iPLA2fl isoform is also located near the center of the protein and separates an N-terminal domain with protein binding activity from a C-terminal catalytic domain (12). Smad proteins participate in controlling cell proliferation and apoptosis and form heterooligomers with signaling partners, via the proline-rich middle linker domain in the case of Smad4 (79).
G. Other Features Another feature of iPLA2/8 is its ability to bind calmodulin, a regulatory protein involved in a variety of cellular calcium-dependent signaling pathways. Both iPLA~fl expressed in Sf9 cells from the rat eDNA and native iPLA2fl in rat brain
GROUPVIACa2+-INDEPENDENTPHOSPHOLIPASEA2
17
can be purified by ealmodulin-affinity column (41, 42). These results suggest that iPLA~2~ is able to bind calmodulin in a Ca2+-dependent manner. Removal of Ca 2+ leads to the dissociation of iPLA2fi from ealmodulin-affinity matrices. In the presence of Ca ")+, the aetivi~ of iPLA2fi is inhibited by calmodulin ill a concentration-dependent manner (41), suggesting that Ca '2+ and ealmodulin negatively regulate iPLA2~ activity. Cahnodulin is a protein capable of recognizing positively charged, amphiphilie oe-helical peptides rather than a clearly defined amino acid sequence motif (80); thus, it is relatively difficult to identify cahnodulin-binding domains from sequence analyses. The amino acid sequences of a number of cahnodulin-binding proteins have been determined, and in many cases the loeations of the binding domains have been mapped by deletion nmtagenesis or chemical methods (80). Similar studies might permit identification of the eahnodulin-binding domain of iPLA2~. Phosphorylation is an important posttranslational modification for regulating the function of proteins. To date, no phosphorylation of iPLA2~ has been reported, although PROSCAN results indicate that iPLA.2~ contains eonsensns phosphorylation sites for calcium/calmodulin-dependent protein kinase II, protein kinase A, protein kinase C, protein kinase G, and casein kinase I!.
IV. Gene Structure, Alternative Splicing, and Chromosomal Localization Recently, we reported the cloning of the human iPLA2~ gene by screening a human Lambda FIX II genomic library and determination of its structure by combining sequencing and PCR approaches (12). Subsequently, Larsson et al. (53) reported the analysis of human iPLA2fi gene from two genomie clones (H $228A9 and H $447C4, accession numbers AL022322 and AL021977, Sanger Center, Hinxton, Cambridgeshire CB10 1SA, UK). The human iPLA2fi gene spans about 70 kb and consists of at least 17 exons, ranging from 74 to 811 bp in size, and 16 introns, ranging from 0.2 kb to 23 kb (Fig. 5). The 5'-untranslated region was identified as exon la and part of lb. Exon la is contained in done HS447C4, while exon lb and the rest of the exons are contained in clone HS228A9. The translational stop codon of iPLA2~, the 3'-untranslated region, and the polyadenylation signal were located in exon 16. Analysis of the exon/intron boundary sequences indicated that the 5'-donor and 3~-aceeptor sequences at splicing sites conform to the generally recognized consensus sequences (12, 53). Human islets express mRNA speeies encoding two iPLA2]~ isoforms, as do human U937 promonocytie cells (12). The 162-bp in-frame insertion in the
'T t",i
&
0
"l:::l
<
~ tH II I ~ / / II l l l t t ' / /
II
,,,,~,~/
!i
III I~lltt/
IHmft /
I HU~l~" i ' i
-~
II
E
q ©
1 U~-I Hill
Cl
I ttoo~
q
[..,,
>
>
llH~ui~_~ "~ i H u ~ . ~ - ~
~
III pUlI'I~ "9 I U ~ - -
i.
.~
o~~o ,oqx
If!
<.-
GROUPVIACa2+-INDEPENDENTPHOSPHOLIPASEA2
19
eighth ankyrin repeat of SH-iPLA2/3 corresponds exactly to exon 8 of the human iPLA2/~ gene. This indicates that mRNA encoding the SH-iPLA2/3 isoform arises from an exon-skipping mechanism of alternative splicing (81). Severn splicing variants of human iPLA2/3 have been identified from EST clones and reflect insertions of 52, 53, and 168 bp, respectively, that do not occur in the transcripts encoding LH- and SH-iPLA2/~ isoforms (11). Indeed, these insertions arise from introns and are designated E8b, E9b, and E13a, accordingly. Analyses of the exon/intron boundary sequences of these alternative-splice sites demonstrate consensus splicing site sequences in the corresponding introns (12, 53). These alternative splicing events may yield three putative transcripts, as elucidated in Fig. 5. These putative transcripts contain a polyadenylation signal encoded by exon 16 and therefore acquire a poly(A) tail required for translational competence. The iPLA2-2 transcript includes E13a and encodes a C- terminally truncated protein due to the introduction of a premature translational stop codon x~thin E13a. This variant retains the GXSXG sequence. The ankyrin-iPLAe-1 transcript includes E9a and encodes a C-terminally trnncated protein due to the introduction of a premature translational stop eodon within Ega. The transcript ankyrin-iPLA2-2 results from skipping of exon 2 and inclusion of ESa and E9a. This transcript encodes a truncated protein that has a deletion close to the N terminus and a C-terminal truncation f r o m a premature stop codon within the E9a. Neither variant ankyrin-iPLA2-1 nor -2 contains GXSXG sequence, and both are catalytic inactive. Since iPLA2/~cDNAs cloned from rodents are similar to human SH-iPLA2¢I (8-10, 12). a question of interest is whether a long iPLA2/~ isoform also exists in rodents. Larrson et al. (53) reported the identification a PCR fragment containing a sequence corresponding to exon 8 of the hmnan iPLA2/~ gene from the total RNA of rat vascular smooth muscle cells and concluded that the presence or absence of the exon 8 is a tissue-specific and not a species-specific feature. We have mapped the human iPLA2/~ gene to chromosome 22q13.1 using fluorescence in situ hybridization (FISH) (12). The two human genomic clones HS228A9 and HS447C4 represent the DNA fragments from chromosome 22q12.3-13.32 between the genetic markers DS426 and DS272 (53). Allelic losses of chromosome arm 22q are frequently observed in human meningiomas and carcinomas of the colon, ovary, and breast. Recent studies of loss of heterozygosity (LOH) in human breast carcinomas revealed that 40-66% of these tumors were associated with LOH on chromosome 22q13.1, suggesting that one or more tumor suppressor genes associated with breast cancer reside in this region (82, 83). Studies of allele loss in human head and neck carcinomas also indicated the possibility that tumor suppressor genes reside on human chromosome 22q13.1 (84). These results raise the possibility that iPLA2/~ represents a candidate for a tumor suppressor gene located on chromosome 22q13.1.
20
ZHONGMINMAAND JOHN TURK
V. Tissue Distribution and Expression The mRNA encoding iPLA2fl has been found in all human, mouse, and rat tissues studied, with different relative abundance (8-12, 53). The tissue distribution in rodents has been examined by Northern blot analysis using mouse/rat multiple tissue blot (8). The data obtained from this study revealed that a single 3.2-kb iPLA2fl transcript could be identified in all examined tissues, with highest expression in testis followed by liver and kidney. Analyses of human tissues identified multiple bands of different sizes (53). Among these, a 3.2-kb band, which might represent the full-length human iPLA,2fl transcripts, was abundantly expressed in testis, brain, spinal cord, and thyroid. On the other hand, the 3.2-kb band may contain multiple transcripts that include alternative splicing variants because the "-~200-bp differences among these variants are difficult to distinguish by Northern blots. Characterization of the regional distribution of various iPLA2fi enzymes in organs has been reported (42, 85). Although several of groups of PLA,2 are expressed in rat brain, Ca2+-independent PLA2 activity accounts for the dominant PLA2 activity in all brain regions, including cerebral cortex, cerebellum, hippocampus, hypothalamus, and striatum (42). Northern blot analyses indicated that mRNA species of iPLA2fl, iPLA,2F, and cPLA2F are expressed in brain (42). Because iPLA2F and cPLA2F are largely associated with membranes, the Ca2+-independent PLA2 activity measured in the cytosolic fraction of each brain region probably represents the iPLA2fl activity (42).
VI. Enzymology of Group VIA PLA2 A. PhospholipaseA2 and PhospholipaseA1 Activities of Group VIA PLA2 Enzymatic activities of iPLA2fl have been characterized with the protein purified from cultured cells (7, 8) and tissue (42) and with recombinant enzyme expressed from iPLA2fl eDNA in various host cells (40, 41 ). Using singly labeled (1-pa1mitoyl-2-[1- 14 C]palmitoyl-sn-glycerol-3-phosphorylcholine) and doubly labeled substrates (1,2-[1-14C]palmitoyl-sn-glycerol-3-phosphorylcholine), purified iPLA2fl has been demonstrated to exhibit both phospholipase A2 and A1 activities. The enzyme preferentially hydrolyzes sn-2 over sn-1 fatty acid substituents by a factor of 5 for 1,2-dipalmitoyl phosphatidylcholine (PC) in mixed micelles (7, 8).
B. Selectivityof Group VIA PLA2for Phospholipids The fatty acid selectivity of iPLA2fl was evaluated by using different substrates. When purified iPLA2fl from P388D1 macrophage-like cells was assayed
GROUP VIA Ca2+-INDEPENDENT PHOSPHOLIPASE A:~
21
using mixed micelles of 100 #M phospholipid and 400/zM Triton X-100, the enzyme showed no preference for either sn-2 arachidonic acid- or sn-1 alkyl ether-containing phospholipids (7). Under the same conditions, purified rat brain iPLA2/~ revealed the following substrate preference toward the fatty acid chain in the sn-2 position of PC: lionleoyl > palmit@ > oleoyl > arachidonoyl (42). In the case ofiPLA2fl purified from CHO cells (8), the enzyme also failed to exhibit any significant preference for particular fatty acid sn-2 substituents, although different rates of hydrolysis of unsaturated t~attyacids were observed in the assay with different Triton X-100 concentrations. When the activity assay was performed with 50 #M or 500 #M Triton X-100, different hydrolysis rates were observed with some substrates. The overall rates of hydrolysis were, at least for some substrates, sensitive to lipid presentation, e.g., 1,2-dipalmitoyl PC was hydrolyzed four times more rapidly when substrate was dispersed in 500 vs. 50 #M Triton, whereas 1-hexadecyl-2-arachidonyl-PCwas hydrolyzed eight times more rapidly when substrate was presented in 50 vs. 500 #M Triton. These findings indicate that the apparent fatty acid selectivity ofiPLA.~fl depends upon substrate presentation (8). Rat brain iPLA2fl has been reported to hydrolyze PC substrates with an sn-2 linoleate residue five times more rapidly these with an sn-2 arachidonate substituent (42). Yang et al. (42) have proposed that iPLA2fl could be an important enzyme in linoleate metabolism. Purified recombinant iPLA2/~ exhibits little preference between substrates containing choline vs. ethanolamine head groups (7, 8, 40), although choline substrates are hydrolyzed more rapidly than other head-group classes under some assay conditions (8). Purified rat brain iPLA2fl also showed no significant head-group preference (42).
C. Lysophospholipase,PAF Acetylhydrolase, and TransacylaseActivitiesol"Group VIA PLA2 Purified iPLA2fl from 388D1 cells and recombinant iPLA2fl expressed from the CHO cell cDNA exhibited detectable lysophospholipase activity in a Triton X-100 mixed-micelle assay (7, 8). The kinetic studies byWolfand Gross (40) using the purified recombinant iPLA2fl overexpressed in Sf9 cells from CHO cDNA using L-or1-palmitoyl-2- [1-14C]arachidonyl-PC as substrate demonstrated that the recombinant protein exhibited calcium-independent phospholipase A1/A2 and lysophospholipase activities at similar levels when assayed in the absence of Triton X-100. Assay conditions therefore have an important impact on the apparent expression of various lipase activities of iPLAefl. With the combination of 50 #M Triton X-100 and 50% glycerol, the lysophospholipase activity of iPLA2fl appeared equivalent to its PLA2 activity (19). The iPLA2fl activity varied with Triton-X 100 concentration, and optimal activity was observed at a Triton/phospholipid molar ratio of 4:1 (7).
22
ZHONGMINMAANDJOHNTURK
The sn-1 ether-linked substrate PAF can be hydrolyzed by iPLA2/3 in assays employing 500/zM or 50/zM Triton X-100, indicating that the phospholipase A2 activity of iPLA2/3 is not restricted to long-chain fatty acids (7, 8). The observed PAF acetylhydrolase activity was about 5% of the corresponding to the iPLA2 activity (7, 8). Like the lysophospholipase activity of iPLA2/3, the apparent PAF acetylhydrolase activity was also influenced by assay conditions, suggesting that the enzyme activity is strongly dependent on substrate presentation (19). The iPLA2/3 also exhibited both lysophospholipid/transacylase and phospholipid/transacylase activities that were susceptible to inhibition by BEL (19).
VII. Potential Cellular Functions A. Signaling Function in Insulin-Secreting Cells Stimulation of islets with glucose leads to insulin secretion and hydrolysis of arachidonate from/3-cell membrane phospholipids (86, 87). Nonesterified arachidonate may participate as a second messenger in glucose-stimulated insulin secretion (see Refs. 88-91). The mechanism of glucose-stimulated arachidonate release from the membrane phospholipids is incompletely understood. Many studies (92-101) have demonstrated that PLAz activation is involved in insulin secretion since glucose and other secretagogues such as carbachol and CCK-8 stimulate PLA2 activation in islets. Inhibition of PLA2 activities suppresses insulin secretion, release of AA, and accumulation of lysophospholipids. Recent studies (102, 103) in insulin-resistant mice indicate that islet P LA2 activation is potentiated in these animals, providing evidence that islet PLA2 activation occurs in hyperinsulinemic mice. Studies in isolated islets, insulinoma cells, and whole animals thus suggest that activation of PLA,2 is involved in secretagogue-induced insulin secretion. Several PLA2s have been identified in islets (9, 104-107) with various sensitivities to PLA2 inhibitors. Any individual or multiple PLA2 enzymes might play signaling roles in islet/3 cells. The suicide substrate BEL inhibits iPLA2/3 at concentrations that do not inhibit Ca2+-dependent sPLA2s or cPLA2 activities (17, 35). BEL is thus a useful tool for distinguishing iPLA2/3 from other PLA2s (18), although BEL also inhibits the MgZ+-dependent phosphatidic acid phosphohydrolase (PAPH-1) (108) and iPLA2F (group VIB PLA2) (6). Interestingly, glucose-induced release of arachidonate from islets requires glucose to be metabolized ( 88, 109, 11 O) but can occur without calcium influx (86, 88), suggesting that the PLA2 responsible for AA release in islet/3 cells may be activated by glucose metabolism and is Ca2+-independent. Treatment of isolated islets or insulinoma cells with BE L suppresses both hydrolysis of arachidonate from membrane phospholipids and glucose-stimulated insulin secretion (111-116), suggesting
GROUPVIACa2+-INDEPENDENTPHOSPHOLIPASEA2
23
that iPLA2 may be responsible for glucose-stimulated AA release. These observations have led to the hypothesis that iPLA2 in fi cells may participate in glucosestimulated insulin secretion by releasing AA upon glucose stimulation (88). Characterization of islet iPLA2 activity led to the cloning of iPLA2fi from rat and human islets (9, 12). BEL also inhibits the Mg2+-dependent PAPH-1 activity (108). To circumvent this problem, Balsinde et al. (117) have used antisense oligonucleotides to suppress iPLA2~ expression in P388D1 cells; however, similar approaches have not been successful in insulinoma cells, perhaps because of the high level of iPLA2fl expression in these cells (118). Nevertheless, the PAPH-1 inhibitor propranolol failed to suppress glucose-induced arachidonate release, as measured by isotope dilution gas chromatography/mass spectrometry (GC/MS), from isolated islets under conditions where such release was effectively suppressed by BEL (118). This indicates that PAPH-1 is not the BELsensitive target involved in glucose-stimulated arachidonate release and insulin secretion, but that iPLA2fl might be. Substantial evidence indicates that PLA2 is involved in secretagogue-induced insulin secretion and AA release, and the studies with BEL suggest that iPLA2fl participates in these processes. In order to examine the role of iPLA2fi further, we have achieved stable overexpression of iPLAefl in INS-1 cells, an insulinoma cell line (119). We found that iPLA2floverexpressing INS-1 cells exhibit increased secretory responses to insulin seeretagogues compared to parent IN S-1 cells or INS-1 cells transfected with vector only (119a). This enhanced insulin secretory response is inhibited by BEL but not by propranolol. These data support the hypothesis that iPLA2/~ participates in glucose-stimulated insulin secretion. Islet PLA2 activation also has been reported to be involved in insulin secretion induced by cholecystokinin-8 (CCK-8) (95, 98). CCK-8 was found to induce accumulation of lysophosphatidylcholine (LPC) and AA in a Ca2+-independent manner. Inhibition of islet iPLA2 activity with BEL reduced CCK-8-induced AA release and insulin secretion, suggesting that iPLA2 contributes to the insulinotropie action of eholecystokinin-8 in rat islets (99). It is also possible that PLA2 enzymes in addition to the group VIA PLA2s are also involved in insulin secretion. Our hypothesis is that secretagogue-induced activation of iPLA2 in islet fl ceils results in accumulation of nonesterified AA and that this amplifies the glucose-induced rise in cytosolic [Ca2+]. This could result in activation of cPLA2, which is also expressed by islet fl cells, and cPLA2 could further amplify the release of AA. Islets also express group IB sPLA2 in their secretory granules, and this enzyme might participate in the process of fusion of insulin secretory granules and plasma membranes in the final steps in exocytosis (120). It is not known whether islets express the recently cloned membrane-associated iPLA2F (group VIB PLA2) (6), but this enzyme is also BEL-sensitive and is another candidate for the BEL-sensitive target(s) that participate in glucose-induced insulin secretion.
24
ZHONGMIN MA AND JOHN TURK
B. Apoptosis Apoptosis, the process of programmed cell death, is associated with changes in glycerophospholipid metabolism. In some cells induced to undergo apoptosis, arachidonate release parallels the reduction in cell viability and DNA fragmentation (121,122). cPLA2 has been implicated in AA release during TNFc~-induced apoptosis (123-125). Fas-induced apoptosis of human promonocytic U937 cells is also accompanied by fatty acid release. Recently, it was reported that both Fasinduced AA and oleic acid release from U937 cells are mediated by iPLA2 but not by cPLA2 (69). In this study, it was observed that, during apoptosis, cPLA2a was converted to a 78-kDa fragment with concomitant loss of catalytic activity. Cleavage of cPLA2a correlated with increased caspase-3-1ike protease activity in apoptotic cells. Neither cPLA20t nor secretory PLA2 inhibitors suppressed AA release, but inhibitors of iPLA2 suppressed AA release and delayed cell death induced by Fas, suggesting that Fas-induced AA release is mediated by iPLA2. In a subsequent study (70), these investigators provided additional evidence that iPLA2 mediates enhanced release of fatty acids from apoptotic cells. Proteolytic cleavage of iPLA2fl by the action of caspase-3 results in loss of the N terminus through the first ankyrin repeat and in production of a form of iPLA2fl that is more active than the uncleaved form. In fact, iPLAzfl does contain a caspase-3 cleavage consensus site DVTDlS3Y. Apoptosis in islet fl cells is induced by stimuli that induce Ca2+-store depletion. This process requires hydrolysis of AA from membrane phospholipids and its conversion to 12-tipoxygenase metabolites by a mechanism that does not require a rise in cytosolic [Ca.~+ ] (126). Ca 2+ -store-depletion-induced hydrolysis of arachidonate from islet and smooth muscle cell phospholipids also does not require a rise in cytosolic [Ca2+] and is mediated by a BEL-sensitive PLA2 (41,127). The PAPH-1 inhibitor propranolol does not block Ca2+-store-deple tion-induced release of arachidonate from islet phospholipids (41). Activation of iPLA2 during Ca2+-store depletion may therefore participate in Ca2+-store depletion-induced apoptosis of fi cells. Some observations raise the possibility that iPLAz may participate in interleukin-lfi (IL-lfi) actions, including induction ofapoptosis. IL-I~ induces apoptosis of human islet ~ cells through Fas-mediated events (128). It was found that IL-I~ induces nitric oxide (NO) production, inducible nitric oxide synthase (iNOS) expression, and accumulation of nonesterified arachidonate and a 12-1ipoxygenase product in islets (129) by a BEL-sensitive mechanism (106). Inhibition of iPLA2 by BEL attenuates prostaglandin generation induced by IL-lfi in renal mesangial cells (130). Recently, it was reported that iPLA2 regulates iNOS induction in cardiac myocytes through its product lysophosphatidic acid (131). When neonatal ventricular myocytes were treated with BEL, IL-l~-induced PGE2 production, AA release, NO production, and iNOS expression were all inhibited. In addition, iPLA2 has been reported to be involved
GROUPVIACa2+-INDEPENDENTPHOSPHOLIPASEA2
25
in potassium-regulated IL-1/~ processing (132). Both IL-1/3 maturation and formation of glycerophosphocholine were blocked by BEL, suggesting that iPLA,2 is involved in IL-1/3 processing. Several features of iPLA2fl fuel interest in the possible participation of the enzyme in apoptosis. These include the presence of the caspase-3 cleavage site consensus sequence DVTD18a(y/N) and the fact that the cleavage at this site removes 183 amino acid residues from the N terminus of iPLA2fl and yields a protein with enhanced catalytic activity (70). In addition, the human long isoform ofiPLA2fl contains a proline-rich region encoded by exon 8 with homology to the proline-rich middle linker domain of Smad4 (Fig. 4). Smad4 is the product of a tumor suppressor gene, participates in apoptosis in some settings, and interacts with its signaling partners via the proline-rich middle linker domain that exhibits homology to the region encoded by exon 8 in human iPLA2/3. Finally, the location of the human iPLA2fl gene on chromosome 22q13.1 occurs at a site known to contain human tumor suppressor gene(s).
C. Membrane PhospholipidRemodeling A role ofiPLA2/~ in membrane phospholipid remodeling has been proposed by Dennis and colleagues using the routine P388D1 macrophage-like cell line as a model system (15, 117, 133). There are two pathways for incorporation of fatty acids into cellular phospholipids. One is the deacylation/reaeylation cycle of membrane phospholipid remodeling, and the other is the de novo synthesis pathway. Most araehidonate in cellular phospholipids is incorporated via a deacylation/reaeylation cycle. Macrophages and maerophage cell lines exhibit the ability to incorporate AA into their membrane phospholipids in a Ca2+-independent manner (134, 135). In P388Dt maerophage-like cells, BEL was found to inhibit AA esterifieation in a dose-dependent and saturatable manner, and the decrease in AA incorporation was associated with inhibition of cellular iPLA2 activity and a reduction in LPC content (133). Based on this study, it was proposed that iPLA2 acts to generate LPC aceeptors for arachidonate incorporation into membrane PC and therefore plays a housekeeping role in phospholipid remodeling (15, 133). In addition, studies using antisense inhibition of iPLA2fl expression in the same system supported this proposal (117). In this model, the function of iPLA2fl is to regulate the deaeylation/reacylation cycle, through which the cells incorporate AA and other unsaturated fatty acids into their membrane phospholipids, by providing the lysophospholipid aeeeptor employed in the acylation reaction. By regulating AA esterifieation through generation of LPC acceptor molecules, iPLA2 might also play a major role in regulating AA turnover in cells (136). While iPLA2 may well play this role in P388D1 cells, it appears not to do so in islets or insulinoma cells (118). Under conditions where islets exhibit essentially no measurable iPLA2 activity after treatment with 10 mM BEL, incorporation
26
ZHONGMIN MA AND JOHN "lURK
of [3H]AA into phospholipids is not suppressed, but rather is enhanced. This is also true for INS-1 insulinoma cells. Under these conditions, the PAPH-1 inhibitor propranolol does not influence [3H]AA incorporation into either islet or insulinoma cell phospholipids, indicating that the effects of BEL on [3H]AA incorporation do not reflect PAPH-1 inhibition. Measurement of islet LPC levels by electrospray ionization mass spectrometry (ESI/MS) with an internal standard indicates that treatment of islets with BEL induces only a 25% decline in LPC content. This modest reduction in islet LPC levels does not limit arachidonate incorporation into islet PC and does not interfere with the subsequent transfer of arachidonate to phosphatidylenthanolamine (PE) (118). Moreover, ESI/MS analyses of incorporation of arachidonate mass into PC of INS-1 insulinoma cells indicate that inhibition of iPLA2 with BEL does not retard, but slightly accelerates this process (118). This is also the case in differentiated human U937 promonocytic cells (137). We have also observed that incorporation of neither [3H]AA nor arachidonate mass, as assessed by ESI/MS, into PC is accelerated in INS-1 insulinoma cells that overexpress iPLA2/~ after stable transfection with rat iPLA2/~ cDNA (119a). Treatment of such iPLA2/~-overepxressing cells with BEL also fails to suppress arachidonate incorporation into PC under these conditions. These observations imply that iPLA2/~ does not play a general role in arachidonate incorporation into cellular PC, although it appears to do so in murine P388D1 macrophage-like cells (117, 133). These cells, however, exhibit some atypical features of arachidonate incorporation and contain only about 3% arachidonate in their phospholipids (135) compared to 25% for natural macrophages (138, 139) and 27% for differentiated U937 promonocytic cells, which are also of monocyte-macrophage lineage. The paucity of arachidonate in the phospholipids of murine P388D1 macrophage-like cells suggests that these cells may be deficient in the arachidonate incorporation mechanisms that are employed by cells capable of maintaining a high content of arachidonate in their phospholipids. In contrast, islets exhibit among the highest arachidonate content in phospholipids of any known tissue (112, 113, 140).
D. Membrane Homeostasisand Other Functions iPLA2 has been suggested to participate in membrane homeostasis by regulation of PC catabolism (141). PC is the most abundant phospholipid in mammalian cell membranes and is essential for cell viability. The cellular levels of this lipid are tightly controlled. Changes in the rate of PC synthesis are generally balanced by changes in PC catabolism. CDP:phosphocholine cytidylyltransferase (CCT) is the rate-limiting enzyme in PC biosynthesis. Overexpression of CCT in Hela cells (141) and CHO cells (142) results not only in increased rates of PC synthesis but also in increased rates of PC breakdown to its catabolite glycerophosphocholine. There is thus little increase in net PC accumulation
GROUPVIACa2+-INDEPENDENTPHOSPHOLIPASEA2
27
(141). The increased PC breakdown under these conditions was associated with increased levels of iPLA2 activity and in iPLA2fl immunoreactive protein (142). Treatment with BEL blocks the accumulation of glycerophosphocholine (141). These observations suggest that upregulation of iPLA2fl occurred as compensatory response to overexpression of CCT in order to maintain a homeostatic relationship between PC biosynthesis and degradation. iPLA2 has been implicated in AA release and eicosanoid synthesis. It has been reported that iPLA2 is involved in the protein kinase C-dependent AA release induced by zymosan in P388D1 macrophage-like cells (143). In this system, inhibition of iPLA2 with either BEL or an antisense oligonucleotide resulted in a decrease in zymosan-induced prostaglandin E2 (PGE2) generation (143). Others subsequently reported that zymosan can induce arachidonate release from a P388D1 subclone without the participation of iPLA2 (144). In isolated ductal cells of rat submandibular gland, activation of iPLA2 by P2X7 agonists ATP or the ATP analog Bz-ATP [2',3'-O-(4-benzoylbenzoyl)adenosine 5'-triphosphate] was observed (145). ATP or Bz-ATP induces the release of [3H]AA to the extracellular medium in a time- and dose-dependent manner. In the absence of extracellular calcium, the release of [3H]AA in response to the purinergic agonists was completely blocked by BEL. In addition, ATP and BzATP stimulated Ca2+-independent secretion of the serine protease kallikrein from these cells, and this secretory event was also blocked by BEL. These investigators concluded that the P2X7 receptor in ductal cells is coupled to kallikrein secretion through iPLA2 activation (145). In human granulocytes, BEL was found to inhibit leukotriene synthesis induced by 0.1-0.15 #M of calcium ionophore A23187 or opsonized zymosan (146). Treatment with propranolol did not attenuate leukotriene synthesis under these conditions, indicating that PAPH-1 was not involved in leukotriene generation (146). A role for iPLA2fl in prostaglandin generation has been suggested in the iPLA2floverexpressing human embryonic kidney 293 cells (147, 148). iPLA2/3 overexpression resulted in increased spontaneous fatty acids release (147) and in the A23187-induced fatty acid release (148). iPLA2fl-derived AA was not metabolized to PGE2. When cotransfected with hamster iPLA2fl cDNA and human cyclooxygenase-1 (COX-l) cDNA, the transfectants released an amount of PGE2 that was substantially higher than that released by cells cotransfected with iPLA2fl cDNA and cyclooxygenase 2 (COX-2), suggesting that A23187-induced AA release by iPLA2/~ is preferentially metabolized by COX-1 compared to COX-2 (148). Production oflysophosphatidic acid (LPA) by iPLA2 has also been suggested to play several important roles in cells. It was reported that iPLA2 is required for ot2-adrenergic-indueed preadipocyte spreading (•49). BEL specifically blocks this response without affecting spreading induced by LPA or serum. The effect of BEL to inhibit c~2-adrenergic agonist-induced preadipocyte spreading was
28
ZHONGMINMAANDJOHNTURK
reversed by LPA but not by AA and other fatty acids. These observations suggest that iPLA2 is involved in ot2-adrenergic control of preadipocyte spreading via its product LPA (149). Studies in parotid acinar cells (150) suggest that iPLA2 might also be involved in exocytotic membrane fusion. In these cells, exocytosis of amylase stimulated by isoproterenol was dose-dependently inhibited by BEL but not by AACOCF3 at concentrations up to 30/zM. BEL also inhibited amylase release evoked by forskolin and membrane-permeable cAMP analogs, but it did not inhibit cAMPdependent protein kinase activity in parotid acinar cells (150).
VIII. Future Perspectives Emerging evidence suggests that iPLA2fl could play a variety of roles in different cell types. Salient structural features of the enzyme include a GXSXG serine lipase consensus motif in which the central serine is required for catalysis, an ATP-binding domain, an ankyrin-repeat domain that probably mediates homomultimer formation and could mediate heteromultimer formation with regulatory proteins, and a bipartite nuclear localization signal sequence. In addition, multiple splicing variants of the iPLA2fl exist, and the long isoform of human iPLA2fl contains a region encoded by a distinct exon of the human iPLA2fl gene. This region shares homology with the proline-rich middle linker domain of Smad4, and this domain mediates interaction of Smad4 with other proteins. That iPLA2fl activity can be regulated by formation ofheterooligomefic complexes is suggested by the fact that coexpression of truncated, catalytically inactive splicing variants with full-length iPLA2 results in alternation of iPLA2fl activity. These observation suggest that iPLA,2fl may be subject to complex regulatory mechanisms that differ among cell types. Further studies are required to determine how iPLA2fl is activated by extracellular signals and how it is regulated within the cell. Protein evolution demands conservation of key residues to maintain structural integrity but allows for sequence variation that provides functional specificity. The existence of an ankyrin-repeat domain in iPLA2fl suggests that the enzyme could interact with other proteins, such as integral membrane proteins or cytosolic proteins. Such binding could regulate iPLA2fl function either positively or negatively. This structural feature is unique to iPLA2fl among the PLA2 family but is common in some regulatory proteins such as transcriptional factors and cell-cycle regulators. This suggests that iPLA2~ may be involved in similar functions, such as regulating nuclear membrane homeostasis during the cell cycle. Such a role would be consistent with the presence of a bipartite nuclear localization signal sequence within the iPLA2fl molecule. Changes in cell nuclei are hallmarks of apoptosis. Nuclear membrane degradation resulting from activation of iPLA2fl could explain its suggested involvement in
GROUP VIA Ca2+-INDEPENDENT PHOSPHOLIPASE A2
29
apoptosis. D e t e r m i n i n g w h e t h e r iPLA2fl translocates f r o m cytosol to n u c l e u s d u r i n g apoptosis is an i m p o r t a n t f u t u r e objective.
REFERENCES 1. 2. 3. 4. 5. 6. 7. 8.
E. A. Dennis, Trends" Biochem. Sci. 22, 1-2 (1997). R. E Irvine, Biochem. J. 204, 3-16 (1982). B. Samuelsson, Science 220, 568-75 (1983). K. Jalink, P. L. Hordijk, and W. H. Moolenaar, Biochim. Biophys. Acta 1198,185-196 (1994). D. A. Six and E. A. Dennis, Biochim. Biophys. Acta 1488, 28-39 (2000). D. J. Mancuso, C. M. Jenkins, and R. W. Gross,]. Biol. Chem. 275, 9937-9945 (2000). E. J. Ackermann, E. S. Kempner, and E. A. Dennis, J. Biol. Chem. 269, 9227-9233 (1994). J. Tang, R. W. Kriz, N. Wolfman, M. Shaffer, J. Seehra, and S. S. Jones, ]. Biol. Chem. 272, 8567-8575 (1997). 9. Z. Ma, S. Ramanadham, K. Kempe, X. S. Chi, J. Ladenson, andJ. Turk, J. Biol. Chem. 272, 11118-11127 (1997). 10. M. A. Balboa, J. Balsinde, S. S. Jones, and E. A. Dennis, J. Biol. Chem. 272, 8576-8580 (1997). 11. P. K. Larsson, H. E. Claesson, and B. P. Kennedy, J. Biol. Chem. 273, 207-214 (1998). 12. Z. Ma, X. Wang, W. Nowatzke, S. Ramanadham, and J. Tnrk, J. Biol. Chem. 274, 9607-9616 (1999). 13. K.W. Underwood, C. Song, R. W. Kriz, X. J. Chang, J. L. Knopf, and L. L. Lin, J. Biol. Chem. 273, 21926-21932 (1998). 14. R. T. Pickard, B. A. Strifler, R. M. Kramer, andJ. D. Sharp, J. Biol. Chenl. 274, 8823-8831 (1999). 15. J. Balsinde and E. A. Dennis, J. Biol. Chem. 272, 16069-16072 (1997). 16. D. M. Stafforini, T. M. Mcintyre, G. A. Zimmerman, and S. M. Prescott, J. Biol. Chem. 272, 17895-17898 (1997). 17. E. J. Aekermann, F. K. Conde, and E. A. Dennis, J. Biol. Chem. 270, 445-450 (1995). 18. J. Balsinde, M. A. Balboa, P. A. Insel, and E. A. Dennis, Annu. Rev. Pharmacol. Toxicol. 39, 175-189 (1999). 19. Y. C. Lio and E. A. Dennis, BiochinL Biophys. Acta 1392, 320-332 (1998). 20. K. Senda, H. Yoshioka, N. Doke, and K. Kawakita, Plant Cell Physiol. 37, 347-353 (1996). 21. C. May, R. Preisig-Muller, M. Hohne, P. Gnau, and H. Kindl, Biochim. Biophys. Acta 1393, 267-276 (1998). 22. S. Subramani, J. Biol. Chem. 271, 32483-32486 (1996). 23. S. Subramani, Physiol. Rev. 78, 171-188 (1998). 24. E L. Zhang and P. J. Casey, Annu. Rev. Biochem. 65, 241-269 (1996). 25. L.W. Tjoelker, C. Wilder, C. Eberbardt, D. M. Stafforini, G. Dietsch, B. Schimpf, S. Hooper, T. H. Le, L. S. Cousens, G. A. Zimmerman, Y. Yamada, T. M. Mclntyre, S. M. Prescott, and P. W. Gray, Nature 374, 549-553 (1995). 26. K. Hattori, H. Adachi, A. Matsuzawa, K. Yamamoto, M. Tsujimoto, J. Aoki, M. Hattori, H. Arai, and K. Inoue, J. Biol. Chem. 271, 33032-33038 (1996). 27. D. M. Staffbrini, S. M. Prescott, G. A. Zirnmerman, and T. M. Mcintyre, Biochim. Biophys. Acta 1301, 161-173 (1996). 28. M. Murakami, Y. Nakatani, G. Atsumi, K. Inoue, and I. Kudo, Crit. Rev. Inimumd. 17,225-283 (1997). 29. J. D. Clark, L. L. Lin, R. W. Kriz, C. S. Ramesha, L. A. Sultzman, A. Y. Lin, N. Milona, and J. L. Knopf, Cell (Cambridge, Mass.) 65, 1043-1051 (1991).
30
ZHONGMIN MA AND JOHN TURK
30. J. D. Clark, A. R. Sehievella, E. A. Nalefski, and L.-L. Lin, J. Lipid Mediators Cell Signalling 12, 83-117 (1995). 31. C. C. Leslie, J. Biol. Chem. 272, 16709-16712 (1997). 32. Z. S. Derewenda and U. Derewenda, Biochem. Cell Biol. 69, 842451 (1991). 33. Z. S. Derewenda and A. M. Sharp, ?FIBS 18, 20-25 (1993). 34. J. D. Shaw, R. T. Pickard, X. G. Chiou, J. v. Manetta, S. Kovacevic, J. R. Miller, A. D. Varshavsky, E. F. Roberts, B. A. Strifler, D. N. Brems, and R. M. Kramer, J. Biol. Chem. 269, 23250-23254 (1994). 35. S. L. Hazen, L. A. Zupan, R. H. Weiss, D. P. Getman, and R. W. Gross, J. Biol. Chem. 266, 7227-7232 ( 1991). 36. R.T. Pickard, X. G. Chiou, B. A. Strifler, M. R. DeFelippis, P. A. Hyslop, A. L. Tebbe, Y. K. Yee, L. J. Reynolds, E. A. Dennis, R. M. Kramer, andJ. D. Sharp,J. Biol. Chem. 271,19225-19231 (1996). 37. M. Hattori, H. Arai, and K. Inoue, J. Biol. Chem. 268, 18748-18753 (1993). 38. M. Hattori, H. Adachi, J. Aoki, M. Tsujimoto, H. Arai, and K. Inoue, J. Biol. Chem. 270, 31345-31352 (1995). 39. K. Hattori, M. Hattori, H. Adachi, M. Tsujimoto, H. Arai, and K. Inoue, J. Biol. Chem. 270, 22308-22313 (1995). 40. M.J. Wolf and R. W. Gross, J. Biol. Chem. 271, 30879-30885 (1996). 41. W. Nowatzke, S. Ramanadham, Z. Ma, F.-F. Hsu, A. Bohrer, and J. Turk, Endocrinology 139, 4073-4085 (1998). 42. H. C. Yang, M. Mosior, B. Ni, and E. A. Dennis, J. Neurochem. 73, 1278-1287 (1999). 43. S. K. Hanks, A. M. Quinn, and T. Hunter, Science 241, 4~52 (1988). 44. R. K. Wierenga and W. G. J. Hol, Nature (London) 302, 842 (1983). 45. W. Hemmer, M. McGlone, I. Tsigelny, and S. S. Taylor, J. Biol. Chem. 272, 16946-16954
(1997). M. P. Kamps and B. M. Sefton, Mol. Ceil Biol. 6, 751 (1986). L. Breeden and K. Nasmyth, Nature (London) 329, 651-654 (1987). S. E. Lux, K. M. John, and V. Bennett, Nature (London) 344, 36-42 (1990). P. Bork, Proteins Struct. Funct. Genet. 17, 363~374 (1993). L. L. Peters and S. E. Lux, Seminars Hematol. 30, 85-118 (1993). P. Michaely and V. Bennett, Trends. Cell Biol. 2, 127-129 (1992). S. Naisbitt, E. Kim, J. c. Tu, B. Xiao, C. Sala, J. Valtschanoff, R. J. Weinberg, P. F. Worley, and M. Sheng, Neuron 23, 569-582 (1999). 53. P. K. A. Larsson Forsell, B. P. Kennedy, and H.-E. Claesson, Eur J. Biochem. 262, 575-585
46. 47. 48. 49. 50. 51. 52.
(1999). 54. 55. 56. 57. 58. 59. 60. 61. 62. 63.
S. Liu and J. Mchowat, Am. J. Physiol. 275, H1462-H1472 (1998). H. Meves, Prog. Neurobiol. 43, 175-186 (1994). B. N. Becker, H. F. Cheng, and R. C. Harris, Am. J. Physiol. 273, F554-F562 (1997). S. Gorina and N. Pavletich, Science 274, 1001-1005 (1996). V. Mandiyan, J. Andreev, J. Schlessinger, and S. R. Hubbard, EMBOJ. 18, 6890-6898 (1999). S. G. Sedgiwick and S. J. Smerdon, TIBS 24, 311-316 (1999). A. S. Baldwin, Jr., Annu. Rev. Immunol. 14, 649-683 (1996). S. Ghosh, M. J. May, and E. B. Kopp, Annu. Rev. Immunol. 16, 225-260 (1998). P. A. Baeuerle, Cell 95, 729-731 (1998). T. Huxford, D.-B. Huang, S. Malek, and G. Ghosh, Cell (Cambridge, Mass.) 95, 759-770
(1998). 64. M. D. Jacobs and S. C. Harrison, Cell (Cambridge, Mass.) 95, 749-758 (1998). 65. D. Kalderon, B. L. Roberts, W. D. Richardson, andA. E. Smith, Cell (Cambridge, Mass.) 39, 499-509 (1984).
GROUP VIA Cae+-INDEPENDENT PHOSPHOLIPASE A-2
31
66. J. Robbins, S. M. Dilworth, R. A. Laskey, and C. Dingwall, Ceil (Cambridge, Mass.) 64, 615-623 (1991). 67. C. Dingwall and R. A. Laskey, Trends Biochem. Sci. 16, 478-481 (1991). 68. I. W. Mattaj and L. Englmeier, Annu. Rev. Biochem. 67, 265-306 (1998). 69. G. Atsumi, M. Tajima, A. Hadano, Y. Nakatani, M. Murakami, and I. Kudo, J. Biol. Chem. 273, 13870-13877 (1998). 70. G.-I. Atsumi, M. Murakami, K. Kojima, A. Hadano, M. Tajima, and I. Kudo, J. Biol. Chem. 275, 18248-18258 (2000). 71. G. M. Cohen, Biochem. J. 326, 1-16 (1997). 72. Y. Lazebnik, S. Kaufmann, S. Desnoyers, G. Poirier, and W Earnshaw, Nature (London) 371, 346-347 (1994). 73. G. I. Patterson, A. Koweek, A. Wong, Y. Liu, and G. Ruvkun, Genes Dev. 11, 2679-2690
(1997). 74. M. de Caestecker, P. Hemmati, S. Larisch-Bloch, R. Ajmera, A. Roberts, and R. Lechleider, J. Biol. Chem. 272, 13690-13696 (1997). 7,5. S. A. Hahn, M. Schutte, A. T. Hoque, C. A. Moskaluk, L. T. da Costa, E. Rozenblum, C. L. Weinstein, A. Fischer, C. J. Yeo, R. H. Hruban, and S. E. Kern, Science 271,350-353 (1996). 76. M. Schutte et. al., Cancer Res. 56, 2527-2530 (1996). 77. S. Thiagahngam, C. Lengauer, F. S. Leach, M. Schutte, S. A. Hahn, J. Overhauser, J. K. Willson, S. Markowitz, S. R. Hamilton, S. E. Kern, K. W. Kinzler, and B. Vogelstein, Nat. Genet. 13, 343-346 (1996). 78. G. Lagna, A. Hata, A. Hemmati-Brivanlou, and J. Massague, Nature (London) 383, 832-836
(1996). 79. 80. 81. 82. 83. 84. 85. 86. 87.
M. Kretzschmar and J. Massague, Curt. Opin. Genet. Dev. 8, 103-111 (1998). K. T. O'Neil andW. F. Degrado, TIBS 15, 59-64 (1990). M. Mckeown, Annu. Rev. Cell Biol. 8, 133-155 (1992). E Allione, F, Eisinger, P. Parc, T. Noguchi, H. Sobol, and D. Birnbaum, Int. J. Cancer 75, 181-186 (1998). A. Iida, K. Kurose, R. Isobe, E Akiyama, G. Sakamoto, M. Yoshimoto, E Kasumi, Y. Nakamura, and M. Emi, Genes Chromosom. Cancer 21, 108-112 (1998). A. Miyakawa, X. L. Wang, H. Nakanishi, F. L. hnai, M. Shiiba, T. Miya, Y. Imai, and H. Tanzawa, Int. J. Oncol. 13, 705-709 (1998). G. Y. Molloy, M. Rattray, and R. J. Williams, Neurosci. Lett. 258, 139-142 (1998). B. A. WoH, S. M. Pasquale, and J. Turk, Biochemistry 30, 637"2-6379 (1991). B. A. Wolf, J. Turk, W. R. Sherman, and M. L. McDaniel, J. Biol. Chem. 261, 3501-3511
(1986). J. Turk, R. W. Gross, and S. Ramanadham, Diabetes 42, 367-374 (1993). P. M. Jones and S. J. Persaud, J. Endocrinol. 137, 7-14 (1993). D. Piomelli, C u ~ Opin. Cell Biol. 5, 274-280 (1993). W. A. Khan, G. C. Blobe, and Y. A. Hannun, Cell Signal. 7, 171-184 (1995). S. G. Laychock, Cell Calcium 3, 43-54 (1982). W. Y. Fujimoto and S. A. Metz, Endocrinology 120, 1750-1757 (1987). Y. C. Jolly, C. Major, and B. A. Wolf, Biochemistry 32, 12209-12217 (1993). R. J. Konrad, Y. C. Jolly, C. Major, and B. A, Wolf, Biochem. J. 297, 283-290 (1992a). R. J. Konrad, Y. C. Jolly, C. Major, and B. A. Wolf, Biochim. Bi~rphys. Acta 1135, 215-220 (1992b). 97. A. C. Loweth, J. H. Scarpello, and N. G. Morgan, Biochem. Biophys. Res. Commun. 218, 423-427 (1996). 98. E. Simonsson, S. Karlsson, and B. Ahren, Regul. Pept. 65, 101-107 (1996). 99. E. Simonsson, S. Karlsson, and B. Ahren, Diabetes 47, 1436-1443 (1998). 88. 89. 90. 91. 92. 93. 94. 95. 96.
32
ZHONGMIN MA AND JOHN TURK
100. J. Turk, B. A. Wolf, J. B. Lefkowith, W. T. Stump, and M. L. McDaniel, Biochim. Biophys. Acta 879, 399-409 (1986). 101. J. Turk, R. Gross, and S. Ramanadham, in "Advances in Lipobiology" (R. Gross, ed.), pp. 215239. JAI Press, Greenwich, CT, 1996. 102. E. Simonsson, S. Karlsson, and B. Ahren, Biochem. Biophys. Res. Commun. 272, 539-543 (2000). 103. E. Simonsson and B. Ahren, Int. J. Pancreatol. 27, 1-11 (2000). 104. M. Chen, Z. Yang, A. Naji, and B. A. Wolf, Endocrinology 137, 2901-2909 (1996). 105. R. W. Gross, S. Ramanadham, K. K. Kruszka, X. Han, and J. Turk, Biochemistry 32, 327--336 (1993). 106. Z. Ma, S. fiamanadham, J. Corbett, A. Bohrer, R. W. Cross, M. L. Mcdaniel, and J. Turk, J. Biol. Chem. 271, 1029-1042 (1996). 107. Z. Ma, S. Ramauadham, Z. Hu, and J. Turk, Biochim. Biophys. Acta 1391, 384-400 (1998). 108. J. Balsinde and E. A. Dennis,/. Biol. Chem. 271, 31937-31941 (1996). 109. J. Turk, M. Mueller, A. Bohrer, and S. Ramanadham, Biochim. Biophys. Acta 1125, 280-291 (1992). 110. R.J. Konrad, Y. C. Jolly, C. Major, and B. A. Wolf, Mol. Cell. Endocrinol. 92, 135-140 (1993). 111. S. Ramanadham, R. Gross, and J. Turk, Biochem. Biophys. Res. Commun. 184, 647-653 (1992). 112. S. Ramanadham, A. Bohrer, R. W. Gross, and J. Turk, Biochemistry 32, 13499-13509 (1993). 113. S. Ramanadham, A. Bohrer, M. Mueller, P. Jett, R. W. Gross, and J. Turk, Biochemistry 32, 5339-5351 (1993). 114. S. Ramauadham, R. W. Gross, X. Han, and J. Turk, Biochemistry 32, 337-346 (1993). 115. S. Ramanadham, M. J. Wolf, P. A. Jett, R. W. Cross, and J. Turk, Biochemistry 33, 7442-7452 (1994). 116. S. Ramanadham and J. Turk, Cell Calcium 15, 259-264 (1994). 117. J. Balsinde, M. A. Balboa, and E. A. Dennis,]. Biol. Chem. 272, 29314-29321 (1997). 118. S. Ramanadham, F.-F. Hsu, A. Bohrer, Z. Ma, and J. Turk,]. Biol. Chem. 274, 13915-13927 (1999). 119. M. Asfari, D. Janjic, P. Meda, G. Li, P. A. Halban, and C. B. Wollheim, Endocrinology 130, 167-178 (1992). 119a. Z. Ma, S. fiamanadham, M. Wohltmann, A. Bohrer, F.-F. Hsu, and J. Turk, J. Biol. Chem. 276, 13198-13208 (2001). 120. S. Ramanadham, Z. Ma, H. Arita, S. Zhang, and J. Turk, Biochim. Biophys. Acta 1390, 301312 (1998). 121. I. Kudo, M. Murakami, S. Hara, and K. Inoue, Biochim. Biophys. Acta 1170, 217-231
(1993). 122. M. Koller, P. Wacht|er, A. David, G. Muhr, and W. Konig, Inflammation 21,463-474 (1997). 123. C. Voelkel-Johnson, T. E. Thorne, and S. M. Laster, J. Immunol. 156, 201-207 (1996). 124. M. Hayakawa, N. Ishida, K. Takeuchi, S. Shibamoto, T. Hori, N. Oku, F. Ito, and M. Tsujimoto, J. Biol. Chem. 268, 11290-11295 (1993). 125. D. Wissing, H. Mouritzen, M. Egeblad, G. G. Poirier, and M. J~i~ttel~i,Proc. Natl. Acad. Sci. U.S.A. 94, 5073-5077 (1997). 126. Y. Zhou, D. Tend, E Drayluk, D. Ostrega, M. Roe, L. Philipson, and K. Polonsky, J. Clin. Invest. 101, 1623-1632 (1998). 127. M. J. Wolf, j. Wang, J. Turk, and B. W. Gross,]. Biol. Chem. 272, 1522-1526 (1997). 128. A. C. Loweth, G. T. Williams, R. E L. James, J. H. B. Scarpelo, and N. G. Morgan, Diabetes 47, 727-732 (1998). 129. M. L. McDaniel, G. Kwon, J. R. Hill, C. A. Marshall, and J. A. Corbett, Proc. Soc. Exp. Biol. Med. 211, 24-32 (1996).
GROUP VIA Ca2+-INDEPENDENT PHOSPHOLIPASE A2
33
130. S. Akiba, M. Hayama, and T. Sato, FEBS Lett. 437, 225-228 (1998). 131. E. Isenovic and M. LaPointe, Hypertension 35, 249-254 (2000). 132. I. Walev, J. Klein, M. Husmann, A. Vagela, S. Strauch, H. Wirtz, O. Weichel, and S. Bhakdi, J. Immunol. 164, 5120-5124 (2000). 133. J. Balsinde, I. D. Bianco, E. J. Ackermann, E K. Conde, and E. A. Dennis, Proc. Natl. Acad. Sci. U.S.A. 92, 8527-8531 (1995). 134. J. Balsinde, B. Fernandez, and J. A. Solis-Herruzo, Eur. J. Biochem. 221, 1013-1018 (1994). 135. J. Balsinde, S. E. Barbour, I. D. Bianco, and E. A. Dennis, Proc. Natl. Acad. Sci. U.S.A. 91, 11060-11064 (1994). 136. J. J. Daniele, G. D. Fidelio, and I. D. Bianco, Prostaglandins Other Lipid Mediat. 57,341-350 (1999). 137. F.-F. Hsu, Z. Ma, M. Wohhmann, A. Bohrer, W. Nowatzke, S. Ramanadham, and J. Turk, J. Biol. Chem. 275, 16579-16589 (2000). 138. W. A. Scott, J. M. Zrike, A. L. Hamill, J. Kempe, and Z. A. Cohn, J. Exp. Med. 152, 324-335 (1980). 139. E. E. Kroner, B. A. Peskar, H. Fischer, and E. Ferber,J. Biol. Chem. 256, 3690-3697 (1981). 140. S. Ramanadham, F. F. Hsu, A. Bohrer, W. Nowatzke, Z. Ma, and J. Turk, Biochemistry 37, 4553~1567 (1998). 141. I. Baburina and S. Jackowski, J. Biol. Chem. 274, 9400-9408 (1999). 142. S. E. Barbour, A. Kapur, and C. L. Deal, Biochim. Biophys. Acta 1439, 77-88 (1999). 143. S. Akiba, S. Mizunaga, K. Kume, M. Hayama, and T. Sato, J. Biol. Chem. 274, 19906-19912 (1999). 144. J. Balsinde, M. A. Balboa, and E. A. Dennis,J. Biol. Chem. 275, 22544-22549 (2000). 145. E. Alzola, A. Perez-Etxebarria, E. Kabre, D. J. Fogarty, M. Metioui, N. Chaib, J. M. Macarulla, C. Matute, J. P. Dehaye, and A. Marino, J. Biol. Chem. 273, 30208-30217 (1998). 146. P. K. A. Larsson Forsella, G. Runarssonb, M. Ibrahima, M. Bj6rkholmb, and H.-E. Claessona, FEBS Lett. 434, 295-299 (1998). 147. M. Murakami, S. Shimbara, T. Kambe, H. Kuwata, M. Vi Winstead, J. A. Tischfield, and I. Kudo, J. Biol. Chem. 273, 14411-14423 (1998). 148. M. Murakami, T. Kambe, S. Shimbara, and I. Kudo, J. Biol. Chem. 274, 3103-3115 (1999). 149. C. Pages, A. Bey, M. Lafontan, P. Valet, and S. Saulnier-Blache, Biochem. Biophys. Res. Commun. 265, 572-576 (1999). 150. T. Takuma and T. Ichida, J. Biochem. (Tokyo) 121, 1018-1024 (1997).
CooA: A Heme-Containing Regulatory Protein That Serves as a Specific Sensor of Both Carbon Monoxide and Redox State GARY P. ROBERTS, *'1 M A R C V. T H O R S T E I N S S O N , * R O B E R T L. KERBY,* W I L L I A M N . LANZILOTTA, t AND THOMAS POULOS t
*Department of Bacteriology University of Wisconsin-Madison Madison, Wisconsin 53706 t Department of Biochemistry and Molecular Biology Program in Macromolecular Structure University of California-Irvine Irvine, California 92697 I. II. III. IV. V.
VI. VII. VIII. IX. X. XI.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CO Oxidation by RhodospiriUum rubrum and Other Microorganisms . . . . . The coo Genes and Their Regulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General Behavior of CooA as a Transcriptional Activator Responding to the Redox State and the Presence of CO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Structure of CooA and Its Implications . . . . . . . . . . . . . . . . . . . . . . . . . . A. Comparison of CooA Structure to That of CRP . . . . . . . . . . . . . . . . . . . . B. The H e m e Region of CooA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Model for Activation of CooA by CO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CooA as a Redox Sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CooA as a CO Sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cooperativity of Ligand Binding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transcriptional Activation by CooA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DNA Recognition Properties of CooA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Future Direction and Open Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
36 36 38 40 42 42 47 50 52 53 54 56 58 59 60
CooA, the heme-containing carbon monoxide (CO) sensor from the bacterium RhodospiriUum rubrum, is a transcriptional factor that activates expression of certain genes in response to CO. As with other hemc proteins, CooA is 1To whom correspondence should be addressed. Progress in Nucleic Acid Research and Molecular Biology, Vol. 67
35
Cop~,Tight O 2001 by Academic Press. All rights of reproduction in any form reserved. 0079-6603/01 $35.00
36
GARYE ROBERTSET AL. unable to bind CO when the Fe heme is oxidized, consistent with the fact that some of the regulated gene products are oxygen-labile. Upon reduction, there is an unusual switch of protein ligands to the six-coordinate heme and the reduced heine is able to bind CO. CO binding stabilizes a conformation of the dimeric protein that allows sequence-specific DNA binding, and transcription is activated through contacts between CooA and RNA polymerase. CooA is therefore a novel redox sensor as well as a specific CO sensor. CooA is a homolog of catabolite responsive protein (CRP), whose transcriptionally active conformation has been known for some time. The recent solution of the crystal structure of the CO-free (transcriptionally inactive) form of CooA has allowed insights into the mechanism by which both proteins respond to their specific small-molecule effectors. © 2001 Academic Press.
I. Introduction In the past several years, the field of biological sensing of small gaseous molecules such as NO (1-7), 02 (8-14), and H2 (15), in addition to CO (16-18), has shown dramatic progress and involves both prokaryotes and eukaryotes. These sensors must be able to perform some degree of selectivity in sensing their effectors, so that they can trigger the proper biological response and avoid being inhibited or inappropriately triggered by the "wrong" small molecules. Rhodospirillum rubrum can oxidize CO to CO2, and the expression of genes involved in that process are regulated by CooA (CO-oxidation activator). This heine-containing dimer functions both as a redox sensor and as a specific CO sensor. Part of this process involves a highly unusual switch of protein ligands upon reduction of the heme, as well as the utilization of proline as an axial ligand, which has not previously been seen (19). As a member of the CRP/FNR superfamily of regulatory proteins, many of which respond to small-molecule effectors, the solution of its structure in the effector-free form has provided insight into the mechanism by which CooA, and presumably some other members of this family, become activated in response to their effector molecules. The combination of structural information with the results of a variety of mutagenic, spectroscopic, and functional analyses has also revealed important features that underlie its sensing of redox and CO. The analysis of CooA provides an important addition to the understanding of the selectivity in such sensing systems, as well as suggesting the mechanism by which many members of the CRP/FNR superfamily become active for DNA binding.
II. CO Oxidation by Rhodospirillumrubrum and Other Microorganisms CO is found throughout the environment and is the product of chemical decomposition, biological processes, atmospheric reactions, and human
CooA:A CO-SENSINGTRANSCRIPTIONALFACTOR
37
activity (20). In aquatic environments, the most significant source of CO is thought to be abiotic photooxidation of biological molecules (21, 22), and direct measurements indicate its presence throughout the water column at nanomofar concentrations (21, 23). The removal of CO is through oxidation by radical chemistry in the upper atmosphere or through microbial metabolism, with the latter processes encompassing adventitious mechanisms in which CO is considered a fortuitous substrate, or biochemical systems in which the metabolism of CO (often as an enzyme-bound intermediate) is the connection between oneand two-carbon metabolites of an anabolic or catabolic process. These types of metabolism depend on one of two distinct biochemical mechanisms (24). Aerobic, "carboxydotrophic," organisms exclusively elaborate a molybdenumcontaining hydroxylase that catalyzes the catabolic oxidation of CO and water to CO2 plus reducing equivalents, which ultimately yield H2 through the activity of a hydrogenase. Expression of these systems, such as the 12-gene cox cluster of Oligotr~rpha carboxidovorans, requires the presence of CO, and the cox genes are found adjacent to cbb genes that encode components of the Calvin cycle (25). As yet, the regulatory mechanism of the cox (and homologous) systems has not been described. It will be interesting to compare this CO-dependent regulation, which enables aerobic transcription, with CooA, which controls a low-redox process. Anaerobic CO-oxidizing microbes span a tremendous diversity and encompass methanogenic, acetogenic, and sulfate-reducing organisms. The oxygenlabile enzymatic oxidation of CO to CO2 routinely assayed in vitro is catalyzed by a Ni-containing enzyme that performs, in vivo, the fundamental steps in the interconversion of single-carbon intermediates and acetyl-CoA. While the level of the enzyme may be modestly influenced by the particular carbon substrate upon which the organism is cultivated, its expression does not require the presence of exogenous CO (24). The observation that purple nonsulfur bacteria can anaerobically metabolize CO was first made by Uffen with an organism that is now known as Rubrivivax gelatinosus (26), although R. rubrum is now the best characterized example of the group. Phototrophs express a "hybrid" system: the central Ni-containing enzyme catalyzing CO oxidation is termed carbon monoxide dehydrogenase (abbreviated CODH), and the R. rubrum enzyme is certainly evolutionarily related to those of strict anaerobes, based on predicted primary protein sequences (27). Yet the R. rubrum enzyme, like that of aerobic carboxydotrophs, solely catalyzes CO oxidation to CO,2 and reducing equivalents, which in turn yield He via electron carriers and a specific hydrogenase. The 12-gene coo regulon, aside from that encoding the regulatory protein, is transcriptionally dependent upon the presence of CO. Carboxydothermus hydrogenoformans, a nonphototrophic anaerobe isolated from a volcanic swamp, probably contains a similar CO-oxidation system (28). Perhaps aside from organisms isolated from burning coal piles (29) or volcanic environments (28) where exceptional levels of CO can be expected, the
38
GARYE ROBERTSET AL.
biological niche for most CO-utilizing aerobes and anaerobes is obscure: Measured bulk CO levels, normally in the nanomolar range in aqueous environments and in the range of parts per billion in the atmosphere, are generally well below the Km for CODHs of isolated organisms (30). This is also the case for the R. rubrum coo system. It does not appear to be for "detoxification," both because mutants lacking the coo system are highly tolerant of CO under lab conditions and because R. rubrum is surprisingly capable of coupling the thermodynamically poor anaerobic conversion of CO to CO2 plus H2 to grow at a rate only 20% slower than phototrophic growth in the same medium (31). This complex, specific, and regulated metabolism implies either that R. rubrum naturally experiences periods of exceptional CO exposure or that its native environments are so energy-limiting as to make this system advantageous.
III. The coo Genes and Their Regulation All known coo genes fall into three contiguous transcripts on the R. rubrum chromosome (27, 32, 33), as depicted in Fig. 1. coos encodes the CODH activity and cooF encodes a ferredoxin that forms a tight complex with CooS and promotes its association with the membrane. This membrane association apparently supports efficient electron transfer to the products of the cooMKLXUH operon. Based on sequence similarity and some biochemical characterization (32), the products of the first five genes of this operon form a membrane-associated complex that passes reductant to CooH. In the course of this electron transfer, useful energy is generated, presumably by the generation of a proton gradient, that supports growth on CO as sole energy source (31). CooH is an unusual NiFe hydrogenase in that it is highly tolerant of CO (34), consistent with its role in this CO-oxidizing system, and that it lacks a cleavable C terminus. This C terminus is removed in many Ni-containing hydrogenases upon Ni insertion, but in CooH, the site of normal cleavage is replaced by a stop codon; it is essentially "precleaved" (32). The CooCTJ products are apparently involved in Ni
co /
~
cooM
--
--
I( L X U H
-
C o o A
active~
cooF S C T J
C
inactive
cooA nadC B
F[c. 1. Transcriptional organization of the coo regu]on in R. mbrum. The dark horizontal arrows
indicatethe directionof transcription,nadBC, ORF,and ahpC are not involvedin CO oxidation.
CooA:A CO-SENSINGTRANSCRIPTIONALFACTOR
39
processing for CooS based on mutant phenotype, similarity to other genes, and substantial biochemical and physiological characterization (33, 35, 36). Both the cooMKLXUH and cooFSCTJ operons appear to be completely under the control of CooA, the CO sensor that is the focus of this review, but cooA has a separate low-level promoter so that it is present in the cell under all growth conditions examined (Y. He and G. P. Roberts, unpublished data). The expression of the coo genes is under the control of CooA, which activates transcription only when it is reduced and binds CO (17, 37, 38). This regulation of coo gene expression can be rationalized in light of our knowledge of the regulated gene products. For example, CO is the only known substrate of CooS, the C O D H encoded by this system (39) [although CooS can also run the reverse reaction (40)], and the regulation of coo expression in the absence of CO appears to be extremely tight in R. rubrum, with C O D H activity, C O D H antigen, and cooFSCTJ mRNA being undetectable in the absence of CO. In the presence of low levels of CO, however, the genes are highly expressed (41), consistent with the potential utilization of this system as sole energy source in the cell. The affinity of CooA (K,I ~ 4 # M ) for CO (M. V. Thorsteinsson et al., unpublished data) is in the same range as the Km of CooS ("~32 #M) (42), so that CooA causes Coos to be synthesized when CO levels allow a reasonable level of enzymatic activity. CooS has also been found to be O2-sensitive, as has CooH, so it is not surprising that the coo genes are also expressed only under anaerobic conditions. Less obviously, Coos is maximally active at a redox poise of - 3 0 0 mV and below (J. Heo and P. W. Ludden, personal communication), and the behavior of CooA is consistent with this fact. CooA is competent to bind CO only when its heme is reduced, and the reduction and oxidation midpoint potentials have been reported to be - 3 2 0 and - 2 6 0 mV, respectively (43), reflecting the different initial CooA species present in each redox titration. Although this comparison is certainly simplistic and we have not determined the internal redox state of R. rubrum cells expressing the coo genes, it serves to suggest that regulation of CooA activity might be sufficient to explain the observation that reducing culture conditions are necessary for consistent high-level coo expression in R. rubrum (31). In the absence of CO or in the presence of oxidizing conditions, only cooA is expressed. However, under reducing conditions in the presence of CO, cooFSCTJ and cooMKLXUH become very highly expressed, with the former being expressed at a level approximately fivefold higher than the latter. Oddly, in the presence of CO, some portion of the cooFSCTJ transcript appears to read through into cooA, and Western analysis has shown an increase in CooA accumulation in R. rubrum in the presence of CO (Y. He and G. P. Roberts, unpublished data). The physiological reason for this readthrough, if any, remains unknown. Given the ability of the coo system to provide the sole energy source for the cell, one might expect its expression to be regulated by the energy status of the
40
GARYE ROBERTSET AL.
cell. More specifically, one might predict that the cooFSCTJ operon would be expressed under anaerobic conditions in the presence of CO, since the expression of these genes should support "detoxification" of CO, but that the cooMKLXUH operon would have have an additional layer of regulation that reflects energy needs. However, as we have shown that virtually all of the reductant generated by CooS flows through CooH, the hydrogenase, it appears possible that the proteins encoded by the latter operon are necessary for the efficient consumption of the reductant (31). Also, as both operons are expressed in the presence of light, an excellent energy source, there seems to be no reason to propose additional regulatory inputs, nor is there any molecular evidence for the involvement of other regulatory factors. In general, the highest expression, as measured by protein accumulation, appears to be that ofcooFS, suggesting that some attenuation reduces the transcription into cooCTJ (Y. He and G. P. Roberts, unpublished data). Relatively little has been done to study the mechanism by which the coo genes are turned off upon the exhaustion of CO. The addition of 02 to a cooexpressing culture leads to the fairly rapid cessation of coo expression (41), as would be expected, but unanswered questions remain about the precise response to CO exhaustion. Expression would presumably cease only when no CO-bound CooA remained in the cell; however, this exhaustion might take a very long time, as CooA is a fairly abundant regulatory protein (Y. He and G. P. Roberts, unpublished data) and the kinetic rate of dissociation for CO bound to heme proteins (e.g., sperm whale myoglobin) is typically extremely long (with a rate of dissociation of "~1 rain) (44). This "problem" would be exacerbated by the increase in CooA levels in the presence of CO due to transcriptional readthrough. The resolution of this paradox will require further physiological analysis of R. rubrum under these changing conditions.
IV. General Behavior of CooA as a Transcriptional Activator Responding to the Redox State and the Presence of CO CooA is a homolog of both CRE a cAMP sensor in E. coli that regulates genes whose products are involved in carbon utilization (reviewed in Refs. 45 and 46), and FNR, an E. coli protein involved in regulation of gene expression in response to anaerobiosis (11, 13, 47, 48). Numerous other homologs have been identified through sequence comparison, with biological roles such as control of virulence (49, 50), although CRP and FNR are by far the best studied. Interestingly, although both FNR and CooA are sensors of the redox state of the cell, they do so by completely different mechanisms, and CooA actually appears to be more similar to CRP in general behavior. Both CooA and CRP
CooA:A CO-SENSING TRANSCRIPTIONALFACTOR
41
are homodimers under all conditions (17, 51, 52) and undergo a conformational change upon effector binding. With C RP, binding of the effector cAMP induces a conformational change within the dimer, which is thought to cause the DNAbinding domains to reorient themselves so that they can simultaneously interact with one-half of a palindromic DNA target sequence (see Ref. 53 for a recent method and a summary of previous approaches). In contrast, FNR appears to be primarily regulated through a monomer-dimer transition, the formation of the DNA-binding dimerie form being caused by the formation of Fe4S4 centers only under anaerobic conditions (47, 48). Dimeric CooA contains a protoheme moiety in each monomer that is the site of CO binding (17, 38, 54, 55). It can exist in three states: (1) When the heine is oxidized, CooA is incapable of binding CO or DNA; (2) when the heine is reduced, CooA cannot bind DNA, but can bind CO; and (3) in the reduced, CO-bound form, CooA is competent to bind DNA. The details of these transitions, as well as the structural features that affect this behavior, are described in detail below. Upon reduction, a required step for CO binding, CooA undergoes a conformational change, as detected on native gels (54), which in part reflects a highly unusual ligand switch (38, 54) by which one of the protein ligands to the heine (Cys-75) is replaced by another (His-77). As detailed in Sections V,B and VI, this ligand switch involves a rearrangement of the heme with respect to the protein. It is likely that a role of this switch is to set the proper redox poise of the protein, although this issue has not been well-studied. Upon CO binding to reduced CooA, another conformational change is apparent on native gels (54), and it is this change that leads to a form of CooA that is competent to bind DNA and to promote transcriptional activation in vitro (17, 54). Further information on the DNA sequences bound and the interaction of CooA with RNA polymerase is provided in Sections IX and X. The crystal structure of the reduced, effector-free (reduced but lacking CO) form is described in the following section, as are hypotheses concerning the conformational change upon CO binding. Briefly, the binding of CO displaces the Pro-2 ligand to the heme, thereby triggering the conformational activation. CO, but no other small molecule, leads to the activation of CooA so that it is a specific CO sensor. This is particularly significant, as another well-studied heinecontaining sensor, FixL, responds to a variety of small molecules in vitro (9, 56, 57). Similarly, soluble guanylyl cyclase might be able to respond to CO under some conditions (16, 58), in addition to its well-studied response to NO (1-5). In the case of CooA, however, 02 oxidizes the heme, whereas cyanide, azide, and imidazole fail to bind at all to either oxidation state of the protein (,54). While NO is capable of binding the heine, it displaces both protein ligands and CooA fails to become active (59), suggesting that activation requires the displacement of the appropriate protein ligand together with the retention of the other.
42
GARY P. ROBERTS ET AL.
The cooperativity of CO binding to CooA (see Section VIII) is likely to be of physiological significance, and it is interesting to consider effector binding in CRP. CRP appears to bind its effector, cAMP, with negative cooperativity (60, 61 ), although claims of positive cooperativity have also been made (62, 63). At physiologically reasonable levels of cAMP, one would expect CRP to accumulate with a single cAMP bound; it is therefore unsurprising that this form of CRP appears to have the highest affinity for its DNA target site (60). Rather recently, the structure of CRP was determined with four cAMP molecules bound per CRP dimer (64), although the physiological significance of this form remains unclear and will not be considered further here. However, neither the role of this negative cooperativity for cAMP/CRP in vivo nor its mechanism is understood. In the case of CooA, the functionality of a dimer with a single CO bound is unknown.
V. The S~ructureof CooA and Its Implications The structure of the effector-free form (reduced but without CO) of CooA has been solved at 2.6 A resolution (19) and reveals several striking features. First, each monomer of CooA contains a heme with a novel ligand, which is Pro-2 at the N terminus of the other protein monomer. Proline has not previously been identified as an axial ligand to a heme, and its properties as a ligand are therefore of substantial interest. Second, the structure of the reduced CooA dimer is highly asymmetric (Ref. 19 and Fig. 2A; see color insert) with the two different monomer forms designated A and B. It is clear that this asymmetry is a property of the packing in the crystal lattice. Form B makes no contacts with other dimers in the crystal lattice, while form A is confined to the observed conformation by contacts from three other molecules in the crystal lattice (19). It is therefore our working hypothesis that effector-free CooA in solution is likely a fully symmetric dimer of form B. Although it is clear that there are at least three physiologically significant structures of CooA (oxidized, reduced, and reduced + CO), only the structure of the reduced, effector-free form has been solved. Nevertheless, a number of important insights into the total behavior of CooA can be drawn, in part because of the existence of the structure of the effector-bound form of CRP, both in the presence (65, 66) and absence (67) of DNA. The following sections draw comparisons between the structures of effector-free CooA and effector-bound CRP, then focus on the residues in the immediate vicinity of the heme, and finally propose a mechanism of activation of CooA by CO.
A. Comparison of CooA Structure to That of CRP Because the CooA structure represents its effector-free form, it is useful to compare this CooA structure to the various structures of CRP, all of which
,-,
D
FIG. 2. (A) Structure of the reduced form of CooA (19). The monomer shown in turquoise is referred to in the text as form A, while the monomer in brown is form B. The red molecule is the heme on each monomer. (B) Structure of cAMP-bound CRP (65). The red molecule is cAME (C) C Helix movement upon activation. The panel shows the comparison of the C helix of effectorbound CRP (in orange) compared to that of effector-free CooA (in blue). The relative positions of the heme (in blue) and the cAMP (in orange) are also shown. (D) The vicinity of the heine of the reduced form of CooA. Residues 74-78 are displayed, with His-77 centered over the heme and serving as the ligand. The position of Cys-75 is indicated to the left of His-77.
CooA:A CO-SENSINGTRANSCRIPTIONALFACTOR
43
are in the effector-bound form (including CRP bound to DNA). Figure 2B (see color insert) shows the effector-bound form of CRP, which shows only a modest rotational asymmetry between the two subunits (67), although DNA binding renders the dimer highly symmetric (65, 66). The site of cAMP binding is in the center of the effector-binding domain, immediately adjacent to the long ot helices, termed the C helices (Fig. 3), that form the dimer interface. Indeed, there is an interdigitation of residues in this portion of CRP, such that a residue (Set-128) in one monomer makes contact with the cAMP bound to the other monomer (67). The conformational change in CRP upon cAMP binding is unknown, although a variety of approaches indicate that the change is significant (53, 68, reviewed in Ref. 45). In the absence of cAMP, CRP has much poorer affinity for its specific DNA target (60). Although the structure of effector-bound CooA has not been solved, that structure is likely to be rather similar to that of effector-bound CRP, as both proteins bind structurally similar DNA sites and have at least some similarities in their interactions with RNA potymerase (Section IX). There are several fairly
FIG. 3. Left: Monomerof form B of CooA;right: monomerof CRE For each structure, the letters are adjacentto the appropriateu helices,whilethe numbersare adjacentto the appropriate fl sheetsof the two monomers.
44
GARYE ROBERTSET AL.
striking differences between effector-bound CRP and form B of effector-free CooA: (1) The C helix of CooA is extended by means of a fusion of the C and D helices, relative to that of CRP. (2) The relative positions of the C helices within the effector-binding domain in CooA is distinct from that of CRP. (3) There are a number of different contacts made by residues in the "hinge" region, which is between the DNA-binding and effector-binding domains of the two proteins. (4) There is a completely different orientation of the F helices, which create the sequence-specific contacts with DNA, in the two proteins. (5) The positions of the effector-binding sites are similar, but not identical, in CRP and CooA. Each of these issues is addressed in turn below. 1. EXTENSION/FUSIONOF THE C HELIX OF EFFECTOR-FREE CooA Form B of CooA achieves its extended configuration, relative to that of CRP, because of the extension of the long C helix that serves as the dimer interface in both CooA and CRP (Fig. 2A,B). That extension is actually due to the orientation of helix D of the DNA-binding domain (Fig. 3) relative to its position in CRP, such that the D helix fuses with the C helix. In effectorbound CRP, the D helix lies close to the upper surface of the effector-binding domain in an orientation relative to that of the C helix that is more than 140 ° different than in form B of CooA. Effector-bound CRP thus provides a number of contacts between the DNA-binding and effector-binding domains, with both the D and E helices of the DNA-binding domain providing contact with/3 sheet 5 (Fig. 3) of the effector-binding region. In contrast, there is relatively little surface contact between these two domains in form B of CooA, the only apparent interaction being between the loop between the E and F helices of the DNA-binding domain and the tip of the loop formed by/3 sheets 4 and 5 (the "4/5 loop"), an element of the effector-binding domain. Although it is not clear that effeetor-free CRP has a similar extended structure, that hypothesis is consistent with the data, as neutron scattering experiments (68) have indicated that effector-free CRP is significantly extended relative to effector-bound CRP. If both the effector-bound and the effector-free forms of CRP and CooA were shown to be similar, it would imply that the mechanism of activation of the two proteins was also likely to be similar. It is tempting to hypothesize an additional sort of stabilizing interaction that might also be present in a dimer of form B CooA namely, interaction between the two DNA-binding domains themselves. Such an interaction is difficult to assess at present because the structure is defined only through Ala-213, so that the position of the C-terminal 9 residues is unknown (19). However, the structure suggests that these residues in each monomer might be oriented so that they could interact. In any event, it is clear that there are substantial differences in the positions of interaction among the various domains within the dimers of
CooA:ACO-SENSINGTRANSCRIPTIONALFACTOR
45
effector-bound CRP and effector-free CooA that reflect differences in activation state rather than primary protein sequence. 2. THE RELATIVEPOSITIONS OF THE C HELICESWITHINTHE EFFECTOR-BINDING DOMAINCHANGESUPONACTIVATION Besides the dramatic conformational differences noted above, there is a more subtle but very significant difference in the relative position of the two C helices with respect to each other in the two proteins (19). While each C helix of CooA can be aligned with its homolog helix in CRP, the relative positions of the two C helices in the two proteins is quite different (Fig. 2C; see color insert). There is a bending and rotation of the C helices with respect to each other such that (in effector-bound CRP as compared to effector-free CooA) they move apart at the bottom of the C helix with no relative movement at the fulcrum, represented by Leu-130 of CooA (Leu-134 of CRP). This has the effect of repositioning the two C helices such that they have different regions of closest contact in the two proteins. Indeed, along the entire length of the C helices in both CooA and CRP, there is a suboptimal leucine zipper and the repositioning upon activation has the effect of changing the regions where these two leucine zippers interact. The general issue is discussed further in Section V,C,1. 3. DIFFERENTIALCONTACTSMADE BYRESIDUES IN THE HINGE REGION OF CRP AND CooA The "hinge" region of CRP (Fig. 3), centered on Asp-138, is the boundary between the DNA-binding and the effector-binding domains of that protein (67). The fusion of the C and D helices (relative to their position in CRP) in form B of CooA directly affects the CooA region analogous to the hinge region of CRP, and it is not surprising that residues in the vicinity of that reorientation make rather different contacts in the two proteins. More interestingly, the specifics of these contacts appear to be important in stabilizing each of the respective conformations (19). Two residues are particularly relevant: Phe-132 of CooA (homologous to Phe-136 of CRP) and Arg-138 of CooA (homologous to Arg-142 in CRP). In CooA, Phe-132 interacts with the F helix of the DNAbinding domain, while Phe-136 of CRP makes contact with the 4/5 loop of the effector-binding domain. The region of interaction in the 4/5 loop is retained in CooA, suggesting that such an interaction might also be relevant to effectorbound CooA. The other residue, Arg-138, interacts with the 4/5 loop in CooA (of the other monomer), but its homolog in CRP, Arg-142, interacts with the DNA backbone. The actual functional importance of these interactions remains to be examined, but the very different role of these residues in the effectorbound and -free forms implies that their reorientation upon activation is central to the activation mechanism.
46
GARYP. ROBERTSET AL.
4. THE POSITION OF THE F HELICES THAT CONTACT DNA ARE DRAMATICALLYDIFFERENT IN EFFECTOR-FREE CooA AND EFFECTOR-BOUND CRP Another implication of the CooA structure (irrespective of which form of CooA is analyzed) is the fact that the F helices, which are necessary for direct interaction with the specific DNA target sequence in CRP and must certainly perform a similar role in CooA, are both oriented approximately 180 ° away from their position in CRP. That is, in effector-free CooA, they "face" the effector domain, which would make DNA interaction virtually impossible. Given that both CRP and CooA show a very low affinity for DNA in the absence of effector (17, 45), one might assume that this low affinity reflects a small fraction of the proteins that have reoriented their DNA-binding domains spontaneously and transiently without effector. It is a formal possibility, however, that low affinity for nonspecific DNA reflects interactions with other regions of the proteins. In any event, the positioning of the DNA-binding helices in effector-free CooA would seem to be of biological importance and will need to be further examined. 5. THE POSITIONS OF THE EFFECTOR-BINDING SITES ARE SIMILAR, BUT NOT IDENTICAL, IN CRP AND CooA
The heroes of CooA lie somewhat below the analogous position in CRP where cAMP binds (Fig. 2A, B), although we believe that it is important that both the heme and cAMP interact directly with portions of the C helix. A hypothesis for the role of effector-mediated perturbation of the C helices in activation is discussed in Section V,C. The heme occupies a pocket that does not exist in CRP, but seems to be accommodated by the deletion of eight residues in CooA relative to CRP (approximately residues 73-80 of CRP) (19). It also appears that there is a small rotation, as viewed down the axis of the C helix, of part of the effector domain away from the C helix to make space for the heine. From a protein evolutionary standpoint, it would be interesting to know if CooA evolved from a protein that lacked heine and therefore what structural changes were essential to create a pocket for heine binding. In addition to the structural differences between CooA and CRP noted above, a number of other differences can be detected between the two forms of these proteins (19) that are reflective of the substantial conformational differences. Some of these differences no doubt reflect critical differences between effector-bound and effector-free CRP/CooA, while others merely reflect differences between CooA and CRP themselves. Further experimentation will be necessary to distinguish between these two possibilities for any observed difference in the two structures in hand. The structural differences also fail to identify the mechanism by which binding ofeffector triggers the change between the two conformations, and the hypotheses for this trigger are described in Section V,C.
CooA: A CO-SENSING TRANSCRIPTIONAL FACTOR
47
B. The Heme Region of CooA The structure of reduced, effector-free CooA confirmed the role of His-77 as one ligand to the ferrous heme iron, as suggested by previous mutagenic and spectroscopic studies (38, 54). Another aspect of the structure of this side of the heme was surprising, however, in that Cys-75, which certainly replaces His-77 as the ligand in the oxidized form (69), is approximately 4.8 ~_ from the heme Fe (Fig. 2D; see color insert). This implies that there must be a conformational change in the protein, a movement of the heme, or both, in the course of the reversible oxidation and reduction of the berne. Under the assumption that there is at least some heine movement, the present structure predicts that the heme would "move into" the effector domain upon oxidation. Because the heine remains low-spin, six-coordinate under both conditions, this implies that the Pro-2 ligand on the other side of the heme is either extremely flexible in its position or that a ligand switch takes places on that side as well upon oxidation/reduction. Currently, we favor the former hypothesis because of the results described in Section V,B,3. His-77 appears to be relatively free of contacts with other residues in the reduced form, with the exception of Cys-75 and Asn-42; the role of this latter residue in activation of CooA is unknown, but its location adjacent to His-77 suggests that it is likely to be of some importance. The axial ligand on the other side of the heme in reduced CooA was one of the major surprises revealed by the structure (19). That ligand is Pro-2, the N-terminal residue from the other monomer of the dimer. Proline has not been previously identified as a heine ligand, because the secondary amine would be unavailable to serve as a ligand when involved in a typical peptide bond. However, as the N-terminal residue (following posttranslational removal of Met-1), the N ofproline appears to serve as a reasonably strong-field ligand, consistent with the observation that known strong-field small-molecule ligands such as cyanide are unable to displace it. The spectroscopies of MCD, RR, and EPR were not able to distinguish the Pro ligand from the very common His ligation, suggesting that Pro appears as a "neutral nitrogen donor" in such spectroscopic analysis (55, 70, 71 ). As described below, while Pro is a highly unusual ligand, it does not appear essential for CO responsiveness of CooA, and the precise role of this region of CooA for a response to CO needs to be determined (72). The exchange of N-terminal arms between subunits of the CooA dimer is an example of"domain swapping" whereby subunits of an oligomeric protein are linked together by the exchanged domains. The presence of a proline at the position where the intercalating domain extends from its monomer, presumably Pro-14 in CooA, is often crucial for proper configuration of the exchanged arm (73). It is probably of functional importance that the N terminus of CooA, ending with Pro-2, appears to be relatively unconstrained in the structure. This lack of constraint might provide the flexibility necessary to allow the
48
GARYE ROBERTSET AL.
movement of the heme that almost certainly takes place during oxidation reduction (19, 38, 54). The importance of these ligands for CooA function has been explored by a variety of mutagenic approaches and analyses, which will be briefly summarized. 1. CYS-75 Is NOT IMPORTANTFOR THE RESPONSE TO CO Biochemical and spectroscopic analyses of variants altered at position 75 have shown that Cys-75 (1) is important for heme stability in oxidized CooA (54); (2) is not critical for a functional response to CO, as C75S and C75A CooA have significant activity in vivo (38, 54); (3) is presumably important for setting the proper redox poise of CooA; and (4) can also be partially replaced by an unknown adventitious ligand in the oxidized state of variants that lack Cys-75 (54). Interestingly, Cys, Ser, and Ala appear to be the only acceptable residues at position 75 that yield functional CooA, as these were the only substitutions found when the codon was randomized and the resultant library was screened for a response to CO in vivo (M. Conrad and G. P. Roberts, unpublished data). The similar size of these residues suggests that this might be the critical requirement, and preliminary results suggest that larger residues at this position might affect protein stability, possibly through effects on heine stability. Besides serving as an axial ligand for oxidized CooA, the residue at position 75 therefore appears to have some additional structural constraints. 2. HIS-77 Is CRITICAL FOR ACTIVATION OF CooA IN RESPONSE TO CO CooA variants altered at position 77 (1) accumulate berne-containing CooA that is reasonably stable (38, 54); (2) bind CO but are unable to activate transcription in vivo and are unable to bind DNA in vitro (using an assay of fluorescence polarization with tagged target DNA; M. V. Thorsteinsson et al., unpublished data); (3) are perturbed in their UV/Vis spectra in the reduced but not the oxidized form (38, 54); (4) are able to bind cyanide in the reduced form (in contrast to all forms of wild-type CooA, which fail to bind cyanide) (74); (5) typically bind cyanide with positive cooperativity, although the nature of the cooperativity depends on the residue at position 77, as well as on the small molecule ligand (Ref. 74 and M. V. Thorsteinsson et al., unpublished data); (6) are missing a spectroscopic species detected in the oxidized form of wild-type CooA (when Cys-75 is the ligand), consistent with an interaction between His-77 and Cys-75 in wild-type CooA (54); (7) are perturbed in their redox-mediated ligand switch, where they appear to be significantly more difficult to reduce (54, 70); and (8) accumulate to varying degrees as six-coordinate, low-spin species in the reduced form, which implies the presence of at least one adventitious ligand that can replace His-77 (54, 70, 74).
CooA:A CO-SENSINGTRANSCRIPTIONALFACTOR
49
These various results indicate that His-77 is important for the proper poise of the redox-mediated ligand switch and absolutely critical for the conformational change that alows DNA binding in response to CO binding. As it has very recently been shown that Pro-2 is displaced by CO binding in CooA (73a) the importance of His-77 is presumably to properly position the heme for which the trans ligand has been released by CO. Note that variants lacking His at position 77 are sixcoordinate, low-spin to varying degrees when reduced (70, 71), so there must be an adventitious ligand on that side of the heme that can replace the missing His-77. Although this adventious ligand is apparently strong enough to create a low-spin heme, it must position the heine incorrectly upon CO binding, and this mis-positioning presumably is critical for activation, but of modest influence on eooperativity. The identity of the adventitious ligand is presently unknown. 3. PRO-2 IS OPTIMAL, BUT NOT CRITICAL, FOR HEME ACCUMULATION AND THE RESPONSE TO CO Given the uniqueness of Pro-2 as a heme ligand, we expected that this residue would be central to CooA function. However, variants altered in the Pro-2 region (including those with longer and shorter N termini, as well as with different terminal residues) often accumulate reasonable amounts ofheme-containing CooA that responds to CO both in vivo and in vitro (72). These variants appear generally normal in UV/Vis spectra for both the reduced and reduced + CO forms, suggesting that some adventitious ligand is replacing Pro-2 in at least the former case, when it is known to serve as a heine ligand. However, these variants are significantly altered in the oxidized form, where they have a mix of five- and six-coordinate species. When one example, P2Y, was purified and characterized, it displayed an affinity for CO that was slightly greater than that of wild-type CooA and a modest decrease (10-fold) in affinity for DNA in vitro in the presence of CO (72). These results support the following conclusions about Pro-2 in CooA. (1) There is apparently an adventitious ligand that can replace Pro-2 in the reduced form of CooA. The structure does not suggest an obvious candidate, other than the N terminus created by the replacement of Pro-2. This possibility is attractive, since it should have properties generally similar to Pro and would be present in all variants. However, a deletion of residues Pro-3 and Arg-4 has approximately the same properties as other Pro-2 region variants (R. L. Kerby et al., unpublished data) and it is unclear if the N terminal region is sufficiently flexible to reach the heine iron when shortened by two residues. (2) Pro-2 is likely to be the ligand in the oxidized form of CooA, as the spectroscopy of this form is severely perturbed in Pro-2 variants. (3) CooA can function with a wide variety of residues at the N terminus, consistent with the great flexibility of the protein in the vicinity of the heme and the relative unimportance of Pro-2 in the CO-activated state.
50
GARYEROBERTSETAL.
Surprisingly, given the apparent lack of a critical importance of Pro-2 to activation, some other residues near the N terminus seem to be more critical for the accumulation of CO responsive CooA in vivo (72). Arg-4 is of some importance for accumulation of heme-containing CooA, and because its amino group is positioned very close to one of the proprionate groups of the heme, it is possible that this interaction is important for heme stability (19). Phe-5, which fills a hydrophobic pocket near the bottom of the effector-binding domain, is also necessary for CooA accumulation, and this is probably due to effects on protein stability. Finally, Ash-6 lies at the end of an a helix and forms a hydrogen bond with Asn-9, which might be the basis for its importance. The effects of the alteration of Pro-2 on activation of CooA by CO are particularly surprising in light of the demonstration that it is the ligand displaced by CO (73a). This suggests that either the displaced Pro-2 has relatively little role in activation or else some other residue can serve that role satisfactorily. The obvious model is therefore that the precise positioning of the heme by His-77 ligation is critical for activation. We therefore favor a hypothesis that the removal of the Pro-2 "tether" frees the heme to interact with the C helices and affects the relative positioning of these helices, and that this serves as at least one portion of the activation mechanism.
C. Model for Activation of CooA by CO The primary and secondary structures of CooA and CRP are similar enough to provide a basis for hypotheses about the nature of activation of both proteins by their effectors. It is highly likely, for example, that CooA assumes a structure upon CO binding that is rather similar to that of cAMP-bound CRP, as CooA must interact both with a similar DNA sequence and with several similar regions of RNA polymerase to stimulate transcription activation (see Section IX). For both proteins, it is easiest to consider only two states, active and inactive, in an equilibrium. Without effector, this equilibrium is substantially toward the inactive form, as evidenced by the very low, but detectable, activity of both proteins in the absence of effector (17, 45). The binding of effector shifts that equilibrium either by destabilizing the inactive form, stabilizing the active form, or both. An important fact that must be addressed in any hypothesis of activation is that the DNA-binding domains in both proteins lie far from the site of effector binding so that there cannot be a direct interaction between the effector-binding site and the DNA-binding domain. The comparison of the structures of CRP and CooA suggests that the C helix positioning is central to the response to effector for both proteins. Of the differences between the CRP and CooA structures noted in Section V,A, the difference in positioning of the C helices provides the most obvious direct link between the sites of effector binding and the rest of the proteins. The general hypothesis is therefore that effector binding causes a shift in the relative position
CooA:A CO-SENSINGTRANSCRIPTIONALFACTOR
51
of the C helices, which changes in regions of closest proximity along the leucine zipper that forms the dimer interface. This change in C helix position would potentiate the changes (noted in Section V,A) in the hinge region, eliminating some of the stabilizing factors in the inactive form of the proteins and allowing new interactions that stabilize the active form. This is actually a slightly more specific version of the model of activation proposed previously for CRP in which cAMP binding was proposed to change the "alignment" of the effector-binding domains, although the nature of the effector-free form of CRP could not be guessed from the available data (45). This hypothesis for activation appears to be consistent with the available data on CRP, particularly the data on effector-independent variants (so-called CRP* variants) and those variants altered for cooperative binding of effector. Mutations leading to the CRP* phenotype fall in two general regions (45, 46): in the vicinity of the hinge itself and in the vicinity of the cAMP binding site. Those in the former category presumably "bypass" the normal activation process by either stabilizing the active form or destabilizing the inactive form of the protein, which is consistent with the proposed model. The class of CRP* mutants affecting the cAMP binding site is exemplified by changes in Thr-127 and Ser-128, which affect both the cooperativity of cAMP binding and effector-independent activity (63, 75, 76). It is striking that many of these variants do not significantly affect cAMP affinity, so that the position of these residues at the cAMP-binding site might be coincidental to the phenotypes observed. Instead, it is likely that these changes affect the quality of the leucine zipper in this region. The T127L substitution in CRP is particularly effective at altering the phenotype and also improves the predicted leucine zipper motif. In the case of CooA, we have also found that altering the leucine zipper in the 121-126 region of CooA (homologous to the 125-130 region of CRP) toward that ofa"stronger" leucine zipper has the effect of creating effector-independent CooA variants (termed CooA* by analogy), consistent with the above hypothesis (R. L. Kerby, unpublished data). Indeed, one of the strongest CooA* variants we have obtained is a double mutant, C123L/M 124T, which affects the homologs of CRP residues 127 and 128. This variant has the property of being active under all conditions (oxidized, reduced, and reduced plus CO), although the precise level of activity does show some dependence on the condition examined. The number of well-characterized CooA* variants is low and does not provide a critical test of the hypothesis. Unfortunately, the putative CooA* variant M131L, which has been published several times (38, 77, 78), may not be informative. This variant, which affects the hinge region, has been proposed to be approximately 18 times as active as wild-type CooA in the presence of CO, which would be extraordinary given the apparently efficient activation of wild-type CooA by CO. However, we have created the identical substitution in our lab, and it has the more reasonable phenotype of only a modest CO activation (R. L. Kerby and G. P. Roberts,
52
GARYP. ROBERTSET AL.
unpublished data), consistent with other variants at that position that have also been published (77). We assume that the causative mutation for the high levels of activity is not the M 131L substitution, but rather a mutation in the lacZ reporter gene or elsewhere in the reporter strain. In general then, what data exist for CooA are consistent with the proposed activation mechanism. It is also interesting that FNR has some phenotypically interesting mutations in the C helix region as well. FNR apparently undergoes a monomer--dimer transition in the course of activation, reflecting the synthesis of Fe4S4 clusters in the effector-binding domain (47, 48); thus, its activation mechanism might be expected to be fundamentally unlike that of CRP and CooA. Nevertheless, a D154A substitution (where D154 is the homolog of T127 of CRP and C123 of CooA) or a D154V substitution creates effector-independent variants, in this case independent of the need for reducing conditions and the resultant Fe4S4 clusters (13, 48, 79). While an Ala would not be expected to create a strong leucine zipper, the Asp found in wild-type FNR should be highly deleterious for such a zipper. How might CO binding to the heine of CooA cause this perturbation of the C helices? Our best current hypothesis is that CO binding allows a repositioning of the heine and this repositioning results in the movement of the bottoms of the C helix away from each other, which in turn potentiates the various conformational changes noted above. The specifics of the role of heme positioning in C helix movement is a major issue in understanding the behavior of this protein.
VI. CooA as a Redox Sensor CooA will not bind CO until the heme is reduced, and the midpoint potential of that reduction has been reported to be -320 to -260 mV for reduction and oxidation, respectively (43). As described in Section I1, this potential makes physiological sense in light of the properties of CooS, the CODH of the system. CooA sets the proper redox poise in part through an unusual ligand switch, whereby Cys-75 is replaced as a ligand by His-77. While Cys-75 as thiolate is presumably a better ligand to the oxidized heme than is the imidazole of His-77, and His-77 should be a better heme ligand in the reduced state than is the thiolate of Cys-75, the details of this ligand-switch mechanism are poorly understood. Irrespective of these details, however, the mechanism certainly must be different from the O~/redox sensors that have been found, such as FNR (13, 47, 48, 80), FixL (10, 56, 57), OxyR (81), and DOS [a putative 02 sensor from E. coli, (•4)], as these either do not involve a heme or, in the case of FixL and DOS, actually sense 02 by directly binding it. In this article we have referred to "redox-sensing" somewhat loosely, implying rather more than we actually know. Specifically, the "signal molecule"
CooA:ACO-SENSINGTRANSCRIPTIONALFACTOR
53
that is actually sensed by CooA in the cell is unknown; thus while the regulation "reflects" the redox state of the cell, it is highly likely that a specific small molecule is actually sensed. There certainly are small molecule redox couples, such as NAD/NADH, with midpoint potentials that might be appropriate for CooA. We cannot rule out some sort of heine reductase intermediary, although the fact that no such factor is encoded by the coo regulon of R. rT~brum, as well as the functionality of the coo system when moved to other organisms such as R. sphaeroides and E. coli (82), suggests that this is less likely. In vitro, CooA is readily reduced by small molecules such as methyl viologen, sodium dithionite, and titanium citrate, and is oxidized by potassium ferricyanide; however, physiologically significant small molecules need to be assessed for their ability to reduce CooA in vitro. Although the precise mechanism of redox response of wild-type CooA is unclear, some very interesting variants with different responses to oxidizing conditions have been found. Perhaps the most striking is M124R (M124K being generally rather similar). Cells containing M124R CooA behave like cells with wild-type CooA under anaerobic conditions (no CooA activity) and anaerobic conditions with CO (high CooA activity) (R. L. Kerby, M. V. Thorsteinsson, and G. P. Roberts, unpublished data). However, when grown aerobically (and without CO), cells with M124R CooA display significant levels of CooA activity, in striking contrast to cells with wild-type CooA. Consistent with these observations, purified reduced M124R CooA shows normal UV/Vis spectra in the presence and absence of CO and normal DNA binding only in the presence of CO. When oxidized, however, purified M124R CooA binds DNA in vitro in the absence of CO, and UV/Vis and EPR analyses indicate that there is a mix of five- and six-coordinate species in the oxidized state. Because this proportion changes toward the six-coordinate form at elevated pH, it has been possible to show that the five-coordinate form is the source of the activity. There are no obvious hints from the crystal structure why this substitution would have these properties; but irrespective of mechanism, M124R CooA is, in a sense, a sensor of"oxidizing conditions." The fact that CooA senses redox by a mechanism unlike that of other studied sensors, coupled with the availability of CooA variants altered in redox sensing, makes this a highly interesting aspect of CooA function that merits substantially more attention in the future. This variant is suggestive of the fascinating range of behaviors that CooA variants will be capable of and whose eventual analysis will address larger issues of heine protein function.
VII. CooA as a CO Sensor As noted previously, wild-type CooA responds exclusively to CO, as other small molecules (such as cyanide, azide, and imidazole) either fail to bind to the
54
GARYE ROBERTSET AL.
heme (54) or, in the case of NO, displace both heine ligands and fail to cause an active conformation of CooA (59). The failure of NO, which displaces both protein ligands to the berne, to activate CooA is consistent with the hypothesis that the precise positioning of the heine is essential to the activation process, and certainly the protein ligand that remains bound trans to the CO would be expected to be important for this positioning. The failure of CooA to bind other small molecules might reflect any of several factors: (1) The small molecules are not of sufficient field strength to replace the protein ligands; (2) the small molecules might be stericallyprecluded from reaching or binding to the heme because of a restricted heme pocket; or (3) charge repulsion might restrict the ability of charged small molecules to bind. The first possibility appears to play a role, as CO and NO, the only two small molecules that bind to wild-type CooA, have greater field strength than any of the other small molecules. However, the heme appears to be generally exposed, suggesting that steric hindrance is not likely to be a major factor in specificity. Similarly, the only obvious charges in the vicinity of the heme, other than the heine proprionates, are the negative charge provided by Cys-75 on one side and the positive charge provided by Arg-4 on the other side. Only the former would seem to be a candidate for repelling cyanide or azide. Obviously, an understanding of the structure of the CO-bound form of CooA will be necessary to further clarify the basis for CO specificity. The mere fact that wild-type CooA does not respond to other effectors does not preclude the the possibility of CooA variants with altered specificity. For example, CooA variants that lack His-77, a normal ligand in the reduced form, are now capable of binding cyanide, while retaining their ability to bind CO. H77Y also shows an extremely low level of activation by cyanide in the very sensitive in vivo assay, although efforts to detect this low activity in vitro have been unsuccessful (71). Nevertheless, the results are encouraging that the combination of targeted mutagenesis and phenotypic screens will allow the identification of CooA variants with altered specificity, which will provide important information concerning the molecular basis for ligand specificity.
VIII. Cooperativily of Ugand Binding It has been difficult to determine the precise cooperativity of wild-type CooA in CO binding: The very low solubility of CO combined with the high affinity of CO for CooA makes the analysis of free CO, and hence the precise nature of cooperativity, technically difficult. However, our present analysis of the data is most consistent with the hypothesis that wild-type CooA is positively cooperative for binding CO, and there is no doubt that CooA variants
CooA:A CO-SENSING TRANSCRIPTIONALFACTOR
55
altered at position 77 are perturbed in that cooperativity (M. V. Thorsteinsson et al., unpublished data). The molecular basis for the implied communication between the two heme molecules has not been identified, but the examination of the crystal structure (Fig. 2A) suggests that the likely pathway involves the lower portions of the C helices of the two protein monomers, which are adjacent to and between the hemes. It seems to be a reasonable hypothesis that CO binding to one heme would affect its positioning relative to the C helix of the other monomer and that this perturbation of the C helix could be transmitted to the heme of the other monomer. While this general scheme is very much like that proposed for activation of CooA by CO, there must be some important differences for the following reason. CooA variants altered at position 77, and therefore lacking the normal His-77 ligand to the reduced form, are still able to bind CO, 'although the degree of cooperativity and CO affinity depends on the nature of the residue at position 77. In addition, these variants all show no detectable activation by CO, demonstrating that activation and cooperativity are distinct. Cooperative binding of CO is not sufficient to support activation and it remains unclear whether or not it is necessary. CooA variants altered at position 77 are also able to hind cyanide and the majority are generally able to do so with positive cooperativity (74). Interestingly, there is no clear pattern among the position 77 variants in terms of their degree of cooperativity in binding either CO and cyanide (74; M. V. Thorsteinsson et al., unpublished data). This indicates that the precise degree of cooperativity is a function of both the small molecule and the residue at position 77. Although the exact mechanism of cooperative CO binding by CooA remains to be determined, it is useful to compare the situation to that of CRE There has been some debate on the matter, but it is generally agreed that CRP is negatively cooperative for binding its effector, cAMP, under physiological conditions (60, 61). Interestingly, the form of CRP with a single molecule of cAMP bound has a higher affinity for DNA, suggesting that it is the physiologically significant form (60). The physiological significance of CRP with two cAMP molecules bound is unclear. We currently do not know if CooA with a single CO molecule bound is active in DNA binding. Either positive cooperativity or activation by a single effector molecule is easy to rationalize for a system that is maximized for sensitivity to the effector. It is more difficult to understand a role for negative cooperativity in this sort of sensing system, unless the doubly bound form is coincidentally less active and the cooperativity is simply a mechanism to reduce the likelihood of that species being present. As the mechanisms of cooperativity in both proteins is unraveled, the comparison between the two homologs, with effector responses of opposite cooperativity, will certainly be informative.
56
GARYE ROBERTSETAL.
IX. Transcriptional Activation by CooA The transcription start sites, as well as the sites of CooA binding, have been determined for the two naturally occurring CooA-regulated promoters in R. rubrum (32, 83). The relative positioning of the sites makes it clear that these are analogous to the Class II sites used by CRP in which the CRP binding site overlaps with the - 3 5 region to stabilize RNA polymerase (RNAP) binding (Fig. 4). Although the nature of the RNAP that CooA utilizes in R. rubrum is unknown, CooA has been heterologously expressed in both Rhodobacter sphaeroides and E. coli, where it supports redox- and CO-dependent gene expression from one of its natural promoters (82). Purified CooA has also been shown to function in in vitro transcription assays with purified Ecr 7° ofE. coli, as well as the RNAP with the major housekeeping sigma from R. sphaeroides (82), which has allowed significant analysis of its behavior, albeit in a heterologous system. While the results might not represent the function of CooA in its normal host, R. rubrum, they certainly indicate the capabilities of this CooA in transcription activation. In these analyses, CooA was shown to be necessary and sufficient for redox- and CO-dependent transcription. This activation required the carboxyl terminal domain of the ot subunit of RNAP for transcription activation. Footprinting analysis showed that CooA facilitated the DNA binding of normal E~r7°, but not that of Ecr 7° lacking the carboxyl terminus of the ~ subunit. An analysis of the residues of the ot subunit involved in interaction with CooA suggested that they are similar, but not identical, to the residues that interact with CRP at Class II promoters (82, 84, 85). It therefore appears that despite the relatively low level of identity between CooA and CRP, CooA behaves rather similarly to CRP in its interaction with RNA polymerase at a Class II promoter in this heterologous system.
FIG. 4. AcartoonofCooAbindingatoneofits twonaturalpromotersinR, rubrum, aCTD and c~NTD refer to the carboxyl and amino domains of the ~ subunit of RNA polymerase, respectively. AR1 is the surface of CooA that interacts with the ~CTD, AR2 interacts with c~NTD, and AR3 interacts with the ~r subunit. Adapted from S. Busby and R. H. Ebright, J. Mol. Biol. 292, 199-213 (1999) by permission of the publisher Academic Press London.
CooA:A CO-SENSINGTRANSCRIPTIONALFACTOR
57
Another aspect of transcription activation is the nature and role of the regions on the transcriptional activators that interact with polymerase; these are termed "activating regions," and abbreviated AR (Fig. 3). The AR regions have been defined through a variety of analyses of CRP (86) and FNR (87), which have generally similar but not identical properties in terms of interaction with RNA polymerase. For Class II promoters, where the activator protein binds immediately adjacent to the RNA polymerase, AR1 is found on the upstream monomer and interacts with the C-terminal domain of the ot subunit of RNAP, which extends away from the rest of RNAP and reaches around the activator, to contact both the activator and, typically, the adjacent DNA (Fig. 4); this interaction increases the initial binding of RNAP to DNA. In CRP, AR1 is represented by residues 155-164 (the region between the D and E helices in Fig. 3), which lie in the DNA-binding domain; in FNR, this region, together with a nearby region of the effector-binding domain, has been implicated in this type of interaction (87, 88). AR2 is found on the downstream monomer, near the bottom of the effector-binding domain, and interacts with the N-terminal domain of the ot subunit of RNAP to facilitate the isomerization from the closed to the open complex (86); it is unclear if AR2 is functional in FNR. Finally, AR3, which falls at the tip of the 4/5 loop (Fig. 3), is defined as a region that interacts with ~r7° and also facilitates isomerization (89). It is functional in wild-type FNR, but apparently not in wild-type CRP, although such a region can be "created" by certain mutations in the vicinity of the tip of the 4/5 loop. Before the structure of CooA was solved, comparison of the primary sequences of CRP and CooA suggested that only AR3 might be conserved (37, 82), although the requirement for the C-terminal domain of ~ noted above was strongly suggestive of the presence of an AR1 region as well (82). The comparison of the three-dimensional structures of CRP and CooA suggests that there are similar residues in roughly similar positions for both AR2 and AR3. Consistent with this view, CooA variants have been found at positions Lys-26, and Thr-97 (AR2, analogous to CRP residues Lys-22 and Lys-101, respectively) and Val-57 (AR3, analogous to CRP residue Lys-52) that are inactive in promoting expression in vivo, yet retain their normal affinity for DNA upon purification, which is the expected phenotype for variants with defects in AR regions (j. LeDuc and G. P. Roberts, unpublished data). The AR1 region of the CooA structure remains poorly defined and neither gain-of-function nor loss-of-function variants have been assigned to that region to date, although the demonstration of a requirement for the C terminus of ot is strong evidence of such a region in CooA. Based on these results, it is our working hypothesis that CooA has regions analogous to all three AR regions of CRP, although the details of their role might well be mechanistically different.
58
GARYE ROBERTS ET AL.
X. DNA Recognition Properties of CooA The D N A sequence that is bound by CooA is represented by the four "half-sites" found adjacent to the two known CooA-regulated promoters in R. rubrum. Although a "consensus" sequence for CooA binding can be suggested, it is obviously based on a very small data set. In C R P and FNR, the n u m b e r of biologically relevant promoters regulated by each protein has allowed a very good indication of a consensus for each, although natural binding sites rarely come particularly close to those consensus sequences, presumably because such a high-affinity site would interfere with proper regulation (86, 90).
-A-TGTGA
A
...... TCACA-T-
CRP
A-A-TTGAT--A-ATCAAT---
FNR
-A-TGTCA
COOA
...... CGACA-A-
T
B
G'
G184 R180 L
K188~
T
G,A E l 8 1 % o R P / V I
T
-
R185 M~89 ~T
T "..... $212~ L T
c Q178 CooA A _
T18~ ~' 2 ~
G216.f/~
G*.-..E209-~ A
• 8~ \
FNR
AT 1 ~ 1 ~
~)-i L
FIG. 5. (A) CooA, CRP, and FNR "consensus target sequences"; as noted in the text, the CooA target is poorly defined because there are only two known binding sites. (B) Helical wheel presentations of CRP, FNR, and CooA. These are views down the F helix of each protein, with the numbers within the circles indicating the relative position in the helix and the numbers on the outside indicating the specific protein residues. Known or proposed interactions between residues and bases are indicated. Adapted from Ref. 45.
CooA:A CO-SENSINGTRANSCRIPTIONALFACTOR
59
The data in Fig. 5 indicate the "consensus" sites for the three proteins and a "helical wheel" representation of the relative position of the residues in the F helix of each protein. Although there has been some mutagenesis of the F helix of CRP to indicate some of the critical residues for interaction, the "rules" for nucleotide recognition for such a helix have not been worked out (45). Two observations can be made, however. Arg-180 of CRP (at helical position 1) has been proposed to make specific contact with the G in the second position of the CRP consensus (91, 92). Both this residue (Arg-177) and the nucleotide have been conserved in CooA, but neither is present in FNR, consistent with the residue's proposed recognition role. Similarly, the second residue of CRP and FNR has been proposed to be critical for the recognition of a G (the fourth position in the CRP consensus, and the third in the FNR consensus), and CooA has both altered the residue and appears to utilize a C at that position in its consensus (83). As suggested by Fig. 5B, it remains unclear which residues specify several of the highly conserved bases in the CRP and FNR target sequences, and it will be informative to have a more complete analysis of the residues responsible for specific DNA binding in this protein family.
XI. Future Direction and Open Questions The analysis of CooA has already revealed some very interesting insights into important biological questions, such as heme ligation strategies, redox sensing, cooperativity, and the nature of activation in the CRP superfamily of proteins. Nevertheless, a number of open questions remain whose answers will further our understanding of the above issues and address others. (1) What are the structures of oxidized and the CO-bound forms of CooA? The latter structure can be guessed at by analogy with CRP structures, but the former structure, which would provide more details about the ligand switch, is completely unknown. (2) Is the hypothesis about the role of C-helix reorientation upon activation correct? What is the direct cause of the C helix reorientation upon CO binding? Is it positioning of the heme itself? (3) What is the molecular basis for cooperativity in CO binding? (4) Is the form of CooA with a single bound CO molecule active for DNA binding? (5) What is the molecular basis for the specificity of CooA for activation only by CO? Can CooA variants be identified that have dramatic activation in response to cyanide or NO? (6) What are the adventitious ligands that replace Cys-75, His-77, and Pro-2 when these ligands have been mutationally eliminated? These results would tell us something about the flexibility of the heme region. Elimination of these adventitious ligands would also allow us to draw better conclusions about the importance of the normal ligands. (7) How important are the features that have been proposed to stabilize the effector-free and effector-bound forms of CooA? In other words, the recent crystal structure
60
GARY E ROBERTS ET AL.
o f C o o A allowed a proposal o f a n u m b e r of different interactions in the hinge region b e t w e e n the two forms o f CooA, but the significance o f these has not b e e n tested. (8) Can we develop a b e t t e r u n d e r s t a n d i n g o f the molecular basis o f the redox poise of C o o A and the role o f the ligand switch in that poise? (9) W h a t are the implications of these various features of C o o A on the response of R. r u b r u m to C O ? T h e bulk o f the analysis o f CooA has b e e n p e r f o r m e d either in E. coli or with purified p r o t e i n in vitro, and it will b e interesting to reexamine selected CooA variants in the normal host at physiologically normal levels o f CooA.
ACKNOWLEDGMENTS This work was supported by the College of Agricultural and Life Sciences, University of Wisconsin-Madison; by support from NIH grant GM53228 (to G. P. R), NIH NRSA Fellowship (to M. V. T.), NSF grant MCB9807798 (to T. L. P.), and USDA Fellowship USDA-98353056549 (to W. N. L.). The authors thank Hwan Youn, Yiping He, Mary Chamberlain Conrad, and Jason LeDuc for permission to include unpublished data. Hwan Youn, Kerridwen McNamara, Mary Chamberlain Conrad, and Jason LeDuc also provided helpful suggestions and comments on the text.
REFERENCES 1. D. S. Bredt and S, H. Snyder, Annu. Rev. Biochem. 63, 175-195 (1994). 2. D. L. Garbers and D. G. Lowe,J. Biol. Chem. 269, 30741-30744 (1994). 3. G. Deinum, J. R. Stone, G. T. Babcock, and M. A. Marietta, Biochemistry 35, 1540-1547
(1996). 4. L.J. Ignarro, j. N. Degnan, W. H. Baricos, P. J. Kadowitz, and M. S. Wolin, Biochim. Biophys. Acta 718, 49-59 (1982). 5. E. A. Dierks, S. Hu, K. M. Vogel, A. E. Yu, T. G. Spiro, and J. N. Burstyn, J. Am. Chem. Soc. 119, 7316-7323 (1997). 6. I. E. Tosques, J. Shi, and J. P. Shapleigh,J. BacterioI. 178, 4958-4964 (1996). 7. A.V. Kwiatkowski,W. P. Laratta, A. Toffanin, and J. P. Shapleigh,J. Bacteriol. 179, 5618-5620 (1997). 8. M. A. Gilles-Gonzalez, G. Gonza]ez, and M. P. Perutz, Biochemistry 34, 232-236 (1995). 9. K. R. Rodgers, G. S. Lukat-Rodgers, and J. A. Brown, Biochemistry 65, 9539-9548 (1996). 10. E. K. Monson, G. S. Ditta, and D. R. Helinski, J. Biol. Chem. 270, 5243-5250 (1995). 11. D.O. Gostick, J. Green, A. S. Irvine, M. J. Gasson, andJ. R. Guest, Microbiology 144, 705-717 (1998). 12. N. Hasegawa, H. Arai, and Y. Igarashi, FEMS Micro. Lett. 166, 213-217 (1998). 13. G. Unden and J. Schirawski,Mol. Microbiol. 25, 205-210 (1997). 14. V. M. Delgado-Nixon, G. Gonza]ez, and M. A. Giles-Gonzalez, Biochemistry 39, 2685-2691 (2000). 15. P. M. Vignais, B. Dimon, N. A. Zorin, A. Colbeau, and S. Elsen, J. Bacteriol. 179, 290-292
(1997).
CooA: A CO-SENSING TRANSCRIPTIONAL FACTOR
61
16. A. Verma, D. J. Hirsch, C. E. Glatt, G. V: Ronnett, and S. H. Snyder, Science 259, 381~384
(1993). 17. D. Shelver, R. L. Kerby, Y. He, and G. P. Roberts, Proc. Natl. Aead. Sci. U.S.A. 94, 11216-
11220 (1997). 18. S. Aono, H. Nakajima, K. Saito, and M. Okada, Biochem. Biophys. Res. Commun. 228, 752-756
(1996). 19. W. N. Lanzilotta, D. J. Sehuller, M. V. Thorsteinsson, R. L. Kerby, G. P Roberts, and T. L. Poulos, Nat. Struct. Biol. 7, 876480 (2000). 20. R. Conrad, Adv. Microbial Ecol. 10, 231-283 (1988). 21. J. w. Swinnerton, v. j. Linnenbom, and R. A. Lamontagne, Science 167, 984-986 (1970). 22. Y. Zuo and R. D. Jones, Water Res. 31,850-858 (1997). 23. R. D. Jones, Deep-Sea Res. 38, 625-635 (1991). 24. J. G. Ferry, Annu. Bey. MicrobioI. 49, 305-33 (1995). 25. B. Santiago, U. Sehiibel, C. Egelseer, and O. Meyer, Gene 236, 115-124 (1999). 26. R. L. Uffen, Proc. Natl. Acad. Sci. U.S.A. 73, 3298-3302 (1976). 27. R. L. Kerby, S. S. Hong, S. A. Ensign, L. J. Coppoc, P. w. Ludden, and G. P. Roberts, J. Bacteriol. 174, 5284~5294 (1992). 28. V. A. Svetlichny, T. G. Sokolova, M. Gerhardt, M. Ringpfeil, N. A. Kistrikina, and G. A. Zavarzin, System. Appl. Microbiol. 14, 254-260 (1991). 29. D. Gadkari, G. MSrsdirf, and O. Meyer, J. Bacteriol. 174, 6840-6843 (1992). 30. R. Conrad, Microbiol. Rev. 60, 609-640 (1996). 31. R. L. Kerby, P. W. Ludden, and G. P. Roberts, J. Bacteriol. 177, 2241-2244 (1995). 32. J. D. Fox, Y. He, D. Shelver, G. P. Roberts, and P. V~ Ludden, J. Bacteriol. 178, 6200-6208
(1996). 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.
47. 48. 49. 50. 51. 52. 53.
R. L. Kerby, P. W. Ludden, and G. P. Roberts,J. Bacteriol. 179, 2259-2266 (1997). J. D. Fox, R. L. Kerby, G. P. Roberts, and P. W. Ludden, J. Bacteriol. 178, 151,5-1524 (1996). R. K. Watt and P. W. Ludden,J. Biol. Chem. 273, 10019-10025 (1998). R. K. Watt and P. W. Ludden, J. Bacteriol. 181, 4554-4560 (1999). D. Shelver, R. L. Kerby, Y. He, and G. P. Roberts,J. Bacteriol. 177, 2151-2156 (1995). S. Aono, K. Ohkubo, T. Matsuo, and H. Nakajima, J. Biol. Chem. 273, 25757-25764 (1998). M. R. Hyman, S. A. Ensign, D. J. Arp, and P. W, Ludden, Biochemistry 28, 6821-6826 (1989). S. A. Ensign, Biochemistry 34, 5372~5381 (1995). D. Bonam, L. Lehman, G. P. Roberts, and P. W, Ludden,J. Bacteriol. 171, 3102--3107 (1989). S. A. Ensign, M. R. Hyman, and P. W. Ludden, Biochemistry 28, 4973-4979 (1989). H. Nakajima and S. Aono, Chem. Lett., 1233-1234 (1999). R.J. Rohlfs, A. J. Matthews, T. E. Carver, J. s. Olson, B. A. Springer, K. D. Egeberg, and S. G. Sligar, J. Biol. Chem. 265, 3168-3176 (1990). A. Kolb, S. Busby, H. Buc, S. Garges, and S. Adhya, Annu. Rev. Biochem. 62, 749-795 (1993). S. Adhya, S. Ryu, and S. Garges, in "Subcellular Biochemistry. Vol. 24: Proteins: Structure, Function, and Engineering" (B. B, Biswas and S. Roy, eds.), p. 303. Plenum Press, New York, 1995. N. Khoroshilova, H. Beinert, and P. Kiley, Proc. Natl. Acad. Sei. U.S.A. 92, 2499-2503 (1995). B. Lazazzera, H. Beinert, N. Khoroshilova, M. C. Kennedy, and P. Kiley, J. Biol. Chem. 271, 2762-2768 (1996). K. S. Mclver, A. S. Heath, B. D. Green, andJ. R. Seott, J. Bacteriol. 177, 6619-6624 (1995). I. Uehida, S. Makino, T. Sekizaki, and N. Terakado, Mol. Microbiol. 23, 1229-1240 (1997). H. Aiba, S. Fugimoto, and N. Ozaki, Nucleic Acids Res. 10, 1345-13 61 (1982). P. Cossart and B. Gicquel-Sanzey, Nucleic Acids Res. 10, 1362-1378 (1982). N. Baichoo and T. Heyduk, Biochemistry 36, 10830-19836 (1997).
62
GARY P. ROBERTS ET AL.
54. D. Shelver, M. F. Reynolds, M. V. Thorsteinsson, R. L. Kerby, S. Chung, R. B. Parks, J. N. Burstyn, and G. P. Roberts, Biochemistry 38, 2669-2678 (1999). 55. T. Uchida, H. Ishikawa, S. Takahashi, K. Ishimori, I. Moilshima, K. Ohkubo, H. Nakajima, and S. Aono, J. Biol. Chem. 273, 19988-19992 (1998). 56. W. Gong, B. Hao, S. S. Mansy, G. Gonzalez, M. A. Gilles-Gonzalez, and M. K. Chan, Proc. Natl. Acad. Sci. U.S.A. 95, 15177-15182 (1998). 57. W. Gong, B. Hao, and M. K. Chan, Biochemistry 39, 3955-3962 (2000). 58. J. R. Stone and M. A. Marietta, Chem. Biol. 5, 255-261 (1998). 59. M. E Reynolds, D. Shelver, K. M. Vogel, R. B. Parks, M. V. Thorsteinsson, R. L. Kerby, T. G. Spiro, G. P. Roberts, and J. N. Burstyn, Biochemistry 39, 388-396 (2000). 60. M. Takabashi, B. Blazy, and A. Baudras, Biochemistry 19, 5124-5130 (1980). 61. T. Heyduck andJ. C. Lee, Biochemistry 28, 6914-6924 (1989). 62. J. Malecki, A. Polit, and Z. Wasylewski, J. Biol. Chem. 275, 8480-8486 (2000). 63. I. Gorshkova, J. L. Moore, K. H. MeKenney, and F. P. Schwarz, J. Biol. Chem. 270, 2167921683 (1995). 64. J. M. Passner and T. A. Steitz, Proc. Natl. Acad. Sci. U.S.A. 94, 2843-2847 (1997). 65. G. Parkinson, C. Wilson, A. Gunasekera, Y. Ebright, R. Ebright, and H. Berman, J. Molec. Biol. 260, 395-408 (1996). 66. S. C. Schultz, G. C. Sheilds, and T. A. Steitz, Science 253, 1001-1007 (1991). 67. I. T. Weber and T. A. Steitz, J. Molec. Biol. 198, 311-326 (1987). 68. S. A. Kumar, N. S. Murthy, and J. s. Krakow, FEBS Lett. 109, 121-124 (1980). 69. M. E Reynolds, D. Shelver, R. L. Kerby, R. B. Parks, G. P. Roberts, and J. N. Burstyn, J. Am. Chem, Soc. 120, 9080-9081 (1998). 70. K. M. Vogel, T. G. Spiro, D. Shelver, R. L. Kerby, M. Thorsteinsson, and G. P. Roberts, Biochemistry 38, 2679-2687 (1999), 71. I.K. Dhawan, D. Shelver, M. V. Thorsteinsson, G. P. Roberts, and M. K. Johnson, Biochemistry 38, 12805-12813 (1999). 72. M. V. Thorsteinsson, R. L. Kerby, M. Conrad, H. Youn, C. R. Staples, W. N. Lanzilotta, T. J. Poulos, J. Serate, and G. P. Roberts, J. Biol. Chem. 276, 39332-39338 (2000). 73. M. Bergdoll, M.-H. Remy, C. Cagnon, J.-M. Masson, and P. Dumas, Structure 5, 391-401 (1997). 73a. K. Yamamoto, H. Ishikawa, S. Takahashi, K. Ishimoil, I. Morishima, H. Nakajima, and S. Aono, J. Biol. Chem. 276, 11473-11476 (2001). 74. M. V. Thorsteinsson, R. L. Kerby, and G. P. Roberts, Biochemistry 39, 8284-8290 (2000). 75. ]. Moore, M. Kantorow, D. Vanderzwaag, and K. McKenney, J. Bacteriol. 174, 80304035 (1992). 76'. H. Aiba, T. Nakamura, and H. Mitani and H. Moil, EMBOJ. 4, 3329-3332 (1985). 77. H. Nakajima, T. Matsuo, T. Tawara, and S. Aono, J. Inorg. Biochem. 78, 63-68 (2000). 78. S. Aono, T. Matsuo, T. Shimono, K. Ohkubo, H. Takasaki, and H. Nakajima, Biochem. Biophys. Res. Commun. 240, 783-786 (1997). 79. P. J. Kiley and W. S. Reznikoff, J. Bacteriol. 173, 16-22 (1991). 80. S. B. Melville and R. P. Gunsalus, Proc. Natl. Acad. Sci. U.S.A. 93, 1226-1231 (1996). 81. M. Zheng, E Aslund, and G. Storz, Science 279, 1718-1721 (1998). 82. Y. He, T. Gaal, R. Karls, T. Donohue, R. Gourse, and G. E Roberts, J. Biol. Chem. 274, 10840-10846 (1999). 83. Y. He, D. Shelver, R. L. Kerby, and G. P. Roberts, J. Biol. Chem. 271,120-123 (1996). 84. G. S. Lloyd, S. J. Busby, and N. J. Savery, Biochem. J. 330, 413-420 (1998). 85. A. Attey, T. Belyaeva, N. Savery, J. Hoggett, N. Fujita, A. Ishiharna, and S. Busby, Nucleic Acids Res. 22, 4375-4380 (1994). 86. S. Busby and R. H. Ebright, j. Mol. Biol. 292, 199-213 (1999).
CooA: A CO-SENSING TRANSCRIPTIONAL FACTOR 87. 88. 89. 90. 91. 92.
63
A. Bell and S. Busby, Mol. MicrobioI. 11,383-390 (1994). B. Li, H. Wing, D. Lee, H. C. Wu, and S. Busby, Nucleic Acids Res. 26, 2075-81 (1998). V.A. Rhodius and S. J. Busby, J. Mol. Biol. 299, 295--310 (2000). A. Gunasekera, Y. W. Ebright, and R. H. Ebright, J. Biol. Chem. 26"/, 14713-14720 (1992). S. Spiro and J. R. Guest, Molec. Microbiol. 1, 53~55 (1987). S. Spiro, K. Gaston, A. I. Bell, R. E. Roberts, S. J. Busby, and J. R. Guest, Molec. Microbiol. 4, 1831-1838 (1990).
The msDNAs of Bacteria BERT LAMPSON,* M A S A Y O R I I N O U Y E , tA AND SUMIKO INOUYE t
*Department of Health Sciences East Tennessee State University Johnson City, Tennessee 37614 t Department of Biochemistry Robert Wood Johnson Medical School University of Medicine and Dentistry of New Jersey Piscataway, New Jersey 08854 I. The Structure of m s D N A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Primary Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. C o n s e r v e d Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. m s D N A Is C o m p l e x e d with Protein . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Processing of m s D N A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. The Retron: A Genetic E l e m e n t That Codes for the Production of m s D N A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. G e n e Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. The Retron Insertion Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Transcription of the Retron E l e m e n t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . l l I . Synthesis of m s D N A by Reverse Transcription . . . . . . . . . . . . . . . . . . . . . . . . A. Transcription and F o l d i n g of the P r i m e r - T e m p l a t e RNA . . . . . . . . . . . . B. Priming, Branch Formation, and the Start of e D N A Synthesis . . . . . . . . C. Polymerization and Termination of e D N A Synthesis . . . . . . . . . . . . . . . . IV. Bacterial Reverse Transcriptase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Prevalence, Diversity, and Origin of m s D N A . . . . . . . . . . . . . . . . . . . . . . . . . A. N e w m s D N A s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Old m s D N A s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI. m s D N A and the Host Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Mutation F r e q u e n c i e s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Repetitive Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Potential Uses for m s D N A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Use of m s D N A as an Antisense D N A . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. D i r e c t e d Synthesis of e D N A in Vivo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII. C o n c l u d i n g Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
66 66 68 70 70 71 71 72 74 75 76 77 77 80 82 82 84 84 84 85 86 86 88 88 89
1To w h o m c o r r e s p o n d e n c e should be addressed. Progress in Nucleic Acid Research and Molecular Biology,Vol. 67
65
Copyright • 2001 by Academic Press. All rights of reproduction in any form reserved. 0079-6603/01 $35.00
66
BERT LAMPSON ET AL. msDNAs are small, structurally unique satellite DNAs found in a number of Gram-negative bacteria. Composed of hundreds of copies of single-stranded DNA--hence the name multicopy single-stranded DNA--msDNA is actually a complex of DNA, RNA, and probably protein. These peculiar molecules are synthesized by a reverse transcription mechanism catalyzed by a reverse transcriptase (RT) that is evolutionarily related to the polymerase found in the HIV virus. The genes, including the RT gene, responsible for the synthesis of msDNA are encoded in a retron, a genetic element that is carried on the bacterial chromosome. The retron is, in fact, the first such retroelement to be discovered in prokaryotic cells. This report is a comprehensive review of the many interesting questions raised by this unique DNA and the fascinating answers it has revealed. We have learned a great deal about the structure of msDNA: how it is synthesized, the structure and functions of the RT protein required to make it, its effects on the host cell, the retron element that encodes it, its possible origins and evolution, and even its potential usefulness as a practical genetic tool. Despite the impressive gains in our understanding of the msDNAs, however, the simple, fundamental question of its natural function remains an enduring mystery. Thus, we have much more to learn about the msDNAs of bacteria. © ~00] AcademicPress.
I. The Structure of msDNA A. Primary Structure m s D N A was first identified in the myxobacterium Myxococcus xanthus as a small, single-stranded D N A satellite o f 162 nucleotides (1). It was quickly realized that this D N A molecule is associated with a single-stranded R N A (2, 3). However, the R N A molecule is j o i n e d to the 5' e n d o f the D N A molecule b y a unique 2',5' branch linkage. In o t h e r words, at a specific point within the middle o f the R N A chain, the D N A strand is j o i n e d at the 2' position o f an internal guanosine residue o f the R N A (2, 4). This p r o d u c e s a u n i q u e R N A molecule with a b r a n c h i n g single-stranded D N A (Fig. 1). T h e unusual 2',5' p h o s p h o d i e s t e r
FIG. 1. Conserved secondary structures of msDNAs. As illustrated by Ec73, msDNAs are composed of a single-stranded DNA joined to a single-stranded RNA (boxed sequence). The RNA chain is linked to the 5' end of the DNA strand at a specific internal guanosine base in the RNA (circled G in RNA). The bond is from the 2' position of the internal guanosine of the RNA, linking the 5' end of the DNA chain, msDNAs vary considerably in their primary base sequence (in both the RNA and DNA strands). However, all msDNAs share the following secondary structures. (A) Both the single-stranded RNA and especially the single-stranded DNA fold into stable stem-loop structures. (B) All msDNAs contain a specific internal guanosine residue (circled G) in the RNA which forms the branched, 2'~5' phosphodiester bond with the DNA. (C) A small RNA-DNA hybrid region forms via base pairing between the 3' ends of the DNA and RNA strands. Some msDNAs undergo later processing steps into smaller mature forms. In some cases (Mx162 and Sa163), part of the 5' end of the RNA chain is removed (arrow). For other msDNAs (Ec78 and Ec83), cleavage near the 5' end of the DNA (arrowhead) removes part of the DNA along with the entire ItNA strand.
THE msDNAs OF BACTERIA
67
fc-~
C
C C A A A
~
G-C
•
c-G
~- GGCATAGCT-3 ' :::::
0
C
~ ....
W
5 ' -TTGAGCA J
I ,
A
~G-C ~ -60 C-G T-A
•.
A
k
(A)
IT-A.~ A
m,,.C c
(B)
I
20- C-G T-A T-A G-C A-T C-G
G A G
®2_
C
I
C A ~-C_GJ G-C -40 G-C ToG G-C rG-C~ T C ~A-T 2 G-C T-A C-G G-C
A
j oU I m
C
3'-AUCGA--~
, ~
.
u
A
U
20 A-U 40-G'U 60 U-A g-C G-C G-C UG
RNA
k..AJ
__
GeU
GeU C-G C-G A-U U-A ,,G-C ~,
A U ~-G..)
msDNA-Ec73
(A)
68
BERT LAMPSONET AL.
linkage of msDNA resembles, somewhat, the branch linkage in lariat RNAs which form during splicing ofintrons in eukaryotic cells. In fact, the 2',5' linkage in msDNA can be cleaved by the debranching enzyme (2',5'-phosphodiesterase), purified from HeLa cells, which normally functions to remove lariat RNAs (2, 4). But msDNA is like no other naturally occurring nucleic acid in that this branch linkage joins RNA to a separate molecule of DNA. Currently, about 12 msDNAs have been sequenced and characterized such that their basic structure is known (or at least surmised) with good confidence (Table I). Most of these are from Escherichia coli, but also include InsDNA from Vibrio cholerae and several myxobacteria. All msDNAs are quite small. The largest, Sa163, is composed of a single-stranded DNA of 163 nucleotides, while msDNA-Ec48 contains a DNA of just 48 nucleotides. The single-stranded RNA chains range in size from 49 (Mx65) to 119 (Ec48) bases. This makes these satellite molecules presumably too small to encode for a functional protein or to be autonomously replicating elements. Nevertheless, some bacteria apparently produce huge quantities ofmsDNA. In M. xanthus, for example, msDNA-Mx162 is calculated to exist at 500-700 copies per genome (1). With a genome size estimated at "~9 million base pairs (5), msDNA-Mx162 may represent as much as 1% of the cell's total DNA content. For the most part, msDNAs display a great deal of variation in their primary DNA sequence and RNA sequence. Some, however, are clearly similar, such as the msDNAs from Stigmatella (Sa163), M. xanthus (Mx]62), and Melittangium (ML162) (4, 6, 7). Likewise, the msDNA from V. cholerae (Vc95) shows several regions of identical DNA and RNA sequence with the E. coli msDNA-Ec78 (8). For some bacteria such as M. xanthus and Nannocystis exedens the same bacterial cell produces two different msDNAs. For example, Mx162 and Mx65 are two unrelated msDNAs that coreside in M. xanthus cells (9, 10).
13. Conserved Structures Although msDNAs vary greatly in their primary structure (nucleotide sequence), all msDNAs share similar conserved secondary structures. These include very stable stem-loop structures (Fig. 1A). Both the single-stranded RNA and the single-stranded DNA will fold up into these conserved hairpin structures. In fact, the DNA stem of msDNA-Mx162 can be cleaved with restriction enzymes (e.g., HaelII) that require a double-stranded DNA substrate (1, 4). Another shared secondary structure is the short RNA-DNA hybrid at the 3' ends of the DNA and RNA strands (Fig. 1C). Consistent with this structure, the RNA in the hybrid region has been shown to be susceptible to cleavage by RNase H (3). Finally, the other conserved structure believed to be present in all msDNAs is the unique 2',5' branch linkage that joins the end of the DNA strand to an internal residue in the RNA strand (Fig. 1B). This branch structure always occurs at a specific internal guanosine within the triribonucleotide AGC
~-~
t.D
= =
"~
=¢
o
-a
~ ~.~.~
~-~
o
%
Z
© ,II
Z
~
Z
~
N
Z
Z
Z
~
~
~
Z
Z
© 0
~
~
o~
o"
=o ,'o
o.~ ~6>E
o
,-o'~
•-~
Z
,-n 8
70
BERT LAMPSONET AL.
of the RNA chain. These conserved secondary structures occur because of the way msDNAs are synthesized by reverse transcription, as described below.
C. msDNA Is Complexed with Protein msDNA gets its name from its original discovery as a small satellite DNA existing apart from the chromosome. In all likelihood however, msDNA is actually a complex of DNA, RNA, and protein, msDNA is probably an extrachromosomal, nucleoprotein complex of a high molecular weight. The complex is composed of msDNA (single-stranded DNA linked to single-stranded RNA) bound to reverse transcriptase and perhaps other host proteins. For example, when purified msDNA (Mx162) is allowed to incubate with a small amount of cell-free extract from M. xanthus, gel-shift assays show a complicated pattern of (one or more) proteins bound to the msDNA. The protein(s) appears to bind primarily to the stem region of the folded msDNA molecule (11). Likewise, when RT-Ec67 is purified from E. coli, a 65,000-molecular-weight protein with reverse transcriptase activity copurifies with msDNA as a high-molecular-weight complex. This complex is quite large at 600,000-700,000 molecular weight (12). It is not known how this complex is formed. In addition to RT, other host proteins may also bind or interact with msDNA. Some msDNAs, for example, have been shown to utilize the host cell's RNase H during their synthesis (13, 14). When msDNA is overproduced in high copy number, it appears that the host cell's mismatch repair proteins, such as MutS, may recognize and bind to the msDNA molecule. Thus, msDNA is probably a large nucleoprotein complex composed of the single-stranded DNA-RNA molecule folded into a complicated hairpin secondary structure which is then bound to the RT protein as well as other host proteins such as RNase H.
D. Processingof msDNA Some msDNAs appear to undergo post-production processing reactions to convert a precursor form into a (presumably) more stable mature form of the molecule. For example, pulse-chase experiments clearly show that msDNAMx162 is initially produced with a 77-base single-stranded RNA linked to the DNA chain. Over a 1-h chase period, however, the msDNA is converted into a smaller, mature form. Conversion to the smaller form involves a processing event in which about 16 nucleotides are removed from the 5' end of the RNA chain to give a mature form of msDNA with only a 61-base RNA molecule (3, 4). Two msDNAs discovered from clinical strains ofE. coli, Ec78 and Ec83, apparently are not linked to an RNA molecule. These RNA-free, single-stranded DNAs, however, are derived from a precursor molecule that resembles a typical msDNA. That is, the DNA strand is joined to a single-stranded RNA via a 2',5' branched linkage. These mature RNA-free msDNAs result from a processing event in which the DNA chain is cleaved near its 5' end, releasing the RNA chain
THE msDNAsOF BACTERIA
71
from the molecule (15, 16). In the case of Ec83, an endonucleolytic cleavage of the DNA strand between the fourth and fifth nucleotides removes the entire 84-base RNA strand plus four deoxyribonucleotides joined at the branching guanosine (Fig. 1, arrowhead) (17). The nature of the enzymes that carry out these processing steps on msDNA is not known. The retron Ec83 that codes for the RNA-free msDNA appears to be a typical retron element with respect to its gene organization and structure. A plasmid subclone of the retron element in E. coli K-12 also produces the RNA-free form of msDNA. Lim (15) has speculated that either the RT or the precursor form of the msDNA itself may provide the endonucleolytic activity to process this molecule. At this time, however, it cannot be ruled out that some unusual sequence or structural feature of these msDNAs allows an unidentified host nuclease to recognize and carry out this processing step.
II. The Retron: A Genetic Element That Codes for the Production of msDNA Retrons were the first retroelement to be discovered in the prokaryotes and have since been joined by the group II introns as RT encoding elements. Retrons are defined genetically based solely on their only known activity. That is, they code for the information needed to produce msDNA.
A. Gene Organization The operon required to produce msDNA is called a retron. Because msDNA is not an autonomously replicating satellite DNA, production requires two retron-encoded loci. One, designated msd, encodes a DNA sequence identical to the DNA strand of msDNA. The other locus (located upstream ofmsd) is designated msr and encodes a sequence (when converted into RNA) that corresponds to the RNA strand of msDNA. The location and orientation of these two "genes," msd and nmr, is such that they are situated in opposite directions and converge with a slight overlap of their 3' ends (see Fig. 3A). This unusual arrangement is due to the way in which msDNA is synthesized by reverse transcription, as discussed later. In close proximity to msd is an open reading frame (ORF) which codes for a RT. This ORF is required to produce msDNA, since deletion or interruption of this gene completely blocks production of msDNA. The genes msd, msr, and the RT (also designated ret) are the only information needed by the host cell to produce msDNA, since a plasmid subclone of these three genes will also produce msDNA (18). The retron element was initially defined as these three genes--msr, msd, and ret--plus an upstream promoter because these three genes are transcribed together as an operon (see below). The retron
72
BERT LAMPSON ET AL.
element, however, also includes all unique flanking DNA associated with the msr-msd-ret operon in those examples where the junctions of the element are known. For example, in some wild isolates ofE. coli, the retron-Ecl07 is found inserted into the chromosome between the pyrE and ttk genes at "~82 min (19). Unique flanking DNA associated with Ec107 is revealed by comparing the DNA sequence of this intergenic region between pyrE and ttk in strains that contain the retron element with wild-type E. coli K-12, which lacks this element. For Ec107, this unique DNA at the ends of the retron lie just 70 bp upstream of the msr gene and 47 bp downstream of the RT gene to give an element of only 1.3 kb (Fig. 2). Ecl07 is the smallest retron thus far defined, therefore, the minimum genetic unit that constitutes a retron element is just the genes msr, msd, and ret plus a promoter sequence upstream ofmsr and some short terminal sequences. Some retron elements are larger and contain enough information to encode a second ORF. For instance, retron Ec73 is about 2.4 kb and encodes a second ORF, ORF316, which lies between the msd and ret genes in the element (Fig. 2). However, the function of this second ORF is unknown. The amino acid sequence of ORF316 shows no significant similarity with any known proteins; and when ORF316 was deleted from the retron element, it had no effect on production of msDNA (20). Retron elements that have been sequenced vary considerably in size, primary DNA sequence, and even the amino acid sequence of the RT gene.
8. The Retron Insertion Site Characterization of the precise site on the chromosome where a retron element has inserted requires that the ends or junctions of the retron be known. This has been determined for a number of retron elements from E. coli but remains unknown for retron elements in other bacteria like the myxobacteria. As mentioned earlier, retron Ec107 is inserted by itself within an intergenic region of the chromosome. Apparently, when the retron inserted, it replaced a 34-bp sequence which is lost from this region (between the pyrE and ttk genes) in the E. coli strains containing this element (19). The retron is not flanked by directly repeated chromosomal sequence, as is seen in the duplication of the target site in many transposons found in E. coli. Ec107 does contain, however, short inverted repeat sequences at both the 5' and 3' termini of the element. The mechanism employed by the retron Ec107 to insert into this site on the E. coli chromosome is not known. There are no significant DNA homologies between the intergenic sequence of K-12 (which lacks a retron element ) and the terminal sequences of the retron element, suggesting that homologous recombination is not involved. For those strains ofE. coli that contain retron Ec107, there is only one copy of the element, at the same position, on the chromosome. In fact, all retron elements thus far described appear to be found in only one copy per bacterial cell chromosome. No msDNA-producing element has been described to exist on a plasmid.
2.1 Kb Ec67 mir-msd
or RT 2.2 Kb
Ec86 ORF
~,~
=:~ Ec73 2.4 Kb ~..~ ~.~,~;.,,~, DNA ~ _ ----.-m- ~ ORF RT r n s r - m s d 316
(
msr-rnsd
RT
j
orgl~tmae DNA
100/00 rain.
selC
82.s 82
19 - "
E. c o i l Chromosome
25 -
50
Ec107
msr-msd
1.3 Kb
RT
FIG. 2. Retron element insertion sites in the E. call genome. The precise junctions between retron terminal sequences and host DNA have been determined for four E. coli retrons. Retron elements (black bar) encode all the information needed to produce msDNA, This information is organized into an operon composed of the genes msr (the RNA sequence of msDNA), msd (the DNA sequence of msDNA), and one ORF encoding a reverse transcriptase (RT). Some retron elements may contain an extra ORF of unknown function. Retron elements range in size from 1.3 kb (Ecl07) to 2.4 kb. Except for Ecl07, all E. coli retrons are found inserted into a larger element composed of prophage-like DNA (open bar). The prophage DNA is, in turn, integrated into the E. coli chromosome (stippled bar) and flanked by directly repeated target sequences (open arrows). For example, retron Ec73 is carried by a P4-1ike prophage integrated into the selC gene (for selenocysteinyl tRNA) at 82.3 min on the E. coli chromosome. A phage 186--related prophage integrated at 19 min may contain either retron Ec67 or Ee86. Retron Eel07 is an exception. It is integrated directly into an intergenic region between the pyrE and ttk genes at 82 min on the chromosome. Note that the E. coli chromosome shown here is a composite representing several different strains. For those strains that have them, the chromosome contains only a single copy of one type of retron element. Note also that prophage DNAs are much larger in size than is shown in drawing.
74
BERT LAMPSON ET AL.
In E. coli, the retron Ecl07 is exceptional in one regard. Unlike this element, all the other retrons that have been characterized (Ec48, Ec67, Ec73, Ec83, Ec86, and possibly Ec78) are associated with prophages. For example, retron Ec73 is found inserted into the right end of a 12.7-kb prophage sequence that lies at 82.3 rain on the chromosome ofE. coli strain C1-23 (Fig. 2) (20). The prophage resembles the phage P4, a parasitic phage of the P2 family. In another interesting association with a prophage, phage 186 (a member of the P2 family) has apparently picked up two different retrons, Ec67 and Ec86. As a result of two independent events, these two different retron elements have inserted into precisely the same site of the 34-kb prophage sequence which resides at 19 rain on the chromosome ofE. coli (Fig. 2) (21-23). A careful comparison of the DNA sequence of phage 186 (which lacks a retron) with prophages containing retrons Ec67 and Ec86 indicates that, in each case, the retron element inserted into a region between the cos site and a DNA packaging gene (ORF 341). About 180 bp of phage DNA is lost upon insertion of the retrons, which is reminiscent of Ec107 (24, 25). Apparently, loss of target site sequence is a common event in the insertion of each of these three different retrons. Again, the precise mechanism that leads to insertion of retrons Ec67 and Ec86 is not known. A speculative model has been proposed in which retron insertion involves formation of a junction with target DNA at two spatially separate sites (nicks in the DNA) followed by removal of target DNA between these two sites upon completion of the insertion event. The model further proposes that the left retron junction may form as a consequence ofa RT priming reaction at an aberrant nick in the target DNA (presumably caused by the phage-encoded integrase). Insertion of the retron at this nick would proceed by reverse transcription of a mRNA copy of the element using the nicked target site as a primer (25). This intriguing model is based on a comparison of phage 186 with the prophages containing the retrons Ec67 (~bR67) and Ec86 (~bR86). It should be emphasized, however, that the prophages ~bR67 and ~bR86, although clearly similar to phage 186, are not the same as phage 186. For example, the prophage ~bR67 contains the retronEc67 plus several additional ORFs not found in phage 186 (21). Likewise, the prophage ~bR86 is actually a defective prophage, since SOS induction of the prophage failed to produce any plaques or detectable phage particles in E. coli B even though lethal phage genes were being expressed (26). Thus, it is difficult to know exactly what the target DNA for these retrons actually looked like prior to retron insertion, and therefore it is difficult to identify the exact junction between the retron DNA and phage DNA.
C. Transcription of the Retron Element $1 nuclease protection, primer extension, and lacZ fusion experiments provide good evidence that all three retron genes--msr, msd, and ret--are transcribed together in one long mRNA. In the case of retron-Mx162 (M. xanthus),
THE msDNAsOF BACTERIA
75
transcription begins about 75 bp upstream of msr (4), while in retron-Ecl07 transcription begins right at the first nucleotide ofmsr (19). lacZ fusion studies with retron-Ec86 indicate that transcription extends through msr and msd to include the RT ORF (27). Transcription is probably controlled by a promoter encoded within the retron element upstream of msr. This is because a cloned fragment of the retron containing msr-msd-ret and DNA upstream of these genes will produce msDNA regardless of the orientation in which the insert fragment is found in the plasmid vector (19). In addition, putative -35 and -10 promoter consensus sequences have been identified within the upstream region of the retron. Beyond the rudimentary information described above, little is known about the transcription and regulation of retron genes. Only one study has looked at changes in the expression of retron genes and msDNA production (28). A chromosome-lacZ reporter construction under the control of the retron promoter from Eel07 showed a marked increase in gene expression when E. coli cells entered stationary phase. Increased expression from the retron promoter also occurs when cells are starved for nutrients, particularly phosphate; this expression is enhanced by increased levels of the guanine nucleotide ppGpp, but is not affected by the stationary sigma factor (28). Increased levels of msDNA production occurred primarily due to entry of cells into stationary phase rather than nutrient starvation alone. Although more information is needed to understand the complex regulation of this element, the enhanced expression of retron genes during stationary phase is consistent with the idea, discussed below, that large amounts of msDNA increase the mutation frequency of the host cell. This may provide some advantage for the host during periods of nutritional or other enviromnental stresses.
III. Synthesis of msDNA by Reverse Transcription msDNA can be thought of as a small eDNA that is produced by reverse transcription from a long mRNA template. In the case of msDNA, however, part of the RNA template remains joined to this eDNA upon completion of its synthesis. Another unusual aspect of msDNA synthesis is the use of RNA as both template and primer for initiation and synthesis of the eDNA. In fact, it is the unique priming mechanism used by the retron RTs that essentially explains the unique 2',5' branch structure of the msDNA molecule. As we will see, this unique primer for eDNA synthesis is the 2'-OH group of a specific guanosine residue within the RNA template. It should also be noted that one RT molecule synthesized from a mRNA uses the same RNA, in principle, to synthesize only one molecule of msDNA. In the description of ms DNA synthesis to follow, eDNA refers to the DNA chain that is found in the mature msDNA
76
BERTLAMPSONET AL.
molecule. The precursor RNA that will be reverse-transcribed to make eDNA is called the primer-template RNA to distinguish this longer RNA from the RNA chain actually found in the mature msDNA molecule (which is sometimes referred to as msdRNA).
A. Transcriptionand Folding of the Primer-Template RNA Production of the satellite molecule, msDNA, begins with the transcription of the retron operon msr-msd-ret into a long mRNA. Since this mRNA transcipt includes the RT gene (ret), it is presumably translated first to produce the polymerase, which then uses the upstream msr-msd region of the mRNA as the template as well as the primer for msDNA synthesis. Once the long mRNA is translated, only a small subregion of the RNA at the 5' end (encompassing the msr-msd genes) is needed to serve as the template-primer for eDNA synthesis. The RNA template must fold into essential secondary structures in order to function as a self-primer and template for cDNA synthesis. The most important of these secondary structures is an intramolecular base pairing between two short sequences in the RNA designated al and a2 (Fig. 3A). This stem structure lies directly adjacent to the internal gnanosine base in the folded RNA that will contain the 2'--5' branch linkage with the DNA chain (eDNA) of msDNA. This gnanosine base will be referred to as the branching G residue. Apparently, this stem structure juxtaposes the branching G residue (circled in Fig. 3A) within the folded RNA such that its 2'-OH group can be used as a primer to initiate eDNA synthesis by the RT. Site-directed mutation experiments indicate that maintaining the double-stranded stem structure is what is important, not the primary sequence per se of either al or a2 (29). Likewise, the branching G residue is a specific internal gnanosine, since alternative downstream G residues are not used even if a longer al-a2 stem structure is engineered into the RNA (29). Another, equally important secondary structure within the primer-template RNA is a region that lies downstream of the branching G residue from the a l - a2 stem (Fig. 3B). This region is designated the primer recognition region of the primer-template RNA and contains one or two small stem-loop structures that are also critical for eDNA synthesis. In a series of in vitro experiments, the primer recognition sequence of retron Ee67 was switched with the primer recognition region from retron Ec73 (30). The results indicate that the primary sequence of this region must be specific for eDNA synthesis. In other words, the RT-Ec67 will recognize the specific Ec67 primer region sequence only within the primer-template RNA to initiate eDNA synthesis. Mutational studies also indicate that the secondary stem-loop structure(s) within the primer recognition sequence is also essential for eDNA synthesis (30).
THE msDNAsOF BACTERIA
77
B. Priming, Branch Formation, and the Start of cDNA Synthesis As described above, the retron mRNA transcript must be folded into a stable secondary structure to function as a primer-template for production of cDNA. The folded primer-template RNA contains three important functional regions: (1) the al-a2 stem which positions the branching G residue to serve as a primer; (2) the primer recognition region sequence plus a stem-loop structure(s) which is specifically recognized by the cognate RT; and (3) a template region which will be reverse-transcribed to make the cDNA (Fig. 3B). An example of a folded primer-template RNA for Ec86 is shown in Fig. 3B. The primer recognition region of this RNA actually contains two stem-loop structures labeled a and b. When tested for the binding to RT-Ec86, the RNA requires the second stemloop structure (b) but not the first one (a) (41). Using the SELEX (systematic evolution ofligands by exponential enrichment) method, primer-template RNAs with randomized sequences were subjected to an enrichment scheme using purified retron RTs (41). RNAs enriched from a random pool using the RTEc86 were found to contain the second stem-loop structure (b in Fig. 3B), including a 3-U loop sequence identical to that found in the wild-type primertemplate RNA. In addition, these enriched RNAs also contain the conserved 4-base sequence UAGC, around the branching G residue (underlined). These results indicate that the RT-Ec86 recognizes a specific stem-loop structure (b) and the branching G residue located upstream in the primer-template RNA. With a properly folded primer-template RNA, the RT can now initiate the synthesis of cDNA by incorporating the first deoxyribonucleotide using the 2'-OH group of the branching G residue as a primer. A fully in vitro experiment using in vitro synthesized primer-template RNA and purified RT, indicates that the initiation of cDNA synthesis using an internal 2'-OH group as a primer is carried out by the RT alone with no other retron encoded or host encoded factors needed (31). Thus, the ability to use a 2'-OH primer to produce msDNA appears to be a unique property of retron-encoded RTs.
C. Polymerizationand Terminationof cDNA Synthesis Completion of the priming reaction, with the incorporation of the first deoxyribonucleotide (via the branch linkage), begins the polymerization of cDNA. Further incorporation of bases and extension of cDNA occur in a templatespecific manner (incorporation of the first deoxribonucleotide is also templatedependent). Again, all of these reactions are carried out by the RT. As the cDNA chain is elongated, working in concert with this polymerization reaction is the removal of the template region of the primer-template RNA (Fig. 3A) (32). Trailing slightly behind the advancing polymerization of cDNA, cleavage and removal of the RNA template takes place in a processive manner
78
BERT LAMPSON ET AL.
A D mSd
met (RT)
Transcription
.,® I I I I a2
mRNA
b
"e~ v
/I/st
ms(l Folding
Primer-Template RNA b2 h
. Q at
b4
FIG. 3. Synthesis of msDNA by reverse transcription. (A) Synthesis of msDNA begins with the transcription of the retron-encoded msr-msd-ret operon into a long mRNA. Two sets of inverted repeat sequences--a1 plus a2 and bl plus b2~in the mRNA (5' region) are crucial for the next step. This step is the proper folding of the RNA into several critical secondary structures which 'allow this RNA to serve as both a primer and template for cDNA synthesis. The properly folded primertemplate RNA can now be used by the RT to initiate cDNA synthesis (i.e., the DNA portion of msDNA), cDNA synthesis continues, along with the removal of part of the RNA template by hostderived RNase H, until a specific termination point is reached within the primer-template RNA.
79
THE msDNAs OF BACTERIA
.t,
B
r"'~A
Ec86 Prlnwr-Template R N A
b
u
(2)
%-2" li'oG %2 U-A G-C G-C
U-&
&-U
(1) ............ d"'%.
f"
//
o,d
o
---I",
,c
u
%
- - ' " ' P' =t g / . b , , ' 3 ~ r
,~
5'
-
\
/ ,-,.-.--.,a~/--
T~
a
a
~
/~$ACC4ZGU~#G-C-A-G-U-C-U-U-U-U-U-U-G-C-C-C
"~"~... ........... /
ua
.-~
..G.
"trcoceuooouoo cGoA~ocAue"°'c AAAGGACCAACC C~CUCUCGUAG
"c,tv
%.2"
Q
xuz
I ~'
k,/"
(3) (B) The primer-template RNA structure required for msDNA synthesis. The primer-template RNA (from Ee86 as an example) is derived from a longer mRNA transcript of the mar-msd-ret operon (see Fig. 3A). This RNA folds up into several important secondary structures which divide the RNA into three functional regions. (1) The al and 'a2 sequences in the BNA pair up to form a stem structure adjacent to the internal guanosine (circled G) base that will form the 2' branch linkage with DNA (branching G). The al-a2 stem 'along with a guanosine base opposite the branching G in the folded RNA (G with a double underline) is critical for the proper positioning of the branching G residue so that it can serve as a primer for eDNA synthesis (67). (2) The primer recognition sequence is a region just downstream of the branching G residue which is specifically recognized by the cognate retron RT. For example, in the Ec86 RNA, stem-loop structure (b) has been demonstrated to be essential for msDNA synthesis while structure (a) is dispensible (41). Furthermore, the SELEX method with pure RT-Ee86 and random RNAs indicates that the retron RT specifically recognizes the boxed sequence in stem-loop structure (b) as well as the branching G residue. (3) The template region of the RNA is the part of the RNA that will serve as a template ~br eDNA synthesis. Italicized letters represent the beginning of eDNA synthesis from the 2' branching C residue, cDNA synthesis terminates at the position shov
80
BERT LAMPSON ET AL.
by an RNase H activity. With the possible exception of Ec67, no retron-encoded RT appears to contain an RNase H domain. Thus, a ribonuclease that can selectively cleave RNA in an RNA-DNA duplex must be supplied by the bacterial host cell. Indeed, experiments with E. coli rnh mutants show that production of the mature, wild-type form of msDNA (for both Ec73 and Ecl07) requires the host cell-derived RNase HI (13). Extension of the cDNA chain, with the concomitant removal of template RNA, continues along until the eDNA reaches a specific termination site in the primer-template RNA. For example, in the case of Ec73, cDNA production extends no further than 73 bases (Fig. 3B, arrowhead), with the last RNase H cleavage site only 5 bases lagging behind this termination point (Fig. 3B, open triangle). This leaves a mature rnsDNA molecule containing a 73-base cDNA joined to an RNA chain containing the a2 sequence, the primer region sequence, and 5 bases of the template region which forms the small RNA-DNA hybrid at the 3' ends of each chain. Termination of cDNA synthesis is very specific when msDNA is produced in vivo. When msDNA is produced in cell-free or in vitro systems, however, this specificity for termination is often not maintained. In vitro systems often produce variable msDNAs with both shorter and longer eDNAs than seen in the wild-type molecule (31). What factors determine precise termination of cDNA synthesis are not known. RNase H may play some role, since msDNA formed in E. coli mutant strains lacking a functional RNase H enzyme often produce msDNAs with different termination points (13). It is interesting that, in contrast to the primer recognition region, the template region of the primer-template RNA does not need to be of any specific sequence. For example, in the retron Ec73 the template sequence can be exchanged with a heterologous template sequence from Ec67 (by exchanging their respective msd genes). As long as the primer region remains the same, the Ec73 retron will produce a chimera msDNA containing the Ec73 msdRNA, but having the Ec67 cDNA chain (30). In fact, the template sequence does not even need to be on the same RNA molecule as the primer region. The above-described in vitro system for producing msDNA has demonstrated that a short primer RNA can be used to prime the synthesis of cDNA off of a separate template RNA molecule (33).
IV. Bacterial Reverse Transcriptase Reverse transcriptases are a surprisingly diverse group of proteins found in nearly every kind of eukaryotic organism and in many prokaryotes. Widely assorted genetic elements employ this polymerase for their replication or mobility, including RNA viruses, DNA viruses, transposons, introns, and even mitochondrial plasmids (34, 35). Even among the priming reactions used to initiate cDNA
THE msDNAs OF BACTERIA
81
synthesis there is unusual diversity. For example, retroviral RT uses the 3'-OH at the terminus of a tRNA molecule (36) while the RT of hepatitis B virus uses a protein primer, and, surprisingly, the RT encoded by the Mauriceville rnitochondrial plasmid uses no primer at all (37). The amino acid sequences of RTs are also quite varied. The sequence of RT-Ec67, for example, contains fewer than 16% identical amino acids with the sequence of the HIV-RT (18). Even the amino acid sequences of two retron RTs, Ec86 and Ec73, share only 22% identical residues. Despite these differences, however, a careful alignment of the amino acids among all known RTs clearly shows that the retron RTs are related to the RTs of retroviruses (38). The most highly conserved amino acids among all RTs fall into seven regions or domains which correspond to conserved secondary structures within the protein, and these seven regions---including the well known RT signature sequence, the YXDD box--are also found in the retron sequences (Fig. 4) (34, 38).
fingers
palm
fingers
palm
thumb
~////////~~Y/////////// ! I
I
i I
i I
I I
I I
I I
I ,
Y KG R KQ
NAH
D FF
G SP
YADD
N K
VTG
III
III
III
llll
II
III
III III
X
Y
N-terminal extension (in mYXObacterlalRTsonlY)
Structural Domains of R~ron RTs FIG. 4. Structural domains found within the retron reverse transcriptases. All RTs, from both eukaryotic and prokaryotic elements, contain seven blocks of highly conserved amino acids (open boxes labeled 1-7). Some of the most conserved amino acids found among retron RTs are shown above each block (single-letter abbreviations). These blocks of conserved amino acids correspond to structural regions within the folded protein which resemble a hand and are designated fingers, palm, and thumb (hatched, crosshatched, and stippled bars, respectively, shown at the top). Highly variable or nonconserved amino acids among all RTs are shown as black boxes. However, within these sequences are two regions, designated X and Y, which are shared by most retron RTs and contain conserved amino acids unique to these proteins. Retron RTs from the myxobacteria contain an extended region of unknown function (135-170 amino acids) at the N terminus that lies beyond the polymerase region (shown as a shaded box).
82
BERT LAMPSON ET AL.
The defining activity of retron-encoded RTs is, of course, their ability to use a 2'-OH group as a primer for cDNA synthesis. Not surprisingly, retron RTs form a distinct and deep branch in various phylogenetic trees that have been constructed to compare the relationship among these diverse polymerases (39). Interesting questions about the retron RTs arise: What amino acid segments and corresponding structural regions of the protein are involved in the unusual 2'-OH priming reaction? And are these amino acids and structures unique? Using the powerful in vitro system for the production of msDNA, recent experiments are beginning to provide some answers. Based on the crystal structure of HIV1-RT, retron RTs are thought to resemble a human hand wrapped around the al-a2 stem-helix of the primer-template RNA, something like the hand of a guitar player with fingers and thumb clasped around the frets on the neck of the instrument (39). Thus, the retron protein can be divided into structural domains loosely resembling that of a hand, with fingers, palm, and thumb regions (Fig. 4). For example, the highly conserved pair of aspartate residues in the YADD box lie within the palm structural domain; these residues are believed to form the catalytic core responsible for the DNA polymerase activity (40). In addition to the fingers, palm, and thumb region shared by all RTs, the bacterial enzymes have two domains, designated X and Y, which appear to be unique. Experiments using the in vitro system have been done in which the X and Y regions were either deleted or exchanged with those from other RTs. Based on these experiments, it appears that the C-terminal end of the retron protein, which includes the Y region and the thumb structure, plays a major role in specific recognition and binding to its cognate primer-template RNA. For example, the C-terminal al region of RT-Ec86 was found to specifically recognize the conserved stem-loop structure (b) and the branching G residue in the primer-template RNA for Ec86 (41). This specific binding may sandwich the RNA stem between the thumb and fingers of the RT, allowing the protein to bend the primer RNA and exposing the 2'-OH group of the branching G residue to initiate cDNA synthesis (41).
V. Prevalence, Diversity, and Origin of msDNA msDNA was first discovered in the myxobacteria, and msDNA elements appear to be very old residents in the genome of these bacteria. In contrast, the E. coli retrons appear to be new or recent additions to the genome of this species of bacteria.
A. New msDNAs Several populations ofE. coli have been surveyed for the presence ofmsDNA and, by inference, its corresponding retron element. In each case, msDNA is found in only a small minority of the strains. For example, msDNA is produced
THE msDNAsOF BACTERIA
83
in only 29 out of 87 EPEC strains surveyed (42), while only 7 out of 113 clinical strains (5%) were found to contain msDNAs (43). And only 11 out of 75 (15%) ECOR strains contain an msDNA element (44). This is most significant because the ECOR collection is the best representation of the broad genetic diversity of natural populations of E. coli (45). Seven distinct msDNAs have been found among the various E. coli strains with little or no similarity between them. msDNAs have also been found among other members of the gamma subgroup of the purple (proteobacteria) bacteria, including Salmonella, Klebsiella, Proteus, and Vibrio. Rhizobium and Bradyrhizobium species, which are members of the alpha subgroup of the purple bacteria, also contain msDNA; and, like the populations of E. coli, only a small percentage of strains contain these elements (6 of 41 Rhizobium strains and 4 of 21 Bradyrhizobium strains) (46). All told, some 18 different bacterial genera have been shown to contain msDNA, including all ten major genera of myxobacteria. One explanation for so few msDNA-producing strains in a population ofE. coli is that retrons are a recently acquired DNA from an outside source which was picked up by just a few strains. Indeed, other evidence points to a recent arrival of retron elements to the genome orE. coli. Analysis of the codons found in retron-encoded RT genes indicates an atypical codon usage pattern. For example, in the RT-Ec67gene, 65% of arginine codons are AGA or AGG, while these codons are rarely used (2.7%) among 199 E. coli genes (19). Another bit of evidence comes from the retron Ec107. This type of element is by far the most common retron found among the ECOR collection (8 out of 11 strains that produce msDNA contain the Ec107 element). The degree of nucleotide sequence diversity between individual elements can be used as a measure of how long a retron element has resided in the E. coli genome. For example, a comparison of the nucleotide sequence of 12 individual Ecl07 elements reveals very little sequence variability (9 out of 12 are identical). This is unusual because the Ec107 retrons are found in strains from different phylogenetic branches of the ECOR reference population and would be expected to have undergone significant sequence divergence (47). Together, the evidence strongly suggests that certain strains of E. coli have recently picked up and incorporated retrons from some outside source. An obvious vehicle for the delivery of these elements into the E. coli chromosome is a bacteriophage. Indeed, all known retrons from E. coli, with the exception of Ecl07, are associated with prophage DNA in the chromosome. Ec73, for example, is encoded as part of a 12.7-kb prophage integrated at 82 min in the E. coli chromosome (Fig. 2) (20). This retron-containing prophage, designated qbR73, is related to phage P4, a kind of parasitic phage of another phage called P2. In fact, using P2 as a helper virus, the qbR73 prophage can be induced to excise from the chromosome and form infectious virions (48). The newly formed virions can then be used to infect and lysogenize a new host strain, with the newly formed lysogen capable of producing msDNA.
84
BERT LAMPSONET AL.
B. Old msDNAs A population of 20 different isolates of the soil bacterium M. xanthus has also been examined for msDNAs and retron elements; here, however, the findings are much different from those in E. coli. Whereas msDNA elements rarely occur in more than 10% of all strains of E. coli, Salmonella, or Rhizobium, all 20 strains of M. xanthus examined produce msDNA. While four different msDNAs are found among the retron-containing strains of the ECOR collection, all 20 strains of M. xanthus appear to produce an msDNA identical to Mx162 (49, 50). And whereas the retron RT proteins from E. coli appear to use rare codons frequently, this is not the case with the retron RTs found in M. xanthus. The RT-Mx162, for example, has an overall codon usage very similar to that of other known M. xanthus proteins (10, 51). For the retron element Mx162, at least, the evidence suggests a long and ancient residence in the genome of M. xanthus. Indeed, when the degree of divergence is compared between Mx162 and a closely related retron, ML162, from the myxobacterium Melittangium lichenicloa, these two elements appear to have accumulated mutations at the same rate as any other genes from these two genera. In other words, based on the silent-site substitution rate for the RT genes, it appears likely that Mx162-type DNA sequences have existed in the genome perhaps as long as 150 million years. This would be far enough back in time for a common ancestor of the myxobacteria to have acquired these genes and then faithfully passed them down during the divergence to the various genera that exist today (7).
Vl. msDNA and the Host Cell An enduring puzzle about msDNA is its normal function. What function is intended for such a strange molecule? Whether intended by its unusual structure or not, we do know something about the effects of msDNA on the host cell that produces this molecule.
A. Mutation Frequencies Overexpression of msDNA on a multicopy plasmid vector apparently produces an increase in the frequency of certain mutations in E. coli (,52). For example, if a lac- indicator strain carries a plasmid that can be induced to overexpress the retron Ec86, a substantial increase in the frequency of lac+ revertants appears compared to controls (52, 53). Overexpression of the retron genes resuits in the production of several thousand copies of msDNA, which then causes an increase in the number of mutations, particularly frame-shift mutations in the host cell (54). This increased mutation frequency can be suppressed if the same host cell also contains a plasmid that produces the mismatch repair protein MutS. Indeed, the ability of msDNA to increase the mutation frequency of its
THE msDNAsOF BACTERIA
85
E. coli host cell is clearly linked to the presence of mismatched base pairs that lie in the stem region of the long, folded DNA chain (see Fig. 1A). Overexpression of msDNA-Ec78, which lacks any mismatched base pairs, does not cause an increase in mutations. Thus, the mechanism by which msDNA increases the mutation frequency is due to the host cell's repair proteins, primarily MutS, binding to mismatched bases in the secondary structure of the msDNA chain and essentially being titrated out or sequestered from their normal function. The MutS repair protein also functions in E. coli as a recombination barrier by removing donor DNA that differs significantly from the recipient's DNA. This is due to mismatched bases that form between the donor and recipient DNA, especially in an interspecies genetic exchange, during the recombination process (55). Here again, if the recipient strain contains a plasmid overexpressing msDNA, there is a significant increase in the interspecies recombination rate for this strain (53). Although these effects on the host cell's mutation and recombination frequencies are interesting and potentially significant, they are observed only when msDNA is overexpressed. No effect on the host cell is seen in the natural situation in which a single copy of the retron in the chromosome is expressed from its native promoter. There may, however, be natural conditions in which a native retron element could affect the host cell in a similar manner. For example, as noted earlier, for retron Ecl07 there is a marked increase in expression from the rctron's promoter when cells are in stationary phase. Likewise, there is also an increase in the levels of msDNA in stationary phase (28). In addition, other studies have noted a significant depletion in the amount of MutS and a greatly reduced capacity to carry out mismatch DNA repair in E. coli cells that have entered stationary phase (56). Together, these increased levels of msDNA and reduced levels of MutS during stationary phase may produce conditions that result in the increased mutation and recombination events observed when msDNA is overexpressed from a plasmid. Thus, the presence of a retron element may contribute to the host cell's ability to evolve under stress conditions, in particular the so-called adaptive mutational changes that occur when cells are subjected to long-term starvation or stationary phase.
B. Repetitive Sequences An interesting discovery was made during the study of a retron element from the myxobacterium Nannocystis exedens. Here, repetitive DNA sequences were found in the chromosome of this bacterium which appear to be derived from msDNA (57). Nine different repeat sequences were characterized and found to be variable in size (22-56 bases), contain only part of the 162-base msDNANel60 sequence, and appear to be dispersed throughout the host chromosome. Because msDNA is the only prokaryotic DNA (aside from the rare case of a group II intron) known to be synthesized by reverse transcription, this suggests that these repeat sequences were generated in a similar manner. It would appear
86
BERTLAMPSONET AL.
that the retron element from N. exedens is acting like the SINE-like retroposon found in many eukaryotic genomes. SINE elements have generated large numbers of repeated DNA sequences by reverse transcription of various RNAs like tRNA (34, 35). Perhaps the retron from N. exedens has produced a similar effect on this bacterial chromosome, generating reverse transcribed sequences derived from msDNA that have subsequently become incorporated into the chromosome as repeated, but truncated copies. The actual mechanism that produced these msDNA repeat sequences in N. exedens remains to be determined. But it is interesting that such a reverse transcription mechanism may be affecting the genome of a bacterium. SINE-type retroelements have apparently had an enormous effect on the structure and organization of most eukaryotic genomes. As much as 10-15% of some mammalian genomes are composed of these SINE-type repeat sequences (58). Several classes of repeat sequences, such as ERIC and REP, have been extensively documented in a variety of bacterial chromosomes (59, 60). But the mechanism used to duplicate these sequences is largely unknown. Have some of these various repeated sequences been duplicated by reverse transcription of an RNA transcript (copy)? Can the retron-encoded RT carry out this activity? These fascinating questions should yield important information about the potential of msDNA to change its host genome.
VII. Potential Uses for msDNA Although its natural function remains elusive, several practical uses and applications of msDNA could, in theory, be exploited. For example, the DNA strand of msDNA could be changed to contain a binding sequence for a regulatory protein. Then, just as overproduction of msDNA can bind up the host MutS protein, so this engineered msDNA could be used to titrate out a regulatory protein like a transcriptional repressor and allow a target gene to be expressed. Another potential application of msDNA is to use this multicopy molecule as an antisense DNA directed against the expression of a particular gene. In fact, some success has been achieved at engineering msDNA as an antisense tool.
A. Use of msDNA as an Antisense DNA As discussed in Section III, experiments with the cell-free synthesis of msDNA indicate that a large portion of the upper stem-loop formed in the DNA chain can be removed without significantly affecting the production of this molecule. This results in the synthesis of an msDNA with a truncated DNA strand (61). This finding suggests that a large portion of the DNA sequence of msDNA could be replaced with a heterologous DNA sequence which would then be copied many times by the retron RT system. Indeed, these properties of msD NA have allowed the retron Ec73 to be designed to produce an msDNA with
THE msDNAs OF BACTERIA
A
87
5 '- T T T C A T T A T T A A T A C C C T C T A G A T T
(antisense sequence)
)
J
_AATACc~ TT "~T A T T
C T A
.,~¢oRI
J k__.~a'
Antisense msDNA
msDNA
B
msDNA primer RNA (msr)
mRNA with matching a l seq. al '-~'
plus
/ cDNA Synthesis (in vivo) al FIC. 5. Potential applications of msDNA. (A) msDNA can be engineered to contain a heterologous sequence with antisense activity directed against a mRNA in vivo (base sequence on loop region of msDNA stem-loop structure). An E c o R I restriction site (arrowheads) constructed in the stem region of msDNA ~dlows the antisense sequence to be cleaved offand released in vivo. (B) The primer RNA (msr) for msDNA can be modified such that its a2 sequence matches a sequence (al) in a target mRNA transcript. This modified primer RNA can be produced as a separate RNA in the host cell. The modified primer RNA is then allowed to hybridize with the matching al sequence on a target mRNA, and then the retron RT can initiate the production of an antisense eDNA in vivo.
88
BERT LAMPSON ET AL.
heterologous DNA sequences that can act as antisense DNAs in vivo (62). In this system roughly the top half of the stem-loop structure of the DNA strand of Ec73 is replaced with a new sequence which forms a large loop at the top of the truncated stein (Fig. 5A). This new sequence is complementary to the translation initiation region on the mRNA for the lpp gene ofE. coli. lpp codes for the major outer-membrane lipoprotein and is one of the most abundant proteins in E. coli (63). When overexpression of this engineered, antisense msDNA is induced, the levels of the major lipoprotein in the cell are reduced by as much as 77%. The potential versatility of the msDNA system as an antisense tool is demonstrated by other artificial constructions. For example, another antisense msDNA was constructed to contain an EcoRI site in the stem region of the DNA strand (Fig. 5A). When this msDNA is produced in an E. coli strain expressing an endogenous EcoRI endonuclease, the loop region is cleaved off the msDNA, releasing the single-stranded antisense sequence (62). In addition, msDNA can be produced in eukaryotic cells by introducing specially designed retrons with eukaryotic promoters into yeast cells (64) and even mammalian cells (65). Thus, it may be possible to exploit the msDNA system as a vector for the production and delivery of small antisense oligodeoxyribonucleotides in a variety of different cell types.
B. Directed Synthesisof cDNA in Vivo In yet another, potentially more powerful, antisense technique, it may be possible to use the msDNA system to produce long cDNAs against a specific mRNA target in vivo. Normally msDNA is synthesized by reverse transcription from a single, folded RNA strand that serves as both a primer and template (see Section III). However, using the powerful in vitro system, it was discovered that msDNA (i.e., eDNA) can be produced using separate primer (msr) and ten:plate (msd) RNAs (33). As long as the two individual RNAs anneal to each other via the al and a2 sequences, the corresponding retron RT can initiate eDNA synthesis. This property of the msDNA system could be exploited to produce long cDNAs in vivo. For example, the a2 sequence in the primer RNA could be changed so that it is complementary to a sequence within a specific mRNA of a target cell. As long as the cognate retron RT is also supplied, the newly engineered primer RNA could anneal via the a2 sequence to the target mRNA and initiate eDNA synthesis in vivo using the mRNA as a template (Fig. 5B). Such a process could be an effective antisense technique to block the expression of a target mRNA (66).
VIII. Concluding Remarks Although their intended function remains obscure, the msDNAs of bacteria have, nevertheless, revealed a fascinating genetic system, msDNAs have
89
T H E msDNAs OF BACTERIA
provided us with a new perspective on retroelements and their evolution and the biological importance of reverse transcription. As the first example of an RTencoding element found in prokaryotic cells, retrons may be among the oldest of the retroelements, being a key ancestor of the many different RT-containing vectors found in eukaryotic cells. Still, many questions remain about the origin ofretrons, their mobility, and how have they disseminated and diversified among the different bacterial groups where they are found today. Among the most interesting discoveries concerning the msDNAs is their potential effects on the host cell. By binding and thus interfering with the host mismatch repair system, msDNAs may change the rate of mutations that arise in the host cell. This may be a subtle change, but, nevertheless, it could be biologically significant because it allows the cell to survive environmental extremes or adapt to a new ecological habitat such as a human host (i.e., a virulence factor). Indeed, Shimamoto's group has recently demonstrated that out of 21 V. cholerae natural isolates, only the 12 pathogenic strains contain the retron Vc95 (personal communication). This finding raises an intriguing question of a possible role for this retron in pathogenicity or a possible association of the retron with the V. cholerae pathogenicity island. However, much more needs to be learned about the interaction of msDNA with the parent cell. For example, under what natural conditions would msDNAs have a mutagenic effect on the host cell? Can msDNA and the retron RT act like SINE elements and generate repetitive sequences in the host chromosome? The study of a structurally bizarre DNA of unknown function may, at times, seem to be a foolish pursuit. But the consistent ability of msDNA to surprise and reveal the unexpected has been its greatest lure. The greatest reward, perhaps, may be the potential practical applications for this DNA. msDNAs are a surprisingly versatile system for constructing artificial DNA sequences and could prove to be powerful tools for the production of antisense DNAs in a variety of cells, both bacterial and eukaryotic, msDNA may find other uses, perhaps as a tool to control the expression of a regulon (by inactivating a repressor for example) or as a novel type of mutagen.
ACKNOWLEDGMENTS
B. C. L.wassupportedbya grantfromthe OfficeofResearchat EastTennesseeStateUniversity. S. I. was supportedby a grant from the U.S. PublicHealth Service(GM 26843).
REFERENCES 1. T. Yee, T. Furuichi, S. Inouye,and M. Inouye,Cell (Canubridge, Mass. ) 38, 203 (1984). 2. T. Furuichi, A. Dhundale, M. Inouye,and S. Inouye,Cell (Cambridge, Mass.) 48, 47 (1987). 3. T. Furuichi, S. Inouye,and M. Inouye,Cell (Cambridge, Mass.) 48, 55 (1987).
90
BERT LAMPSON ET AL.
4. A. Dhundale, B. Lampson, T. Furuichi, M. Inouye, and S. Inouye, Cell (Cambridge, Mass.) 51, 1105 (1987). 5. H.-W. Chen, I. M. Keseler, and L. J. Shimkets,J. Bacteriol. 172, 4206 (1990). 6. M.-Y. Hsu, C. Xu, M. Inouye, and S. Inouye,J. Bacteriol. 174, 2384 (1992). 7. S. A. Rice and B. C. Lampson, J. Bacteriol. 177, 37 (1995). 8. T. Shimamoto, M. Kobayashi, T. Tsuchiya, S. Shinoda, H. Kawakami, S. Inouye, and M Inouye, Mol. Microbiol. 34, 631 (1999). 9. A. Dhundale, M. Inouye, and S. Inouye,J. Biol. Chem. 263, 9055 (1988). 10. S. Inouye, P. J. Herzer, and M. Inouye, Proc. Natl. Acad. Sci. U.S.A. 87, 942 (1990). 11. M. Viswanathan, M. Inouye, and S. Inouye, J. Biol. Chem. 264, 13665 (1989). 12. B. C. Lampson, M. Viswanathan, M. Inouye, and S. Inouye, J. Biol. Chem. 265, 8490 (1990). 13. T. Shimamoto, M. Shimada, M. Inouye, and S. Inouye, J. Bacteriol. 177, 264 (1995). 14. T. M. O. Lima and D. Lim, Plasmid 33, 235 (1995). 15. D. IAm, Mol. Microbiol. 6, 3531 (1992). 16. T. M. O. Lima and D Lira, Plasmid 38, 25 (1997). 17. K. Kim, D. Jeong, and D. Lim, J. Bacteriol. 179, 6518 (1997). 18. B. C. Lampson, K. J. Sun, M.-Y. Hsu, J. Vallejo-Ramirez, S. Inouye, and M. Inouye, Science 243, 1033 (1989). 19. ]~:J. Herzer, S. Inouye, and M. Inouye, Mol. Microbial. 6, 345 (1992). 20. J. Sun, M. Inouye, and L. Inouye, J. Bacteriol. 173, 4171 (1991). 21. M.-Y. Hsu, M. Inouye, and S. Inouye, Proc. Natl. Acad. Sci. U.S.A. 87, 9454 (1990). 22. D. Lira andW. K. Maas, Mol. Microbiol. 4, 2201 (1990). 23. D. Lira, Mol. Microbiol. 5, 1863 (1991). 24. D. Lira, Plasmid 34, 58 (1995). 25. I. B. Dodd and J. B. Egan, Virology 219, 115 (1996). 26. J. Kirchner, D. Lira, E. M. Witkin, N. Garvey, and V. Roegner-Maniscalco, Mol. Microbiol. 6, 2815 (1992). 27. D. Lira andW. K. Maas, Cell (Cambridge, Mass.) 56, 891 (1989). 28. P. J. Herzer, J. Baeteriol. 178, 4438 (1996). 29. M.-Y. Hsu, S. Inouye, and M. Inouye, J. Biol. Chem. 264, 6214 (1989). 30. T. Shimamoto, M.-Y. Hsu, S. Inouye, and M. Inouye, J. Biol. Chem. 268, 2684 (1993). 31. T. Shimamoto, M. Inouye, and S. Inouye,J. Biol. Chem. 270, 581 (1995). 32. B. C. Lampson, M. Inouye, and S. Inouye, Cell (Cambridge, Mass.) 56, 701 (1989). 33. T. Shimamoto, H. Kawanishi, T. Tsuchiya, S. Inouye, and M. Inouye, J. Bacteriol. 180, 2999 (1998). 34. T. H. Eickbush, in "The Evolutionary Biology of Viruses" (S. S. Morse, ed.), p. 121. Raven Press, New York, (1994). 35. A.J. Flavell, Comp. Biochem. Physiol. ll0B, 3 (1995). 36. M. Shimada, H. Hosaka, H. Takaku, J. S. Smith, M. J. Roth, S. Inouye, and M. Inouye, J. Biol. Chem. 269, 3925 (1994). 37. M. E Singer, J. Biol. Chem. 270, 24623 (1995). 38. Y. Xiong and T. H. Eiekbush, EMBOJ 9, 3353 (1990). 39. S. Inouye and M. Inouye, Virus Genes 11:2/3, 81 (1996). 40. T. A. Steitz, J. Biol. Chem. 274, 17395 (1999). 41. S. Inouye, M.-Y. Hsu, A. Xu, and M. Inouye, J. Biol. Chem. 274, 31236 (1999). 42. D. Lira, T. A. Gomes, andW. K. Maas, MoI. Microbiol. 4, 1711 (1990). 43. J. Sun, P. J. Herzer, M. P. Weinstein, B. C. Lampson, M. Inouye, and S. Inouye, Proc. Natl. Acad. Sci. U.S.A. 86, 7208 (1989). 44. E J. Herzer, L. Inouye, M. Inouye, and T. S. Whittam,J. Bacteriol. 172, 6175 (1990).
THE msDNAs OF BACTERIA
91
45. R. K. Selander, D. A. Caugant, and T. S. Whittam, in "'Escherichia coli and Salmonella typhimurium" ( F. C. Neidhardt, ed.), p. 1625. American Society for Microbiology,Washington,
D.C., (1987). S. A. Rice, J. Bieber, J.-Y. Chun, G. Stacey, and B. C. Lampson, J. Bacteriol. 175, 4250 (1993). T. Kawaguchi, P. J. Herzer, M. Inouye, and S. Inouye, Mol. Microbiol. 6, 355 (1992). S. Inouye, M. G. Sunshine, E. W. Six, and M. Inouye, Science 252, 969 (1991). A. Dhundale, T. Furuichi, S. Inouye, and M. Inouye, J. Bacteriol. 164, 914 (1985). B. C. Lampson, M. lnouye, and S. Inouye, J. Bacteriol. 173, 5363 (1991). S. Inouye, M.-Y. Hsu, S. Eagle, and M. Inonye, Cell (Cambridge, Mass.) 56, 709 (1989). W. K. Maas, C. Wang, T. Lima, G. Zubay, and D. Lim, Mol. Microbiol. 14, 437 (1994). W. K. Maas, C. Wang, T. Lima, A. Hach, and D. Lim, Mol. Microbiol. 19, 505 (1996). J.-R. Mao, S. Inouye, and M. Inouye, FEMS Micro. Lett. 144, 109 (1996). C. Rayssiguier, D. S. Thaler, and M. Radman, Nature (London) 342, 369 (1989). G. Feng, H.-C. T. Tsui, and M. E. Winkler, J. Bacteriol. 178, 2388 (1996). B. C. Lampson and S. A. Rice, Mol. Microbiol. 23, 813 (1997). P. Nouvel, Genetica 93, 191 (1994). 59. S. Fujitani, T. Komano, and S. Inouye,J. Bacteriol. 173, 2125 (1991). 60. J. R. Lupski and G. M. Weinstock, J. Bacteriol. 174, 4525 (1992). 61. M. Shimada, S. Inouye, and M. Inouye, J. Biol. Chem. 269, 14553 (1994). 62. J.-R. Mao, M. Shimada, S. Inouye, and M. Inouye, J. Biol. Chem. 270, 19684 (1995). 63. J. Coleman, P. J. Green, and M. Inouye, Cell (Cambridge, Mass.) 37, 429 (1984). 64. S. Miyata, A. Ohshima, S. Inouye, and M. Inouye, Proc. Natl. Acad. Sci. U.S.A. 89, 5735 (1992). 6,5. O. Mirochnitchenko, S. Inouye, and M. Inouye, J. Biol. Chem. 269, 2380 (1994). 66. M. Inouye, J. R. Mao, T. Shimamoto, and S. Inouye, Ciba Found. Syrup. 209, 224 (1997). 67. M.-Y. Hsu, S. G. Eagle, M. Inouye, and S. Inouye,J. Biol. Chem. 267, 13823 (1992).
46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58.
Cisplatin: From DNA Damage to Cancer Chemotherapy SETH M. COHEN AND S T E P H E N J. L I P P A R D 1
Department of Chemistmj Massachusetts Institute of Technology Cambridge, Massachusetts 02139 I. Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Kinetic and Thermodynamic Aspects of Protein Binding to Cisplatin-DNA Adducts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Proteins That Bind to Cisplatin-Modified D N A . . . . . . . . . . . . . . . . . . . . B. H M G - D o m a i n Protein Binding to Cisplatin Adducts: Thermodynamic Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. H M G - D o m a i n Protein Binding to Cisplatin Adducts: Kinetic Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III. Structure of an H M G D o m a i n - C i s p l a t i n - D N A Ternary Complex . . . . . . . . A. Cisplatin-DNA Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. H M G i D o m A - C i s p l a t i n - D N A Structure . . . . . . . . . . . . . . . . . . . . . . . . . C. Mutant H M G - D o m a i n Protein Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Other Consequences of Cisplatin Treatment of Cells . . . . . . . . . . . . . . . . . . . A. Telomere Shortening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. TATA-Binding, H1, and AAG Proteins . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Apoptosis and Ubiquitination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Development of New Platinum Compounds . . . . . . . . . . . . . . . . . . . . . . . . . . A. Mononuclear Platinum Complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Polynuclear Platinum Complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Combinatorial/Parallel Synthesis Approaches . . . . . . . . . . . . . . . . . . . . . . D. Screening Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Steroid Hormones, Cisplatin, and Cancer . . . . . . . . . . . . . . . . . . . . . . . . . VI. Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
94 98 98 99 100 104 104 107 110 112 112 114 116 117 117 119 120
122 122 125 126
Cisplatin [c/s-DDP, cis-diamminedichloroplatinum(II)] is a potent antieancer drug that has been used successfully to treat tumors of the head, neck, lungs, and genitourinary tract. The biological activity of eisplatin was discovered serendipitously more than 30 years ago, and since that time research efforts have focused on elucidating its mechanism of action. The present review provides a historical perspective of our attempts to understand this complex phenomenon and the results of recent work that guides our current activities in this field. Continued 1Author to w h o m correspondence should be addressed. Telephone: (617) 253-1892. Fax: (617) 258-8150. E-maih [email protected]. Progressin NucleicAcid Research and MolecularBiology,Vol.67
93
Copyright© 2001 by AcademicPress. All rightsof reproductionin an),formreserved. 0079-6603/01 $35.0(I
94
SETH M. COHEN AND STEPHEN J. LIPPARD efforts to understand the mechanism of genotoxicity of cisplatin are expected to lead to the discovery of new drugs and combinations for the improvement of cancer chemotherapy. © 2001 AcademicPress.
Cisplatin [cis-diamminedichloroplatinum(II), cis-DDP] is a relatively simple inorganic compound that has had a major impact on the treatment of cancer (Fig. 1). The biological activity of cisplatin was discovered serendipitously about 125 years after its first reported synthesis (1). Its utility in the treatment of cancer is a paradigm for the use of metals in medicine (2). Today, cisplatin and carboplatin (Fig. 1), an analog of cisplatin, are among the most widely prescribed anticancer drugs, with annual sales exceeding $500 million (3). Several recent reviews are available describing many of the chemical and biological studies performed to elucidate the genotoxic mechanism of cisplatin. These reports cover a broad range of subjects including the identification of DNA as the target (4), structures of cisplatin-DNA adducts (5), interactions of various proteins with cisplatin-modified DNA (6, 7), cellular consequences of protein binding (8), and the development of new platinum anticancer drugs (3). The sum of these investigations, which are thoroughly examined in these reviews, provides the basis for several hypotheses to explain how the drug elicits its antitumor effect. Because these many excellent reviews have appeared recently (3-8), this article focuses on the most current discoveries from our laboratory. A brief background and perspective are provided, followed by several subjects, including protein recognition of cisplatin-damaged DNA, structural studies on proteincisplatin-DNA ternary complexes, newly uncovered cytotoxic consequences of cisplatin treatment, and parallel synthesis efforts toward the development of novel platinum drugs. This range of topics both communicates the latest developments in understanding the mechanism of action of cisplatin and outlines future prospects for cisplatin-related research.
I. Background The potential utility of platinum compounds as anticancer agents was first recognized following the demonstration of their ability to prevent bacterial cell division (9). Subsequent investigations identified cisplatin as the most active compound (10-12), and a decade of clinical trials ultimately led to FDA approval in 1978. Cisplatin is used to treat a variety of cancers, including tumors of the head, neck, lung, and genitourinary tract. Its greatest success is against testicular cancer, a previously lethal disease with historically low survival rates before the introduction of cisplatin (13).
BIOCHEMISTRY OF CISPLATIN
95 0
H3N
CI
H3N
cisplatin
0%
"'/
CI/
carboplatin
~NH3
trans-DDP
[Cr]extracellular = - 100mM
CellMembrane
[Cl']intracellular
=
-2-30 mM
DNA
Aquation
H3N~ /OH2~ 2+
/Pt
RNA
H3N ~OH2 Proteins//
Glutathione
FIG. 1. Stru•tures•f•is••atin(c/s-DDE•eft)••arb•••atin(•enter)•andtheinactivetransis•mer trans-DDP (right). The general scheme for the in vivo reactivity of cisplatin is depicted. Cisplatin and carboplatin are administered intravenouslywhere extracellular chloride levels are high. Upon entering cells, the large drop in chloride ion concentration causes the platinum compounds to undergo aquation reactions. Exchange of the anionic chloride or 1,1-cyclobutanedicarboxylateligands for water molecules gives a platinum complex with a positive charge that migrates to polyanionic nucleic acids. The platinum complex will react with many cellular targets, including proteins and glutathione, but the ~1% of the platinum that reaches the nuclear DNA is responsible for the genotoxicityof the compound. O f interest to inorganic chemists was the observation that the trans isomer, trans-diamminedichloroplatinum(II) or t r a n s - D D P (Fig. 1), has no activity against tumors. A simple difference in stereochemistry conveys dramatically different levels of activity. While the clinical potential of cisplatin was being evaluated, our laboratory was investigating the interaction of heavy-metal
96
SETH M. COHEN AND STEPHEN J. LIPPARD
complexes with nucleic acids (14, 15 ). One objective of these experiments was to sequence DNA in the electron microscope using a "beads-on-a-string" concept afforded by specific derivatization of nucleotides with electron-dense heavymetal labels (16). During these investigations, platinum compounds were identified that bound to DNA through only covalent, covalent and intercalative, or only intercalative modes (17). The last class of compounds included several platinum terpyridine complexes (18, 19) that afforded the first metallointercalators. The descendents of these early compounds constitute a substantial area of active chemical research today (20). When sufficient evidence had accumulated to identify DNA as the relevant biological target of cisplatin (4), our experience in platinum-nucleic acid chemistry prompted us to pursue the stereochemistry issue. Both cis- and trans-DDP shorten and unwind supercoiled circular DNA (21, 22), binding almost exclusively at the N7 nitrogen atom in purine bases (23), with a preference for guanine over adenine (24). Several types of DNA adducts occur with both platinum compounds, including monofunctional, intrastrand crosslinks, and interstrand crosslinks. The active compound, c/s-DDP, forms predominantly 1,2-intrastrand d(GpG) and d(ApG) crosslinks and, to a lesser extent, 1,3-intrastrand and interstrand crosslinks. The inactive compound, trans-DDP, forms predominantly 1,3-intrastrand and interstrand crosslinks. These discoveries suggested that the toxicity of cisplatin originated from the 1,2-intrastrand crosslinks, which the geometry of trans-DDP does not allow (Fig. 2). Indeed, subsequent work indicated that a positive response of patients to cisplatin correlates with the number of 1,2-intrastrand crosslinks found in tissue samples (25).
FIG. 2. Scheme of different crosslinks formed on DNA by platinum drugs. Several types of adducts are formed, including (in order from left to right) 1,2-intrastrand crosslinks, interstrand crosslinks, monofunctional adducts, and DNA-protein crosslinks. Cisplatin forms predominantly 1,2-intrastrand crosslinks, and these adducts are believed to be the most important lesions for the genotoxiceffectsof the drug.
BIOCHEMISTRYOF CISPLATIN
97
Once the basis for the stereoehemical preference for cis-DDP had been established, the focus of our work turned to investigating the structure, biochemical consequences, and cellular responses of cisplatin damage to DNA. Cisplatin 1,2-intrastrand erosslinks inhibit both replication (26, 2 7) and transcription (28-31). Of primary importance was the discovery of structure-specific recognition protein 1 (SSRP1), a mammalian cellular factor that binds with specificity to cisplatin-modified DNA (32, 33). This protein contains a high-mobility group (H MG) D NA-binding motif (34). Additional studies demonstrated that a range of HMG-domain proteins, including transcription and chromatin architectural factors, could bind to cisplatin-damaged DNA (35-37). HMG-domain proteins specifically recognize the 1,2-intrastrand crosslinks formed by eisplatin, but not adduets formed by trans-DDP, supporting the hypothesis that 1,2-intrastrand adduets were significant in the mechanism of action. The ability of HMG-domain proteins to potentiate the action of cisplatin in vivo is supported by experiments showing that yeast knockout mutants lacking the HMG-domain protein Ixrl, intrastrand crosslink recognition protein 1, are more resistant to eisplatin than are wild-type cells producing the protein (38-40). The dual discoveries that HMG-domain proteins could bind with high affinity to eisplatin 1,2-intrastrand crosslinks and that this recognition could elicit an observable effect in cells treated with cisplatin were evaluated within the framework of various hypotheses for how these proteins might mediate the genotoxieity of the drug (41, 42). One possible pathway by which cisplatin might effect genotoxicity is through "repair shielding" (38). Cisplatin adduets are removed primarily by enzymes of the nucleotide excision repair (NER) pathway (43, 44). When HMG-domain proteins such as HMG1 bind to eisplatin lesions, however, NER can be blocked (43, 44), allowing the damage to persist. The resulting cisplatin-DNA adduets block replication and transcription. Another mechanism by which cisplatin can damage the cell is through the "hijacking" of nuclear factors essential for cellular function. Binding of such factors to cisplatin-modified DNA diverts them from their natural sites, preventing these proteins from performing their critical tasks. SSRP1 has recently been identified as one component of a heterodimeric protein complex that facilitates chromatin transcription (FACT), allowing RNA polymerase II to transcribe through nucleosomes (45). Because SSRP1 binds to cisplatin-DNA adduets, FACT may recognize the damage through its SSRP1 HMG domain, diverting the protein complex from its normal sites. The result would be inhibition of transcription elongation by RNA polymerase II. The repair-shielding and protein-hijacking hypotheses are two possible ways that nuclear proteins may trigger cell death following eisplatin damage. In the following sections we describe recent results that address these hypotheses and introduce evidence suggesting that other pathways may also contribute to the biological activity of cisplatin.
98
SETH M. COHEN AND STEPHEN J. LIPPARD
II. Kinetic and Thermodynamic Aspects of Protein Binding to Cisplatin-DNA Adducts A. Proteins That Bind to Cisplatin-Modified DNA More than a dozen proteins have been identified that bind to cisplatinmodified DNA (Table I), recognizing specifically the 1,2-intrastrand crosslink. SSRP1, the first such protein to be identified, belongs to the HMG family of DNA-binding proteins. HMG-domain proteins are broadly divided into two groups, sequence-specific and sequence-neutral (34). Sequence-specific HMGdomain proteins are usually found only in cells of a particular origin, whereas sequence-neutral proteins are more widely distributed. The biological functions of HMG-domain proteins are diverse (46) and include chromatin remodeling, transcription initiation, and RNA polymerase nucleosome bypass (45). Essentially all structure-specific HMG-domain proteins bind to cisplatin 1,2intrastrand crosslinks. The multifaceted nature of this protein family suggests that many different critical cellular functions may be interrupted as a result of HMG protein binding to cisplatin-DNA adducts.
TABLE I SELECTEDPROTEINSTHAT BIND TO CISPLATIN-MODIFIEDDNA
Abbreviation
Protein name
AAG (MPG, ANPG)
3-Methyladenine DNA glycosylase Histone linker protein High mobility group protein 1 High mobility group protein 2 Intrastrand crosslink recognition protein 1 Lymphoid enhancer-binding factor 1 Replication protein A Sex-determining factor Y Structure-specific recognition protein 1 TATA-binding protein Upstream binding factor Xeroderma pigmentosum A complementing protein Y-box binding protein 1
H1
HMG1 HMG2 Ixrl LEF- 1 RPA (SSB) SRY SSRP1 TBP hUBF XPA YB-1
Function
HMGdomain
Repair protein
No
Chromatin structure Chromatin structure, remodeling Chromatin structure, remodeling Transcription regulation
No Yes Yes Yes
Transcription factor
Yes
Nucleotide excision repair Transcription factor Transcription elongation
No Yes Yes
Transcription initiation Ribosomal RNA transcription factor Nucleotide excision repair
No Yes No
Transcription factor
No
Reprinted with permissionfrom E. R. Jamiesonand S. J. Lippard, Chem. 1999 AmericanChemicalSociety.
Rev.
99, 2467-2498 (1999). Copyright©
BIOCHEMISTRYOF CISPLATIN
99
In addition to HMG-domain proteins, several other protein classes bind to cisplatin-damaged DNA. Included are proteins involved in DNA damage recognition, particularly those involved in transcription-coupled NER. An extensive review of this subject has recently been published (6). Transcription factors, including YB-1 (47) and the TATA-binding protein (TBP), also bind to cisplatindamaged DNA (48, 49). TBP is particularly interesting because the protein is essential for RNA polymerase II transcription initiation in eukaryotes (50). The chromatin histone protein H 1, critical for DNA packaging and storage, also binds to cisplatin lesions (51). Clearly, many important proteins outside the HMG family also recognize cisplatin damage, providing even more potential pathways by which the drug may exert its antitumor effect. Work in our laboratory has focused largely on the HMG-domain protein family because of the high selectivity for cisplatin crosslinks and importance in various cellular functions.
B. HMG-Domain Protein Binding to Cisplatin Adducts: Thermodynamic Considerations Several studies have been performed to elucidate the strength and specificity of HMG-domain protein binding to cisplatin-DNA adducts. The majority of these studies have focused on HMG1, a nuclear protein that contains two tandem HMG domains designated domain A and domain B (Fig. 3). The functions of HMG1 are not fully understood, but most likely involve activation of transcription and chromatin remodeling (46). Electophoretic mobility shift assay (EMSA or bandshift assay) investigations have revealed many important features of HMG1, HMG1 domain A (HMGldomA), and HMG1 domain B (HMGldomB) binding to cisplatin crosslinks. All three proteins bind specifically to 1,2-intrastrand d(ApG) and d(GpG) adducts, but not to 1,3-intrastrand crosslinks or to interstrand crosslinks. The affinity of these proteins for cisplatin-DNA adducts is influenced by the sequences flanking the crosslink. In a sequence containing N1G* G'N2, where G'G* represents a 1,2-intrastrand crosslink, the affinity of HMG-domain protein binding is modulated by the bases in the N1 and N2 positions (52). The selectivity of binding was initially examined in 15-bp oligonucleotides, having 9 of the 16 possible flanking sequences, where N1, N2 = A, C, or T. Synthetic difficulties hampered the preparation of oligonucleotides containing more than two guanosine nucleosides in a row, but later efforts provided access to the remaining seven sequences having either natural guanosine or unnatural 7-deazaguanosine, an isostere that lacks the reactive N7 nitrogen atom in the purine base, in the flanking positions (53). This series of experiments reveals several interesting features of HMGdomain protein binding (Fig. 4). First, the flanking sequence selectivity of
100
SETH M. COHEN AND STEPHEN J. LIPPARD
i0
HMGIdomA: HMGIdomB: LEF-I: hSRY:
298 H I K K P - - - L N A F M L Y M K E ~ A E C T
50
- -LKESAAINQILGRRWHAL
710
810
SAKEKGK F EDMAKADKARYEREMKTY I PPKGETKKKF A A D D K Q P Y E K K A A K L K E K Y E K D IAAYRAK SREEQAKYYELARKERQLHMQLYPGWSARDNYGKKKKRKREK TEAEKWPFFQEAQKLQAMHREKYPNYKYRP
76 box A
40
III 610
1
30
57QDRVKRP-- -MNAF IVWSRDQRRKMALENP - -RMB_NSEI SKQLGYQWKML
g HMGIdomA: HMGIdomB: LEF-I: hSRY:
20
IMGKGD PKKP~GKMS SYAFFVQTCREEHKKKHPDASVNF S ~F SKKC S ERWKTM 86KKKFKDPNAPKR P___ PSAFFLFC SEYRPKI KGEHP--GLS IGDVAKKLGEMWNNT
I II
I
164
88
bo×B
185 --
NH3 ÷ C O 2-
214
acidic C-tail
FIG. 3. General structure (lower right) and sequence alignments of several HMG-domain proteins. The structure is representative of the common a-helical L-shaped motif found in HMGdomain proteins. Each of the three helices is labeled together with the N and C termini. Sequence alignments compare four HMG domains: domains A and B of the architectural protein HMG1, the domain of the transcription factor LEF-1, and the domain of the sex-determining transcription factor hSRY. A map of the protein HMG1 is shown at the bottom. Adapted with permission from S. U. Dunham and S. J. Lippard, Biochemistry36, 11428-11436 (1997). Copyright © 1997 American Chemical Society.
H M G l d o m A is much more p r o n o u n c e d than that of H M G l d o m B . Second, H M G l d o m A binding is most strongly influenced by the base pair at the N2 position, with A/T base pairs being preferred over C / G base pairs in both the N1 and N2 positions. Finally, studies using mutant H M G - d o m a i n s show that base-specific contacts (vide infra) are important for protein affinity, but not sequence selectivity. Because H M G - d o m a i n proteins b e n d the helix upon binding to D N A (54, 55), a preference for A/T over C / G base pairs might be the result of the increased flexibility of A/T base pairs (56). This conclusion is supported by thermodynamic studies of AG*G*A and CG*G*C 15-bp oligonucleotides, which indicate the former duplex to be about 4-fold less destabilized than the latter (57).
C. HMG-Domain Protein Binding to Cisplatin Adducts: Kinetic Considerations As indicated previously, H M G - d o m a i n proteins may play an important role in shielding cisplatin adducts from N E R (43, 44). The ability of H M G 1 to block
BIOCHEMISTRY OF CISPLATIN
101
0.70
r [] 0.60
mmaA|l
0.50 0.40 O 0.30
0.20 0.10 0.0
A,A A,T A,C A,G__T,A T,T T,C T,G C,A C,T C,C C,G G__,AG,T G_,C G,G__
Flanking Bases (N~, N 2)
5'-CCT 3'-GGA
C T C N I G * G * N 2 T C T TC-3' G A G N3C C N 4 A G A AG-5'
FIG. 4. Sequence selectivity of HMG-domain proteins for eisplatin adducts determined by electrophoretic mobility shift assay (EMSA) (52, 53). Normalized quantitation of HMG-domain binding to cisplatin adducts with different flanking sequences is shown for HMGldomA in open bars and HMGldomB in filled bars. The sequence of the 15-bp oligonucleotide used for these studies is shown at the bottom of the figure. G'G* represents a cisplatin 1,2-intrastrand crosslink and G represents 7-deazaguanosine. (9 represents the ratio of protein-bound DNA to free DNA.
NER requires that it be kinetically and thermodynamicallycompetent to bind to eisplatin-DNA adduets in eompetition with repair proteins. Bandshift titrations demonstrate that binding of HMG-domain proteins ean occur with subnanomolar Ka values in vitro, demonstrating high thermodynamic affinity for cisplatin adduets (52, 58-61). Also, bandshift studies with HMG1 and the repair protein RPA, both of which independently bind to eisplatin-modified DNA, indicate that HMG1 bound to eisplatin-modified DNA blocks formation of the RPA complex (62). One possible explanation for the ability of HMG1 to prevent RPA binding is that HMG1 more rapidly recognizes and binds to eisplatin-DNA lesions. Recent stopped-flow experiments (63), using either a fluoreseently labeled base or an oligonueleotide probe exhibiting fluoreseenee resonance energy transfer (FRET) (64), evaluated the kinetics of HMG-domain binding to cisplatinmodified DNA. In the latter probe a 20-bp oligonueleotide was labeled with a fluorescent donor-aeeeptor pair (64). The 5' end of the platinated strand was
109.
SETH M. COHEN AND STEPHEN J. LIPPARD
linked to a rhodamine acceptor and a fluorescein donor was installed on the 5' end of the complementary strand (Fig. 5). These dyes can undergo FRET upon excitation of the fluorescein at 480 nm, which then undergoes nonradiative energy transfer to the rhodamine, which in turn can relax by fluorescent emission (imax = 580 rim). The ratio ofrhodamine to fluorescein emission increases as the dyes are brought closer together. Because HMG1 binding to cisplatin-modified DNA increases the bending of the DNA double helix (54), this binding event brings the dyes closer together. The change in FRET allows for direct monitoring of protein binding, affording both structural and kinetic information. With the use of a suitable model, changes in the fluorescent lifetime of the fluorescein donor permit the determination of the helix bend angle. Coupling of this methodology with stopped-flow techniques provided the first insight into the kinetics of HMG-domain binding to cisplatin-modified DNA. The experiments indicate that HMGldomB binds to the cisplatinmodified FRET probe with a second-order kon value of 1.1 • 0.1 × 109 M-is -1 and a first-order koff value of 30 + 4 s-~. These kinetic data confirm the high affinity of HMGldomB for cisplatin adducts, the calculated Kd value being 27 ~: 4 nM. More importantly, the results indicate that HMGldomB binds to cisplatin-modified DNA extremely rapidly, with a rate constant that is near the diffusion limit. Although the complex is kinetically labile, both kon and koffbeing
[~COOH
[~COOH
/ ~ N,,'~,O H
/'~.N,,~O H
FI= FluoresceinDonor
Rh= RhodamineAcceptor 5'-Rh-TCT 3' - A G A
CCT GGA
TCT AGA
GGT CCA
CTC GAG
TTC TC 3' AAG AG-FI-5'
5'-Rh-TCT 3' - A G A
CCT TCT G*G*T CTC TTC TC - 3' GGA AGA C C A GAG AAG AG-FI-5'
FIG. 5. Structure of the rhodamine acceptor (top left) and donor fluorescein dyes (top right) appended to DNA for fluorescence resonance energy transfer (FRET) experiments (64). The sequences of the platinated and unplatinated oligonucleotide probes used in the FRET studies are shown at the bottom of the figure. G* G* represents a cisplatin 1,2-intrastrand crosslink.
103
BIOCHEMISTRY OF CISPLATIN
rather large, it is clear that HMG-domain proteins can bind rapidly to cisplatin adducts, as would be required to protect them from repair proteins. These FRET results were further elaborated by fluorescence studies using a platinum-modified oligonucleotide probe that contained a nucleoside modified with a fluorescein tag (63). Seven 16-bp oligonucleotide probes were prepared containing a single cisplatin 1,2-intrastrand d(GpG) crosslink and a fluoresceinmodified deoxyuridine located at various positions along the platinated strand (Fig. 6). The largest changes in fluorescence were obtained when the fluoresceinmodified deoxyuridine was placed two bases to the 5' side of the cisplatin adduct. Larger fluorescence changes occurred with HMGldomA and full-length HMG1 than with HMGldomB. The experiments with fluorescein-modified deoxyuridine demonstrate that HMGldomA, HMGldomB, and full-length HMG1 bind to cisplatin-modified DNA with nearly diffusion-limited kol~values(63). The association rate constant of HMGldomA is independent of oligonucleotide length, but depends on the sequences immediately flanking the DNA adduct (Fig. 6). The stopped-flow
Ho.
0
¢,..~-~C00 H
U FL
o
o I
5 ' - C C T cuFLc TG*G*A C C T TCC-3' 3'-GGA GAG
A C C T G G A AGG-5'
5 ' - C C T cuFLc TG*G*T C C T TCC-3' 3 ' - G G A G A G A C C A G G A AGG-5' 5 ' - C C T cuFLc TG*G*C C C T TCC-3' 3 ' - G G A G A G A C C G G G A AGG-5'
FIG. 6. Top: Chemical structure of the fluorescein-modified nucleoside used in fluorescence stopped-flow kinetics experiments (63). Bottom: Sequence of three of the seven oligonucleotide probes used to evaluate the kinetics of HMG-domain protein binding to cisplatin-modified DNA. G* G* represents a cisplatin 1,2-intrastrand crosslink.
104
SETH M. C O H E N AND STEPHEN J. LIPPARD
TABLE II KINETIC PARAMETERS FOR HMG-DOMAIN PROTEINS BINDING TO CISPLATIN-MODIFIED PROBES CONTAINING FLUORESCEIN-MODIFIED DEOXYURIDINE (63) (T = 4°C) Protein HMG1 HMG1 HMG1 HMG1 HMG1 HMG1
domain domain domain domain domain domain
Sequence a A A A B B B
TG*G*A TG*G*T TG*G*C TG*G*A TG*G*T TG*G*C
kon (M-Is -1) 9.0 4- 2.0 2.54-1.0 4.04-1.0 1.9 4-1.3 2.4 4-1.5 2.7 4-1.5
x x x x x x
107 107 107 107 107 107
koff (s -1)
Kd
30 4-11 354-4 264-5 86 4- 56 73 4- 34 77 4- 50
333 4-134 nM 1.44-0.6 ttM 6504-205 nM 453 4- 428 nM 304 4- 237 nM 609 4- 430 nM
%equence shown is only the flanking sequence around the platinum crosslink. The complete sequence of the 16-bp oligonucleotide probes are given in Fig. 6.
fluorescence studies confirm earlier bandshift results indicating that the base to the 3' side of the crosslink strongly modulates the affinity of binding for HMGldomA, and that HMGldomB is substantially less perturbed by the nature of the flanking sequence (52). In addition, the data suggest that, beyond the bases directly flanking the lesion, the effect of sequence context is negligible. The studies reveal that changes in kon, not koff, control the affinity of HMG-domain proteins for cisplatin-DNA adducts (Table II). These kinetic studies show that HMG-domain proteins rapidly form kinetically labile, but thermodynamically stable, complexes with cisplatin-DNA lesions. The association rate constants are near the diffusion limit, suggesting that HMG proteins will bind as fast as, or more rapidly than, any other nuclear proteins. The results clearly establish that HMG-domain proteins are both kinetically and thermodynamically competent to shield platinum 1,2-intrastrand crosslinks from recognition by repair proteins.
III. Structure of an HMG Domain-Cisplatin-DNA Ternary Complex A. Cisplatin-DNA Structures The interpretation of biochemical experiments can be greatly facilitated by structural information. Once DNA, and more specifically the d(GpG) dinucleotide base pair, was established as the primary target for cisplatin, efforts were made to elucidate both the solution and solid-state structures of cisplatin 1,2-intrastrand crosslinks. The first X-ray structural insights into the nature of the platinum adduct came with the determination of cisplatin bound to the
BIOCHEMISTRY OF CISPLATIN
105
dinucleotide d(pGpG) (65). The use of pGpG versus GpG was important because the former allows a neutral species to form upon platination and facilitates important hydrogen-bonding interactions that stabilize the complex. The structure of the dinucleotide adduct confirms that the platinum atom binds to the N7 nitrogen atoms of the guanine bases (Fig. 7). The bases are oriented in a head-to-head fashion with both 06 atoms on the same side of the platinum coordination plane. Once the core structure of the 1,2-intrastrand adduct was known, further structural studies focused on longer fragments of DNA. The X-ray structure of a dodecanucleotide duplex containing a single 1,2-intrastrand erosslink was solved, revealing how platination affects the global double-stranded DNA structure (66, 67). This structure displays a large bend in the DNA helix, a shallow and widened minor groove, and a kink in the DNA at the site of platinum binding. Surprisingly, the local structure around the platinum atom and the two bound guanosines is quite different for the dodecanucleotide duplex when compared with the earlier d(pGpG) structure (vide infra). Several other investigations have elucidated the solution structure of duplexes modified with cisplatin, utilizing two-dimensional NMR methods (68) as well as long-range electron-proton restraints afforded by a nitroxide spin-labeled platinum complex (69). The solution studies confirm the overall features found in the solid
FIG. 7. X-Ray crystal structure of the cisplatin d(pGpG) dinucleotide erosslink (65). Carbon atoms are shaded in light gray and heteroatoms are shaded in dark gray. The drug is bound to the dinucleotide by the N7 nitrogen atoms of the guanine bases. The bases are bound in a head-tohead orientation. A hydrogen bond between a phosphate oxygen and a platinum ammine ligand is depicted by a dashed line.
106
SETH M. COHEN AND STEPHEN J. LIPPARD
FIG. 8. Diagrams of the NMR solution structure (left) and X-ray solid-state structure (right) of cisplatin-modified DNA duplexes (66-68). The overall structures are similar, demonstrating a bend in the DNA with a widened, flattened minor groove.
state (Fig. 8), although there are some differences in the specific geometric parameters of each structure (Table III). Additional studies have afforded structures of other platinum-modified oligonucleotides including hairpins (70), 1,3-intrastrand crosslinks (71, 72), interstrand crosslinks (73 -75), and most recently the structure (B. Spingler, D. A. Whittington, and S. J. Lippard, TABLE III GEOMETRIC PARAMETERS FROM SEVERAL CISPLATIN-DNA 1,2-INTRASTRAND CROSSLINK STRUCTURE DETERMINATIONS
Structure 12-mer ll-mer 12-mer 16-mer
duplex duplex duplex duplex c
Method
Minor groove width (A)
Dihedral angle a (deg)
Average helical twist (deg)
DNA bend (deg)
Reference
NMR NMR b X-ray X-ray
9.4-12.5 9.0-12 9.5-11 5.5-12
47 58 30 75
27 26 32 33
78 81 39, 55 61
68 69 66, 67 77
'~Dihedral angle between the cisplatin-modified guanine bases. blncludes long-range electron-proton distance restraints. CCisplatin adduct bound by HMGldomA protein. Reprinted with permission from E. R. Jamieson and S. J. Lippard, Chem. Rev. 99, 2467-2498 (1999). Copyright © 1999 American Chemical Society.
BIOCHEMISTRY OF CISPLATIN
107
unpublished data) of a dodecanucleotide modified with the cisplatin analog,
1,2-(R,R)-diaminocyclohexaneplatinum(II) (76).
B. HMGI DomA-Cisplatin-DNA Structure Although a wealth of structural information is now available for cisplatinD N A aclducts, until recently there were no geometric details about their interactions with nuclear proteins. In order to investigate the features dictating H M G domain protein recognition of cisplatin-DNA adducts, an X-ray crystal structure was determined of H M G l d o m A bound to a 16-bp oligonucleotide containing a single 1,2-intrastrand d(GpG) crosslink (77). The structure was solved by multiple isomorphous displacement methods using two iodinated derivatives. The structure confirmed many expected structural traits, but also contained several surprising features. The H M G domain binds to the minor groove of the D N A double helix (Fig. 9) opposite the platinum adduct located in the major groove, an anticipated finding. The protein is in its normal L-shaped conformation, in good agreement with an N M R solution structure determination (78). An overlay of the structure of protein bound to cisplatin-modified D N A (77) and free in solution (78) gives
FIG. 9. X-Ray crystal structure of HMGldomA bound to a cisplatin-modified DNA duplex. The protein is shown as a light ribbon, the DNA as a darker wire-frame with a ribbon backbone, and the platinum drug as a light wire-frame rendering. The a helices of HMGldomA are labeled I, II, and III. The space-filling portion of the figure is the intercalating Phe 37 residue that fills the hydrophobic notch created between the platinum-modified bases. Note that the protein binds asymmetrically to the adduct from the minor groove of the DNA duplex. Reprinted with permission from U.-M. Ohndorf, M. A. Rould, Q. He, C. O. Pabo, and S. J. Lippard, Nature (London) 399, 708-712 (1999).
108
SETH M. COHEN AND STEPHENJ. LIPPARD
an rmsd of 2.1 A, with the largest differences occurring at the C terminus of helix I, the N terminus of helix II, and the loop between helices I and II (Fig. 9). The structure of the DNA double helix shows distortions similar to those found in cisplatin-modified duplexes (66, 68) and to DNA bound by other HMG-domain proteins (79, 80). Unlike other HMG-domain DNA complexes, however, the bend in the duplex is not congruent with the angle between helix UII and helix III of the protein. Instead, H M G l d o m A binds in an asymmetric fashion, biased toward the 3' side of the platinum adduct. The asymmetric positioning of H M G l d o m A persists in solution, as confirmed in a hydroxyl-radical footprinting experiment (vide infra). The protein makes several hydrogen-bond, salt-bridge, and van der Waals contacts with the DNA duplex through helices I and II. Helix III interacts with the oligonucleotide only through a single water-mediated contact. A complete contact map for the protein-DNA interactions is presented in Fig. 10. The structure of the bound HMG domain reveals an unexpected intercalation of the side chain of Phe 37, which is located in the loop region between
51
31
FIG. 10. Contact map showingthe protein-DNA interactions for HMGldomA bound to a cisplatin-modified16-bpoligonucleotide.Solidlines representhydrogen-bondingcontacts (including water-mediated),dashedlines representvan der Waalshydrophobiccontacts,and the triangular wedge represents intercalationbetweenthe two platinatedbases. Noticethat $41 to the adenosine 3' of the adduct is the onlybase-specificcontact. Reprintedwith permissionfrom U.-M. Ohndorf, M. A. Rould, Q. He, C. O. Pabo, and S. J. Lippard,Nature (London)399, 708-712 (1999).
BIOCHEMISTRYOF CISPLATIN
109
helices I and II, into the "hydrophobic notch" in the minor groove formed between the two cisplatin-bound guanosines. The intercalating residue additionally destacks the adjacent guanines, with a large roll of 61 ° between the bases. The aromatic phenylalanine side chain forms a n-zr stacking interaction with the guanine base on the 3' side of the lesion, at a distance of 3.5 ,~, and an edgeto-face contact with the guanine base on the 5' side of the lesion. This residue is extremely important for protein binding. Mutation from phenylalanine to tryptophan lowers the affinity approximately 5-fold, and replacement by alanine nearly abrogates protein binding (81). Another interesting interaction is a base-specific contact between Ser 41 and the N3 nitrogen atom of the adenine base immediately 3' to the platinum crosslink (Fig. 10). This is the only base-specific contact in the structure, and mutation of this residue from serine to alanine reduces the affinity of the protein by 5-fold compared to the wild type protein. The alanine mutant shows the same flanking sequence selectivity as the wild-type protein (vide supra), indicating that the interaction is important for protein affinity but not specificity (53). This result is surprising, being the only base-specific interaction and involving one of the flanking bases, yet the contact does not contribute to sequence selectivity. Close inspection of the platinum coordination geometry in the ternary complex reveals the ligand environment to be relaxed relative to that in the structure in duplex DNA alone (66, 67). Superposition of the platinum center from the HMG-bound structure and the metal center in the dodecanucleotide duplex (66, 67) shows striking differences (Fig. 11). Most notably, the dihedral angle
FIG. 11. Superpositionof the platinumcoordinationcenter from the HMGldomA-cisplatinDNA ternarycomplexwiththe d(pGpG) structure (right) and the structure of cisplatinbound to a dodecanucleotide(left).The ternarycomplexis shownin light grayin both overlaydiagrams.The structures clearlyrevealthat the ternarycomplexis more similarto the d(pGpG)structurethan the dodecanucleotidestructure. Reprintedwithpermissionfrom U.-M. Ohndorf,M. A. Rould,Q. He, C. O. Pabo, and S. J. Lippard,Nature (London)399, 708-712 (1999).
110
SETH M. COHEN AND STEPHEN J. LIPPARD
between the two bases is "-,30° in the dodecamer structure, but "~75° in the protein-bound structure. In contrast, comparison of the protein-bound structure with that of the [Pt(NH3)2{d(pGpG)}] adduct (65) shows a very similar geometry, with a dihedral angle in the latter being ~77 °. This analysis suggests that the platinum center in duplex DNA is in a strained geometry owing to restrictions imposed by the relatively rigid DNA double helix. After HMG-domain protein binding, this stress is alleviated, allowing the metal center to take on a more relaxed geometry with coordination parameters closely resembling those of platinum bound to an unconstrained dinucleotide. The change in geometry lowers the free energy of complex formation and stabilizes the overall interaction.
C. Mutant HMG-Domain Protein Studies The detailed molecular structure of the HMGldomA-cisplatin-DNA ternary complex prompted a series of mutagenesis experiments to elucidate which interactions modulate protein binding. Some of these mutants, already mentioned, demonstrate that base-specific contacts are not essential for protein sequence selectivity and that the intercalating phenylalanine residue is crucial for high-affinity binding. The primary focus of the additional mutagenesis studies was to elucidate the significance of the loop region between helices I and II, and further to explore position 37 occupied by the intercalating phenylalanine residue. Specifically, the effects on protein affinity and positioning relative to the TABLE IV BINDING AFFINITIESAND ORIENTATIONSOF HMG-DOMAIN PROTEINS WITH CISPLATIN 1,2-INTRASTRANDCROSS-LINKS (81) Protein
Kd (nM)
Orientation a
DumA D u m a AVN DumA F37W DumA A16F DumA A16F F37A DumA F37A DumA $41A DumB DumB iVN DumB I37F DumB F16A DumB 137A DumB F16A 137A
1.5 -4-0.5 5.0 -4-0.9 9.9 + 2.1 16.5 + 5.7 167 -t- 32 > 1000 5.8 -4-0.7 38.7 4- 9.9 39 4- 2.6 19.5 4-13 119 -4-39 85 4- 17 > 1000
Asymmetric n/a n/a Asymmetric Symmetric rda rda Symmetric n/a n/a Asymmetric Symmetric rda
aSymmetric binding displays equal protection, as determined by hydroxyl-radical footprinting, to bases on both sides of the cisplatin-DNA 1,2-intrastrand crosslink. Asymmetricbinding displays protection to more of the bases to the 3' side of the cisplatin-DNA 1,2-intrastrand crosslink.
BIOCHEMISTRYOF CISPLATIN
111
DNA damage were investigated. Six mutants of H M G l d o m A and five mutants of H M G l d o m B (Table IV) were prepared to address these issues (81). In the loop region between helices I and II, HMGldomA contains seven amino acids whereas H M G l d o m B contains only five (see Fig. 3). Because in H M G l d o m A this loop region contains the critical intercalating Phe 37, it was speculated that the difference in loop length might explain the higher affinity of H M G l d o m A versus H M G l d o m B for cisplatin-modified DNA (52, 58). Two mutants were prepared, HMGldomA dxVN and H M G l d o m B iVN, in which the extra valine and asparagine amino acids were deleted or inserted, respectively. The affinity of HMGldomA AVN is about 3-fold less than that of wild-type HMGldomA, and H M G l d o m B iVN demonstrated an affinity for cisplatin-modified DNA that is essentially unchanged compared to wild-type H M G l d o m B (Table IV). Bandshift experiments with these mutants indicate that the length of the loop region between helices I and II is not a critical factor in determining protein affinity. Several mutations were made at position 37 in both HMGldomA and H M G l d o m B to determine the effect on protein affinity. Two position-37 mutants of H M G l d o m A were prepared, H M G l d o m A F37A and HMGldomA F37W. The tryptophan mutant displays 5-fold lower affinity than the wild-type protein and the alanine mutant has >1000-fold loss in binding affinity. Two mutants of H M G l d o m B were also prepared at position 37, H M G l d o m B I37F and HMGIdomB I37A. The former displays approximately 2-fold higher affinity than wild-type protein, but H M G l d o m B I37A has about 2-fotd lower affinity than wild-type HMGldomB. These results suggest that position 37 can also act as an intercalator for H M G l d o m B and that the native isoleucine residue may have some intercalative ability. All four of these mutants clearly demonstrate the importance of intercalation at position 37 in determining protein binding affinity. Another set of mutant proteins was investigated to evaluate a second possible intercalating position in helix I at position 16 in domain A. Several sequencespecific and sequence-neutral HMG-domain proteins utilize intercalators at or near this site (80). Both single and double mutants were expressed to determine whether position 16 might be involved in intercalation and how it might affect protein binding, especially with respect to the influence of the residue at position 37. The mutant H MGldomA A16F has two potential intercalators but decreased affinity relative to wild-type protein, indicating that in domain A these intercalators do not work in a cooperative manner. The double mutant HMGldomA A16F F37A has lower affinity than HMGldomA, but shows restored binding relative to H M G l d o m A F37A. This result indicates that, in the absence of an intercalator at position 37, position 16 can augment binding to cisplatin adducts. Two mutants o f H M G l d o m B , H M G l d o m B F16A and H M G l d o m B F16A I37A, show approximately 3-fold and > 1000-fold reduced affinity, demonstrating that both of these residues contribute to H M G l d o m B binding.
112
SETH M. COHEN AND STEPHEN J. LIPPARD
The wild-type proteins and some of the mutants described were also analyzed by hydroxyl-radical footprinting to determine how the intercalating residues affect the positioning of the protein on the platinated DNA duplex. Two types of positioning were observed: asymmetric, where the protein preferentially protects bases to the 3' side of the adduct, and symmetric, where the protein protects bases equally well on either side of the adduct. In the crystal and in solution, H M G l d o m A binds in an asymmetric fashion (77). The results of footprinting studies on the mutant proteins are summarized in Table IV (81). Proteins with an intercalating residue at position 37 bind asymmetrically, whereas proteins with an intercalator at position 16 bind in a symmetric fashion. For H M G l d o m A position 37 is dominant, so that the H M G l d o m A A16F mutant binds in an asymmetric fashion. These studies demonstrate that intercalating residues influence both the affinity and orientation of HMG-domain protein binding to cisplatin-DNA 1,2-intrastrand crosslinks.
IV. Other Consequencesof Cisplatin Treatment of Cells The role of HMG-domain proteins has been the focus of many investigations to determine the cytotoxic mechanism of cisplatin (8). Yet, it is quite possible that other cellular events may contribute to the efficacy of the drug. This section describes studies that focus on pathways of cisplatin toxicity not directly associated with H MG-domain proteins.
A. TelomereShortening Telomeres are repeating DNA sequences found at the ends of chromosomes
(82). They are responsible for protecting the chromosome from nuclease degradation, end-to-end fusions, and other deleterious events (83-85). Telomeres are shortened every time cell division occurs, eventually resulting in senescence and termination of the cell line. In immortalized cells, including cancer cells, telomeres do not shorten in a normal fashion following cell division (86). Activation of the telomerase gene, which encodes for an enzyme involved in maintaining telomere length, is critical for sustaining cancerous cells (87-90). The human telomere sequence consists of several thousand nucleotides of the repeating sequence (TTAGGG)n. This G-rich sequence is potentially a good target for cisplatin binding. The possibility of preferential platination of telomere sequences was evaluated by studies of a plasmid comprising approximately 25% human telomere repeat. These experiments revealed only a 2.6-fold increase in platination of the telomere sequence relative to the remainder of the DNA, a result consistent with a statistical distribution of platinum (91). This finding suggests that if telomeric DNA is selectively targeted by cisplatin in cells, it probably is not due to the G-rich sequences in telomere repeats.
BIOCHEMISTRYOF CISPLATIN
113
H i g h D o s e Cisplatin
L o w D o s e Cisplatin AA
w
v
w
Nucleotide Excision Repair System
-41.m
m
A
TRF A•
f
3'
I
R~aI | Hint
----
~
1
3'
I l~aI
Nucleotide Excision 1 Hinfl Repair System
-4k-
~
3'
-UCell cycle arrest Replication and cell division are blocked
~
~
3
'
Replication and cell division are completed except telomeres shortenedTRF
-A--
intact TRF
I ~
3
1
~ -~m
'
] 3r "l"
-- Z
i intactTRF ,
~l l
3'
--w-• ..~
cisplatin DNA adduct excised short strand damaged with cisplafin
m ------
~ CCC'f~
sex'and strand
FIG. 12. Proposedscheme for dose-dependentcisplatin-mediatedtelomeredamage in HeLa cell lines. At high doses of cisplatin (left) most cells die before DNA replicationis completeand no effecton telomerelength (terminalrestrictionfragment,TRF) is observed.At low dosesofcisplatin (right) cells replicateDNA, but cisplatinadducts cannotbe removedfromtelomeresby nucleotide excisionrepair (NER). Replicationis blockedbythe DNA adductslocatedat the telomeres,resulting in incompletereplicationand a shorteningoftelomeres.ReprintedwithpermissionfromT. Ishibashi and S. J. Lippard,Proc. Natl. Acad. Sci. USA 95, 4219-4233 (1998). Treatment of HeLa cells with cisplatin results in degradation and shortening of telomeres in a dose-dependent fashion (92). At high doses, cells die without exhibiting shortened telomeres. However, at low doses, cells can undergo division, with the resulting daughter cells having dramatically shortened telomeres, which ultimately results in cell death. Cells surviving at least 10 days of exposure to low levels of cisplatin do not have shortened telomeres. Figure 12 shows a proposed scheme to account for how cisplatin might result in telomere shortening at the low doses. Although targeting of telomeres by cisplatin is unlikely based on sequence context alone (vide supra), the possibility remains that telomeres could be preferentially damaged owing to their unique location and the structure at the ends of chromosomes (93). Selective damaging of specific regions of the genome by cisplatin is possible and has been demonstrated at promoters during activated transcription (94). Because telomeres are not transcribed, they will not be repaired by the NER system, perhaps making even low levels of telomere damage sufficient to cause senescence and cell death.
114
SETH M. COHENAND STEPHENJ. LIPPARD
In addition to damaging and shortening the telomeres by directly coordinating to DNA, several studies indicate that cisplatin can also interfere with telomerase activity. Telomerase is a ribonucleoprotein complex that adds "I-TAGGG repeats to the telomere ends. Although there are non-telomerase-dependent mechanisms for maintaining telomere length, many cancer cells lines show high levels of telomerase activity when compared with normal somatic cells, which display essentially no such activity (86). A study using esophageal cancer cells revealed that cell lines with high levels of telomerase activity were more sensitive to cisplatin, whereas sensitivity to the anticancer drug 5-fluorouracil showed no correlation with telomerase activity (95). A study using testicular cancer cells indicated that cisplatin reduced telomerase activity in a dose-dependent fashion (96). Other DNA-damaging agents, including doxorubicin, bleomycin, methotrexate, melphalan, and trans-DDP, had no such effect on telomerase activity. Evidence for the direct interaction of cisplatin with the telomerase enzyme was found in vitro (T. Ishibashi and S. J. Lippard, unpublished data). Telomerase-active HeLa cell extracts are inactivated by cisplatin in a dose- and time-dependent manner. In addition, reverse transcriptase polymerase chain reaction (RT-PCR) shows that cisplatin reacts with the RNA component of the telomerase enzyme. These studies indicate that cisplatin can affect both telomeres and the telomerase enzyme; however, the importance of these interactions to the overall cytotoxicity of the drug remains to be evaluated.
B. TATA-Binding,H1, and AAG Proteins Many other important nuclear proteins, which do not belong to the HMGdomain family, also bind to cisplatin 1,2-intrastrand crosslinks. One example is the TATA-binding protein (TBP), which is intimately involved in transcription initiation (50). TBP occurs in eukaryotes that recognize the consensus sequence T. A. T. a/t. A. a/t (56) located in the promoter region upstream of transcribed genes. TBP binds to this "TATA box" as a part of a multiprotein complex transcription factor IID (TFIID), thereby initiating a cascade of protein-binding events leading to transcription by RNA polymerase II and the production of mRNA. TBP has two domains, a nonconserved N-terminal domain (97) and a highly conserved C-terminal domain that constitutes the DNA-binding portion. The conserved C terminus recognizes the TATA box from the minor groove, using two pairs of intercalating residues to bind and kink the DNA (98, 99). Several studies show that, both in reconstituted transcription systems and in cells, TBP can be titrated away from its natural binding site to cisplatin-damaged DNA (48, 49). Transcription in the reconstituted system can be restored by the addition of excess TBP. These experiments suggest that platinum-damaged DNA may divert this critical protein from its natural function. A recent study focused on the details of the TBP interaction with cisplatinmodified DNA. The importance of intercalation in the binding of TBP to
BIOCHEMISTRYOF CISPLATIN
115
platinated D NA was evaluated by investigating a series of oligonucleotide probes containing a TATAbox flanked by G*G* adducts. The results indicate that, when, 1,2-intrastrand crosslinks were placed at or near the sites of TBP intercalation, the affinity of the protein for the oligonucleotide increased as much as 175-fold (100). The binding kinetics demonstrate that the increased affinity is a consequence of a reduction in the dissociation rate constant. These experiments clearly demonstrate that the intercalating residues in TBP, like those in HMG-domain proteins, can recognize the hydrophobic notch created by cisplatin adducts, thereby increasing the stability of the protein-DNA complex. Additional studies have determined the affinity of TBP for isolated cisplatin adducts, outside the context of the TATA box (S. M. Cohen and S. J. Lippard, unpublished data). Preliminary determination of the reverse rate constant by EMSA suggests that TBP has an affinity for cisplatin adducts comparable to that for the TATA element. Most of the approximately 10,000 transcribed genes in mammalian cells use a TATA box for transcription initiation (50). Tumors from cancer patients treated with cisplatin contain between 10,000 and 100,000 platinum adducts per cell (101, 102). Such numbers suggest that TBP could be effectively removed from its natural binding sites by cisplatin adducts in vivo, resulting in additional pathways by which the drug might kill cells. Such a pathway would represent a classic manifestation of the transcription factor hijacking hypothesis. In addition to transcription factors like TBP, structural proteins in the nucleus have also been shown to bind cisplatin-modified DNA. The histone linker protein H1 binds to DNA globally platinated with cis-DDP, but not trans-DDP (51). This protein is an abundant and important nuclear factor that coats DNAlinking nucleosome core particles. H1 binds to cisplatin-modified DNA even in the presence of HMG1, suggesting comparable affinities. It is noteworthy that several chromatin-binding proteins--including H1 which helps determine structure, SSRP1 which facilitates transcription, and HMG1 which remodels chromatin--bind specifically to cisplatin-damaged DNA. The involvement of so many interrelated nuclear factors suggests that the mechanism of cisplatin genotoxicity could be multifactorial and complex. Proteins of the NER complex recognize and remove cisplatin adducts from DNA (6). The 3-methyladenine DNA glycosylase (AAG) family of mammalian repair proteins also recognizes cisplatin-DNA crosslinks (103). Binding of AAG may be facilitated by tyrosine intercalation, an interaction the enzyme displays with other substrates. Although able to bind to cisplatin-DNA adducts, AAG could not excise the lesions and was incapable of repairing 1, N6-ethenoadenine (EA) adducts in the presence of cisplatin-damaged DNA. By analogy to transcription factor hijacking, cisplatin may participate in repair factor hijacking, diverting AAG from its natural substrates leaving other lesions, such as eA damage, to mediate cell death. This hypothesis was proposed to explain the synergistic effects of cisplatin with 1,3-bis-(2-chloroethyl)-l-nitrosourea (BCNU).
116
SETH M. C O H E N AND STEPHEN J. LIPPARD
C. Apoptosisand Ubiquitination Cisplatin damage to DNA triggers a host of cellular events, ranging from protein recognition of cisplatin-DNA crosslinks to telomere shortening. A question remains, however: How do these events contribute to a cascade that leads to cell death? Two morphological patterns, necrosis and apoptosis, characterize cell death (104, 105). Necrosis results when a cell is traumatically damaged, for example, by puncturing the outer membrane. Apoptosis, coined "programmed cell death" or "cell suicide," is a controlled pathway that requires new protein synthesis. Cells exposed to cisplatin exhibit double-stranded DNA cleavage, blebbing of the cell surface, and cell shrinkage, all of which are consistent with apoptosis as the means of cell death (106). DNA cleavage produces "-~180-bp fragments, suggesting internucleosomal scission by an endonuclease. Flow cytometry experiments reveal that cells exposed to cisplatin largely arrest in the second growth phase (G2) of the cell cycle, indicating that blocked DNA replication, which occurs in the synthesis phase (S) of the cell cycle prior to G2, is the not the cause of cell death (107-109). The flow cytometry data are consistent with other experiments demonstrating that cell growth can be inhibited by cisplatin at doses much lower than those required to inhibit replication (110). Transcription can also be blocked by cisplatin-DNA adducts. The arresting of cells in the G2 phase suggests that proteins necessary for mitosis are not being synthesized, implicating transcription inhibition as the means by which apoptosis is triggered. Transcription blockage could well be a consequence of RNA polymerase II stalling at cisplatin-DNA lesions or cisplatin-DNA-protein ternary complexes. RNA pol II transcription is impeded by UV radiation- or cisplatin-induced damage leading to phosphorylation on its carboxyl-terminal domain (111), which results in ubiquitination and degradation in proteasomes (112). Ubiquitinated proteins are covalently modified by the stepwise activity of several enzymes (113). This modification signals a targeted response by the cell and frequently results in degradation of the modified protein inside proteasomes. Other forms of DNA damage do not elicit this cascade. UV radiation-induced reduction in RNA pol II activity is alleviated in repair-competent cells, but not in repair-deficient xeroderma pigmentosum cell lines (114). These observations suggest one possible mechanism for cytotoxicitywhereby cisplatin treatment results in transcription inhibition and degradation of RNA pol II owing to ubiquitination. RNA pol II levels cannot be restored because shielding by HMG-domain proteins, or other proteins that have a high affinity for cisplatin-DNA adducts, protects cisplatin lesions from nucleotide excision repair. As a consequence, there is a continual decline in RNA pol II and transcriptional activity, which ultimately may lead to apoptosis. Although more experiments are required to explore this hypothesis, it is an intriguing scenario
BIOCHEMISTRYOF CISPLATIN
117
that incorporates both the phenomena of transcription inhibition by and repair shielding of cisplatin-DNA cross-links. Other possible pathways need to be explored to determine how cisplatin leads to apoptosis. In addition to ubiquitination and shortening of telomeres, the relationship between cisplatin and p53 has not been resolved. The latter subject has been thoroughly discussed in a recent review (6).
V. Developmentof New Platinum Compounds Although cisplatin and carboplatin (vide infra) have been very successful for the treatment of cancer, the compounds are not ideal drugs. Neither is orally active and both must be administered intravenously. Cisplatin has several detrimental side effects, including nephrotoxicity, neurotoxicity, and severe nausea (3). Patients also experience both inherent and acquired resistance to the drug (115, 116). As a consequence of these shortcomings, several research efforts have focused on producing compounds with equal or greater potency, lower toxicity, efficacy against cisplatin-resistant tumors, and oral availability (3). In the following section we discuss novel discovery methods and a better understanding of cisplatin activity which are leading the way toward developing new platinum drugs and chemotherapeutic regimens.
A. Mononuclear PlatinumComplexes To date, the FDA has approved only one cisplatin analog. Carboplatin (Figs. 1 and 13) has a 1,1-dicarboxylatocyclobutane unit as the labile ligand replacing the two chloride ions in cisplatin. This chelating dicarboxylate dianion dissociates more slowly than the chloride ligands, resulting in a less reactive compound with substantially diminished side effects. Because of the reduced toxicity, carboplatin has been used very successfully and is gradually replacing cisplatin in the clinic. Although not approved in the United States, oxaliplatin is used in parts of South America, Asia, and Europe (117). In this compound, the ammine ligands of cisplatin are replaced by cis-l,2-(R,R)-diaminocyclohexane and the chloride groups by a chelating oxalate leaving group. The compound has been pursued owing to its efficacy against colorectal cancer and some cisplatin-resistant tumors
(118-130). Some Pt(IV) compounds are being investigated because of their oral activity. These compounds, designated JM216 and JM221, are octahedral platinum(IV) complexes that are reduced in vivo to square planar platinum(II) compounds with kinetically inert ammine and cyclohexylamine ligands and a pair of labile chloride leaving groups (3). The resulting platinum(II) metabolites then react in a fashion analogous to cisplatin. These compounds are currently undergoing
118
SETH M. COHEN AND STEPHEN J. LIPPARD
H2
N~ /sO
\/
,
0
~ N2
CarboplaUn
Oxallplatln o
.,N\I/°.
o
.3N\I/°,
/N
o
JM216
-
0
JM221
FIG. 13. Chemical structures of several mononuclear platinum compounds that have been investigated as anticancer drugs. The octahedral compounds (bottom) are Pt(IV) and are intended as orally active alternatives to cisplatin. Of the compounds shown, only carboplatin (upper left) is approved for use in the United States.
clinical trials and may alleviate the need for intravenous administration of platinum drugs. The effect of the cyclohexylamine on the efficacy has not been fully resolved, however. Because JM216, JM221, and oxaliplatin have demonstrated some promise against cisplatin-resistant cell lines, the biological effects of different amine ligands have been investigated. Several studies have demonstrated that such spectator or carrier ligands can affect the ability of a platinum adduct to block replication by various DNA polymerases (121-125). Apparently, HMG-domain proteins, including HMG1 and human upstream binding factor (hUBF), bind to oxaliplatin and JM216/221 lesions with lower affinity than to cisplatin-DNA adducts (123, 126). This difference suggests that HMG-domain proteins may not play a critical role in the genotoxicity of oxaliplatin and JM216/221, thereby explaining the lack of cross-resistance with cisplatin-resistant cell lines. More studies are required to elucidate the effects of various spectator ligands and the pathways that might mediate the toxicity of these complexes. Apart from the compounds described above, few cisplatin analogs have progressed through stage-three clinical trials (3). As an alternative to the conventional design of platinum drugs, which contain a pair ofcis leaving groups, several mononuclear compounds with trans positioning of the ligands have been pursued
119
BIOCHEMISTRY OF CISPLATIN
as possible drug candidates. Many such compounds use aromatic amines as the inert spectator ligands (3). These compounds have met with mixed success, showing some utility against cisplatin-resistant cell lines. To date, no mononuclear trans platinum compound has made significant progress in the clinic.
B. Polynuclear Platinum Complexes As an alternative paradigm to mononuclear platinum compounds, several polynuclear platinum complexes have been prepared for use as antitumor drugs (Fig. 14). Compounds containing two, three, and four platinum centers (•27) in both trans and cis configurations have been investigated (128-131). Of these compounds, monofunetional complexes in which the leaving groups are trans to the linking ligand have received the greatest attention. Recently, one in particular, BBR3464, a trinuclear bifunetional platinum compound (Fig. 14), has entered phase I clinical trials (132). These compounds form a variety of DNA adduets, with 1,3- and longer-range interstrand erosslinks being the major ones (128). Not surprisingly, these lesions are not tightly bound by HMG-domain proteins (vide supra), indicating that their toxicity, like that of oxaliplatin and
-•2+
H3N~ /Cl Cl~ /NH3 /Pt Pt N /N H3N NH2(CH2)6H2N NH3
~ 2+ Cl~ /NH3 H3N"\ //el /Pt Pt H3N ~NH2(CH2)6H2 N/ ~NH3
1,1/C,C
1,1/t,t
CI,~ /NH3 H3NN /CI .Pt Pt CI / \NH2(CH2)sH2 N/ ~CI
Cl~ ,/NH3 H3N~ ,/C----II1+ /Pt Pt H3N ~NH2(CH2)6H2 N/ ~Ct
2,2/c,e
1,2/t,c
Cl~ /NH3
H3NN /NH2(CH2)6H2N~/NH3--] 4+ /et et /Pt H3N ~NH2(CH2)6H2 N/ ~NH3 H3N ~CI 1,0,1/t,t,t FIG. 14. Chemical structures of several multinuelear platinum compounds that have been investigated as anticancer drugs. The naming scheme indicates the number of leaving groups on each metal center ( I, 2, etc.) followed by the stereochemistry of each metal center (eis = e, trans = t). The trinuclear compound (bottom) is BBR3464, which is in phase I clinical trials.
120
SETH M. COHENAND STEPHENJ. LIPPARD
JM216/221, is most likely mediated by a different pathway from that of cisplatin (133). The potential utility of these compounds in practical medicine is currently uncertain.
C. Combinatorial/Parallel SynthesisApproaches Combinatorial and parallel synthetic approaches have assumed a prominent position in drug discovery. Such efforts initially focused on peptide-based drugs, and subsequently combinatorial synthesis was adapted for organic small molecule discovery (134, 135). For inorganic synthesis, combinatorial methods have been applied to the discovery of solid-state materials (136, 137) and catalyst design (138). Recently, we have applied parallel synthesis to platinum drug discovery (139). Conceptually, this process should also be useful for the discovery of inorganic drugs (140) of various types, such as imaging agents (141) and antimalarials (142), although the focus of our work was on platinum anticancer compounds. Our general synthetic scheme and robotic apparatus for parallel synthesis of cisplatin analogs are depicted in Fig. 15. The synthesis allows for the variation of both the spectator and leaving group ligands. In a 96-well reaction block, either K2PtC14 or K[Pt(NH3)C13] is activated with KI, forming [PtI4]2- or [Pt(NH3)I3]-, respectively. Subsequently, the spectator ligand(s) are added followed by a slight excess of AgNO3, which precipitates AgC1 and AgI. The soluble products are then filtered into vials, located below the reaction block, containing various leaving-group ligands. The samples are lyophilized and analyzed by atomic absorption (AA) spectroscopy for platinum content. In several months, roughly 3600 reactions were run and the products screened for their ability to inhibit transcription (vide infra). To put this value in perspective, approximately 3000 platinum compounds had been synthesized and evaluated for activity in the entire chemical literature prior to this study (3). Various spectator ligands were examined, including aliphatic amines, aromatic amines, phosphines, and sulfurcontaining compounds. Also, several types of leaving groups were screened, including chlorides, monodentate acids, and bidentate acids. Of the 3600 reaction mixtures, 14 hits were obtained and further evaluated in a concentration-dependent manner. Of these 14, four displayed good activity in the concentration-dependent study. Of the four compounds, cis-[ (isopropylamine)2PtC12], cis-[(cyclobutylamine)2PtC12], and cis-[ammine (cyclobutylamine)PtC12] had been previously identified as potential drug candidates (143-145). The fourth compound, cis-[ammine(2-amino-3-picoline)PtC12], represents a novel lead in platinum drug discovery, bearing some resemblance to recently investigated platinum compounds (3, 146, 147). The parallel synthesis of platinum compounds is an alternative approach to the discovery of new anticancer drugs. The methodology can be easily adapted to produce other kinds of compounds, and some having trans geometries have
cl\ /cl
Pt CI/ XCI L~ ~1 Pt L/
4KI
•
I\ /~
2L
I/%1 8AgNO3
L~ ~ OH2 Pt
\,
xsZ
•
L\ /I Pt L/ \ l
L~ /Z Pt L/
\Z
ReactionBlock
Co~l~~tionVial
FIG. 15. Synthetic scheme (top) and apparatus (bottom) for parallel synthesis of platinum drug leads. Tetrachloroplatinate(II) is activated with KI, followed by addition of 2 equivalents of spectator ligand (L). Silver chloride is then added, precipitating AgCI and AgI and producing the aquated platinum species. The soluble aqua complex is t~ltered from the reaction block, through a Teflon flit, and into a collection vial containing leaving-group ligands (Z). The samples are lyophilized and analyzed for platinum content by atomic absorption (AA) spectroscopy. In approximately one year, 3600 reactions were performed by using this methodology. The compounds were tested against HeLa cell lines using a fluorescent dye reporter assay.
122
SETH M. COHENAND STEPHENJ. LIPPARD
already been prepared as a proof of concept (139). A parallel synthesis effort requires an accurate and rapid screening process in order to be successful. The next section discusses the high-throughput screens used to evaluate these new platinum drug candidates.
D. Screenin9 Methods To evaluate the large number of compounds produced by parallel synthesis, a rapid, mechanism-based assay was devised. Cisplatin, unlike many other DNAdamaging agents, causes a dose-dependent diminution in transcription activity (148). To identify new drug candidates that kill cells by a related mechanism, an assay was developed to screen for inhibition of transcription in Jurkat and HeLa cell lines. The assay is based on CCF2/AM molecules containing fluorescein, coumarin, and/~-lactam components (Fig. 16) (149). The CCF2/AM dye diffuses passively through cellular membranes, whereupon intracellular esterases remove several protecting groups, resulting in a charged species that is trapped inside the cell. Excitation of the coumarin moiety at 409 nm results in FRET to the fluorescein, which emits light at 520 nm and gives the cells a green color. If the cells are expressing a fl-lactamase, the linker is cleaved, separating the two dyes and preventing FRET. With the molecule cleaved, the coumarin dye emits at 447 nm, giving the cells a blue color. By monitoring the ratio of green and blue cells, a readout for measuring the inhibition of gene expression is obtained. Screening for transcription inhibition in this manner requires approximately 28 h and allows for the analysis of 72-96 compounds per day without robotics. BlaM HeLa cells, stably transfected with a vector encoding for TEM-1 fl-lactamase from E. coli, were placed in 96-well plates. The cells were treated with different platinum compounds and then exposed to the CCF2/AM dye. A ratiometric response was obtained by using a fluorescent plate reader monitoring at 530 nm and 460 nm. Compounds that demonstrated a 530:460 ratio greater than that of cisplatin controls were further evaluated in a concentrationand time-dependent fashion (139).
E. Steroid Hormones, Cisplatin, and Cancer In addition to discovering new platinum compounds with genotoxic activity, the knowledge gained about the mechanism of action of cisplatin has also been used to develop new combination therapies for increasing its efficacy. A recent report demonstrated that HMG1 mRNA levels were upregulated by estrogen treatment of MCF-7 breast cancer cells (150). This finding allowed us to examine directly the postulated role of HMG1 in mediating cisplatin cytotoxicity
(151). Treatment of MCF-7 breast cancer cells with estrogen or progesterone upregulates HMG1 levels, as demonstrated by immunofluorescence. Treatment of MCF-7 cells with either estrogen or progesterone followed by exposure to
BIOCHEMISTRY OF CISPLATIN
19,3
B,oy.~o.~o,
,oo.~y.O.~y.oAo
c,~.~.~
~ O
/
/
/
T v CO2AM
a "O.,.f-~,/O...~O
/
,~V.y.~.~,.,
° °~"NH s.
409 nm
\ o,~% \
\
CCF2/AM
7°°'_\
GreenLight \
ilLA
~-I.a~amase
+
o o.-~..
.o.c.~S~
~-#.
~.~.co2
/
cd /
°02FIG. 16. Scheme of the fluorescent dye system used to screen fnr new platinum anticancer drug candidates. Ceils expressing l%lactamases cleave CCF2/AM and the cells appear blue from the coumarin emission at 447 nm. The fluorescence of the cleaved fluorescein unit is quenched due to the free thiol group. If genotoxic agents are blocking transcription of ~-lactamases, however, the dye cannot be cleaved and undergoes FRET with an emission from the fluorescein dye at 5"20nm, giving the cells a green color. Transcription'a] activity can be calculated ratiometrically from the green versus blue emission. Reprinted with permission from E. R. Jamieson and S. J. Lippard, Chem. Rev. 99, 2467-2498 (1999). Copyright © 1999 American Chemical Soeie~.
124
SETH M. COHEN AND STEPHEN J. LIPPARD
cisplatin causes a 2-fold i n c r e a s e in sensitivity to the d r u g (Fig. 17). C o t r e a t m e n t with half an e q u i v a l e n t each o f e s t r o g e n a n d p r o g e s t e r o n e gives a synergistic effect, r e s u l t i n g in a 4-fold sensitization to cisplatin. A 2-fold sensitization t o w a r d c a r b o p l a t i n also can b e achieved, b u t r e q u i r e s that the cells b e p r e t r e a t e d the p l a t i n u m c o m p o u n d owing to the slower kinetics o f D N A binding.
100
"~ 10-. ---
none
A o
progesterone both
]J ±
"1~ 1
1 [cisplatin] (laM) 100
10-.
none
1
~
~ 1~ 116 [carboplatin] (pM)
210
FIG. 17. Cytotoxieity assay plots demonstrating sensitization of MCF-7 breast cancer cells to cisplatin resulting from steroid hormone treatment (151). The top plot shows that the cells were ~2-fold more sensitive to cisplatin when exposed to estrogen or progesterone and ~4-fold more sensitive when subjected to both steroid hormones. The bottom plot shows that a similar sensitization was also obtained with carboplatin, although a different regimen was required because of its slower reaction kinetics. No sensitization was seen for trans-DDP or with steroid hormone receptornegative HeLa cells, consistent with HMG1 upregulation being the source of sensitization (data not shown). Reprinted with permission from Q. He, C. H. Liang, and S. J. Lippard, Proc. Natl. Acad. Sci. USA. 97, 5768-5772 (2000).
BIOCHEMISTRY OF CISPLATIN
125
No sensitization toward trans-DDP or calieheamicin, another cytotoxic agent, is obtained upon steroid hormone treatment. This work illustrates how a better understanding of the mechanism of cisplatin can lead to potentiation of its activity. By knowing specific biochemical interactions at the molecular and cellular levels, a strategy was devised that could predictably enhance the activity of the drug. Cisplatin, estrogen, and progesterone are all FDA-approved drugs. A phase I clinical trial combining progesterone-carboplatin chemotherapy for women suffering from ovarian cancer has recently commenced (unpublished).
VI. Concluding Remarks The success of cisplatin for treating cancer is unrivaled in the history of inorganic medicinal chemistry. After a serendipitous discovery, years of chemical research have begun to reveal how the drug elicits its genotoxic and cytotoxic effects. The information gathered about the mechanism of action is now being used to develop new treatment strategies and discover new platinum drugs. Despite great strides in this field, many questions remain. Although a strong body of evidence supports the involvement of HMG-domain proteins in mediating cisplatin efficacy, exactly how these proteins trigger cytoxicity remains to be determined conclusively. HMG-domain sensitization may involve both repairshielding and protein-hijacking mechanisms. The roles of other nuclear proteins, such as TBP and AAG, and the precise nature of the interaction of these proteins with cisplatin-damaged DNA have yet to be elucidated. The effects of telomere damage and the importance oftelomere maintenance represent another intriguing route by which cisplatin may ultimately cause cell death. Tile question of why cisplatin is particularly effective against testicular cancer remains unanswered. The pursuit of these remaining questions is critical to understanding one of the most successful anticancer drugs to date. More detailed insights into the mechanism of cisplatin will continue to provide new clues for fighting cancer and perhaps to fulfill the promise of developing other inorganic drugs that rival cisplatin efficacy with reduced side effects. The story of this fascinating drug clearly has many more interesting lessons to teach to bioinorganic and medicinal chemistry communities.
ACKNOWLEDGMENTS
The authors thank Q. He, Dr. T. Ishibashi,Dr. E. R. Jamieson,Dr. U.-M.Ohdorf,A.E Silverman, Dr. B. C. Spingler, and Dr. C. J. Ziegler for permission to cite unpublished data. The research undertaken in our laboratoryand describedhere was supported by grants from the National Cancer Institute (grant CA 3499"2).S. M. C. is the recipient of a National Institutes of Health postdoctoral fellowship.
126
SETH M. COHEN AND STEPHEN j. LIPPARD REFERENCES
1. 2. 3. 4. 5.
S. J. Lippard, Science 218, 1075-1082 (1982). C. Orvig and M. J. Abrams, Chem. Rev. 99, 2201-2203 (1999). E. Wong and C. M. Giandomenico, Chem. Rev. 99, 2451-2466 (1999). J. Reedijk, Chem. Rev. 99, 2499-2510 (1999). A. Gelasco and S. J. Lippard, in "Topics in Biological Inorganic Chemistry" (M. Clarke and P. J. Sadler, eds.), pp. 1~t3. Springer-Verlag, Heidelberg, 1999. 6. D. B. Zamble and S. J. Lippard, in "Cisplatin--Chemistry and Biochemistry of a Leading Anticancer Drug" (B. Lippert, ed.), pp. 73-110. Verlag Helvetica Chimica Acta, Zurich, 1999. 7. J. Zlatanova, j. Yaneva, and S. H. Leuba, FASEBJ. 12, 791-799 (1998). 8. E. R. Jamieson and S. J. Lippard, Chem. Rev. 99, 2467-2498 (1999). 9. B. Rosenberg, L. Van Camp, and T. Krigas, Nature (London) 205, 698-699 (1965). 10. B. Rosenberg, L. Van Camp, E. B. Grimley, andA. J. Thomson,J. Biol. Chem. 242, 1347-1352 (1967). 11. B. Rosenberg, L. Van Camp, J. E. Trosko, and H. V. Mansour, Nature (London) 222, 385-386 (1969). 12. B. Rosenberg and L. Van Camp, Cancer Res. 30, 1799-1802 (1970). 13. G. J. Bosl and R. J. Motzer, N. Engl. J. Med. 337, 242-253 (1997). 14. D. J. Szalda, F. Eckstein, H. Sternbach, and S. J. Lippard, J. Inorg. Biochem. 11, 279-282 (1979). 15. j. K. Barton and S. J. Lippard, in "'Metal Ions in Biology" (T. G. Spiro, ed.), pp. 31-113. John Wiley & Sons. New York, 1980. 16. K. G. Strothkamp and S. J. Lippard, Proc. Natl. Acad. Sci. U.S.A. 73, 2536--2540 (1976). 17. S. J. Lippard, P. j. Bond, K. C. Wu, and W. R. Bauer, Science 194, 726-728 (1976). 18. K. W. Jennette, j. T. Gill, J. A. Sadownick, and S. J. Lippard, J. Am. Chem. Soc. 98, 6159-6168 (1976). 19. M. Howe-Grant and S. J. Lippard, Inorg. Synth. 20, 101-105 (1980). 20. K. E. Erkkila, D. T. Odom, and J. K. Barton, Chem. Rev. 99, 2777-2795 (1999). 21. G. L. Cohen, W, R. Bauer, J. K. Barton, and S. J. Lippard, Science 203, 1014-1016 (1979). 22. H. M. Ushay, T. D. Tullius, and S. J. Lippard, Biochemistry 20, 3744-3748 (1981). 23. G. L. Cohen, J. A. Ledner, W. R. Bauer, H. M. Ushay, C. Caravana, and S. J. Lippard, J. Am. Chem. Soc. 102, 2487-2488 (1980). 24. J. P. Caradonna, S. J. Lippard, M. J. Gait, and M. Singb, J. Am. Chem. Soc. 104, 5793~5795
(1982). 25. P. j. M. Van De Vaart, J. Belderbos, D. De Jong, K. C. A. Sneeuw, D. Majoor, H. Bartelink, and A. C. Begg, Int. J. Cancer 89, 160-166 (2000). 26. R. B. Cieearelli, M. J. Solomon, A. Varshavsky, and S. J. Lippard, Biochemistry 24, 7533-7540
(1985). 27. A. L. Pinto and S. J. Lippard, Proc. Natl. Acad. Sci. U.S.A. 82, 4616~1619 (1985). 28. Y. Corda, C. Job, M. F. Anin, M. Leng, and D. Job, Biochemistry 30, 222-230 (1991). 29. Y. Corda, C. Job, M. 17.Anin, M. Leng, and D. Job, Biochemistry 32, 8582-8588 (1993). 30. C. Cullinane, S. J. Mazur, J. M. Essigmann, D. R. Phillips, and V. A. Bobr, Biochemistry 38,
6204-6212 (1999). 31. J. A. Mello, S. J. Lippard, and J. M. Essigmann, Biochemistry 34, 14783-14791 (1995). 32. J. tt. Toney, B. A. Donahue, P. J. Kellett, S. L. Brubn, J. M. Essigmann, and S. j. Lippard, Proc. Natl. Acad. Sci. U.S.A. 86, 8328-8332 (1989). 33. S. L. Bruhn, P. M. Pil, J. M. Essigmann, D. E. Housman, and S. j. Lippard, Proc. Natl. Acad. Sci. U.S.A. 89, 2307-2311 (1992). 34. R. Grosschedl, K. Giese, and j. Pagel, Trends. Genet. 10, 94-100 (1994).
BIOCHEMISTRY OF CISPLATIN
127
35. S. L. Bruhn, D. E. Housman, and S. J. Lippard, Nucleic Acids Res. 21, 1643-1646 (1993). 36. P. M. Pil and S. J. Lippard, Science 256, 234-237 (1992). 37. R. S. Farid, M. E. Bianchi, L. Falciola, B. N. Engelsberg, and P. C. Billings, Tox. Appl. Pharma.
141,532-539 (1996). S.J. Brown, P. J. Kellett, and S. J. Lippard, Science 261, 603~05 (1993). M. M. McA'Nulty, J. P. Whitehead, and S. J. Lippard, Biochemistr~d 35, 6089~i099 (1996). M. M. McA'Nulty and S. J. Lippard, Mutat. Res. 362, 75-86 (1996). S.J. Lippard, Structural and Biological Consequences of Platinum Anticancer Drug Binding to DNA, The Robert A. Welch Foundation Conference on Chemical Research, Houston, Texas (1993). 42. B. A. Donahue, M. Augot, S. F. Bellon, D. K. Treiber, J. H. Tone},, S. J. Lippard, and J. M. Essigmann, Biochemistnj 29, 5872~5880 (1990). 43. J.-C. Huang, D. B. Zamble, J. T. Reardon, S. J. Lippard, and A. Sancar, Proc. Natl. Acad. Sci. U.S.A. 91, 10394-10398 (1994). 44. D. B. Zamble, D. Mu, J. T. Reardon, A. Sancar, and S. J. Lippard, BiochemistrTj 35, 1000410013 (1996). 45. G. Orphanides, W.-H. Wu, W. S. Lane, M. Hampsey, and D. Reinberg, Nature (Ixmdon) 400, 284-288 (1999). 46, M. Bustin, D. A. Lehn, and D. Landsman, Biochim. Biophys. Acta 1049, 231-243 (1990). 47. T. Ise, G. Nagatani, T. lmamura, K. Kato, H. Takano, M. Nomoto, H. Izumi, H. Ohmori, T. Okamoto, T. Ohga, T. Uchiumi, M. Kuwano, and K. Kohno, Cancer Res. 59, 342-346 (1999). 48. F. Coin, P. Frit, B. Viollet, B. Salles, and J.-M. Egly, Mol. Cell. Biol. 18, 3907-3914 (1998). 49, P. Vichi, F. Coin, J.-P. Renaud, W Vermeulen, J. H. J. Hoeijmakers, D. Moras, and J.-M. Egly, EMBOJ. 16, 7444-7456 (1997). 50. G. Orphanides, T. Lagrange, and D. Reinberg, Gene,s Dev. 10, 2657-2683 (1996). 51. J. Yaneva, s. H. Leuba, K. Van Holde, and J. Zlatanova, Proc. Natl. Acad. SCI. U.S.A. 94, 13448-13451 (1997). .52. S. U. Dunham and S. J. Lippard, BiochemisttTj 36, 11428-11436 (1997). 53. S. M. Cohen, Y. Mikata, Q. I-te, and S. J. Lippard, Biochemistry, 39, 11771-11776 (2000). 54. C. S. Chow, J. P. Whitehead, and S. J. Lippard, Biochemistry 33, 15124-15130 (1994). 55. P. M. Pil, C. S. Chow, and S. J. Lippard, Proc. Natl. Acad. Sci. U.S.A. 90, 9465-9469 (1993). 56. Z. S. Juo, T. K. Chiu, P. M. Leiberman, I. Baikalov, A. J. Berk, and R. E. Diekerson, J. Mol. Biol. 261,239-254 (1996). 57. D. S. Pilch, S. U. Dunham, E. R. Jamieson, s. j. Lippard, and K. J. Breslauer, J. Mol. Biol. 296, 803-812 (2000). 58. C. S. Chow, C. M. Barnes, and S. J. Lippard, Biochemistry 34, 2956--2964 (1995). 59. D. K. Treiber, X. Zhai, H.-M. Jantzen, and J. M. Essigmann, Proc. Natl. Acad. Sci. U.S.A. 91, 5672-5676 (1994). 60. E. E. Trimmer, D. B. Zamble, S. J. Lippard, and J. M. Essigmann, Biochemistry 37, 352-362 (1998). 61. P. C. Billings, R. J. Davis, B. N. Engelsberg, K. A. Skov, and E. N. Hughes, Biochem. Biophys. Res. Commun. 188, 1286-1294 (1992). 62. S. M. Patrick and J. j. Turchi, Biochemistry 37, 8808-8815 (1998). 63. E. R. Jamieson and S. J. Lippard, BiochemistrTj 39, 8426-8438 (2000). 64. E. R. Jamieson, M. P. Jacobson, C. M. Barnes, C. S. Chow, and S. J. Lippard, J. Biol. Chem. 274, 12346-12354 (1999). 65. S. E. Sherman, D. Gibson, A. H.-J. Wang, and S, J. Lippard, Science 230, 412-417 (1985). 66. P. M. Tak'a]mra,C. A. Frederick, and S. J. Lippard,J. Am. Chem. Soc. 118, 12309--12321(1996). 67. P. M. Takahara, A. C. Rosenzweig, C. A. Frederick, and S. J. Lippard, Nature (London) 377, 649-652 (1995). 38. 39. 40. 41.
128
SETH M. COHEN AND STEPHEN J. LIPPARD
68. A. Gelasco and, S. J. Lippard, Biochemistry 37, 9230-9239 (1998). 69. S. U. Dunham, S. U. Dunham, C. J. Turner, and S. J. Lippard, J. Am. Chem. Soc. 120, 53955406 (1998). 70. M. Iwamoto, S. Mukundan, Jr., and L. G. Marzilli, J. Am. Chem. Soc. 116, 6238-6244 (1994). 71. J.-M. Teuben, C. Bauer, A. H.-J. Wang, and j. Reedijk, Biochemistry 38, 12305-12312 (1999). 72. C. J. Van Garderen and L. P. A. Van Houte, Eur J. Biochem. 225, 1169-1179 (1994). 73. H. Huang, L. Zhu, B. R. Reid, G. P. Drobny, and P. B. Hopkins, Science 270, 1842-1845 (1995). 74. E Paquet, C. P6rez, M. Leng, G. Lancelot, and J.-M. Malinge, ]. Biomol. Struct. Dyn. 14, 67-77 (1996). 75. F. Coste, J.-M. Malinge, L. Serre, W. Shepard, M. Roth, M. Leng, and C. Zelwer, Nucleic Acids Res. 27, 1837-1846 (1999). 76. H. Bleiberg, Brit. ]. Cancer 77, 1~3 (1998). 77. U.-M. Ohndorf, M. A. Rould, Q. He, C. O. Pabo, and S. J. Lippard, Nature (London) 399, 708-712 (1999). 78. C. H. Hardman, R. W. Broadhurst, A. R. C. Raine, K. D. Grasser, J. O. Thomas, and E. D. Laue, Biochemistry 34, 16596-16607 (1995). 79. M. H. Werner, J. R. Huth, A. M. Gronenborn, and G. M. Clore, Cell (Cambridge, Mass.) 81, 705-714 (1995). 80. J. J. Love, X. Li, D. A. Case, K. Giese, R. Grosschedl, and P. E. Wright, Nature (London) 376, 791-795 (1995). 81. Q. He, U.-M. Ohndorf, and S. J. Lippard, Biochemistry, 39, 14426-14435 (2000). 82. E. H. Blackburn, Nature (London) 350, 569-573 (1991). 83. C.W. Greider, Cell (Cambridge, Mass.) 97, 419-422 (1999). 84. J. D. Griffith, L. Comeau, S. Rosenfield, R. M. Stansel, A. Bianchi, H. Moss, and T. de Lange, Cell (Cambridge, Mass. ) 97, 503-514 (1999). 85. B. van Steensel, A. Smogorzewska, and T. de L~ge, Cell (Cambridge, Mass.) 92, 401-413 (1998). 86. N. W. Kim, M. A. Piatyszek, K. R. Prowse, C. B. Harley, M. D. West, P. L. C. Ho, G. M. Coviello, W. E. Wright, S. L. Weinrich, and J. w. Shay, Science 266, 2011-2015 (1994). 87. C, M. Counter, H. W. Hirte, S. Bacchetti, and C. B. Harley, Proc. Natl. Acad. Sci. U.S.A. 91, 2900-2904 (1994). 88. T. de Lange, Proc. Natl. Acad. Sci. U.S.A. 91, 2882-2885 (1994). 89. W. C. Hahn, C. M. Counter, A. S. Lundberg, R. L. Beijersbergen, M. W. Brooks, and R. A. Weinberg, Nature (London) 400, 464-468 (1999). 90. D. Hanahan and R. A. Weinberg, Cell, (Cambridge, Mass.) 100, 57-70 (2000). 91. J. N. Burstyn, W. J. Heiger-Bernays, S. M. Cohen, and S. J. Lippard, Nucleic Acids Res. 28, 4237-4243 (2000). 92. T. Ishibashi and S. J. Lippard, Proc. Natl. Acad. Sci. U.S.A. 95, 4219-4223 (1998). 93. T. de Lange, EMBOJ. 11,717-724 (1992). 94. A. Haghighi, S. Lebedeva, and R. A. Gjerset, Biochemistry 38, 12432-12438 (1999). 95. A. Asai, Y. Kiyozuka, R. Yoshida, T. Fujii, K. Hioki, andA. Tsubura, AnticancerRes. 18, 14651472 (1998). 96. A. M. Burger, J. A. Double, and D. R. Newell, Eur. J. Cancer 33, 638-644 (1997). 97. R. Kuddus and M. C. Schmidt, Nucleic Acids Res. 21, 1789-1796 (1993). 98. J. L. Kim, D. B. Nikolov, and S. K. Burley, Nature (London) 365, 520-527 (1993). 99. Y. Kim, J. H. Geiger, S. Hahn, and P. B. Sigler, Nature (London) 365, 512-520 (1993). 100. S. M. Cohen, E. R. Jamieson, and S. j. Lippard, Biochemistry 39, 8259-8265 (2000). 101. A. M. J. Fichtinger-Schepman, A. T. van Oosterom, E H. M. Lohman, and F. Berends, Cancer Res. 47, 3000-3004 (1987).
BIOCHEMISTRY OF CISPLATIN
129
102. E. Reed, S. H. Yuspa, L. A. Zwelling, R. F. Ozols, and M. C. Poirier, J. Clin. Invest. 77, 545-550 103. 104. 10,5. 106. 107. 108. 109. I10. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135.
(1986). M. Kartalou, L. D. Samson, and J. M. Essigmann, Biochemistry 39, 8032~8038 (2000). A. H. Wyllie, J. F. R. Kerr, and A. R. Currie, Int. Rev. Cytol. 68, 251-306 (1980). A. H. Wyllie,J. Pathol. 153, 313-416 (1987). M. A. Barry, C. A. Behnke, and A. Eastman, Biochem. Pharma. 40, 2353-2362 (1990). C. M. Sorenson and A. Eastman, Cancer Res. 48, 6703-6707 (1988). C. M. Sorenson and A. Eastman, Cancer Res. 48, 4484~488 (1988). C. M. Sorenson, M. A. Barry, and A. Eastman,]. Natl. Cancer Inst. 82, 749-755 (1990). B. Salles, J.-L. Butour, C. Lesca, and J.-P. Macquet, Biochem. Biophys. Res. Commun. 112, 555-563 (1983). A. Mitsui and P. A. Sharp, Proc. Natl. Acad. 8ci. U.S.A. 96, 6054-6059 ([999). D. B. Bregman, R. Halaban, A. J. van Gool, K. A. Henning, E. C. Friedberg, and S. L. Warren, Proc. Natl. Acad. Sci. U.S.A. 93, 11586-11590 (1996). M. Hochstrasser, Annn. Rev. Genet. 30, 405-439 (1996). J. N. Ratner, B. Balasubramanian, J. Corden, S. L. Warren, and D. B. Bregman,J. Biol. Chem. 273, 5184-5189 (1998). L. R. Kelland, S. Y. Sharp, C. F. O'Neill, F. I. Raynaud, P. J. Be'ale, and I. R. Judson, J. Inorg. Biochem. 77, 111-115 (1999). R. P. Perez, Eur. ]. Cancer 34, 1535-1542 (1998). M.A. Graham, G. F. Loc~vood, D. Greenslade, S. Brienza, M. Bayssas, and E. Gamelin, Clin. Cancer Res. 6, 1205-1218 (2000). G. Los, P. H. A. Mutsaers, M. Ruevekamp, and J. G. McVie, CancerLett. 51,109-117 (1990). W. Schmidt and S. G. Chaney, Cancer Res. 53, 799-805 (1993). D. Fink, H. Zheng, S. Nebel, P. S. Norris, S. Aebi, T.-P. Lin, A. Nehm~, R. D. Christen, M. Haas, C. L. MacLeod, and S. B. Howell, Cancer Res. 57, 1841-1845 (1997). J. F. Hartwig and S. J. Lippard, J. Am. Chem. Soc. 114, 5646-4654 (1992). A. Vaisman, M. Varchenko, A. Umar, T. A. Kunkel, J. I. Risinger, J. c. Barrett, T. C. Hamilton, and S. G. Chaney, Cancer Res. 58, 3579-3585 (1998). A. Vaisman, S. E. Lira, S. M. Patrick, W. C. Copeland, D. C. Hinkle, J. j. Turchi, and S. G. Chaney, Biochemistry 38, 11026-11039 (1999). E. L. Mamenta, E. E. Poma, W. K. Kaufmann, D. A. Delmastro, H. L. Grady, and S. G. Chaney, Cancer Res. 54, 3500-3505 (1994). G. R. Gibbons, W. K. Kaufmann, and S. G. Chaney, Carcinogenesis 12, 2253-2257 (1991). X. Zhai, H. Beckmann, H.-M. Jantzen, and J. M. Essigmann, Biochemistry 37, 16307-16315 (1998). B. A. J. Jansen, j. van der Zwan, J. Reedijk, H. den Dulk, and J. Brouwer, Eur J. Inorg. Chem., 1429-1433 (1999). J. Kasp~rkov& K. J. Mellish, Y. Qu, ~d Brabee, and N. Farrell, Biochemistry 35, 16705-16713 (1996). Y. Qu, M. J. Bloemink, J. Reedijk, T. W. Hambley, and N. Farrell, j. Ant. Chem. Soc. 118, 9307-9313 (1996). Y. Qu, H. Rauter, A. P. S. Fontes, R. Bandarage, L. R. Kelland, and N. Farrell, J. Med. Chem. 43, 3189-3192 (2000). Y. Zou, B. Van Houten, and N. Farrell, Biochemistry 33, 5404-5410 (i994). V. Brabec, j. Kasp~irkovfi,O. Vrfina, O. Nov~.kov&J. w. Cox, Y. Qu, and N. Farrell, Biochemistry 38, 6781-6790 (1999). J. Kasp~irkovfi,N. Farrell, and V. Brabec, J. Biol. Chem. 275, 15789-15798 (2000). E.J. Martin and R. E, Critchlow, J. Comb. Chem. 1, 32~,5 (1999). R. E. Dolle and K. H. Nelson, Jr.,]. Comb. Chem. 1,235-282 (1999).
130
SETH M. COHEN AND STEPHEN J. LIPPARD
136. E. Danielson, J. H. Golden, E. W MeFarland, C. M. Reaves, W. H. Weinberg, and X. D. Wu, Nature (London) 389, 944--948 (1997). 137. E. Danielson, M. Devenney, D. M. Giaquinta, J. H. Golden, R. C. Haushalter, E. W. McFarland, D. M. Poojary, C. M. Reaves, W. H. Weinberg, andX. D. Wu, Science 279, 837-839 (1998). 138. M. B. Francis, T. F. Jamison, and E. N. Jacobsen, Curt Opin. Chem. Biol. 2, 422-428 (1998). 139. C. J. Ziegler, A. P. Silverman, and S. J. Lippard, J. Biol. Inorg. Chem. 5, 774-783 (2000). 140. Z. Guo and P. J. Sadler, Angew. Chem. Int. Ed. 38, 1512-1531 (1999). 141. E Caravan, J. J. Ellison, T. J. McMurry, and R. B. Lauffer, Chem. Rev. 99, 2293-2352 (1999). 142. V. Sharma and D. Piwnica-Worms, Chem. Rev. 99, 2545--2560 (1999). 143. A. R. Khokhar, Y. J. Deng, S. Al-Baker, M. Yoshida, and Z. H. Siddik, J. Inorg. Biochem. 51, 677-687 (1993). 144. B. J. S. Sanderson, L. R. Ferguson, and W. A. Denny, Murat. Res. 355, 59-70 (1996). 145. F. D. Rochon and P. C. Kong, Can. J. Chem. 64, 1894-1896 (1986). 146. Y. Chen, Z. Guo, S. Parsons, and P. J. Sadler, Chem. Eur J. 4, 672-676 (1998). 147. Y. Chen, J. A. Parkinson, Z. Guo, T. Brown, and P. J. Sadler, Angew. Chem. Int. Ed. 38, 2060-2063 (1999). 148. K. E. Sandman, S. S. Maria, G. Zlokarnik, and S. J. Lippard, Chem. Biol. 6, 541-551 (1999). 149. G. Zlokarnik, P. A. Negulescu, T. E. Knapp, L. Mere, N. Burres, L. Feng, M. Whitney, K. Roemer, and R. Y. Tsien, Science 279, 84-88 (1998). 150. K. Y. Chau, H. Y. P. Lain, and K. L. D. Lee, Exp. Cell Res. 241,269-272 (1998). 151. Q. He, C. H. Liang, and S. J. Lippard, Proc. Natl. Acad. Sci. U.S,A. 97, 5768-5772 (2000).
Molecular and Cellular Biology of the Human Reduced Folate Carrier 1 LARRY H. MATHERLY 2
Developmental Therapeutics Program KarT~umosCancer Institute Department of Phar~nacology Wayne State University School of Medicine Detroit, Michigan 48201 I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Background: Folate Metabolism and the Biochemical Rationale for Folate Antagonists as Anticancer Drugs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Metabolic Roles of the Natural Folates . . . . . . . . . . . . . . . . . . . . . . . . . . . B. The Use of Folate Antagonists in Cancer Therapy . . . . . . . . . . . . . . . . . . 111. Functional Properties of RFC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1V. Properties of the hRFC Protein and Isolation and Characterization of hRFC cDNAs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Identification of a Highly Glycosylated hRFC Protein in Transport-Upregulated K562 Human Erythroleukemia Cells . . . . . . . . . B. Cloning of the hRFC cDNAs That Restore Transport Function to Transport-Impaired Hamster and Human Cells . . . . . . . . . . . . . . . . . . C. Does Disparate hRFC Expression in Transport-Impaired K562 Cells Reflect a Role for Non-hRFC Transport Components and/or Posttranslational Modifications of the hRFC Protein? . . . . . . . . . . . . . . . Transcriptional and Posttranseriptional Regulation of hRFC Expression... A. hRFC Transcript Heterogeneity Involves Variable Splicing of Alternative Upstream Exons and lnternal Deletions from Full-Length hRFC Transcripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Identification of Dual hRFC Promoters: Relationships to Tissue-Specific hRFC Gene Expression . . . . . . . . . . . . . . . . . . . . . . . . . . C. Regulation of hRFC Levels and Function by Wild-Type p53 . . . . . . . . . VI. Molecular Mechanisms of Transport-Mediated MTX Resistance . . . . . . . . . A. General Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Molecular Mechanisms of RFC-Mediated MTX Resistance . . . . . . . . . .
132 133 133 135 138 139 139 141
146 147
147 151 152 153 153 154
1Abbreviations: Acute lymphoblastic leukemia, ALL; N~-(4-amino-4-deoxy-10-methylpteroyl)N~-(4-azido-5-1zSI-salieylyl)-L-lysine, APA)2Sl-ASA-Lys; chinese hamster ovary, CHO; (6R)5,10-dideaza-5,6,7,8-H4PteGlu, DDATHF; dihydrofolate reductase, DHFR; 10-ethyl-10-deazaaminopterin,10-EDAM; folate receptors, FRs; hemagglutinin, HA; human RFC, hRFC; N-hydroxysuccinimide, NHS; methotrexate, MTX; multidrug resistance proteins, MRPs; nmltitargeted antifolate, MTA; open reading frame, ORF; 5'-rapid amplification of eDNA ends, 5'RACE; reduced folate carrier, RFC; tetrahydrofolate, H4PteGlu; transmembrane domain, TMD. 2Address correspondence to: Larry H. Matherl~, Ph.D., Karmanos Cancer Institute, 110 East Warren Avenue, Wayne State University School of Medicine, Detroit, M1 48201. Phone: (313) 833-0715, ext. 2407; FAX: (313) 832-7294; email: [email protected]. Progressin NucleicAcidResearch and MolecularBiology,Vol.67
131
Copyright© 2001by AcademicPress. Allfightsof reproductionin anyfor~nreserved. 0079-6603/01 $35.00
132
LARRYH. MATHERLY C. Insights into the Structural-Functional Determinants of RFC from Studies of Mutant Carriers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
156 157 158
The natural folates are water-soluble members of the B class of vitamins that are essential for cell proliferation and tissue regeneration. Since mammalian cells cannot synthesize folates de novo, tightly regulated and sophisticated cellular uptake processes have evolved to sustain sufficient levels of intracellular tetrahydrofolate cofactors to support the biosynthesis of purlnes, pyrimidines, serine, and methione. Membrane transport is also a critical determinant of the antitumor activity of antifolate therapeutics (methotrexate, Tomudex) used in cancer chemotherapy, and impaired uptake of antifolates is a frequent mode of drug resistance. The reduced folate carrier is the major transport system for folates and classical antifolates in mammalian cells and tissues. This review summarizes the remarkable advances in the cellular and molecular biology of the human reduced folate carrier over the past decade, relating to its molecular structure and transport function, mechanisms of transcriptional and posttranscriptional regulation, and its critical role in antifolate response and resistance. Many key in vitro findings have now begun to be extended to studies of reduced folate carrier levels and function in patient specimens, paving the way for translating basic laboratory studies in cultured cells to improvements in human health and treatment of disease. The results of research into the human reduced folate carrier should clarify the roles of changes in expression and function of this system that accompany nutritional folate deficiency and human disease, and may lead to improved therapeutic strategies for enhancing drug response and circumventing resistance in cancer patients undergoing chemotherapy with antifolates.
© 2001 Academic Press.
I. Introduction The natural folates are water-soluble members of the B class of vitamins that are essential for cell proliferation and tissue regeneration (1). Chemically, folates are hydrophilic anionic molecules that normally show only minimal capacities to cross biological membranes by diffusion alone. Accordingly, mammalian cells have evolved tightly regulated and highly sophisticated uptake processes for facilitating m e m b r a n e traverse of the natural folates. Internalization of folate cofactors by mammalian cells generally involves two primary systems, the highaffinity folate receptors (FRs) (2-5) and the reduced folate carrier (RFC) ( 4 8 ) , which are distinguishable by their unique patterns of tissue expression, their uptake mechanisms and transport kinetics, and their divergent specificities for transport substrates. It appears that at least one of these uptake systems must be
HUMANREDUCEDFOLATECARRIER
133
present in mammalian cells to provide sufficient levels of folate cofactors for the synthesis of nucleotide and amino acid precursors necessary for cell proliferation. Furthermore, folate transport systems participate in specialized tissue functions essential for in vivo folate homeostasis, such as absorption of folates across the intestinal luminal epithelium (9, 10), folate reabsorption in the proximal tubules of the kidney (11, 12), and transplacental folate uptake (13, 14). In addition to its important role in folate homeostasis, membrane transport is a critical determinant of the antitumor activity of antifolate therapeutics [e.g., methotrexate (MTX), Tomudex] used in cancer chemotherapy, and impaired cellular uptake of antifolates is a frequent mode of drug resistance (2, 4-8). Although both the FRs and RFC can contribute to antifolate uptake and resistance, RFC appears to be of greater pharmacologic importance to current treatment modalities for cancer. This reflects its higher transport capacity, its ubiquitous tissue expression, and its particular specificity for 4-amino antifolate drugs such as MTX. In this review, I summarize our recent studies of the human RFC (hRFC). Although RFC was first kineticallycharacterized more than 30 years ago in mouse leukemia cells (15-17), the past decade has seen a near-exponential increase in o u r collective understanding of the biochemical and molecular properties of this essential membrane transport system. This progress was, in large part, catalyzed by the cloning of rodent RFCs in 1994 (18, 19) and, shortly thereafter, hRFC cDNAs in a number of laboratories including my own (20-24). The cloning of hRFC soon fostered a number of other important advances in the field, including the characterization of hRFC gene structure (25-27), the identification of dual hRFC promoters with unique patterns of cis regulatory elements (25, 26), and the development of specific antibodies for characterizing the structure and function of the hRFC protein (28). Many key in vitro findings are now being extended to studies of hRFC levels and function in patient specimens (29-31), paving the way for translating basic laboratory studies in tissue culture models to improvements in human health and treatment of disease. In this review, I summarize the remarkable recent advances in the cellular and molecular biology of this physiologically important and fascinating reduced folate membrane transport system.
II. Background: Folate Metabolism and the Biochemical Rationale for Folate Antagonists as Anticancer Drugs A. Metabolic Roles of the Natural Folates Tetrahydrofolate (H4PteGlu) cofactors are essential to cell survival because these derivatives participate in one-carbon transfer reactions leading to the biosynthesis of purine nueleotides, thymidylate, serine, and methionine (Fig. 1).
134
LARRY H. MATHERLY
Folates are predominantly found within cells as polyglutamates, characterized by the presence of additional (generally 2-8) F-linked glutamyl residues (32). The glutamate condensation reaction, catalyzed by folylpoly-F-glutamyl synthetase, confers enhanced cellular retention and increased rates of carbon transfer over monoglutamyl folates (32). Folate-dependent biosynthetic steps involve one-carbon units attached to the H4PteGlu 5 or l0 positions at the oxidation levels of formaldehyde, formate, or methanol. Whereas carbon transfers during serine, methionine, or purine biosynthetic reactions generate unsubstituted H4PteGlu, during thymidylate synthase catalysis, 5,10-methylene H4PteGlu is consumed as it is oxidized to dihydrofolate. Dihydrofolate is reduced back to H4PteGlu bythe housekeeping enzyme dihydrofolate reduetase (DHFR), whereupon the latter is free to associate with one-carbon units (derived from formate or serine) and participate in one-carbon transfer reactions (Fig. 1). Both D H F R and thymidylate synthase are critical to cell replication and are subject to stringent transcriptional and posttranseriptional controls (33-37).
DihydrofolateReductase l H4PteGlu*
ThymidylateSynthase[
.~Glycine I, 5'10"CH2"H4PteGIu
+ dUMP
""A. ~ H PteGlu
Formate
10"CHO'H4PteGlu 4
5,10-CH+-H4PteGlu
5-CH3-H4PteGIu | Homocysteine B 1 2
5-CHO-H~PteGIu S-AdenosylMethionine FIG. 1. Metabolism of natural folates. The pathways for the interconversion and biosynthetic utilization of the natural folate cofactors are illustrated. Important enzyme targets for antifolates used in cancer chemotherapy are noted and include thymidylate synthase (E.C.2.1.1.45), dihydrofolate reductase (E.C. 1.5.1.3), AICAR formyl transferase (E.C.2.1.2.3), and GAR formyl transferase (E.C.2.1.2.2). Abbreviations: H4PteGlu, tetrahydrofolate; H2PteGlu, dihydrofolate; GAR, glycinamide ribonucleotide; AICAR, aminoimidazoleearboxamideribonucleotide. Adapted from L. H. Matherly and S. P. Muench, Evidence for a localized conversionof endogenous tetrahydrofolates to dihydrofolateas an important element in antifolate action in murine leukemia cells. Biochem. Pharmacol. 39, 2005-2014 (1990); copyright o 1990, with permission from Elsevier Science.
HUMANREDUCEDFOLATECARRIER
135
Some degree of coordinate regulation is necessary, since if DHFR activity were limiting or altogether absent, the continued generation of dihydrofolate during thymidylate biosynthesis would eventually deplete cellular 5,10-methylene H4PteGlu pools. Likewise, other H4PteGlu forms, following their conversion to 5,10-methylene H4PteGlu and, subsequently, to dihydrofolate, would be depleted. This provides a compelling rationale for the therapeutic use of DHFR inhibitors as antibacterial, antiprotozoal, and antineoplastic agents (38). Although somewhat oversimplified (39), the net result of DHFR inhibition by these drugs is a "depletion" of pools of reduced folates, an accumulation of dihydrofolate, and a suppression of folate-dependent biosyntheses of DNA, RNA, and protein precursors. The central role of the folate antagonists in cancer therapy is further considered in the following section.
B. The Use of Folate Antagonists in Cancer Therapy The antifolates represent some of the oldest, yet still most useful agents in the chemotherapeutic armamentarium (40, 41). In 1948, Farber et al. (42) reported that aminopterin was effective in inducing remissions in childhood acute lymphoblastic leukemia (ALL). Despite its greater antitumor activity (43), aminopterin was soon replaced by MTX in clinical use because the considerable host toxicity accompanying its use compromised therapeutic efficacy (43). Furthermore, toxicity with MTX was more predictable, possibly owing to its greater purity and stability (44). Although a large number of antifolates have been synthesized and tested in the intervening years, today MTX remains the only antifolate to achieve widespread clinical use for cancer, as an essential component of multidrug regimens for treating ALL, choriocarcinoma, non-Hodgkin's lymphoma, osteosarcoma, breast cancer, and head and neck cancer (40, 41). In recent years, there has been a renewed focus on the development of antifolate drugs affording improved selectivity over MTX [10-ethyl-10-deazaaminopterin (10-EDAM or Edatrexate)] (45) or capable of circumventing MTX resistance by inhibiting other targets such as thymidylate synthase [ZD1694 (Tomudex), GW1843U89, MTA Cmultitargeted antifolate")] (46-48), and/or the folatedependent purine biosynthetic enzymes aminoimidazolecarboxyamide ribonueleotide (MTA) (48) and glyeinamide ribonucleotide [(6R)-5,10-dideaza-5,6,7,8H4PteGlu (DDATHF or Lometrexol)] (49) formyl transferases. These antifolate structures are depicted in Fig. 2. The agents are currently in various stages of clinical development. ZD1694 has been approved for use in Europe for treating advanced eoloreetal cancer and is currently in clinical trials in the United States. Like their folate counterparts, the "classical" antifolates, including MTX and these "new-generation" antifolate drugs, are converted to poly-y-glutamate conjugates (8, 39, 50). The polyglutamyl forms of MTX and related antifolates
,,p
0 U
U
U
o
,,p U
~
~
C
~
¢J
3[:
=
~
g
~
s
as
o
0 U ¢J Z U
i
I
"U
°~
-r
ff
_ _
Z
~
137
HUMAN REDUCED FOLATECARRIER
5,10-Dideaza-5,6,7,8-Tetrahydrofolate
H3C~N\~
~
9
C02H
GW1843U89
H2r~
.X_
Y" W ) o
@ LY231514 (MTA) FIG.2. (Continued)
are at least equivalent to their monoglutamyl forms as inhibitors of their principal intracellular targets. For some agents (e.g., ZD1694, DDATHF, MTA), the affinities of the enzyme targets for polyglutamyl antifolates exceed those for the parent nonpolyglutamyl drugs by orders of magnitude (8, 39, 46, 48, 49). Moreover, because long-chain (>2 glutamates) antifolyl polyglutamates are retained within cells over unmetabolized monoglutamyl antifolate, their accumulation within cells is accompanied by a sustained inhibition of intracellular targets long after the extracellular (i.e., plasma) antifolate concentrations have declined to low levels (8, 39). Finally, MTX polyglutamates can act as inhibitors of
138
LARRYH. MATHERLY
enzyme targets other than DHFR, including thymidylate synthase and aminoimidazolecarboxyamide ribonucleotide formyl transferase, thereby interfering with the metabolic utilization of exogenous H4PteGlu cofactors [i.e., (6R,S)-5-formyl H4PteGlu or leucovorin] during leucovorin "rescue" from MTX toxicity (39).
III. Functional Properties of RFC MTX is actively transported into mammalian cells by the same RFC-mediated process normally used by 5-methyl H4PteGlu and other reduced folates (4-8, 15-17) (Fig. 9.). For H4PteGlu cofaetors, adequate rates of intracellular delivery are essential for DNA synthesis and sustained cell proliferation. For MTX, membrane transport is critical to antitumor action because of its role in generating sufficient unbound intracellular antifolate for maximal DHFR inhibition and synthesis of MTX polyglutamates (4, 6, 8, 39). A number of other antifolate drugs (aminopterin, 10-EDAM, ZD1694, GW1843U89, DDATHF, MTA) are also excellent transport substrates for RFC (4, 5, 45-49). The functional properties of RFC-mediated transport have been exhaustively characterized using both rodent and human (generally tumor) models in culture (4-8, 15-17), generally incubated with radiolabeled folate analogs (i.e., 3H-MTX or 3H-aminopterin) not subject to the extensive metabolic interconversions of the natural folates. RFC transport properties are remarkably uniform between species (6, 7), with caveats, as noted in Section IV,C. Thus, RFC-mediated uptake is saturable at low (41-5 ~M) concentrations for most transport substrates (5-8). Furthermore, transport is temperature-dependent, maximal at neutral pH, sodium-independent, and highly sensitive to competitive inhibition by folate analogs (5-8). Transport appears to involve a physical transposition of the carrier protein within the plasma membrane since efflux of intracellular folates from cells via RFC can be directly coupled to MTX influx, a phenomenon termed" trans-stimulation" (51). An analogous mechanism is believed to account for the uphill (concentrative) transport of (anti)folates into cells, during which divalent anions (e.g., phosphate, adenine nucleotides) are extruded via RFC down large concentration gradients from the intracellular to the extracellular compartments (8, 52). Although the mechanistic details are not well established, this may involve binding of intracellular anions to RFC at the folate binding site, followed by a more rapid orientation of"loaded" (bound) carrier over "unloaded" (unbound) carrier from the inside to the outside membrane surface, where it is available to bind (and transport) extracellular substrates. Transport of MTX via RFC can occur bidirectionally (see above); however, in reality, the actual flux via RFC is small compared to other modes of efflux (8, 53, 54). Indeed, the major mechanism(s) of MTX efflux from mammalian
HUMAN REDUCED FOLATE CARRIER
139
cells appears to involve one or more exit pumps, altogether separate from RFC, such as the family of multidrug resistance proteins (MRPs) which are sensitive to cellular energetics (8, 53-57). MRPs 1, 2, and 3 can transport MTX from transfected cell lines overexpressing these proteins, thereby decreasing drug accumulation and conferring MTX resistance (55-57). However, the extent to which the individual endogenously expressed MRPs participate in the efflux of the natural folates and folate analogs is far less certain. Likewise, the extent to which changes in drug efflux contribute to the maintenance of cellular and tissue folate pools, or to the development of antifolate resistance, is unestablished. It is interesting that the parent structure, folic acid, binds poorly to RFC compared to MTX and reduced folates (4-8, 52, 58). This suggests that cellular uptake of this cofactor form is largely independent of RFC. Other uptake mechanisms for folic acid have been suggested, including a poorly characterized "low-affinity/high-capacity" transporter (4-8, 58), a transport system with optimal activity at low pH (4, 5, 7, 59, 60), and a family (i.e., or, fi, y) of high-affinity glycosylphosphatidylinositol-anchored membrane FRs (2-5). The latter exhibit high (nM) affinities for folic acid, 5-methyl H4PteGlu, and certain newer antifolates (e.g., DDATHF) compared to MTX and other 4-amino antifolates (2-5). However, FRs are also characterized by far slower transport "turnover" rates than RFC (61, 62). Although neither the physiologic nor the pharmacologic role of this unique family of membrane receptors is entirely certain, in tissues or tumors in which they are expressed in high levels (e.g., ovarian carcinomas), the FRs may easily become a significant transport route at physiologic concentrations of circulating folates (20-50 nM) and at low (nM) plasma concentrations of particular antifolates. However, uptake via the RFC probably predominates at higher pharmacologic concentrations of antifolate drugs when both systems are present.
IV. Properties of the hRFC Protein and Isolation and Characterization of hRFC cDNAs A. Identification of a Highly Glycosylated hRFC Protein in Transport-Upregulated K562 Human Frythroleukemia Cells For many years, biochemical studies of the RFC protein were limited owing to its low levels of expression in cultured mammalian cells. To circumvent this, Sirotnak et al. devised a novel strategy to select L1210 murine leukemia cells that upregulate this transport system (63, 64). Selection was based on the notion that carrier-mediated uptake is rate-limiting to H4PteGlu cofactor utilization in rapidly dividing cells and involved growing cell cultures on
140
LARRYH. MATHERLY
progressively decreasing and growth-restricting concentrations of (6R, S)-5formyl H4PteGlu. Under these conditions, only cell variants that possessed elevated transport capacities for reduced folates were capable of sustained growth. On this basis, L1210 cell lines (designated R82 and R83) were developed for which markedly increased capacities for folate cofactor and antifolate transport via RFC (14- and 40-fold, respectively) were accompanied by similarly increased expressions of a cell surface protein capable of binding 3H-aminopterin at 0° (63, 64). This implied that the elevated transport capacities under these selection conditions probably resulted from an increased synthesis of RFC protein. In 1991, we used an analogous approach to select a transport-upregulated K562 human erythroleukemia cell line (designated K562.4CF) (65). Like the R82 and R83 L1210 lines, K562.4CF cells were selected in a low concentration of folate [i.e., 0.4 nM (6R, S)-5-formyl H4PteGlu] and exhibited striking increases in initial influx rates (4.4- and 6.8-fold, respectively) and steady-state accumulations (3.1- and 4.2-fold, respectively) of 3H-5-formyl H4PteGlu (0.1 #M) and 3H-MTX (0.5 #M). Based on earlier reports by Henderson et al. (66) describing the use of N-hydroxysuccinimide(NHS) 3H-MTX for covalently labeling RFC from L1210 cells, we performed analogous experiments in K562.4CF cells (65). Although an "~42-48-kDa plasma protein, identified as RFC, was specifically labeled with this reagent in L1210 cells (63, 64, 67), in K562.4CF cells, NHS-3H-MTX was incorporated into a broadly migrating band at ~76-85 kDa. Covalent incorporation of 3H into this band increased up to 7-fold in K562.4CF plasma membranes over wild-type K562 cells (65). Since labeling was completely blocked by transport substrates [unlabeled MTX, (6S)-5-formyl H4PteGlu, and, to a lesser degree, folic acid] in direct proportion to their affinities for RFC, it seemed a near certainty that the broadly migrating NHS-3H-MTX-labeled protein in K562.4CF cells was hRFC. We reasoned that the anomalous migration profile of the hRFC protein on SDS gels and the apparent size discrepancy between the functionally homologous murine and human carriers might be attributed to differences in their glycosylation (65). To directly examine this possibility, K562.4CF cells were labeled with NHS-3H-MTX, plasma membranes were prepared, and detergentsolubilized membrane proteins were digested with a commercial glycosidase (endo-fi-galactosidase). Endo-/3-galactosidase treatment resulted in a significant sharpening of the broad radioactive hRFC band and a shift to a substantially lower molecular weight (estimated size of 57,800 4- 7730; n = 5), confirming its glycosylation. Conversely, endo-fl-galactosidase caused no change in the migration of the murine RFC. Similar results were subsequently reported by Freisheim and coworkers for transport-upregulated CCRF-C E M cells (CEM-7A) treated with APA-lZSI-ASA-Lys [N~-(4-amino-4-deoxy-10-methyl pteroyl)-N~-(4-azido-5JzSI-salicylyl)-L-lysine], a photoaffinity ligand for RFC
HUMANREDUCEDFOLATECARRIER
141
(68), and in HL60 cells treated with NHS ~H-aminoptern (69). Thus, it seemed probable that differences in the extent of glycosylation between the homologous human and murine RFCs, at least in part, were responsible for the disparity in their molecular masses. The functional significance of hRFC N-glycosylation is considered in greater detail in the following section.
B. Cloning of the hRFC cDNAs That RestoreTransport Function to Transport-Impaired Hamster and Human Cells In 1989, Flintoff and colleagues reported that H4PteGlu cofactor and MTX membrane transport activity could be restored in transport-impaired chinese hamster ovary (CHO) cells by transfections with genomic DNAs from drugsensitive human or hamster cells (70). This approach was subsequently extended to transfections with CHO genomic DNA cosmid clones (71), thus setting the stage for the eventual cloning of RFC. In 1994, Dixon et al. used expression cloning to isolate a cDNA from L1210 cells, reportedly for the murine RFC, that could restore MTX transport function and MTX sensitivity to transport-impaired ZR-75-1 human breast carcinoma cells (18). This was soon followed by a report that the homologous cDNA clone from a CHO cell cDNA library restored MTX transport and binding activity to transport-impaired CHO cells (19). Although the true identity of the cloned transporter was initially controversial, attention nonetheless quickly shifted to the cloning and characterization of the homologous human cDNA(s). In 1995, four reports on the human cDNAs were published almost simultaneously (20-23). This was followed in 1997 by the description of a nearly identical cDNA clone from human intestine (24). In our laboratory, a family of human cDNA clones was isolated from a cDNA library prepared from our own hRFC transport-upregulated K562.4CF subline (see above) by hybridization screening with the putative murine RFC cDNA (21). Of the three unique cDNAs (designated KS6, KS32, and KS43; 1.4 kb, 2.5 kb, and 2.8, respectively; Fig. 3, upper panel) characterized in detail, the longest form (KS43; Genbank # U19870) contained a 98-bp 5' untranslated region (UTR), an open reading frame (ORF) of 1776 bp, preceded by a Kozak consensus sequence, and a 3' UTR of 864 bp, followed by a poly(A) sequence. The predicted molecular mass of the translated protein is 64,873 Da, a close approximation to the estimated size of the enzymically deglycosylated NHS-3H-MTX-labeled hRFC from intact K562.4CF cells in our earlier study (65). With minor variations (see below), the KS43 cDNA is identical to cDNAs reported for other human tissues, including those from placenta (Genbank Accession # U15939) (23), lymphoblasts (U175566) (20), intestine (AF004354) (24), and testis (NM003056) (22).
LARRYH. MATHERLY
142 0
1
2
I
I
3 I
kb
KS6
~ATG
I
I
~ A T G
KS32
KS43
I1
1ATG
I
KS6 -93 G C C G C G T G T C G G G G C ~ T G C G G G G T G T C T C G G G G C C CT G G G G T G A G T G C G G G C G C G G G C C G A G G T T ~ C CTGT GAGGCACAGTGTCA C
C
T
T
C
G
T
C
C
C
C
T
C
C
E
G
A
C
~
T
G
C
A
C
G
T
G
G
C
~
.
G
G
A
T
G
G
~
~
G
T
G
. .
KS32 -3"/3 TCGACCGC GCCCAGGCACGTGTTGCTTCGGGACGGGCCAGCCGGTGGGT~CCTGC~CAC~GT~ GGCGGGCCCCTGGCACATCTCCAC~CCATTGTC~CTGTGC~GCT~£GT~iGT~CT TCCCCTTGGAGCAGCTG~TGCGGATC~~TCCC~GTCTCAG~ CTGGA~CGCT~C'Z~TCCCC~GTGGGTCCTTCCA~C~CGCT&CAGC~~ TCCCCCT&GGGTCCTGTCCATC~CTCCTGTCCTGGG~.GTCTGGCCTGCCTGGCCT~ ~C.~CT.a~GCTGGCCC C~TC CAG~J~TGCATC~~CAGTG~CA C C T T C G ~ % " ~ ' ~ C T C ~ ~ . A C G 3 ~ . G C C ~ r ~ _ ~ G C A G C ~ C r . . . G C ~ g C C C T C ~ G C G ~
. .
KS43 -98
GGGCCGGGTCCGGGAGCCC~GCCGCCCCGCCGAGTCGCAGGCACJIG~TCA
C C T ~ G T C C C ~ C ~ T G C A C G ~ G G A T G G ~ - _ . , C C C ~ C A G C C C A G C G T G
. .
FIG. 3. hRFC cDNAs and heterogeneous 5' UTRs. Upper panel: A schematic of the 1~aultiple hRFCs reported by Wong et al. (21) and described in the text. The solid white regions correspond to the hRFC ORF, the solid black regions indicate the constant upstream sequence up to position -51, and the striped or dotted regions represent the unique 5' UTRs, beginning at position -52, as described in the text. Lower panel: Heterogeneous 5' UTRs for the hRFC transcripts, as determined from sequencing the KS6, KS32, and KS43 cDNAs. The constant regions up to position -51 are italicized and the ATG translational start is underlined.
The predicted KS43 amino acid sequence shows homologies of 64-66% with the mouse (18), hamster (19), and rat (Genbank Accession # U38180) proteins (Fig. 4). Interestingly, the single N-glycosylation site predicted for the KS43encoded protein (at asparagine 58) was conserved in the hamster and the rat, but not the mouse, proteins. All the rodent sequences contained at least one additional consensus N-glycosylation site not present in the h u m a n transporter. By hydropathy analysis ( T M P R E D ) (72), the KS43-encoded protein conformed to
HUMAN REDUCED FOLATECARRIER Human Rat
~vPs~z~P~Laswa~r.v~crYGr~xapensex~p~z~r~ MVPT~QVGEKQACESPRQDRHLKSWR|LVFYLCFFGFM£QLRPGESFZTPYLL|~_RNFT
I N~'PTGQVAEKQACHHPRQDRBLKSWRCLVFYLCFFGFNIQLRPGHSPZTPYLZ~NFT
Hamster Mouse Human Rat Hamster Mouse
143
MVPTG~VAEKQA~EPRQD~LKSWRCLVFYLCFFGFMAQLRPGHSFZTPFLL~-R|FT
ss n~WnZZPX~SX~aWWZ~D~H~ZPWV~C~Srvcw~SVVHMQ
59 |BOVTMEIIPVLPYeHLAVLVPZFLLTDYLRYKPZLILQCLSFMCVWLLLLLGTeVVHM0 59 KHQVTNHIIPMLPYSHLAVLVPVFLLTDYLRYKPVLVLQCLeFVCVWLLLLLGTSVVHMQ
Human Rat Hamster Mouse_
121 119 119 119
Human Rat Hamster Mouse_
,,,
Human Rat HameteE House
236 236 238 236
Human Rat HameCsr Mouse
295 TD_T~L~MG~VDAASTLLB~ZT~FTAGFV~ZRWALWS~LVZ~VZAZQAGLVFCMF 297 ~LH~YNGAVDAASTLLBAZTSFBAGFVKZRWALWSKLVZAJJVZA~QAGLVFCMY 295 T D S S L ~ Y M G A V D A A S T L L S A I T S F S A G F L m I R ~ W S K L V I A | V I A I Q A | L V F C M F
Human Rat
Hamster HOUSe Human
Hamster Mouse Human Rat Hamster Mouse_
Human Rat
avs 179 179
LMBLFYSVTMAARZAY88YZFeLVRP|RYQRV~SRAAVLLGVF~SVLGQLLVTVG|V LMHVFYSZT~AHZATHeYZFeLV'dPeRY~RMASTSR~VLLGVF~eVLGQVLVTLQJ~ LMHVFYSVT~HZAYSeYZFSLVRPSRYQRMASYSHAAVLLGVFm~BVLG~VL~L~|| LMEVFYSVTJLLARZAYSSYZFSLV|PSRYQRMASTSRAAVLLGVF|SSVLG~|LVTVG|Z
:SL~rZLTSr4~SLrLHR~nSLr~NHSALVma~LPCSLDQMHP~PG~PX ZSLGFZIFSL|LSLFLKRP~LFYNRSAL~LPCHLDQMHPGPGRPE ~SLGFZLFSLVLSLFLHRPERSLFFNRe|L|R|~ALPCELDGMHPGP|RPE
~ L ~ | L ~ | C | D S| L | R M L ~ L~D_|LR|P QLRLW|L~WVFNSAGYYLV|YYVHl L W | | P~LHRML~|TCHDeFLVHMLSHLVJJNVHOPOLRLMCLWWVFNSAGYYLZBYYVHVLWHZ P|KLHHVL~CH~FL~LSHLV|HLHQP|VHLMCLWWVFNSa~YYLI~YVHVL~ ||KLDRML~|TCRDSFLVRMLSHLVRI~QPQLRLWCLWWVFNS~YYLZ|I'fYVHVLMRe
351 Q/~ZWVCYVTFVLFRGAYQFLVPZATFOZABSLSKELCALVFGZNTFLATALK 352 ~ W V ~ M T Y V L P R G A Y Q F L V P I A T F Q I A S S L B K H L C A L V F G I H T F L A T A L K 412 T|zTlizvs~Ip:GL~LliVeH0rBZ,Y B V T F L Z L S Z Z ~ L D O L ~ ~ O | 405 T~TLVVSDKHGLGLQV~FRIYJJMTFL~LSIZ~GLDGLHY||RGHMQPL~_AAQJJ 43.2 T ~ TLVVSDKRGLGL~QF|Z Y|VYFMVLSVI CF||K~LDGVRYC~ERGRHQPL ~ Q Z 405 T | I TLVVSDKRGLGLQVH|QFRl Y|ZYFLML S I|CF~GLDGLHYC|RGRH0 P L ~ 0 E 472 46 S 472 465
L~E~|~QAL SVQD ~ Z 4 ~ P ~ P | L | P H D | ~ L Z~ L~SPLH~pVQ&Z 8 L Q D G ~ ' , ~ Q P S ~ P ~ L L P B D G ~ V I D ~ [ I P . A ~ A K A | I ~ , q P L~ V Q ~ M Q D ~ P 8 ~ PQLLP E D G ~ H D ~ L P ~ A K & ~ LR~PL~VQAZeL~D~Pe~PQLL~RDG~HD~GD~L~AHAm
53.2
Hamster Mouse_
FIG. 4. Aminoacid homologiesfor the human, mouse, hamster, and rat hRFCs. Aminoacid homologiesare basedon the GenbanksubmissionsofeDNAsequencedatafor the human (U19870), mouse (L23755),hamster (U03031),and rat (U38180)RFC cDNAs.
a model expected for an integral membrane protein, with 11 or 12 stretches of 17-25 mostly hydrophobic, u-helix-promoting amino acids, an internally oriented C terminus, and a single outwardly facing N-glyeosylation consensus site at asparagine 58 (Fig. 5). An analogous 12 transmembrane domain (TMD) structure was predicted for the rodent proteins (7, 18). There is a high degree of sequence conservation of the primary amino acids for several of the putative TM Ds (i.e., TMDs 1-5, 7, and 8), and there is a nearly complete lack of homology in the large central loop connecting TM Ds 6 and 7 and in the carboxyl-terminal region (Figs. 4 and 5). Moreover, the human KS43 eDNA-encoded protein contains 72-79 more amino acids than the homologous rodent proteins.
144
LARRYH. MATHERLY
FIG. 5. Topologystructure of the human reduced folatecarrier. A topologymodelfor hRFC is showndepicting 12 TMDs, internally oriented amino and carboxyltermini, and an externally oriented N-glycosylationsite at asparagine58.
The two additional cDNA clones (KS6, KS32; Fig. 3) isolated from K562.4CF cells differed from KS43 in several notable ways. For instance, their 5' noncoding regions were of differing lengths (93 and 373 bp, respectively) from that for the KS43 cDNA (98 bp). All these forms were completely nonhomologous starting at position -52 (Fig. 3, lower panel). Furthermore, the KS6 cDNA included only the most 5' region (~1300 bp) of the hRFC ORF, and KS32 contained a 626-bp deletion starting at position 1568 in the ORF and an additional 4 bp (TGTG) in the 3' UTR immediately upstream from the poly(A) tail at position 2718. As a result of the internal deletion in the KS32 cDNA, the in-frame translational stop codon for KS43 (at position 1774) is lost and the KS32 coding sequence continues into the 3' noncoding region until it reaches a new stop codon at position 2205. The net result is that the predicted KS32 ORF contains 14 bp (including the stop codon) not used in KS43, and encodes 4 new carboxyl-terminal amino acids (LRCS). The predicted molecular mass for the 526 amino acid KS32 protein is 58,394 Da. Most significantly, both the KS43 and KS32 cDNAs were able to restore 3H-MTX transport function and MTX sensitivity to transport-impaired hamster cells, although an interesting human/hamster "hybrid" transport phenotype was observed (21). Upon transfection of transport impaired human K562 cells (designated K500E) with the KS43 cDNA, 3H-MTX uptake exceeded that for parental cells by 2.2-2.6-fold, and for "mock" transfected cells by 22- to 26fold (74). Interestingly, the restored transport exhibited typical hRFC properties, including characteristic sensitivities to competitive inhibition by transport substrates (GW1843U89, folic acid, 5-formyl H4PteGlu, Tomudex; Fig. 2) and to irreversible covalent inhibition by unlabeled NHS-MTX, and a capacity for trans-stimulation by elevated pools of intracellular reduced folates (73). The
HUMAN REDUCED FOLATECARRIER
]45
KS43-encoded protein was detected by photoaffinity labeling (73) or on Western blots as a broadly migrating species of similar size ('-~85-92 kDa; Fig. 6) to that previously identified in K562.4CF by NHS-3H-MTX labeling. As before, the broadly migrating protein band shifted to ",65 kDa upon enzymic deglycosylation, confirming the glycosylation of the single consensus site at position asparagine 58. Collectively, the functional and structural features of the KS43encoded protein convincingly established its identity as hRFC. A functional ~65-kDa protein was also detected on Western blots of plasma membrane proteins from K500E cells transfected with Gln-58 KS43 cDNA in which asparagine 58 (AAC) was replaced by glutamine (CAG) (Fig. 6) (74). Wild-type (Asn-58) and Gln-58 hRFC proteins containing a hemagglutinin (HA) epitope (YPYDVPDYASL) at the carboxyl terminus were similarly functional and, by immunofluorescence staining of permeabilized cells with rhodamineconjugated anti-HA antibody, both were localized to the cell surface (74). Thus,
/ -97.4 hRFC
-68
GIn58hRFC
5
10
FIG.6. Westernblot of hRFC and dg-hRFC. Data are shownforwesternblot of plasma membrane proteins from K562transfectantsexpressingwild-typehRFC (K43-6)and hRFC for which a glutamine is substituted for asparagine at position 58 (GIn5ShRFC;KEDG-1). Proteins (5 and 10/zg, respectively)were fractionatedon a 7.5% gel in the presence of SDS and electroblotted onto a PVDF membrane. Detectionwas with anti-hRFC antibody"(28) and an enhanced chemiluminescencekit. Molecularweight standards (in kDa) are shown.
146
LARRYH. MATHERLY
the expressed hRFCs in our transfected cells specifically localize to plasma membranes, and the extent of N-glycosylation appears to play no significant role in either transport function or surface targeting of the carrier. Our glycosylation experiments also shed light on the membrane topology of hRFC since they confirm an extracellular orientation for asparagine 58 (Fig. 5). Because immunofluorescence detection of the carboxyl HA-hRFC required cell permeabilization, an internal cytosolic orientation of the hRFC carboxyl terminus is implied (74). An identical result was obtained for HA-hRFC containing HA epitopes in the large loops connecting TMDs 4 and 5 (G158), and 6 and 7 (E226), stably expressed in K500E cells (X. Liu and L. H. Matherly, manuscript in preparation). Conversely, permeabilization was not required for HA insertions into the loops connecting TMDs 3 and 4 (N120), 5 and 6 ($183), and 7 and 8 (E294), suggesting an extracellular orientation for these regions (X. Lin and L. H. Matherly, manuscript in preparation). Most recently, Ferguson and Flintoff (75) used an identical HA-epitope insertion strategy to partially map the topology of hRFC expressed in CHO cells. Their major findings are also consistent with the topology model depicted in Fig. 5 and include cytoplasmic orientations for proline 20 (N terminus), serine 225 (loop between TMDs 6 and 7), and glycine 492 (carboxyl terminus), and an external orientation for proline 297 (loop between TMDs 7 and 8) (75).
C. Does Disparate hRFC Expressionin Transport-Impaired K562 Cells Reflect a Role for Non-hRFC Transport Components and/or Posttranslational Modifications of the hRFC Protein? As described earlier, when expressed in transport-impaired cells, the mouse
(18, 76), hamster (19), and human (20-24) RFC cDNAs all confer an array of properties typical of the classical endogenously expressed carrier. However, certain characteristics of the restored transport are clearly inconsistent with those expected for "normal" RFC function. For instance, in a series of human (K562) transfectants from our laboratory, the restored transport approximated only 3-30% of the high levels expected from affinity labeling or Western blotting assays of the hRFC protein (73, 74). Similar results were obtained in highly MTXresistant CCRF-CEM cells transfected with the KS43 hRFC eDNA (A. Gifford, M. Norris, M. Haber, and L. H. Matherly, manuscript in preparation), and in transport-impaired K562 cells expressing a tetracycline/doxycycline-inducible KS43 hRFC construct, over a wide range of hRFC levels (B. C. Ding and L. H. Matherly, manuscript in preparation). For the latter, a linear relationship between hRFC protein and transport activity was evident at low levels of doxyeyclineand hRFC expression. However, activity soon plateaued with higher concentrations of doxycycline, even as hRFC protein continued to increase. In
HUMANREDUCEDFOLATECARRIER
147
still other studies, significant differences in transport properties were described for endogenously expressed versus transfected RFCs. These include relative bidirectional transport fluxes (77), transport substrate specificities (21, 78), and pH dependencies (78-80) for RFC-mediated uptake. Collectively, these findings suggest the possibility that membrane transport by exogenously expressed RFC may be modulated by endogenous cellular factors. This may reflect the involvement ofnon-RFC transport components, similar to the sodium-dependent D-glucose cotransporter (SGLT1), for which a putative accessory RS1 protein was identified in cotransfection experiments (81, 82). For folate transport, the notion of an "'accessory" transport protein is not new, and was previously advanced by Underhill et al. (83) from studies of reversion analysis and somatic cell hybrids between transport-defective CHO lines. Furthermore, a number of putative non-RFC transport proteins have been implicated in RFC function. These include a 46-kDa murine "reduced folate transporter" (84), a 66-kDa nmrine phosphoprotein (85), an unidentified 38-kDa cytosolic or peripheral membrane protein in L1210 and CCRF-CEM cells treated with APA-leSI-ASA-Lys(68), and a 92-kDa NHS-aH-MTX-binding glycoprotein in K562 cells (GP-MTX) (86). These results suggest a functional and structural complexity of reduced folate membrane transport, including the possibility that RFC may exist as a nmlticomponent system. Of course, other explanations can be envisaged to explain the discrepancy between transport activity and RFC protein in the K562 and CCRFCEM transfectants, or the altered binding affinites in hRFC-transfected CHO cells (21). These include effects on protein folding, plasma membrane insertion/orientation, and/or posttranslational modifications of hRFC. An additional possibility is that hRFC transport is subject to regulatory controls responsive to folate cofactors or other cellular metabolites (6,5, 69, 87). This may occur directly at the level of the hRFC or be mediated by a secondary membrane or nomnembrane component(s), as well.
V. Transcriptional and Posttranscriptional Regulation of hRFC Expression A. hRFC Transcript Heterogeneity InvolvesVariable Splicing of Alternative Upstream Fxons and Internal Deletions from Full-LengthhRFC Transcripts Our findings of heterogeneous hRFC cDNAs differing in their 5' UTRs (Fig. 3) and the presence of ORF and 3' UTR deletions (21) were hardly expected. Multiple hRFC transcripts were confirmed in CCRF-CEM cells by 5' RACE (rapid amplification of eDNA ends) analysis (26). Two groups of transcripts were detected in CCRF-CEM cells, one containing a variable-length up
148
LARRYH. MATHERLY
stream sequence, identical to the original KS43 cDNA 5' UTR, and another consisting of variants containing the outermost segment, the middle portion, or the full-length KS32 5' UTR. Within each group, a range of 5' termini was detected, suggesting multiple transcriptional starts. However, the 5' UTR for KS6 was not detected in CCRF-CEM cells by 5' RACE (26). These findings were corroborated by the results of primer extension analysis of CCRF-CEM and K562.4CF transcripts, in which numerous extension products were detected (26). We initially considered several possibilities to explain this extraordinary level of hRFC transcript diversity. These included the existence of(l) multiple polymorphic hRFC genes or (2) a single hRFC gene for which the sole primary transcript was variably spliced. Alternatively, (3) multiple hRFC transcripts with different 5' UTRs may be generated by separate noncoding exons from unique promoters within a single hRFC gene, with an additional diversity ofhRFC transcript forms resulting from alternative splicing. To distinguish between these possibilities, we characterized the structure of the hRFC gene from human leukocyte and placental genomic libraries. The human hRFC gene contains seven exons, including two alternative noncoding exons (exons i and 2), flanked by separate promoters (termed A and B or, previously, "Pro32" and "Pro43," respectively; see below), and spanning ~-29 kb (Fig. 7A) (27). The intron-exon junctions all exhibit a GT/AG consensus sequence with the exception of the splice donor for exon 3, which contains a GC. The size of the hRFC gene is somewhat larger than either the homologous murine (~23 kb) (88) or hamster (15.3 kb) (89) RFC genes. Although there is little similarity between the intron sizes for the human and rodent RFC genes, the intron--exon junctions are highly conserved. Exon 7 in the hRFC gene is larger than the homologous murine and hamster exons, reflecting the extended carboxyl terminus ofhRFC (Fig. 4). The upstream organization of the hRFC gene is depicted in Fig. 8 (upper panel). The unique 5' UTRs for the KS43/KS32/KS6 transcripts can be localized to separate noncoding exons immediately upstream from a large ('~3.4 kb) intron and exon 3 containing the translation start site. Both the KS43 and KS6 5t UTRs are included in a single noncoding exon (exon 1) and are separated by only 6 bp, suggesting that they most likely represent splice variants. The KS32 5' UTR is localized to exon 2. The truncated KS32 5' UTRs are both flanked by consensus GT donor elements, suggesting that they most likely represent splice variants. Most importantly, these results establish that hRFC trancripts with diverse 5' UTRs can be encoded by a single gene locus. The hRFC gene structure determined in our laboratory is virtually identical to that deduced by Williams et al. (90); however, there are unexplained differences from that reported by Tolner et al. (25), involving the acceptor splice junctions for exons 6 and 7 and the donor splice junction for exon 6 and in the
HUMAN REDUCED FOLATE CARRIER
149
A ATG
Exon ProB
1
2 Pro A
~,3 I1
4
6
5
I2
][3
15
14
1 kb
Exon 6
B
,I
KS43 KS32
Exon 7
Intron 5 Pl
....
7a
7b
7c 7d
6 7a ~
7b
7c 7d
....
6 7a 7c 7d ~ 6 7d
KS1
.... lkb
F1G.7. hRFC gene structure and the predicted downstream structures ofthe full-length(KS43) and 2 variant hRFC transcripts (KS1, KS32) resulting from exon 7 deletions. Pane] A: A schematic is shown depicting the structure of the hRFC gene as previously described (27). The ATG in exon 3 represents the predicted translatiomd start codon. The lengths of the exons (El-E7) are 165, 401, 238, 760, 202, 142, and 1426 bp, respectively. The approximate sizes of the introns, Ii-I5, are 3.4, 3.8, ~110 bp, 4.8, and ~10 kb, respectively. The upstream region is highlighted in greater detail in Fig. 8. Panel B: The 3r ORF transcript heterogeneity of human RFC cDNAs is depicted. The schematic (top) shows the downstream region of the hRFC gene including a portion of exon 6, intron 5, and exon 7 (depicted as segments a-d). Also shown are the predicted structures for the full-length (KS43) and two variant RFC transcripts (KS32 and KS1), characterized by deletions of variable portions of exon 7. As described in the text, for KS32, segment 7b (626 bp, 1568-2193)is deleted. KS1 contains a deletion of segment 7a (275 bp, 1294-1568), 7/0 (1569-2280), and 7c (88 bp, 2194-2281), and contains onlythe 3' end ofexon 7 or segment 7d (457 bp, 2282-2738) fused to exon 6. By utilizing new stop codons at positions 2205 and 2456, respectively, KS32 would contain an additional four amino acids (LRCS) and KS1 an additiona] 58 amino acids (NEELHVASLSLWKSHLRL AADTLSSEGS SGSGPRSWFLSPTLRAALHGPVCPSEVCPS) not present in KS43. Adapted from L. Zhang, S. C. Wong, and L. H. Matherly, Structure and organization of the human reduced folate carrier gene. Biochim. Biophys. Acta 1142, 389-393 (1998); copyright © 1998, with permission from Elsevier Science.
e s t i m a t e d sizes for i n t r o n s 3 (110 b p vs. 3100 bp), 4 (4800 b p vs. 2000 bp), a n d 5 (10,000 b p vs. 5900 bp). T h e bases for these differences are unclear. As n o t e d in Section III,B, t h e KS32 e D N A isolated in o u r laboratory (21) also c o n t a i n e d an i n t e r n a l d e l e t i o n o f b o t h o p e n r e a d i n g frame a n d 3' U T R s e q u e n c e . F r o m t h e d e d u c e d h R F C g e n e structure, this w o u l d result b y d e l e t i n g 626 b p
150
-2016
LARRY H. MATHERLY
999bp
PromoterB lntron2
PromoterA -1088 ~ -1043 GGACGGACCCGCCCACCCCGCAGCCGCGCGCCCGCCGCGCCGCCTT
Basal Promoter B -5o,
[ CRE/AP1 [
-45s
CGC~CCCCAGCCTGCCCTCCGCGTCATCCTGGGGCGCCAAGTCCCA
Basal Promoter A FIG. 8. Upstream organization of the hRFC gene and identication of basal promoters. A schematic is shown depicting the structure of the upstream region of the hRFC gene, including exons 1-3 and the dual promoters (hRFC-A and -B), as described in the text. Also shown are the basal hRFC promoters, including the highly conserved GC box and the CRE/AP1 element.
from exon 7 (positions 1568-2193; Fig. 7B). A separate internally deleted hRFC form was identified by RT-PCR of total RNAs from an assortment of human leukemia cell lines (K562, CCRF-CEM, and REH) (27). This truncated hRFC form (designated KS1) contained a 988-bp nucleotide deletion (positions 12942281) and includes only the 3' end ofexon 7 (457 bp, positions 2282-2738) fused to exon 6 (Fig. 7B). By utilizing a new stop codon (at 2456) in the 3' UTR, the putative hRFC variant protein would contain 58 amino acids not present in the full-length hRFC clone. This resulting truncated --~54-kDa protein is missing the carboxyl-terminal hRFC sequence including TMD 12 and, unlike the KS32 protein, appears to be nonfunctional (91). The origins of the KS32/KS1 hRFC variants are not entirely certain. However, since conserved splice donor (GT-) and acceptor (-AG) elements are present in the deleted segments of the KS1 variant, this form can be envisaged to arise by alternative splicing of the full-length hRFC transcript. However, for KS32, there are no recognizable cryptic splice donor or acceptor consensus sequences in the deleted segment to support the notion that this form is also a splice variant. Although the signficance of these forms is not established, to date, they have been detected only at low levels (27), or in cells (K562.4CF, CEM-7A) selected for growth in low concentrations of folates and overexpressing hRFC (65, 91).
HUMANREDUCEDFOLATECARRIER
151
B. Identification of Dual hRFC Promoters: Relationships to Tissue-Specific hRFC Gene Expression The juxtaposition of the 5' flanking regions to exons 1 and 2 (Fig. 8, upper panel) suggested that each of these could be a separate promoter. Although computer analysis of these highly GC-rich DNA fragments failed to identify either canonical TATA or CAAT motifs within the first 100 bp upstream of the transcription initiation sites, a number of conserved potential transcription elements could nonetheless be identified (Spl, AP1, CRE, MZF1, E2E AP2, etc.). In order to confirm promoter activities and identify the minimal regions required for basal transcription, we fused each of these 5' flanking regions (positions -2016 to -959 designated RFC-B; and -958 to -277, designated RFC-A) to a firefly luciferase reporter gene (in pGL3-Basic), for transient expression in HT1080 human fibrosarcoma and HepG2 human hepatoma cells (26, 92). Both the hRFC-A and -B reporter gene eonstructs showed potent, orientationdependent promoter activities. By 5' deletion analysis, the major promoter activities were localized within 47-bp region for RFC-A (positions -501 and -455) and within 46 bp for RFC-B (positions -1088 and -1043), upstream of the major transcriptional start sites for exons 2 and 1, respectively (Fig. 8). From database analysis, the RFC-B minimal promoter contained a highly conserved GC box (eccgcec) and a CRE/API-like element (ecgegtcatcct) was identified in promoter A (Fig. 8). When these elements were mutated and the mutant hRFC-B (cccgaac) and -A (ccgcgtcctcct) minimal promoters expressed in HT1080 and HepG2 cells, promoter activities decreased 60-90%, thus confirming important transactivating roles for these regions. Protein binding to these elements was established by electromobility gel shift assays with nuclear extracts prepared from HT1080 and HepG2 cells and by competition with competitive oligonucleotides and by snpershift assays (92). Accordingly, DNA-protein complexes involving Sp1 and Sp3 were bound to the minimal hRFC-B promoter in both HT1080 and HepG2 cells; for the RFC-A promoter, DNA-protein complexes with CREB-1 (HT1080 and HepG2), cJun (HT1080), and ATF-1 (HepG2) were identified. Sp1 and Sp3 would seem to be reasonable candidates to regulate constitutive RFC-B transcription, given their ubiquitious expression (93). Sp3 exists as three iso~brms, generated by different translation starts, all of which can (to varying degrees) act as activators or repressors of transcription, depending on the cell and promoter context (93). Accordingly, changes in the intracellular levels of Spl and the Sp3 isoforms could conceivably exert profound effects on transcription from the hRFC-B promoter and the generation of KS43 hRFC transcripts. The downstream hRFC-A promoter contains a consensus CRE/AP1 element, recognized by dimerie complexes of members of the b-ZIP superfamily (CREB-1, c-Fos, cJun, ATF-1) (94-96). In our studies, CREB-1 was able
152
LARRYH. MATHERLY
to form a DNA-protein complex with the basal hRFC-A promoter in both HepG2 and HT1080 cells. Conversely, an ATF-1 promoter complex was identified only in HepG2 and a cJun complex was detected only in HTI080 cells (92). These findings suggest a tissue-specific regulation of the hRFC-A minimal promoter. Furthermore, they imply a combinatorial regulation of the hRFCA promoter via the binding of a diversity of b-ZIP homo- and heterodimeric complexes. The complexity of the bZIP superfamily and the intricate relationships among these factors underscore the regulation of hRFC-A, and may explain a previous report of effects of second-messenger pathways on hRFC levels or function (97). Of course, the level of transcription ultimately achieved depends on a large number of factors, including the levels of the individual transcription factors and the presence of upstream and downstream inhibitory and activating elements, as well as the overall promoter architecture. Clearly, the mechanisms that determine differential promoter activities and alternate promoter usage between cell or tissue types is an area that needs to be further explored. Thus, the wide diversity of hRFC transcript forms results from multiple transcriptional starts and variable splicing of two alternative 5' noncoding exons, each transcribed from its own promoter and containing a unique distribution of cis elements. This promoter duality should provide a potentially powerful means of ensuring adequate levels ofhRFC transcripts (and protein), in response to tissue requirements for folate cofactors, or to other exogenous tissue- or cell-specific signals. Another level of transcriptional control could arise from differences in tissue levels or distributions of the AP1/CREB/ATF transfactors (promoter A) or relative levels of Spl and Sp3 isoforms (promoter B). Finally, differences in translation efficiencies for mature mRNAs or rates of mRNA turnover could reflect the synthesis of hRFC mRNA isoforms with unique 5' noncoding exons with differing lengths and/or secondary structures (98).
C. Regulationof hRFCLevelsand Function by Wild-Type p53 Following the intriguing finding of Thottassery et al. (99) in H35 rat hepatoma cells that loss of wild-type p53 function leads to an enhanced sensitivity to MTX via a mechanism independent of DHFR, we initiated an interesting series of experiments to explore the possibility that hRFC may be a physiologically important target for this tumor suppressor (100). Two complementary approaches were used: (1) REH lyanphoblastic leukemia cells were treated with the anticancer drug daunorubicin (0.5/zM) to induce endogenous wild-type p53; (2) p53-null K562 cells (K562 ptet'°'vp53) were engineered to express wild-type p53 under control of a tetracycline (doxycycline)-inducible promoter so that p53 protein and p53 regulated processes could be studied in a tightly controlled
HUMANREDUCEDFOLATECARRIER
153
manner in direct response to the addition of doxyeycline. Accordingly, we found that induction of p53 resulted in a potent and striking loss (3-5-fold) of hRFC transcripts, decreased levels of immunoreactive hRFC protein, and a suppression of MTX transport (100). Although the mechanism for this p53 effect on hRFC levels and function is not entirely certain, a number of key features have been established. (1) p53 induction results in increased levels of p21 protein, indicating that it is transcriptionally active. However, the effects on hRFC are likely not due to a cell-cycle blockade by p21, since induction of the eyclin-dependent kinase 4/6 inhibitor p15 in an analogous tetracycline-inducible model resulted in a nearly identical loss of S-phase fraction without any significant effect on hRFC levels or MTX transport. (2) The turnover of hRFC transcripts in the presence of actinomycin D was unaffected by p53 induction, implying that the effects of p53 on hRFC are likely to be transcriptional. Finally, (3) we observed a potent inhibition of hRFC-B promoter activity by p53 in transiently transfeeted K562 ptet'°rdp53 cells which could be localized to the 46-bp Spl- and Sp3-dependent hRFC-B minireal promoter (positions -1088 and -1043); however, no effect was observed on the formation of DNA/Spl and Sp3 protein complexes by gel shift analysis. Collectively, our results establish a potent transcriptional suppression of hRFC by p53, apparently independent of its well-established effects on cellcycle progression and Sp1/Sp3 binding to the highly conserved GC-box element in the hRFC-B minimal promoter. Most significantly, they suggest that for the nearly 50% of tumors that express wild-type p53, p53 induction by any of an assortment of DNA damaging agents (e.g., chemotherapy, irradiation) could possibly attenuate the effectiveness of antifolate therapy owing to reduced hRFC levels and decreased drug uptake. Likewise, decreased uptake of reduced folates by this mechanism may exacerbate the effects of DNA damaging agents, due to perturbations of nucleotide pools required for DNA repair. The role of hRFC in MTX resistance is further considered in the following section.
VI. Molecular Mechanisms of Transport-Mediated MTX Resistance A. General Considerations Membrane transport of MTX and related agents into tumor cells is key to their chemotherapeutic effectiveness because the level of drug achieved within cells is an important determinant of the extent of drug binding to intraeellular enzyme targets (i.e., DHFR). Membrane transport is also critical to the metabolism of antifolates to polyglutamyl forms required for drug retention and, for certain analogs (DDATHF, Tomudex), high-affinity binding to intracellular enzymes
154
LARRYH. MATHERLY
(Section II,B). Thus, it is not surprising that impaired membrane transport is a common mode of antifolate drug resistance (2, 4-8). Decreased drug uptake generally involves alterations in RFC. For MTX, impaired RFC function has been reported in cultured murine and human tumor cells (28, 67, 73, 101-105, 107, 108) and in murine tumor cells derived in vivo during MTX chemotherapy (109). In addition, a wide range ofhRFC expression was described in leukemia blasts from patients with acute lymphoblastic leukemia and low levels of hRFC were detected at relapse, suggesting an important role for hRFC in clinical resistance (29, 30). For osteosarcomas, decreased hRFC expression was also associated with a poor response to chemotherapy including MTX (31). For MTX, the relationships among drug uptake, binding to DHFR, and polyglutamate synthesis can be complicated because MTX resistance is multifaceted and can only rarely be attributed to a single mechanism. For instance, increased levels of DHFR (110, 111), kinetically altered DHFR (112, 113), and/or decreased synthesis of MTX polyglutamates (114, 115) may accompany impaired MTX influx, frequently making it difficult to evaluate the importance of one mode of resistance over another. In other cases, essentially identical metabolic or pharmacologic effects are achieved by different modes of resistance (i.e., increased DHFR and impaired MTX transport). Impaired MTX transport that renders tumor ceils insensitive to conventional doses of drug should, at least in part, be circumvented by increasing the extracellular concentrations of antifolate, thereby forcing drug into cells by a mutated RFC with reduced substrate affinity, by an alternative transport route altogether, or, on a limited basis, by simple passive diffusion. Regardless of the mode of cell entry, a point is eventually reached for which increasing extracellular drug doses are no longer accompanied by significant changes in the levels of intracellular drug. This is due to the saturability of the membrane carrier(s), the electrical restrictions on drug accumulation, and/or the presence of a high-capacity efflux pump(s) (8). The net effect is that relatively small increases in cellular DHFR or seemingly small losses in drug uptake can quickly result in a requirement for an intracellular MTX concentration that is unattainable in vivo. These considerations exemplify the pharmacologic elements associated with impaired MTX transport and MTX resistance. In the following sections, I describe the results of recent studies of the molecular changes in RFC structure, function, and expression accompanying the development of MTX resistance.
B. Molecular Mechanisms of RFC-Mediated MTX Resistance Transport alterations can manifest as reduced rates of RFC transloeation (reduced Vmo,0, decreased affinities for transport substrates (increased Kt), or a composite of these effects. With the cloning of cDNAs for the rodent (18, 19) and human (20-24) RFCs, a number of laboratories, including my own,
HUMANREDUCEDFOLATECARRIER
155
began to explore the underlying bases for defective MTX uptake and consequent resistance. For the murine RFC, a number of RFC mutants were identified for which transport is completely abolished (e.g., A130P) (76) or which exhibit altered substrate-binding specificities ($297N, I48F, E45K) (101, 103, 106). Interestingly, for two distinct MTX-resistant CCRF-CEM lymphoblastic leukemia sublines from separate laboratories (104, 105), an identical lysine-to-glutamate substitution at position 45 (E45K) in hRFC was associated with impaired MTX uptake and resistance, as well as a markedly increased uptake of reduced folates and folic acid resulting in elevated intracellular folate pools. In spite of the high frequency at which this E45K variant occurs in cultured cells, we have not yet detected this variant in more than 200 ALL blast specimens from children at diagnosis or relapse (116). A similar transport phenotype was characterized in DDATHF-resistant L1210 cells for which a replacement of isoleucine 48 by phenylalanine (I48F) and tryptophan 105 by glycine (W105G) in the murine RFC resulted in near-normal transport of antifolate (MTX, DDATHF) but a strikingly increased uptake of folic acid (108). Losses of RFC expression accompanying MTX resistance have been widely reported in cultured lines but have, in general, been poorly characterized (28, 73, 76, 106, 107). In our laboratory, the nearly complete loss ofhRFC transcripts and protein in highly transport-impaired K562 (designated K500E) cells was associated with decreased transcripts from the upstream hRFC-B promoter (73, 117). This was independent of changes in hRFC D NA sequence, including both the hRFC coding region and upstream region, and was not due to differences in the levels of transcription factors required for hRFC-B transcription (117). Rather, the only detectable alteration that might be causal involved losses of two hRFC alleles and a translocation of the lone remaining hRFC allele, which resulted in transcriptional silencing (117). In still other reports, losses of full-size hRFC protein were described in spite of seemingly normal levels ofhRFC transcripts (28, 107, 108). For a highly MTXresistant ('-~250-fold) CCRF-CEM subline in our laboratory, we found that losses of hRFC protein were due to an early translation termination resulting from a 4-bp (CATG) insertion at position 191 (the donor splice junction for exon 3) and an increased (~ 12-fold) rate of protein turnover of a doubly mutated (G44R, S127N) hRFC protein (28). Similar findings of frameshift mutations (107, 108) and increased rates of turnover (108) of mutant RFCs have been reported elsewhere. Interestingly, an identical CATG insertion at position 191, presumably due to aberrant splicing, has been identified in hRFCs from ALL blasts (116). For one patient with low levels of MTX transport, 60% of the hRFC transcripts contained a CATG insertion at position 191 (T. R. Witt and L. H. Matherly, unpublished observation). Thus, it appears that a wide range of molecular alterations can account for the MTX transport-impaired phentoype, including hRFC mutations, aberrant
156
LARRYH. MATHERLY
mRNA splicing, gene deletions and translocations, and increased rates of carrier turnover. Evidence is beginning to accumulate that some of these mechanisms may be relevant to clinical resistance to MTX, as well. Studies of mutant RFCs with altered substrate binding or impaired rates of membrane traverse have begun to implicate functionally important residues or domains in the carrier protein. This is further considered in Section VI,C.
C. Insights into the Structural-Functional Determinants of RFC from Studies of Mutant Carriers The characterization of the molecular alterations in RFC accompanying MTX resistance has begun to shed light on amino acid residues or domains that are critically important to transport substrate binding and/or membrane translocation. Most notable is TMD1, for which mutations in highly conserved residues (Fig. 4) are all accompanied by profound effects on transport substrate binding [e.g., G44R (28), E45R (I03-105), I48F (106)]. Mutants E45R and I48F are highly substrate-specific since affinities and uptake for folic acid (103-106) and/or 5-formyl H4PteGlu (103-105) are increased whereas binding of MTX is decreased or unaltered. For E45, increased binding affinities for 5-formyl H4PteGlu and folic acid were preserved with some substitutions (E45Q, E45R) but not others (E45D, E45L, and E45W) (118), suggesting that size rather than charge is the predominant structural consideration at this position. Indeed, these findings suggest that glutamate 45 does not directly participate in substrate binding, but plays an important structural role in generating an optimum conformation for binding of folic acid and 5-formyl H4PteGlu. Interestingly, serine 46, also located in TMD1, has also been implicated as important to RFC function in L1210 cells since $46N results in decreased rates of membrane translocation with minimal effects on transport substrate binding (102). Other regions of potential functional and structural importance suggested by mutant studies include TMD8 [i.e., $309 ($313 in hRFC) and $297 in murine RFC] (102, 120) and TMD 3 [i.e., S127 in hRFC and A130 in murine RFC (A132 in hRFC)] (28, 76). For the $309F murine RFC, uptake of both MTX and reduced folates was decreased compared to wild-type; however, this effect was significantly less when 5-CHaH4PteGlu and 5-CHOH4PteGlu were used as substrates (119). Whereas the $297N mutation in the external loop between TMD7 and -8 in the murine RFC results in a decreased affinity for MTX compared to aminopterin and 10-deazaaminopterin or 10-ethyl-10-dazaaminopterin (without a Vm~ effect), this residue is not conserved among species (Fig. 4). Very recent studies from our laboratory have used deletion and insertion mutagenesis to explore determinants of hRFC fucntion. Thus, removal of the 16 amino-terminal residues or the 105 carboxyl-terminal amino acids preserved transport activity, whereas loss of the loop between TMDs 6 and 7 (K204 to
HUMANREDUCEDFOLATECARRIER
157
R263 or D215 to R263) abolished activity (X. Liu and L. H. Matherly, manuscript in preparation). Insertion of the HA epitope into the carboxyl- (P20) or aminoterminal regions (G492) (75); the loops connecting TMDs 3 and 4 (Q120), 4 and 5 (G158), 5 and 6 (S183), or 7 and 8 (E294) (X. Liu and L. H. Matherly, manuscript in preparation); or the putative cytoplasmic loop between TMDs 6 and 7 ($225 or E226) (75, X. Liu and L. H. Matherly, manuscript in preparation), all preserved transport activity for hRFC. HA insertion into the loop between TMDs 10 and 11 (N385) reduced both binding and uptake of MTX, whereas insertion into the loop between TMDs 11 and 12 (P427) selectively abolished uptake without an effect on substrate binding (75).
VII. Concluding Remarks This chapter attempts to provide an overview of the tremendous advances and insights gained over the past decade into the cellular and molecular biology of hRFC. Although many of the experimental details and examples are based on studies from my own laboratory, I have, nonetheless, attempted to provide sufficient coverage of the topic to demonstrate both the vitality of this competitive research area and its potential importance to a range of health problems resulting from folate deficiency and impaired cofactor delivery. Also described is the importance of hRFC to the membrane transport of a number of antifolate drugs used for cancer chemotherapy and how alterations in hRFC structure and function, or its expression, are key elements in drug resistance. In spite of the progress made in many of these areas, a large number of important questions remain unresolved or unexplored. For instance, there is only scant understanding of the energetics of RFC transport or the tertiary and/or multimerie character of the transporter. Moreover, very few details have been provided concerning the translocation process itself, critical determinants of transport substrate binding, or the regions that comprise the putative membrane channel. Given the growing awareness that folate pools are compartmentalized in mitoehondria and the possible importance of drug targeting to this organelle (120), the biochemical and molecular properties of the folate/antifolate transport process(es) in mitochondria need to be investigated. Although a potential role for the family of MRPs in the efflux of folates and antifolates has been suggested from transfection studies, the role of the endogenously expressed MRPs in the maintenance of intracellular folate pools and the accumulation of cytotoxic concentrations of antifolates has not been established. As structural determinants for folate and antifolate transport become dearer, approaches to drug design must incorporate constraints dictated by the requirements for binding by both the hRFC and the MRP efflux pumps. In this fashion, it may be possible to identify antifolate inhibitors with greater cellular
158
LARRYH. MATHERLY
retention based on their increased affinities for uptake by h R F C and/or their lack of transport activity with the major efflux pumps. Although suggestions of a complex transcriptional and posttranscriptional regulation o f h R F C are beginning to emerge, in relation to the use of alternative upstream exons transcribed from unique promoters with distinct basal transcriptional elements, much remains to be learned regarding tissue-specific regulatory controls that respond to folate requirements and/or other cell- or tissue-specific signals. The results o f such studies should provide insights into the roles of changes in h R F C gene expression in relation to nutritional folate deficiency and h u m a n disease states associated with decreased accumulations of folate cofactors, including cardiovascular disease, neural tube defects, and cancer.
ACKNOWLEDGMENTS This work was supported by grants CA53535and CA77641from the National Cancer Institute, National Insitutes of Health. I thank Ms.Teah R. Witt for editorial assistance in the preparation of this article.
REFERENCES 1. E. L. R. Stokstad, in "Folic Acid Metabolism in Health and Disease" (M. F. Picciano, E. L. R. Stokstad, and J. E Greogory,eds.), pp. 1-21. Wiley-Liss,New York, 1990. 2. M. Ratnam and J. H. Freisheim, in "Folic Acid Metabolism in Health and Disease" (M. F Picciauo, E. L. R. Stokstad, and J. F. Greogory,eds.), pp. 91-120. Wiley-Liss,New York, 1990. 3. A. C. Antony,Annu. Rev. Nutr 16, 501~521 (1996). 4. G. Jansen, in "Anticancer Development Guide: Antifolate Drugs in Cancer Therapy" (A. L. Jackman, ed.), pp. 293-321. Humana Press, Totowa, NJ, 1999. 5. L. H. Matherly, in "Drug Transport in Antimierobial and Anticancer Chemotherapy" (N. H. Georgopapadakou, ed.), pp. 453-524. Marcel Dekker, New York, 1995. 6. E M. Sirotnak, Cancer Res. 45, 3992-4000 (1985). 7. F. M. Sirotnak and B. Tolner, Annu. Rev. Nutr. 19, 91-122 (1999). 8. I. D. Goldman and L. H, Matherly, Pharv~tcol. Ther 28, 77-100 (1985). 9. J. Selhub, G. J. Dhar, and I. H. Rosenberg, in "Membrane Transport of AntineoplasticAgents" (I. D. Goldman, ed.), Int. Encyl. Pharmacol. Ther., Vol. 118, pp. 147-168. Pergamon, Oxford, 1986. 10. H. M. Said, F. K. Ghishan, and R. Redha, Am. J. Physiol. 252, G229-G236 (1987). 11. H. C. Lee, R. Shoda, J. A. Krall, J. D. Foster, J. Selhub, and T. L. Rosenberry, Biochemistry 31, 3236-3246 (1992). 12. J. Selhub and I. H. Rosenberg, Proc. Natl. Acad. Sci. U.S.A. 75, 3090-3093 (1978). 13. A. C. Antony, C. Utley, K. C. Van Home, and J. F. Kolhouse,]. Biol. Chem. 256, 9684-969"2 (1981). 14. J. H. Sweiryand D. L. Yudlievich, Biochim. Biophys. Acta 821,497-501 (1985). 15. D. Kessel and T. C. Hall, CancerRes. 27, 1539-1543 (1967). 16. F. M. Sirotnak, S. Kurita, and D. J. Hutchison, Cancer Res. 28, 75-80 (1968).
HUMAN REDUCED FOLATE CARRIER
159
17. I. D. Goldman, N. S. Liehtenstein, and V. T. Oliverio, J. Biol, Chem. 243, 5007-5017 (1968). 18. K. H, Dixon, B. C. Lampher, J. Chiu, K. Kelley, and K. H. Cowan, J. Biol. Chem. 269, 17-20 (I994). 19. F. M. R. Williams, R. C. Murray, T. M. Underhill, and W. F. Flintoff, J. Biol. Chem. 269, 5810-5816 (1994). 20. F. M. R Williams and W. E Flintoff, J. Biol. Chem. 270, 2987-2992 (1995). 21. S. C. Wong, S. A. Proefke, A. Bhushan, and L. H. Matherly, J. Biol. Chem. 270, 17468-17475 (1995). 22. J. A. Moscow, M. Gong, R. He, M. K. 8gagias, K. H. Dixon, S. L. Anzick, P. S. Mettzer, and K. H. Cowan, Cancer Bes. 55, 3790-3794 (1995). 23. P. D. Prasad, S. Ramamoorthy, F. H. Leihaeh, and V. Ganapathy, Biochem. Biophys. Res. Commnn. 206, 681-687 (1995). 24. T. T. Nguyen, D. L. Dyer, D. D. Dunning, S. S. Ruhin, K. E. Grant, and H. M. Said, Gastroenterology 1112, 783-791 (1991). 25. B. Tolner, K. Roy, and F. M. Sirotnak, Gene 211, 331-341 (1998). 26. L. Zhang, S. C. Wong, and L. H. Matherly, Bioehem. J. 332, 773-780 (1998). 27. L. Zhang, S. C. Wong, and L. H. Matherly, Biochim. Biophys. Acta 1442, 389-393 (1998). 28. S. C. Wong, L. Zhang, T. L. \Vitt, S. A. Proefke, A. Bhushan, and L. H. Matherly, J. Biol. Chem. 274, 10388-10394 (1999). 29. R. Gorlick, E. Goker, T. Trippett, P. Steinherz, Y. Elisseyeff, M. Mazumdar, W. F. Flintoff, and J. R. Bertino, Blood 89, 1013-1018 (1997). 30. L. Zhang, J. w. Taub, M. Williamson, S. C. Wong, B. Hukku, J. Pullen, Y. Ravindranath, and L. H. Matherly, Clin. Cancer Res. 4, 2169-77 (1998). 31. W. Guo, J. H. Healey, P. A. Meyers, M. Ladanyi, A. G. Huvos, J. R. Bertino, and R. Gorlick, Clin. Cancer Res. 5, 621-627 (1999). 31a. L. H. Matherly and S. P. Muench, Biochem. Pharmacol. 39, 2005-2014 (1990). 32. B. Shane, Vitamins Hormones 45, 263-335 (1989), 33. J. c. Azizkhan, D. E. Jensen, A. J. Pierce, and M. Wade, Grit. Rev. EukatTjotic Gene Expression 3, 229-254 (1993). 34. C. H. Jenh, P. K. Geyer, and L. F. Johnson, Mol. Cell. Biol. 5, 2527-2532 (1985). 35. E. Chu, D. M. Koeller, J. L. Casey, J. c. Drake, B. A. Chabner, P. C. Elwood, S. Zinn, and C. J. Allegra, Proc. Natl. Acad. Sci. U.S.A. 88, 8977-8981 (1991). 36. E. Chu, C. H. Takimoto, D. Voeller, J. L. Grem, and C. J. Allegra, Biochemistry 32, 4756-4760 (1993). 37. E. A. Ereikan-Abali, D. Banerjee, M. C. Waltham, N. Skacel, K. W. Seotto, and J. R. Bertino, Biochemistry 36, 12317-12322 (1997). 38. J. c. Hardman, A. G. Gilman, and L. E, Limbird, "The Pharmacological Basis of Therapeutics" 9th ed., McGraw-Hill, New York, 1995. 39. L. H. Matherly, R. L. Seither, and I. D. Goldman, Pharmacol. The~ 35, 27-56 (1987). 40. E. Chu and C. j. Allegra, in "Cancer Chemotherapy and Biotherapy" (B. A. Chabner and D. L. Longn, eds.), 2nd ed., pp. 109-148. Lippincott-Raven, Philadelphia, 1996. 41. J. Jolivet, K. H. Cowan, G. A. Curt, N. J. Clendenin, and B. A. Chabner, N. Engl. J. Med. 309, 1094-1104 (1983), 42. S. Farber, L. Diamond, R. D. Mercer, R. H. Sylvester, and V. A. Wolt}, N. Engl. J. Med. 238, 787-793 (1948). 43. A. Goldin, J. M. Venditti, S. R. Humphreys, D. Dennis, N. Mantel, and S. W. Greenhouse, J. Natl. Cancer Inst. 15, 1657-1664 (1955). 44, L.M. Glode, S. W. Pitman, W. D. Ensminger, A. Rosowsky, N. Papathanosopoulos, and E. Frei, Cancer Res'. 39, 3707-3714 (1979). 45. S. C. Grant, M. G. Kris, C. W. Young, and F. M. Sirotnak, Cancer Invest. 11, 36-45 (1993).
160
LARRY H. MATHERLY
46. L. R. Hughes, T. C. Stephens, F. T. Boyle, and A. L. Jackman, in "Antieaneer Development Guide: Antifolate Drugs in Cancer Therapy" (A. L. Jackman, ed.), pp. 147-165. Humana Press, Totowa, NJ, 1999. 47. G. K. Smith, J. w. Bigley, I. K. Dev, D. S. Dueh, R. Ferone, and W. Pendergast, in "Anticancer Development Guide: Antifolate Drugs in Cancer Therapy" (A. L. Jackman, ed.), pp. 203-227, Humana Press, Totowa, NJ, 1999. 48. C. Shih and D. E. Thornton, in "Anticancer Development Guide: Antifolate Drugs in Cancer Therapy" (A. L. Jackman, ed.), pp. 183-201. Humana Press, Totowa, NJ, 1999. 49. L. G. Mendelsohn, J. F. Worzalla, and J. M. Walling, in "Anticancer Development Guide: Antifolate Drugs in Cancer Therapy" (A. L. Jaekman, ed.), pp. 261-280. Humana Press, Totowa, NJ, 1999. 50. R. G. Moran, Semin. Oncol. 26, 24-32 (1999). 51. I. D. Goldman, Biochim. Biophys. Acta 233, 624-633 (1971). 52. I. D. Goldman, Ann. N.Y. Acad. Sci. 186, 400-422 (1971). ,53. G. B. Henderson and E. M. Zevely, J. Biol. Chem. 259, 1526-1531 (1984). 54. E M. Sirotnak and D. E O'Leary, Cancer Res. 51, 1412-1417 (1991). 55. M. Kool, M. van der Linden, M. de Haas, G. L. Scheffer, J. M. L. de Vree, A. J. Smith, G. Jansen, G. Peters, N. Ponne, R. J. Scheper, R. P. J. Oude Elferink, E Baas, and P. Borst, Proc. Natl. Acad. Sci. U.S.A. 96, 6914-6919 (1999). 56. T. Hirohashi, H. Suzuki, and Y. Sugiyama, J. Biol. Chem. 274, 15181-15185 (1999). 57. J. H. Hooijberg, H. J. Broxterman, M. Kool, Y. G. Assaraf, G. Peters, P. Noordhuis, R. J. Scheper, P. Borst, H. M. Pinedo, and G. Jansen, CancerRes. 59, 2532-2535 (1999). 58. E M. Sirotnak, L. J. Goutas, D. M. Jacobsen, L. S. Mines, J. R. Barrueco, Y. Gaumont, and R. L. Kisliuk, Biochem. Pharvnacol. 36, 1659-1667 (1987). 59. G. B. Henderson and B. P. Strauss, CancerRes. 50, 1709-1714 (1990). 60. E. E. Sierra, K. E. Brigle, M. J. Spinella, and I. D. Goldman, Biochem. Pharmacol. 53, 223-231
(1997). 61. G. R. Westerhof, S. Rijnboutt, J. H. Schornagel, H. M. Pinedo, G. J. Peters, and G. Jansen, Cancer Res. 55, 3795-3802 (1995). 62. M. J. Spinella, K. E. Brigle, E. E. Sierra, and I. D. Goldman, J. Biol. Chem. 270, 7842-7849
(1995). 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74.
F. M. Sirotnak, D. M. Moccio, and C. H. Yang,]. Biol. Chem. 259, 13139-13144 (1984). C. H. Yang, E M. Sirotnak, and L. S. Mines, J. Biol. Chem. 263, 9703-9709 (1988). L. H. Matherly, C. A. Czajkowski, and S. M. Angeles, CaneerRes. 51, 3420-3426 (1991). G. B. Henderson and E. M. Zevely,]. Biol. Chem. 259, 4558-4562 (1984). J. D, Schuetz, L. H. Matherly, E. H. Westin, and I. D. Goldman,J. Biol. Chem. 263, 9840-9847 (1988). J. H. Freisheim, M. Ratnam, T. P. MeAlinden, K. M. R. Prasad, F. E. Williams, G. R. Westerhof, J. H. Schornagel, and G. Jansen, Adv. Enzyme Regul. 32, 17-31 (1992). C. H. Yang, J. Pain, and E M. Sirotnak, J. Biol. Chem. 267, 6628-6634 (1992). I: M. Underhill and W. E Flintoff, Mol. Cell. Biol. 9, 1754-1758 (1989). T. M. Underhill, F. M. R. Williams, R. C. Murray, and W. F. Flintoff, J. Cell. Mol. Gen. 18, 337-349 (1992). K. Hoffman and W Stoffel, Biol. Chem. Hoppe-Seyler 374, 166 (1993). S. C. Wong, R. McQuade, S. A. Proefke, A. Bhushan, and L. H. Matherly, Biochem. Pharmacol. 53, 199-206 (1997). S. C. Wong, L. Zhang, S. A. Proefke, and L. H. Matherly, Biochim. Biophys. Acta 1375, 6-12
(1998). 75. P. L. Fergeson and W. F. Flintoff, J. Biol. Chem. 274, 16269-16278 (1999).
HUMAN REDUCED FOLATE CARRIER
].61
76. K. E. Brigle, M. j. Spinella, E. E. Sierra, and I. D. Goldman,J. Biol. Chem. 270, 22974-22979
(1995). 77. R. Zhao, R. Seither, K. E. Brigle, I. G. Sharina, P. J. Wang, and I. D. Goldman, J. Biol. Chem. 272, 21207-21212 (1997). 78. C. K. Kumar, T. T. Nguyen, F. B. Gonzales, and H. M. Said, Am. J. Physiol. 274, C289-294 (1998). 79. j. H. Chiao, K. Roy, B. Tolner, C. H. Yang, and F. M. Sirotnak,J. Biol. Chem. 273,11165-11170 (1997). 80. H. M. Said, T. T. Nguyen, D. L, Dyer, K. If. Cowan, and S. A. Rubin, Biochim. Biophys. Acta 1281, 164-172 (1996). 81. M. Veyhl, J. Spangenberg, B. Pusehel, R. Poppe, C. Deke], G. Fritzseh, W. Haase, and H. Koepsell, J. Biol. Chem. 268, 25041-25053 (1993). 82. H. Koepsell and J. Spangenberg, J. Membr. Biol. 138, 1-11 (1994). 83. T. M. Underhill and W. F. Flintoff, Sore. Cell Mol. Genetics 15, 49-59 (1989). 84. J. H. Chain, C. H. Yang, K. Roy, J. Pain, and F. M. Sirotnak, J. Biol. Chem. 270, 29698-29704 (1995). 85. A. Bhushan, D. Wroblewski, Y. Xuan, T. R. Tritton, and H. P. Hacker, Biochem. Pharmacol. 51,477-482 (1996). 86. L.H. Matherly, S. M. Angeles, and C. A. Czajkowski, J. Biol. Chem. 267, 23253-23260 (1992). 87. G, Jansen, G. R. Westerhof, M. J. A. Jarmuszewski, I. Kathmann, G. Rijksen, and J. H. Schornagel, J. Biol. Chem. 265, 18272-18277 (1990). 88. K, E. Brigle, M. J. Spinella, E. E. Sierra, and I. D. Goldman, Bioehim. Biophys. Acta 1353, 191-198 (1997). 89. R. C. Murray, E M. R. Williams, and W. E Flintoff, J. Biol. Chem. 272, 19174-19179 (1996). 90. F. M. R. Williams and W. F. Flintoff, Sore. Cell. Mol. Genetics 24, 143-156 (1998). 91. S. Drori, H. Sprecher, G. Shemer, G. Jansen, I. D. Goldman, and Y. G. Assarat, Eur J. Biochem. 267, 690-701 (2000). 92. J. R. Whetstine and L. H. Matherly, J. Biol. Chem. 276, 6350-6358 (2001). 93. S, Philipsen and G. Suske, Nucleic Acids Res. 27, 2991-3000 (1999). 94. M. Karin, Z. G. Liu, and E. Zandi, Curt Opin. Cell Biol. 9, 240-246 (1997). 95. P, Angel and M. Karin, Biochim. Biophys. Acta 1072, 129-157 (1991). 96. A. J. Shaywitz and M. E. Greenberg, Annu. Rev. Biochem. 68, 821-861 (1999). 97. C. K. Kmnar, M. P. Moyer, P. K. Dudeja, and H. M. Said, J. Biol. Chem. 272, 6226-6231 (1997). 98. M. Kozak, J. Biol. Chem. 266, 19867-19870 (1991). 99. J, v. Thottasserty, G. P. Zambetti, K. Arimori, E. G. Sehuetz, and J. D. Sehuetz, Proc. Natl. Acad. Sci. U.S.A. 94, 11037-11042 (1997). 100. B. C. Ding, J. R. Whetstine, and L. H. Matherly, J. Biol. Chem. 276, 8713-8719 (2001). 101. K. Roy, B. Tolner, J. H. Chiao, and F. M. Sirotnak, J. Biol. Chem. 273, 2526-2531 (1998). 102. R. Zhao, Y. G. Assaraf, and I. D. Goldman, J. Biol. Chem. 273, 7873-7879 (1998). 103. R. Zhao, Y. G. Assaraf, and I. D. Goldman, J. Biol. Chem. 273, 19065-19071 (1998). 104. G. Jansen, R. Mauritz, S. Drori, H. Sprecher, I. Kathmann, M. Bunni, D. G. Priest, P. Noordhuis, J. H. Schornagel, H, M. Pinedo, G. J. Peters, and Y. G. Assaraf, J. Biol. Chem. 273, 30189-30198 (1998). 105. M. Haber, A. J. Gifford, L. H. Matherly, and M. D. Norris, Abstr Am. Assoc. Cancer Bes. 40, 675 (1999). 106. A. Tse, K. Brigle, S. M. Taylor, and R. G. Moran, J. Biol. Chem. 273, 25953-25960 (1992). 107. M. Gong, J. Yess, T. Connolly, S. P. Ivy, T. Ohnuma, K. H. Cowan, and J. A. Moscow, Blood 89, 2494-2499 (1997).
162
LARRY H. MATHERLY
108. H. Sadlish, R. C. Murray, E M. R. Williams, and W E Flintoff, Biochem. J. 346, 509-516
(2000). 109. F. M. Sirotnak, D. M. Moccio, L. E. Kelleher, and L. ]. Goutas, Cancer Res. 41, 4442-4452
(1981). 110. J. L. Biedler, A. M. Albrecht, D. J. Hutchinson, and B. A. Spengler, CancerRes. 32, 153-161 111. 112. 113. 114. 115. 116. 117. 118. 119. 120.
(1972). R. T. Schimke, CancerRes. 44, 1735-1742 (1984). W. 17. Flintoffand K. Essani, Biochemistry 19, 4321-4327 (1990). P.W. Melera, J. A. Lewis, J. L. Biedler, and C. Hession,J. Biol. Chem. 255, 7014-7028 (1980). G. Pizzorno, Y. M. Chang, J. J. McGuire, and J. R. Bertino, Cancer lies. 49, 5275-5280 (1989). D. E. McCloskey, J. J. McGuire, C. A. Russell, B. G. Rowan, ]. R. Bertino, G. Pizzorno, and E. Mini,/. Biol. Chem. 266, 6181-6187 (1991). A. J. Gifford, J. R. Whetstine, K. S. Bark, J. s. Kulkarni, J. w. Tanb, M. Haber, M. D. Norris, and L. H. Matherly, Abstr. Am. Assoc. Cancer Res. 41,758 (2000). B.C. Ding, T. L. Witt, B. Hukku, H. Heng, L. Zhang, and L. H. Matherly, Biochem. Pharmacol. 61,665~75 (2001). R. Zhao, E Gao, P. J. Wang, and I. D. Goldman, Mol. Pharmacol. 57, 317-2323 (2000). R. Zhao, E Gao, and I. D. Goldman, Biochem. Pharmacol. 58, 1615-1624 (1999). J. S. Kim and B. Shane, ]. Biol. Chem. 269, 9714-9720 (1994).
Gene Targeting via Triple-Helix Formation BRIAN R CASEY AND P E T E R M. GLAZER 1
Departnwnts of Therapeutic Radiology and Genetics" Yale University School of Medicine New Haven, Connecticut 06,520 I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII. XIII, XIV. XV.
General Problems in Oligonueleotide-Based Gene Therapy Strategies , . . Delivery of Triplex-Forming Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Minor-Groove Binders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Peptide Nucleic Aeids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic Chemistry of TFOs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Improvements in TFO Chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . TFOs as Molecular Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Chromatin Barrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . TFOs as Antigene Agents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . TFO-Mediated Upregulation of Gene Expression . . . . . . . . . . . . . . . . . . . . TFO-Mediated Mutagenesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . TFO-Mediated Recombination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . TFO-Direeted Sequence Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Repair Systems Implicated in TFO-Indueed DNA Alterations . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
164 164 166 166 167 170 172 174 177 178 180 181 185 185 188 189
A report on a recent workshop entitled "Gene-Targeted Drugs: Function and Delivery" conveys a justified optimism for the eventual feasibility and therapeutic benefit of gene-targeting strategies (1). Although multiple approaches are being explored, this chapter focuses primarily on the uses of triplex-forming oligonucleotides (TFOs). TFOs are molecules that bind in the major groove of duplex DNA and by so doing can produce triplex structures. They bind to the purine-rieh strand of the duplex through Hoogsteen or reverse Hoogsteen hydrogen bonding. They exist in two sequence motifs, either pyrimldine or purine. Improvements in delivery of these TFOs are reducing the quantities required for an effective intracellular concentration. New TFO chemistries are increasing the half-life of these oligos and expanding the range of sequences that can be targeted. Alone or conjugated to active molecules, TFOs have proven to be versatile agents both in vitro and in vivo. Foremost, TFOs have been employed in antigene strategies
ICorrespondenee should be addressed to Peter M. Glazer: telephone (203) 737-2788; fax: (203) 737-2630; E-maih [email protected]. Progress in Nucleic Acid Research and Molecular Biology, Vol. 67
Copyright O 2001 hy Academic Press.
163
All rights of reproduction in any form reserved.
0(179-6603/01 $35.00
164
BRIAN P. CASEYAND PETER M. GLAZER as an alternative to antisense technology. Conversely, they are also being investigated as possible upregulators of transcription. TFOs have also been shown to produce mutagenic events, even in the absence of tethered mutagens. TFOs can increase rates of recombination between homologous sequences in close proximity. Directed sequence changes leading to gene correction have been achieved through the use of TFOs. Because it is theorized that these modifications are due to the instigation of DNA repair mechanisms, an important area of TFO research is the study of triple-helix recognition and repair. © 2001 AcademicPress.
I. General Problems in Oligonucleotide-Based Gene Therapy Strategies Despite great hope for specificity of action, any molecular-based therapeutic strategy can have unintended effects. As human antisense trials are already a reality, it is not premature to study possible immune responses to the delivery vehicle, conjugated chemical agents, or to the nucleic acid itself. Even naked DNA, which might be thought to be the least antigenic of the three, can elicit negative reactions in animals. Bacterial DNA or DNA of unusual geometry has been shown to provoke antibody production (2). In addition a charged oligo may be sequestered, and so rendered inactive, by charged cellular proteins (2). Phosphorothioate molecules, for example, have documented protein-binding tendencies (3). The length of the oligo also poses a problem. Although an increase in oligo length assures stability of the triplex through increases in hydrogen bonding and base stacking, it does not necessarily assure greater specificity. For any given sequence of an oligo, some erroneous targeting always occurs owing to the binding of a subset of that sequence. Because control and active oligos are necessarily different in at least one important variable (sequence, base type, backbone chemistry, or sugar conformation or chemistry), it is difficult to design a control that would have the same potential for side effects (3). IfTFOs are to be used in conjunction with agents that can damage DNA, there is the additional concern of promiscuous actions by the appended moiety.
n. Delivery of Triplex-Forming Molecules As the therapeutic value of TFOs so heavily will depend upon the ability to deliver sufficient quantities to the nucleus of cells, this review cannot neglect a brief discussion of some of the currently available delivery choices. The attraction of a vectorless delivery system is the significant lessening of the possibility of an immune reaction. Passive uptake of oligonucleotides, albeit reasonably effective, usually occurs through endocytosis. Known inhibitors of
GENE TARGETING VIA TRIPLE-HELIX FORMATION
165
endocytosis can greatly reduce the uptake of oligos (4). Other modes of delivery are being sought because the short half-life in cells of phosphodiester molecules can, to a large extent, be attributed to endosomal destruction. Cationic lipid mixtures have been shown to increase the rates of association of phosphorothioate molecules to cells 25-fold and, more specifically, to increase the amount that can successfully reach the nucleus (5). Streptolysin O produces reversible pores in the cell membrane. Using a concentration of 150 units/ml, reversible permeabilization of 75% of a population of leukemia cells has been detected, and uptake of chimeric fluorescent methylphosphonodiester/phosphodiester oligonucleotide molecules has been increased at least 1000-fold over passive uptake levels (6). Effectively removing the membrane barrier, either through the use of streptolysin O or through electroporation methods, may be unsuitable for therapeutic purposes. The use of carrier molecules has also been attempted. The highly branched, and hence highly valenced, polyethylenimine molecule is easily protonated and serves as a good cationic cartier through the negatively charged outer cell membrane. The protonation appears to serve a critical buffering function as well, which allows survival of oligos in endosomes (7). Another way the endosoreal pathway can be bypassed involves the use of fusion peptides that have hydrophobic domains for crossing the membrane as well as hydrophilic nuclear localization signals for addressing the oligo. Such a construct has succeeded in transporting 90% of a plasmid into the cytoplasm in mammalian cells in an hour without cytotoxic effects. Under temperature conditions not favorable to the endosomal pathway it was still effective (8). Among the distinct advantages of using a viral delivery system are the ability to escape the endosomal system and the ability to target specific cell types at near-perfect efficiency. Immunohistochemistry of nuclear lysates from cells infected by oligo-linked adenovirus constructs revealed high levels of nuclear localization. The concentration of internalized oligo was predicted to have reached 20/zM (9). The issue of half-life remains. Despite the protection from endosomes, the half-life of even modified oligos is still typically only about half a day. A pharmacological study of the fate of phosphorothioate molecules injected into mice found, on the positive side, widescale distribution of the oligo among all tissues but the brain, and on the negative side, almost 50% degradation and excretion within 2 days. Different tissues revealed different propensities to degrade the oligo (10). A study done in mice demonstrated an 8-h stability of a purine-rich phosphodiester TFO modified with only a 3' propylamine (11). The enhancement of nuclease resistance should secure a greater chance for success of molecular therapies if repeated or prolonged TFO action is beneficial; however, unforeseen detrimental effects of prolonged TFO existence may arise.
166
BRIANP. CASEYAND PETERM. GLAZER
III. Minor-Groove Binders As an alternative to sequence-specific targeting via the major groove, small minor-groove binding molecules have also been exploited to try to circumvent the concentration, sequence, and ion limitations of TFOs. In 1992 it was reported that a 5-bp mixed sequence (5'-TGTCA-3') was successfully targeted by either of two peptide derivatives of netropsin and distamycin a. Binding of dimers of these peptides to the minor groove is thought to occur through hydrogen bonds formed between nitrogen atoms on the peptides and the N3 of purine, 02 of pyrimidines, and N2 of guanine (12). Many potential antitumor drugs are mutagenic agents that work by binding in the minor groove and producing lesions and/or by inhibiting repair processes (13). If this mutagenic property" could be better directed toward specific genome locations, treatments could perhaps be more effective. Binding rules to duplex pairs based upon the specific apposition of pyrrole and imidazole rings have been derived. In accordance with these rules, a six-nncleotide sequence was successfully targeted at subnanomolar concentrations, and with high specificity, a one-nucleotide change reduced binding by at least an order of magnitude (14). The practical relevance of such small molecules for gene manipulation efforts became clear with the demonstrated relief of repression of a cytomegalovirus gene. In vitro pyrrole-imidazole compounds were able to reduce the binding of a repressor and thus increase transcription (15).
IV. Peptide Nucleic Acids Peptide nucleic acids (PNAs) are another example of a sequence-specific DNA-binding molecule. These hybrid molecules contain the bases seen in nucleotides, and thus have the same capacity for engaging in Watson-Crick or Hoogsteen hydrogen bonding, but are supported on a peptide backbone. Their chemistry and construction can be found in a 1996 review (16). They are useful for gene-targeting efforts because of their nuclease resistance, neutral charge, and slow rates of dissociation. As with TFOs, various chemical improvements have been made to broaden the ability of PNAs to bind targets under less than optimal conditions (17, 18). One species of PNAs, bis-PNAs, has two domains that form clamps by binding the same patch of DNA. PNA clamps are characterized by strand displacement and by extremely slow off rates. One PNA domain strand invades, often with the help of attached positively charged residues such as lysine, and binds via the standard Watson-Crick hydrogen bonds to a complementary purine sequence. The other linked domain binds via Hoogsteen bonds in the same manner as a pyrimidine-motifTFO. Which domain binds first has been disputed.
GENE TARGETINGVIATRIPLE-HELIXFORMATION
167
One study looking at kinetic constants concluded that the Watson-Crick domain binds first and furnishes the specificity of binding. The overall binding process seemed to be dictated by the kinetics of duplex formation (19). Another study taking advantage of the need for protonation of the third-strand cytosine came to the opposite conclusion. The binding rates and specificity were pH-dependent and thus due to the Hoogsteen binding domain. If the cytosines were substituted with pseudoisocytosine, the specificity rates went up at neutral pH (20). In any case, active transcription through the PNA target region greatly increases the rate of clamp formation (21). PNAs have been tested for their ability to interfere with normal DNA processes and for their mutagenic character. (For a review of therapeutic applications of PNAs, see Ref. 22. ) A PNA concentration as low as 1/zM has been shown to block in vitro transcription elongation (23). As a testament to the stability of the PNA clamp, the activity level of a viral helicase, which is known to disrupt TFO-formed triplexes, has been reduced to half in in vitro assays using only nanomolar amounts of PNAs (24). Ten-mer PNA clamps introduced through streptolysin O exposure into mouse fibroblast cells induced mutations 10 times above background levels when targeted against incorporated supF genes. Most mutations consisted of one base-pair insertion or deletions around the clamp site, suggestive of strand slippage events (25).
V. Basic Chemistry of TFOs While attempting to solve the structure of DNA, James Watson pondered the idea that life may depend on a triad. The initial data before him, it seemed, equally fit a three-stranded model (26). Just a few years after the publication of Watson and Criek's "suggested" double helix, a new type of DNA assembly was discovered. Work with strands of polyuridine and polyadenine unexpectedly led to complexes possessing a 2 : 1 ratio which were further stabilized in the presence of divalent cations (27). Although it was then established that such triplex structures can be artificially created, the degree to which they actually form in vivo remains uncertain. In vitro studies indicate that triple-helix formation is slower by several orders of magnitude than that of duplexes (28). Nevertheless, intramoleeular triplexes, which do not involve foreign strands, have since been shown to form spontaneously in vitro with sequences taken from the promoter regions of natural genes. As a result, potential roles for triple helices in gene regulation have been posited (29). Further indirect evidence that such structures may form under natural conditions, and hence are readily recognized, comes from an in vitro study indicating that bacterial transposons may use triplex structures as guides in choosing integration sites (30), and from the demonstration that triplex-binding proteins exist in mammalian nuclear cell extracts (31).
168
BRIAN P. CASEYAND PETER M. GLAZER
The discovery of triple-strand binding naturally led to an investigation of sequence requirements. Using columns with fixed single strands, the binding rules were worked out by observing which duplexes could be retained on the columns under different pH and temperature conditions (32). It was discovered that a potential target duplex must contain a strand with a run ofpurines, for only a purine-rich strand can provide the necessary hydrogen bonding ability to form triple helices (33). From these experiments, it is known that degeneracy exists in the binding rules: A can bind to A, U, and T, while G can bind G or C (32). Three TFO sequence motifs are possible: one in which the third strand consists A
,~x-forming nucleotide
FIG. 1. Basic features of an oligonucleotide-based triple helix interaction. (A). Ribbon model demonstrating the relative position of a TFO (black strand) in the major groove of DNA (gray and white strands). Solid black lines: Traditional Watson--Crick base pairs. Hoogsteen hydrogen bonds are omitted. (B). The pyrimidine triple helix binding motif. Above: Binding of a TFO in a parallel orientation to the polypurine strand of DNA. Below: Base triplets involved via Hoogsteen interactions. (C). The purine binding motif. Above: Binding of a TFO to the polypurine strand of DNA in an antiparallel orientation. Below: Three canonical base triplets in this motif with reverse Hoogsteen bonds shown. Reprinted with permission from P. Chan and P. M. Glazer, Triplex DNA: Fundamentals, advances, and potential applications for gene therapy, J. Mol. Med. 75(4), 267-282 (1997).
GENE
TARGETING
B
VIA TRIPLE-HELIX
3'( 5'
169
FORMATION
5'
~ i i i i i i i i i i i i i i i i i i i 1 i ~
3' ~
3'
5'
Triplex-forming oligonucleotide Polypurine strand of DNA Polypyrirnidine strand of DNA
/R R
CH=
~""-- N '
h
H
H
N.~ ~
C
R
R
A.A:T
G-G:C
51 5~
i
3' ~
i'tl
z i i iii
~ r
i a i j'l'l
T-A:T
i
i j i~
3'
5'
Triplex-forming oligonucleotide Polypurine strand of D N A Polypyrimidine strand of D N A
.CH3 "
R ...N/~
Protonation
I '1
o
at N3
required
.,. n
CH~
H I-I N
N"
~...
1T!
. .N..._ ~,-N.R -~ o
~\~n
R
C+.G:C F I e . 1.
2.
N-,',"H"N"~O N'R
R
"H
3I
Iif
.....
T.A:T (Continued)
of the purines G and A, another in which the pyrimidines C and T compose the oligo, and finally a mixed purine/pyrimidine TFO with G and T as the constituent nucleotides. All varieties bind in the major groove without strand invasion. Evidence for duplex unwinding at the site of DNA triplex formation is lacking (34). However, duplex invasion with a phosphodiester molecule is possible if an invading oligo is tethered to a triplex-forming molecule (35). Experiments carried out using oligos designed to deliver cleaving agents to a duplex/triplex junction yielded patterns indicating that purine and mixed purine/pyrimidine third strands bind antiparallel (in reference to their 3' to 5' phosphate orientations) to the purine target via reverse Hoogsteen hydrogen bonds, whereas pyrimidine strands bind in a parallel manner through Hoogsteen bonds (Fig. 1). Molecular modeling has revealed that in the purine motif, because of the need to keep the
170
BRIANP. CASEYAND PETERM. GLAZER
bases in the anti conformation, the ribophosphate backbone is placed approximately equidistant from each duplex strand. The same is not true in the case of a pyrimidine motif (36). Kinetic and thermodynamic studies have uncovered properties essential for triplex formation. Activation energy calculations have shown that three to five triplets must be bound before subsequent rapid formation of a stable triplex is possible. Centrally located mismatches apparently cause no difference in on rates but do increase the offrates (28). Studies substituting 5-methyl cytosine for C in pyrimidine TFOs have shown the importance of stacking interactions for the stabilization of triple helices (0.1-0.4 kcal/mol over a pH range of 5.8-7.6) (37). (For a discussion of the kinetic properties of pyrimidine triplex formation, see Ref. 38.) Each TFO motif is plagued by a major limitation. The purine motif and the mixed motif with a high G content require that the concentration of monovalcnt cations, particularly K+, be extremely low, below physiological amounts. This is because such ions can stabilize unwanted TFO secondary structure, like G quartets, thereby lessening the supply ofoligos poised for triplex formation. Gel shift assays performed with different monovalent ions revealed similar patterns of inhibition of triplex formation and concomitant increase in self-aggregating products, depending on ion concentration and ion radius (39). The closer in character ions are to K+, the more inhibitory they turn out to be. Another in vitro study involving the purine motif produced inhibition of triplex formation by potassium without effects on dissociation rates. Inhibition in this case was apparently not due to the shuttling ofofigos into G-quartet formation (40). Pyrimidine TFOs are limited by the requirement of cytosine protonation needed for hydrogen bonding to the duplex. The implication of this proton requirement is that unmodified pyrimidine TFOs must bind under nonphysiological acidic pH. All motifs require divalent metal cations to neutralize charge repulsion between backbones to allow not only triplex formation but also triplex maintenance. These divalent cations must not be Ni2+ or Cd 2+, which bind to the bases and so are detrimental to hydrogen binding (41). Polyamine compounds such as spermine or spermidine can also serve the function of masking backbone charge (42, 43). At a 500-/zM concentration, spermine doubles the association rate at low ion concentration and at neutral pH (42). Conducive to intracellular formation, polyamines can be found in the millimolar range in eukaryotic cells.
VI. Improvements in TFO Chemistry In order to optimize triplex formation, a number of innovations in TFO chemistry have been developed and tested. Intercalators such as coral)me have been reported to improve triplex stability and function (44). Intercalators have been conjugated to either the 5' or 3' ends of oligos for the purpose of anchoring
GENE TARGETINGVIATRIPLE-HELIXFORMATION
171
the TFO once the triplex has been formed (45). Intercalators placed in the middle of a TFO have overcome destabilizing mismatches (46). The evolution ofintercalator design has led to the advent of a five aromatic ring system, derived from benzo[f]quino[3,4,-b]quinoxaline, which has a preference for intercalating in triple helices (47). A major limitation of triplex technology that is steadily being overcome is the need for a sufficiently long stretch of purines in the target duplex. The target sequence range has been considerably broadened through the use of base analogs or through the trick of strand crossing. 4-(3-Benzamidophenyl)imidazole 'allows the binding, through van der Waals interactions, with inversions otherwise destabilizing for the pyrimidine TFOs (48). In addition, interruptions in a purine motif can be skipped by using azole nucleosides which avoid steric hindrance and yet provide base-stacking stability (49), (See also Ref. 50; and for a discussion of targeting inversions, see Ref..51 .) Short purine runs neighboring short pyrimidine runs can be targeted if complex oligos are designed. Purine stretches in both duplex strands can become bound by short oligo components held together by flexible linkers. In certain sequence contexts 5'-pyrimidine-purine-3' constructs are twice as effective at triplex formation as 5'-purine-pyrimidine-3' constructs (52). Assisted targeting has also been tried. Mixed sequences have been targeted using an artificial third strand/rec A filament. The use of this filament is not expected to result in a formation like a TFO mediated triplex (44). The pyrimidine dependence on low pH has been circumvented primarily through the use of analogs. If the cytosine on the third strand is methylated at the 5 position, triplex formation can occur at physiological pH (53, 54). Because a run of five methylcytidines can lead to lower triplex stability, however, other replacement analogs---both purine and pyrimidine--have been developed, ineluding the purine analog 8-oxo-2'-deoxyadenosine (55). The tendency of purine strands to form undesirable secondary structures has been combated by using analogs and by engineering intended secondary structures into the TFOs. In one study the inhibitory effects of high potassium concentrations on purine-motif triplex formation was eliminated by the use of 7-deazaxanthine without a compromise in Kd values. The degree of K+ resistance using this compound may depend on sequence context (56). Resistance to potassium up to a concentration of 200 nM is possible through the use of 6-thioguanine. Sulfur's larger radius and reduced electronegativity ostensibly produce the steric hindrance and lowered affinity for potassium necessary for prevention of the rise of G quartets (57). Ironically, creation of partially self-complementaryoligos may provide protection against G-quartet formation. Such secondary structures would still leave a stretch of unpaired nucleotides which could initiate triplex formation (58). Many types of chemical substitutions, including changes in backbone linkages, backbone components, and base types, as well as the attachment of positively charged molecules, have been shown to lower repulsions between the
172
BRIAN P. CASEY AND PETER M. GLAZER
backbones and increase nuclease resistance. Phosphorothioates, which have a nonbridging oxygen substituted with a sulfur, have a much prolonged half-life. Phosphoramidate backbones, which contain an amino group joining the phosphorus atom to the 3 position, have been shown in vitro to bind better to duplex than phosphodiester molecules in the pyrimidine but not the purine motif (59). Morpholino oligonucleotides, which are neutral in charge, contain phosphorodiamidate linkages, and have morpholine rings instead of deoxyribose, have been shown to be superior in binding compared to phosphoroamidate oligos in the absence of Mg 2+ (60). The analog 5-(1-propynyl)-2'-deoxyuridine has also been shown to decrease the need for divalent ions, possibly by virtue of increased stability through improved stacking interactions (61). Levels of Mg2+ as well as of other cationic molecules have been rendered irrelevant by the conjugation of spermine onto polypyrimidine oligos (62).
VII. TFOs as Molecular Tools Because the chemistry of TFOs allows the attachment of various non-nucleic acid molecules, TFOs have been used as targeting vehicles. Because TFOs confer sequence specificity, modifying and cleaving enzymes can have their sites of action restricted. Such sequentially acting oligo/enzyme compounds have demonstrated both the sequence specificity of TFOs and the soundness of using such linked macromolecules to focus protein action. The ability to produce predicted cleaving events offered hope for the development of a unique mapping tool and for a way of expediting the isolation of desired regions. Also, TFOs may be good tools for the study of basic biological processes. The generation of a strand break by other means has already resulted in significant information about recombination (63). Through a number of means, desired cleavage events, have been achieved using TFOs. EDTA-Fe tethered to a TFO produces strand-breaking free radicals (64). Pyrimidines tethered to phenanthroline in the presence of copper ions and a reducing agent likewise has produced specific double-stranded breaks (65). Double-stranded breaks can also be caused by isotopic radiation, as has been demonstrated by the appending of a 5-125I-dCMP onto an oligo, whose binding to a plasmid followed by transfection into cells led to a 10,000-fold induction of mutagenesis (66). Also, the duplex/triplex junctions are known to be excellent substrates for photoendonucleases, at least in vitro (67). Cleavage of a single-stranded DNA molecule via a fusion protein containing the active site for a nuclease is also possible (68). TFO coverage of target regions can block cutting by restriction enzymes. Indeed, the restriction enzyme protection assay is an accepted test for triplex formation. TFOs can block methylation sites, thereby allowing restriction enzyme-mediated isolation of designated sections of genomic DNA (69).
173
GENE TARGETING VIA TRIPLE-HELIX FORMATION
The specificity of action ofa TFO cleaving tool was established in a paper by Posvic et al. in 1992. They achieved nonenzymatic cleavage by using two TFOs with appended N-bromoacetyl electrophile moieties that flank a target region. These TFOs were constructed so as to alkylate guanines two base pairs from the 5' end of each TFO. After depurination with piperidine, a double-stranded break that left ligatable overhangs occurred in a single site on a 4-kb plasmid or on a 340-kb yeast chromosome. The efficiency was calculated to be around 85% (70), The ability of TFOs to direct covalent modification in vitro has been repeatedly demonstrated. Alkylation of adenines at the N3 position in the minor groove has been achieved via a tethered 5-methylcyclopropapyrroloindole residue (71). To increase success rates, intercalating agents have been used in linkers to help thread minor-groove acting agents to their site of action (72). Alkylation of the N7 position of guanine in the major groove has been accomplished by the attachment of a chlorambucil to purine TFOs. Under conditions not favorable to triplex formation, no such modification occurred (44). The intercalator psoralen has also been guided by TFOs to preferred sites. Psoralen is a planar tricyclic photoactivatable D NA intercalator that can crosslink DNA by forming a covalent bridge between two thymines on opposite DNA duplex strands (73). Psoralen can be linked to the 5' or 3' ends of oligos via carbon linkers (Fig. 2). IfTFO binding sites are located right next to 5'-TpA-3'
5'i AGGAAGGGGG 3'GCTG!~GCTTCC~,~GCTTAGG/~GGGGGTGGTGGTS' 3'
CGACTTCGAAGGTTCGAATCCTTCCCCCACCACCA ~' ~50~
160
170
18~ ~
~'
f
E cos
supF
cX
cos
FIG. 2. Strategy for triple-helix formation with psoAG10. The sequence of the psoralen-linked oligopurine, psoAG1O, is shown positioned opposite the target sequence binding site for triplex formation in the supF gene, an E. coli amber suppressor tRNA. The psoralen moiety tethered at the 5t terminus was designed to react with the thyrnidines at bps 166-167. Reprinted with permission from 17.P. Gasparro et al., Site specific targeting ofpsoralen photoadducts with a triple helix forming oligonucleotide, characterization of psoralen monoadduct and crosslink formation, Nucleic Acids" Res. 22(14), 2845-2852 (1994).
174
BRIAN P. CASEYAND PETER M. GLAZER
dinucleotide sequences, psoralen can be incorporated at that site at a high frequency by a psoralen-TFO construct upon UVA irradiation. Substantial improvements in the targeting of mutations have been reported using psoralen TFOs (74). In vitro studies have yielded up to 90% adduct formation and, more specifically, about 80% crosslink formation using TFOs conjugated with psoralen (75). The degree of freedom of action of the psoralen is determined by the length of the linker (76). For the purpose of studying the effect of a particular psoralen adduct in the absence of the delivering oligo, disulfide linkers that can be reduced have been designed so that the oligo is released from the site (77). One of the first papers to demonstrate such directed action and to posit practical ramifications was published in 1991 (78). Targeted intercalation next to a reporter gene via this method was subsequently shown. Gel shift assays and HPLC analysis confirmed the preference for monoadduct formation on the purine strand in a 4:5'-furanside monoadduct versus a 3,4-pyroneside monoadduct formation (79).
VIII. The Chromatin Barrier The ease with which a TFO can gain access to a chromatinized target sequence is a matter open to debate. On this issue the in vitro and in vivo evidence seem to be contradictory. If22-mer triplex complexes, but not 10-mer, are preformed on linearized DNA fragments highly receptive to both TFOs and nucleosomes, footprinting assays reveal an inability to reconstitute nucleosomes. Conversely, if the nucleosomes are preformed and the DNA then exposed to 22-mer TFOs, no displacement of the histone assembly is seen unless performed under destabilizing high salt conditions (80). From hydroxyl-radical cleavage patterns derived in another study, it was discovered that both parallel and antiparallel TFO binding tended to be restricted to parts of the target DNA that were in weakest contact with histone proteins (81). These results notwithstanding, mutagenesis and crosslinking data gathered using psoralen-conjugated TFOs point to the accessibility of chromatinized targets. Polypurine tracts of HIV-1 provirus integrated in mammalian cells permeabilized by digitonin and incubated with special binding buffer have been crosslinked by psoralen-TFOs. Comparison between different cell lines with different sites of viral incorporation has revealed the importance of target location for the frequency of success (82). A native gene, an allele of the MHCII locus, in a genomic prep was covalently modified using a chlorambucil TFO with an 80% efficiency at 0.5 #M (44). In 1998, the alkylation of an endogenous target was reported. A 12-mer purine oligo linked to a nitrogen mustard entered streptolysin O-treated HT-29 adenocarcinoma cells and successfully modified a chemokine receptor target, the CCR5 H W receptor gene, as
GENE TARGETINGVIATRIPLE-HELIXFORMATION
175
determined by ligation-mediated PCR (83). A subsequent report has cautioned that ligation-mediated PCR has been found to produce artifacts if precautions are not taken (84). Other work using mutagenesis as an assay also demonstrated TFO targeting of chromosomal sites. Using a mouse fibroblast cell line containing multiple copies of a supF reporter gene optimized for triplex formation by the insertion of a polypurine sequence, Vasquez et al. were able to detect localized mutagenesis. Targeted mutagenesis by a TFO molecule was detected by the isolation of genomic DNA, packaging the supF containing lambda phage vector, and infecting bacteria. The lawn of bacteria on IPTG and X Gal containing plates revealed either blue or white plaques depending upon the ability of the packaged vector DNA to suppress an amber mutation in the lacZ gene of the bacteria and so to produce a colored metabolite. In a forward assay the supF gene would be functional until TFO-induced mutagenesis prevented the proper tRNA folding, ultimately causing the production of white plaques (Fig. 3). Using this system and relying upon passive uptake of oligos at a 2 #M concentration led to a 6- to 10-fold increase in mutagenesis. Sequence analysis revealed mostly insertions and deletions around the triplex binding site (85). The accessibility issue has been addressed in other work as well. A collagenase gene in human fibroblast cells was crosslinked in situ at two sites by psoralen TFOs that were lipofected into the cells at a concentration of 250 nM. Confirmation was by single-strand ligation PCR. This method involves rescuing the genomie DNA, cleaving around the TFO binding site, extending a primer up to the point of blockage, ligating a known sequence to the end of the PCR product, then PCRing with primers complementary to regions just inside the first primer and to the ligated end, and checking for abbreviated products. Interestingly, the addition of the transcription inducer phorbol ester failed to improve targeting frequency as judged by an absence of an increased frequency of abbreviated primer extension products (86). The ultimate test of targeting ability was carried out in live mice with the use of unpsoralenated 3'-end modified TFOs. A 5-fold induction of mutagenesis was seen in the supFG1 gene in these transgenic mice, which were given intraperitoneal injections of 1 mg of oligo per day for 5 consecutive days. At the time of sacrifice, all major tissues were examined and all except brain were found to contain approximately similar mutation rates. Controls included scrambled oligos, an untargeted control reporter gene, and the supposedly inaccessible brain tissue. In all cases, as expected, mutation rates were at background levels (87). How can the in vitro and in vivo results be reconciled? Both types of studies involved similar kinds of TFOs, with similar lengths and binding affinity. Like the nucleosome reeonstitution studies, the mouse data contained nonpsoralenated TFOs. It appears that the in vitro studies must not be adequate reflections of the
176
BRIAN R CASEY AND PETER M. GLAZER
o O~~~o_CHz.O/~..O-I~-O.'
5' AGGAAGC43GGGGGTGGTGGGGGAGGGGGAG
......
3'
o
Mouse cells with multiple
psoralen-oligo transfection
-.~-i"::.~..:~....
o
Triplex formation
in vivo
ctor DNA
o
~
s , ~
~
[~
supFG1
cos
o
3,
--
cl
cos
],;.;, r /
High molecular
Lambda in vitro packaging extracts
Plate and screen
on
bacterial lawn
i
i
Isolate and analyze mutant phage
Sequence supFGl gene
FIG. 3. Experimental system for detecting chromosomal mutations targeted by TFOs in mouse cells. TFOs targeted to supF reporter genes incorporated in mouse fibroblast cells induce mutagenesis, detectable by recovery of the lambda phage shuttle vector, and infection oflacZ amber indicator bacteria. Reprinted with permission from K. M. Vasquez et al., Chromosomal mutations induced by triplex-forming oligonucleotides in mammalian cells. Nucleic Acids Res. 27(4), 1176-1181 (1999).
GENE TARGETINGVIATRIPLE-HELIXFORMATION
177
in vivo situation, which is more dynamic and where phasing is variable. Might the presence of H 1 or other cell components absent in the in vitro studies make an unexpected difference? Is the confinement of the in vitro studies to short stretches of DNA too artificial? As for the absence of a difference in targeting efficiency in tile presence of a transcriptional activator, perhaps the particular region studied under the given circumstance was for the applied concentration of oligo maximally accessible.
IX. TFOs as Antigene Agents Unlike antisense strategies, which rely upon the successful elimination of numerous messenger molecules, TFOs bear the promise of shutting down the very source of those messages. In the late 1980s and early 1990s, a series of in vitro studies was performed that examined the ability of TFOs to block transcription initiation or elongation. As determined by footprinting assays, TFOs without linked binding agents occluded polymerase binding to a bacteriophage artificial operator sequence (an inserted triplex-amenable polypurine sequence) (88). Another study showed interference with transcription to be dependent on the position and number of TFO binding sites (89). Hope for increasing the number of possible endogenous targets comes partly from the finding that target regions for the purpose of inhibition may also be located away from the proximal promoter region. A 13-bp sequence found in the coding region of the bla gene of the Tn3 transposon has been bound and functional mRNA production prevented. Transcription could be resumed if the TFO was melted off of the template (90). Longer TFOs, and hence stronger binders, have been found to be better inhibitors of elongation. Enhancers of transcription have also been occluded. The transcription factor Spl has been prevented from binding its normal site by prior triplex formation (91). The demonstration of TFO-mediated RNA level reduction in intact cells was soon forthcoming. Progesterone response elements residing in the tyrosine aminotransferase gene were targeted by a mixed purine/pyrimidine TFO which was linked to cholesterol. Monkey kidney CV-1 cells were transfected with plasraids bearing the target region upstream of the CAT reporter gene. Following incubation with these TFOs at a concentration of 20/zM and then with progesterone, CAT activity was reduced 50% (92). The polypurine tract of HIV-1 placed in a plasmid in the 5' untranslated region ofa luciferase reporter gene has been targeted by phosphoramidate oligos. Transcription elongation was blocked if the target was episomal or integrated in the genome of HeLa cells. Inhibition of the production of functional luciferase was seen at the single cell level (93). Overall protein reduction was around 50%. Controls for this work included
178
BRIANE CASEYAND PETER M. GLAZER
scrambled oligos and mismatched target sequences that expressed comparable basal levels ofluciferase mRNA, and hence should have been equally accessible. As further indication that the effect was triplex-mediated, the effectiveness was improved by the inclusion of triplex-stabilizing intercalators (93). The attainment of a desired phenotypic effect was also observed in experiments targeting the C-myc oncogene. A 10/zM concentration of a 22-mer G/T oligo was able to reduce the RNA and protein levels within 24 h to 50%. Flow cytometry indicated a --~20% increase in the amount of transfected cells trapped in S phase. At such a dose a 5- to 6-fold increase in leukemia cell mortality was seen after 4 days (94). In another example, tumor progression in mice was slowed by the targeting of the promoter of the IGF-1 receptor (95).
X. TFO-Mediated Upregulation of Gene Expression The ability to modulate a chosen gene's expression would be of benefit both to basic research and to medicine. For research purposes, it would be useful to ectopically turn on genes to discover phenotypes that may result from subtle overexpression. For instance, developmental biology questions dealing with the finetuning of embryonic stage genes could then be addressed. Current methods that involve transfection with additional copies of the gene, even if under the control of inducible promoters, are easy but relatively crude. Copies that randomly incorporate themselves into chromosomes may be silenced or have negative effects on neighboring genes. For therapeutic purposes, it would be ideal to be able to stimulate the production ofgene products whose absence leads to illness. TFOs could potentially serve as weapons in the arsenal against cancer if they could switch on genes that promote apoptosis. Alternative strategies for increasing gene expression often lead to the upregulation of an entire class or classes of genes rather than single, specific genes. For example, some natural or artificially synthesized chemicals such as hormones can directly or indirectly upregulate myriad genes responsible for proliferation. Extreme treatments of cells, such as heat shocking or serum starvation triggers, heighten transcription of many survival genes. The means by which triplex-forming molecules could turn on gene expression include nucleosome repositioning, mutation of repressor binding sites, formation of artificial transcription bubbles, recruitment of transcription factors, and physically dragging transcription factor-activating domains to promoter regions. The basal transcriptional apparatus is known to compete with nucleosomes. If relieved of histones, a template is more accessible and more likely to be traversed by a polymerase. The binding of a TFO to a chromatinized target appears to be in competition with histones. Once bound, however, a TFO might advantageously change the phasing and so alleviate nucleosome
GENE TARGETING VIA TRIPLE-HELIX FORMATION
179
mediated repression. The ability of preformed triplexes to restrict nucleosome placement has been shown in vitro (96). A therapeutic goal for researchers involved with sickle cell anemia and thalassemia disorders is the upregulation of F-globin, which is normally expressed only during early stages of development./%Globin is mutated in patients suffering from either disorder, but amelioration of symptoms can occur when y-globin expression is elevated. An 11-met purine oligo with a tethered psoralen has been designed to crosslink a site within the F-globin gene promoter. When the target is located in a plasmid subsequently transfeeted into human fibroblast cells, resultant mutations within the promoter cause a 4-fold increase in y-globin mRNA. In vitro binding assays showed that Oct 1 repressor binding to the promoter was reduced with such mutations (97).),-Globin mRNA levels were also increased by the employment of two pyrimidine-motifPNA clamps, which were shown to bind and unwind an upstream region. In vitro transcription assays revealed the start points to be at the TFO binding site as well as at the normal start site (98). The inspiration for this latter work had come from previous experiments with E. coli (99). A more ambitious strategy involving transcription factor recruitment has also been attempted. A two-domain TFO has been designed with a single-stranded portion fbr triplex formation and a double-stranded section containing binding sites for the transcription factors SRF and ELK. While footprinting and competition assays indicated triplex formation on a plasmid and sequential transcription factor binding, no evidence for biological activity has been presented (100). An alternative construct in which multiple copies of the activating domain of the herpes simplex virus protein 16 (HSV VP16) were conjugated to a polypurine TFO, a modest upregulation (3- to 4-fold) was seen of a reporter gene situated on a plasmid when tile triplex was extracellularly formed (101 ). Because of large error bars and the lack of a no-oligo control, it is not clear whether the activation represented, at least in part, relief of TFO-produced repression. In addition to pharmacokinetic and toxicological questions raised by oligonueleotide-based medicines in general, the aim of upregulating a gene's expression prompts a host of particular concerns. For how long can, and should, such an efi~ct last? Will it require significant basal level expression and/or replication around the target in order for TFOs to bind? Can such TFOs turn on only one gene in isolation from all others, and can it be made to do so in a particular cell type? To address these sorts of questions, gene chip experiments seem appropriate. Secondary effects need to be investigated as well. What will be the response of the cell? Does the TFO directly or indirectly cause the cell to produce countermeasures that may negate the primary stimulation? Is the targeted gene under autoregulation so that the production of more protein actually leads to repression? Most importantly, does the overexpression of the gene result in the expected phenotype?
].80
BRIANP. CASEYAND PETER M. GLAZER
Xh TFO-MediatedMutagenesis A number of mutagenesis studies from our lab and others have confirmed the ability to target a particular site for psoralen damage using TFOs as targeting reagents. A TFO linked to psoralen was found to produce damage in the supF reporter gene in a concentration- and UVA-dose--dependent manner. In experiments to target the supF gene contained in an SV40 vector, psoralen-TFO damage was targeted in vitro, followed by passage of the vector DNA through monkey COS cells, yielding mutation frequencies in the range of 5-10% (102). Analysis revealed mostly point mutations located at the predicted site ofpsoralen intercalation at the duplex/triplex junction. Assays requiring intracellular triplex formation have also detected targeted mutagenesis. In 1995, it was shown that treatment of COS cells already containing the SV40 vector with a concentration of 2 #M TFO yielded a mutation frequency of 2.1% (versus a background level of 0.03%). Most of the mutations were T-to-A transversions at the target ApT site, consistent with the mutagenic activity ofpsoralen adducts. A low frequency of deletions was also seen (103). In experiments to study TFO-mediated targeting of a chromosomal gene, mouse fibroblast cell lines containing multiple copies of the supFG1 gene were transfected with psoralen-conjugated purine TFOs designed to bind to a 30-bp polypurine sequence in the gene. The result was an induction of mutations by a factor of at least 6 compared with a scrambled control oligo. However, the types of mutations generated were different from the signature psoralen mutation pattern ofT : A to A : T transversions. These mutations were proposed to be due to strand slippage events occurring during triplex-induced repair synthesis. The poly G tract in the target gene would be prone to dislocation and misalignment during repair synthesis, which could lead to the deletions and insertions noted (85). A recent report on psoralen-TFO mutagenesis in yeast has looked at the impact of backbone chemistry on mutagenesis efficiency. In the reporter system used, the production of mutations at the psoralen target site reverts an ochre mutation and thus restores a ura3 selectable gene. If the triplexes were preformed using phosphodiester, psoralen-linked TFOs and the complex then transfected into ura3(-) cells, reversion rates were on the order of 15%. Even though phosphoramidate oligos have greater affinity for their targets and are better able to resist degrading enzymes (104), their use in this situation led to only about a 1% mutation frequency. The relative efficacy of each was reversed when the mutation was in an endogenous, chromosomal ura3 gene. In this case TFOs with phosphoroamidate backbones were 40 times more effective in generating revertants, but at a low absolute frequency (40 x 10-7). While mutation frequencies seem low, this is in part due to the fact that other psoralen-induced or triplex-induced mutations could not be detected by this assay. For a negative
GENE TARGETINGVIATRIPLE-HELIXFORMATION
181
control, an isogenic strain was used that had mismatches only in the TFO binding site but retained the same TpA psoralen target site within the ochre codon (105). In other efforts to target a chromosomal gene in mammalian cells, the hypoxanthine phosphoribosyl transferase gene (hprt) was knocked out by TFO molecules in CHO cells, allowing survival in selection media. A region around the intron 4-exon 5 boundary was chosen as a third-strand target site for a pyrimidine TFO, with the tethered psoralen directed at a TpA within the splice acceptor region. Induced mutations were detected at a frequency at least 5-fold above background, in the range of 10-4, when a 10-#M concentration of a 23-mer pyrimidine TFO containing 2'-O-methyl sugar modifications, methylcytosine instead of cytosine, internal intercalators, and a 5'-linked psoralen was electroporated into the cells. Only those oligos possessing either a pyrene or acridine conjugate necessary to improve binding affinity via intercalation were effective. PCR and sequencing analysis confirmed that the mutations were generated at the triplex target site (106).
XII. TFO-Mediated Recombination Easy, efficient gene replacement is a long-term goal ofgene therapy research. The introduction of foreign sequence information usually results in nonspecific incorporations through end joining or nonhomologous recombination events. Inducers of DNA damage, including UV radiation (107), carcinogenic compounds (108), and the crosslinking agent psoralen (109), are known to stimulate recombination events in eukaryotic cells. Despite some discerned sequence preferences, these agents inflict damage in a nearly random manner. Success in gene replacement might then be expected to depend on the development of constructs that are simultaneously sequence-specific and recombination-provoking. Preliminary results from experiments designed to detect increases in intermolecular recombination due to site-directed psoralen crosslinking yielded modest results. Two SV40 shuttle plasmids, one containing a defective copy of the supF reporter gene bound by a psoralen-TFO and the other containing a wild-type copy but missing the ability to replicate in cells, were transfected into human Jurkat cells. Onlyifhomologous recombination occurred could functional supF (and consequently lacZ) be made in indicator bacteria electroporated with the rescued plasmid. Recombination rates on the order of 0.05% were observed. As a point of comparison, a linearized donor plasmid was also cotransfected as one control condition. Under these circumstances a 2% recombination rate was obtained. The conclusion of the authors was that psoralen damage was either poorly corrected or at least not frequently corrected by a recombination pathway (110). Caveats include the possibility that the psoralen photoproduct production
182
BRIAN E CASEY AND PETER M. GLAZER
in these experiments may have been suboptimal, and that the experiment required the successful transfection of both constructs into each cell, which may have been a low frequency event. Our group has studied intramolecular recombination provoked by intermolecular triplex formation. For these studies an SV40 shuttle vector was created that contained two tandem mutated copies of the supFreporter gene flanking a triplex binding site but with a slight overlap in the 3' end of the first copy (Fig. 4). When psoralen triplexes were allowed to form on the plasmid prior to transfection into COS cells, recombination occurred 14% of the time relative to a background level of 0.02%. When intracellular triplex formation was attempted, recombination rates were about 0.58% while mutation rates (which could also be measured in the assay) were similar at 0.77%. Recombination always produced O II
/
CH2- O ~ O ' P '
O
O
3 5
"0"
riplex
formation
~ AGGAAGGGGGGGGTGGTGGGGGAGGGGGAG - -TTAGGAAGGGGGGGGTGGTGGGGGAGGGGGAG - -AATCCTTCCCCCCCCACCACCCCCTCCCCCTC-
b
pSR~7
,.I
. . . .
3' ( p s o - A G 3 0) - - 5' - 3'
........
pSu p F2
amp
SV40 origin
pBRa27
origin
SV40 T antigen
FIG. 4. Schematic representation of the pSupF2 vector designed to study recombination induced by triplex-targeted DNA damage. The SV40-based shuttle vector contains two tandem mutant supF genes. The upstream mutant supF gene contains a C-to-G point mutation at nucleotide position 163. The downstream mutant gene, supF2, contains a G-to-A point mutation at nucleotide position 115. At the 3' end of supF1 is an engineered polypurine sequence creating a high-affinity triplex binding site. A purine-rieh oligonucleotide was designed to form a triple helix in the antiparallel triplex motif at this site. The oligonucleotide was conjugated at its 5 ~end to 4'-hydroxymethyl-4,5',8trimethylpsoralen via the 4'-hydroxymethyl position through a two-carbon linker. Reprinted with permission from A. F. Faruqi et al., Recombination induced by triple helix-targeted DNA damage in mammalian cells, Mol. Cell. Biol. 16(12), 6820-6828 (1996).
GENE TARGETING VIA TRIPLE-HELIX FORMATION
183
a nonparental supF sequence. This was confirmed by observing redistribution of silent marker mutations engineered into the supFG1 genes. Furthermore, one copy was usually preferentially altered, consistent with a gene conversion mechanism. This copy tended to be the one with overlap with the triplex binding site and hence the psoralen damaged copy. Reversion mutations were ruled out as an explanation for the high level of correction because psoralen-TFOs crosslinked onto plasmids that contained only a single mutant copy produced minimal levels of regained function. Through controls it was discovered that the high level of recombination was made possible by the presence of the crosslinks and of the third strand (111). In a subsequent paper our lab has provided evidence of TFO-induced recombination in the absence ofpsoralen. The same system described above was used, and with the triplexes intracellularly formed in COS cells. A 5-fold stimulation of intramolecnlar recombination was witnessed without psoralen, as opposed to a "-~20-fold stimulation when psoralen accompanied the TFO. The experiments were also carried out in XPA-deficient cell lines, which lack the XPA damage recognition factor that can initiate nucleotide excision repair (NER). The results indicated that NER was essential for the triplex-stimulated recombination, indicating a requirement for NER-mediated strand breaks for the triplex-induced recombination (112). Our group has also obtained encouraging data that suggest not only that recombination can be induced by a TFO but also that a bifunctional oligonucleotide can provide the sequence information used by cellular machinery. A supFG1 reporter gene in an SV40 shuttle vector was targeted with an oligonucleotide designed to have two domains. A triple-helix forming single-stranded 30-nt portion (AG30) was tethered to a short (40 bp or nt) donor fragment, either double- or single-stranded, designed to provide the homologous sequence information, except for a single base pair difference at the position to be changed (Fig. 5). /Q values were minimally altered by the burden of an attached 40bp duplex via a flexible linker. The theory behind the design was that the formation of a triple helix by the TFO portion might provoke cellular DNA repair or recombination processes while the donor domain might be used by the machine~y as a template for information transfer. TFO-donor conjugates in both purine and pyrimidine motifs were tested for the ability to generate single or multiple base changes in the target gene, detectable by either a forward mutation assay or a reversion assay. When the AG30/donor DNA duplex was bound onto the plasmid before transfection into monkey COS 7 cells, reversion frequencies of ~1% were detected. When the targeting was mediated by intraeelhdar triplex formation, a successful correction frequency of 0.04%, or a 50-fold improvement over background, was obtained (113). The method of oligo transfection affects the levels of success, with a 10-ibld improvement to 0.5% with cationic lipids instead of eleetroporation as the transfection method (unpublished data). When the TFO domain and the donor domain
184
BRIAN P. CASEY AND PETER M. GLAZER
Homologous donor domain with wild type sequence
Linker region
3 ' 'I'CCCTCG-C-,I~&GCTTCC
J )e
5 ' AGC]GI~3C-G"I~I~'~I~GGg~L~ 9
T T
Triplex forming domain (AG30)
~~G~OO~;I~OGOOO~3, 3 'T C C C T C G - G - A A G C T T C C 5 'A G G G A G C - 4 ~ - T T C G A A G G
121
......... .........
160 N
-~ t44
AGGAAGC4~GGGGGTGGTGGGGGAGGGGGA~' TCCTTCCCCCCCCACCACCCCCTCCCCCT~'
167
/ /
-,.
\ "~
f~
\
/
supFGl 144
pBR327
pSupFG1/G144C
amp
......
/ \
"
~, ;, ; ,' ; I ; i ;'~ ~ i ; 6 a,,,;, ;, ; l l l r l l l l l l i l l l l l l l l l l l l l l l l l l ]
SV40 origin
pB.R327 ongm
1
SV40 T antigen
FIG. 5. Schematic diagram depicting the binding and positioning of single-stranded tethered donor TFOs to the s u p F gene. G at position 144 of the donor domain is intended to correct the inactivating G 144C mutation. Reprinted with permission from E Chan et al., J. Biol. Chem. 274(17), 11541-11548 (1999).
were used as separate, unlinked molecules, recombination was still detected, but at a reduced frequency, indicating a synergistic effect in the bifunctional construct. For TFO-mediated information exchange to become a therapeutic technique, success rates must be higher. This will necessitate an understanding of the mechanism of information transfer and the ability to manipulate the mechanism(s). The questions that must be addressed include the following: Which parts of the oligo constructs are recognized, by what pathways, and by which components of these pathways? It is clear from the initial work that NER is at least partly responsible for the sequence change, as cells deficient in XPA undergo fewer recombination events (113). Can a correlation be drawn between TFO binding affinity and the frequency of provoked information exchange? What is the optimal length and composition of the oligo? These and other questions remain to be answered. Intracellular delivery is another critical issue, as our group has found that substantially increased levels of TFO-provoked recombination are produced when delivery is improved by direct intranuclear microinjection. Cell lines have been used that bear tandem copies of defective herpes simplex virus thymidine kinase (TK) genes flanking, not overlapping, a 30-bp polypurine stretch in a TK(-) background. When treated with a 30-mer purine TFO designed to
GENE TARGETINGVIA TRIPLE-HELIXFORMATION
185
bind to the polypurine site, recombination resulting in a functional TK gene was detectable by selection in hypoxanthine-aminopterin-thymidine (HAT) medium. Microinjection of about 70,000 oligonucleotide molecules per cell increased the incidence of HAT-resistant colonies to 1%, or 2500 times the background level (versus the lesser 6-fold increase when the oligos were delivered by cationic lipids) (114). Southern analysis of the genomic DNA from the HAT-resistant colonies revealed that functional copies of the TK gene were produced by conservative gene conversion events (114).
XIII. TFO-Directed Sequence Change The ability to actually change a DNA sequence in a directed fashion would open up new gene therapy possibilities. Current transgenic techniques rely upon a low rate of successful homologous recombination on the order of one in a thousand specific events and necessitate positive and negative selection of cells in culture. TFOs, because they can induce predictable mutations, present new options. An early attempt used a TFO to target the mutagen psoralen to a specific base pair in the supF gene in an intact lambda genome. After crosslinking and passage through bacteria, mutations arose at levels at least 100 times that seen in an untargeted gene. This report found that slightly more than half of the mutations were T-to-A transversions at the site of psoralen intercalation (115). Similar findings were obtained when looking at intracellular TFO-directed mutagenesis on a shuttle vector in COS cells, with mutation frequencies of up to 2%, as noted above (103). Following upon this observation, Fresco and colleagues have focused on the mutation in the fl-globin gene associated with sickle cell anemia. Sickle cell anemia is caused by a single A : T to T : A transversion. Unfortunately, the region in the/3-globin gene near the sickle mutation lacks an uninterrupted polypurine sequence. To overcome this, the Fresco study used modified oligos that possessed not only base analogs (5-methylcytosine and 5-propynyluracil) but also components that allowed binding to short purine stretches on alternate strands, as well as the ability to bind one strand of the target by Watson-Crick pairing. Specific binding and photoproduct formation at the target base pair within a plasmid containing the target gene were clearly detected (116). However, the intracellular activity of this TFO remains to be tested.
XIV. Repair Systems Implicated in TFO-Induced DNA Alterations The phenomena of TFO-induced mutagenesis, recombination, or directed sequence change are most likely the consequence of at least two events: the
186
BRIANE CASEYAND PETER M. GLAZER
production of an unusual strueture on the DNA which is treated as damage, and the actual execution of the damage reeognition and repair proeess(es). The full details of both events are currently unknown, although hints as to the nature of the latter event are emerging and offer insight into the probable nature of the first. Definitive answers regarding both events would help in the design of the next generation of TFO agents. TFOs, even without tethered mutagens, have been shown to provoke mutagenesis (85) and recombination (112). In the absence of conjugated moieties, it is unelear what form the damage might take. Of the common types of DNA alterations known to be recognized by a cell (nicks, double-stranded breaks, base damage, pyrimidine dimers, mismatches, and bulky adduets), it is most likely that the last of these explains how the triplex is sensed. A bulky adduet is a single- or multinueleotide generic obstruction that can alter the structure of the double helix, which fits the description of a TFO-indueed triplex. Replication or transeription may be blocked because of such an aberrant structure, or the altered helix strueture itself could trigger repair. Several studies have demonstrated the ability of TFOs to hinder RNA polymerase progression, raising the possibility that triplex formation could be resolved in some instances by transeription-eoupled repair. As for the repair systems responsible for recognizing triplexes, evidence points toward involvement of the NER pathway. In an in vitro assay for repair in HeLa cell extracts, high-affinity triplex formation could induee general repair synthesis on a plasmid substrate (117). Comparisons have been made between the mutagenie effects of TFOs in normal human fibroblast cells and in those deficient in XPA, a NER recognition protein known to bind to junctions between unwound and duplex DNA. Both NER and its related pathway, transcriptioncoupled repair, have been implicated in the recognition of triplex damage caused by TFOs. Intracellular triplex formation in COS cells on a supFG1 reporter plasmid led to induced mutagenesis at and around the triplex site, mostly one- or multiple-point mutations and some large deletions. Cells deficient in either XPA or CSB, a protein essential for transcription-coupled repair (TCR), had reduced mutagenesis, indicating a role for both NER and TCR in the induced mutagenesis. Psoralen-TFOs not only can stimulate repair, but, depending on their length, can also inhibit or influence the pattern of repair and mutagenesis. Reduction in the level of excision products was seen in an in vitro repair assay using HeLa cell extracts when the psoralen oligo bound was 30 nucleotides long, but not when it was 10. In live COS cell experiments, the respective mutation frequencies caused by the 30-mer versus the 10-mer were 2.8% and 5.2%, respectively, when the triplex was preformed before transfeeting the plasmid. Analysis of the mutations revealed that the mutations caused by the psoralen 30-mer were shifted one base relative to the position of the transversions resulting from the
A. Repair (1) pso-AG 10 monoadduct ,I,
(3) pso-AG30 monoadduct $
,L excision
,I, [ excisionblocked by thirdstrand _
~ repairsynthesis
I
(4) pso-AG30 crosslink B_e~_a_ir_on__ugp_e_r_slrg._n_c!. $
,I, \
(2) pso-A G 10 crosslink 165
,I,~
excisionblocked by third strand
J,
167
~ excision
R_ep~_irQnJQw_er_st_r#_n_d 166
/I 167
1-
166
~'167
excision
--
~ repair synthesis
166
167
T
,~,
166~ 7
repair synthesis 167~ 166~/
B. Bypass replication (1) pso-AGIO monoadduct I--
r bypass synthesis
166 l--
(2) pso-AG30 monoadduct
bypass synthesis 166 I
FIG. 6. Model tbr psoralen adduet repair (A) and bypass replication (B) in the presence of a triple helix. The stiek diagrams indicate the potential repair pathways for oligonueleotide-direeted monoadducts and erosslinks. The psoralen-eonjugated oligonueleotides are represented by the smaller third strands in each diagram, being connected to the duplex by either one line (monoadduet) or two lines (crosslink). The small arrows mark predicted sites of endonuclease incisions based on the reported properties of the nueleotide excision repair complex in mammalian cells. Displaced arrows are meant to suggest possible inhibition of the endonuclease activity by the third strand. DNA synthesis, either as a component of the repair reaction or in trans-lesion bypass replication, is represented by the dashed lines. Reprinted with permission from G. Wang and R M. Glazer, Altered repair of targeted psoralen photoadducts in the context of an oligonucleotide-mediated triple helix, J. Biol. Chem. 270(38), 22595-22601 (1995).
188
BRIAN P. CASEYAND PETER M. GLAZER
presence of the 10-mer in association with the psoralen conjugate. This may be the result of prevention by the 30-mer of the incision of the preferred strand (118). To explain the origin of the mutations, a model for repair of these crosslinks has been offered (Fig. 6). Other work has also shown an effect of a triple helix on repair. A similar length-dependent inhibition was obtained with a HeLa cell extract experiment that used a Southern blotting technique to examine the identities of excision products (119). Here again, endonuclease activity was prevented in the presence of the TFO but not when the TFO was removed before exposure to the extracts. To further study the influences of triplexes on repair processes in living cells, a luciferase assay was set up in HeLa cells whereby expression was dependent on repair of the crosslinked/triplex region. Psoralen in the absence of a TFO was ostensibly removed at a 6- to 7-fold higher frequency than psoralen linked to a TFO (119). While many indicators seem to point to NER as essential to mutagenesis, is it actually the immediate cause of the misincorporations? The answer seems to be No, and the culprit is likely to be an error-prone repair or lesion bypass system. Yeast knockout experiments furnish persuasive evidence. Plasmids containing a selectable marker crosslinked in vitro by a psoralen-TFO and then transformed into yeast have low survival and a 1% mutation frequency, mostly base substitutions and single base insertions. When the rad18 gene is rendered nonfunctional, and so the error-prone repair pathway is compromised, mutagenesis falls to background levels. The same is not true when radl, responsible for the incisions during NER, is missing. In this case, mutation increases in screens designed to detect insertions but decreases in screens better able to detect substitutions (120). Recombinational repair appears to process triplex lesions in an error-free manner. Elimination of the tad51 gene, which encodes the strand exchange protein, raises mutation levels (120). In human cells, psoralenTFO crosslinks were found to cause an increased level of mutation in cells from patients with xeroderma pigmentosum variant (XPV) (77). XPV cells are now known to be defective in the human rad30 homolog, and thus have abnormality in translesion bypass polymerase activity (121, 122).
XV. Conclusion Triplex technology is beginning to deliver on its promise of allowing the manipulation of gene expression and alteration of gene sequences. While still burdened by concerns about efficient delivery and polypurine target site restriction, improvements in TFO chemistry and transfection methods have greatly enhanced the prospects for ultimate success. It remains to be seen whether other sequence-specific agents are versatile and effective in modifying genomes.
GENE TA11GETING VIA TRIPLE-HELIX FORMATION
189
F r o m early e x p e r i m e n t s in w h i c h T F O s first d e m o n s t r a t e d their b i n d i n g abilities to r e c e n t e x p e r i m e n t s w h i c h have p r o v e n their ability to target e n d o g e n o u s genes in live animals, c o n f i d e n c e in t h e i r p o t e n t i a l has b e e n growing. At t h e very least, T F O s are good m o l e c u l a r tools for basic r e s e a r c h a n d m a y assist in the discovery of i m p o r t a n t facts r e g a r d i n g D N A repair a n d r e c o m b i n a t i o n . But it is n o t w i t h o u t good r e a s o n that these o n e - t i m e in v i t r o curiosities are e m e r g i n g as p o t e n t i a l t h e r a p e u t i c agents.
REFERENCES 1. M. J. Waring, Biochim. Biophys. Acta 1378, 1t63--65 (1998). 2. ~d V. Vlassov, I. E. Vlassova, and L. ~d Pautova, Prog. Nucleic. Acid Res. Mol. Biol. 57, 95-143
(1997). 3. L. M. Neckers and K. Lyer, in "Applied Antisense Oligonueleotide Technology" (A. M. Stein ed.), pp. 147-156. Wiley-Liss, New York, 1998. 4. L. A. Yakubov et al., Proc. Natl. Acad. Sci. U.S.A. 86, 6454-6458 (1989). 5. C. F. Bennett, M. Y. Chiang, H. Chan, J. E. Shoemaker, and C. K. Mirabelli, Mol. Phamnacol. 41, 1023-1033 (1992). 6. D. G. Spiller and D. M. Tidd, Antisense Res. Dev. 5, 13-21 (1995). 7. O. Boussifet al., Proc. Natl. Acad. Sci. U.S.A. 92, 7297-7301 (1995). 8. M. C. Morris, P. Vidal, L. Chaloin, F. Heitz, and G. Divita, Nucleic Acids" Res. 25, 2730-2736
(1997). 9. S.W. Ebbinghaus et al., Gene Ther. 3, 287-297 (1996). 10. S. Agrawal, J. Temsamani, and J. Y. Tang, Proc. Natl. Acad. Sci. U.S.A. 88, 7595-7599 (1991). 11. J. G. Zendegui, K. M. Vasquez, J. H. Tinsley, D. I. Kessler, and M. E. Hogan, Nucleic Acids' Res. 20, 307-314 (1992). 12. W. S. Wade, M. Mrksich, and P. B. Dervan, J. Am. Chem. Soc. 114, 8783-8794 (1992). 13. P. R. Turner and W. A. Denny, Murat. Res. 355, 141-169 (1996). 14. J. W. Trauger, E. E. Baird, and P. B. Dervan, Nature (London) 382, 559-561 (1996). 15. L. A. Dickinson et al., Biochemistr~j 38, 10801-10807 (1999). 16. P. E. Nielsen, Methods Enzynwl. 267, 426-433 (1996). 17. J. Lohse, O. Dahl, and P. E. Nielsen, Proc. Natl. Acad. Sci. U.S.A. 96, 11804-11808 (1999). 18. M. Egholm et al. , Nucleic Acids" Res. 23, 217-222 (1995). 19. V. V. Demidov, M. ~ Yavnilovich, B. P. Belotserkovskii, M. D. Frank-Kamenetskii, and P. E. Nielsen, Proc. Natl. Acad. Sci. U.S.A. 92, 2637-2641 (1995). 20. H. Kuhn, V. V. Demidov, P. E. Nielsen, and M. D. Frank-Kamenetskii, J. Mol. Biol. 286, 1337-1345 (1999). 21. H. J. Larsen and P. E. Nielsen, Nucleic Acids Res. 24, 458-463 (1996). 22. P. E. Nielsen, M. Egholm, 11,H. Berg, and O. Buchardt, Anticancer Drug Des. 8, 53-63 (1993). 23. J. C. Hanvey et al., Science 258, 1481-1485 (1992). 24. L. Bastide, P. E. Boehmer, G. Villani, and B. Lebleu, Nucleic Acids Res. 27, 551~554 (1999). 25. A. F. Faruqi, M. Egholm, and P. M, Glazer, Proc. Natl. Acad. Sci. U.S.A. 95, 1398-1403 (1998). 26. J. Watson, "The Double Helix," pp. 77, 89. Atheneum, New York, 1968. 27. G. Felsenfeld, J. Am. Chem. Soc. 79, 2023-2024 (1957). 28. M. Rougee et al., Biochemistry 31, 9269-9278 (1992). 29. G. Xu and A. G. Goodridge, J. Biol. Chem. 271, 16008-16019 (1996).
190
BRIAN E CASEY AND PETER M. GLAZER
30. J. E. Rao, P. S. Miller, and N. L. Craig, Proc. Natl. Acad. Sci. U.S.A. 97, 3936-3941 (2000). 31. A. L. Guieysse, D. Praseuth, and C. Helene, J. Mol. Biol. 267, 289-298 (1997). 32. A. G. Letai, M. A. Palladino, E. Fromm, V. Rizzo, and J. R. Fresco, Biochemistry 27, 9108-9112
(1988). 33. R. H, Sharer, Prog. Nucleic Acid Res. Mol. Biol. 59, 55-94 (1998). 34. V. I. Lyamiehev, S. M. Mirkin, M. D. Frank-Kamenetskii, and C. R. Cantor, Nucleic Acids Res.
16, 2165-2178 (1988). 35. H. B. Gamper Jr., Y.-M. Hour, M. R. Stamm, M. A. Podyminogink, and R. B. Meyer, J. Am. Chem. Soc. 120, 2182-2183 (1998). 36. P. A. Beal and E B. Dervan, Science 251, 1360-1363 (1991). 37. S. E Singleton and P. B. Dervan, Biochemistry 31, 10995-11003 (1992). 38. L. J. d. Maher, P. B. Dervan, and B. J. Wold, Biochemistry 29, 8820-8826 (1990). 39. W. M. Olivas and L. J. Maher, 3rd, Biochemistry 34, 278-284 (1995). 40. A. J. Cheng and M. W. Van Dyke, Nucleic Acids Res. 21, 5630-5635 (1993). 41. S. W. Blume et al., Nucleic Acids Res. 27, 695-702 (1999). 42. K.J. Hampel, P. Crosson, and J. s. Lee, Biochemistry 30, 4455-4459 (1991). 43. S. F. Singleton and E B. Dervan, Biochemistry 32, 13171-13179 (1993). 44. E. S. Belousov et al., Nucleic Acids Res. 25, 3440-3444 (1997), 45. J. S. Sun et al., Proc. Natl. Acad Sci. U.S.A. 86, 9198-9202 (1989). 46. 8. Kukreti, J. S. Sun, T. Garestier, and C. Helene, Nucleic Acids Res. 25, 4264-4270 (1997). 47. C. Escude et aI., Proc. Natl. Acad. Sci. U.S.A. 95, 3591-3596 (1998). 48. L. C. Griffen, L. C. Kiessling, P. A. Beal, P. Gillespie, and P. B. Dervan,J. Ant. Chem. Soc. 114,
7976-7982 (1992). 49. K. Jayaraman, R. H. Darland, T. S. Rao, G. R. Revankar, V. Bodepudi, N. Chaudhary, and J. Guy-Car{by, Nucleosides Nucleotides 14, 951-955 (1995). 50. O. A. Amosova and J. R. Fresco, Nucleic Acids Res. 27, 4632-4635 (1999). 51. D. M. Gowers and K. R. Fox, Nucleic Acids Res. 27, 1569-1577 (1999). 52. S. D. Jayasena and B. H. Johnston, Nucleic Acids' Res. 20, 5279~5288 (1992). 53. J. S. Lee, M. L. Woodsworth, L. J. Latimer, and A. R. Morgan, Nucleic Acids Res. 12, 6603~3614
(1984). 54. T. J. Posvic and P. B. Dervan, J. Am. Chem. Soc. 111, 3059-3061 (1989). 55. P. S. Miller, Prog. Nucleic Acid Res. Mol. Biol. 52, 261-291 (1996). 56. A. F. Faruqi, S. H. Krawczyk, M. D. Matteucei, and E M. Glazer, Nucleic Acids Res. 25,
633~340 (1997). 57. W. M. Olivas and L. J. Maher, 3rd, Nucleic Acids Res. 23, 1936-1941 (1995). 58. F. Svinarchuk, D. Cherny, A. Debin, E. Delain, and C. Malvy, Nucleic Acids Res. 24, 3858-3865
(1996). 59. J. M. Dagle and D. L. Weeks, Nucleic" Acids' Res. 24, 2143-2149 (1996). 60. L. Lacroix, P. B. Arimondo, M. Takasugi, C, Helene, and J. L. Mergny, Bioehem. Biophys. Res. Commun. 270, 363-369 (2000). 61. L. Lacroix et al., Biochemistry 38, 1893-1901 (1999). 62. C. H. Tung, K. J. Breslauer, and S. Stein, Nucleic Acids Res. 21, 5489~5494 (1993). 63. M. Jasin, Trends Genet. 12, 224-228 (1996). 64. H. E. Moser and P. B. Dervan, Science 238, 645-650 (1987). 65. J. C. Francois et al., Proc. Natl. Acad. Sci. U.S.A. 86, 9702-9706 (1989). 66. K. Mezhevaya, T. A. Winters, and R. D. Neumann, Nucleic Acids' Res. 27, 4282-4290 (1999). 67. L. Perrouault et al., Nature (London) 344, 358-360 (1990). 68. R. Landgraf, C. H. Chen, and D, S. Sigman, Biochemistry 33, 10607-10615 (1994). 69. S.A. Strobel, L. A. Doucette-Stamm, L. Riba, D. E. Housman, and P. B. Dervan, Science 254,
1639-1642 (1991).
GENE TARGETING VIA TRIPLE-HELIX FORMATION
191
70, T. J. Posvic, s. A. Strobel, and P. B. Dervan, J. Am. Chem. Soc. 114, 5934-5941 (1992). 71. E. A. Lukhtanov et al., Nucleic Acids Res. 25, 5077~5084 (1997). 72. R. O. Dempcy, I. E Kutyavin, A. G. Mills, E. A. Lukhtanov, and R. B. Meyer, Nucleic Acids Res. 27, 2931-2937 (1999). 73. E. Ben-Hut and M. M. Elkind, Mutat. Res. 18, 315-324 (1973). 74. Z. Sandor and A. Bredherg, Nucleic Acids. Res. 22, 2051-2056 (1994). 75. K. M. Vasquez, T. G. Wensel, M. E. Hogan, and ]. H. Wilson, Biochemistry 35, 10712-10719
(1996). 76. M. Raha, L. Lacroix, and P. M. Glazer, Photochem Photobio167, 289-294 (1998). 77. M. Raha, G. Wang, M. M. Seidman, and E M. Glazer, Proc. Natl. Acad. Sci. U.S.A. 93, 2941-
2946 (1996). 78. M. Takasugi et al., Proc. Natl. Acad. Sci. U.S.A. 88, 5602-5606 (1991). 79. F. P. Gasparro, P. A. Havre, G. A. Olack, E. J. Gunther, and E M. Glazer, Nucleic Acids Res.
22, 2845-2852 (1994). 80. M. L. Espinas, E. Jimenez-Garcia, A. Martinez-Balbas, and F. Azorin, J. Biol. Chem. 271, 81. 82. 83. 84. 85.
31807-31812 (1996). P. M. Brown, C. A. Madden, and K. R. Fox, Biochemistry 37, 16139-16151 (1998). C. Giovannangeli et al., Proc. Natl. Acad. Sci. U.S.A. 94, 79--84 (1997). E. S. Belousov et al., Nucleic Acids Res. 26, 1324-1328 (1998). N. A. Becket and L. J. Maher, 3rd. Antisense Nucleic Acid Drug Dev. 9, 313-316 (1999). K. M. Vasquez, G, Wang, P. A. Havre, and P. M. Glazer, Nucleic Acids Res. 27, 1176-1181
86. 87. 88. 89. 90.
D. It. Oh and P. C. Hanawalt, Nucleic Acids, Res. 27, 4734-4742 (1999). K. M. Vasquez, L. Narayanan, and P. M. Glazer, Science 290, 530-533 (2000). L. J. d. Maher, Biochemistry 31, 7587-7594 (1992). L. J. d. M',flmr, P. B. Dervan, and B. Wold, Biochemistry 31, 70-81 (1992). G. Duval-Valentin, N. T. Thuong, and C. Helene, Proe. Natl. Acad. Sci. U.S.A. 89, 504~508
(1999).
(1992). L. J. d. M~flmr, B. Wold, and P. B. Dervan, Science 245, 725-730 (1989). N. H. Ing et al., Nucleic Acids Res. 21, 2789-2796 (1993). M. Faria et al., Proc. Natl. Acad. Sei. U.S.A. 97, 3862-3867 (2000). E. M, McGuffie, D. Pacheco, G. M. Carbone, and C. \L Catapano, CancerRes. 60, 3790-3799 (2O00). 95. E Rininsland et al., Proe. Natl. Acad. Sci. U.S.A. 94, 5854-5859 (1997). 96. L. Westin, E Blomquist, J. F. Milligan, and O. Wrange, Nucleic Acids Res. 2g, 2184-2191 91. 92. 93. 94.
(1995). 97. X. S. Xu, P. M. Glazer, and G. Wang, Gene 242, 219-228 (2000). 98. G. Wang et al., Nucleic Acids Res. 27, 2806-2813 (1999). 99. N. E. Mollegaard, O. Buehardt, M. Eghohn, and E E. Nielsen, Proc. Natl. Acad. Sci. U.S.A.
91, 3892-3895 (1994). F. Svinarchuk et al., Nucleic Acids Res. 25, 3459-3464 (1997). S. Kuznetscwa et al., Nucleic Acids Res, 27, 3995-4000 (1999). E A. Havre and P. M. Glazer,]. Virol. 67, 7324-7331 (1993). G. Wang, D. D. Le~, M. M. Seidman, and P. M. Glazer, Mol, Ceil. Biol. 15, 1759-1768 (1995). S. Gryaznov et al., Nucleic Acids Res. 24, 1508-1514 (1996). F. X. Barre et al., Proe. Natl. Acad. Sci. U.S.A. 97, 3084-3088 (2000). A. Majumdar et al., Nat. Genet. 20, 212-214 (1998). T. Tsujimura, V. M. Maher, A. R. Godwin, R. M. Liskay, and J. J. McCormick, Proc. Natl. Acad. Sei. U.S.A. 87, 1566-1570 (1990). 108. Y. Y.Wang, V, M. Maher, R. M. Liskay, and J. J. McCormick, Mol. Cell. Biol. 8, 196-202 (1988).
100. 101. 102. 103. 104. 105. 106. 107.
192
BRIAN P. CASEY AND PETER M. GLAZER
109. W. A. Saffran, C. R. Cantor, E. D. Smith, and M. Magdi, Murat. Res. 274, 1-9 (1992). 110. Z. Sandor and A. Bredberg, Biochim. Biophys. Acta. 1263, 235-240 (1995). 111. A. E Faruqi, M. M. Seidman, D. J. Segal, D. Carroll, and P. M. Glazer, Mol. Cell. Biol. 16,
68204828 (1996). 112. A. F. Faruqi, H. J. Datta, D. Carroll, M. M. Seidman, and P. M. Glazer, Mol. Cell. Biol. 20,
990-1000 (2000). 113. P. P. Chan et al.,J. Biol. Chem. 274, 11541-11548 (1999). 114. Z. Luo, M. A. Macris, A. F. Faruqi, and P. M. Glazer, Proc. Natl. Acad. Sci. U.S.A. 97, 9003-9008
(2000). 115. E A. Havre, E. J. Gunther, 17. E Gasparro, and P. M. Glazer, Proc. Natl. Acad. Sci. U.S.A. 90,
7879-7883 (1993). 116. S. Broitman, O. Amosova, N. G. Dolinnaya, and J. R. Fresco,J. Biol. Chem, 274, 21763-21768
(1999). 117. G. Wang, M. M. Seidman, and E M. Glazer, Science 271,802-805 (1996). 118. G. Wang and P. M. Glazer, J. Biol. Chem. 276, 22595-22601 (1995). 119. A. L. Guieysse, D. Praseuth, C. Giovannangeli, U. Asseline, and C. Helene, J. Mol. Biol. 296,
373--383 (2000). 120. F.X. Barre, C. Giorannangeli, C. Helene, and A. Harel-Bellan, Nucleic Acids Res. 27, 743-749
(1999). 121. C. Masutani et al, Nature (London) 399, 700-704 (1999). 122. R. E. Johnson, C. M. Kondratick, S. Prakash, and L. Prakash, Science 285, 263-265 (1999).
Searching New Targets for Anticancer Drug Design: The Familiesof Ras and Rho GTPases and Their Effectors SALVADOR AZNAR AND JUAN CARLOS LACAL 1
Instituto de Investigaciones Biomddicas CSIC, 28029 Madrid, Spain 196 198 199 201 201 204 205 207 209 211 212 214
Rho Proteins and T h e i r Effectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. W i s k o t t - A l d r i c h Syndrome Protein (WASP) . . . . . . . . . . . . . . . . . . . . . . . B. I Q G A P . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. ACK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. p21-Activated Kinase (PAK) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. R O C K Family: Multifunctional Effectors . . . . . . . . . . . . . . . . . . . . . . . . . F. Phospholipase D (PLD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . G. Protein Kinase N (PKN) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Ras Proteins and T h e i r Effeetors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Raf Kinase Pathway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Phosphatidylinositol-3-OH Kinase (PI3K) . . . . . . . . . . . . . . . . . . . . . . . . . C. R a l G D S F a m i l y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I l l . Pharmacological Approaches to Reverting Transformation by GTPases of the Ras and Rho Families . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. R O C K lnhibitors with Antitumor and Antimetastatic Activity . . . . . . . . . B. P L D as a Target for C a n c e r T r e a t m e n t . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Choline Kinase Inhibitors Have in Vivo Antitumor Activity . . . . . . . . . . . D. Inhibition of Raf Kinase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Inhibition of Ras and Rho Farnesylation . . . . . . . . . . . . . . . . . . . . . . . . . . F. O t h e r Potential Targets for Drllg Design . . . . . . . . . . . . . . . . . . . . . . . . . . IV. F u t u r e Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
215 216 216 217 218 219
221 223 223
The Ras superfamily of low-molecular-weight GTPases are proteins that, in response to diverse stimuli, control key cellular processes such as cell growth and development, apoptosis, lipid metabolism, eytoarchitecture, membrane traflicking, and transcriptional regulation. More than 100 genes of this superfamily grouped in six subfamilies have been described so far, pointing to the complexities and specilicities of their cellular functions. Dysregulation of members of at least two of these families (the Ras and the Rho families) is involved in the 1Corresponding [email protected],
author.
progress in NucleicAc.~idResearch and MolecularBiology,Vol.67
Telephone
+34-91.585.4607;
fax
+34-91.585.4606;
E-mail:
CopyrightO 2001 layAcademicPress.
193
Allrightsof reprodu~on in any form reserved, 0079-6603/01 $35.00
194
SALVADOR AZNAR AND JUAN CARLOS LACAL events that lead to the uncontrolled proliferation and invasiveness of human tumors. In recent years, the cloning and characterization of downstream effectors for Ras and Rho proteins have given crucial clues to the specific pathways that lead to aberrant cellular growth and ultimately to tumorigenesis. A direct link between the functions of some of these effectors with the appearance of transformed cells and their ability to proliferate and invade surrounding tissues has been made. Accordingly, drugs that specifically alter their functions display antineoplasic properties, and some of these drugs are already under clinical trials. In this review, we survey the progress made in understanding the underlying molecular connections between carcinogenesis and the specific cellular functions elicited by some of these effectors. We also discuss new drugs with antineoplastic or antimetastatic activity that are targeted to specific effectors for Ras or Rho proteins. © 2001AcademicPress.
The members of the Ras superfamily of small GTPases are proteins that cycle between an active GTP-bound state and an inactive GDP-bound state (Fig. 1). This activation/inactivation cycle is well regulated by three families of proteins: guanine exchange factors (GEFs), which catalyze the exchange of GDP for GTP; GTPase-activating proteins (GAPs), which enhance their intrinsic GTPase activity; and guanine nucleotide dissociation inhibitors (GDIs), whose function is to inhibit nucleotide exchange, thus stabilizing the protein in a specific state (for a thorough review, see Refs. 1 and 2).
GTP
GDP
SIGNAL
INACTI~
ACTIVE
Pi FIG. 1. Ras proteins cycle between the inactive GDP-bound state and the activated GTP-bound state. The activating loop is regulated by specific exchange factors (GEFs). The inactivating loop is built-in in the GTPase, but it is greatly enhanced by specific GTPase activating proteins (GAPs). Members of the Rho family have additional regulators, guanine nucleotide dissociation ip.hibitors (GDIs), whose function is to inhibit nucleotide exchange, thus stabilizing the protein in a specific state.
N E W TARGETS FOR ANTICANCER D R U G DESIGN
195
More than 100 members of the Ras superfamilyof GTPases have been cloned so far. These GTPases can be subdivided into at least six families: Ras, Rho, Rab, Arf, Ran, and Rad/Gem (1). Each family is composed of several members with distinct expression, cellular localization, and effects. Ras and Rho proteins have been directly related to cell growth control and carcinogenesis. Whereas these GTPases appear to be mediating the housekeeping functions of normal cells, dysregulatory processes such as point mutations or overexpression lead to malignant phenotypes, including metastasis. In this sense, it has been estimated that 25-30% of human cancers have a mutated version of a Ras protein--most frequently K-Ras, followed by N-Ras, and, at a much lower rate, H-Ras. Mutations in different Ras genes have been found in pancreatic cancers, cholangiocarcinomas, adenocareinoma of the lung, squamous head and neck tumors, and acute leukemia, among others (1). Not surprisingly, Ras is the most widely studied oncogene in human carcinogenesis and one of the best targets for intelligent drug design. The relationship of Rho proteins to cell trans~brmation and human cancer is building up strongly. Although it was shown very early that RhoGTPases have transforming properties both in vivo and in vitro, this finding was overlooked owing to the discovery of their role in cell architecture. More recent studies have shown not only the oncogenic potential of Rho proteins, but also the fact that Ras-dependent transformation requires functional Rho proteins, including RhoA, Racl, and Cdc42Hs. Also, expression of oncogenic Rho proteins is sufficient to induce full metastatic potential. Moreover, severa] studies have shown that RhoGTPases might be commonly overfunctional in human cancers, and are overexpressed in colon, pancreas, and lung carcinomas. Finally, several proteins that enhance RhoGTPases exchange, GEFs, and some of their effector proteins are found to induce transformation and metastasis in vivo (3). Both Ras and Rho GTPases mediate key cellular processes that are essential for proper development and attachment of the cell in a specific tissue. This implies that cell-cell and cell-extracellular matrix (ECM) adhesion, actin cytoskeleton reaorganization, cell growth, and apoptosis are tightly regulated by these small GTPases upon well specified stimuli. However, the negative aspect of these multifunctional proteins arises in the context of scenarios that cause their constitutive activation (i.e., point mutations or overexpression) and render them insensitive to regulatory signals. In this ease, these GTPases trigger myriad signals that lead to uncontrolled cell growth, enhanced angiogenesis, inhibition of apoptosis, and genetic instability, all which result in tmnor growth. In addition, permissive loss of cell contact and cell-ECM adhesion, enhanced motility, and regulation of proteases that degrade the cell basement take place, allowing cancer cells to invade the bloodstream or lymph vessels and to attach and proliferate at different tissues.
196
SALVADORAZNARANDJUANCARLOSLACAL
Thus, in the past years much effort has been expended in elucidating the mechanisms that underlie these effects. Recent findings have led us to understand that Ras and Rho GTPase-mediated transformation can no longer be seen as separate events, and that the pathways elicited by both families form a complex weave that finally results in tumorigenesis and metastasis. In pursuit of these pathways, many proteins that act downstream of these GTPases have been cloned. Structural and functional studies have resulted in the classification of these effector proteins into two main groups: kinases (serine/threonine kinases, tyrosine kinases, and lipid kinases) and nonkinases. Only recently has the function of most of these effectors begun to be understood. However, their role in cytoskeleton organization with respect to cell adhesion, motility, and invasion, and their role in transformation have provided enough information to develop drugs with putative antineoplasic activity targeted to inhibit their functions. Some of these drugs are already in clinical trials. Extensive and thorough reviews of the role of the Ras superfamily in transformation and metastasis are available in the literature (1-3). In this review, the specific functions of effectors to Ras and Rho GTPases in transformation and metastasis will be discussed. In addition, a general overview of their potential use as targets for development of new drugs with antineoplasic properties will be presented. Finally, drugs designed to target Ras- and Rho-dependent signaling will be discussed.
I. Rho Proteins and Their Effectors The Rho family of proteins has been intensively investigated in the last few years. The Rho family includes RhoA, RhoB, RhoC, Cdc42, TC10, Racl, Rac2, Rac3, RhoG, Rho6/Rndl, Rho7/Rnd2, Rho8/Rnd3, RhoE, RhoD, and RhoH. Most of the studies rely on RhoA, Racl, and Cdc42 proteins as prototypes. These GTPases are structurally and functionally related small GTPases, with a broad spectrum of critical cellular functions such as cell growth, cytoskeleton organization, cell motility, development, apoptosis, lipid metabolism, and transcriptional regulation. However, the specific signaling network involved in each case is not fully understood. RhoA, Racl, and Cdc42 exert their biological effects through a large number of effectors that are either specific to one or shared by at least two members (Table I). Known effectors for RhoA include rhotekin, rhophilin, citron, and myosin binding subunit (MBS). Effectors that are shared with Racl include kinectin and pl40mDial/2, neither of which has intrinsic kinase activity. RhoA effectors with serine/threonine kinase activity include the ROCK family, CRIK (citron Rho-interacting kinase)/citron kinase, and the PKN family. Finally, lipid-related
197
NEW TARGETS FOR ANTICANCER DRUG DESIGN TABLE I EFFECTORS FOR RHO PROTEINS AND THEIR BIOLOGICAL FUNCTIONS Biological effects
Other Rho Effectors
GTPase
Actin
Cell-cell
specificity
reorganization
adhesion
+ +
+
Transcription regulation
Transformation
Metastasis
+
+
Apoptosis
Rho ROCK family PKN family
Rac (PRK2)
RaedCde42 WASP family IQGAP1/2 PAK ~amily ACK1/2 POSH POR1 Borg family MRCKc~/~ M LK-3 MEKK-1/4 PI3K p70S6 kinasc p67PHOX CIP4
HSF1, SRF
+ +
Citron kinase Citron Rhotekin Rhophilin raDial/2 DAG0 kinase PLD PIP5 kinase
SRF
SRF SRF AP-1
Rac/Cdc42
Cdc42/Racl Cdc42/Racl Cdc42/Rac1 Cdc42
+ + + NF-xB
Rac Cdc42/FC 10 Cdc42 Cdc42/Rac 1 Cdc42/Racl Rac/Cdc42(?)
+
+
Cdc42
enzymes include kinases such as PIP5K (phosphatidylinositol-4-phosphate 5-kinase), DAGOK (diacylglycerol kinase 0) (4), and phospholipase D (PLD). Effectors for Cdc42 with no kinase activity include WASP and N-WASP, Borgs (binder of RhoGTPases), CIP4 (Cdc42-interacting protein 4), and IQGAP1/2. Those with intrinsic kinase activity are ACK-1/2, which are tyrosine kinases and serine/threonine kinases, such as the PAK family, MRCKa/~6, p70 s6 kinase, and members of the MAPKs, MEKK4, and MLK3. In the case of Rac, nonkinase effectors include WAVE (a third member of the WASP family); POR1, POSH, pl40Sra-1, p67PHOX; and those shared with Cdc42, IQGAP1/2, and Borgs. Serine/threonine kinase effectors shared with Cdc42 are the PAK family, MLK3 and MEKK4, and p70 s6 kinase. In addition, Rac has a lipid kinase effector, PI3K (phosphatidylinositol 3-kinase), and
198
SALVADORAZNARANDJUANCARLOSLACAL
both GTPases have been reported to bind to phospholipase D (PLD). The best characterized effectors--in particular, those related to cancer generation and those with potential for anticancer drug development--are considered below.
A. Wiskott-Aldrich Syndrome Protein (WASP) Wiskott-Aldrich syndrome protein (WASP), originally found as a protein with no catalytic activity, contains a CRIB homology domain (Cdc42/Rac interactive binding) (5) specific for Cdc42 (6). Although mainly cytosolic, it contains a putative nuclear localization signal and has been determined by cell fractionation to be present at low levels in the nucleus (7, 8). The gene was isolated by positional cloning and was determined to be mutated in Wiskott-Aldrich syndrome (WAS) patients (7). Patients with this disease display severe thrombocytopenia as a consequence of defective platelet formation (9), eczema, and severe immunodeficiency involving B and T cells, resulting in persistent infections (10). In addition, WAS patients show a high tendency to develop lymphoid malignancies. WASP, which binds to the active form of Cdc42 (11), is implicated in actin polymerization (10) and podosome assembly and disassembly (12) necessary for cell migration. It is proposed that most of the pathologies observed in WAS patients could be a consequence of impaired actin cytoskeletal rearrangements, which are necessary for proper platelet formation and size, as well as for correct T and B cells cytoarchitecture. A recently cloned novel member of the WASP family, WASP-family verprolin-homologous protein (WAVE) (13), shares similar C-terminal structures with the other two members of this family but exhibits no similarities at the N terminus. This region is important for the regulatory aspects of these proteins, suggesting that modulation of WAVE is distinct to that of its family counterparts. In fact, WAVE is a specific effector to Racl rather than Cdc42. WAVE expression causes actin filament clusters, binds to profilin, and binds to constitutively active Racl (13). Furthermore, a dominant negative WAVE mutant with impaired actin reorganization function suppresses Racl-induced membrane ruffling/lamelipodia. This effect is observed for another Rac effector, POR1 (14, 15), suggesting that both POR and WAVE mediate Rac-induced lamelipodia. In a variety of cancer cell lines WAVE is found to be hyperphosphorylated, which correlates with an increase in membrane ruffles. This is presumably carried out by mitogen-activated protein kinase (MAPK), since WAVE is phosphorylated by MAPK upon PDGF treatment of Swiss3T3 cells (16). However, the implications of WAVE in cancer development and metastasis remain unknown. Whether hyperphosphorylation of WAVE is a cause or an effect of tumor progression and/or metastasis should be clarified. Furthermore, WAS patients
NEWTARGETSFOR ANTICANCERDRUGDESIGN
199
show a high frequency of lymphoid malignancies, but the specific role for WASP signaling in this process is unclear.
B. IQGAP Rho GTPases not only transform cells but also are necessary for the invasive phenotype of tumors (2). Under normal conditions, these proteins dynamically control and regulate, together with the extracellular matrix (ECM), cell-cell and cell-ECM adhesion. So far, two sets of effectors to Rac and Cdc42 have been found to mainly control this process: IQGAP and ACK. However, the direct link, if such exists, between metastasis and these effectors awaits description. IQGAP interacts with both Cdc42 and Racl, and together with myotonic dystrophy kinase-related cdc42-binding kinase (MRCK) localizes to membrane ruffles and cell-cell contacts (17, 18). Although it contains many conserved domains, such as SH3 binding domain, WW domain, calmodulin-binding domain, and a RasGAP-like motif, the regulation of these motifs is mostly unknown. It binds directly to actin filaments, crosslinking them, and this ability seems to be partially regulated by calmodulin via its calmodulin-binding domain (19). Whereas in yeast this protein seems to be involved in the recruitment of actin filaments to the budding cell, in mammals it appears to have an important role in Cdc42/Racl-mediated cell-cell adhesion (17). IQGAP1 accumulates in cell-cell adhesions in an E-cadherin//3-catenin-dependent man her (18), and together with MRCK, mediates Racl/Cdc42Hs-induced cell-cell contacts, All three prototypes ofRho GTPases--RhoA, Racl, and Cdc42Hs--regulate E-cadherin activity (17). Apparently, both Rat1 and Cdc42 regulate this protein directly, whereas RhoA may regulate its activity indirectly by reorganizing the actin cytoskeleton. In the context of Racl and Cdc42Hs, this regulation appears to be via IQGAP1 as well as MRCK (for Cdc42). Consistent with a role in adherens junctions, IQGAP1 interacts with ~-catenin, which in turn associates with the cytoplasmic tail of E-cadherin (20). Binding of IQGAP to E-cadherin takes place when both Rac and Cdc42 are in their inactive GDP-bound state. IQGAP constitutively binds to and sequesters -catenin/E-cadherin, preventing their binding to ~-catenin. This results in the disruption of the eadherin-catenin complex (E-cadherin/~-catenin/fi-catenin) necessary for cell-cell adhesion. Once Cde42 is active, it sequesters IQGAP1, which can no longer prevent the ~-catenin/ot-eatenin interaction leading to the formation of the cadherin/catenin complex (18-20). Thus, Rael and Cdc42 positively regulate E-eadherin activity in part by inhibiting the negative regulation of IQGAP1 over the adhesion complex. The relevance of cell-cell contact is most prominent when dynamic cellcell rearrangements take place. This is of particular importance in mammalian
200
SALVADORAZNARANDJUANCARLOSLACAL
embryo development since E-cadherin-mediated cell contacts in the blastoeyst are necessary for proper development to take place (21). In addition, cell-cell adhesion plays a crucial role in cell scattering. Various stimuli, such as hepatoeyte growth factor (HGF), phorbol esters, or v-Src, disrupt cadherin-dependent cell-cell adhesion, resulting in cell scattering and enhanced motility. Interestingly, RhoA, Racl, and Cdc42 participate in this process in the context of both phorbol esters and HGF (22-24). Most importantly, dysregulation of eadherins and eatenins (i.e., changes in the expression or mutations that disrupt their interactions) leads to enhanced migration and invasiveness of tumor cells. For instance, inactivation of endogenous E-cadhefin, either by deletion mutants that exert dominant negative effects or by treatment with E-cadherin-specific antibodies, leads to loss of intracellular adhesion and an invasive phenotype (25, 26). Other mechanisms that lead to abnormal cell adhesion include posttranslational modification of different components such as phosphorylation, competitive protein interactions between components that form the stable adhesion, and the activation state of Rho and Racl (27). However, there is an apparent contradiction with respect to the role of Rho GTPases in tumor invasion, since they either induce cell motility, migration, and metastasis of human cancer cell lines or trigger downstream cascades that ultimately lead to increased cell-cell adhesion. For instance, the inactivation of either Racl or RhoA in MDCK cells leads to a loss of cell contacts (28, 29), and increased Racl activity reduces invasiveness of epithelial cells transformed by different oncogenes (30). In addition, Tiaml (an exchange factor for Racl) and Racl exert an antiscatter/migratory effect via IQGAP1, but have metastatic properties in vivo (31, 32). Interestingly, Tiaml induces cell motility in rastransformed MDCK cells on a collagen matrix, but restores cell adhesion when placed on fibronectin or laminin (31). Furthermore, constitutively active Racl and Cdc42 stimulate ot2fll-mediated motility on collagen (33). Thus, signals from the ECM seem to ultimately dictate the fate of the cell with respect to adhesion and, depending on the extracellular matrix Rho GTPases, promote either invasiveness or enhanced cellular contacts. Furthermore, a unidirectional hierarchy whereby Rac is placed upstream of Rho in an inhibitory manner has been described for several cell lines (34, 35). Upon HGF treatment, Rac becomes rapidly and transiently activated and induces enhanced cell-cell adhesion, whereas Rho stimulates cell scattering and motility in a prolonged manner. The exact mechanism of this interaction between these GTPases and the physiological context in which they take place remain unclear. But it may be that oncogene signaling, together with specific matrix components in contact with the transformed cell(s), determines whether the cell remains attached to the original tissue or detaches and infiltrates into the surrounding stromal ceils to metastasize.
NEW TARGETSFOR ANTICANCERDRUG DESIGN
201
C. ACK As mentioned earlier, another set of effeetors that appear to be involved in cell-cell adhesion is the ACK1/2 family. ACK-1 was originally cloned as a nonreceptor tyrosine kinase that specifically binds to Cdc42, but not Racl or Rho, in vivo. So far, two mammalian members have been cloned, ACK-1 and ACK-2 (36, 37), both sharing extensive primary structure similarities with a tyrosine kinase domain, an SH3 domain, a proline-rich C-terminal tail, and a Cdc42-binding domain (CRIB motif). Serum, EGF, or bradykinin treatment of COS-7 cells leads to rapid binding of ACK-1 to Cdc42GTP and activation of its kinase domain. Thus, this effector might be a converging point between receptor tyrosine kinase and G-coupled receptor signaling (37). ACK-2 is activated in a Cdc42-dependent manner by cell adhesion via integrin complexes requiring binding to integriu fi 1 (38). Recently, it has been observed that increased in vivo anticancer drug resistance of small cell lung cancer cells (SCLC) occurs as a consequence of enhanced adhesion of these cells to the extracellular matrix (EC M) (39). fl 1-Integrin-stimulated tyrosine kinase signaling suppresses chemotherapy-induced apoptosis and promotes tumorigenicity of both tumoral and metastatic cells (which, once attached to a new tissue, are surrounded by an extensive stroma of ECM). Cdc42 is overexpressed in lung cancer (40) and ACK-2 is involved in Cdc42-mediated cell adhesion by interacting with fl 1-integrin. Thus, this effector could have a role in metastasis and enhanced drug resistance of lung cancer cells. Furthermore, the minimum fragment of ACK-1 that binds Cdc42-GTP efficiently inhibits v-Ha-Ras-induced transformation (41, 42). In these works, evidence is provided that Cdc42 acts downstream of Ras upon EGF or NGF challenge of fibroblasts, and that Cdc42 is a key player in Ras transformation.
D. p21-Activated Kinase (PAK) PAK-family effeetors for Rael and Cdc42 have intrinsic serine/threonine kinase activity which is stimulated upon binding to active Racl and Cdc42 (43). Four mammalian PAK isoforms have been cloned so far: rat ~-PAK, mainly expressed in the brain (44), and its human homolog hPAK1 (45); rat fi-PAK (46), homologous to mouse mPAK3 (47); ubiquitous human hPAK2/PAK65G/PAK (43, 48); and PAK4, which is the most unrelated to the others (49). In addition, homologs to the mammalian PAKs are found in S. cerevisae, S. pombe, C. elegans, and Drosophila (50). The functions of PAK in Rac1/Cdc42-mediated cytoskeleton rearrangements, cell-cycle regulation, transcriptional control, apoptosis, transformation, and metastasis are starting to be unraveled. Different isoforms of PAK, activated by Rac and Cdc42, appear to exert specific cytoskeletal changes in response to
202
SALVADORAZNARANDJUANCARLOSLACAL
diverse stimuli. PAK1-3 are not involved in Cde42- or Rael-indueed microspike and lamellipodia formation, but induce other eytoskeletal changes that affect aetin (51, 52). For instance, PAK/hPAK2 phosphorylates ealdesmon and desmin in vivo, inhibiting their downregulatory action over aetin-myosin ATPase activity, leading to enhanced contraction of smooth muscle (53). In addition, in S. cerevisae, PAK homologs Ste20 and Cla4 regulate Cde42-dependent aetin polarization throughout the cell cycle (54). On the other hand, PAK4 is implicated in Cde42Hs-mediated reorganization of the aetin eytoskeleton and in the formation of filopodia (49). Membrane localization of PAK is Rae/Cde42-dependent and in itself seems to increase its activity slightly. In this sense, the Rac/Cdc42 pathway is involved in nerve growth factor (NGF)-indueed neurite outgrowth of neuronal PC12 cells via PAK membrane targeting (55). NGF induces membrane targeting of PAK as well as Rae and Cde42 activation. Interestingly, this effect is independent of PAK1 serine/threonine kinase activity, suggesting that PAK can orchestrate different cellular effects either by its intrinsic kinase activity or by interactions with other proteins at the membrane. The role of PAK in neuronal shape and growth is of particular importance, since PAK3 has been found to be mutated in its kinase domain in nonsyndromie X-linked mental retardation (56). In addition, PAK1 binds via its SH3 domain to the adaptor Nck in response to tyrosine phosphorylation stimuli, which targets it to the membrane (57). Recently, a new family of PAK-binding proteins, Cool/PIX/p85SPR, has been cloned (58-60). So far, two structurally similar members of the Cool family, p50 and p85, which arise from alternative splicing, and two members of the PIX (PAK-interaeting exchange factor) family, u-PIX and fl-PIX (identical to p85Cool), have been cloned. They contain DH (Dbl homology domains), PH (Pleekstrin homology domains), and SH3 domains through which they bind to PAK. Binding of p50 to PAK inhibits Dbl-induced PAK activation, whereas binding of p85 has no effect on PAK activity. These proteins, in turn, associate with a family of Cool-interacting proteins, Cool-associated tyrosine phosphorylated, Cat-1 and -2 (59), which bind to p85 Cool-l, but not p50, and are tyrosine-phosphorylated in a cell adhesion- and cell cycle~lependent manner. Consistent with having a DH domain, both PIX members, when membranetargeted, show GEF (guanine exchange factor) activity for Racl, but not Cdc42, in vivo (59), suggesting a positive feedback loop that increases the intracellular concentration of active Racl. However, the specific regulatory roles of these proteins are unknown. Different studies have suggested a role for PAK in transformation downstream of small GTPases. Ras induces activation of PAK in NIH3T3 fibroblasts, probably via various effeetors, since two mutants of Ras that fail to activate PI3K do not activate PAK when expressed alone but do so efficiently when eotransfected (61). Several other groups have also reported a role for PAK in Rac- and
NEW TARGETSFOR ANTICANCERDRUGDESIGN
203
Ras-mediated signaling and transformation (62, 63). For instance, a mutant p65PAK that lacks the kinase domain but retains the ability to bind to Rae or Cdc42 inhibits Ras-induced TPA response element (TRE)-dependent transcription and transformation in rat 3Y1 and Rat-1 cells (62, 64). And combinations of Rac/Raf, Ras/Raf, and Rho/Raf show synergism in both ERK activation and transformation in a PAK-dependent manner. Furthermore, PAK3, downstream of Rac, upregulates Raf-1 activity by phosphorylating it on a single residue (65). This mechanism is seemingly within the context of Ras, since overex-pression of Racl alone does not result in detectable induction of Rafl/MAPKs, and Ras-activated PI3K leads to Rac activation, resulting in enhanced PAK3 and Raf-1 activities. However, these findings are somewhat controversial since Rat-induced transformation in NIH3T3 fibroblasts is independent of PAK1, which seems to be necessary only for transcription of the cyclin D promoter (66). Taking into account that p65PAK and PAK1 are identical, these differences could not be explained as a result of specific effects triggered by different members of the PAK family. However, it is possible that p65PAK is necessary for transformation in the context of the Ras-Rac pathway, but not for Rat-mediated transformation independently of Ras. In this case, a dominant active mutant of PAK itself does not elicit Schwann cell transformation, but does inhibit transformation induced by Ras in these neuronal cells (67). On the other hand, hyperactive Rat3 and PAK are necessary for human breast cancer cell DNA synthesis and tumor growth in a JNK-independent manner (63). Whether this effect is dependent on Ras is yet to be determined. A role for PAK in JNK and p38 activation by Rac and Cdc42 has been proposed. This pathway is quite controversial, since constitutively active mutants of PAK can stimulate both JNK and p38 activities (45, 68-70) while mutants of Rac that fail to bind PAK remain capable of inducing JNK activity. Moreover, an attenuation rather than a synergism of JNK upregnlation is observed when constitutively active forms of Cdc42 or Rac are coexpressed with hPAK1 (66, 71). In addition, no direct phosphorylation by PAK of any of the members of MAPK that lead to JNK or p38 activation has been observed. Differential specificities and signaling pathways elicited by Rho GTPases in different cells could account for these opposing observations. For instance, RhoA can only activate the JNK pathway in human embryonic kidney (HEK) cells 293T in a PAK-independent manner (72). Furthermore, PAK could be activated by other members of the family of Rho GTPases, which might lead to JNK/p38 activation (73). All three Rho GTPase prototypes, RhoA, Racl, and Cdc42, as well as HaRas efficiently induce nuclear transloeation and activation of NFKB (74, 75). Although the specific function of NFKB is controversial, it seems to be a transforming factor in Ras/Rho GTPase signaling (76-80). Accordingly, NFKB activation is necessary for Dbl- and Dbs-induced transformation via Rho GTPases
(74, 75,81).
204
SALVADORAZNARANDJUANCARLOSLACAL
In addition to PAKs, both Rac and Cde42 can use different effectors in response to specific stimuli to activate the JNK/p38 MAP kinase pathways. One candidate is a Rac target dubbed POSH (Plenty of SH3 domains) (82) which, when expressed in COS-1 cells, mediates Racl-induced JNK activation and NFKB translocation to the nucleus. In addition, two proteins termed MEKK4 and MLK3 are activated by both Rac and Cdc42, ultimately leading to JNK and p38 activation (71, 72, 83, 84). Interestingly, MLK3 links Rho GTPases to NFKB activation and is involved in Rac/Cdc42-mediated transformation of NIH3T3 cells in a MEK-dependent fashion (85). In addition to their role in transformation, MLK3 and MEKK4 might be regulating pathways that lead to invasiveness and metastasis of tumor cells. This is predicted from the fact that activation of endogenous p38 is necessary for the metastatic phenotype of breast carcinoma cells, by promoting transcription and mRNA stabilization of the urokinase plasminogen activator uPAR gene (86, 87). Interestingly, SB203580, a pyrimidazole derivative that specifically inhibits p38, abrogates the metastatic capacity of these cells (86). However, its role in Rac1/Cdc42-induced metastasis is unclear since neither induces transcription of the uPAR gene (88).
E. ROCK Family: Multifunctional Effectors RhoA is a key protein for cell-cycle regulation, apoptosis, differentiation, and transcriptional control. Not surprisingly, Rho is found to cause transformation and metastasis in vivo, and has been found to be overexpressed in several human cancers (89, 90). Major insights on the specific Rho-triggered pathways that lead to these distinct but interconnected effects have come from studies with its known effectors. Among them, ROCK kinases are most important. Two members of this family of serine/threonine kinases have been cloned so far: ROCKII/Rho kinase/ROKa and ROCK1/p160a°CK/ROK3 (91-95). These kinases, concomitantly with other Rho effectors, mediate most of the cytoskeletal changes elicited by Rho GTPase, mainly neurite retraction of neuronal cells, stress fiber formation, and focal adhesion assembly (92-94, 96-98). Recently, an essential role for ROCK family members in Rho-induced transformation and metastasis has been described (99, 100). ROCK has been found to be essential for Rho- and Ras-mediated transformation (99), further strengthening the interdependence of both GTPases in oncogenesis (101). A ROCK-specific inhibitor (102, 103), Y-27632, is capable of inhibiting both growth in soft agar and focus formation of Rho- and Rastransformed cells. Also, it inhibits anchorage-independent growth of two out of four colorectal tumor cell lines tested. Interestingly, this drug is not capable of inhibiting Rho-mediated serum response factor (SRF) activation, transcription of c-fos promoter, or reentry into S phase upon serum stimulation, suggesting that these Rho-induced effects might not be indispensable for transformation.
NEWTARGETSFOR ANTICANCERDRUGDESIGN
205
Moreover, others have reported that ROCK activity is required for Rho-mediated transformation in a manner independent of its cytoplasmic effects (104, 105). Interestingly, no inhibitory effect for Src-transformed cells is observed, indicating that neither Ras nor Rho-ROCK pathways are necessary for Src-mediated transformation. Besides its antineoplastic activity, Y-27632 also inhibits transcellular invasion of rat MM1 hepatoma-induced tumors in vivo (100). Accordingly, a dominant negative mutant of ROCK substantially abrogates the invasive phenotype of these cells, and a dominant active mutant of ROCK elicits tumor invasion in a serum- and Rho-independent manner. Furthermore, Y-27632 treatment of rat MM1 hepatoma cells results in inhibition of Rho-mediated actomyosin cytoskeletal changes, hallmarks of Rho activity. Thus, ROCK-mediated transformation is independent of cytoskeletal changes, whereas the invasive phenotype elicited by ROCK is based on actin reorganization. Given that transformation and cytoskeletal changes induced by Rho GTPases are elicited by independent mechanisms (104, 105), it seems that ROCK could be orchestrating both events downstream of Rho through divergent pathways. Interestingly, treatment of cells MM1 with C3-exoenzyme completely abrogates Rho-induced invasivehess but only partially inhibits ROCK-mediated invasiveness, suggesting that ROCK may be triggering a positive feedback loop that leads to increased Rho activity.
F. Phospholipase D (PLD) Phospholipase D is a ubiquitous enzyme that catalyzes the hydrolysis of phosphatidylcholine (PC) to phosphatidic acid (PA) and choline (Cho). It has been implicated in various cellular processes, such as cell shape, secretion by membrane trafficking, and cell growth (106). Furthermore, a role for PLD in both the structural features and signals that lead to tumorigenesis and metastasis has been described. There are two known human isoforms of PLD--PLD 1 and PLD2 which are differentially regulated (107). PLD1 appears to be regulated by growth factors, protein kinase Ca (PKCot), Src, Ras, members of the Rho GTPase family (Rho, Cdc42, Racl), and ARF (108-116). Whereas activation of PLD 1 by Ras/Rho GTPases occurs by both their proto-oncogenic and oncogenic forms, it has been reported that PLD2 is constitutively active. Also, it is postulated to exert downstream effects by in vivo downregulation of its activity (107). The interaction of PLD1 with RhoA was initially observed in in vitro studies (114), representing a potential effector for this GTPase. Recently, a direct in vivo interaction, with binding sites between both these proteins, has been reported (117, 118). RhoA membrane targeting is necessary for PLD binding, since a nonisoprenylated recombinant RhoA cannot induce PLD 1 activity (119). In vivo direct interaction between PLD1 and Cdc42 has been reported as well (112).
206
SALVADORAZNARANDJUANCARLOSLACAL
Rho GTPases, like Ras, mediate serum and growth factor stimulation of PLD1 activity (111,115, 120-122). However, the specific role of these GTPases upon particular growth factors remains unclear. Rac is upstream of PLD1 in EGF-stimulated fibroblasts independent of ARF, PKC, or RhoA (111). In addition, PDGF induces PLD1 activity, presumably through two independent mechanisms via PKC~ and Ras (121). But treatment of PDGF-challenged fibroblasts with C3-exoenzyme partially inhibits PLD1 activity, suggesting a possible role of RhoA in this process. Furthermore~ activation of PLD by PDGF or phorbol esters seems to occur via both ARF1 and ARF6 (123). In addition, phospholipids play an important role in PLD1 regulation. Increased phosphatidylinositol 4,5-bisphosphate (PtdlnsP2) levels stimulate PLD1 activity in HL60 cells (124). Accordingly, PLD1 contains a PH domain that specifically binds to phosphatidylinositol 4,5-bisphosphate and is necessary for its activity. Binding to PtdlnsP2 is independent of phosphatidylcholine~ and point mutations or deletion of its PH domain completely inhibit enzyme activity and proper cellular localization. Furthermore, PLD activation by m3 muscarinic acetylcholine receptor is mediated by Rho and its effector ROK (125). This activation was measured in the presence of phosphatidylinositol 4,5bisphosphate and is phosphorylation-dependent. Hence, Rho may be activating PLD by direct interaction, via ROK, and by recruiting PLD to the membrane by interaction of its PH domain with Ptdlns2. Whether all three processes occur simultaneously or are preferentially elicited upon specific stimuli is currently unknown. As mentioned above, PLD activity is upregulated in oncogene signaling (115, 120, 121,126), indicating that PLD may facilitate the transformed phenotype. Interestingly, cells overexpressing the epidermal growth factor receptor (EGFR) not only exhibit high levels of PLD] activity, but also become transformed by overexpression of PLD1 (129). This activation is Ras- and Ral-dependent and independent of both PKCot and PKC~ (128). Moreover, PLD activation in v-Raf-transformed cells is inhibited by dominant negative mutants of both Ral and Rho, suggesting a role for the latter in Ras-mediated activation of PLD1 (129). Hence, PLD1 might not only be necessary for oncogene signaling, but also may constitute an oncogene in itself in specific cell systems. However, these results must be taken with caution, since the phenomenon is only observed in a particular cell system and overexpression of PLD1 is detrimental to other cell types (127) or nontransforming to NIH3T3 cells (V. Penalva and J. C. Lacal, unpublished results). Commensurate with a role for PLD in human cancer, this enzyme has been implicated in metastasis of both human tumor cells and experimental tumors (130-132). Metastatic cells have a high affinity for the basement membrane and have the capacity to degrade it via different enzymes, mainly metalloproteases that become activated upon ECM protein-induced signaling (133). For instance,
NEWTARGETSFOR ANTICANCERDRUGDESIGN
207
laminin, a major constituent of the ECM, increases gelatinase A [matrix metalloproteinase 2 (MMP-2)] levels in ras-transformed rat fibroblasts via PLD (130), and inhibition of PLD abrogates invasiveness of these cells. Moreover, laminindependent PA production by PLD in HT1080 metastatic cells causes the release of MMP-2, which results in the invasive and metastatic phenotype of these cells (131). In addition, PLD is essential for the induction of urokinase-type plasminogen activator (uPAR), a protein involved in the degradation of the ECM, in v-Src- and v-Ras-induced tumorigenesis, and in metastasis (132). Not surprisingly, a role for Ras and Rho GTPases in uPAR regulation has been described (134). Both oncogenic Ras and Rho, but not Rac and Cdc42, induce the transcriptional activation of the human uPAR promoter. Thus, Rho and Ras regulated by signals from the ECM, such as laminin or fibronectin (135), might be promoting metalloproteases and uPAR at the transcriptional level in part by PLD. In this sense, ot-thrombin induces nuclear translocation of both RhoA and PLD, which results in PLD activation and nuclear envelope signal transduction (NEST) (136). Furthermore, a role for both RhoA and PLD in promoting transcription of AP-1 sites in T cells in a P KC-dependent manner has been described (137, 138). Interestingly, the proximal region of the urokinase plasminogen receptor contains an AP- 1 site and a KB site specific for NFK B (139, 140). Hence, Rho GTPases might cause enhanced invasiveness and metastasis of tmnor cells via a PLD-uPAR pathway. All the above results suggest that PLD may be a converging or integrated step for both Ras- and Rho-dependent signaling. Since Ras-indueed transformation depends upon Rho signals, it is tempting to speculate that PLD may be a link in this crosstalk.
G. Protein Kinase N (PKN) PKNs are serine/threonine kinases related to PKC that bind to the GTPbound form of RhoA in vivo (141-143). However, PRK2 also binds to active Racl (144). Recently, a novel member of this family ofkinases, PKNfi, has been cloned (145). PKNfi shares high sequence homology to PKN in its kinase domain but, unlike PKN/PRK1, contains proline-rich sequences that are consensus regions for SH3-domain binding. Interestingly, this isoforln is not expressed in normal hmnan adult tissue, but is found highly expressed in human tumors. A Drosophila homolog, termed Ds-Pkn, has been found; this homolog binds active Rho and Rac, and is required for the proper changes in epidermis that occur during dorsal closure of the embryo (145). PKN is involved in the regulation of the cytoskeleton induced by RhoA. PRK2 mediates stress fiber formation induced by RhoA (143), and membrane ruffles elicited by the insulin receptor in Rat-1 and 3T3 are mediated by active PKN as well (146). Furthermore, PKN family members interact with different proteins involved in actin reorganization. In this sense, PKN binds to a-aetinin, which crosslinks aetin, in a phosphatidylinositol 4,5-bisphophate-dependent
208
SALVADORAZNARANDJUANCARLOSLACAL
manner (147). In addition, PKN associates with the neurofilament L, and phosphorylates vimentin and type 111 intermediate filament protein GFAP (glial fibrillary acidic protein) (148, 149). PKN mediates, together with other signals such as the Src-Ras-Raf cascade and PKC-dependent signals, NGF-induced vgfgene expression (150). This suggests that this Ser/Thr kinase might be involved in transcriptional regulation. In fact, upon different cellular stresses such as heat shock, or serum starvation, PKN translocates from the cytoplasm or perinuclear regions to the nucleus and directly phosphorylates and activates heat shock transcription factor 1 (HSF1), which promotes transcription of the et/fl-crystallin gene (151, 152). Also, MEKK2 (MEK Kinase 2) interacts directly with PKN, further suggesting a role for PKN in transcriptional regulation (153). Recently, PKN has been found to promote RhoA-induced transcription of atrial natriuretic factor (ANF) gene via a serum response element (SRE) and, to a lesser extent, an Sp1 site (154). Interestingly, Rac activation of SRF is dependent on both RhoA and PLA2/arachidonic acid pathway (155, 156). Thus, PKN might be mediating arachidonic induction of SRE In addition to PKN, two other Rho effectors are involved in SRF activation--Rho kinase and p140mDial/2 (157, 158). RhoA, Racl, and Cdc42 activate the serum response factor (SRF), which transcriptionally activates reporters that contain the serum response element (SRE), (159). However, whereas both Rac and Rho seem to be capable of inducing SRE chromosomal templates to some extent (160), only oncogenic Cdc42 seems to do it efficiently (161). Surprisingly, the specific functions of SRF in Rho GTPase signaling are mostly unknown. With respect to Rho-mediated transformation, the implication of SRF is controversial. Mutants of Rho in its effector binding domains present contradictory results (104, 105). Although ROCK, Dbl, or RhoB can activate SRF, this activation is independent of their transforming activity (99, 162). Hence, the role of SRF in transformation and its possible activation by PKN must be determined in future studies. Like PAK2, which is cleaved and activated by caspases during apoptosis, PKN is proteolytically cleaved and activated by caspase-3 upon Fas ligand, staurosporin, or etoposide treatment of Jurkat and U937 cells (163-165). PRK2 cleaved in early stages of apoptosis binds to and prevents phosphorylation of Akt in vivo at serine 473 and threonine 308 (166). Phosphorylation of Akt is necessary for its full activation; thus PRK1 inhibits Akt downstream signaling and abrogates its anti-apoptotic effects. Since RhoA-induced tumors show a high apoptotic index in vivo (32), PKN might mediate this effect, at least in part. Furthermore, this pathway might be essential for the antineoplasic activity of farnesyltransferase inhibitors (FTls) (167). The mechanisms whereby PKN ceases to trigger its pro-apoptotic effect in Rho-mediated tumor development are of interest, since the resistance of
NEWTARGETSFOR ANTICANCERDRUGDESIGN
209
tumor cells to cell death is one of the hallmarks of carcinogenesis. Detached metastatic cells anchor to different tissues that lack the survival signals proper to the original one. Hence, these cells must display inhibitory pathways that silence the apoptotic machinery, thereby surviving adverse situations, a process that has been termed anoikis. The state of the PKN family in anoikis needs to be studied in hmnan tumors since proteins involved in this process could constitute targets for antimetastasic compounds.
II. Ras Proteins and Their Effectors The Ras family is composed of several members, including the three human Ras proteins (Harvey-, Kirsten-, and N-Ras) and several related proteins with different degrees of homology in their primary sequence with Ras proteins. This family includes Rap1A and -B, Rap2A and -B, RallA and -1B, TC-21, Rheb, R-Ras, M-Ras, Rit, and Rin. Ras genes are the most extensively studied human oncogenes to date. These proteins have been implicated in a large number of key cellular functions including cell-cycle control in response to many extracellular stimuli, differentiation, cell architecture, cell adhesion, development, apoptosis, senescence, and transcriptional and translational regulation (reviewed in Refs. 1 and 3). Furthermore, the study of Ras signaling is yielding a profound understanding of the steps that take place in oncogenesis. This family of oncogenes has been implicated in 30% of all human malignancies, being mutated in some cancers in up to 90% of the cases (168). Surprisingly, it was originally found that dysregulation of Ras activity in human tumors takes place mainly as a consequence of single-base-pair changes that render the protein nonsusceptible to inactivation (i.e, loss of intrinsic GTPase activity and resistance to GAP activity, or enhanced GDP/GTP exhange) (169-170). Furthermore, like Rho GTPases, ras genes are overexpressed in human tumors with no mutations in the primary sequence, with an average increase in expression of 2- to 10-fold with respect to control tissue (1). A large number of Ras effectors that bind its GTP-bound form have been identified in the past decade (Fig. 2). An obvious application of the identification of effector molecules for Ras is the design of strategies to specifically interfere with downstream signaling that may contribute to a transformed phenotype. Of the effectors for Ras proteins so far identified, the most extensively studied ones are Raf serine/threonine kinases, the RalGDS family of exchange factors, and the lipid kinase PI3Ks. Several other proteins have been shown to interact directly with activated Ras, such as p120GAE NF1GAP, MEKK1, AF-6, Nore-1, Rin-1, and Canoe; however, the roles of some of these are unknown or controversial, and their involvement in Ras-mediated transformation, if any, is not yet determined.
E
/
,,, ~ ~ ~--......
~D
~o
--I
NEWTARGETSFOR ANTICANCERDRUGDESIGN
211
In this section, we focus on Raf, RalGDS, and PI3K signaling, especially with respect to transformation and their potential in anticaneer drug development.
A. Raf Kinase Pathway Raf was tile first identified effector to Ras, and its study represents an essential contribution to the understanding Ras signaling (171-175). Membrane recruitment of Raf- 1 by Ras enhances Raf- 1 kinase activity by a complex and not yet fully identified mechanism that might involve tyrosine kinases (175). Raf-1 activation by Ras leads to a sequential activation of downstream kinases that ultimately results in the activation of p42/p44 ERK/MAPK (176). These proteins, in turn, migrate to the nucleus, phosphorylating and activating several transcription factors that regulate transcription of early-immediate genes such as c-fos (177, 178). Furthermore, induction ofc-myc by Raf-1 has been observed (179); it activates NFzB downstream ofphorbol esters, serum growth factors, and oncogenes (180), and modulates CBP/p300 activity in T cells (181). In addition, active ERKs interact with other cytoplasmic proteins such as p90 ribosomal $6 kinase or PLA2 (phospholipase A2) and RKIP (Raf kinase inhibitor protein) (182). The three mammalian Raf proteins that have been identified so far--c-Raf (Raf-1), A-Raf, and B-Raf--are susceptible to different signals and show different specificities (183). In this sense, whereas Raf-1 and A-Raf are strongly activated by v-Src and weakly by oncogenic H-Ras in fibroblasts, B-Raf is strongly activated by Ras, but not Src (184). However, Raf-1 is strongly activated by oncogenic N-Ras, suggesting that both members of the Ras superfamily might regulate MAP kinase cascades differently (185). Furthermore, this regulation might be cell-type-specific, since Raf-1, and not A-Raf, is strongly activated by Ha-Ras in human embryonic kidney cells (HEK 293) (186). In any case, extensive evidence implicates this cascade in Ras-mediated transformation. For instance, overexpression of constitutively active mutants of Raf-1 (Raf-CAAX) transforms 3T3 fibroblasts (187). Furthermore, a fragment of c-Raf consisting of the minimal region necessary for Ras binding, suppresses v-Ha-Ras-induced transformation in fibroblast (188). However, Raf-1 is not capable of transforming intestinal epithelial cells and is not necessary for Ras transformation in this system, suggesting that its oncogenic potential might be cell-type-specific (189). Surprisingly, transformation of these cells by Ras is dependent on ERK activation (190), suggesting a more complex Ras signaling with respect to MAP kinases. Moreover, whereas mutation ofc-Raf-1 in human cancers has not been detected, it is amplified in different lung cancer cell lines (191). Interestingly, no distinguishable differences are observed in tumors developed by wild-type Raf-1 versus a constitutively active mutant, although tumor development is delayed and tile incidence is lower in the case of the former. Thus, increased levels of c-Raf-1 are capable of inducing lung cancer in mice, suggesting that Raf-1 levels might be an important factor for tumor development in vivo.
212
SALVADORAZNARANDJUANCARLOSLACAL
The roles of Raf-1 in oncogene signaling and in tumor behavior are starting to be understood. Raf-1 is presumably modulated in tumors where Ras is mutated, and in this sense a 4-fold increase in its kinase activity has been observed in mouse liver tumors (192). Interestingly, Raf-1 interacts with Bcl-2 which promotes the anti-apoptotic effect of the latter (193). Interaction of Bcl-2 with Raf-1 does not lead to phosphorylation of the former, but instead is a way of recruiting Raf-1 to the mitochondrial membrane where it binds to and phosphorylates pro-apoptotic BAD (194). Furthermore, a third protein, BAG-l, seems to modulate this pathway by interacting with both Bcl-2 and Raf-1, enhancing the intrinsic kinase activity of latter (195). This pathway might be critical for the resistance of tumor cells to chemotherapy or irradiation-induced apoptosis that is so readily observed. In fact, treatment of four different cell lines derived from human tumors with a drug that targets Raf-1 expression (see below) induces apoptosis in vitro by a mechanism independent of p53 (196). As mentioned above, Raf-1 activates NFKB and this transcription factor is necessary for its oncogenic potential, since inhibition of NFxB signaling suppresses Raf-l-induced transformation (197). Raf-1 does not regulate IKBcd/~ directly, but occurs via MEKK1, suggesting that this Ser/Thr kinase is a key player in Ras/Raf transformation. In addition, the Raf/MAPK pathway has been related to the appearance of hypercalcemia of malignancy by stimulating the release of PTHRP gene expression (198). Thus, activation of Raf by Ras proteins is a critical step in the regulation of cell proliferation and transformation in some cell lines. In addition to its role in cell proliferation, Raf has been shown to induce cell-cycle arrest, differentiation, and apoptosis. Cell-cyle arrest may be mediated by modulating the activity of CKIs (cyclin-dependent kinases inhibitors) (199, 200).
B. Phosphatidylinositol-3-OH Kinase (PI3K) The members of this family of enzymes are lipid kinases that catalyze the phosphorylation of the 3' position of phosphatidylinositol 4,5-bisphosphate (Ptdlns[4,5]Pz) to yield phosphatidylinositol 3,4,5-triphosphate (PtIns[3,4,5]P3). Several isozymes have been cloned so far that respond to either growth factor receptor tyrosine kinases or G-coupled heterotrimeric receptors (1). Two heterodimeric forms of PI3K, ot and/3, consist of a catalytic subunit of 110 kDa (pll0a and pll0/~); these isoforms interact and are modulated by different adaptor subunits of 85 kDa. These two isoforms are regulated by receptor tyrosine kinases (RTKs). On the other hand, PI3KF does not interact with adaptor subunits and responds to stimuli coupled to heterotrimeric G-coupled receptors. Ras binds to and activates the catalytic subunit of PI3Ka in response to diverse growth factors such as NGF or EGF, and a direct activation of PI3K by oncogenic Ras takes place as well (201, 202). In addition, activation of
NEW TARGETSFOR ANTICANCERDRUG DESIGN
213
endogenous Ras induces a subtle increase in PI3K activity which leads to some, but not all, of the effects observed under full activity (203). Downstream of PI3K, its lipid product interacts with several intracellular proteins via their Pleckstrin homology domains (PH), which results in the control of metabolic regulation, cell proliferation control, and cell survival. In this sense, Ptdlns(3,4,5)P3 interacts with protein kinase B (PKB/Akt), recruiting it to the membrane, where it can interact with several other proteins that phosphorylate it and activate its kinase domain (204). PKB not only interacts with glycogen synthase-3, phosphofructokinase, and GLUT4, thus regulating intracellular metabolism (205-208); but it also exerts an anti-apoptotic signal by modulation of NFKB, p53 or BAD (209-211). Another set of proteins whose PH domain is membrane-recruited by Ptdlns [3,4,5]P3 is the p70 ribosomal $6 kinases (212). Four isoforms of this kinase have been found so far, and they regulate translation of polypyrimidinecontaining mRNAs. In addition to the aforementioned effects, PI3K has been proposed as one of the links between Ras and Rho signaling. This comes from the observation that RacGEFs (Vav and Sos) become activated upon binding of the PH domain to the lipid product of PI3K (213), and that Ras and PI3K are necessary for Rac activation (214, 215). In addition, PI3K activation by Ras induces lamellipodia, a well-known cytoskeletal effect of Rac (203). Interestingly, PI3K appears to be downstream of Cdc42 and Racl as well (33), and its activity is necessary for collagen-mediated cell motility and invasion via both GTPases. Whether this is a consequence of a positive feedback loop within the context of Ras, or a specific relationship between Rho GTPases and PI3K independent of Ras is still unclear. PI3K has been extensively implicated in cancer. For instance, transformation of hematopoietic cells by the fusion oncogene BCR/ABL requires PI3K activity, since mutants of BCR/ABL incapable of binding to PI3K are devoid of its transforming potential (216). Moreover, PI3K is constitutively active under a naturally occurring oncogenic epidermal growth factor receptor (EGFRvlII), and treatment of EGFRvlII-transformed cells with LY294002, a PI3K-specific inhibitor, suppresses their oncogenic phenotype (217). Most importantly, PI3K has been implicated in carcinogenesis, angiogenesis, and metastasis in vivo (218-225). In this sense, a natural deletion mutant cloned from a transformed cell line, p65-PI3K, which includes the initial 571 residues of the wild-type p85ot subnnit, is capable of constitutively activating PI3K in vivo and transforming cells, hence constituting an oncogene (218). It was later shown that transgenic mice with T cells expressing this oncogene showed infiltrating lymphoproliferative disorder and autoimmune renal disease (221). This was confirmed with a more in vivo approach in which the PI3KCA gene (encoding for the p110c~ catalytic subunit) was found to be amplified in approximately 40% of ovarian cancers and correlated with the transformed phenotype (219). Furthermore, treatment of ovarian
214
SALVADORAZNARAND JUANCARLOSLACAL
cancer cells with LY294002 induced apoptosis and decreased proliferation of tumors (222, 224). Recently, this same gene has been found to be amplified in cervical cancer (220). In addition, PI3K has been implicated in small cell lung cancer cells (222).
C. RalGDS Family Ral GDP dissociation stimulators (RalGDS) are a family of exchange factors for Ral GTPases that bind Ras in their active state (226-228). These include RalGDS itself, Rgl (RalGDS-like) (228), Rgl2 (229), and Rlf (RalGDS-like factor) (230). Stimulation of RalGDS by growth factors leads to Ral activation in a Ras-dependent manner (231). However, not much is known about the physiological roles of RalGDS downstream of Ras. One putative downstream target for Ral proteins, RIP (Ral-interacting protein) is a GAP (GTPase-activating protein) specific for Rac and Cdc42, thus constituting together with PI3K a second possible link between Ras and Rho signaling (232-235). Furthermore, there is a link between RalGDS and PLD activation both in the context of Ras and Src, as described above. Even though Ral is the main downstream target of RalGDS, this protein might signal through other downstream proteins as well. This comes from the observations that Ral and RalGDS induce distinct effects when overexpressed in murine fibroblasts. For instance, whereas RalGDS can induce anchorageindependent growth and tumorigenicity of NIH3T3 cells, Ral fails to do so (235). On the other hand, a dominant negative mutant of Ral can suppress Rasmediated transformation, suggesting that this pathway is necessary for full Ras transforming potential, but not sufficient to exert an oncogenic effect on its own (236). Moreover, both RalGDS and Ral can induce c-fos transcription by an apparently different mechanism (237). Several other reports have implicated RalGDS family members in Rasmediated transformation (238-242). In keeping with a role for RalGD S members in Rho GTPase signaling, recently, oncogenic Rgr has been found to transform cells in a manner dependent on the Raf pathway as well as Rho signaling to the nucleus (242). Whereas Ras is necessary for Rgr full oncogenic potential, Rho is essential for its transcriptional activation. In addition, Rgr induces both JNK and p38 activation, suggesting a role for Rac and Cdc42 in the regulation of these kinase cascades. However, no attempts have been made so far to develop drugs targeted to RalGDS or Ral signaling. This might be expected since little is known about the specific mechanism through which these proteins induce transformation independently, or in the context of Ras. Furthermore, no systematic work has been done to assess whether RalGDS is implicated in tumorigenesis in vivo, nor its state in human cancers. Thus, further work will have to be done to specify the specific role(s) of this pathway in oncogenesis and to determine whether they constitute putative targets for new drugs with antineoplastic activity.
215
NEW TARGETS FOR ANTICANCER DRUG DESIGN
III. PharmacologicalApproaches to Reverting Transformation by GTPases of the Ras and Rho Families Ras and Rho proteins have emerged as prominent players in human carcinogenesis. As a consequence, signaling pathways governed by these GTPases are potential candidates for a therapeutic intervention in human cancer, provided that they represent crucial steps in the generation or maintenance of tumor cells (Fig. 3). In addition to the GTPases themselves, a number of effectors for Ras and Rho proteins are considered targets for intelligent drug development. We have selected a few examples of approaches that are either under development for a specific blockade of Ras- and Rho-mediated transformation, or may be developed in the near future. The complexity of the signaling cascades elicited by GTPases of the Ras and Rho families presages the difficulties that will be found in this endeavor. But despite the increasing complexity of the signaling network that is being unveiled for both Ras and Rho proteins, some progress has been made, allowing us to be optimistic.
~k,,RhoA, Rat1, Cdc42J
• o27~.~ PKN
~•
(
',
LD ' t,
,,
~ •
,' II
S
RAF
I*
ChoK
FIG. 3. Signaling pathway and effector candidates fbr drug development. Ras and Rho signaling pathways arc potential candidates for anticancer drug development. Pharmaceutical intervention is pursued in sevend of these pathways for both Ras- and Rho-dependent signaling. Some of these targets are specific for each family, Others impinge on both signaling routes and may constitute more efficient antitumor drugs. In addition, Ras processing and membrane location are also targets, ahhough the mechanisms of action remain controversial,
216
SALVADORAZNARAND JUANCARLOSLACAL
A. ROCK Inhibitors with Antitumor and Antimetastatic Activity As described earlier, the inhibition of ROCK action is of interest for both tumor progression and invasion. Evidence has been reported for a correlation of antitumoral activity achieved by ROCK inhibition by Y-27632. Several compounds related to Y-27632 are under study (102). These inhibitors show Ki values for ROCK kinases approximately 100 times lower than those of other serine/threonine kinases, such as PKN, citron kinase, MLC kinase, and cAMPdependent kinase. The in vitro inhibition exerted by this drug seems to follow competitive kinetics with ATP at concentrations as low as 10 -5 and with complete inhibition at 10 -7. However, given that the intracellular concentration of ATP is millimolar, a second mode of inhibition in vivo has been proposed. Upon binding to ROCK, Y-27632 competes with ATP for the active site of the enzyme, but also induces a conformational change that might interfere with ROCK binding to targets or regulatory proteins (102). The design of new related compounds with lower Ki values is being developed, and another compound, Y-30141, is 10 times more active than its relative toward ROCK inhibition. The in vivo antineoplasic action of ROCK-inhibiting drugs is still not completely understood, but it seems that the drug uptake is by a carrier-mediated facilitated diffusion without concentration in the cell (102). A continuous exposure of syngeneic rats with this drug is necessary to achieve a considerable reduction of invasiveness of tumor cells but, most importantly, without adverse side effects (100). Hence, the development of drugs with enhanced ROCK-inhibitory properties promises to be a useful tool for both antineoplastic and antimetastatic treatment and awaits clinical trials.
B. PLD as a Target for Cancer Treatment Consistent with the reported role for PLD 1 in carcinogenesis, different drugs with in vitro antitumor activity inhibit PLD activity. For instance, suramin, xanthogenate derivatives, or aminosteroids inhibit PLD with an IC50 on the order of 1 IxM (243). Surprisingly, the antitumor phospholipid analog, hexadeeylphosphocholine (HePC)/mitelfosine, induces a rapid activation of PLD in human breast fibroblasts (244). Similar effects are observed with a variety of cell types (244a). Thus, activation of PLD rather than inhibition (an effector for both Ras and Rho signaling) seems to accompany the antitumor actMty of the drug. Mitelfosine is a cytostatic drug commonly used, along with other conventional chemotherapeutics, in topical treatment of cutaneous metastases derived from breast carcinoma, with little nonspecific cytotoxieity (245). Other works suggest that this compound and other derivatives of the family of alkylphosphocholines (APCs), also induce apoptosis of tumor cell lines but not of normal cells (246, 247). This effect is thought to occur via the SAPK/JNK pathway, which is
NEW TARGETSFOR ANTICANCERDRUG DESIGN
217
rapidly activated upon HePC treatment and enhances radiation-induced apoptosis of tumor cells. In addition, HePC increases ceramide levels and exposure of HePC-treated cells with fuminosin B1, a specific inhibitor for ceramide synthesis, abrogates HePC-induced apoptosis (248). These compounds also induce apoptosis in some human leukemia cells with no hematological toxicity (249). Studies of the mechanism of action of PLD using HePC suggest that this enzyme contains two substrate-binding sites, one without catalytic activity that binds the substrate, and a second pocket which exerts the intrinsic catalytic activity of the enzyme (250). HePC presumably competes with phosphatidylcholine for both binding sites and, once bound, is hydrolyzed with much slower kinetics than the natural substrate. This is seemingly in contradiction with an activation of PLD by HePC (244), since the drug competes with its natural substrate, inhibiting the release of choline and phosphatidic acid. However, these contradictory results seem to have a trivial explanation. Monitoring PLD activity over a prolonged treatment with HePC in NIH3T3 and MDCK cells demonstrates that whereas short drug exposure (1 h) induces a rapid activation of PLD in a PKC-independent manner, exposure of cells to HePC for 24 h completely abrogates PLD activity, rendering it insensitive to both PKC and phorbol ester stimulation (244a). Although the specific mechanism through which this drug can initially activate, then inactivate PLD later, is still unknown, this provides the basis for its antitumoral activity, consistent with the reported role of Ras, Rho, and PLD in tumorigenesis. The applicability of PLD-inhibitory drugs as antineoplasic or antimetastatic compounds is starting to be revealed. But, further research is necessary to design new drugs with enhanced activities. The recent determination of the crystal structure of bacterial Streptomyces PLD (251) may help to further our understanding of its mode of action, and allow us to design more potent inhibitory drugs with enhanced antitumor, and possibly antimetastatic, properties.
C. Choline Kinase Inhibitors Have in Vivo Antitumor Activity Choline kinase (ChoK) is a eytosolie enzyme that catalyzes the phosphorylation of choline to yield phosphorylcholine (PCho) (252). In the past few years, this enzyme has evoked interest because it appears to be involved in the regulation of cell growth by certain growth factors (253, 254). Moreover, increased levels of PCho as a consequence of elevated ChoK activity in ras-transformed cell lines has been observed (255-257). This holds true for human tumors where PCho and ChoK levels are increased compared to the normal tissue (260, 261). Accordingly, a family of drugs targeted to inhibit ChoK exerts antitumoral activity both in vitro and in vivo (260, 261). The characteristic of this signaling pathway and its implication as a target for anticancer therapy have attracted some interest recently (262).
218
SALVADORAZNARANDJUANCARLOSLACAL
ChoK might play a key role in the signaling pathways induced by oncogenes. Cell lines transformed with ras, src, raf, and mos show elevated levels of PCho as a consequence of increased endogenous ChoK activity (260, 263, 264), and the same is observed in naturally occurring human tumors (258, 259, 265). By contrast, overexpression of the v-fi~s oncogene has no effect on ChoK levels. In addition to Harvey-, Kirsten-, and N-ras, cell lines transformed by mutated RhoA proteins have increased basal ChoK activity (R. de Molina and J. c. Lacal, unpublished). This effect does not seem to be a mere consequence of oncogenic signaling, but rather an essential player in tumor development, since inhibition of ChoK abrogates growth factor-induced mitogenesis and addition of PCbo to cells induces entry into S phase (253, 260, 266, 267). Recently, a new family of compounds derived from hemicholinimn-3 (HC-3) has been shown to be highly specific for ChoK inhibition and to exert antitumoral activity in cell lines and in mice (260-262). Thus, addition of HC-3 derivatives to transformed or tumoral cells leads to an antiproliferative effect by inhibiting choline kinase, which prevents entry into S phase without affecting the ERK1/2 pathway, PI3K, PI-PLC, or PC-PLD (260, 261). Furthermore, unlike the parent molecule, these drugs can be administered daily to mice at a dose up to 5-35 mg/kg of body weight for up to 10 weeks without adverse deadly side effects (261). Treatment of these mice, previously injected with either the P388 murine tumor cell line or human tumoral cell lines HT29 and A431, results in 80% inhibition of tumor growth with an increased life span value of at least 125% (261,262). Taken together, these results suggest that ChoK plays a crucial role in the onset of carcinogenesis. The synthesis of new compounds derived from HC-3 with enhanced antineoplastic activity based on their inhibitory effect over this enzyme has proven to be a useful strategy. However, much work needs to be done not only to further our understanding of the specific implications of ChoK in tumor development, but to improve the antitumoral activity of compounds targeted to its activity. Our recent findings suggest that these drugs not only induce cytostatic effects, but also might be triggering apoptosis of tumoral cells without any effect on their normal counterparts (A. Rodriguez-Gonzalez and J. c. Lacal, unpublished data). This is of great interest because it may be possible to achieve not only tumor growth inhibition but also tumor disappearance in vivo. Moreover, the fact that many human tumors display high ChoK activity and elevated levels of PCho points to a probable general role of these compounds in cancer treatment (258, 259, 265).
D. Inhibition of Raf Kinase Different approaches have been made to develop drugs targeted to Raf-1 kinase activity with possible antineoplastic activity. One such approach based on antisense oligodeoxynucleotide is already in phase I trial and is yielding promising results (268-273). A 20-mer phosphorothioate antisense oligodeoxynucleotide
N E W TARGETS FOR ANTICANCER D R U G DESIGN
219
(termed ISIS 5132 or CGP 69846A) targeted to the 3'-untranslated region of human Raf-1 mRNA has been synthesised; this oligodeoxynucleotide specifically inhibits Raf-1 gene expression and suppresses tumor progression of a variety of human tumor cell lines (breast, small cell lung, large cell lung, colon, and squamous lung carcinoma) in mice (273). Moreover, combination of ISIS 5132 with conventional chemotherapy drugs (e.g., tamoxifen, cisplatin, adriamyein, or mitomycin C) results in a synergiste antitumor effect with complete tumor regression in transplanted syngeneic mice (272). Furthermore, when injected intravenously, a new liposome-based delivery method for ISIS 5132, termed LE-5132 (for liposome encapsulated), has proved to be more effective than the naked compound toward nude mice injected with squamous cell carcinoma cells (SQ-20B) (271). Whereas both LE-5132 and radiation produced a eytostatie effect in SQ-20B cells, a combination of the two induced a marked tumor regression in all cases. Accordingly, this compound has been approved for phase I clinical trials (268-270). The drug is well tolerated at doses up to 6.0 mg/kg, with fatigue and fever as side effects, when administered over 3 weeks (2 70). Corresponding with its effect~ Raf-1 levels in peripheral blood mononuclear cells (PBMC) decrease in a dose dependent manner, which correlates in some patients with clinical benefits (269). Recently, continuous administration of ISIS 5132 during 21 days at a 4.0-mg/kg dose has been carried out, with monitoring of side effects and tumor behavior (268). As previously described, the side effects were minimal; but most importantly, 2 out of 30 patients were stabilized with respect to tumor progression, and one patient with ovarian carcinoma displayed a 97% reduction in carcinoembryonic antigen-125. No other drugs targeted to Raf-1 activity (i.e., expression of intrinsic kinase activity) have reached clinical trials. Whether ISIS 5132 is to become a common therapy in cancer treatment, either alone or together with conventional drugs, is a question only the future can answer. Recently, a combination of taxol and AS101 in nude mice was surprisingly found to activate and depend on Ras/Raf-1/MAPK (ERK1/2) activity (274). Apparently, the cell-cycle arrest and proapoptotic effects of this drug combination take place via the Raf-dependent regulation of p21 (waf) and Bel-2, respectively. Should drugs targeted to Raf-1 prove a reliable therapy, further research will surely be done, either to enhance its activity (by improving its delivery methods) or to synthesize new compounds targeted to Raf kinases.
E. Inhibition of Ras and Rho Farnesylation Ras and Rho proteins require membrane targeting in order to be functional (27,5). Membrane localization is achieved by a complex meehanism that involves C-terminal prenylation, protease cleavage, carboxymethylation, and palmitoylation (275). A strategy was set up to design efficient inhibitors of the enzymes responsible for Ras prenylation, thereby preventing Ras proteins from becoming
220
SALVADORAZNARANDJUANCARLOSLACAL
membrane-associated and halting its oneogenic potential (276). The initial approach was to inhibit the enzyme farnesyl transferase (FT), which is responsible for processing the Harvey-Ras protein. Several types of FT inhibitors (FTIs) have been developed in recent years, but the best characterized ones belong to the family ofpeptidomimetic drugs. FTIs designed toward Ras processing were found to be highly successful nontoxic antiproliferative and antitumor drugs, and reversed the degenerative growth of tumors by inducing apoptosis of tumoral cells in vivo (277, 278). However, later findings demonstrated that the most frequent ras oncogenes found in human tumors, Kirsten- and N-ras (which account for more than 90% of cases), follow different processing mechanisms from those in Harvey-ras (275). Furthermore, cell lines transformed by oncogenes that are dependent or independent of Ras signaling were found to be sensitive to FTIs. The mystery of the mechanism of action of FTIs remained until recently. Originally, these drugs were thought to exert their antitumoral activity by inhibiting Ras isoprenylation (279); in the past few years, however, it has become apparent that their activity could be linked to inhibition of farnesylated RhoB
(280-282). RhoB shares over 90% homology with RhoA. However, its cellular localization is different, and it mediates different cellular processes. RhoB localizes mainly to early endosomes and perinuclear region, and it mediates receptor trafticking (283, 284). Interestingly, it becomes farnesylated (F) and geranylgeranylated (GG), two processes that are under the control of different enzymatic entities. Both RhoB-F and RhoB-GG end up attached to endosomal vesicles (285). RhoB binds to and activates PRK1 (286). Upon FTI treatment, an accumulation of RhoB-GG takes place, shifting its cellular localization from the endosome to the plasma membrane. The mechanism that causes this shift is unknown, but this cellular relocalization mediates cell growth arrest and induction of apoptosis by farnesyltransferase inhibitors (280, 281). Given its involvement in Rho-mediated apoptosis, PRK1 might be a key player in FTI-induced effects. Furthermore, PRK1 has recently been shown to interact with PDK1 (3-phosphoinositidedependent kinase-1), a protein that phosphorylates Thr 308 and partially activates Akt in vivo (287). Interaction of PRK1 with PDK1 is thought to allow the former to phosphorylate Set 473 as well, resulting in full activation of Akt. Interestingly, both Bcl-XL and Akt activation block farnesyltransferase effects (288). Thus, PRK1 seems to be triggering an antiapoptotic effect, in contradiction to what is observed upon FTI treatment. However, given that RhoB-GG is targeted to the membrane, it has been proposed that it might elicit some effects similar to those of RhoA. Accordingly, overexpression of RhoB-GG or FTI treatment induces stress fiber formation (281), and ectopic expression of RhoBGG inhibits Akt activation by Ras (167). Thus, upon RhoB-GG relocalization at the membrane, PRK1 might no longer interact with PDK1, preventing full Akt activity; and/or its effects might resemble that of PRK2, which fully inhibits Akt. Whether this is the case remains to be seen.
NEWTARGETSFOR ANTICANCERDRUGDESIGN
221
F. Other Potential Targets for Drug Design In addition to the above-mentioned targets, an increasing number of candidates are showing potential for drug discovery in this field. Here, we briefly describe those of most interest, including some that could be developed into real drug targets within a short period. The involvement of WASP in tumor progression is still under investigation. Should a definite link between tumor development and WASP members be made, different approaches could be taken when developing potential anticancer/antimetastatic drugs targeted to WASP family members. A compound capable of inhibiting its downstream activity by impairing its ability to bind downstream targets (Arp2/3, Nck, Grb2) could be used. In addition, the precise mechanism of WASP autoinhibition and activation is known (289). The C-terminal region of WASP interacts with its GTP-binding domain, occluding residues that would otherwise interact with the Arp2/3 complex. Upon Cdc42GTP binding, a conformational change takes place that liberates the residues responsible for the interaction with the actin-polymerizing machinery. Thus, drugs that compete for this essential region could be useful as therapeutic tools. The fact that Rac is overexpressed in metastatic human tumors suggests that IQGAP1 activity must be downregulated (32, 40, 290-292). However, the mechanisms whereby IQGAP1 is presumably inhibited, either by fibronectin or laminin extracellular matrix-dependent signaling or by opposing RhoA effects, are totally unknown. IQGAP contains an SH3 domain, suggesting that it may be under the regulation of diverse signaling pathways. In addition, it contains a RasGAP-like motif with no apparent activity. This motif could be linking IQGAP1 to other small GTPases, signaling in an inhibitory manner. Furthermore, it contains a WW motif that binds proline-rich consensus epitopes involved in protein-protein interactions in a diversity of signaling cascades (293, 294). Hence, the precise mechanism of activity of IQGAP1 in Racl-induced tumor invasion and metastasis will provide clues on possible treatments to restore its fimction, promoting cell-cell adhesion and reduced invasiveness of cancer cells. Drugs targeted to inhibit proteins that downregulate this effector may offer a plausible strategy for antimetastasis treatment. As described above, the specific mechanism through which ACK mediates transformation is currently unknown. Nevertheless, chemical inhibitors targeted to inhibit its intrinsic kinase activity could constitute drugs with potential antineoplastic activity, For instance, the use of tyrphostin-based chemical inhibitors specific to tyrosine kinases has begun to emerge for in vivo treatment of cancer cells (295, 296). Moreover, a combination of tyrphostin treatment with conventional chemotherapeutics such as cisplatin or doxorubicin results in enhanced antineoplasic effects, with reduced nonspecific toxicity for the latter compound (297). Although recently cloned, there is already a peptide inhibitor derived
222
SALVADORAZNARAND JUANCARLOSLACAL
from the minimal sequence of ACK required for Cdc42-binding with possible in vivo anticancer properties. The efficacy of this inhibitor as an anticancer drug is currently under study. Furthermore, its kinase activity could be targeted by tyrosine kinase inhibitors, which could lead to decreased cytotoxicity of anticancer conventional drugs as well as suppression of tumor growth and metastasis. Signals downstream of serine/threonine kinases activated by Rac and Cdc42 lead to transformation. Although their mechanisms of action are beginning to be elucidated, the overall essential pathways that lead to carcinogenesis are unknown. Furthermore, structural data are needed in order to design possible inhibitors with putative anticancer properties targeted to these kinases. Even though the ldnase domain of PAK is predicted to function as a general serine/threonine kinase domain and the crystal structure of Cdc42 bound to the GTPase binding domain of PAK has been described (298), no crystal structure of PAK is available. Furthermore, neither its kinetics nor the crystal structure bound to a specific substrate has been determined. Drugs targeted to serine/threonine kinases are available, such as balanol derivatives, which could be structurally modified and tested as inhibitors of PAK activity (299, 300). In addition, specific inhibitors targeted to MAPKs are common tools in molecular biology, with good knowledge on their mechanism of action (301); however, their potential as anticancer therapeutics is unknown. PI3K is a promising candidate for novel anticancer drugs. Its intimate implication in a number of cancers as an oncogene--and the fact that both its kinetics and signaling are quite well understood--makes it a more than plausible target for cancer therapy. However, few drugs targeted to PI3K activity are being tested for antineoplastic activity, and to date no PI3K inhibitor is in clinical trials (302, 303). Flavonoid derivatives are being developed with enhanced inhibitory activity over both PKC and PI3K (304). Two already classic PI3K inhibitors, LY294002, and wortmanin/viridin, have antitumoral activity, both in vitro and in vivo. But in the case of the latter, its mode of action is unclear (i.e., it shows a high promiscuity over other proteins), and it is quite reactive and unstable (305, 306). In addition, inositol derivatives such as inositol 1,3,4,5,6pentakisphosphate and inositol 1,4,5,6-tetrakisphosphate have been shown to suppress PI3K downstream signaling, not by inhibition of the kinase itself, but by preventing PKB interaction with Ptlns[3,4,5]Pa (307). These compounds prevent anchorage-independent growth of mammary carcinoma and small cell lung cancer cells. However, all these compounds are in early preclinical trials, and hence far from approval for treatment of patients. Finally, it has been reported that farnesyltransferase inhibitors might in part exert their in vitro apoptotic activity in tumor cells by inhibiting the PI3K/Akt/BAD pathway (308). As indicated above, the mode of action of these drugs appears to be modulation of RhoB activity in which PKN/PDK could be playing a role in the antitumoral
NEW TARGETSFOR ANTICANCERDRUG DESIGN
223
effects observed. Interestingly, PI3K and PDK interact and are involved in the Akt survival pathway. Thus, whether the role of PI3K/Akt in FTI treatment lies within the Ras/RhoB context is still unclear and deserves further investigation. Interestingly, knockout mice of the catalytic subunit ofPI3K F develop spontaneous malignant epithelial tumors in colorectum (309) Thus, drugs targeted to inhibit PI3K activity in general might have detrimental effects, and it might be necessary to design inhibitors specific to those isotypes that have been directly related to cancer, such as the pll0cl.
IV. Future Perspectives The concept of transformation by ras oncogenes has changed dramatically since the discovery of the intricate network among Ras and Rho proteins and their effectors. Evidence has been accumulated that demonstrates an active role of members of the Rho family of GTPases and their regulators and effectors in the development of human cancers. As a consequence, attempts to interfere selectively with Ras-dependent signaling needs to take into account the possible involvement of Rho proteins as well. It may be considered that because the complexity of the signaling web is still far from being fully understood, our efforts to design specific drugs that will efficiently and specifically interfere with the transforming phenotype will fail. However, the amount of information gathered so far in this field may be sufficient to offer some hope of a successful approach. Our task is to attempt to dissect those signaling pathways that are critical for cell transformation and those that are redundant or irrelevant. In time, the key elements in cell transformation will be unmasked and become excellent targets for drug design. In fact, some of the drugs now available against a few of the known effectors for Ras and Rho show evidence of sensitive and selective steps for cancer prevention. This is the case for ROCK, Rafl, FTIs, PLD, and ChoK, which still need some fine-tuning in the design. Some other drugs, while not currently rendering satisfactory results, will be extremely useful in achieving this goal. Finally, once we focus our attention to those relevant identified targets, the eftbrt to design successful drugs will be easier. BEFERENCES 1. 2. 3. 4.
M. Malumbresand A. Pellieer,Rev. Oncologia 1, 66-76 (1999). L. Van Aelstand CrislynD'Souza-Sehorey,Genes Dev. 11, 2295-2322 (1997). R. Hernandez-Alcoceba,L. del Peso, and l" C. Lacal,Cell Mol. Life Sci. 57, 65-76 (2000). B. Houssa,J. de Widt, O. Kranenburg,W. H. Moolenaar,and W. J. van Blitterswijk,J. Biol. Chem. 274, 6820-6822 (1999). 5. P. Burbelo,D. Dreschel,and A. Hall,J. Biol. Chem. 270, 29071-29074 (1995).
224
SALVADOR AZNAR AND JUAN CARLOS LACAL
6. M. Symons, J. Derry, B. Karlak, S. Jiang, v. Lemallieu, E McCormick, U. Francke, and A. Abo, Cell (Cambridge, Mass. ) 84, 723-734 (1996). 7. J. Derry, J. Ochs, and U. Francke, Cell (Cambridge, Mass. ) 78, 635-644 (1993) Erratum: Cell (Cambridge, Mass. ) 79 (1994). 8. O. Rivero-Lezcano, A. Marcilla, J. Sameshima, and K. Robbins, Mol. Cell Biol. 15, 5725-5731
(1995). 9. C. Mullen, K. Anderson, and M. Blaese, Blood 82, 2961-2966 (1993). 10. H. Ochs, S. Slitcher, L. Harker, B. Von, R. Clark, and R. Wedgwood, Blood 55, 243-252 (1980). 11. E Aspenstrom, U. Lindberg, and A. Hall, Cu~ Biol. 6, 70-75 (1996). 12. S. Linder, D. Nelson, M. Weiss, and M. Aepfelbacher, Proc. Natl. Acad. Sei. U.S.A. 96, 9648-9653 (1999). 13. H. Mild, S. Suetsugu, and T. Takenawa, EMBOJ. 17, 6932-6941 (1998). 14. L. Van Aelst, T. Joneson, and D. Bar-Sagi, EMBOJ. 15, 3778--3786 (1996). 15. C. D'Souza-Shocrey, R. L. Boshans, M. McDonough, E Statll, and L. Van Aelst, EMBO J. 16, 5445-5454 (1997). 16. H. Mild, M. Fukuda, E. Nishida, and T. Takenawa, J. Biol. Chem. 274, 27605-27609 (1999). 17. S. Kuroda, M. Fukata, M. Nakagawa, and K. Kaibuchi, Biochem. Biophys. Res. Commun. 262, 1-6 (1999). 18. S. Kuroda, M. Fukata, M. Nakagawa, K. Fujii, T. Nakamura, T. Ookubo, I. Izawa, T. Nagase, N. Nomura, H. Tani, I. Shoji, Y. Matsuura, S. Yonehara, and K. Kaibuchi, Science 281,832-835 (1998). 19. L. Machesky, Curt Biol. 8, R202-R205 (1998). 20. M. Fukata, S. Km'oda, M. Nakagawa, A. Kawajiri, N. Itoh, I. Shoji, Y. Matsuura, S. Yonehara, A. Kikuchi, and K. Kaibuchi, J. Biol. Chem. 274, 26044-26050 (1999). 21. T. Fleming and M. Johnson, Annu. Bey. Cell Biol. 4, 459-485 (1988). 22. A. Ridley, P. Comoglio, and A. Hall, Mol. Cell Biol. 15, 1110-1122 (1995). 23. K. Takaishi, T. Sasald, I". Kameyama, S. Tsuldta, and Y. Takai, Oncogene 11, 39-48 (1995). 24. K. Tal(aishi, T. Sasald, M. Kato, W. Yamoehi, S. Kuroda, T. Nakamura, M. Takeichi, and M. Takai, Oneogene 9, 273-279 (1994). 25. M. Takeiehi, Cu~. Opin. Cell Biol. 5, 806-811 (1993). 26. K. Vlcminckx, L. Vakaet, M. Marcel, W. Fiers, and F. Van Roy, Ceil (Ca~rd~ridge, Mass.) 66, 107-119 (1991). 27. W. Birchmeier and J. Bchrens, Biochim. Biophys. Acta 1198, 11-26 (1994). 28. V. Brag& L. Machesky, A. Hall, and N. Hotchin, J. Cell Biol. 137, 1421-1431 (1997). 29. K. Takaishi, T. Sasaki, H. Kotani, H. Nishioka, and Y. Takai, J. Cell Biol. 139, 1047-1059 (1997). 30. P. Hordijk, J. Ten Klooser, R. Van der Kammen, F. Miehiels, L. Oomen, and]. Collard, Science 278, 1464-1466 (1997). 31. E. Sanders, S. van Delft, J. ten Klooster, T. Reid, R. van der Kammen, F. Michiels, and J. Collard,.]. Cell Biol. 143, 1385-1398 (1998). 32. L. del Peso, R. Hernandez-Alcoceba, N. Embade, A. Carnero, E Esteve, C. Pajc, and J. C. Lacal, Oncogene 15, 3047-3057 (1997). 33. P. Keely, J. Westwick, I. Whitehaed, C. Der, and L. Parise, Nature (London) 390, 632-635
(1997). 34. E. Evers, G. Zondag, A. Malliri, L. Price, J. ten Klooster, R. Van der Kammen, and J. Collard, Eur. J. Cancer 36, 1269-1274 (2000). 35. E. Sander, J. ten Klooster, S. van der Kammen, and j. Collard, J. Cell Biol. 147, 1009-1022
(1999).
NEW TARGETS FOR ANTICANCER DRUG DESIGN
225
36. E. Manser, T. Leung, H. Salihuddin, L. Tan, and L. Lim, Nature (Lonckm) 363, 364-367
(1993). 37. W. Yang and R. A. Cerione, J. Biol. Chem. 272, 24819-'24824 (1997). 38. W. Yang, Q. Lin, J. L. Guan, and R. A. Cerione, J. Biol. Chem. 274, 8524-8530 (1999). 39. T. Sethi, R. C. Rintoul, S. M. Moore, A. C. MacKinnon, D. S~ter, C. Choo, E. R. Chilvers, I. Dransfield, S. C. Donnelly, R. Strieter, and C. Haslett, Nat. Med. 5, 662-668 (1999). 40. G. Fritz, I. Just, and B. Kaina, Int. J. Cancer 81,682-687 (1999). 41. M. S. A. Nur-E-Kamal, J. M. Kamal, M. M. Qureshi, and H. Maruta, Oncogene 18, 7787-7793 (1999). 42. D. Derossi, G. Chassaing, and A. Prochiantz, Trends" Cell Biol. 8, 84-87 (1998). 43. M. Teo, E. Manser, and L. Lim,J. Biol. Chem. 270, 26690-26697 (1995). 44. E. Manser, T, Leung, H. Salihuddin, Z. Zhao, and L. Lim, Nature (London) 367, 40-46 (1994). 45. J. Brown, L. Stowers, M. Baer, J. Trejo, S. Coughlin, and J. Chant, Curt. Biol. 6, 598-605 (1996). 46. E. Manser, C. Chong, Z. Zhao, "12 Leung, G. Michael, C. Hall, and L. Lira, J. Biol. Chem. 270, 25070-25078 (1995). 47. S. Bagrodia, S. Taylor, C. Creasy, J. Chernoff, and R. Cerione, ]. Biol. Chem. 270, 2273122737 (1995). 48. G. Bollag Martin, F. McCormick, and A. Abo, EMBOJ. 14, 1970-1978 (1995). 49. A. Abo, J. Qu, M. S. Cammarano, C. Dan, A. Fritsch, V. Baud, B. Belisle, and A. Minden, EMBOJ. 17, 6527-6540 (1998). 50. L. Lim, E. Manser, T. Leung, and C. Hall, Eur. J. Biochem. 242, 171-185 (1996). 51. N. Lamarche, L. Tapon, P. Stowers, P. Burbelo, T. Aspenstrom, T. Bridges, J. Chant, and A. Hall, Cell (Cambridge, Mass. ) 87, 519-529 (1996). 52. T. Joneson, M. McDonough, D. Bar-Sagi, and L. Van-Aelst, Science 274, 1374-1376 (1996). 53. D. B. Foster, L. Hua Shen, J. Kelly, P. Thibault, J. van Eyk, and A. Mak, J. Biol. Chem. 275, 1959-1965 (2000). 54. S. Holly and K. Blumer,]. Cell Biol. 147, 845-856 (1999). 55. R. H. Daniels, P. S. Hall, and G. M. Bokoch, EMBOJ. 17, 754-764 (1999). 56. K. Allen, J. G. Gleeson, S. Bagrodia, M. W. Partingtin, J. C. MacMillan, R. A. Cerione, J. c. Mulley, and C. A. Walsh, Nat. Genet. 20, 25~30 (1998). 57. W. Lu and B. J. Mayer, Oncogene 18, 797-806 (1999). 58. S. Bagrodia, S. J. Taylor, K. A, Jordon, L. Van Aelst, and R. Cerione, J. Biol. Chem. 273, 23633-23636 (1998). 59. E. Manser, T.-H. Loo, C.-G. Koh, Z.-S. Zhao, X.-Q. Chen, L. Tan, I. Tan, T. Leung, and L. Lira, Mol. Cell 1, 183-192 (1998). 60. W. K. Oh, J. C. Yoo, D. Jo, Y. H. Song, M. G. Kim, and D. Park, Biochem. Biophys. Res. Commun. 235, 794-798 (1997). 61. Y. Tang, J. Yu, and J. Field, Mol. Cell Biol. 19, 1881-1891 (1999). 62. S.-I. Osada, M. Izawa, T. Koyama, S.-I. Hirai, and S. Ohno, FEBS Lett. 404, 227-233 (1997). 63. J. P. Mira, V. Bernard, J. Groffen, L. C. Sanders, and U. G. Knaus, Proc. Natl. Acad. Sci. U.S.A. 97, 185-189 (2000). 64. Y. Tang, Z. Chen, D. Ambrose, J. Liu, J. B. Gibbs, J. Chernoff, and J. Field, Mol. Cell Biol. 17, 4454-4464 (1998). 65. A. J. King, H. Sun, B. Diaz, D. Barnard, W. Miao, S. Bagrodia, and M. Marshall, Nature (London) 396, 180-183 (1998). 66. J. Westwick, Q. Lambert, G. Clark, M. Symons, L. Van Aelst, R. Pestell, and C. Der, Mol. Cell Biol. 17, 1324-1335 (1997).
226
SALVADOR AZNAR AND JUAN CARLOS LACAL
67. Y. Tang, S. Marwaha, J. L. Rutkowski, G. I. Tenekoon, E C. Philips, and J. Field, Proc. Natl. Acad. Sci. U.S.A. 95, 5138--5144 (1998). 68. S. Bagrodia, R. Derijard, R. Davis, and R. Cerione, J. Biol. Chem. 270, 27995-27998 (1995). 69. J. Frost, S. Xu, M. Hutchison, S. Marcus, and M. Cobb, Mol. Cell Biol. 16, 3707-3713 (1996). 70. S. Zhang, J. Han, M. Sells, U. Chernoff, U. Knaus, R. Ulevitch, and G. Bokoch,]. BioL Chem. 270, 23934-23936 (1995). 71. H. Teramoto, O. Coso, H. Miyata, T. Igishi, T. Miki, and S. Gutkind, J. Biol. Chem. 271, 27225-27228 (1996). 72. H. Teramoto, P. Crespo, O. A. Coso, T. Igishi, N. Xu, and J. s. Gutkind, J. Biol. Chem. 271, 25731-25734 (1996). 73. W. Lu and B. J. Mayer, Oncogene 18, 797-806 (1999). 74. R. Perona, S. Montaner, L. Saniger, I. Sanchez-Perez, R. Bravo, and J. c. Lacal, Genes Dev. 11,463-475 (1997). 75. D. J. Sulciner, K. Irani, Z. X. Yu, V. J. Ferrans, C. P. Goldschmidt, and T. Finkel, Mol. Cell Biol. 16, 7115-7121 (1996). 76. T. S. Finco, J. K. Westwick, J. L. Norris, A. A. Beg, C. J. Der, and A. S. Baldwin, Jr., J. Biol. Chem. 272, 24113-24116 (1997). 77. M. Arsuna, E Mercurio, A. L. Oliver, S. S. Thorgeirsson, and G. E. Sonenshein, Mol. Cell Biol. 20, 5381-5391 (2000). 78 B. Banmann, C. K. Weber, J. Troppmai, S. Wbiteside, A. Israel, U. R. Rapp, and T. Wiryb, Proc. Natl. Acad. Sci. U.S.A. 97, 4615-4620 (2000). 79. H. Jo, R. Zhang, H. Zhang, T. A. McKinsey, J. Shao, R. D. Beauchamp, C. W. Ballard, and P. Liang, Oncogene 19, 841-849 (2000). 80. J.Y. Reuther, G. W. Reuther, D. Cortez, A. M. Pendergast, and A. S. Baldwin, Jr., Genes Dev. 12, 968-981 (1998). 81. I. P. Whitehead, Q. T. Lambert, J. A. Glaven, K. Abe, K. L. Rossman, G. M. Mahon, J. M. Trzaskos, R. Kay, S. L. Campbell, and C. J. Der, Mol. Cell Biol. 19, 7759-7770 (1999). 82. N. Tapon, K.-I. Nagata, N. Lamarche, and A. Ha/l, EMBOJ. 17, 1395-1404 (1998). 83. P. Gerwins, J. L. Blank, and G, L. Johnson, J. Biol. Chem. 272, 8288~295 (1997). 84. L. Tibbles, Y. Ling, E Kiefer, J. Chan, N. Iscove, J. R. Woodget, and N. J. Lassam, EMBOJ. 15, 7026-7035 (1996). 85. S. P. Henner, T. G. Hofman, A. Ushmorov, O. Dienz, I. Wing-Lan Leung, N. Lassam, C, Sheidereit, W. Droge, and M. L. Schmitz, Mol. Cell Biol. 20, 2556-2568 (2000). 86. S. Huang, L. New, Z. Pan, J. Han, and G. R. Nemerow, J. Biol. Chem. 275, 12266-12272 (2000). 87. L. Montero and Y. Nagamine, Cancer Res. 59, 5286-5293 (1999). 88. S. M. Muller, E. Okan, and P. Jones, Biochem. Biophys. Res. Commun. 270, 89~898 (2000). 89. R. Perona, P. Esteve, B. Jimenez, R. P. Ballestero, S. Ramon y Caja/, and J. c. Laeal, Oncogene 8, 1285-1292 (1993). 90. L. del Peso, R. Hernandez-Alcoceba, N. Embade, A. Carnero, P. Esteve, C. Paje, and J. c. Lacal, Oncogene 15, 3047-3057 (1997). 91. T. Ishizaki, M. Maekawa, K. Fujisawa, K. Okawa, A. Iwamatsu, A. Fujita, N. Watanabe, Y. Saito, A. Kakizuka, N. Morii, and S. Narumiya, EMBOJ. 15, 1885-1893 (1996). 92. T. Leung, E. Manser, L. Tan, and L, Lim, J. Biol. Chem. 270, 29051-29054 (1995). 93. T. Leung, X.-Q. Chen, E. Manser, and L. Lim, Mol. Cell Biol. 16, 5313--5327 (1996). 94. O. Nakagawa, K. Fujisawa, T. Ishizaki, Y. Saito, K. Nakao, and S. Narumiya, FEBS Lett. 392, 189-193 (1996). 95. T. Matsui, M. Amano, T. Yamamoto, K. Chihara, M. Nakafuku, M. Ito, T. Nakano, K. Okawa, A. Iwamatsu, and K. Kaibuchi, EMBOJ. 15, 2208-2216 (1996).
NEW TARGETS FOR ANTICANCER DRUG DESIGN
9,27
96. T. Ishizaki, M. Naito, K. Fujisawa, M. Maekawa, N. Watanabe, Y. Saito, and S. Narumiya, FEBS Lett. 404, 118-124 (1997). 97. M. Amano, K. Chihara, K. Kimura, Y. Fukata, N. Nakamura, Y. Mastuura, and K. Kaibuchi, Science 275, 1308-1311 (1997). 98. M. Hirose, T. Ishizaki, N. Watanabe, M. Uehata, O. Kranenburg, W. H. Moolenaar, F. Matsumnra, M. Maekawa, H. Bito, and S. Narumiya, J. Ceil Biol. 141, 1625-1636 (1998). 99. E. Sahai, T. Ishizaki, S. Narumiya, and R. Treisman, Cur~ Biol. 9, 136-145 (1999). 100. K. Itoh, K. Yoshioka, H. Akedo, M. Uehata, T. Ishizaki, and S. Narumiya, Nat. Med. 5, 221-225
(1999). 101. R. G. Qiu, j. Chen, F. McCormick, and M. Symons, Proc. Natl. Acad. Sci. U.S.A. 92, 11781-
11785 (1995). 102. T. Ishizaki, M. Uehata, I. Tamechika, J. Keel, K. Nonomura, M. Maekawa, and S. Narumiya, Mol. Pharmacol. 57, 976-983 (2000). 103. M. Uehata, T. Ishizaki, H. Satoh, T. Ono, T. Kawahara, T. Morishita, H. Tamakawa, K. Yamagami, J. Inui, M. Maekawa, and S. Narumiya, Nature (London) 389, 990-994 (1997). 104. E. Stdlai, A. S. Alberts, and R. Treisman, EMBOJ. 17, 1350-1361 (1998). 105. M. Zohar, H. Teramoto, B. Z. Katz, K. M. Yamada, and J. S. Gutkind, Oncogene 17, 991-998
(1998). 106. J. H. Exton, Ann. N.Y. Acad. Sci. 905, 61-68 (2000). 107. W. C. Colley, T.-C. Sung, R. Roll, J. Jenco, s. M. Hammond, Y. Altshuller, D. Bar-Sagi, A. J. Morris, and M. A. Frohman, Curt Biol. 7, 191-201 (1997). 108. K. Ohguchi, Y. Banno, S. Nakashima, and Y. Nozawa, Biochem. Biophys. Res. Commun. 211,
306-311 (1995). 109. M. Schmidt, U. Rumenapp, C. Bienek, J. Keller, C. Von Eichel-Streiber, and K. H. Jakobs, J. Biol. Chem. 271, 2422-2426 (1996). 110. K. C. Malcohn, C. M. Elliott, and J. H. Exton, J. Biol. Chem. 271, 13135-13139 (1996). 111. J. A. Hess, A. H. Ross, R. G. Qiu, M. Symons, and J. H. Exton,J. Biol. Chem. 272, 1615-1620
(1997). 112. S. J. Walker, W. J. Wu, R. A. Cerione, and H. A. Brown, J. Biol. Chem. 275, 15665-15668
(2000). 113. J. S. Han, H. C. Kim, J. K. Chung, H. S. Kang, J. Donaldson, andJ. K. Koh, Biochem. Mol. Biol. Int. 45, 1089-1103 (1998). 114. K. C. Malcolm, A. H. Ross, R. G. Qiu, M. Symons, and J. H. Extort, J. Biol. Chem. 269,
25951-25954 (1994). 115. A. Carnero, A. Cuadrado, L. del Peso, and J. c. Lacal, Oncogene 9, 1387-1395 (1994). 116. I. Guilleman and J. H. Extort, Biochim. Biophys. Acta 1405, 161-170 (1998). 117. M. Yamazaki, Y. Zhang, H. Watanabe, T. Yokozeki, S. Ohno, K. Kaibuchi, H. Shibata, M. Mukai, Y. Ono, M. A. Frohman, and Y. Kanaho, J. Biol. Chem. 274, 6035-6038 (1999). 118. C. D. Bae, D. S. Min, I. N. Fleming, andJ. H. Exton,J. Biol. Chem. 273,11596-11604 (1998). 119. H. Kuribara, K. Tago, T. Yokozeki, T. Sasaki, Y. Takai, N. Morii, S. Narumiya, T. Katada, and Y. Kanaho, J. Biol. Chem. 270, 25667-25671 (1995). 120. L. del Peso, L. Lucas, P. Esteve, and J. c. Lacal, Biochem. J. 322, 519-528 (1997). 121. L. Lucas, L. del Peso, P. Rodriguez, V. Penalva, and J. C. Lacal, Oncogene 19,431-437 (2000). 122. J. C. Lacal, FEBS Lett. 410, 73-77 (1997). 123. K. Shome, Y. Nie, and G. Romero, J. Biol. Chem. 273, 30836-30841 (1998). 124. M. N. Hodgkin, M. R. Masson, D. Powner, K. M. Saqib, C. P. Ponting, and M. J. Wakelam, Curt Biol. 10, 43-46 (2000). 125. M. Schmidt, M. Voss, P. A. Oude Weernink, J. Wetzel, M. Amano, K. Kaibuchi, and K. H. Jakobs, J. Biol. Chem. 274, 14648-14654 (1999).
228
SALVADOR AZNAR AND JUAN CARLOS LACAL
126. L. del Peso, R. Hernandez, P. Esteve, and J. C. Lacal, J. Cell Biol. 61,599-608 (1996). 127. Z. Lu, T. Joseph, T. Sukezane, P. Frankel, M. Zhong, S. Bychenok, L. Xu, L. A. Feig, and D. A. Foster, Mol. Cell Biol. 20, 462-467 (2000). 128. A. Hornia, Z. Lu, T. Sukezane, M. Zhong, T. Joseph, P. Frankel, and D. A. Foster, Mol. Cell Biol. 19, 7672-7680 (1999). 129. P. Frankel, M. Ramos, J. Flom, S. Bychenok, T. Joseph, E. Kerkhoff, U. R. Rapp, L. A. Feig, and D. A. Foster, Biochem. Biophys. Res. Commun. 255, 502-507 (1999). 130. R. Reich, M. Blumenthal, and M. Liscovitch, Clin. Exp. Metastasis 13, 134-140 (1995). 131. B. T. Williger, R. Reich, M. Neeman, T. Bercovici, and M. Liscovitch, J. Biol. Chem. 270, 29656-29659 (1995). 132. J. A. Aguirre-Ghiso, P. Frankel, E. F. Farias, Z. Lu, H. Jiang, A. Olsen, L. A. Feig, E. B. de Kier Joffe, and D. A. Foster, Oncogene 18, 4718-4725 (1999). 133. "v: P. Terranova, J. E. Williams, L. A. Liotta, and G. R. Martin, Science 226, 982-985 (1984). 134. S. M. Muller, E. Okan, and P. Jones, Biochem. Biophys. Res. Commun. 270, 892-898 (2000). 135. S. Bourdoulous, G. Orend, D. A. MacKenna, R. Pasqualini, and E. Ruoslahti, J. Cell Biol. 143, 267-276 (1998). 136. J. j. Baldasare, M. B. Jarpe, L. Alferes, and D. M. Raben, J. Biol. Chem. 272, 4911-4914 (1997). 137. F. Mollinedo, C. Gajate, and I. Flores, J. Immunol. 153, 2457-2469 (1994). 138. J.-H. Chang, J. c. Pratt, S. Sawasdikosol, R. Kapeller, and S. Burakoff, Mol. Cell Biol. 18, 4986-4993 (1998). 139. J. Dang, D. Boyd, H. Wang, H. Allgayer, W. E Doe, and Y. Wang, Eur. J. Biochem. 264, 92-99 (1999). 140. Y. Wang, j. Dang, H. Wang, H. Allgayer, G. A. Murrell, and D. Boyd, Eur. J. Biochem. 267, 3248-3254 (2000). 141. G. Watanabe, Y. Saito, P. Maudale, T. Ishizaki, N. Fujisawa, N. Morii, H. Mukai, Y. Ono, A. Kakizuka, and S. Narumiya, Science 271,645-648 (1996). 142. L. A. Quilliam, Q. T. Lambert, Y. L. Mickelson, J. K. Westwick, A. B. Sparks, B. K. Kay, N. A. Jenkins, D. J. Gilbert, N. G. Copeland, and C. J. Der, J. Biol. Chem. 271, 28772-28776 (1996). 143. V. Sylvie and J. Settleman, Mol. Cell Biol. 17, 2247-2256 (1997). 144. K. Oishi, H. Mukai, H. Shibata, M. Takahashi, and Y. Ona, Biochem. Biophys. Res. Commun. 261, 808-814 (1999). 145. Y. Lu and J. Settleman, Genes Dev. 13, 1168-1180 (1999). 146. L. Q. Dong, L. R. Landa, M. J. Wick, L. Zhu, H. Mukai, Y. Ono, and F. Liu, Proc. Natl. Acad. Sci. U.S.A. 97, 508~5094 (2000). 147. H. Mukai, M. Toshimori, H. Shibata, H. Takanaga, M. Kitagawa, M. Miyahara, M. Shimakawa, and Y. Ono, J. Biol. Chem. 272, 474~4746 (1997). 148. H. Mukai, M. Toshimori, H. Shibata, M. Kitagawa, M. Shimakawa, M. Miyahara, H. Sunakawa, and Y. Ono, J. Biol. Chem. 271, 9816-9822 (1996). 149. K. Matsuzawa, H. Kosako, N. Inagaki, H. Shibata, H. Mukai, Y. Ono, M. Amano, K. Kaibuchi, Y. Matsuura, I. Azuma, and M. Ingaki, Biochem. Biophys. Res. Commun. 234,621-625 (1997). 150. S. R. J. Salton, C. Volonte, and G. D'Arcangelo, FEBS Lett. 360, 106-110 (1995). 151. H. Mukai, M. Miyahara, H. Sunakawa, H. Shibata, M. Toshimori, M. Kitagawa, M. Shimakawa, H. Takanaga, andY. Ono, Proc. Natl. Acad. Sci. U.S.A. 93, 10195-10199 (1996). 152. M. Kitagawa, H. Mukai, M. Takahashi, and Y. Ono, Biochem. Biophys. Res. Commun. 252, 561-565 (1998). 153. W. Sun, S. Vincent, J. Settleman, and G. L. Johnson@ Biol. Chem. 275, 24421-244288 (2000). 154. M. R. Morissette, V. P. Sah, C. C. Glembotski, and J. H. Brown, Am. J. Physiol. Heart Circ. Physiol. 278, H1769-1774 (2000).
NEW TARGETS FOR ANTICANCER DRUG DESIGN
29~9
155. B. C. Kim, C. J. Lira, and J. H. Kim, FEBS Lett. 415, 325-328 (1997). 156. H. Mukai, M. Kitagawa, H. Shibata, H. Takanaga, K. Mori, M. Shimakawa, M. Miyahara, K. Hirao, and Y. Ono, Biochem. Biophys. Res. Commun. 204, 348-356 (1994). 157. M. Chihara, M. Amano, N. Nakamura, T. Yano, M. Shibata, T. Tokui, H. Ichikama, R. Ikebe, M. Ikebe, and K. Kaibuchi, J. Biol. Chem. 272, 25121-25127 (1997). 158. T. Tominaga, E. Sahai, P. Chardin, F. McCormick, S. A. Courtneidge, andA. S. Alberts, Mol. Cell 5, 13-25 (2000). 159. C. S. Hill, J. Wynne, and R. Treisman, Cell (Cambridge, Mass.) 81, 1159-1170 (1995). 160. S. Montaner, R. Perona, L. Saniger, and J. C. Lacal, J. Biol. Chem. 274, 8506-8515 (1999). 161. A. S. Alberts, O. Geneste, and R. Treisman, Cell (Cambridge, Mass.) 92, 475-487 (1998). 162. P. F. Lebowitz, W. Du, and G. C. Prendergast, J. Biol. Chem. 272, 16093-16095 (1997). 163. T. Rudel and G. M. Bokoch, Science 276, 1571-1574 (1997). 164. N. Lee, H. MaeDonald, C. Reinhardt, R. Halenbeek, A. Roulston, T. Shi, and L. T. Williains, Proc. Natl. Acad. Sci. U.S.A. 94, 13642-13647 (1997). 165. M. Takahashi, H. Mukai, M. Toshimori, M. Miyamoto, and Y. Ono, Proc. Natl. Acad. Sci. U.S.A. 95, 11566-11571 (1998). 166. H. Koh, K. H. Lee, D. Kim, S. Kim, J. W. Kim, and J. Chung,]. Biol. Chem. 275, 34454-34458 (2000). 167. G. C. Prendergast, Curt Opin. Cell Biol. 12, 166-173 (2000). 168. J. L. Bos, Cancer Res. 49, 46824689 (1989). 169. C. J. Tabin, S. M. Bradley, C. I. Bargmenn, R. A. Weinberg, A. G. Papageorge, E, M. Scolnick, D. Dhar, R. Lowy, and E. H. Chang, Nature (London) 300, 143-147 (1982). 170. E. P. Reddy, R. K. Reynolds, E. Santos, and M. Barbacid, Nature (London) 300, 149-152 (1982). 171. S. A. Moodie, B. M. Willmnsen, M. j. Weber, and A. Wolfinan, Science 260, 1658-1661 (1993). 172. P. H. Warne, P. Rodiriguez-Vieiana, and J. Downward, Nature (London) 364,352-355 (1993). 173. X. F. Zhang, J. Settleman, J. M. Kyriakis, E. Takeuehi-Suzuki, S. J. Elledge, M. S. Marshall, J. T. Bruder, U. R. Rapp, and J. Avruch, Nature (l~mdon) 364, 308-313 (1993). 174. D, Stokoe, S. G. McDonald, K. Cadwallader, M. Symmons, and J. F. Hancock, Science 264, 1463-1467 (1994). 17.5. J. M. Kyriakis, H. App, X. Zhang, D. I. Banerjee, U. B. Brautigan, U. R. Rapp, and J. Avrueh, Nature (London) 358, 417-421 (1992). 176. A. Khokhlatchev, V. B. Canagarajah, J. Wilsbacher, M. Robinsin, M. Atkinson, E. Goldsmith, and M. H. Cobb, Cell (Cambridge, Mass.) 93, 605~615 (1998). 177. R. Treisman, Curn Opin. Cell Biol. 8, 205-215 (1996). 178. R. Marais, Y. Light, H. F. Paterson, C. S. Mason, and C. J. Marshall, J. Biol. Chem. 272, 4378-4383 (1997). 179. J. Troppmair, J. Hartkamp, and U. R. Rapp, Oncogene 17, 685~90 (1998). 180. A. Kalkuhl, J. Troppmair, A. Buchmann, S. Sttinchcombe, C. L. Buenemann, U. R. Rapp, K. Kaestner, and M. Schwarz, Hepatology 27, 1081-1088 (1998). 181. C~ Hagemann and U, R. Rapp~ Exp. Cell Res. 253, 34-46 (1999). 182. H. G. Wang, T. Miyashita, S. Takayama, T. Sato, T. Torigoe, S. Krajewski, S. Tanaka, L. Hovey, J. Troppmair, and U. R. Rapp, Oncogene 9, 2751-2756 (1994). 183. C. K. Weber, J. R. Slupsky, C. Hermann, M. Schuler, U. R. Rapp, and C. Block, Oncogene 19, 169-176 (2000). 184. M. Hamilton and A. Wolfman, Oncogene 16, 1417-1428 (1998). 185. E. Kerkhoffand U. R. Rapp, Mol. Cell Biol. 17, 2576-2586 (1997). 186. B. Baumann, C. K. Weber, J. Troppmair, S. Whiteside, A. Israel, U. R. Rapp, and T. Wirth, Proc. Natl. Acad. Sci. U.S.A. 97, 4615-4620 (2000).
230
SALVADOR AZNAR AND JUAN CARLOS LACAL
187. E. Kerkhoffand U. R. Rapp, Mol. Cell Biol. 17, 2576-2586 (1997). 188. M. Friedman, A. Tikoo, M. Varga, A. Murphy, M. S. Nur-e-Kamal, and H. Maruta, J. Biol. Chem. 269, 30105-30108 (1994). 189. L. M. Gangarosa, N. Sizernore, R. Graves-Deal, S. M. Oldham, C. J. Der, and R. J. Coffey, j. Biol. Chem. 272, 18926-18931 (1997). 190. S. M. Oldham, A. D. Cox, E. R. Reynolds, N. S. Sizemore, R. J. Coffey, Jr., and C. J. Der, Oncogene 16, 2565--2573 (1998). 191. E. Kerkhoff, L. M. Fedrorov, R. Siefken, A, O. Walter, T. Papadopoulos, and U. R. Rapp, Cell Growth Differ 11, 185-190 (2000). 192. A. Kalkuhl, J. Troppmair, A. Buchmann, S. Sttinchcombe, C. L. Buenemann, U. R. Rapp, K. Kaestner, and M. Schwarz, Hepatology 27, 1081-1088 (1998). 193. H. G. Wang, T. Miyashita, S. Takayama, T. Sato, T. Torigoe, S. Krajewski, S. Tanaka, L Hovey, J. Troppmair, and U. R. Rapp, Oncogene 9, 2751-2756 (1994). 194. H. G. Wang, U. R, Rapp, andJ. C. Reed, Cell (Cambridge, Mass.) 87, 629~638 (1996). 195. H. G. Wang, S. Takayama, U. R. Rapp, and J. C. Reed, Proc. Natl. Acad. Sci. U.S.A. 93, 7063-7068 (1996), 196. Q. C. Lau, S. Brusselbach, and R. Muller, Oncogene 16, 1899-1902 (1998). 197. B. Baumann, C. K. Weber, J. Troppmair, S. Whiteside, A. Israel, U. R. Rapp, and T. Wirth, Proc. Natl. Aead. Sci. U.S.A. 97, 4615-4620 (2000). 198. F. Aklilu, J. Gladu, D. Goltzman, and S. A. Rabbani, Cancer Res. 60, 1753-1760 (2000). 199. J. Zhu, D. Woods, M. McMahon, and J. M. Bishop, Genes Dev. 12, 2997-3007 (1998). 200. E. Kerldaoff and U. R. Rapp, Oncogene 17, 1457-1462 (1998). 201. P. Rodriguez-Viciana, P. H. Warne, B. Dhand, I. Vanhaesebroeek, I. Gout, M. J. Fry, M. D. Waterfield, and J. Downward, Nature (London) 370, 527-.521 (1994). 202. P. Rodriguez-Viciana, P. H. Warne, M. D. Waterfield, andJ. Downward, EMBOJ. 15, 22422251 (1996). 203. D. H. Van Weering, J. de Rooij, B. Marte, J. Downward, J. L. Bos, and B, M. Burgering, Mol. Cell Biol. 18, 1802-1811 (1998). 204. P. J. Coffer, j. Jin, and J. R. Woodgett, Biochem. J. 335, 1-13 (1998). 205. D. A. Cross, D. R. Alessi, P. Chen, M. Andjelkouch, and B. D. Hemmings, Nature (London) 378, 785-789 (1995). 206. Q. Wang, R. Somwar, P. J. Bilan, Z. Liu, J. Jin, J. R. Woodgett, and A. Klip, Mol. Cell Biol. 19, 4008-4018 (1999). 207. M. Shaw and P. Chen, FEBS Lett. 461, 120-124 (1999). 208. M. Delcommenne, C. Tan, ~ Gray, L. Rue, J. Woodgett, and S. Dethar, Proc. Natl. Acad. Sci. U.S.A. 95, 11511-11216 (1998). 209. J. A. Romashkova and S. S. Makarov, Nature (London) 401, 86-90 (1999). 210. P. Sabbatini and E McCormick, J. Biol. Chem. 274, 24263-24269 (1999). 211. A. Khwaja, Nature (London) 401, 33-34 (1999). 212. J. Downward, Science 279, 673-674 (1998). 213. P. T. Hawkins, A. Eguinoa, R. G. Qui, D. Stokoe, T. Cooke, R. Waiters, S. Wennstrom, L. Claesson-Welsch, T. Evans, M. Symons, and L. Stephens, Curt. Biol. 5, 393~03 (1995). 214. A. S. Nimnual, B. A. Yatsula, and D. Bar-Sagi, Science 279, 260-563 (1998). 215. J, Han, K. Luby-Phelps, B. Das, X. Shu, Y. Xia, R. D. Mosteller, U. M. Krishna, J. R. Falck~ M. A. White, and D. Brock, Science 279, 558--560 (1998). 216. T. Skorski, A. Bellacosa, M. Nieborowska-Skorska, M. Majewski, R. Martinez, J. K. Choi, T. Trotta, P. Wlodarski, D. Perrotti, T. O. Chan, M. A. Wasik, P. N. Tsichlis, and B. Calabretta, EMBOJ. 16, 6151-6161 (1997). 217. D. K. Moscatello, M. Holgado-Madruga, D. R. Emlet, R. B. Montgomery, and A. J. Wong, J. Biol. Chem. 273, 200-206 (1998).
NEW TARGETS FOR ANTICANCER DRUG DESIGN
231
218, C. Jimenez, D. R. Jones, E Rodriguez-Viciana, A. Gonzalo-Garcia, E. Leonardo, S. Wennstrom, C. Von Kobble, J. L. Torah, L. Borlado, V. Calvo, S. G. Copin, J. P. Albar, M. L. Gaspar, E. Diez, M. A. Marcos, J. Downward, C. A. Martinez, L Merida, and A. C. Carrera, EMBOJ. 17, 743-753 (1998). 219. L. Shayesteh, Y. Lu, W. L. Kuo, R. Baldoechi, T. Collins, D. Pinkel, B. Powell, G. B. Mills, and J. w. Gray, Nat. Genet. 21, 99-102 (1999). 220. Y. Y. Ma, S. J. Wei, Y. C. Lin, J. C. Lung, T, C. Chang, J. Whang-Peng, J. M. Liu, D. M. Yang, W. K. Yang, and C. Y. Shen, Oncogene 19, 2739-2744 (2000). 221. L. R. Borlado, C. Redondo, B. Alvarez, C. Jimenez, L. M. Criado, J. Flores, M. A. Marcos, C. A. Martinez, and A. C. Carrera, FASEBJ. 14, 895-903 (2000). 222. S. M. Moore, R. C. Rinoul, T. R. Walker, E. R. Chilvers, C. Haslett, and T. Sethi, Cancer Res. 58, 5239~5247 (1998). 223. L. Hu, O. Zaloudek, G. B. Mills, J. Gray, and R. B. Jaffe, Clin. CancerRes. 6, 880-886 (2000). 224. A. Bardelli, M. L. Basile, E. Audero, S. Giordano, S. Wennstromm, S. Menard, P. M. Comogolio, and C. Ponzetto, Oncogene 18, 1139-1146 (1999). 225. B. H. Jiang, J. z. Zheng, M. Aoki, and P. K. Vogt, Proc. Natl. Acad. Sci. U.S.A. 97, 1749-1753 (2000). 226. M. Spaargaren and J. R. Bisehiff, Proc. Natl. Acad. Sci. U.S.A. 91, 12609-12613 (1994). 227. E Hofer, S. Fields, C. Schneider, and G. S. Martin, Proc. Natl. Acad. Sci. U.S.A. 91, 1108911093 (1994). 228. A. Kikuehi, S. D. Demo, Z. H. Ye, Y,W. Chen, and L. T. Williams, Mol. Cell Biol. 14, 7483-7491
(1994). 229. S. N. Peterson, L. Trabalzini, T. R. Brtva, T. Fischer, D. L. Altschuler, P. Ma~elli, E. G. Lapertina, C. J. Der, and G. C. White, J. Biol. Chem. 271, 29903-29908 (1996). 230. R. M. F. Wolthius, B. Bauer, L. J. van't Veer, A. M. Wittinghofer, B. M. T. Burgering, and J. L. Bos, Oncogene 12,353-362 (1996). 231. R. M. Wolthius, F. Zwertkruis, T. C. Moen, and J. L. Bos, Curt. Biol. 8, 471-474 (1998). 232. S. B. Cantor, T. Urano, and L. A. Feig, Mol. Cell Biol. 15, 4578-4584 (1995). 233. M Julien-Flores, O. Dorseuil, F. Romer, F. Letourner, S. Saragosti, R. Berger, A. Tavitian, G. Gacon, and J. H. Camonis, J. Biol. Chem. 270, 22473-22477 (1995). 234. S. H. Park and R. A. Weinberg, Oncogene 11, 2349-2355 (1995). 235. L. J. Bos, Biochim. Biophys. Acta 1333, M19-31 (1997). 236. T. Urano, R, Emkey, and L. A. Feig, EMBOJ. 15, 810-816 (1996). 237. M. Ozaki, S. Kishida, T. Hinoi, T. Hasegawa, M. Tamada, and T. Kataoka, Oncogene 14, 515-521 (1997). 238. M. Ozaki, S. Kishida, H. Murai, "12Hinoi, and A. Kikuchi, Cancer Res. 15, 2387-2392 (1996). 239. R. Khosravi-Far, M. A. White, J. K. Westwick, P. A. Solski, M. Chrzanowska-Wudnicka, L. Van Aelst, M. H. Wigler, and C. J, Der, Mol. Cell Biol. 16, 3923-3933 (1996). 240. M. A. White, T. Vale, J. tt. Camonis, E. Schaefer, and M. H. Wigler, J. Biol. Chem. 27, 16439-16442 (1996). 241. R. M. Wolthius, N. D. Ruiter, R. H. Cool, and J. L. Bos, EMBOJ. 16, 6748~6761 (1997). 242. I. Hernandez-Mufioz, M. Malumbres, P. Elonardi, and A. Pellieer, Oncogene 19, 2745-2757 (2000). 243. C. Gratas and G. Powis, Anticancer Res. 13, 1239-1244 (1993). 244. T. Weider, Z. Zhang, C. C. Geilen, C. E, Orfanos, A. E. Giulano, and M. C. Cabot, Cancer Lett. 100, 71-79 (1996). 244a. L. Lucas, R. Hernandez-Alcoceba, V. Penalva, and J. K. Lacat, Oneogene 20, 1110-1117 (2001). 245. S. Clive, J. Gardiner, and R. C. Leonard, Cancer Chemother. Pharmacol. 44 (suppl.), $29-30
(1999).
232
SALVADOR AZNAR AND JUAN CARLOS LACAL
246. G.A. Ruiter, S. E Zerp, H. Bartelink, W, J. van Blitterswijk, and M. Verheij, Cancer Res. 59, 2457-2463 (1999). 247. B. Erdlenbrunch, V. Jendrossek, M. Marx, A. Hunold, H. Eibl, and M. Lakomek, Anticancer Res. 18, 2551-2557 (1998). 248. T. Wieder, C. E. Orfanos, and C. C. Geilen, J. Biol. Chem. 273, 11025-11031 (1998). 249. S. M. Konatantinov, M. Topashka-Ancheva, A. Benner, and M. R. Berger, Int. J. Cancer 77, 778--786 (1998). 250. N. Dittrich, R. Haftendorn, and R. Ulbrich-Hofmann, Biochim. Biophys. Acta 1391, 265-272 (1998). 251. I. Leiros, F. Secundo, C. Zambonelli, S. Servi, and E. Hough, Struct. Fold Des. 8, 655-667 (2000). 252. J. Wittenberg and A. Kornberg, J. Biol. Chem. 202, 431~t44 (1953). 253. A. Cuadrado, A. Carnero, E Dolfi, B. Jimenez, and J. c. Lacal, Oncogene 2959-2968 (1993). 254. B. Jimenez, L. del Peso, S. Montaner, P. Esteve, and J. C. Lacal,J. Cell Biochem. 57, 141-149
(1995). 255. 256, 257. 258. 259. 260. 261. 262. 263. 264. 265. 266. 267. 268. 269. 270.
J. c. Lacal, J. Moscat, and S. A. Aaronson, Nature (London) 330, 269-271 (1987). J. C. Lacal, Mol. Cell Biol. 19, 333-340 (1990). I. G. Macara, Mol. Cell Biol. 9, 325-328 (1989). K. K. Bhakoo, S. R. Williams, C. L, Florian, H. Land, and M. Noble, Cancer Res. 56, 46304635 (1996). A. Rodriguez-Gonzalez, A. Ramirez de Molina, and J. C. Lacal, unpublished results. R. Hernandez-Alcoceba, L. Saniger, J. Campos, M. C. Nufiez, E Khaless, M. A. Gallo, A. Espinosa, and J. C. Lacal, Oncogene 15, 2289-2301 (1997). R. Hernandez-Alcoceba, F. Fernandez, and J. C. Lacal, Cancer Res. 59, 3112-3118 (1999). J. C. Lacal, L Drugs 4, 419-426 (2001). D. Teegarden, E. J. Taparowski, and C. Kent, J. Biol. Chem. 265, 6042--6047 (1990). S. Ratman and C. Kent, Arch. Biochem. Biophys. 323, 313--322 (1995). K. Nakagami, T. Uchida, S. Ohwada, Y. Koibuchi, Y. Suda, T. Sekine, and Y. Morishita, Jpn. J. Cancer Res. 90, 419-424 (1999). Z. Kiss and T. Chung, Biochem. Biophys. Res. Commun. 218, 505~509 (1996). T. Chung, K. S. Crilly, W. H. Anderson, J. j. Murhejee, and Z. Kiss, J. Biol. Chem. 272, 3064-3072 (1997). C. C. Cunningham, J. T. Holmlund, J. H. Schiller, R. S. Geary, T. J. Kwoh, A. Dorr, and J. Nemunaitis, Clin. Cancer Res. 6, 1626-1631 (2000). P. J. O'Dwyer, J. P. Stevenson, M. Gallagher, A. Cassella, I. Vasilevska, B. P. Monia, J. Holmlund, E A. Dorr, and K. S. Yao, Clin. CancerRes. 5, 3977-3982 (1999). J. P. Stevensonm, K. S. Yao, M. Gallagher, D. Friedland, E. P. Mitchell, A. Cassella, B. Monia, T. J. Kwoh, R. Yu, J. Holmlund, E A. Dorr, and O'Dwyer, Clin. Oncol. 17, 2227-2237
(1999). 271. P. C. Gokhale, D. McRae, B. P. Monia, A. Bagg, A. Rahman, A. Dritschilo, and U. Kasid, Antisense Nucleic Acid Drug Dev. 9, 191-201 (1999). 272. T. Geiger, M. Muller, B. P. Monia, and D. Fabbro, Clin. CancerRes. 3, 1179-1185 (1997). 273. B. P. Monia, J. F. Johnston, T. Geiger, M. Muller, and D. Fabbro, Nat. Med. 2, 668-675
(1996). 274. Y. Kalechman, D. L. Longo, R. Catane, A. Shani, M. Albeck, and B. Sredni, Int. J. Cancer 86, 281-288 (2000). 275. A. D. Cox and C. J. Der, Curr. Opin. Cell Biol. 4, 1006--1008 (1992). 276. J. B. Gibbs, A. oliff, and N. E. Kohl, Cell (Cambridge, Mass.) 77, 175--178 (1994). 277. R. E. Barrington, A. Subler, E, Rands, C. A. Omer, P. J. Miller, J. E. Hundley, S. J. Koester, D. A. Troyer, D. J. Bearss, M. W Conner, J. B. Gibbs, K. Hamilton, K. S. Koblan, S. D. Mosser,
NEW TARGETS FOR ANTICANCER DRUG DESIGN
278. 279. 280. 281. 282. 283. 284. 285. 286. 287. 288. 289. 290.
291. 292. 293. 294. 295.
296. 297.
298. 299. 300. 301. 302. 303. 304. 305. 306.
233
T. J. O'Neill, M. D. Sehaber, E. T. Senderak, J. j. Windle, A. Oliff, and N. E. Kohl, Mol. Cell Biol. 18, 85-92 (1998). R. Mangues, T. Corral, N. E. Kohl, W. F. Symmans, S. Lu, M. Malumbres, J. B. Gibbs, A. Oliff, and A. Pellicer, Cancer Res. 58, 1253-1259 (1998). J. Gibbs, A. Oliff, and N. E. Kohl, Cell (Combridge, Mass.) 77, 17.5-178 (1994). A.X. Liu, W. Du, J. P. Liu, T. M. Jessel, and G. C. Prendergast, Mol. Cell Biol. 20, 6105-6113 (2000). W. Du, E Lebowitz, and G. C. Prendergast, Mol. Cell Biol. 19, 1831-1840 (1999). W. Du and G. C. Prendergast, Cancer Res. 57, 5492-5496 (1999). G. Zaleman, V. Closson, G. Linares-Cruz, F. Lerebours, N. Honore, A. Tavitian, and B. Olofsson, Oncogene 10, 1935-1945 (1995). S. Ellis and H. Mellor, Trends'. Cell Biol. 10, 85-88 (2000). E Adamson, C. J. Marshall, A. Hall, and E A. Tilbrook, J. Biol. Chem. 267, 20033-20038 (1992). H. Mellor, E Flynn, C. D. Nobes, A. Hall, and E J. Parker, j. Biol. Chem. 273, 48114814 (1998). A. Balendran, A. Casamayor, M. Deak, M. Paterson, P. Gaffney, R. Currie, C. E Dowries, and D. R. Alessi, Cur~: Biol. 9, 393-404 (1999). W. Du, A. Liu, and G. C. Prendergast, Cancer Res. 59, 42084212 (1999). A, Kim, L. Kakalis, N. Abdul-Manan, G. Liu, and M. Rosen, Nature (London) 404, 151-158 (2000). J. Mira, V. Bernard, J. Groffen, L. Sanders, and U. Knaus, Proc. Natl. Aead. Sci. U.S.A. 97, 185-189 (2000). E Jordan, R. Bragao, M. Boavida, C. Gespaeh, and E. Chastre, Oneogene 18, 6835-6839 (1999). L. Boutguignon, H. Zhou, L. Shao, and Y. Cben, J. Cell Biol. 150, 177-192 (2000). M. Sudol, H. Chen, C. Bougeret, A. Einbond, and P. Bork, FEBS Lett. 369, 67-71 (1995). G. Pirozzi, S. McConnell, A. Uveges, J. Carter, A. Sparks, B. Kay, and D. Fowlkes, ]. Biol. Chem. 272, 14611-14616 (1997). M. Zhao, H. Koyoi, Y. Yamamoto, M. Ito, M. Towatari, S. Omura, T. Kitanmra, R. Ueda, H. Saito, and T. Naoe, Leukemia 14, 374-378 (2000). M. Okabe, Y. Uehara, T. Noshima, T. Itaya, Y. Kunieda, and M. Kurosawa, Leukemia Res. 18, 867-873 (1994). A. Novogrodsky, M. Weisspapir, M. Patya, A. Meshorer, and A. Vaniehl4n, Cancer Res. 58, 2397-2403 (1998). A. Morreale, M. Venkateson, H. R. Mott, D. Owen, D. Nietlispach, E N. Lowe, and E. D. Lane, Nat. Struct. Biol. 7, 384~388 (2000). J. Styawan, K. Koide, T. C. Diller, M. E. Bunnage, S. S. Taylor, K. C. Nicolau, and L. L. Brunton, Mol. Pha~nacol. 56, 370-376 (1999). K. Koide, M. E. Bunnage, L. Gomez Paloma, J. R. Kanter, S. S. Taylor, L. L. Brunton, and K. C. Nicolau, Chem. Biol. 2, 601~08 (1995). Z. Wang, B. J. Canagarajab, J. C. Kassisa, M. H. Cobb, P. R. Young, S. Abdel-Meguid, J. L. Adams, and E. J. Goldsmith, Structure 15, 1117-1128 (1998). L. Seymour, Cancer Treat. Rev. 25, 301~312 (1999). M. D. Garrett and E Workman, Eur J. Cancer 35, 2010-2030 (1999). L. Gamet@ayrastre, S. Maneti, M. P. Gratacap, J. Tulliez, H. Chap, and B. Payrastre, Gen. PharmacoI. 32, 279-286 (1999). R. M. Schultz, R. L. Merriman, and S. L. Andis, Anticancer Res. 15, 1135-1139 (1995). L. E. Lemke, G. D. Paine-Murrieta, C. W. Taylor, and G. Powis, Cancer Chemother Pharmacol. 44, 491497 (1999).
234
SALVADOR AZNAR AND JUAN CARLOS LACAL
307. G. Razzini, C. E Berrie, S. Vignati, M. Broggini, G. Mascetta, A. Brancaccio, and M. Falasca, FASEB]. 14, 1179-1187 (2000). 308. K. Jiang, D. Coppola, N. C. Crespo, S. V. Nicosia, A. D. Hamilton, S. Sebti, and J. Q. Cheng, Mol. Cell Biol. 20, 139-148 (2000). 309. T. Sasaki, J. lrie-Sasaki, Y. Horie, K. Bachmaier, J. E. Fata, M. Li, A. Suzuki, D. Bouchard, A. Ho, M. Redston, S. Gallinger, R. Khokha, T. W. Mak, P. T. Hawkins, L. Stephens, S. W. Scherer, M. Tsao, and ]. M. Penninger, Nature (London) 406, 897-902 ((2000).
Major Histocompatibility Class I Folding, Assembly, and Degradation: A Paradigm for Two-Stage Quality Control in the Endoplasmic Reticulum M A R K R. FARMERY 1 AND N E I L J. B U L L E I D "2
University of Manchester School of Biological Sciences 2.205 Stopford Building Oxford Road, Manchester M13 9PT United Kingdom I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Primary Quality Control of Glycoprotein Folding in the ER . . . . . . . . . . . . . A. De- and Reglucosylation of Nascent Polypeptides . . . . . . . . . . . . . . . . . . B. C'alnexin and Calreticulin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Primary Quality Control and the Degradation of Misfolded Proteins . . . D. ERp57 and Quality Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . III. Secondary Mechanisms and Protein-Specific Quality Control . . . . . . . . . . . . IV. Quality Control in the ER: General Considerations . . . . . . . . . . . . . . . . . . . . V. Quality Control and MHC Class I Folding and Assembly . . . . . . . . . . . . . . . A. Primary Quality Control and MHC Class I Biogenesis . . . . . . . . . . . . . . . B. Protein-Specific Factors and MHC Class I Assembly . . . . . . . . . . . . . . . . VI. Conclusions and Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
236 239 239 241 243 246 247 249 252 254 260 263 264
Protein folding in living cells is a complex process involving many interdependent factors. The primary site for folding of nascent proteins destined for secretion is the endoplasmic reticulum (ER). Several disease states, including cystic fibrosis, are brought about because of irregularities in protein folding. Under normal cellular conditions, "quality control" mechanisms ensure that only correctly folded proteins are exported from the ER, with incorrectly folded or incompletely assembled proteins being degraded. Quality control mechanisms can be divided into two broad processes: (1) Primary quality control involves general mechanisms that are not specific for individual proteins; these monitor 1Present address: Karolinska Institute-Sumitomo Pharmaceuticals Center for Alzheiiner's Research, Karolinska Institute, KFC Novum, S-141 86 Huddinge, Sweden. 2To whom correspondence should be addressed. Telephone +44 (0)161 275 5103; fax +44 (0)161 275 5082; E-mail [email protected]. Progressin NucleicAcidResearch and MolecularBiology,Vol.67
235
Copyright© 2001by AcademicPress. All rightsof repr~xluetionin anyformreserved. 0079-6603/01$35.00
236
MARKR. FARMERYAND NElL J. BULLEID the fidelity of nascent protein folding in the ER and mediate the destruction of incompletely folded proteins. (2) Partially folded or assembled proteins may be subject to secondary quality control mechanisms that are protein- or proteinfamily-specific. Here we use the folding and assembly of major histocompatibility complex (MHC) class I as an example to illustrate the processes of quality control in the ER. MHC class I, a trimeric complex assembled in the ER of virally infected or malignant cells, presents antigenic peptide to cytotoxic T lymphocytes; this mediates cell killing and thereby prevents the spread of infection or malignancy. The folding and assembly of MHC class I is subjected to both primary and secondary quality control mechanisms that lead either to correct folding, assembly, and secretion or to degradation via a proteasome-associated mechanism. © ~001 Academic Press.
I. Introduction Protein folding in the living cell is a complex and dynamic process, prone to error. Numerous mechanisms have evolved to ensure that newly synthesized proteins achieve their native, functional form (1). Here, we examine one such mechanism, using the folding and assembly of the type I membrane glycoprotein major histocompatibility complex (MHC) class I as an example. Beginning with a general discussion on quality control in the ER, we aim to unify current dogma regarding glycoprotein folding and assembly by introducing and extending the model of a two-stage protein folding quality control in the ER that was recently proposed in an excellent review article (2). The ER is a primary site of protein synthesis and biogenesis in mammalian cells and is the first organelle encountered during the transition of a protein through the secretory pathway. Secretory proteins emerging from the ER usually comprise a compact folded native conformation and have undergone correct posttranslational modification (3). Common side products of protein biogenesis in the ER are misfolded and incompletely assembled proteins. Unlike correctly folded and assembled proteins, these are retained and degraded (4). These intermediate protein forms are sorted not only from other proteins in the ER but from conformational variants of themselves as well. In other words, quality control (QC) in the ER ensures the secretion of correctly folded and assembled multisubunit proteins and describes a process of conformation-dependent molecular sorting of newly synthesized proteins (2) In order to enter the secretory pathway, nascent, unfolded proteins are cotranslationally translocated across the membrane of the ER. After entering the lumen of the ER, proteins begin to fold with the assistance of molecular chaperones and other folding factors (1). The ER environment is naturally oxidizing, favoring disulfide bond formation. The specific ER-resident enzymes involved
QUALITYCONTROLIN THE ENDOPLASMICRETICULUM
237
in protein folding, include peptidyl prolyl cis-trans-isomerase and members of the protein disulfide isomerase (PDI) family. The milieu of the ER lumen also includes molecular chaperones of the Hsp70 and Hsp90 class heat-shock proteins such as immunoglobulin heavy chain binding protein (Bip/Grp78) and Grp94 (1). A group of chaperones with lectin-like properties may also be encountered, depending on the nascent polypeptide; these are the type I membrane protein calnexin (CNX) and its soluble homolog calreticulin (CRT) (5). The thiol-dependent reductase ERp57 (also known as ER-60, Erp60, Erp61, Grp58, P58, HIP-70, or Q-2) has been shown to form a complex with either CNX or CRT during the folding of glycosylated membrane and soluble proteins in the ER, but the precise role of this protein remains to be determined (6). Current hypotheses surrounding the function of ERp57 will be discussed. However, it is the coordinated activity of these ER-resident proteins that mediates the controlled folding and assembly of nascent polypeptides and ensures the fidelity of ER quality control. Quality control in the ER comprises several overlapping mechanisms that share a high degree of redundancy. These mechanisms can be divided into two subsets. Primary QC controls the folding of all proteins, whereas secondary QC is specific for selected proteins and protein families. Primary QC is dependent on general biophysical properties shared by incompletely folded proteins. These include exposure of hydrophobic surface patches, mobile loops, and a lack of structural compactness. During protein biogenesis, these features are transiently exposed, resulting in exposure of nascent proteins to the primary QC machinery (2). The main strategies involved in primary QC mechanisms involve retention in the ER and selective ER-associated degradation (ERAD) of incompletely or incorrectly folded molecules. In this chapter, we will primarily consider quality control mechanisms involved in glycoprotein biosynthesis. The retention of glycoproteins during primary QC is dependent on interaction with lectin-like chaperones. Other retention mechanisms exist, mediated by exposure of free cysteine residues or aggregation, and these are discussed elsewhere (2, 7). Most soluble and membrane-bound proteins synthesized in the ER undergo asparagine-linked glycosylation. Covalent attachment of oligosaccharides to asparagine residues of nascent polypeptide chains occurs soon after exposure to the lumen of the ER. The subsequent action of three enzymes--uridine diphosphate glucose : glycoprotein glucosyltransferase (UGGT) and glycosidases I and II--results in the generation of monoglucosylated glycosylation intermediates which are recognized by the lectins CNX and CRT. CNX and CRT act in concert to mediate primary QC by associating transiently with these glycoprotein intermediates, influencing folding, oligomerization, and ER retention. UGGT will reglucosylate only incorrectly folded substrates, and this specificity provides a sensing mechanism allowing the cell to recognize incompletely folded
238
MARKR. FARMERYAND NEILJ. BULLEID
proteins. Thus, polypeptide chains that have not reached their native conformation are maintained in a monoglucosylated state bound to either CNX or CRT. The association with these lectins thereby retains the polypeptide within the ER until a native conformation has been reached. The cycle of binding and release from CNX or CRT can be broken either by correct folding or by targeted degradation to prevent a buildup of misfolded proteins within the ER. A variety of protein-specific secondary QC mechanisms assist the final stages of protein transit through the ER. Interaction with these factors occurs once the primary QC machinery no longer retains a protein. The mechanism whereby proteins are selected for transport from the ER to the Golgi apparatus is unclear. The two main models are bulk flow and cargo capture. Bulk flow suggests homogeneous movement of the contents of the ER to the Golgi, whereas cargo capture implies specific receptor-mediated movement of molecules from the ER along the secretory pathway. Neither model adequately explains cargo movement and, in reality, a combination of both mechanisms may exist (2). The accessory proteins mediating secondary QC are involved in maturation, folding, and complex assembly in the ER and influence forward transport to the Golgi, either by retention or by acceleration of protein flow. Typically, accessory proteins involved in secondary QC function as chaperones or assembly factors, escort proteins, or cargo receptors. Crucially, these proteins mediate the folding of specific proteins or protein families, unlike those involved in primary QC, consequently, they are often expressed in specific cell types. Major histocompatibility complex class I is trimeric protein complex assembled in the ER. The folding and oligomerization of MHC class I is mediated by primary and secondary QC mechanisms. Functional MHC class I molecules are expressed on the surface of virally infected cells and present antigen to T cells in the cell-mediated arm of the immune system. MHC class I consists of three subunits: a type I membrane glycoprotein heavy chain, soluble /~2-microglobulin, and an 8- to 12-residue antigenic peptide. Antigenic peptides are generated in the cytosol following proteasome-mediated proteolysis and are transported into the lumen of the ER by specific transport proteins. MHC class I heavy chain undergoes disulfide bond-mediated folding and is associated with j62-microglobulin prior to association with further accessory proteins and antigenic peptide loading. The process of MHC class I folding, assembly, and peptide loading is relatively well characterized and intimately involves primary and secondary QC mechanisms and factors. In this chapter, we describe quality control in the ER and review current knowledge regarding the role of CNX, CRT, and ERp57 in glycoprotein maturation. We discuss ER-associated degradation in the context of the quality control of glycoprotein biosynthesis. In addition, we outline the QC mechanisms that ensure the correct folding, assembly, and cell-surface expression ofpeptideloaded MHC class I molecules.
QUALITYCONTROLIN THE ENDOPLASMICRETICULUM
239
II. Primary Quality Control of Glycopratein Folding in the ER A. De- and Reglucosylation of Nascent Polypeptides Primary quality control of glyeoprotein folding in the ER is dependent on the addition of core oligosaceharides to nascent proteins. The mechanisms involved in addition of oligosaccharide side chains have been understood since the 1980s (8). Branched 14-residue oligosaccharide glycans (Glc3MangGlcNAc2; Fig. 1) are transferred to asparagine residues in the consensus sequence NXS/T by oligosaecharyltransferase associated with the translocation machinery. This occurs eotranslationally as soon as the aceeptor site enters the lumen of the ER. The subsequent processing of added oligosaccharides is intimately linked to quality control in the ER, and removal of glucose residues within the core glycans begins on the growing nascent chain. Several ER-resident enzymes are involved in oligosaccharide processing. Glucosidase I removes the terminal 0t-l,2-1inked glucose, whereas glucosidase II excises the remaining ot-l,3-glucose residues. This is an extremely efficient process, resulting in complete deglucosylation of nearly all N-linked glycoprotein chains. The monoglucosylated glycans that bind CNX and CRT arise as intermediates during the stepwise removal of glucose residues. In an alternative pathway, monoglucosylated glycans are generated by action of the ER lumenal enzyme uridine diphosphate (UDP) glucose:glycoprotein glucosyltransferase (UGGT), which transfers singleglucose residues to fully deglucosylated glycans. This reestablishes the glucose~-l,3-mannose glycosidic bond recognized by CNX and CRT (Fig. 1) (9, 10). Clearly, mechanisms are present within the ER to facilitate addition and removal of glucose residues during quality control. The question then arises as to how misfolded glycoprotein substrates are recognized. Current data imply that specificity arises partly in the action of UGGT. However, this is not fully understood and additional features may be required. The sensing of intermediate or misfolded protein structures is central to the CNX/CRT cycle. Correctly folded proteins are released to higher QC mechanisms, whereas incorrectly folded or immature glycoproteins are retained within the ER and ultimately degraded. Central to the QC process is the ability to sense and retain incompletely assembled proteins. Clearly, the role of UGGT is crucial for glycoprotein quality control. UGGT, a soluble glycoprotein of 170 kDa that resides in the lumen of the ER, is ubiquitously expressed in most cell types, tissues, and species (9, 10) and has the unique ability to glucosylate misfolded, but not native, glycoproteins. Investigation of the action of UGGT has proved challenging because the many structures adopted by nonnative proteins are difficult to isolate. Nevertheless, technical adaptations in several experiments have provided insights into the
~
~
.~ ~
~
~.~,-~
r~
.= = ,
~
= "~ ~ '
~
~.~
~
~.~
~
o
O~ '~
~
,~
"~"~ ~ ~x..e ~ - do
~ '~
® ~ - ~ = "~ ~
~ "~
--
~
mo~O
.~
~
~z.~
~
~ ¢ ~~"
o,a
=
240
~
o
,~
o"c:l
~
~
~'-6
o= =~.= ,.,
QUALITYCONTROLIN THE ENDOPLASMICRETICULUM
241
action of UGGT through the use of well-characterized intermediates, facilitating the identification of the folding stages in which glycoproteins are glucosylated. By utilizing conformers of RNase B, studies have shown that UGGT recognizes partially folded nonnative intermediate forms, but not the native state, of this glycoprotein. In addition, it was also suggested UGGT is able to distinguish between normative conformers as it did not interact with fully unfolded RNase B, suggesting that discrete populations of glycoprotein intermediates are recognized by UGGT (9). In another study (10), it was concluded that the protein recognition elements and oligosaccharide have to be covalently linked and that the innermost GlcNAc residue has to be accessible to UGGT. It is probably this last factor that contributes to the exclusive glucosylation of incompletely folded glycoproteins as, in the native conformation, the innermost GlcNAc becomes inaccessible to macromolecular probes (11). It has been suggested that UGGT distinguishes misfolded protein intermediates through exposure of patches of hydrophobic amino acid side chains in nascent glycoproteins, in a manner similar to that of Hsp70 (12). This was based on biochemical studies with a purified enzyme that was found to bind to hydrophobic amino acids under physiological conditions (11). However, no evidence has arisen in vivo suggesting that such patches trigger glucosylation. In contrast, it is more probable that UGGT has a preference for partially structured confbrmations (10). Thus, a series of ER resident proteins act in concert to mediate association of nascent glycoproteins with CNX and CRT by the generation of glycan intermediates and the recognition of immature or misfolded glycoproteins. UGGT has an essential role in the primary quality control of glycoprotein folding in the ER as it is involved in the recognition of incompletely folded and misfolded protein conformations. It is able to specifically reglucosylate these polypeptides, generating monoglucosylated glycans that then associate with CNX/CRT. These glycoprotein conformers are also generated by the actions of two further ER-resident enzymes, glucosidase I and glucosidase II. It is these enzymes that regulate the association of glycoproteins with CNX and CRT and thereby the retention of nascent glycoproteins in the primary quality control pathway (Fig. 1).
B. Calnexin and Calreticulin Two homologous lectin-like proteins which are localized in the ER are central to the primary quality control of glycoprotein folding. These are the type I transmembrane phosphoprotein calnexin (CNX) (13) and its soluble homolog calreticulin (CRT) (14). In addition, two CNX homologs have been described in the testis, calmegin (15) and calnexin-t (16). All members of the CNX family contain repetitive proline-rich P domains within their lumenal region, which contains a high-affinity calcium-binding site where oligosaccharides are thought to
242
MARKR. FARMERYAND NEILJ. BULLEID
bind specifically (17, 18). The C-terminal cytosolic domains of CNX, calmegin, and calnexin-t contain four defined regions: (1) a juxtamembrane lysine-rich domain; (2) an acidic glutamic acid-rich domain; (3) a phosphorylation domain; and (4) a putative ER retrieval motif. Well-conserved serine phosphorylation sites are located within the phosphorylation domains, three in CNX and four in calmegin and calnexin-t. Analysis of CNX phosphorylation sites by mass spectrometry in vivo revealed that two were within protein kinase CK2 sites and the third was within a PKC/PKD site (17). In addition, it has recently been shown that phosphorylation of CNX can regulate ribosome-binding properties of CNX and, probably, its association with the translocon (18). CRT has both a highly negatively charged calcium-binding domain and a -KDEL ER retention motif at its C terminus (19). CRT is localized primarily in the lumen of the ER, but it has been reported to be present in several cell compartments. Several functions have been described for CRT, including calcium sequestration and RNA binding (19), but it is now clear that, like CNX, it is a lectin-like chaperone that transiently binds to many nascent glycoproteins in the ER (14, 20). Both CNX and CRT function as monomers, although they may be part of larger, dynamic, heterologous complexes comprising molecular chaperones and folding enzymes. Early experiments, utilizing tissue culture ceils treated with glycosylation inhibitors, showed that CNX selectively associated with folding intermediates of glycoproteins (21, 22). In an extension of these studies using glucosidase inhibitors, it was demonstrated that CNX bound to partially trimmed glycoproteins (23). It was subsequently shown in biochemical studies that both CNX (24-27) and CRT (28) specifically associated with monoglucosylated glycoproteins. CNX and CRT are unique lectin-like proteins in that they bind only monoglucosylated core glycans. The precise nature of how CNX and CRT mediate glycoprotein folding remains controversial. Two models have been proposed that address this question. So far, we have discussed a "lectin-only" model in which CNX and CRT associate with glycoprotein exclusively through lectin-oligosaccharide interactions. As we have described, dissociation of CNX and CRT from nascent glycoproteins is mediated by glucosidase II, which removes the single glucose residue, and rebinding occurs when UGGT functions as a folding sensor and adds back a glucose residue, reforming the monoglucosylated intermediate. The role of CNX and CRT in chaperoning protein folding is then facilitated by other factors and folding enzymes that are subsequently recruited, such as ERp57. Recent studies using purified components in vitro have shown that CNX and CRT suppress the aggregation of both monoglucosylated glycoproteins and nonglycosylatedunfolded proteins, and that this effect is enhanced by the action of ATE It would seem that CNX mad CRT have the ability to distinguish native and nonnative protein conformations and behave as classical chaperones in vitro. These observations have supported a "dual-binding" model of CNX and CRT function, which was suggested in the 1990s (29-32). This model incorporates
QUALITYCONTROLIN THE ENDOPLASMICRETICULUM
243
the lectin-binding properties of CNX and CRT and proposes that a second site binds polypeptide segments of unfolded glycoproteins. Dissociation is mediated by a conformational change in the polypeptide-binding site, combined with the action of glucosidase II. As in the lectin-only model, cycles of binding and release continue until the glycoprotein acquires a conformation in which recognition sites for UGGT and CNX/CRT are buried within native structure. What is clear is that the concerted actions of CNX, CRT, UGGT, and glucosidase I and II mediate cycles of binding, dissociation, and de- and reglucosylation--a process that results in glycoprotein retention prior to release to secondary QC mechanisms. The retention of glycoproteins will eventually lead to their degradation. In the next section, we discuss potential mechanisms for regulating this process.
C. Primary Quality Control and the Degradation of Misfolded Proteins Misfolded and incompletely assembled glycoproteins are recognized by the QC machinery in the ER and are targeted for degradation in a process known as ER-associated degradation (ERAD) (4). ERAD is mainly carried out by the 26S proteasome located in the cytosol, but targeting and dislocation from the ER involve several stages. Unassembled or terminally misfolded proteins are recognized by chaperones such as BiP and CNX, or by factors such as mannosespecific lectins, and are transported out of the ER by retro-translocation through the Sec61 channel. They are then, in the case of glycoproteins, deglycosylated and polyubiquitinylated prior to degradation by the proteasome (33, 34). It is not clear how misfolded or incompletely assembled proteins are identified or how this process is regulated, but it is likely to involve the machinery responsible for protein folding. A close relationship must exist between folding and assembly on the one hand and targeting for degradation on the other. It has been suggested that one mechanism whereby proteins can be targeted for degradation is by regulating the trimming of N-linked oligosaccharide chains (35, 37). As we have seen, prior to folding, most glycoproteins interact with either CNX or CRT or both, the interaction with these proteins being regulated by the glucosylation state of the oligosaccharide side chain. The oligosaccharide side chain can also be modified within the ER by the action of mannosidase I which removes terminal mannose residues. Interestingly, inhibition of mannose trimming, either by expression in ER mannosidase-deficient yeast strains (35) or by incubation of mammalian cells in the presence of deoxymannojirimycin (37-39), results in stabilization of polypeptide chains that would normally be degraded. The specific mannosidases involved are the slow-acting ER resident c~-l,2-mannosidases (35). These results have led to the hypothesis that mannose trimming provides a timing mechanism allowing the lifetime of a protein within the ER to be regulated.
244
MARKR. FARMERYAND NEIL J. BULLEID
How is ERAD integrated with quality control? A recent study identified a specific oligosaccharide side-chain structure involved in the degradation of a misfolded yeast glycoprotein (35). This was extended in a study in which the intracellular fate of terminally misfolded otl-antitrypsin was examined in mouse hepatoma cells (40). Central to ERAD is the oligosaccharide structure Man8GlcNAc2. By simultaneously inhibiting proteasome activity, it was also shown that the glycan intermediate GlclMan8GlcNAc2 stably associates with CNX. A picture begins to emerge as to how cycles of association with CNX, glucose, and mannose trimming could function together with the proteasome to regulate ERAD. This is summarized in Fig. 2, which depicts a fully glucosylated glycoprotein undergoing multiple rounds of glucose and mannose trimming, due to the actions of glucosidases I and II and mannosidases I and II, and regulated by association with UGGT and/or calnexin. Intermediate glycoforms are subsequently recognized and targeted either to higher QC processes or to retrotranslocation and degradation. Again, this model (Fig. 2) emphasizes that the role of UGGT as a folding sensor is important not only in the cycle of association and dissociation with CNX/CRT but also for targeting misfolded proteins for degradation. In addition, it also suggests a central role for CNX in the targeting of some glycoprotein substrates for retro-translocation (2). Once a misfolded protein is selected for degradation, it is dislocated from the ER by a process of retro-translocation. This has been shown in a series of genetic and biochemical experiments in both yeast and mammalian cells that have elucidated some of the mechanisms involved (33, 42, 43). The possibility that proteins could be rerouted from the ER to the cytosol was originally inferred from observations that toxic proteins that enter the ER kill the cell by inactivating protein synthesis in the cytosol (44). This idea was subsequently confirmed by showing that mutant ricin A chain was transported from the ER to cytosol in a retrograde manner (45). These findings suggested that toxin molecules could be dislocated from the ER to the cytosol by a preexisting transport mechanism, which could also be used to export unfolded proteins prior to degradation (46). Several studies went on to identify factors involved in dislocation of proteins from the ER. The current model proposes that ERAD substrates are dislocated from the ER to the cytosol via the Sec61 channel (47). The role of the translocon in retrograde transport was uncovered in a combination of studies utilizing yeast genetics and mammalian cell biology (44, 49-51). In some cases, the degradation is dependent upon ubiquitin (37, 52-53) and, in addition, ER chaperones seem to be involved (48). In yeast, these include the BiP homolog Kar2p, CNX, and Sec63p (the yeast homolog of DnaJ) (54, 55). The involvement of other genes, yet to be characterized, has also been demonstrated in yeast ERAD mutants. These are known as DER and H E R D genes (56, 57) and, although their function is unknown, it is postulated that they are involved in the export from the ER
245
QUALITY CONTROL IN THE ENDOPLASMIC RETICULUM
GlclMan 9
V rq Mana
Man7
J Cytosol I Proteasome
=#
I
S
FIG. 2. Regulation of ER-associated degradation (ERAD) of glycoproteins. The folding of nascent glycoproteins (open circle) is mediated by the calnexin (CNX)/calreticulin (CRT) cycle through addition and removal of glucose residues and retention (see Fig. 1). Incorrectly folded proteins are subject to ERAD through the actions of glucosidases I and II (GI and GII, respectively), UDP-Glc : glycoprotein glucosyltransferase (UGGT), and ER mannosidases I and II (Man I and Man II, respectively). For clarity, oligosaccharide side structures are labeled. The fully folded glyeoprotein (closed circle) undergoes mannose trimming (from Man 9 to Man 7), further maturation, and exits the ER. Persistently unfolded proteins are targeted for degradation by the action of mannosidase I, which generates the MansGlcNAc2 form of the oligosaccharide. After reglycosylation by UGGT, the protein reassociates with CNX. The GlctMansGlcNAc2 form is a suboptimal substrate for GII, as compared to the GlclMangGlcNAc2 form (ref), which results in a prolonged interaction with CNX. The misfolded glycoprotein is retrotranslocated from the ER and degraded by the proteasome. This figure is adapted from Ref. 2 and based on a recently proposed model (40).
of both soluble and integral membrane proteins destined for ERAD. In the case of glycoproteins, it has been reported in some cases that deglycosylation occurs prior to degradation. This is thought to be mediated by the enzyme peptide:N-glycanase (PNGase), but the subcellular location of this step remains unclear (51, 58-60). The question remains as to how the ERAD mechanisms of dislocation and degradation are coupled. The driving force behind the retro-translocation of ERAD substrates from the ER to the cytosol is unknown. Several possible
246
MARKR. FARMERYAND NEILJ. BULLEID
mechanisms have been proposed involving familiar components of the ER quality control machinery. Is retro-translocation mediated by a "molecular ratchet" mechanism, similar to the role of BiP in translocation (61)? An analogous role for the cytosolic hsp70 chaperones was suggested and later ruled out following the observation that Ssalp, the cytosolic hsc70 in yeast, is not required for degradation of proot-factor or A1PiZ (62, 63). What is more likely is that the chemical energy of ubiquitination acts as a molecular ratchet to prevent backward motion of the protein once retro-translocation has been initiated, but this remains to be confirmed. Another scenario involves a multiprotein complex acting as the retro-translocation apparatus. In yeast this could comprise BiP, some DER gene products, and Sec63p, and would be allosterically connected to the proteasome to ensure high coupling between retro-translocation and degradation. The ATPase function of BiP provides a driving force from the lumenal side of the translocon, and the proteasome itself, by its AAA-ATPase subunits located at the 19S cap structure (75, 76), energizes the retro-translocation process from the cytosolic face of the ER (48).
D. ERp57 and Quality Control As we have seen, central to primary QC in the ER is the CNX/CRT cycle. Integration of de- and reglucosylation, rounds of CNX/CRT association and dissociation, and degradation result in the release of correctly folded and assembled proteins from the ER. Recently, a third dimension was added to this pathway with the discovery of ERp57 and its ability to influence glycoprotein folding. ERp57 is a member of the protein disulfide isomerase (PDI) family which, in addition to archetypal PDI, also includes ERp72 (CaBP2), P5 (CaBP1), and the pancreas-specific PDIp (66). ERp57 contains two "thioredoxin-like" motifs, which in PDI constitute the thiol/disulfide oxidoreductase active site (67). ERp57 has been shown to have thiol-dependent reductase activity, which indicates that ERp57 may play a role in protein folding. Numerous other functions have been proposed for ERp57, including a caruitine palmitoyl transferase, a cysteine protease, and a hormone-induced protein of the brain (66). A series of crucial experiments addressed the function of ERp57. These experiments showed specific interaction with N-glycosylated soluble and integral membrane glycoproteins, and afforded evidence that this interaction is dependent on glucose trimming (68). Since ERp57 possessed no intrinsic lectin-like properties, the immediate prediction of an involvement with CNX/CRT was proved experimentally, and it was proposed that ERp57 acts in concert with CNX and CRT to mediate glycoprotein folding. It was quickly established that ERp57 forms distinct complexes with CNX and CRT, both within the lumen of the ER and when the purified proteins are mixed in solution. This was subsequently extended to show that ERp57 interacted directly in complex with CNX and CRT in the absence of glycoprotein substrate (68-71).
QUALITYCONTROLIN THE ENDOPLASMICRETICULUM
247
The most recent model proposes that ERp57 functions in the ER lumen as a subunit of discrete complexes formed with CNX and CRT. These complexes modulate glycoprotein folding and are therefore acting as molecular chaperones (71). Indeed, this has been shown experimentally in several studies. As we shall see later, ERp57 is involved, in association with CNX and CRT, in the assembly of MHC class I (72-74). The PDI activity of ERp57 was addressed by studying glycoprotein refolding in vitro. It was found that the disulfide isomerase activity of ERp57 is greatly enhanced in the presence of CNX and CRT (75), providing direct evidence that it is the combination of ERp57 with CNX and CRT that modulates glyeoprotein folding. Finally, a direct role of ERp57 in catalysis of native disulfide-bond formation in glycoproteins has recently been elucidated (76). Here, mixed disulfide species formed in vivo between ERp57 and native viral glycoprotein substrates could be identified. These mixed disulfides represent transient intermediates during the catalysis of disulfide-bond formation and exchange, and they were obtained with ERp57 only when the precursors were N-glycosylated and suitably glucose-trimmed. Where does ERp57 fit into glyeoprotein folding in the ER? We have established that CNX/CRT are crucial players in primary quality control. It seems that ERp57 is recruited by CNX/CRT to mediate glycoprotein folding by functioning as a glycoprotein-specific PDI. However, E Rp57 can also influence the folding ofglycoproteins that lack cysteine residues. This suggests that the role of ERp57 may not be limited to catalysis of disulfide-bond formation and exchange, and that, like PDI, ERp57 may have a wider role as a molecular chaperone (66, 67, 71 ). Although the precise nature of ERp57 remains to be established, what is clear is that intricate processes occur within the lumen of the ER to mediate glycoprotein biogenesis. This is achieved by dynamic interaction among a variety of cytosolic and ER lumenal factors. This results in the release of correctly folded and assembled proteins in a native, transport-competent form to higher QC mechanisms. Alternatively, immature or misfolded proteins are rerouted to the cytosol and degraded by the proteasome. Primary QC mechanisms are vital in ensuring that only correctly folded proteins are released from the ER and transported along the secretory pathway.
III. Secondary Mechanisms and Protein-Specific Quality Control Folded, mature proteins are released from elements of the primary QC machinery in the ER and transported to the Golgi complex. The precise mechanisms involved in protein export from the ER are not fully understood. However, two models have been proposed: bulk flow and cargo capture (see Section I). It is likely that the reality in the cell encompasses both models, but neither
248
MARKR. FARMERYAND NEILJ. BULLEID
adequately explains the dynamic mechanisms occurring during protein export. What is clear is that, after folding has occurred in the ER, a process of sorting occurs as motifs are exposed on the proteins (77, 78) that direct their selective incorporation into coat complex II (COPII) vesicles (79-85). These bud off from the ER (86) and fuse to form the ER-Golgi intermediate compartment (ERGIC). The ERGIC is a collection of mobile tubular-vesicular structures located both near the Golgi and in the cell periphery. It is here that sorting occurs prior to anterograde movement to the Golgi or retrograde movement back to the ER (88, 109, 110). A variety of secondary QC mechanisms modulate the export of specific proteins from the ER. The factors involved are heterogeneous and relatively poorly characterized. However, it is possible to apply a general classification related to function, as summarized in Table I and references therein, although this is not an exhaustive list. The protein-specific factors mediating secondary QC function as molecular chaperones or assembly factors. A good example is HSP47. This protein-specific chaperone transiently associates with procollagen in the ER and dissociates from it in the cis-Golgi compartment (88). Others form potential cargo receptors. In mammalian cells, the mannose-specific transmembrane lectin ERGIC 53 acts as a cargo receptor for the transport of glycoproteins from the ER to the ERGIC (87). Other proteins limit the export of specific proteins from the ER, and these include egasyn and earboxylesterase (89, 90). A final group acts as escort proteins and includes the well characterized receptor-associated protein (RAP), which interacts with proteins of the lowdensity lipoprotein (LDL) receptor family and other transmembrane receptors. RAP prevents aggregation and premature ligand binding by its target molecules by escorting them out of the ER to the Golgi (91). The complex then dissociates, presumably because of the lower pH in the Golgi, and RAP is rerouted back to the ER via the KDEL receptor. All of these proteins are specific for particular target proteins or protein families, and they generally mediate maturation, folding, and assembly of proteins in the ER or accelerate or inhibit forward transport. Secondary QC mechanisms also involve amino acid signal motifs in the case of some membrane proteins. Typically, these are short peptide sequences in cytosolic and transmembrane protein domains that mediate selective export, retention, retrieval or degradation. Indeed, export motifs have been identified in several proteins destined for export from the ER. In transfected cells, the efficient exit of ERGIC-53 from the ER requires two COOH-terminal phenylalanine residues within the cytosolic domain. These mediate ERGIC-53 binding to COPII coats in vitro. The diphenylalanine motif by itself is not sufficient for efficient export of ER reporter proteins, suggesting that this is part of a larger, currently unknown, ER-exit signal (III). Members of the p24 protein family also possess a diphenylalanine motif that mediates binding to COPII coat.
QUALITYCONTROLIN THE ENDOPLASMICRETICULUM
249
These proteins cycle between the ER and post-ER compartments and have been proposed to function as transport receptors (112) and COPI receptors (113). The mechanism mediating binding between the diphenylalanine motif and the COPII coat is unknown. Other proteins contain a diacidic ER-exit motif. This was shown to be required for export of vesicular stomatitis virus glycoprotein (VSV-G). There is no evidence for a direct interaction of VSV-G protein with COPII (78), but this finding does support the existence of selective signal-motif mediated ER export. What are the mechanisms of selective signal-motif mediated ER export, retention and retrieval? One of the first examples identified was that of charged residues in the transmembrane domain of T cell receptor a chain, which were found to serve as a signal for selective ER degradation of the unassembled subunit (115). In another mechanism, the export of immature human highaffinity receptor for IgE from the ER is prevented by an exposed ER retention signal [a coat protein complex (COPI)-interacting KXXX motif] in the cytoplasmic C-terminal domain of the ot subunit. Export occurs when the ot subunit assembles with the F subunit and the KXXX signal is masked (116). In two studies of polytopic transmembrane proteins, the role of the cytosolic signal motif Arg-Lys-Arg (RKR) and the related sequence RXR was investigated. It was shown that mutation of these motifs (RKR) and (RXR) resulted in transport to the cell surface of both immature Drosophila ATP-sensitive K + channel and misfolded AF508 CFTR mutant, respectively (117, 118). As misfolded AF508 CFTR is known to concentrate in the ERGIC (119), RXR represents a signal that interacts with COPI components, resulting in retention. These sequences probably act as retention or retrieval signals that must be masked before protein can be exported from the ER. Conversely, in the ease of export motifs, it is likely that these are unmasked as the protein matures. Secondary QC factors are protein-specific and, in many eases, are thought to have eoevolved with their substrate. A variety of mechanisms share overlapping features. Crueially, these are based on prevention of the exit ofmisfolded proteins from the ER, or involve rerouting back to the ER for further maturation.
IV. Quality Control in the ER: General Considerations The first part of this chapter has provided a broad overview of quality control in the ER specifically related to glycoprotein folding. We have used examples drawn from the study of biochemistry, cell biology, and genetics in both yeast and mammalian systems to illustrate this complex, dynamic, and highly intricate process. This overview is by no means exhaustive, and controversy and conjecture
t-
©
0
"~
0
©
%
~ ~'-=
c
~
~
'-= ~
~ .~ .~
Z
m
Z .,<
~
m
M M ©
N M
~
:-~ ~
c~
~
~
~
~
~
~
cq
,
5
M
~
~
~
~.~.~
~o
< M ©
.~
•-
Z ©
~
~,
"~
<
~
r.
.~_~
Z
©
o
250
~
= o~
=
~
~
~
~
~
¢~
"~
~ .o -~..= .~ ~ ~ . =
-~
=
~ .~ ~ .~ .~. ~
..~
-~
.
.
.
,~
~
~
~.~
.
.
o
~
o
I
0~
.
6 u~
©
E
,d
,8 CD
o
251
252
MARK R. FARMERY AND NEIL J. B U L L E I D
exist at many levels. However, consensus is emerging in areas of study such as the mechanisms involved in the regulation of ERAD and the role of C NX, CRT, and ERp57. Evidently, general QC mechanisms are utilized during initial stages of protein folding, and this is nonspecific. As higher order structures are formed and assembly occurs, protein-specific quality control comes into play and mediates the release of mature proteins from the ER. Quality control mechanisms have wide-ranging implications in both protein biogenesis and pathogenesis of folding-related diseases, such as cystic fibrosis. Further detailed study is required to fully understand the truly dynamic process of quality control. In the next section we examine the folding and assembly of MHC class I. This protein complex matures in the ER and is subject to wellcharacterized primary and secondary QC processes.
V. Quality Control and MHC Class I Folding and Assembly Mammalian cells have evolved numerous protection mechanisms. Cells that have become cancerous or have been invaded by viruses are recognized and killed by cytotoxic T lymphocytes (CTLs), blocking the spread of malignancy or infection. Recognition is reliant on the expression of the major histocompatibility (MHC) class I molecule on the cell surface of an infected or malignant cell. This displays a pathogenic or tumor-specific peptide fragment to a CTL cell surface receptor, and this interaction mediates cell killing (120, 121). The human major histocompatibility complex is encoded in the human leukocyte antigen (HLA) locus, which is divided into seven main regions, encoding MHC class I, II, or III alleles. These alleles can be further subdivided so that, for example, MHC class I heavy chains are encoded by three HLA regions, namely HLA-A, B, or C (122). Mice also have three loci, namely K, D, and L (123). MHC class I molecules are expressed on the surface of nearly every cell in higher eukaryotes. At the cell surface the fully assembled MHC class I molecule consists of a 44-49-kDa class I heavy chain associated noncovalently with a small (12-kDa) accessoryprotein, fl2-microglobulin, and an 8- to 12-amino acid residue antigenic peptide (124). Class I heavy chain is a type I membrane glycoprotein comprising three extracellular domains (al, or2, and ct3), a transmembrane domain, and a cytosolic tail. Two intrachain disulfide bonds are present in the lumenal domain of class I heavy chain, with one in the c~2 domain and the other in the 0t3 domain. These disulfide bonds are formed in the ER where the folding of the molecule is regulated. Class I heavy chains have an unusually high degree of polymorphism, which mediates the variety and specificity of peptide antigen binding. This variability is located in the two outer domains, which form the peptide binding cleft (Fig. 3). Each specific class I
253
QUALITY CONTROL IN THE ENDOPLASMIC RETICULUM oeDtide
b.
a.
FIG. 3. Structure of the lumenal domain of human MHC class I. The structure of assembled trimerie human HLA-A2 MHC class I complex is illustrated (121, 186). (a) Here only the lumenal domain is showa. Heavy chain comprises three distinct regions, al, a2, and a3 domains, with the peptide-binding cleft clearly shown between domains cd and a'2. Two disulfide bonds are present in heavy chain, located in the c~2 and a3 domains. The single N-linked glycosylationsite is in the al domain. Heavy chain is associated noneovalently with ¢12-microglobulin,which has a single disulfide bond. Antigenic peptide lies in the peptide-binding cleft and is presented to cytotoxic T lymphocytes. (b) The complex is rotated to show a view from above, with antigenic peptide lying inside the well-defined peptide-binding cleft.
molecule has a preference for certain amino acid residues at specific positions in the peptides. The side chains of these preferred residues fit well into the pockets of the peptide binding cleft, which have unique shapes and charges owing to the presence of polymorphic residues (124, 152). In humans, heavy chain undergoes asparagine-linked glycosylation at a single consensus site, whereas in mice, either two or three oligosaecharide side chains are added (136). In both humans and mice, N-linked glycosylation occurs within the c~1 domain. H o w are the folding and assembly of M H C class I regulated? Many studies have analyzed the processes occurring during M H C class I assembly. Studies in mammalian cells utilizing biochemistry, genetics, and cell biology have revealed an intricate pathway, mediated by primary and secondary QC mechanisms in the EB. Interactions between key components have been elucidated in vitro using protein biochemistry. Overall, consensus is emerging as to how M H C class I is assembled in the EB. This is summarized in Fig. 4, which illustrates the currently
254
MARK R. FARMERYAND NEIL J. BULLEID
(a)
GlycosylatedMHC class I heavy chain
ER lumen (d)
(c)
Transport to Golgi
Cytosol
Antlgenic pepti~: ~ ,f
FIG. 4. Sequential folding and assembly of MHC class I in the endoplasmie reticulum. (a) Nascent MHC class I heavy chain enters the lumen of the ER and is immediately glycosylated and interacts with calnexin (CNX),which stablizes the molecule as foldingoccurs. (b) Calnexin is displaced as heavy chain assembles with/~2-microglobulin. (c) A peptide-loading complex forms between heavy chain-~2-microglobulin heterodimer, tapasin, the TAP peptide transporter (TAP1TAP2), calreticulin (CRT), and ERp57. (d) As peptide loading occurs, higher order structural rearrangements are thought to occur in the ~1 and c~2 domains of heavy chain, resulting in the formation of a peptide-binding cleft, peptide loading, and dissociationof the peptide-loading complex; this is followedby exit of the heterotrimer from the ER.
accepted pathway of events leading to M H C class I folding and assembly in h u m a n cells. However, key questions remain and these will be highlighted.
A. Primary Quality Control and MHC Class I Biogenesis Quality control of M H C class I folding and assembly begins as soon as nascent heavy chain enters the ER. Class I M H C heavy chain undergoes N-linked glycosylation, immediately suggesting that this protein serves as a substrate for CNX/CRT-mediated quality control. I n d e e d this proved to be the case, as several classic investigations showed. It was observed that assembled class I molecules
QUALITYCONTROLIN THE ENDOPLASM1CRETICULUM
255
were rapidly transported to the cell surface. In contrast, the intracellular transport of free heavy chains or peptide-deficient heavy chain-/~2-microglobulin heterodimers is impaired, which suggested an interaction between immature MHC class I and an unknown factor(s). Chemical crosslinking was used to study early events in MHC class I biogenesis in murine cells, and it was shown that nascent heavy chains interacted with an E R-specific 88-kDa protein (12 7), which was subsequently identified as CNX (128). Further analysis suggested a role for CNX in the regulation of MHC class I transport through the secretory pathway (129). But how was this regulation mediated and what role did CNX play, if any? Several lines of evidence had shown in the 1990s that removal of glucose residues from nascent oligosaccharide side chains was required for glycoprotein association with CNX (20, 22, 23). As we have seen, this is achieved through the activity of glueosidases I and II in the ER (see Section II,A and Fig. 1). By expressing murine MHC class I heavy chain in cell lines deficient in glucosidase II, or wild-type cells treated with the glucosidase II inhibitor castanospermine, it was shown that glucose trimming was also a requirement for efficient heavy chain association with CNX (130). Several other interesting, and paradoxical, observations were made during this study. First, although CNX association was perturbed in the absence of glucosidase II, surface expression of assembled MHC class I molecules was the same as in wild-type cells. Second, in the cells deficient in glucosidase II, and hence CNX association, levels of BiP expression were markedly increased. The authors concluded that alternative QC pathways could exist within the ER that have different requirements for removal of glucose residues from nascent side chains. This was partly supported by the observation that BiP associates with immature human class I heavy chains (135). Returning to our general discussion at the beginning, we see that this hypothesis supports overlapping QC mechanisms in the ER that share a high degree of redundancy. A human leukemic cell line, CEM, and a variant of CEM resistant to natural killer (NK) cell-mediated lysis, called CEM-NK R, were used to study further the assembly of hmnan MHC class I assembly. Characterization of CEM-NK ~ by two-dimensional gel electrophoresis revealed that this cell line is completely deficient in CNX. The surt~ace expression of MHC class I in CEM-NK ~ was compared to CEM and it was found that there was no significant difference between CEM and CEM-NK R, indicating that CNX is not absolutely' required for human MHC class I expression (131). So, what was the role of CNX in MHC class I assembly? In other studies around that time (21), it was observed that CNX selectively associated with nascent, incompletely folded, monomeric glycoproteins in a hepatoma cell line. It was found that dissociation from CNX occurred at different rates and that this was related to the time taken for protein folding. Could CNX be playing a role in MHC class I folding? Two studies investigated this by
256
MARKR. FARMERYAND NEILJ. BULLEID
attempting to map regions where CNX associations may take place and analyzing the formation of intrachain disulfide bonds in class I heavy chain (133, 134). Human B cells were transfected with genes encoding either wild-type HLA-A2 heavy chains or mutant heavy chains lacking sites for glycosylation or deficient in binding to 32-microglobulin. CNX did not associate with wild-type heavy chains but bound strongly to mutant heavy chains unable to bind 32-microglobulin. In Daudi cells, deficient in synthesis of 32-microglobulin, wild-type HLA-A2 heavy chains, but not a nonglycosylated mutant, bound CNX. These results led to the conclusion that CNX first recognizes carbohydrate on substrate binding and then binds more stably to peptide determinants, which disappear upon folding (133). To address this hypothesis, these investigators went on to examine the role of CNX in the initial stages of class I folding by examining disulfide bond formation in vivo (134). Mature class I heavy chain comprises two intrachain disulfide bonds (Fig. 3). These form rapidly after insertion of the protein into the ER membrane, and CNX was seen to associate with both reduced and oxidized forms of class I heavy chain during this process. By treating cells with the membrane-permeable reducing agent dithiothrietol, heavy chain disulfide bond formation was blocked, along with association with CNX. Addition of castanospermine slowed the formation of disulfide bonds but did not decrease the amount of assembled heavy chain32-microglobulin complexes that formed. The authors concluded that CNX could promote disulfide-bond formation in class I heavy chains but does not directly facilitate maturation and subsequent binding to 32-microglobulin in human cells. A slightly different story arises in the study of mouse MHC class I association with CNX. Crosslinking and gel permeation analysis showed that CNX remains tightly bound to class I heavy chain in the presence of 32-microglobulin while in the ER (127). In a subsequent study (135), it was determined that the structural basis tbr this difference resides in species-species characteristics of heavy chains itself. Mouse class I heavy chains contain two or three N-linked glyeosylation sites at positions 86 and 176 and, in the latter case, position 256 (136). It was suggested that this could facilitate binding by multiple CNX molecules or strengthen binding of a single C NX molecule by providing additional attachment sites for lectin binding. The introduction of a second glycosylation site in human heavy chain resulted in increased CNX binding in both the absence and presence of 32-microglobulin (138), leading to the suggestion that in the case of human class I, CNX is displaced by/~2-microglobulin binding. These observations go some way to explaining the differences between mouse and human MHC class I assembly and suggest that the location of N-oligosaccharides within proteins can influence their folding and interactions with chaperones such as CNX. Clearly, CNX interactions with MHC class I heavy chains are species dependent. Mouse heavy chain has a prolonged interaction with CNX because
QUALITY CONTROL IN THE ENDOPLASMIC RETICULUM
257
of additional glycosylation sites when compared with human heavy chain. It is now well established that, under normal conditions in the human cell, nascent class I heavy chain is glycosylated at a single site and interacts with CNX after generation of a monoglucosylated intermediate. This stabilizes the protein as it undergoes disulfide bond formation and is displaced when heavy chain noncovalently associates with/~2-microglobulin. Several questions immediately arise. If CNX dissociates, does CRT have a role in human MHC class I assembly, and what mediates disulfide-bond formation? As we have seen, MHC class I assembles and is expressed on the surface of a CNX-deficient cell line (131, 139), suggesting that CNX is not essential. In addition, it was shown that mouse heavy chain-/32-microglobulin dimers interact with CNX (127), but demonstrating this in the human system was not possible. It was hypothesized that other chaperones may functionally replace CNX, and CRT, the soluble homolog of CNX, was an obvious candidate. Using a radiolabeling and coimmunoprecipitation approach, it was demonstrated that a subset of class I molecules associated with CRT and that this population comprised heavy ehain-/32-inicroglobulin dimers (140). In another study, the role of /32-microglobulin in class I association with CNX and CRT was analyzed in a/~2microglobulin-deficient murine cell line (141). Not only was/32-microglobulin required for CRT interaction with heavy chain, but also deglucosylation of N-linked glycan side chains was shown to be important for dissociation of class I proteins from CRT. This was extended in a study of the size of the glycan chain and whether this would influence associations with both CNX and CRT (142). These data showed that CNX interaction with class I proteins having truncated N-glycans was reduced compared to normal class I molecules, whereas assembly with CRT was unperturbed by N-glycan chain length. Along with these and other studies (133, 143-145), distinct roles for CNX and CRT in class I folding and assembly had finally been established. CNX and CRT interact with different MHC class I intermediates, and this seems to be regulated by N-glycan composition and chain length. A picture emerges of highly regulated MHC class I folding and assembly in the ER being initially dependent upon primary QC mechanisms discussed earlier. As nascent chains fold, they associate with CNX and undergo cycles of glucose removal and readdition. Once an intermediate structure is attained, association with/32-microglobulin displaces CNX. The formation of a fi2-mieroglobulin-heavy chain heterodimer seems to then provide a signal for CRT interaction. CRT remains associated with heavy chain-/32-microglobulin until the final stages of class I assembly in the ER. A third character is cast in the drama of primary quality control in the ER, namely ERp57, and recent data suggest a key role for this protein in MHC class I folding. ERp57 has been shown to form a complex with newly synthesized proteins that, like CNX and CRT, is glycoprotein-specific. These studies used
258
MARKR. FARMERYAND NEILI. BULLEID
a crosslinking approach to identify ERp57 as a crosslinking partner that interacted with glycoproteins, forming a complex with CNX or CRT (66, 68-70). Significantly, ERp57 has also been shown to form a complex with CNX and CRT during MHC class I assembly in intact human cells (73, 74). This observation was extended in a kinetic analysis of rodent M HC class I heavy chain interaction with CNX and ERp57, which suggested that there is a delay in the interaction with CNX in comparison to ERp57. The functional significance of these observations has remained unclear. It has been speculated that ERp57 may be involved in stabilization of MHC class I peptide loading complex, peptide trimming, or ER-associated degradation of misfolded glycoproteins (72). What is more likely is that ERp57 regulates disulfide bond formation in nascent glycoproteins (reviewed in Ref. 71). We investigated the role of ERp57 in class I heavy chain disulfide-bond formation in an in vitro experimental system that allowed timing of interactions to be analyzed (74). This showed that ERp57 associated with both unfolded and folded class I complexes, indicating that heavy chain folding occurs during its interaction with ERp57, supporting a role for ERp57 in disulfide-bond formation. It is not unlikely that ERp57 acts as a glycoprotein-specific PDI. It possesses two thioredoxin motifs (WCGHCK), which are important for the isomerization of protein disulfide bonds (149, 151 ); and it has been shown in vitro that, in conjunction with CNX and CRT, ERp57 can catalyze intrachain disulfide-bond formation (75). Although it remains uncertain, it can be suggested that ERp57 is responsible for catalyzing the formation of the disulfide bonds in the class I complex. It is possible that ERp57 may mediate the formation of a disulfide bond in the heavy chain 0t2 domain, which connects the ~2 helix to the ~ sheet, forming the floor of the peptide binding groove (Fig. 3) (152). Can MHC class I biogenesis be used to elucidate events occurring during ERAD? As we have seen, ERAD is intimately involved in primary quality control. The combination of association with CNX and glucose trimming seems to act as a sensor for the generation of correctly folded glycoprotein intermediates. Misfolded intermediates are identified and targeted for degradation, and this process is associated with the trimming of mannose residues in the oligosaccharide side chain (see Section II,C). We have analyzed the ERAD of MHC class I in a reconstituted in vitro system (38). This study showed that degradation is cytosol- and ATP-dependent. In addition, degradation was inhibited in the presence of proteasome inhibitors. These included the protease inhibitor N-acetyl-L-leucyl-L-leucyl-L-norleucyl (ALLN), which binds to the active sites of the proteasome (175), and lactacystin, that covalentlybinds to the unique threonine active sites of the proteasome (176). When CNX association was inhibited by treatment with castanospermine, the rate of degradation was accelerated, suggesting that CNX stabilized heavy chain. Stabilization of heavy chain was also achieved by inhibiting mannose trimming or by removing the class I heavy chain
QUALITYCONTROLIN THE ENDOPLASMICRETICULUM
259
N-linked glycosylation site. This study supported similar previous studies that examined the degradation of MHC class I and established that misfolded MHC class I is removed from the ER to the eytosol, deglyeosylated, and degraded
(177). Clearly, calnexin and mannose trimming are central to ERAD. But how is dislocation mediated? Clues to the answer of this question came from the study of virally infected cells themselves. It is well established that MHC class I plays a central role in antiviral immune response. As this defense mechanism has evolved in mammalian cells, so viruses have developed a variety of mechanisms to evade MHC class I detection (169, 178). Certain viruses inhibit surface expression of MHC class I complexes. In cells infected by human eytomegalovirus (HCMV), this process involves several mechanisms. The HCMV gene US6 encodes a 22-kDa glycoprotein that binds to TAP/class I complex (see Section V,B) and inhibits the translocation ofpeptide from the eytosol (179). In another mechanism, the rapid degradation of class I molecules is mediated by the HC MV genes US2 and US11, which encode ER-resident glyeoproteins (181, 184). Study of this process has provided clues to the general mechanisms of ERAD. Pulse-chase analysis in cells expressing either US2 or USll revealed that both could induce rapid dislocation of newly synthesized MHC class I from the ER to the cytosol via the Sec61 complex (33). More recent investigations dissected US2/USll-mediated dislocation and degradation further. First, it was shown that the effects of US2/USll on class I dislocation could be inhibited by agents that affect intracellular redox potential and/or free thiol status (183). This suggested that the dislocation process was multifactorial and could involve components in both the eytosol and the ER lumen, such as molecular chaperones, and implied that reduction of intrachain disulfide bonds could be important. A second study supported this hypothesis. Removal of the eytosolic tail of MHC class I revealed that this domain was required for dislocation to the eytosol and subsequent degradation, but not for US2/USll association (184). This suggested that US2/USll alone do not have the capacity to mediate extraction from the ER. The authors presented a model in which US2/US11 associate, via lumenal domains, with MHC class I and deliver the complex to the transloeon. Once inside the translocon, heavy chain unfolds and is then dislocated from the ER. The cytosolic tail is required for this dislocation process. Cytosolic factors, such as Hsp70, could be involved in extraction, either by providing energy for the reaction or by acting as a molecular ratchet to prevent heavy chain from sliding back into the ER lumen. This is followed by deglyeosylation by N-glycanase, as the presence of an oligosaccharide side chain would sterieally hinder entry into the proteasome. In contrast to previous data, ubiquitination does not appear to be required to initiate class I dislocation. Although a ubiquitinated class I intermediate is formed during U S2/US 11-dependent dislocation, this does not arise until dislocation is initiated,
260
MARKR. FARMERYAND NEIL]. BULLEID
and mutagenesis of class I cytosolic tail lysine residues (the sites ofubiquitination) does not prevent dislocation (185). The mechanisms of dislocation from the ER to cytosol remain unclear. What has been established is that dislocation and degradation are tightly coupled. We have shown that, in the presence of proteasome inhibitors, MHC class I substrates that would normally be degraded accumulate in the ER and not in the cytosol (38). This contrasts with the situation in virally infected cells; inhibition of the proteasome results in dislocation of heavy chain into the cytosol, where it appears as a soluble, deglycosylatedprotein (33). However, based on our findings and the model suggested above, it is probable that the proteasome must be associated with the ER membrane, and this is likely to involve accessory factors. How this association is maintained is not known. From these studies and those described in Section II,C, the general process of ERAD can be summarized as follows: (1) Either misfolded proteins or unassembled subunits are recognized and targeted for transport to the site of degradation; (2) polypeptides are dislocated from the ER to the cytosol through the Sec61 translocation apparatus; (3) the polypeptides are either degraded immediately or are ubiquitinated or deglycosylated prior to degradation by the proteasome. Consensus is emerging as to events that occur in the ER lumenal environment during ERAD. However, key questions remain to be answered. Although we have gone some way toward determining how this process is regulated, precise mechanisms remain to be identified. In addition, the role of accessory proteins in targeting substrates to the translocon for dislocation is unclear.
B. Protein-Specific Factors and MHC Class I Assembly We have described how class I heavy chain enters the ER, undergoes disulfide-bond formation, and associates CNX prior to assembly with fi2-microglobulin. This association with calnexin can be viewed as a primary QC system, whereas subsequent association with class I-specific proteins constitutes the secondary QC system. Once formed, the fi2-microglobulin-heavy chain heterodimers interact with both CRT and ERp57 to form a peptide-loading complex. The maturation of this complex involves the MHC class I-specific factors tapasin and transporter associated with antigen processing (TAP). The third component in fully assembled MHC class I is the antigenic peptide. These are generated in the cytosol of infected or malignant cells by proteasomal degradation of foreign antigens and translocated into the lumen of the ER by means of TAP, a specialized transporter and member of the ATP-binding cassette transporter superfamily. The TAP complex (153) consists of two related subunits, TAP 1 and TAP 2, and MHC class I interacts with both of these subunits (156). MHC class 1-specific peptides are generated in the cytosol by the proteasome. Briefly, the multisubunit proteasome contains two sets of three catalytic
QUALITYCONTROLIN THE ENDOPLASMICRETICULUM
261
subunits that, during an immune response, are replaced by interferon-induced homologous subunits (163). This exchange creates an immunoproteasome that displays changes in proteolytic cleavage specificity, resulting in the production of class I-specific peptides. These associate with the MHC-encoded TAP transporter in an ATP-independent manner, and are transported across the ER membrane following ATP hydrolysis (164-166). However, the specificity of peptide binding and transport by TAP varies from species to species, human TAP being the most promiscuous (167, 168). It would seem that this coincides with the spectrum of peptides bound by MHC class I molecules in a given species, suggesting a coevolution between TAP and class I molecules (169). A broader discussion of the generation and transport of class I-specific peptides is beyond the scope of this chapter, but is reviewed elsewhere (170, 171). Class I heterodimers associate with TAP prior to peptide loading (162) and this association is mediated by the class I-specific protein tapasin. The original discovery of tapasin and its interaction with MHC class I was made when it was copurified as part of a complex, including CRT, with TAP (104, 140). Tapasin is a 48-kDa transmembrane glycoprotein that is part of the immunoglobulin superfamily and has a probable ER retention signal located in its cytosolic domain. It is thought that tapasin plays a central intermediary role in bringing together the components of the class I peptide loading complex, as heavy chainfl2-microglobulin heterodimers, CRT, and ERp57 do not associate with TAP in the absence of tapasin (140, 147, 154). Tapasin can bind to heavy chainfl2-microglobulin heterodimers and TAP independently, and it is postulated that tapasin acts as a linker protein that brings together the complete peptide loading complex. Analysis of deletion mutants revealed that the N-terminal domain of tapasin interacts with MHC class I and the C-terminal domain interacts with TAP (155). Once assembled, up to four peptide-loading complexes can associate with a single TAP molecule (104). How peptide loading is achieved and the fimctions of TAP and tapasin are areas of intense study and controversy. The precise role oftapasin in MHC class I assembly has remained elusive, but two studies provided initial insight (157, 158). The first investigation generated a soluble tapasin molecule that is missing the 35 C-terminal amino acids of the protein, which includes that transmembrane domain and the cytosolic tail. When this molecule was expressed in the tapasin-defieient .220 cell line, it did not associate with TAP; therefore, class I heavy chains expressed in the presence of soluble tapasin in .220 do not associate with TAE If TAP were critical for peptide loading, one would expect impaired class I expression under these conditions. Surprisingly, in the presence of these molecules, class I expression was restored to wild-type levels. Class I molecules were also shown to mature twice as quickly in the presence of soluble tapasin as compared to full-length tapasin, presumably because class I molecules are no longer retained in the ER by their association with TAE This study also showed that peptides loaded into class I complexes
262
MARKR. FARMERYAND NEILJ. BULLEID
expressed in the presence of soluble tapasin were indistinguishable from those isolated from class I molecules expressed in the presence of wild-type tapasin. Finally, soluble tapasin could restore class I-mediated CTL recognition. Clearly, this study brings into question the physiological role of the class I association with TAP. The situation is further complicated as not only the role of TAP, but also the significance of tapasin, is questioned in the second study (158). In this ease, the expression and function of three class I molecules were analyzed in .220 cells in the presence and absence of tapasin. These alleles showed striking variability in their association with TAP and dependence on tapasin for their expression and activity. This study partially suggested that, although peptides could be bound in the absence of tapasin, they were of low stability, and that in the presence of tapasin a different peptide repertoire was bound. Both studies were supported by the observation that HLA-A2 is capable of presenting viral antigens in the absence of tapasin (159). However, a proportion of molecules was released at the cell surface in an unassembled, or "peptide-reeeptive" form, indicating that molecules were not assembled as efficiently as in wild-type cells. Evidence emerged later for successive peptide binding stages during class I assembly (160). A T134K point mutation in HLA-A2 results in the assembly of class I molecules that fail to bind to TAP. This mutation also disrupts interaction between class I and CRT. T134K molecules did not present viral antigens to T cells, even though they bound peptides and/~2-microglobulin in vitro. The resulting effect of this mutation was to mediate the rapid exit from the ER of "peptide-empty" MHC class I complexes, unlike empty wild-type molecules, which are retained and degraded. Paradoxically, the rapid exit of empty T134K molecules was dependent on a supply of TAP-derived peptides, suggesting that the mutation acts to prevent class I from optimizing the binding of their peptide ligand, rather than preventing peptide binding. In other words, class I molecules are loaded with class I-specific peptide, but the T134K mutation prevents a retention and editing step which would normally result in the expression of optimal peptide on the cell surface. A secondary QC step that is dependent upon MHC class 1-specific protein factors. The authors went on to suggest that MHC class I assembly with peptide is a two-stage process: initial binding of suboptimal peptides, followed by peptide optimization that depends on ER retention. Recent data also suggest that tapasin plays a role in peptide association with TAP (161). Peptide transport was analyzed in tapasin-deficient .220 cells, which showed that peptide binding to TAP was severely diminished, although the transport rate of bound peptides was normal. Efficient peptide association was restored by transfection of tapasin into .220 cells. The authors suggest that tapasin may stabilize the peptide-binding site on TAP, extending its role in secondary quality control. It can be suggested that this stage of MHC class I quality control
QUALITYCONTROLIN THE ENDOPLASMICRETICULUM
263
involves retention in the ER until optimal peptide loading occurs. A view supported in a study that utilized green fluorescent protein-tagged class I showed that peptide-loaded class I can be retained in the ER for peptide optimization
(173). Clearly, the role of tapasin extends beyond its function as a structural "keystone" in the class I peptide-loading complex. TAP probably encompasses a wider function than peptide transport. We have seen that it is involved in selection of peptides, but in terms of quality control, TAP may also mediate ER retention of MHC class I loaded with suboptimal peptide in association with tapasin. Taken together, current data suggest that tapasin is also involved in the efficiency of peptide loading through peptide editing or exchange, and that this function regulates the transport of assembled MHC class I to the cell surface. What remains unclear is whether tapasin is involved directly in peptide editing, through proteolytic trimming for example, or if this function is carried out by another factor, such as ERp57. However, it is clear that through the coordinated activities of several filnctionally diverse proteins specific for MHC class I, only fully assembled, optimally peptide-loaded complexes are released from the ER to the cell surface.
VI. Conclusionsand Perspectives The concerted efforts of cell biology, genetics, biochemistry, and molecular biology over recent years have provided profound insights into the highly complex world of protein folding in the ER. We have seen that intricate and regulated mechanisms facilitate the efficient production of folded and assembled proteins. Protein folding in the ER is continually monitored, ensuring the fidelity of proteins that are released from the ER, suggesting that folding and assembly on the one hand are balanced with degradation of misfolded proteins on the other. What is clear is that when these processes go wrong, serious pathological states can result, such as cystic fibrosis and Alzheimer's disease. Further dissection of QC processes in the ER will not only facilitate understanding of normal cellular mechanisms but may also allow therapeutic intervention in protein folding-related diseases. Quality control in the ER is by no means fully understood. Many questions are unanswered. The precise roles of components such as ERp57 remain unknown. A lack of clarity surrounds the mechanisms regulating primary quality control, in particular the actions of CNX/CRT and UGGT, and the process of dislocation in ERAD. It is highly probable that many, as yet unknown factors are involved in both primary mechanisms and protein specific quality control. The publication of the sequence of the human genome should facilitate the identification of factors associated with quality control, degradation, and the ER
264
MARK R. FARMERY AND NEIL J. BULLEID
in general. F i t t i n g t h e pieces t o g e t h e r in this m o l e c u l a r jigsaw will c o n t i n u e to p r o v i d e an exciting a n d i n t r i g u i n g c h a l l e n g e for biology in t h e 21st century.
REFERENCES 1. 2. 3. 4. 5.
M.J. Gething and J. Sambrook, Nature (London) 355, 33-45 (1992). L. Ellgaard, M. Molinari, and A. Helenius, Science 286, 1882-1888 (1999). S. M. Hurtley and A. Helenius, Annu. Rev. Cell Biol. 5, 277-307 (1989). R. D. Klausner and R. Sitia, Cell (Cambridge, Mass.) 62, 611-614 (1990). D. N. Hebert, J. F. Simons, J. R. Peterson, and A. Helenius, Cold Spring Harb. Syrup. Quant. Biol. 60, 405-415 (1995). 6. E J. van der Wal, J. D. Oliver, and S. High, Eu~ J. Biochem. 256, 51--59. 7. J. w. Brewer and R. B. Corley, J. Cell Sc/. 109, 2383-2392 (1996). 8. R. Kornfeld and S. Kornfeld, Annu. Rev. Biochem. 54, 631~564 (1985). 9. E. S. Trombetta and A. Helenius, J. Cell Biol. 148, 1123-1129 (2000). 10. M. C. Sousa and A. J. Parodi, EMBOJ. 14, 4196--4203 (1995). 11. M. C. Sousa, M. A. Ferrero-Garcia, and A. J. Parodi, Biochemistry 31, 97-105 (1992). 12. B. Bukau and A. L. Horwich, Cell (Cambridge, Mass.) 92, 351-366 (1998). 13. I. Wada, D. Rindress, P. H. Cameron, W. J. Ou, j. j. Doherty 2d, D. Louvard, A. W. Bell, D. Dignard, D. Y. Thomas, and J. j. Bergeron, J. Biol. Chem. 266, 19599-19610 (1991). 14. M. J. Coppolino and S. Dedhar, Int. J. Biochem. Cell BioL 30, 553-558 (1998). 15. D. Watanabe, K. Yamada, Y. Nishina, Y. Tajima, U. Koshimizu, A. Nagata, and Y. Nishimune, J. Biol. Chem. 269, 7744-7749 (1994). 16. S. Ohsako, Y. Hayashi, and D. Bunick,]. Biol. Chem. 269, 14140-14148 (1994). 17. E. Chevet, C. A. Jakob, D. Y. Thomas, and J. J. Bergeron, Semin. Cell Dev. Biol. 10, 473-480
(1999). 18. E. Chevet, H. N. Wong, D. Gerber, C. Cochet, A. Fazel, P. H. Cameron, J. N. Gushue, D. Y. Thomas, and J. J. Bergeron, EMBOJ. 18, 3655--3666 (1999). 19. K. H. Krause and M. Michalak, Cell (Cambridge, Mass.) 88, 439-443 (1997). 20. D. N. Hebert, B. Foellmer, and A. Helenius, EMBOJ. 15, 2961-298 (1996). 21. W. J. Ou, P. H. Cameron, D. Y. Thomas, and J. J. Bergeron, Nature (London) 364, 771-776
(1993). 22. 23. 24. 25. 26. 27. 28. 29.
C. Hammond, I. Braakman, and A. Helenius, Proc. Natl. Acad. Sci. U.S.A. 91,913--917 (1994). K. P. Kearse, D. B. Williams, and A. Singer, EMBOJ. 13, 3678-3686 (1994). D. N. Hebert, B. Foellmer, and A. Helenius, Cell (Cambridge, Mass.) 81,425~433 (1995). A. Ora and A. Helenius, J. Biol. Chem. 270, 26060-26062 (1995). W. M. Nauseef, S. J. McCormick, and R. A. Clark, J. Biol. Chem. 270, 4741-4747 (1995). K. S. Cannon and A. Helenius, J. Biol. Chem. 274, 7537-7544 (1999). J. R. Peterson, A. Ora, P. N. Van, and A. Helenius, Mol. Biol. Cell 6, 1173-1184 (1995). Y. Saito, Y. Ihara, M. R. Leach, M. E Cohen-Doyle, and D. B. Williams, EMBOJ. 18, 6718-6729
(1999). 30. Y. Ihara, M. F. Cohen-Doyle, Y. Saito, and D, B. Williams, Mol. Cell 4, 331-341 (1999). 31. F. E. Ware, A. Vassilakos, P. A. Peterson, M. R. Jackson, M. A. Lehrman, and D. B. Williams, J. Biol. Chem. 270, 4697-4704 (1995). 32. D. B. Williams, Biochem. Cell Biol. 73, 123-132 (1995). 33. E. J. Wiertz, D. Tortorella, M. Bogyo, J. Yu, W. Mothes, T. R. Jones, T. A. Rapoport, and H. L. Ploegh, Nature (London) 384, 432-438 (1996). 34. M. Pilon, R. Schekman, and K. RSmiseh, EMBOJ. 16, 4540-4548 (1997).
QUALITY CONTROL IN THE ENDOPLASMIC RETICULUM
265
35. C. A. Jakob, P. Burda, J. Roth, and M. Aebi, J. Cell Biol. 142, 1223-1233 (1998). 37. M. de Virgilio, C, Kitzmtiller, E. Schwaiger, M. Klein, G. Kreibich, and N. E. Ivessa, Mol. Biol. Cell 10, 4059-4073 (1999). 38. C. M. Wilson, M. R. Farmery, and N. J. Bulleid, j. Biol. Chem. 275, 21224-21232 (2000). 39. Y. Liu, P, Choudhury, C. M. Cabral, and R. N. Sifers, J. Biol. Chem. 272, 7946-7951 (1997). 40. Y. Liu, P. Choudhury, C. M. Cabral, and R. N. Sifers, J. Biol. Chem. 274, 5861-5867 (1999). 42. K. Su, T. Stoller, J. Rocco, J. Zemsky, and R. Green, J. Biol. Chem. 268, 14301-14309 (1993). 43. M. Knop, N. Hauser, and D. H. Wolf, Yeast 12, 1229-1238 (1999). 44. M. Zhou and R. Schekman, Mol. Cell 4, 925-934 (1999). 45. A. Rapak, P. O. Falnes, and S. Olsnes, Proc. Natl. Acad. Sci. U.S.A. 94, 3783~3788 (1997). 46. T. Suzuki, H. Park, K. Kitajima, and W J. Lennarz, J. Biol. Chem. 273, 21526-21530 (1998). 47. Z. Beb6k, C. Mazzoehi, S. A. King, J. S. Hong, and E. J. Sorscher,]. Biol. Chem. 273, 2987348. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75.
29878 (1998). j. Chillar6n and I. G. Haas, Mol. Biol. Cell 11,217-226 (2000). R. R. Kopito, Cell (Cambridge, Mass.) 21,427-430 (1997). M. Pilon, R. Schekman, and K. R6misch, EMBO]. 16, 4540-4548 (1997). E.J. Wiertz, D. Tortorella, M. Bogyo, J. Yu, w Mothes, T. R. Jones, T. A. Rapoport, and H. L. Ploegh, Nature (London) 384, 432-438 (1996). M. M. Hiller, A. Finger, M. Schweiger, and D. H. Wolf, Science 273, 1725-1781 (1997). T. Biederer, C. Volkwein, and T. Sommer, Science 278, 1806-1809 (1997). D. Qu, J. H. Teckman, S. Omura, and D. H. Perlmutter, J. Biol. Chem. 271, 22791-22795 (1996). R. K. Plemper, S. Bohmler, J. Bordallo, T. Sommer, and D. H. Wolf, Nature (London) 388, 891495 (1997). R. Y. Hampton, R. G. Gardner, and J. Rine, Mol. Biol. Cell 7, 2029-2044 (1996). J. Bordallo, R. K. Plemper, A. Finger, and D. H. Wolf, Mol. Biol. Cell 9, 209-222 (1998). R. Halaban, E. Cheng, Y. Zhang, G. Moellmann, D. Hanlon, M. Michalak, V. Setaluri, and D. N. Hebert, Proc. Natl. Acad. Sci. U.S.A. 94, 6210-6215 (1997). J. B. Huppa and H. L. Ploegh, Immunity 7, 113-122 (1997). H. Yu and R. R. Kopito, J. Biol. Chem. 274, 36852~368528 (1997). K. E. Matlaek, B. Misselwitz, K. Plath, and T. A. Rapoport, Cell (Cambridge, Mass.) 97, 553564 (1999). j. L. Brodsky and A. A. MeCraeken, Semin. Cell Dev. Biol. 10, 507-513 (1999). J. L. Brodsky, E. D. Werner, M. E. Dubas, J. L. Goeckeler, K. B. Kruse, and A. A. McCracken, J, Biol. Chem. 274, 3453-3460 (1999). M. H. Glickman, D. M. Rubin, V. A. Fried, and D. Finley, Mol. Cell Biol. 18, 3149-3162 (1998). K. Leonhard, A. Stiegler, W. Neupert, and T. Langer, Nature (London) 398, 348-351 (1999). j, D. Oliver, H. L. Roderiek, D. H. Llewellyn, and S. High, Mol. Biol. Cell 10, 2573-2525 (1999). R. B. Freedman, rE R. Hirst, and M. F. Tuite, Trends Biochem. Sci. 19, 331~336 (1994). J. D. Oliver, F. j. van der Wal, N. J. Bulleid, and S. High, Science 275, 86~8 (1997). J. G. Elliott, J. D. Oliver, and S. High, J. Biol. Chem. 272, 13849-13855 (1997). F. J. van der Wal, J. D. Oliver, and S. High, Eur J. Biochem. 256, 51-59 (1998). S. High, E J. Lecomte, S. Russell, B. M. Abell, andJ. D. Oliver, FEBS Lett. 476, 38-41 (2000). N. A. Mortice and S. J. Powis, Curt Biol. 8, 713-716 (1998). E. A. Hughes and P. Cresswell, Curt Biol. 8, 709-712 (1998). M. R. Farmery, S. Allen, A. J. Allen, and N. J. Bulleid, J. Biol. Chem. 275,14933-14938 (2000). A. Zapun, N. J. Darby, D. C. Tessier, M. Michalak, J. j. Bergeron, and D. Y. Thomas, f. Biol. Chem. 273, 6009-6012 (1998).
266
MARK R. FARMERY AND NEIL J. BULLEID
76. M. Mollinari and A. Helenius, Nature (London) 402, 90-93 (1999). 77. E Kappeler, D. R. Klopfenstein, M. Foguet, J, P. Paccaud, and H. P. Hanri,J. Biol. Chem. 272, 31801~31808 (1997). 78. N. Nishimura and W. E. Balch, Science 277, 556~558 (1997). 79. W. E. Balch, J. M. McCaffery, H. Plutner, and M. G. Farquhar, Cell (Cambridge, Mass. ) 76, 841-852 (1994). 80. M. Aridor, ]. Weissman, S. Bannykh, C. Nuoffer, and W. E. Balch, J. Cell Biol. 141, 61-70 (1998). 81. M. Aridor, S. I. Bannylda, T. Rowe, and W. E. Balch, J. Biol. Chem. 274, 4389~399 (1999). 82. M. J. Kuehn, J. M. Herrmann, and R. Schekman, Nature (London) 391, 187-190 (1998). 83. S. Springer and R. Schekman, Science 281,698-700 (1998). 84. M. Aridor and W. E. Balch, Trends. Cell Biol. 6, 315 (1996). 85. W. C. Nichols, U. Seligsohn, A. Zivelin, V. H. Terry, C. E. Hertel, M. A. Wheatley, M. J. Moussalli, H. P. Hauri, N. Ciavarella, R. J. Kaufman, and D. Ginsburg, Cell (Cambridge, Mass.) 93, 61-70 (1998). 86. S. Springer, A. Spang, and R. Schekman, Cell (Cambridge, Mass.) 97, 145-148 (1999). 87. H. P. Hauri, F. Kappeler, H. Andersson, and C. Appenzeller, J. Cell Sci. 113, 587~596 (2000). 88. K. Nagata, Trends Biochem. Sci. 21, 22-26 (1998). 89. L. Zhen, H. Baumann, E. K. Novak, and R. T. Swank, Arch. Biochem. Biophys. 304, 402-414
(1993). 90. S. Maelntyre, D. Samols, and P. Dai|ey, J. Biol. Chem. 269, 24496-24503 (1994). 91. G. Bu, H. J. Geuze, G. J. Strons, and A. L. Schwartz, EMBOJ. 14, 2269-2280 (1995). 92. M. Marzioch, D. C. Henthorn, J. M. Herrmann, R. Wilson, D. Y. Thomas, J. J. Bergeron, R. C. Solari, and A. M. Rowley, Mol. Biol. Cell I0, 1923-1938 (1999). 93. K. J. Roberg, M. Crotwe]l, P. Espenshade, R. Gimeno, and C. A. Kaiser, J. Cell Biol. 145, 659-672 (1999). 94. J. Powers and C. Barlowe, J. Cell Biol. 142, 1209-1222 (1998). 95. J. A. Trilla, A. Duran, and C. Roncero, J. Cell Biol. 145, 1153-1163 (1999). 96. L. A. Graham, K. ]. Hill, and T. H. Stevens, J. Cell Biol. 142, 39-49 (1998). 97. P.W. Sherwood and M. Carlson, Proc. Natl. Acad. Sci. U.S.A. 96, 7415-7420 (1999). 98. W. P. Barz and P. Walter, Mol. Biol. Cell 10, 1043-1059 (1999). 99. P.-O. Ljungdalfl, C. J. Gimeno, C. A. Styles, and G. R. Fink, Cell (Cambridge, Mass.) 71, 463-478 (1992). 100. N. D. Dwyer, E. R. Troemel, P. Sengupta, and C. I. Bargmann, Cell (Cambridge, Mass.) 93, 455-466 (1998). 101. A. R. Walmsley, M. R. Batten, U. Lad, and N. J. Bulleid, J. Biol. Chem. 274, 14884-14892 (1999). 102. W. G. Annaert, B. Becker, U. Kistner, M. Reth, and R. Jahn, J. Cell Biol. 139, 1397-1410 (1997). 103. P. Cresswell, Cell (Cambridge, Mass.) 84, 505-507 (1996). 104. B. Ortmann, J. Copeman, P. j. Lehner, B. Sadasivan, J. A. Herberg, A. G. Grandea, S. R. Riddell, R. Tampe, T. Spies, J. Trowsdale, and P. Cresswell, Science 277, 1306-1309 (1997). 105. E. K. Baker, N. J. Colley, and C. S. Zuker, EMBOJ. 13, 4886-4895 (1994). 106. J. F. Fleming, G. M. Spitsen, T. Y. Hui, L. Olivier, E. Z. Du, M. Raabe, and R. A, Davis,J. Biol. Chem. 274, 9509-9514 (1999). 107. A. van der Spoel, E. Bonten, and A. Azzo, EMBOJ. 17, 1588-1597 (1998). 108. Y.-T. Chen, D. B. Stewart, and W. J, Nelson, J. Cell Biol. 144, 687-699 (1999). 109. M. Aridor, S. I. Bannykh, T. Rowe, and W E. Balch, J. Cell Biol. 131,875~93 (1995). 110. j. Klumperman, A. Schweizer, H. Clausen, B. L. Tang, W. Hong, V. Oorschot, and H. P. Hauri, J. Cell Sci. 111, 3411-3425 (1998).
QUALITY CONTROL IN THE ENDOPLASMIC RETICULUM
267
111. E Kappeler, D. R. Klopfenstein, M. Foguet, andJ. E Paccaud, and H. R f Biol. Chem. 272,
31801-31808 (1997). 112. F. SchimmOller, B. Singer-Kruger, S. Schroder, U. Kruger, C. Barlowe, and H. Riezman, EMBO]. 14, 1329-1339 (1995). 113. M. Dominguez, K. Dejgaard, J. Fullekrug, S. Dahan, A. Fazel, J. P. Paecaud, D. Y. Thomas, J. J. Bergeron, and T. Nilsson, J. Cell Biol. 140, 751-765 (1998). 11,5. J. s. Bonifacino, C. K. Suzuki, and R. D. Klausner, Science 247, 79--82 (1990). 116. F. Letourneur, S. Hennecke, C. Demolliere, and P. Cosson,J. Cell Biol. 129, 971-978 (1995). 117. N. Zerangue, B. Schwappach, and Y. N, Jan, Neuron 22, 537-548 (1999). 118. X. B. Chang, L. Cui, Y. X. Hou, T. J. Jensen, A. A. Aleksandrov, A. Mengos, and J. R. Riordan, Mol. Cell 4, 137-142 (1999). 119. C. Hammond and A. Helenius,J. Cell Biol. 126, 41-52 (1994). 120. R. M. Zinkernagel and P. C. Doherty, Nature (London) 251,547-548 (1974). 121. P. J. BjSrkmau, M. A. Saper, B. Samraoui, W. S. Bennett, J. L. Strominger, and D. C. Wiley, Nature (London) 329, 506~512 (1987). 122. T. H. Hansen, K. Ozato, and D. H. Sachs, Adv. Immunol. 34, 39-70 (1993). 123. J. Klein, "Biology of the Mouse Histocompatibility-2 Complex." Springer, New York, 1975. 124. R. N. Germain and D. H. Margulies, Annu. Rev. hnmunol. 11,403~t50 (1993). 127. E. Degen and D. B. Williams,]. Cell Biol. 112, 1099-1115 (1991). 128. N. Ahluwalia, J. j. Bergeron, I. Wada, E. Degen, and D. B. Williams, ]. Biol. Chem. 267,
10914-10918 (1992). 129. M. R. Jackson, M. F, Cohen-Doyle, P. A. Peterson, and D. B. Williams, Science 263, 384-387
(1994). J. P. Balow, J. D. Weissman, and K. P. Kearse, J. Biol. Chem. 270, 29025-29029 (1995). J. E, Scott and J. R. Dawson,]. hnmunol. 155, 143-148 (1995). Q. Zhang, M, Tector, and R. D. Salter,]. Biol. Chem. 270, 3944-3948 (1995). M. Tector and R. D. Salter,]. Biol. Chem. 270, 19638-19642 (1995). E. NSBner and P. Parham, J. Exp. Med. 181,327-337 (1995). L. W. Malloy, Immunol. Res. 6, 11-29 (1987). Q. Zhang and R. D. Salter, J. Immunol. 160, 831-837 (•998). B. K, Sadasivan, A. Cariappa, G. L. Waneck, and P. Cresswell, CoM Spring Harb. Syrup. Quant. Biol. 60, 267-275 (1995). 140. B. Sadasivan, P. J. Lehner, B. Ortmann, T. Spies, and P. Cresswell, hnmunity 5, 103-114 (1996). 141. J. E. van Leeuwen and K. E Kearse, Proc. Natl. Acad. Sci. U.S.A. 93, 13997-4001 (1996). 142. M.J. Bennet and K. P. Kearse, Biochem. Biophys. Res. Commun. 262, 473-478 (1999). 143. A. Vassilakos, M. F. Cohen-Doyle, P. A. Peterson, M. R. Jackson, and D. B. Williams, EMBOJ. 15, 1495596 (1996). 144. F. E. Ware, A. Vassilakos, P. A. Peterson, M. R. Jackson, M. A. Lehrman, and D. B. Williams, ]. Biol. Chem. 270, 4697-4704 (1995). 145. W. K, Suh, E. K. Mitchell, Y. Yang, E A. Peterson, G. L. Waneck, and D. B.,J. Exp. Med. 184, 337-348 (1996). 149. C. F. Bennett, J. M. Balearek, A. Varriehio, and S. T. Crooke, Nature (London) 334, 268-270 (1988). 1,51. S. P. Srivastava, N. Q. Chen, Y. X. Liu, and j. L. Holtzman,]. Biol. Chem. 266, 20337-20344 (1991). 152. F. Cresswell, B. Arunachalam, N. Baugia, T. Dick, G. Diedrich, E. Hughes, and M. Maric, Immunol. Res. 19, 191-200 (1999). 153. B. Lankatt-Buttgereit and R. Tamp6, FEBS Lett. 464, 108-112 (1999). 154. J. C. Solheim, B. M. Carreno, and T. H. Hansen, J. Immunol. 158, 541-543 (1997). 130. 131. 133. 134. 135. 136. 138. 139.
268
MARK R. FARMERY AND NEIL J. BULLEID
155. N. Bangia, E J. Lehner, E. A. Hughes, M. Surman, and P. Cresswell, Eur J. Immunol. 29,
1858-1870 (1999). 156. s. j. Powis, Eur J. Immunol. 27, 2744-2747 (1997). 157. P.J. Lehner, M. J. Surman, and P. Cresswell, Immunity 8, 221-231 (1998). 158. C. A. Peh, S, R. Burrows, M. Barnden, R. Khanna, P. Cresswell, D. J. Moss, and J. McCluskey, Immunity 8, 531--542 (1998). 159. J.W. Lewis, A. Sewell, D. Price, and T. Elliott, Eur. J. Immunol. 28, 3214-3220 (1998). 160. J. w. Lewis and T. Elliott, Curr. Biol. 8, 717-720 (1998). 161. S. Li, K. M. Paulsson, S. Chen, H.-O. SjSgren, and E Wang, J. Biol. Chem. 275, 1581-1586
<2000). 162. B. M. Carreno, J. C. Solheim, M. Harris, I. Stroynowski, J. M. Connolly, and T. H. Hansen, J. lmmunol. 155, 4726-4733 (1995). 163. K. Friih and Y. Yang, Cu~. Opin. Immunol. 11, 76-81 (1999). 164. J.J. NeeOes, 17. Momburg, and G. J. Hammerling, Science 261, 769-771 (1993). 165. P. M. van Endert, Immunol. Res. 15, 265-279 (1996). 166. M. J. Androlewicz and P. Cresswell, Immunity 1, 7-14 (1994). 167. P. M. van Endert, D. Riganelli, G. Greco, K. Fleischhauer, J. Sidney, A. SeRe, and J. F. Bach, J. Exp. Med. 182, 1883-1895 (1995). 168. F. Momburg, J. J. Neefjes, and G. ]. Hammerling, Cu~. Opin. Immunol. 6, 32-37 (1994). 169. K. Friih, A. Gruhler, R. M. Krishna, and G. J. Schoenhals, Immunol. Rev. 168,157-166 (1999). 170. I. A. York, A. L. Goldberg, X. Y. Mo, and K. L. Rock, ImmunoL Rev. 172, 49-66 (1999). 171. Y. Wang, D. Chen, and M. J. Androlewiez, Immunol. Rev. 172, 67-72 (1999). 172. Y. Li, L. Salter-Cid, A. Vitiello, T. Preckel, J. D. Lee, A. Angulo, Z. Cai, E A. Peterson, and Y. Yang,J. Biol. Chem. 275, 24130-24135 (2000). 173. D. Marguet, E. T. Spiliotis, T. Pentcheva, M. Lebowitz, J. Schneck, and M. Edidin, Immunity
11,231-240 (1999). 175. J. Lowe, D. Stock, B. Jap, E Zwickl, W. Baumeister, and R. Huber, Science 268, 533-539
(1995). 176. G. Fenteany, R. F. Standaert, W. S. Lane, S. Choi, E. J. Corey, and S. L. Schreiber, Science
268, 726-731 (1995). 177. E. A. Hughes, C. Hammond, and P. Cresswell, Proc. Natl. Acad. Sci. U.S.A. 94, 1896-1901
(1997). 178. H. L. Ploegh, Science 280, 248-253 (1998). 179. P.J. Lehner, J. T. Karttunen, G. W. G. Wilkinson, and P. Cresswell, Proc. Natl. Acad. Sci. U.S.A.
94, 6904-6909 (1997). 181. T. R. Jones and L. Sun, J. Virol. 71, 2970-2979 (1995). 183. D. Tortorella, C. M. Story, J. B. Huppa, E. J. Wiertz, T. R. Jones, and H. L. Ploegh,J. Cell Biol.
142, 365~376 (1998). 184. C. M. Story, M. H. Furman, and H. L. Ploegh, Proc. Natl. Acad. Sci. U.S.A. 96, 8516-8521
(1999). 185. C. E. Shamu, C. M. Story, T. A. Rapoport, and H. L, Ploegh, J. Cell Biol. 147, 45~57 (1999). 186. M. A. Saper, P. J. Bj6rkman, and D. C. Wiley, J. Mol. Biol. 219, 277-319 (1991).
Index
Antisense DNA msDNAs, 86-87 Antisense oligodeoxynucleotide, 219 APCs, 217 Apoptosis cisplatin, 116-117 iPLAefl, 24-25 AR, 57 Arachidonate glucose-stimulated, 22 Arachidonic acid (AA), 12 A-Raf, 211 Asparagine-linked glycosylation, 237-238 ATP-binding domains iPLA2fl, 8-10
A AA, 12 AAG proteins cisplatin, 114-115 AB-GP205 ankyrin-repeat domains, 12 Acetylhydrolase PAF, 4 group VIA PLAe, 21-22 group VIIIA, 7 ACK 1/2, 201 biologic'a] functions, 197 drug design, 222 Activating regions (AR), 57 Acute lymphoblastic leukemia (ALL), 135 Alkylphosphocholines (APCs), 217 ALL, 135 Alternative splicing iPLAe/3, 17-19 Amiloride-sensitive sodium channel ankyrin-repeat domains, 12 Anaerobic CO-oxidizing microbes, 37 Anemia sickle cell, 179 Ankyrin brain, 12 epithelial, 12 erythroid, 12 iPLAe, 13 lymphocyte, 12 Ankyrin-binding glycoprotein 205 (AB-GP205) ankyrin-repeat domains, 12 Ankyrin proteins ankyrin repeats, 11 Ankyrin-repeat domains integral membrane proteins, 12 iPLAefl, 10-14 alignment, 10 proteins, 11 Anticancer drug development Ras signaling pathway, 215-216 Rho signaling pathway, 215-216
B Bacteria purple nonsulfur CO metabolism, 37 Bacterial reverse transcriptase, 80D81 Bap31 quality control, 251 B-catenin quality control, 251 BEL, 27 suicide substrate iPLA2fl, 6, 22 B-globulin, 179, 186 B-hairpin motifs, 13 Bipartite nuclei localization signal iPLA2fl, 14-15 Borg family biological functions, 197 B-Raf, 211 Brain rat iPLA2fl, 21 Brain ankyfin, 12
269
270
INDEX
Breast cancer MCF-7 eisplatin, 124 Bromoenol lactone (BEL), 27 suicide substrate iPLA2fl, 6, 22 Bypass replication model, 188
C Caenorhabditis elegans, 10 Calmegin C-terminal cytosolic domains, 242 Calmodulin iPLA2/5, 16-17 Calnexin/calreticulincycle, 240 Calnexin (CNX), 237-238 C-terminal cytosolic domains, 242 glycoprotein folding ER, 241-243 major histocompatibility class I, 255-258 Calreticulin/ca]nexin cycle, 24-25240 Calreticulin (CRT), 237-238 glycoprotein folding ER, 241-243 major histocompatibility class I, 258 CAMP, 41 Carbon monoxide (CO) binding CooA, 41 CooA, 40M2 HIS-77, 48-49 model, 50-52 Cys-75, 48 metabolism purple nonsulfur bacteria, 37 oxidation Rhodospirillum rubrum, 36-38 PRO-2 heine, 49-50 sensor CooA, 53-54 source aquatic environments, 37 Carbon monoxide dehydrogenase (CODH), 37 Carboplatin, 117 chemical structure, 118 structure, 94
Carboxydothermus hydrogenoformans, 37 Carboxylestera~se quality control, 251 Caspase-3 cleavage site iPLA,2fl, 15 Catabolite responsive protein (CBP), 36, 40-41 consensus target sequences, 58 effector binding sites, 46 effeetor-bound F helices, 46 hinge region residues differential contacts, 45 monomer, 43 strncture vs, Coo& 42-46 three-dimensional structures, 57-58 Catalytic domains protein kinase family, 8 Cathepsin A quality control, 251 Cdc42 E-eadherin, 199-200 effeetors, 197, i99-200 lung cancer, 201 CDK inhibitors, 13 CDNA hBFC characterization, 139-147 cloning, 140-146 heterogeneous 5r UTRs, 142 KS6, 144 KS32, 144 synthesis msDNAs, 77 polymerization, 7740 termination, 77-80 in vivo, 87 Cell adhesion/signaling molecules ankyrin-repeat domains, 12 Cell death cisplatin, 116-117 iPLA,2fl, 24-25 Cell transformation Rho proteins, 195 CEM-NKr~, 255--256 ChoK inhibitors antitumor activity, 217-219
INDEX Choline kinase (ChoK) inhibitors antitumor activity, 217-219 Chromatin barrier TFO, 174-177 Chromosome 22q13.1, 19 Chromosomes localization iPLA2/~, 17-19 mutations TFOs, 176 ChsTp quality control, 250 Cisplatin, 93-125 AAG proteins, 114-115 apoptosis, 116-117 dinueleotide crosslink, 105 H1,114-115 HeLa cells, 113 MCF-7 breast cancer cytotoxicity assay plots, 124 repair shielding, 97 screening, 122 structure, 94 TBP, 114-115 telomere shortening, 112-114 ubiquitination, 116-117 Cisplatin adducts HMG-dom~tin protein binding kinetics, 100-104 thermodynamics, 99-100 HMG-domain proteins sequence selectivity, 101 Cisplatin-DNA, 95-96 damage, 97 HMG 1 domA platinum coordination center, 109 protein-DNA interactions, 108 structure, 107-110 x-ray crystal, 107 intrastrand crosslink structure determinations geometric parameters, 106 Cisplatin-DNA adducts protein binding, 98-104 Cisplatin-DNA ternary complex HMG-domaiu structure, 104-112 Cisplatin-modified DNA duplex
271 N MR solution structure diagram, 106 x-ray solid-state structure diagram, 106 Citron, 196 biological functions, 197 Citron kinase biological functions, 197 Citron Rho-interacting kinase/eitron kinase (CRIK), 196 Cloned Ca2+ iPLA2 characteristics, 7 CNX, 237-238 C-terminal cytosolic domains, 242 glycoprotein folding ER, 241-243 major histocompatibility class I, 255-258 CO. See Carbon monoxide Coat complex II (COPII), 248 CODH, 37 CooA, 35~30 binding, 56 C helices effector-binding domain, 45 CO, 40-42 binding, 41 model, 50-52 sensor, 53-54 consensus target sequences, 58 cyanide binding, 55 DNA recognition properties, 58-59 effector binding sites, 46 effector-free C helix, 44-45 F helices, 46 future direction, 59--60 heine region, 47~50 hinge region residues differential contacts, 45 HIS-77 CO, 48-49 ligand binding, 54~56 monomer, 43 redox sensor, 52~53 redox state, 40-42 structure, 42-52 vs. CRP, 42-46 three-dimensional structures, 57-58
272
INDEX
CooA (cont.) transcriptional activation, 56~58 transcriptional activator, 40-42 CooA gene, 39 Coo genes regulation, 38-40 CooH, 39, 40 COOI/PIX/p85SPR, 202 CooMKLXUH
operon, 38 Coo regulon Rhodospirillum rubrum
transcriptional organization, 38 CooS, 39, 40 COPII, 248 Coralyne, 171 Cox genes, 37 CPLA2y characteristics, 7 CREB-1, 151-152 CRIK, 196 CRP. See Catabolite responsive protein CRT, 237-238 glycoprotein folding ER, 241-243 major histocompatibility class I, 258 CTLs, 251 Cyanide binding CooA, 55 Cyclin-dependent kinase (CDK) inhibitors, 13 Cys-75 CO, 48 Cytotoxic T lymphocytes (CTLs), 251 D DAG0 kinase biological functions, 197 Daunorubicin, 152-153 Deoxyuridine HMG-domain proteins kinetics, 104 Developmental proteins ankyrin repeats, 11 Dgtlp quality control, 250 DHFR, 134-135 5,10-dideaza-5,6,7,8-tetrahydrofolate structure, 136
Dihydrofolate reductase (DHFR), 134-135 DNA. See also Cisplatin-DNA antisense msDNAs, 86-87 crosslinks platinum drugs, 96 3-methyladenine, 98 ms. See MsDNAs recognition properties CooA, 58--59 TFO repair systems, 186--189 sequence change, 185-186 Domains ankyrin-repeat integral membrane proteins, 12 iPLA2/~, 10-14 alignment, 10 proteins, 11 ATP-binding iPLA2, 8-10 catalytic protein kinase family, 8 structural retron reverse transcriptase, 81 Drosophila melanogaster, 10 Drugs anticancer development, 215-216 design ACK 1/2, 222 WASP, 221 peptide-based combinatorial/parallel synthesis approaches, 120-122 platinum DNA crosslinks, 96 Dual human reduced folate carrier promoters identification, 151-152
Ec48 characteristics, 69 Ec67 characteristics, 69 Ec73 characteristics, 69
INDEX Ec78 characteristics, 69 Ec83 characteristics, 69 Ec86 characteristics, 69 Ecl07 characteristics, 69 E-cadherin IQGAP1, 199-200 Rho, 199-200 Edatrexate, 135 structure, 136 EDTA-Fe TFO, 173 Effector-bound CRP F helices CNA, 46 E ffector-free CooA C helix extension/fusion, 44-45 F helices CNA, 46 Egas)m quality control, 251 Eicosanoid synthesis, 27 Endoplasmic reticulum-associated degradation (ERAD), 243-246 major histocompatibility class I biogenesis, 258-259 regulation, 245 Endoplasmic reticulum (ER), 236-237 protein folding future perspectives, 263-264 quality control, 237, 249-252 Endoplasmic reticulum-Golgi intermediate compartment (ERGIC), 248 quality control, 250 Epithelial ankyrin, 12 ER, 236-237 protein folding future perspectives, 263-264 quality control, 237, 249-252 ERAD, 243-246 major histocompatibility class I biogenesis, 258 -259 regulation, 245 ERGIC, 248 quality control, 250
273 ERp57 ER primary quality control, 246-247 major histoeompatibility class I, 258 Erv14p quality control, 250 Erythroid ankyrin, 12 Escherichia toll, 56, 68 retron insertion site, 72-74 ExPaSy, 14 Expert Protein Analysis System (ExPaSy), 14
F Facilitates chromatin transcription (FACT), 97 FACT, 97 Farnesyl transferase (FT), 220 FixL, 41 Fluorescein dyes structure, 102 Fluorescein-modified nucleoside chemical structure, 103 Folate antagonists, 133-138 cancer therapy, 135-138 metabolism, 133-138 natural metabolic roles, 133-135 Folic acid structure, 136 5-~brmyltetr'ahydrofolate structure, 136 FT, 220
G GAPs, 194 GDIs, 194 GEFs, 194 Gene targeting triple-helix formation, 163-189 Gene therapy oligonucleotide based problems, 164 F -globulin upregulation, 179 Glucose-stimulated arachidonate, 22
274
INDEX
Glycerophosphocholine, 27 Glycoprotein folding ER nascent polypeptides, 239-241 primary quality control, 239-247 protein-specific quality control, 247-249 secondary quality control, 247-249 Glycoprotein misfolding ER primary quality control, 243-246 Group IVC PLA2 (ePLA2F) characteristics, 7 Group VIA Ca2+ iPLA2fl molecular biology, 1-29 Group VIA PLA2 enzymology~ 20-22 lysophospholipase, 21-22 PAF acetylhydrolase, 21-22 phospholipids, 20-21 transacylase, 21-22 Gsf2p quality control, 250 GTPases families, 194 Guanine-exchange factors (GEFs), 194 Guanine nucleotide dissociation inhibitors (GDIs), 194 GW1843U89 structure, 136
H H1, 98 cisplatin, 114-115 Hamster cells transport-impaired hRFC cDNA cloning, 140-146 HC-3, 218 HeLa cells cisplatin, 113 Heme CO PRO-2, 49-50 Heine region CooA, 47-50 Hemicholinium-3 (HC-3), 218 Herpes simplex virus protein 16 (HSV VP16), 179-180
High-affinity folate receptors, 132-133 Highly glycosylated human reduced folate carrier protein transport-upregulated K562 human erythroleukemia cells, 139-141 High mobility group protein 1 domA-cisplatin-DNA platinum coordination center, 109 protein-DNA interactions, 108 structure, 107-110 x-ray crystal, 107 High mobility group protein-domain-cisplatin-D NA ternary complex structure, 104-112 High mobility group protein-domain protein binding cisplatin adducts kinetics, 100-104 thermodynamics, 99-100 High mobility group protein-domain proteins cisplatin adducts sequence selectivity, 101 deox3aaridine kinetics, 104 sequence alignments, 100 structure, 100 High mobility group protein 1 (HMG1), 98 bandshift assay, 99 cisplatin, 122--123 adducts, 99-100 EMSA, 99 MCF-7 breast cancer estrogen, 122-124 High mobility group protein 2 (HMG2), 98 His-77, 54 CooA CO, 48~49 Histone linker protein (H1), 98 cisplatin, 114-115 HMG1. See High mobility group protein 1 HMG2, 98 Host cell msDNAs, 84-86 HRFC. See Human reduced folate carrier HSP47 quality control, 251 HSV VP16, 179-180 HUBF, 98
INDEX
275
Human cells transport-impaired hRFC eDNA cloning, 140-146 Human long iPLA2fl isoform (LH-iPLA2fl) aligument, 16 proline-rich region, 15-16 Human reduced folate carrier-A promoter, 151-152 Human reduced folate carrier-B promoter, 151 Human reduced folate carrier eDNA characterization, 139-147 cloning, 140-146 heterogeneous 5' UTRs, 142 Human reduced folate carrier (hRFC), 131-159 amino acid homologies, 143 exons, 148 expression regulation, 147-153 gene upstream organization, 148-150 promoters dual, 151-152 protein posttranslational modifications, 146-147 properties, 139-147 topology structure, 144 transcript heterogeneity, 147-1,50 transport-impaired K562, 146-147 Western blot, 145-146 wild-ty1~ep53, 152-1.53 Human short iPLAgJ3 isoform (SH-iPLA.2fi),19
I IL-1/3, 24 Independent phospholipase A,2a (iPLA2a) characteristics, 7 Independent phospholipase A2/3 (iPLA2fl) alternative splicing, 17-19 ankyrin-repeat domain, 10-14 alignment, 10 apoptosis, 24-25 ATP-bindingdomain, 8-10 BEL suicide substrate, 6 bipartite mmlei localization signal, 14-15 ctdmodulin, 16-17 caspase-3 cleavage site, 15
characteristics, 7 chromosomal localization, 17-19 cloning, 17 enzymatic activities, 20 future perspectives, 28-29 gene structure, 17-19 group VIA Ca2+ molecular biology, 1-29 /ipase consensus motif FXSXG, 5-8 membrane homeostasis, 26-29 membrane phospholipid remodeling, 25-26 mRNA encoding, 20 phosphorylation, 17 purified recombinant, 21 rat brain, 2l sequence, 4-17 signaling function insulin-seereting cells, 22-23 structural characteristics, 4-17 tissue distribution, 20 Independent phospholipase A-2F(iPLA2F) amino acid sequence, 6 Independent phospholipase A2 (iPLA2) ankyrin, 13 cloned Ca2+ characteristics, 7 Insulin-secretingcells signaling function iPLA2/~, 22-23 Integral membrane proteins ankyrin-repeat domains, 12 Intercalators TFOs, 171 Interleukin-lfl (IL-lfl), 24 Intrastrand erosslink (Ixrl), 98 Invariant chain quality control, 251 Ion channels ankTrin-repeat domains, 12 IPLA2 ankyrin, 13 cloned Ca2+ characteristics, 7 iPLA2ot characteristics, 7 iPLA#L See Independent phospholipase A2~ iPLA2F amino acid sequence, 6 IQGAP, 199-200
276
INDEX
IQGAP1 drug design, 221-222 E-cadherin, 199-200 ISIS 5132, 219 Islet/~ ceils, 24 Ixrl, 98
JM216, 117 chemical structure, 118 JM221, 117 chemical structure, 118 JNK PAK, 203
K K562.4CRF cells, 140-141 KS1 origin, 150 KS6 cDNA, 144 KS32 cDNA, 144 origin, 150 k Laglp quality control, 250 LDL receptor family, 248 LE-5132, 219 LEF-1, 98 Leucine zipper, 51 Leukemia acute lymphoblastic, 135 LH-iPLA2/~ alignment, 16 proline-rich region, 15-16 Ligand binding CooA, 54-56 Lipase consensus motif FXSXG iPLA2/~, 5 4 Low-density lipoprotein (LDL) receptor family, 248
LPA, 27 LST1 quality control, 250 Lymphocyte adhesion antigen CD44 ankyrin-repeat domains, 12 Lymphocyte ankyrin, 12 Lymphoid enhancer-binding factor 1 (LEF-1), 98 LY231514 (MTA) structure, 136 Lysophosphatidic acid (LPA), 27 Lysophospholipase group VIA PLA2, 21-22
M Major histocompatibility class I, 235--264 assembly protein-specific factors, 260-263 biogenesis ERAD, 258-259 primary quality control, 254--260 disulfide bonds, 256 encoding, 251-252 lumenal domain structure, 253 quality control, 251-263 sequential folding and assembly, 254 TAP, 260-263 tapasin, 260-263 MBS, 196 MCF-7 breast cancer cisplatin cytotoxicity assay plots, 124 HMG1 estrogen, 122-124 MDia 1/2 biological functions, 197 Melittanium, 68 Membrane homeostasis iPLA2/3, 26-29 Membrane phospholipid remodeling iPLA2/3, 25-26 Methotrexate (MTX), 133, 135 intracellular transfer, 138-139 polyglutamyl form, 135-138 structure, 136 3-methyladenine DNA, 98
INDEX 5-methyl tetrahydrofolate structure, 136 Microbes anaerobic CO-oxidizing, 37 Microsomal triglyceride transfer protein quality control, 251 Minor-groove binders, 166 Mitelfosine, 217 ML162, 68 characteristics, 69 Mononuclear platinum complexes, 117-119 chemical structure, 118 Morpholino oligonucleotides, 172 MRNA encoding iPLA2fl, 20 Msd, 71, 74 MsDNAs, 65-89 antisense DNA, 86--87 branch formation, 77 cDNA synthesis, 77 characteristics, 69 diversity, 82-84 host cell, 84-$6 mutation frequencies, 84--85 new, 82-83 old, 84 origin, 82-84 potential applications, 88 potential uses, 86-87 prevalence, 82~4 primer-template RNA, 76 priming, 77 processing, 70-71 production, 71-75 protein, 70 repetitive sequences, 85-86 RT, 74-80 structure, 66~68 primary, 66-67 secondary conserved, 66-70 Msr, 71, 74 MTA structure, 136 MTX, 133, 135 intraeellular transfer, 138--139 polyglutamyl form, 135-138 structure, 136 Mutagenesis TFO, 180-181
277 Mutant carriers RFC, 156-157 Mutant HMG-domain protein binding affinities, 110 studies, 110-112 Mutation frequencies msDNAs, 84-85 Mx65 characteristics, 69 Mx162, 68 characteristics, 69 Myosin binding subunit (MBS), 196 Myxococcusxanthus, 66
N Nannocystis exedens, 68 Nascent polypeptides glycoprotein folding ER, 239-241 Natural folates metabolic roles, 133-135 NER, 186-188 pathway, 97 NEST, 207 NFKB, 13, 203 NinaA quality control, 251 Nonesterifled arachidonate, 22 Non-human reduced folate carrier transport components, 146-147 Nuclear envelope signal transduction (NEST), 207 Nucleoside fluorescein-modified chemical structure, 103 Nucleotide excision repair (NER), 186-188 pathway, 97
O ODR-4 quality control, 250 ODR-8 quality control, 250 Oligodeoxynucleotide antisense, 219
278
INDEX
Oligonucleotide based-gene therapy problems, 164 Oligonucleotide based triple helix interaction, 168-169 Oligotropha carboxidovorans, 37 Oxaliplatin chemical structure, 118
P P38 PAK, 203 P21-aetivated kinase (PAK), 201-204 biological functions, 197 drug design, 222 ]NK, 203 membrane localization, 202 p38, 203 PAF acetylhydrolase, 4 group VIA PLA2, 21-22 group VIIIA, 7 PAK, 201-204 biological functions, 197 drug design, 222 JNK, 203 membrane localization, 202 p38,203 PAPH- 1, 23 Parallel synthesis platinum, 120-122 Parotid acinal cells, 28 PDI, 237 Peptide-based drugs combinatorial/parallel synthesis approaches, 120-122 Peptide nucleic acids (PNAs), 166-167 P24 family quality control, 250 Phosphatidylinositol 4,5-biphosphate PDL, 206 Phosphatidylinositol-3-OH kinase (P13K), 212-214 cancer, 213-214 Phosphocholine cytidylyltransferase, 26-27 Phospholipase Al (PLA1) activities, 20 Phospholipase Ae (PLA2) activities, 20
classification, 3-4 molecular biology, 2-3 Phospholipase D (PLD), 205-207 cancer treatment, 216-217 Phospholipids group VIA PLA2, 20-21 Phosphorylation iPLA2fl, 17 PI3K drug design, 222-223 PIP5 kinase biological functions, 197 P13K, 212-214 cancer, 213-214 PKN, 196, 207-209 biological functions, 197 PLA1 activities, 20 PLA2 activities, 20 classification, 3-4 molecular biology, 2-3 Platelet-activating (PAF) acetylhydrolase, 4 group VIA PLA2, 21-22 group VillA, 7 Platinum parallel synthesis, 120-122 Platinum drugs DNA crosslinks, 96 new
development, 117-125 PLD, 205-207 cancer treatment, 216-217 PNAs, 166-167 Polymerization cDNA synthesis, 77-80 Polynuclear platinum complexes, 119-120 chemical structure, 119 Polypeptides nascent glycoprotein folding ER, 239-241 POR1 biological functions, 197 POSH, 204 biological functions, 197 Primer-template RNA, 242 msDNAs, 76
INDEX Priming msDNAs, 77 PRO-2 CO heine, 49-50 Programmed cell death cisplatin, 116-117 iPLA2/~, 24-25 Prolyl 4-hydroxylase quality control, 250 Propranolol, 23 Protective protein quality control, 251 Protein disulfide isomerase (PDI), 237 Protein kinase B, 213 Protein kinase family catalytic domains, 8 Protein kinase N (PKN), 196, 207-209 biologie~dfunctions, 197 Proteins AAG cisplatin, 114-115 aulefrin ankyrin repeats, 11 ankyrin-repeat domain, 11 binding cisplatin-DNA adduets, 98-i04 eatabofite responsive. See Catabolite responsive protein developmental anle/rin repeats, 11 fblding, 236 ER, 263-264 highly glycosylated hRFC transport-upregulated K562 human erythroleukemia cells, 139-141 histone linker, 98 HMG domain. See HMG-domain proteins hRFC properties, 139-147 integral membrane ankyrin-repeat domains, 12 msDNAs, 70 mutant HMG-domain binding affinities, 110 studies, 110 112 protective quality control, 251 Ras. See Ras proteins receptor-associated, 248
279 Rho. See Rho proteins synaptic ankyrin repeats, 11 TATA-binding,98 cisplatin, 114-115 TGPase-aetivating, 194 transcription'a] thctor ankyrin repeats, 11 viral host range ankyrin repeats, 11 WASP-familyverprolin-homologous, 197-199 Protein-specific quality control, 250 PsoAG10 triple-helix, 174 Psoralen adduct repair model, 188 TFO, 173-174 repair, 187 TFO-induced recombination, 182 TFO mutagenesis, 180-181 PSupF2 vector schematic representation, 183 Purified recombinant iPLAefl, 21 Purple nonsulfur bacteria CO metabolism, 37
Q Quality control, 251
R Rae effeetors, 199-200 Rael E-cadherin, 199-200 effeetors, 196-197 Raf kinase inhibition, 219-220 pathway, 211-212 Raf-1 kinase drugs, 219 Ral GDP dissociation stimulators (RalGDS), 214
280 RalGDS, 214 RalGDS family, 214 RAP, 248 quality control, 251 Ras farnesylation inhibition, 220-221 PDL, 207 proteins cycling, 194 effectors, 209-214 future perspectives, 223-224 PI3K, 212-214 RalGDS, 214 signaling pathway anticancer drug development, 215-216 superfamily, 194 Rat brain iPLAz/~, 21 Receptor-associated protein (RAP), 248 quality control, 251 Red cell anion exchanges ankyrin-repeat domains, 12 Redox sensor CooA, 52-53 Redox state CooA, 40-42 Reduced folate carrier (RFC), 132-133 functional properties, 138-139 MTX resistance molecular mechanisms, 154-156 mutant carriers, 156-157 stramtural-functional determinants, 156-157 Repair shielding cisplatin, 97 Repair systems DNA alterations TFO, 186-189 Repetitive sequences msDNAs, 85-86 Ret, 74 Retron, 71-75 insertion site, 72-74 E. coli, 72-74 Retron element transcription, 74-75 Retron reverse transcriptase structural domains, 81
INDEX Reverse transcriptase (RTO), 66 RFC, 132-133 functional properties, 138-139 MTX resistance molecular mechanisms, 154-156 mutant carriers, 156-157 structural-functional determinants, 156-157 Rho farnesylation inhibition, 220-221 proteins cell transformation, 195 effectors, 196-209 future perspectives, 223--224 PDL, 207 signaling pathway anticancer drug development, 215-216 RhoA, 204-205 E-cadherin, 199-200 effectors, 196-197 PDL, 205-206 PKN, 207-209 Rhodamine acceptor structure, 102 Rhodobacter sphaeroide.s, 56 RhodospiriUum rubrum, 35-36 coo regulon, 53 transcriptional organization, 38 CO oxidation, 36-38 Rhophilin, 196 biological functions, 197 Rhotekin, 196 biological functions, 197 RNA primer-template, 242 msDNAs, 76 TFOs, 178 RNAP binding, 56, 57 RNA polyrnerase (RNAP) binding, 56, 57 RNAse H, 80 ROCK, 196, 204-205 biological functions, 197 inhibitors antitumor activity, 216 Y-27632, 216 RTO, 66
INDEX
Sa163, 68 characteristics, 69 SCLC, 201 Secretory phospholipase A2 (sPLA2) classification, 3~t Serum response element (SRE), 208 Serum response factor (SRF), 208 Sex-determining factor Y (SRY), 98 SGLT1, 147 SH-iPLA2fl, 19 Shr3p quality control, 250 Sickle cell anemia, 179, 186 Signal bipartite nuclei localization iPLA2fl, 14-15 Signaling function iPLA2fl insulin-secreting cells, 22-23 Signaling molecules cell adhesion ankyrin-repeat domains, 12 Signaling pathway Ptas antieaneer drug development, 215-216 Rho anticaneer drug development, 215-216 Signal transduetion nuclear envelope, 207 Single-stranded tethered donor TFOs schematic diagram, 184 Small cell lung cancer (SCLC), 201 Sodium channel amiloride-sensitive ankyrin-repeat domains, 12 Sodium-dependent D-glucose eotransporter (SGLT1), 147 SPLA2 classification, 3-4 SRE, 208 SRF, 208 SRY, 98 SSRP1, 97, 98 Stigmatella, 68 Structural domains retron reverse transcriptase, 81 Structure-specific recognition protein 1 (SSRP1), 97, 98
281 Suicide substrate BEL iPLA2fl, 22 Synaptic proteins ankyrin repeats, 11
TAP MHC1, 260-263 Tapasin MHC1, 260-263 quality control, 251 TATA-binding protein (TBP), 98, 194 cisplatin, 114-115 TBP, 98, 194 cisplatin, 114-115 Telomere shortening cisplatin, 112-114 Termination cDNA synthesis, 77-80 Tetracycline, 152-153 Tetrahydrofolate cofactors, 133-134 TFOs. See Triplex forming oligonucleotides TGPase-activating proteins (GAPs), 194 Thalassemia, 179 Thymidylate synthase, 134-135 Tomudex, 135 structure, 136 Toxins ankyrin repeats, 11 Transacylase group VIA PLA2, 21-22 Transcriptional activator CooA, 40-42 Transcriptional factor proteins ankyrin repeats, 11 Trans-DDP
structure, 94 Transporter associated with antigen processing (TAP) MHC1, 260-263 Transport-impaired hamster cells hRFC cDNA cloning, 140-146 Transport-impaired human cells hRFC cDNA cloning, 140-146 Transport-impaired K562 hRFC, 146-147
282 Transport-mediated methotrexate resistance molecular mechanisms, 153-157 Transport-upregulated K562 human erythroleukemia cells highly glycosylated hRFC protein, 139-141 Triple helix oligonucleotide based, 168-169 psoAGl0, 174 Triplex forming oligonucleotides (TFOs), 163-164 antigene agents, 177-178 chemical substitutions, 172 chemistry, 167-171 improvements, 171-172 chromatin barrier, 174-177 chromosomal mutations, 176 cleaving tool, 173 delivery, 164-165 DNA sequence change, 185-186 DNA "alterations repair systems, 186-189 EDTA-Fe, 173 half-life, 165 intercalators, 171 molecular tools, 172-174 motifs limitations, 169-170 mutagenesis, 180-181 recombination, 181-185 psoralen, 182 RNA, 178 strand crossing, 171 upregulation gene expression, 178-180
Ubiquitination cisplatin, 116-117 UGGT, 239-241 glycoprotein folding ER, 239-241 Upstream binding factor (hUBF), 98 Uridine dipohosphate glucose:glycoprotein glucosyltransferase (UGGT), 239-241
INDEX glycoprotein folding ER, 239-241 US2/USll, 259-260 V Vc95, 68 characteristics, 69 Viral host range proteins ankyrin repeats, 11 Virio cholerae, 68 Vma12p-Vma22p complex quality control, 250 W WASP, 198-199 biological functions, 197 drug design, 221 WASP-family verprolin-homologous protein (WAVE), 197-199 WAVE, 197-199 Wild-type p53 hRFC, 152-153 Wiskott-Aldrich syndrome protein (WASP), 198-199 biological functions, 197 drug design, 221 × Xeroderma pigmentosmn (XPA), 98 XPA, 98
Y-27632, 205 ROCK, 216 YB-1, 98 Y-boxbinding protein 1 (YB-1), 98 Z ZD1694, 135 stnleture, 136