HIPPOCAMPUS 21:1–8 (2011)
RAPID COMMUNICATION Differential Regulation of Synaptic Plasticity of the Hippocampal and the Hypothalamic Inputs to the Anterior Thalamus Marian Tsanov,1,2 Seralynne D. Vann,3 Jonathan T. Erichsen,4 Nick Wright,4 John P. Aggleton,3 and Shane M. O’Mara1,2* ABSTRACT: The hippocampus projects to the anterior thalamic nuclei both directly and indirectly via the mammillary bodies, but little is known about the electrophysiological properties of these convergent pathways. Here we demonstrate, for the first time, the presence of long-term plasticity in anterior thalamic nuclei synapses in response to high- and low-frequency stimulation (LFS) in urethane-anesthetized rats. We compared the synaptic changes evoked via the direct vs. the indirect hippocampal pathways to the anterior thalamus, and found that long-term potentiation (LTP) of the thalamic field response is induced predominantly through the direct hippocampal projections. Furthermore, we have estimated that that long-term depression (LTD) can be induced only after stimulation of the indirect connections carried by the mammillothalamic tract. Interestingly, basal synaptic transmission mediated by the mammillothalamic tract undergoes use-dependent, BDNFmediated potentiation, revealing a distinct form of plasticity specific to the diencephalic region. Our data indicate that the thalamus does not passively relay incoming information, but rather acts as a synaptic network, where the ability to integrate hippocampal and mammillary body inputs is dynamically modified as a result of previous activity in the circuit. The complementary properties of these two parallel pathways upon anterior thalamic activity reveal that they do not have duplicate functions. V 2009 Wiley-Liss, Inc. C
KEY WORDS: anterior thalamic nuclei; fornix; mammillothalamic tract; subiculum; LTP
INTRODUCTION The hippocampus and anterior thalamic nuclei form key components of a neural circuit linking medial temporal lobe and medial diencephalic regions required for episodic memory (Aggleton and Brown, 1999; Warburton et al., 2001). Within this circuit, anterior ventral and anterior medial thalamic nuclei both receive: (i) direct inputs from the hippocampus (subiculum) via the fornix, and (ii) indirect hippocampal (subicular) inputs from the medial mammillary bodies via the mammillothala1
Trinity College Institute of Neuroscience, Trinity College Dublin, Ireland; 2 School of Psychology, Trinity College Dublin, Ireland; 3 School of Psychology, Cardiff University, United Kingdom; 4 School of Optometry and Vision Sciences, Cardiff University, United Kingdom Grant sponsor: Welcome Trust (to J.P.A. and S.M.O’M); Grant number: W01048. *Correspondence to: Shane M. O’Mara, Trinity College Institute of Neuroscience, Trinity College Dublin, Dublin 2, Ireland. E-mail:
[email protected] Accepted for publication 23 October 2009 DOI 10.1002/hipo.20749 Published online 30 December 2009 in Wiley Online Library (wileyonlinelibrary.com). C 2009 V
WILEY-LISS, INC.
mic tract (Meibach and Siegel, 1975; Aggleton et al., 1986; Ishizuka, 2001). Although hippocampal projections to the mammillary bodies and anterior thalamic nuclei originate from the same regions of subiculum, they arise from separate neuronal populations (Ishizuka, 2001). The importance of the mammillothalamic tract for memory is emphasized by the repeated finding that damage to this tract is characteristic of thalamic strokes, which induce amnesia (von Cramon et al., 1985; Van der Werf et al., 2003). Current models of how the convergent inputs to the anterior thalamic circuitry might support memory emphasize the flow of hippocampal information through these pathways without explaining how these relays might transform or modulate this information in a way vital for memory. Both synaptic and nonsynaptic mechanisms of neuronal plasticity underpin experience-dependent alterations in brain networks and hypothetically at least, support long-term memory processes (Madison et al., 1991; Moser et al., 1998). The present study, therefore, compared activity-dependent plasticity within the anterior thalamic nuclei (Fig. 1A, inset), following stimulation of either the dorsal fornix (Dfx, Fig. 1B) or the mammillothalamic tract (MTT). Our first goal was to determine whether LTP or LTD can be induced in the separate anterior thalamic nuclei inputs. To examine basal mammillothalamic synaptic transmission, animals underwent implantation of bipolar stimulating electrodes in MTT and a recording electrode in the ipsilateral anterior ventral thalamic nucleus (Fig. 1A). Male 7- to 10-week-old ListerHooded rats (Harlan, UK) were triple housed and maintained on 12:12 h light:dark cycles with food and water provided ad libitum. Under urethane anesthesia (ethyl carbamate: 1.5 g/kg, i.p.), the animals underwent insertion of a monopolar recording electrode (RNEX-300, David Kopf Instruments Ltd.) in the anterior thalamic nuclei (Fig. 1C) and bipolar stimulating electrodes in the mammillothalamic tract (MMT) (Fig. 1D). For the recording electrode, a drill hole was made in the cranium (1 mm in diameter), 1.1–1.3 mm posterior to bregma and 1.4–1.6 mm lateral to the midline, dorsal to the anterior ventral thalamic nucleus (ATN) in the rat (Kruger et al., 1995).
2
TSANOV ET AL.
FIGURE 1. Experimental design for plasticity-inducing recordings in anterior thalamic nuclei of anesthetized animals. (A) Illustration of the position of the bipolar stimulating electrodes in dorsal fornix (Dfx) and mammillothalamic tract (MTT), and monopolar recording electrode in the ventral anterior thalamic nucleus (ATN). The circuitry of hippocampodiencephalic connections, schematically represented in the upper right corner, reveals the parallel hippocampal outputs. Stimulation of Dfx allows slope measurement of monosynaptic onset recording from ATN, shown as the upper left inset, while the stimulation of MTT represents the indirect pathway passing through the mammillary bodies (MB). (B) Coronal section showing stimulating electrode tracks
that reach Dfx. (C) Coronal section revealing the track of recording electrode in ATN. The column on right shows field potentials recorded at different depths in response to MTT stimulation. Each step, marked with the horizontal lines, represents 400 lm. Horizontal bar: 5 ms, vertical bar 1 mV. (D) The stimulation electrodes were alternatively positioned in MTT; the histological section shows stimulation electrode tracks reaching MTT. White arrows point to the targeted structures in (C) and (D) (marked with dashed line) ipsilateral to the recording site; contralateral fiber tracts are also noted. Smaller black arrows point to the location of electrode tips.
A recording electrode was lowered 5.0 mm from the dural surface to reach ATN. A second drill hole was made for a bipolar stimulating electrode with coordinates targeting MTT (2.3–2.5 mm posterior to bregma, 0.8–1.0 mm lateral to midline). Alternatively, for the experiments that monitored the direct hippocampothalamic connection, the stimulation electrodes were positioned for Dfx coordinates (3.3–3.5 mm posterior to
bregma, 0.3–0.5 mm lateral to midline) (Fig. 1B). A stimulating electrode was positioned in the MTT/Dfx ipsilateral to the hemisphere from which ATN recordings were obtained. The depth was 5.0–6.0 mm from the dural surface for MTT and 1.5–2.5 mm for Dfx (Kruger et al., 1995). Final positions of the stimulating and recording electrodes were then determined by maximizing the amplitude of the field potential recorded in
Hippocampus
LONG-TERM SYNAPTIC PLASTICITY IN ANTERIOR THALAMUS the ATN in response to electrical stimulation of MTT/Dfx. Monopolar recordings from ATN were made relative to ground and reference screws inserted into the contralateral parietal and frontal bones. Once verification of the location of the electrodes was complete, recordings were allowed to stabilize for 10 min before experiment. Signals were filtered between 0.1 Hz and 1 kHz, and then amplified (DAM-50 differential amplifier; World Precision Instruments, Hertfordshire, UK). Recordings were digitized online using a PC connected to a CED-1401 plus interface and analyzed using Spike 2 software (CED, Cambridge, UK). Field potential (FP) slope and amplitude were used as a measure of excitatory synaptic transmission in the ATN region. To obtain these measurements, an evoked response was generated in Dfx or MTT by stimulating at low frequency (0.025 Hz) with single biphasic square wave pulses of 0.1 ms duration per half wave, generated by a constant current isolation unit. For each time point measured during the experiments, five records of evoked responses were averaged. The FP slope was measured as the intermediate 90% of the slope value between the first positive and the first negative deflections of the FP. The FP amplitude represents the absolute difference between the value of the first positive and the value of the first negative deflections of the FP (Fig. 1A, up-left inset). In addition, we measured the FP maximal slope through the five steepest points obtained on the negative deflection of the FP. The latency of the field potential was measured in milliseconds from the first positive deflection to the maximal point of the negative deflection. By means of input-output (IO) curve determination, the maximum FP was found, and during experiments, all potentials used as baseline criteria were evoked at a stimulus intensity that produced 40% of this maximum (100–400 lA). The baseline FP data were obtained by averaging the response to stimulation of the MTT/Dfx, to obtain five sweeps at 40 s intervals, every 5 min over a period of 30 min. Electrophysiological data were then expressed as the mean percentage of baseline FP 6 standard error of the mean (S.E.M.). Statistical significance was estimated by using factorial analysis of variance (ANOVA). Using a factorial ANOVA, we estimated the effects of the stimulation protocol and the effect of time on the field potential values compared with the baseline period, composed of the first six time points. The probability level interpreted as statistically significant was P < 0.05. During baseline test-pulse stimulation, the MTT-evoked FP slope and amplitude gradually increased their values within the first 30 min of the recordings. Consistently, for all the animals in this group, the FP slope reached 130–160% of the first recorded value for each experiment (Fig. 2A, n 5 5). The same change was evident for the FP amplitude (data not shown), although both parameters did not always share the same percentage increase when compared with individual recordings. In comparison with MTT stimulation, Dfx baseline recordings had a different profile. The average baseline for Dfx test-pulses did not show long-term alteration of the FP slope (Fig. 2A, n 5 4) and amplitude (data not shown). In this case the ATN FP was evoked predominantly via the direct projections as the
3
low amplitude of the evoked response was not enough to induce a population spike response in MMB and subsequent transsynaptic activation of ATN through MMB. The restricted nature of the synaptic transmission is supported by the monosynaptic latency of the Dfx-evoked FP onset (2–3 ms), which determines the slope measurement in our recordings (Fig. 1A). Our next experiment was designed to identify the trigger for mammillothalamic baseline augmentation. One of the most common models of homosynaptic plasticity proposes the rate of synaptic activation as a major factor in Hebbian modifications (Bienenstock et al., 1982). Neuronal stimulation with a frequency of less than 0.1 Hz is usually regarded as having no effect on the plasticity of hippocampal and cortical synaptic transmission (Lisman, 1994). We therefore used test pulses with a rate of 0.025 Hz to evoke baseline neuronal responses before the application of plasticity-inducing stimulation protocols. Recent findings, however, show that thalamocortical responses in vivo recorded with a baseline frequency of 0.025 Hz can undergo activity-dependent potentiation per se (Tsanov and Manahan-Vaughan, 2007a). In order to distinguish the role of baseline test-pulse rate as a factor in the observed phenomenon, we recorded mammillothalamic synaptic transmission with a baseline frequency lower than 0.025 Hz. The testpulses for this group of animals were given every 120 s (0.008 Hz), monitored for 3 h (slow baseline) and then reset back to 0.025 Hz for another 2.5 h for the rest of the experiment. If the FP augmentation is frequency-driven, then the increase of the thalamic response in this group would be expected to be diminished or absent. Interestingly, the baseline synaptic transmission profile (Fig. 2A) appeared to reach the same amplitude as in the 0.025 Hz frequency group. The FP slope increased during the first 10 time points, showing no significant difference from the 0.025 Hz frequency group, when the data are compared on equalized time scale (Fig. 2A, ANOVA, F < 1, P 5 0.903, n 5 5). A similar pattern was observed for FP amplitude (ANOVA, F < 1 P 5 0.810, n 5 5, data not shown), where the potentiation dynamics were not time-related but appeared to be dependent on the number of pulses delivered to MTT. Analyzing the data of the first 10 time points with the original time scales for 0.008 Hz (Fig. 2A inset below-left) and 0.025 Hz baseline (Fig. 2A inset below-right), we find a significant difference in the pattern of FP augmentation (ANOVA, F(1,9) 5 4.02, P < 0.05, n 5 5). This pattern can be presented also as different correlations between the slope values and time points for each frequency (Pearson’s, r 5 0.5185 for 0.008 Hz and r 5 0.7758 for 0.025 Hz, Fig. 2B). A challenging issue was to explore the mechanism responsible for the pulse-dependent increase of the mammillothalamic responses. Two related pieces of evidence were considered in our investigation: (1) anterior ventral thalamic nucleus possesses substantial fiber/terminal BDNF-immunoreactivity (Conner et al., 1997; Snyder et al., 1997) and (2) exogenous application of BDNF can induce a rapid and persistent enhancement of synaptic transmission in hippocampal and cortical preparations (Kang and Schuman, 1995; Akaneya et al., 1997). Application of BDNF induces a slow-onset, persistent strengthening of synHippocampus
4
TSANOV ET AL.
aptic transmission at hippocampal synapses in vivo (Messaoudi et al., 1998; Ying et al., 2002). Thus, we reasoned that BDNF may be the factor driving the ATN response augmentation. We
tested this hypothesis with i.c.v. application of a human recombinant TrkB-Fc chimera (T 8694; Sigma-Aldrich) (5 lg/ ll), which blocks TrkB ligand signaling (Shelton et al., 1995; Sharma et al., 2006). Stock solution was made as 250 lg/ml in 13 phosphate buffered saline (PBS) containing 0.1% bovine serum albumin (BSA) and stored at 48C until used. TrkB-Fc or control protein (IgG, 5 lg/ll; I4131–10MG; Sigma-Aldrich) was injected in a 5-ll volume over a 5-min period via a Hamilton syringe in the lateral cerebral ventricle (i.c.v). The TrkB-Fc injection was carried out immediately after the first FP recording and after 5 min (sufficient time for diffusion from the lateral cerebral ventricle into the adjacent anterior thalamic nuclei to occur) the baseline recording was continued. Under these conditions, we found a significant impairment of the pulse-dependent augmentation of the ATN FP slope (Fig. 2D, ANOVA, F(1,49) 5 4.45, P < 0.001, n 5 4) and amplitude (ANOVA, F(1,49) 5 2.32, P < 0.01, n 5 4, data not shown), compared to the treated with control protein (IgG, 5 lg/ll) group. Our results suggest a role for TrkB in this phenomenon in agreement with studies, demonstrating that blockade of BDNF-TrkB interaction by TrkB-receptor antibodies (Kang et al., 1997) or anti-BDNF antibodies (Chen et al., 1999) strongly reduces synaptic long-term potentiation LTP (Gartner et al., 2006). The first question regarding synaptic plasticity in ATN was whether subsequent stimuli would result in synaptic facilitation or depression. These effects were examined after the baseline was stabilized and pairs of stimuli were delivered with interstimulus intervals of 20, 30, 40, 50, 100, and 250 ms. The paired-pulse (PP) ratio (Fig. 3) represents the value of the second potential (FP2) over the value of the first one (FP1) for
FIGURE 2. MTT and Dfx express different basal synaptic transmission properties. (A) Basal synaptic transmission of ATN after stimulation of MTT (white dots) and Dfx (gray dots). Averaged values of FP slope reveal gradual augmentation in the group of animals with MTT stimulation (n 5 5) and no change throughout the 4 h recording period in the group with Dfx stimulation (n 5 5). Even when the baseline frequency of MTT test-pulses was reduced from 0.025 Hz to 0.008 Hz (black dots) (n 5 5), no change in the profile of the baseline occurred. The elevation was dependent on the number of the given pulses, but not on the time from the start of the experiment. The lower x-axis represents the time points of the ‘‘slow baseline’’ with test-pulses induced with 0.008 Hz frequency. First 180 min of the 0.008 Hz baseline are presented in the inset below left. The augmentation of the FP has slower onset when compared to the initial 180 min of the standard baseline (inset below right). Gray dashed line in the plots delineates 100% (baseline). (B) FP slope correlates differently with the initial time points of MTT baseline, depending on the given frequency; black symbols, 0.008 Hz, white symbols, 0.025 Hz. (C) Analogs represent FPs evoked at the points marked in the figures. Horizontal bar: 5 ms, vertical bar 1 mV. (D) The TrkB-Fc chimera (5 lg/ll) blocks the onset of field potential augmentation when applied (i.c.v.) after the first recording. A significant reduction of the FP slope is observed in TrkB-Fc-treated animals (black dots; n 5 4) in comparison to the control protein (IgG, 5 lg/ll) treated group (white dots; n 5 4). Hippocampus
FIGURE 3. Synaptic depression and synaptic potentiation in ATN are input specific. Paired pulses lead to facilitation for mammillothalamic (A) and dorsal fornix (B) stimuli. Bars represent mean paired-pulse (PP) ratio of the second over the first response for slope (black) and amplitude (white). On the right sides are shown examples of FP traces at interstimulus intervals of 20 ms—upper pairs, 30 ms—middle pairs and 40 ms—lower pairs. (C) Analysis of the first 100 pulses of 1Hz stimulation in MTT-stimulated group (n 5 6) reveals instant suppression of the FP slope (white dots) compared with the baseline period prior the start of LFS and no change of the FP amplitude (black dots). (D) The same analysis for the Dfx-stimulated group (n 5 5) demonstrates almost parallel suppression of both FP slope (white dots) and amplitude (black dots) in the time course of 1 Hz stimulation. Gray dashed line in the plots delineates 100% (baseline). (E) 1 Hz (900 pulses) stimulation of Dfx induces LTD of the ATN FP slope in the cases with Dfx stimulation (white
dots, n 5 5). In the group of animals with mammillothalamic recordings (black dots, n 5 7), the same protocol failed to induce depression. (F) High-frequency stimulation (HFS) (100 Hz) leads to potent increase of the FP slope for the MTT implanted animals (black dots, n 5 5). In contrast, no change of the slope and significant but weak potentiation of the amplitude followed the same stimulation protocol in Dfx implanted rats (white dots, n 5 7). (G) Analogs represent FPs evoked at the points marked in the figures. Horizontal bar: 5 ms, vertical bar 1 mV. (H) Analyses of the maximal FP slope from the same data, measured by the steepest value of the negative-going potential (white dots) and FP latency, measured by the distance in ms between the initial positive deflection of the FP onset and the lowest point of the current sink (black dots) in Dfx-stimulated group (n 5 7). HFS to ATN afferents results in an increase of the maximal slope and concurrent decrease of the field potential latency.
6
TSANOV ET AL.
the slope (black bars) and the amplitude (white bars). The significance level was evaluated with post hoc Student t-tests by taking the average of four slope/amplitude values of FP1, for a given interval, and normalizing the average of four values for FP2 with respect to this value. Significant facilitation was present for 20, 30, and 40 ms interstimulus intervals the FP2 amplitude of the MTT-evoked responses (t-test, P < 0.05, n 5 4) (Fig. 3A). Only the 30 ms intervals revealed significant facilitation for the MTT-evoked FP2 slope (t-test, P < 0.05), whereas the 20 ms interval evoked tendency of depression (Fig. 3A). Similarly, Dfx paired-pulses induced facilitation for 20 and 30 ms intervals for the FP2 amplitude (t-test, P < 0.05, n 5 4) and 30 ms interval for the FP2 slope (t-test, P < 0.05) (Fig. 3B). Our next step was the exploration prolonged forms of plasticity in ATN. To examine whether low-frequency stimulation (LFS) can induce plasticity of ATN field responses, we used a stimulation protocol known to evoke LTD. Baseline responses were collected for 30 min before the application of an LFS protocol consisting of 900 pulses at a frequency of 1 Hz. The stimulus amplitude was the same as that used for previous recordings. Stimulation of Dfx and MTT resulted in very different changes. To address the manner in which the short-term component of LFS-induced plasticity predicts the long-term profile in both pathways, we compared the FP alterations occurring immediately after the first pulses in the time course of the 1Hz stimulation. Separate pathways have recently been shown to possess different types of short-term plasticity in the same hippocampal region in vivo (Klausnitzer and Manahan-Vaughan, 2008). While mossy fiber-CA3 synapses respond to 1Hz LFS with a potent facilitation, the commissural/associational CA3 synapses are unaffected by the same stimulation protocol (Klausnitzer and Manahan-Vaughan, 2008). In line with these findings, we compared short-term alterations in ATN responses mediated by low frequency MTT and Dfx stimulation. The transition from 0.025 Hz to 1 Hz resulted in immediate changes of the field potential for both groups. The similarity was apparent by the instant depression of FP slope for MTT (Fig. 3C), as for Dfx input (Fig. 3D) with no significant difference between the values of both groups (ANOVA, F(1,105) 5 1.032, P 5 0.17, n 5 5 for Dfx group and n 5 6 for MTT). However, MTT stimulation did not affect FP amplitude (Fig. 3C), while Dfx stimuli induced a gradually developing decrease of the ATN response (Fig. 3D). Repeated measures of the first 100 pulses of the amplitude values showed significant differences between both groups (ANOVA, F(1,105) 5 3.109, P < 0.01, n 5 5 for Dfx group and n 5 6 for MTT group). LFS of Dfx induced immediate depression of FP slope compared with baseline, which continued through the 4 h recording session for the occurrence of stable LTD (Fig. 3E, ANOVA, F(1,49) 5 7.24, P < 0.001, n 5 5). In the group of animals where the FP amplitude was measured after MTT stimulation, the same LFS protocol failed to induce LTD (Fig. 3E, n 5 7) in comparison to the nonstimulated group. Analysis comparing the poststimulation values for the LFS group to the last six time points of the baseline Hippocampus
before the stimulation protocol reveals potentiation of the evoked response (Fig. 3E, ANOVA, F(1,49) 5 5.15, P < 0.05, n 5 7). In order to increase the precision of plasticity detection and to decrease the effect of baseline pulse-dependent augmentation in MTT recordings, we timed the baseline recording for 120 min, enough for the FP values to reach a stable plateau. The large amplitude long-term synaptic modification known as LTP (Larson and Lynch, 1986; Buzsa´ki et al., 1987) was the next target of our hippocampodiencephalic investigation. High frequency stimulation (HFS) consisted of 10 bursts, with each burst containing 10 pulses at 100 Hz, with an interburst interval of 10 s. After 120 min of baseline recording, a HFS was applied to the MTT in one group, and to Dfx in a second group. For the MTT-stimulated group the poststimulation recordings showed a prominent and long-lasting increase of about 160% for FP slope (Fig. 3F, ANOVA, F(1,49) 5 4.02, P < 0.01, n 5 5) and for FP amplitude (ANOVA, F(1,49) 5 3.70, P < 0.01, n 5 5, data not shown) compared with the last six baseline time points. When the data were compared with the nonstimulated baseline group, in which the pulse-dependent elevation of the FP parameters reached about 125% (Fig. 2A), the HFS-induced potentiation appeared as an additional level of synaptic strength increase (FP amplitude: ANOVA, F(1,49) 5 4.16, P < 0.01, n 5 5; FP slope: ANOVA, F(1,49) 5 4.37, P < 0.01, n 5 5). The Dfx-stimulated group demonstrated a restricted post-tetanic plasticity. No detectable change in the FP amplitude occurred ANOVA, F(1,49) 5 1.22, P 5 0.11, n 5 7, data not shown) while a small but significant increase in the FP slope was evident (Fig. 3F,G, ANOVA, F(1,49) 5 2.09, P < 0.05, n 5 7). The dissociation between these two parameters of the FP raises a question concerning the degree to which the different components of LTP are involved (Bliss and Lomo, 1973). In order to present the HFS-induced alterations of all FP parameters, we measured the maximal slope of the FP, as defined by the steepest deviation of the pulse-evoked negative potential. Concurrently, we analyzed the latency of the FP negative peak measured in milliseconds from the FP positive onset. HFS to Dfx evoked an increase of the maximal slope (Fig. 3H, ANOVA, F(1,49) 5 5.12, P < 0.01, n 5 7) as well as a significant decrease of the FP latency (Fig. 3F, ANOVA, F(1,49) 5 2.47, P < 0.05, n 5 7). Both results suggest that the tetanic stimulation might affect the excitability of the recorded neuronal population, a phenomenon described in hippocampal recordings (Andersen et al., 1980). The present study reports, for the first time, that the anterior thalamic nuclei (ATN) are capable of long-term synaptic modification of their responses. This finding corroborates the idea that, far from being a passive receiver, the anterior thalamus plays an active role in amplifying the convergence of hippocampal and mammillary body inputs (Vann and Aggleton, 2004). Furthermore, we distinguished differing and specific short- and long-term plasticity properties for the direct and the indirect pathways. Spatial deficits after mammillary body damage are not as severe as those found after hippocampectomy (Thompson, 1981), and are typically less severe than those associated with
LONG-TERM SYNAPTIC PLASTICITY IN ANTERIOR THALAMUS anterior thalamic damage (Aggleton et al., 1995; Gaffan et al., 2001). The implication is that both the direct subicular—anterior thalamic pathway and the mammillary body—anterior thalamic pathway (which presumably involves indirect subicular influences) support memory processes. The present study, therefore, set out to compare the electrophysiological properties of the two major convergent routes upon the anterior thalamic nuclei (fornix vs. mammillothalamic tract). We found that the plasticity characteristics of both pathways express a tendency to oppose each other. While HFS of mammillothalamic pathway induces large-amplitude, stable LTP of FP slope and amplitude, the direct hippocampothalamic field plasticity after HFS is expressed only with a small amplitude increase of the intrinsic excitability, measured by FP slope. LFS to mammillothalamic tract did not evoke depression of the ATN response, while the same low-frequency protocol to the dorsal fornix was followed by a long-lasting and stable depression of both FP slope and amplitude. These results favor the role of the MMB as an input that elevates the polarity of ATN plasticity, which is concurrently lowered by the direct subicularthalamic input. Inputs from the MMB are also known to target inhibitory interneurons within the anterior thalamus (Wang et al., 1999), suggesting that the activation of GABAergic synapses could affect the shape of field response. Because the fiber density of MTT and Dfx might differ and this could bias the FP representation of the degree of synaptic activation via each input, we normalized the stimulus intensity for each pathway by recording with 40% of the maximum FP amplitude that can be evoked by the afferents activation. This approach has allowed the comparison of synaptic plasticity occurring in other structures with parallel inputs (Doyere et al., 1997; Kosub et al., 2005). Furthermore, we have detected a use-dependent plasticity of mammillothalamic synapses, which is BDNF-dependent. The detected augmentation of baseline recordings was pulse-, but not timing-dependent. This finding is consistent with a recent observation in freely moving rats that sensory thalamocortical synaptic transmission undergoes potentiation mediated by TrkB receptors (Tsanov and Manahan-Vaughan, 2007b). The longterm response to 900 pulses delivered with rate of 1 Hz suggests that MTT responses are not obeying the common model of plasticity, but rather follow a frequency-dependent activation. In addition to the opposing long-term synaptic effects of the hippocampothalamic and mammillothalamic connections, these two pathways also differ in their short-term plasticity properties. Several findings have demonstrated that different hippocampal regions respond dissimilarly to the change of test-pulses frequency from baseline to 1 Hz and these responses reflect the type of plasticity that each region expresses (Salin et al., 1996; Klausnitzer and Manahan-Vaughan, 2008). Pulses delivered with a frequency of 1 Hz induce immediate frequency depression of FP slope and amplitude of the fornix impulses that bypass the MMB. Interestingly, the same low-frequency stimulation fails to evoke frequency decrease of mammillothalamic synaptic weights. Short-term synaptic plasticity, in particular synaptic suppression, is an important component of the nonlinear temporal dynamics that lead to enhancement of neuronal
7
responses (Chance et al., 1998) and an increase in the signal-tonoise (S/N) ratio in neuronal processing (Abbott et al., 1997). We conclude that the rapid depression plasticity in the hippocampodiencephalic circuit is mediated by the anatomically shortest pathway—via the direct fornix projections to thalamus. Our data support and extend previous findings that reveal synaptic plasticity as one of the major properties underlying thalamic function in the adult brain (Rauschecker, 1998).
REFERENCES Abbott LF, Varela JA, Sen K, Nelson SB. 1997. Synaptic depression and cortical gain control. Science 275:220–224. Aggleton JP, Brown MW. 1999. Episodic memory, amnesia, and the hippocampal-anterior thalamic axis. Behav Brain Sci 22:425– 444. Aggleton JP, Desimone R, Mishkin M. 1986. The origin, course, and termination of the hippocampothalamic projections in the macaque. J Comp Neurol 243:409–421. Aggleton JP, Neave N, Nagle S, Hunt PR. 1995. A comparison of the effects of anterior thalamic, mamillary body and fornix lesions on reinforced spatial alternation. Behav Brain Res 68:91–101. Akaneya Y, Tsumoto T, Kinoshita S, Hatanaka H. 1997. Brain-derived neurotrophic factor enhances long-term potentiation in rat visual cortex. J Neurosci 17:6707–6716. Andersen P, Sundberg SH, Sveen O, Swann JW, Wigstrom H. 1980. Possible mechanisms for long-lasting potentiation of synaptic transmission in hippocampal slices from guinea-pigs. J Physiol 302:463–482. Bienenstock EL, Cooper LN, Munro PW. 1982. Theory for the development of neuron selectivity: Orientation specificity and binocular interaction in visual cortex. J Neurosci 2:32–48. Bliss TVP, Lomo T. 1973. Long-lasting potentiation of synaptic transmission in the dentate area of the anesthetized rabbit following stimulation of the perforant path. J Physiol 232:331–356. Buzsa´ki G, Haas HL, Anderson EG. 1987. Long-term potentiation induced by physiologically relevant stimulus patterns. Brain Res 435:331–333. Chance FS, Nelson SB, Abbott LF. 1998. Synaptic depression and the temporal response characteristics of V1 cells. J Neurosci 18:4785– 4799. Chen G, Kolbeck R, Barde YA, Bonhoeffer T, Kossel A. 1999. Relative contribution of endogenous neurotrophins in hippocampal long-term potentiation. J Neurosci 19:7983–7990. Conner JM, Lauterborn JC, Yan Q, Gall CM, Varon S. 1997. Distribution of brain-derived neurotrophic factor (BDNF) protein and mRNA in the normal adult rat CNS: Evidence for anterograde axonal transport. J Neurosci 17:2295–2313. Doyere V, Srebro B, Laroche S. 1997. Heterosynaptic LTD, depotentiation in the medial perforant path of the dentate gyrus in the freely moving rat. J Neurophysiol 77:571–578. Gaffan EA, Bannerman DM, Warburton EC, Aggleton JP. 2001. Rats’ processing of visual scenes: Effects of lesions to fornix, anterior thalamus, mamillary nuclei or the retrohippocampal region. Behav Brain Res 121:103–117. Gartner A, Polnau DG, Staiger V, Sciarretta C, Minichiello L, Thoenen H, Bonhoeffer T, Korte M. 2006. Hippocampal long-term potentiation is supported by presynaptic and postsynaptic tyrosine receptor kinase B-mediated phospholipase C gamma signaling. J Neurosci 26:3496–3504. Ishizuka N. 2001. Laminar organization of the pyramidal cell layer of the subiculum in the rat. J Comp Neurol 435:89–110. Hippocampus
8
TSANOV ET AL.
Kang H, Schuman EM. 1995. Long-lasting neurotrophin-induced enhancement of synaptic transmission in the adult hippocampus. Science 267:1658–1662. Kang H, Welcher AA, Shelton D, Schuman EM. 1997. Neurotrophins and time: Different roles for TrkB signaling in hippocampal longterm potentiation. Neuron 19:653–664. Klausnitzer J, Manahan-Vaughan D. 2008. Frequency facilitation at mossy fiber-CA3 synapses of freely behaving rats is regulated by adenosine A1 receptors. J Neurosci 28:4836–4840. Kosub KA, Do VH, Derrick BE. 2005. NMDA receptor antagonists block heterosynaptic long-term depression (LTD) but not longterm potentiation (LTP) in the CA3 region following lateral perforant path stimulation. Neurosci Lett 374:29–34. Kruger L, Saporta S, Swanson LW. 1995. Photographic Atlas of the Rat Brain: The Cell and Fiber Architecture Illustrated in Three Planes with Stereotaxic Coordinates. New York: Cambridge University Press. 299 pp. Larson J, Lynch G. 1986. Induction of synaptic potentiation in the hippocampus by patterned stimulation involves two events. Science 232:985–988. Lisman JE. 1994. The CaM-kinase hypothesis for the storage of synaptic memory. Trends Neurosci 17:406–412. Madison DV, Malenka RC, Nicoll RA. 1991. Mechanisms underlying long-term potentiation of synaptic transmission. Annu Rev Neurosci 14:379–397. Meibach RC, Siegel A. 1975. The origin of fornix fibers which project to the mammillary bodies in the rat: A horseradish peroxidase study. Brain Res 88:508–512. Messaoudi E, Bardsen K, Srebro B, Bramham CR. 1998. Acute intrahippocampal infusion of BDNF induces lasting potentiation of synaptic transmission in the rat dentate gyrus. J Neurophysiol 79:496–499. Moser EI, Krobert KA, Moser MB, Morris RG. 1998. Impaired spatial learning after saturation of long-term potentiation. Science 281:2038–2042. Rauschecker JP. 1998. Cortical control of the thalamus: Top-down processing and plasticity. Nat Neurosci 1:179–180. Salin PA, Scanziani M, Malenka RC, Nicoll RA. 1996. Distinct shortterm plasticity at two excitatory synapses in the hippocampus. Proc Natl Acad Sci USA 93:13304–13309.
Hippocampus
Sharma SK, Sherff CM, Stough S, Hsuan V, Carew TJ. 2006. A tropomyosin-related kinase B ligand is required for ERK activation, long-term synaptic facilitation, and long-term memory in aplysia. Proc Natl Acad Sci USA 103:14206–14210. Shelton DL, Sutherland J, Gripp J, Camerato T, Armanini MP, Phillips HS, Carroll K, Spencer SD, Levinson AD. 1995. Human trks: Molecular cloning, tissue distribution, and expression of extracellular domain immunoadhesins. J Neurosci 15:477–491. Snyder SE, Li J, Salton SR. 1997. Comparison of VGF, trk mRNA distributions in the developing and adult rat nervous systems. Brain Res Mol Brain Res 49:307–311. Thompson R. 1981. Rapid forgetting of a spatial habit in rats with hippocampal lesions. Science 212:959–960. Tsanov M, Manahan-Vaughan D. 2007a. The adult visual cortex expresses dynamic synaptic plasticity that is driven by the lightdark cycle. J Neurosci 27:8414–8421. Tsanov M, Manahan-Vaughan D. 2007b. Intrinsic, light-independent and visual-activity dependent mechanisms cooperate in the shaping of the field response in rat visual cortex. J Neurosci 27:8422–8429. Van der Werf YD, Scheltens P, Lindeboom J, Witter MP, Uylings HB, Jolles J. 2003. Deficits of memory, executive functioning and attention following infarction in the thalamus; a study of 22 cases with localised lesions. Neuropsychologia 41:1330–1344. Vann SD, Aggleton JP. 2004. The mammillary bodies: Two memory systems in one? Nat Rev Neurosci 5:35–44. von Cramon DY, Hebel N, Schuri U. 1985. A contribution to the anatomical basis of thalamic amnesia. Brain 108:993–1008. Wang B, Gonzalo-Ruiz A, Sanz JM, Campbell G, Lieberman AR. 1999. Immunoelectron microscopic study of gamma-aminobutyric acid inputs to identified thalamocortical projection neurons in the anterior thalamus of the rat. Exp Brain Res 126:369–382. Warburton EC, Baird A, Morgan A, Muir JL, Aggleton JP. 2001. The conjoint importance of the hippocampus and anterior thalamic nuclei for allocentric spatial learning: Evidence from a disconnection study in the rat. J Neurosci 21:7323–7330. Ying SW, Futter M, Rosenblum K, Webber MJ, Hunt SP, Bliss TV, Bramham CR. 2002. Brain-derived neurotrophic factor induces long-term potentiation in intact adult hippocampus: Requirement for ERK activation coupled to CREB and upregulation of Arc synthesis. J Neurosci 22:1532–1540.
HIPPOCAMPUS 21:9–21 (2011)
Hippocampal Signals for Strong Memory When Associative Memory Is Available and When It Is Not Peter E. Wais* ABSTRACT: The paired-associate task has been used with functional magnetic resonance imaging (fMRI) in studies that assessed the role of the medial temporal lobe (MTL) subserving recollection and familiarity. Some researchers have interpreted their results to mean that the hippocampus selectively subserves recollection and not familiarity [cf., Eichenbaum et al., (2007) Annu Rev Neurosci 30:123–152]. Yet many of these results confound recollection and familiarity with strong and weak memories, and it is not clear whether the conclusions represent differences between memory processes or memory strength. In the current study, participants were scanned with fMRI during retrieval in a paired-associate task, and a new approach separated the analysis of memory strength from the analysis of memory processes. The data were sorted by confidence level in an old/new task, and the high-confidence responses were compared in categories when associative memory was highly accurate and when it was not available. The results show that high-confidence memory produced increased activity in the hippocampus, relative to the level for forgotten pairs, both when associative memory was available and when it was not. Two interpretations are discussed for the behavioral results for when associative memory was not available: one account based on familiarity and the other account based on noncriterial recollection. The conclusion is that recognition of the word-pairs was based on familiarity when associative memory was not available. Together with the fMRI results that activity in two regions associated with cognitive control (left ventrolateral prefrontal cortex and left inferior parietal lobule) was greater when responses were based on associative memory than when based on familiarity, the findings suggest that the hippocampus supports strong memory and that cortical regions make an additional contribution to recollection. V 2009 WileyC
Liss, Inc.
KEY WORDS: hippocampus; fMRI; associative memory; recollection; familiarity; prefrontal cortex
INTRODUCTION Recognition is typically described as relying on two processes, recollection and familiarity (Mandler, 1980; Curran and Hintzman, 1995). Recollection refers to remembering an experience with associated details, whereas familiarity refers to a sense of awareness that is absent of associated details about a particular prior experience. Many studies have
Department of Psychology, University of California, San Diego, California Grant sponsors: Innovative Research Grant from the Kavli Institute for Brain and Mind at the University of California, San Diego. *Correspondence to: Peter E. Wais, Department of Neurology, University of California, San Francisco, 600 16th Street Room N474, San Francisco, CA 94158. E-mail:
[email protected] Accepted for publication 4 September 2009 DOI 10.1002/hipo.20716 Published online 15 December 2009 in Wiley Online Library (wileyonlinelibrary.com). C 2009 V
WILEY-LISS, INC.
examined recollection and familiarity using a paired-associate task. In a paired-associate task, recognition that two stimuli studied as a pair are in the intact form in which they were studied is thought to depend on recollection, whereas mistaken recognition that rearranged pairs are in the intact form is thought to indicate familiarity for one or both of the stimuli (Reinitz et al., 1992; Kroll et al., 1996; Giovanello et al., 2003; Stark and Squire, 2003; Kan et al., 2007). A number of studies have used functional magnetic resonance imaging (fMRI) to measure neural activity correlated with judgments in paired-associate tasks (Jackson and Schacter, 2004; Kirwan and Stark, 2004; Law et al., 2005; Staresina and Davachi, 2006; Chua et al., 2007; Habib and Nyberg, 2008; Giovanello et al., 2009). These researchers followed the typical interpretations that recognition of intact pairs depends on recollection of the association formed between stimuli and that recognition of rearranged pairs is based on familiarity for one of the stimuli when recollection of the association is not available. The key regions of interest in these fMRI studies were structures in the medial temporal lobe (MTL), including the hippocampal region (the hippocampal region includes the dentate gyrus, the hippocampus proper and the subiculum) and the perirhinal, entorhinal, and parahippocampal cortices. Based on findings that the hippocampus serves a critical role in the learning and memory of abstract associations (Eichenbaum, 2000; Brown and Aggleton, 2001), the researchers intended their fMRI studies to examine whether MTL regions subserve selective roles for recollection and familiarity (cf., Eichenbaum et al., 2007) or contribute to both recognition processes (cf., Squire et al., 2007). The results in the fMRI studies showed that hippocampal activity during encoding was greater in association with pairs subsequently recognized in the intact form in which they were studied than with pairs subsequently forgotten (Jackson and Schacter, 2004; Chua et al., 2007; Habib and Nyberg, 2008) or with subsequent recognition of one of the studied stimuli when associative memory failed (Kirwan and Stark, 2004; Staresina and Davachi, 2006). The results in the retrieval studies were mixed. Kirwan and Stark (2004) found that hippocampal activity increased during the recollection of paired associations, relative to the recognition of only one of the paired stimuli, but not relative to missing studied pairs. Law et al. (2005) found that hippocampal activity increased in a linear
10
WAIS
trend with increasing memory strength for an association. The results from these studies converge on the conclusion that increased activity in the hippocampus is correlated with binding associative memories, but it is not clear from these data whether the hippocampus subserves only recollection during retrieval (Kirwan and Stark, 2004; Law et al., 2005; Habib and Nyberg, 2008). Two different views about these fMRI data have developed, and these views describe opposing sides in the debate about a possible division of labor in the MTL between structures that subserve recollection and regions that subserve familiarity. For the studies in which the results showed that no MTL regions other than the hippocampus increased activity in correlation with recollection of the intact pair associations, the authors interpreted their results as evidence that the hippocampus selectively subserves recollection and that other MTL regions do not (Jackson and Schacter, 2004; Staresina and Davachi, 2006; Chua et al., 2007). In research that reported activity in other MTL regions that was correlated with recollection of associations in a pattern similar to that in the hippocampus, the authors rejected the view that a division of labor in the MTL involves a selective role for the hippocampus in recollection and the cortex in familiarity-based decisions and novelty detection (Kirwan and Stark, 2004; Law et al., 2005; Habib and Nyberg, 2008). There are several elements underlying the designs of these prior studies that make the interpretation of their results problematic. One problem is that none of the studies followed an approach that equated memory strength for the observations taken to be based on recollection with the observations taken to be based on familiarity and, as such, it is not clear whether the results represent differences between recollection and familiarity, or differences between robust memory and weak memory, or both. The confound of memory strength and recognition processes plagues the interpretation of source memory studies as well (Wixted, 2007) and has hindered the development of a satisfying answer about the roles of MTL structures in recollection and familiarity (Wais, 2008). Another problem is that recognition of the rearranged pairs can be interpreted as the result of a recall to reject strategy or as the result of strong familiarity for the one of the words. Moreover, interpretations for familiarity itself continue to be debated in the cognitive literature. One view interprets recognition responses as based on familiarity when indications for associative or source memory are incorrect, and this holds true even when memory confidence is high (Slotnick and Dodson, 2005; Wixted, 2007). Another view holds that recognition based on recollection produces memory confidence above a high threshold, and, consequently, strong memories are typically based on recollection of either associative information or some noncriterial detail from the learning experience (Yonelinas and Jacoby, 1996; Yonelinas, 1999). The goal for the current study was to address both of the problems above by using a new approach in a paired-associate task. For each word-pair, the first step tested confidence in recognition (which could have been based only on familiarity for the words, or familiarity for the words and recollection of the paired association), and the second step tested accuracy in Hippocampus
recollection for whether the word pairs were intact from study or rearranged. The experimental design avoided the confound of memory strength with memory processes and permitted interpretations about the contribution of recollection and familiarity in recognition decisions on the basis of the participants’ indications about associative memory. Participants gave their responses for old/new judgments and pair-type judgments using specific, six-level, confidence-rating scales. Confidence ratings of this type have been utilized commonly with experiments that have examined recognition memory, and these ratings were interpreted as indications of underlying memory strength (Kelley and Wixted, 2001; Slotnick and Dodson, 2005). Heathcote (2003) found that participants’ use of confidence ratings was consistent as an indication of memory strength criteria that they applied across experiments that manipulated the semantic and orthographic similarity between the words in test pairs. Although the old/new responses given by the participants here can be interpreted as analogous to indications for memory strength associated with their recognition decisions, the results are referred to in terms of confidence ratings. The results for the associative memory decisions are referred to in terms of accuracy of the pair-type judgments. The analysis compares fMRI activity associated with three categories of word pairs that were recognized with high confidence in the old/new task: decisions associated with high accuracy in the judgment of intact pairs (recollection), decisions associated with high accuracy in the judgment of rearranged pairs (recall to reject), and decisions associated with chance judgment (guesses) about pair types (familiarity for words in the absence of recollection for the paired association). The expected advantage of collecting associative memory judgments for intact and rearranged pairs with a six-level, confidence-rating scale was that a proportion of the responses would indicate guesses about the pair-types. The interpretation of the pair-type guesses as indications of familiarity for the words when associative memory was not available would be straightforward. Although the rearranged pair-type judgments made with high accuracy were interpreted as recall-to-reject responses, it is possible that some of those decisions were based on strong familiarity for one of the test words, as described earlier. The key comparison in the analysis was between the intact pair-type judgments based on recollection and the guesses about pair types based on familiarity for one of the target words. The aim was to examine the role of the hippocampus in retrieving strong memories when associative memory was available (accurate pair-type judgments) and when it was not (guesses about pair-type judgment).
MATERIALS AND METHODS Participants Informed consent was obtained from 18 students (11 females) at the University of California, San Diego. All participants
STRONG MEMORY IN THE HIPPOCAMPUS
11
FIGURE 1. Experimental design and behavioral analysis: in this example, the word-pair ‘‘teeth marriage’’ was studied and endorsed as maybe, probably or definitely ‘‘old’’ during the old/new recognition task. For the associative memory task, a response of ‘‘111’’ or ‘‘11’’ indicates that the pair is judged to
be in the intact form in which it was studied, a response of ‘‘1’’ or ‘‘2’’ indicates ‘‘maybe’’ pair-type judgments, and a response of ‘‘22’’ or ‘‘222’’ indicates that the pair is judged to be in a rearranged form. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
were right-handed and learned English as their native language. No participant had history of psychiatric or neurological disorders, or was taking medication that would affect the central nervous system.
The memory test for each participant was conducted in the MRI scanner approximately 1 h after the study session. Participants saw test pairs in five blocks of 40 pairs each (5.0 s per word-pair). Each test block included 30 target pairs from the study session, plus 10 foil pairs. Of the 30 target pairs in each block, 15 pairs were presented in the intact form in which they were studied, and 15 pairs were presented in a rearranged form in which the words had been taken from different study pairs. Both words in the 10 foil pairs were new. The test pairs were presented in a random order to each participant and intermixed with trials from an odd-even digit task described below (fMRI procedure and data analysis). For each pair presented in the test phase, participants first gave a confidence judgment as to whether the pair was old or new (1, definitely new; 2, probably new; 3, maybe new; 4, maybe old; 5, probably old; 6, definitely old) within a 2.5 s limit. Participants then gave a confidence judgment for their pair-type decision (‘‘111’’ definitely intact, ‘‘11’’ probably intact, ‘‘1’’ maybe intact, ‘‘2’’ maybe rearranged, ‘‘2 2’’ probably rearranged, ‘‘2 2 2’’ definitely rearranged) within a 2.5 s limit. Participants were instructed not to enter a pair-type judgment for pairs endorsed as new. The old/new scale and the pair-type decision scale were each presented for 2.5 s beneath the test pair on each trial (Fig. 1). To facilitate fMRI analysis (see later), participants also performed an odd/even classification task (Stark and Squire, 2001) on trials randomly intermixed with the memory task. For this baseline task, the digits 1–9 were presented for 1.25 s each in blocks of 2, 4, 6, or 12 digits. Performance on this task has been associated with levels of MTL activity that can be effectively contrasted as a baseline vs. activity associated with memory tasks (Stark and Squire, 2001). This contrastive approach in fMRI can determine only the relative levels of activity
Stimuli A total of 400 English nouns were selected from the MRC Psycholinguistic Database with the following constraints: word frequency of 50 to 300, length of 5 to 12 letters, and two to four phonemes. The words were sorted into 200 pairs, taking care to minimize semantic association between the words in each pair as much as possible. The word-pairs were randomly assigned to either the target list (150 word-pairs) or the foil list (50 word-pairs). All participants saw the same list of target pairs and lure pairs.
Behavioral Procedure and Data Analysis Participants viewed a list of word-pairs presented on a desktop computer after being instructed to study each pair ‘‘so that you can best remember the words as a pair during a memory test in your next session.’’ Pairs were presented in five blocks of 30 pairs each. In each trial, the words were centered on the computer screen and presented on the same line in black New Times Roman font. Pairs in each block were randomly ordered for each participant. During each 5.0 s trial in the study session, participants viewed the pairs, formed their own associations, and responded yes or no whether they hade made a successful association between the words. Before beginning the study session, participants completed a brief practice session as an orientation to the task (the practice word-pairs were all names of major cities).
Hippocampus
12
WAIS
TABLE 1. Distribution of Behavioral Responses (n518) (a) Mean proportions of Old/New judgments by confidence rating Judgment: New 1 2 Targets 0.03 (0.01) 0.10 (0.02) Lures 0.15 (0.03) 0.30 (0.04)
3 0.09 (0.01) 0.22 (0.03)
4 0.15 (0.02) 0.17 (0.02)
5 0.18 (0.03) 0.10 (0.02)
Old 6 0.44 (0.04) 0.06 (0.02)
(b) Mean proportions of pair type judgments for high-confidence hits Judgment: Rearranged 2 2 2 22 Intact pairs 0.06 (0.02) 0.07 (0.03) Rearranged pairs 0.34 (0.02) 0.27 (0.03)
2 0.11 (0.04) 0.18 (0.03)
1 0.07 (0.03) 0.08 (0.04)
11 0.14 (0.02) 0.07 (0.03)
Intact 111 0.55 (0.02) 0.06 (0.03)
(a) the Old/New judgments for targets and lures are shown as the mean proportion in each confidence rating (standard error of the mean); and (b) the associative memory judgments for high-confidence hits (targets given old/new ratings of 5 or 6) are shown as the mean proportion in each confidence rating for pair-types (1, 11, 111 5 correct for Intact pairs, and 2, 2 2, 2 2 2 5 correct for Rearranged pairs).
associated with the tasks, however, and caution must be taken against interpreting how close those levels may be to zero or a tonic level of activity.
fMRI Scanning Parameters, Procedure, and Data Analysis Imaging was carried out in a GE Signa Excite 3T scanner at the Center for Functional MRI (University of California, San Diego). Functional images were acquired using a gradient-echo, echo-planar, T2*-weighted pulse sequence (TR 5 2.5 s, TE 5 30 ms, 908 flip angle, bandwidth 5 250 MHz, FOV 5 22 cm). A total of 41 slices covering the whole brain were acquired perpendicular to the long axis of the hippocampus (matrix size 5 64 3 64, slice thickness 5 5 mm). Following five functional runs, high-resolution structural images were acquired using a T1-weighted, fast spoiled gradient-echo (FSPGR) pulse sequence (TE 5 3.1 ms, 128 flip angle, FOV 5 25 cm, 172 slices, 1 mm slice thickness, matrix size 5 256 3 256). Between word presentations, participants were given 0, 2, 4, 6, or 12 trials of the 1.25 s baseline task that served to jitter the MR signal acquired for subsequent deconvolution of the hemodynamic response function (hrf ). For each participant, the fMRI data were partitioned into six categories based on the old/new confidence ratings provided on each trial. The six categories were as follows: (a) high-confidence hits (i.e., correct old responses to target pairs that were rated 5 or 6) with correct judgment of intact pairs as in the intact form in which they were studied; (b) high-confidence hits (50 s or 60 s) with correct judgment of rearranged pairs as rearranged from the form in which they were studied; (c) high-confidence hits (50 s or 60 s) with ‘‘maybe’’ pair-type judgments (correct and incorrect pairtype decisions made with low confidence); (d) misses (targets rated 1, 2, 3, or in some cases, 4); (f ) false alarms (foil pairs rated 4, 5, or 6); and (g) correct rejections (foil pairs rated 1, 2, or 3). For each of the six categories, the hemodynamic response (relative to the baseline condition) was estimated for the 15 s following the onset of the presentation of the word pair in the Old/New task by using signal deconvolution and the AFNI Hippocampus
suite of programs (afni.nimh.nih.gov). Data analysis was then performed on the area under the hrf from 0 to 15 s following the presentation of the word-pair (at about 15 s, the hrf returned to baseline) based on the interpretation that pair-type judgments were formed during the old/new task response and the unavoidable auto-correlation between successive TR’s during the old/new and pair-type judgment tasks. The anatomical scans and the fMRI data were normalized to the template of the Talairach-Tournoux brain atlas. Functional data were resampled to 2 3 2 3 2 mm3 voxels and blurred with a 4 mm FWHM Gaussian kernel. These data were used for the whole brain analysis. For the analysis of MTL activity, the region of interest large deformation diffeomorphic metric mapping method (ROI-LDDMM, Miller et al., 2005) of alignment was used to improve cross-participant alignment and increase statistical reliability (Kirwan et al., 2007). Voxel-based, pairwise t-tests (threshold of P < 0.005, twotailed) were then carried out as group analyses across all 18 participants with participants as random effects. Both the whole brain and the MTL analyses were based on the area under the hrf for contrasts of interest (described later). Monte Carlo simulations (AlphaSim application in AFNI) were then used to correct for multiple comparisons and to determine how large a cluster of voxels was needed to be statistically significant (P < 0.05).
RESULTS Behavioral Results The participants affirmed that they formed associations between the words combined as a pair during the encoding task with a mean success rate of 63% 6 2%. In the results from the recognition task, old/new responses for the word pairs were (73% 6 3%) correct, and d 0 5 1.29 6 0.08. The distribution of responses for hits and false alarms across the six-level old/new confidence scale revealed a bias to respond ‘‘old’’ (Table 1: old/new judgements by confidence rating), which is
STRONG MEMORY IN THE HIPPOCAMPUS
13
FIGURE 2. Accuracy of the old/new judgments and pair-type judgments (standard error of the mean): (a) mean percent correct for the responses given at each confidence level in the old/new task (n 5 18); and (b) mean accuracy of the judgments for Intact
and rearranged pairs that have been collapsed into categories with accurate intact and rearranged judgments or guesses about pairtypes (n 5 14). The dotted lines in both figures indicate the level of chance.
not uncommon for similar parametric tests of recognition confidence collected during tests in the fMRI scanner (e.g., Ranganath et al., 2004a; Daselaar et al., 2006). As a result, accuracy of the responses to targets given a confidence level of 4 (maybe old) was at chance overall [Fig. 2a, (49% 6 3%) correct]. The accuracy of responses to targets given a confidence rating of 5 or 6 was much higher (66% 6 3% correct and 91% 6 2% correct, respectively). Typically, a target is considered forgotten if it is incorrectly declared to be new (i.e., if it receives a confidence rating of 1, 2, or 3). However, as indicated earlier, most of the participants in this experiment exhibited a liberal response bias such that ratings of 4 were as likely to be given to targets as to foils. In that case, a confidence rating of 4 for a target indicates a forgotten item as well. In the analyses, old/new confidence ratings of 1, 2, 3, or 4 were taken to reflect forgotten pairs for the 15 of 18 participants who exhibited no better than chance accuracy when responding 4 (maybe old), and responses of 1, 2, or 3 to were taken to reflect forgotten pairs for the remaining 3 participants whose old/new confidence ratings of 4 were associated with above chance accuracy. Hereafter, this approach to analyzing the data will be referred to as the comparison between the high-confidence hits and the forgotten pairs. For the high-confidence hits, overall accuracy associated with intact/rearranged decisions was (72% 6 2%) correct, and association d 0 5 1.27 6 0.10 (Table 1b: pair-type judgements for high-confidence hits). The comparisons of particular interest in the fMRI data were between the neural activity associated with each of the three associative memory categories and the forgotten pairs. This approach provided for the comparison of fMRI activity when old/new confidence was high and availability of
associative memory differed across the recollection categories (i.e., intact and rearranged judgments) and the category when associative memory was weak or not available (i.e., ‘‘maybe’’ judgments). The intended analysis was to equate the level of old/new confidence across observations for the three associative memory categories by considering only observations when target pairs were endorsed as ‘‘definitely old,’’ (i.e., an old/new rating of 6). Analysis of the fMRI data, however, required at least 10 observations from each participant in each of the conditions of interest (the forgotten items and each of the three associative memory categories of recognized pairs). To improve power to analyze the fMRI data for all 18 participants, the target pairs endorsed in the Old/New task as high-confidence hits were grouped in categories of confident Intact judgments (ratings of ‘‘111’’ and ‘‘11’’), confident Rearranged judgments (ratings of ‘‘2 2 2’’ and ‘‘2 2’’) and low-confidence ‘‘maybe’’ judgments (ratings of ‘‘1’’ and ‘‘2’’). Collapsing trials as described earlier to make the fMRI analysis possible yielded categories for which mean old/new confidence was high, if not fully equated. Mean old/new confidence varied for the three categories of pair-type judgments according the accuracy of those judgments: for intact, mean old/new confidence was 5.64 6 0.06 and pair-type judgments were 82% 6 2% correct; for rearranged, mean old/new confidence was 5.14 6 0.12; and pair-type judgments were 72% 6 2% correct; and for ‘‘maybe’s,’’ mean old/new confidence was 4.91 6 0.08 and pair-type judgments were 58% 6 2% correct. For a subgroup of 14 participants, the accuracy of associative memory for high-confidence hits with ‘‘maybe’’ pair-type judgments was at chance and indicated that recognition of these word-pairs was made with high confidence but without Hippocampus
14
WAIS
TABLE 2. Behavioral Results When Old/New Confidence is Equated (n514): Mean Old/New Ratings are Presented for the High-Confidence Hits with Mean Accuracy of the Corresponding Intact and Guess Pair-Type Judgments (Standard Error of the Mean) Pair-type judgment: Mean old/new rating Mean accuracy for pair-type
Guess
Intact
5.66 (0.07) 0.54 (0.03)
5.76 (0.05) 0.89 (0.02)
associative memory from the study session. For this subgroup, pair-type judgments were 89% 6 2% correct for intact pairs; 75% 6 3% correct for rearranged pairs; and 54% 6 3% correct for judgments of ‘‘maybe intact’’ or ‘‘maybe rearranged.’’ The latter category will be referred to hereafter as high-confidence hits with guess judgments (Fig. 2b). Old/New confidence associated with intact, rearranged, and guess pair-type judgments was 5.82 6 0.03, 5.54 6 0.08, and 5.61 6 0.08, respectively. Across the 14 participants in this subgroup, there were sufficient numbers of observations in each scanning run for each participant that it was possible to perform a second-level analysis in which old/new memory strength for high-confidence hits was equated for observations in the guess condition with observations in the intact condition (Table 2).
Interpretation of Associative Memory Categories (n 5 14) Participants gave separate responses for recognition in the old/new task and for recollection in the paired-associate task, which allowed for an assessment of confidence about associative memory that was not confounded with confidence about item memory for one or both of the words. The three categories of pair-type judgments assessed the contribution of recollection underlying recognition of the target pairs. The comparison across the high-confidence hits, therefore, examined conditions when recognition was strong and retrieval of associative memory was made with high confidence and high accuracy vs. when recognition was strong and retrieval of associative memory was made with little confidence and low (or chance) accuracy. The fMRI analyses described later compared the activity associated with the three categories of high-confidence hits against the activity associated with forgotten pairs. The mean accuracy of pair-type decisions increased across the categories of high-confidence hits from guess to rearranged to intact judgments. Mean old/new confidence, however, did not increase in a fashion directly related to increasing recollective strength. These results suggest that recognition confidence associated with rearranged pairs was weaker, on average, than for the recollected intact pairs and no different than when only one of the words in a pair was recognized (i.e., guesses). One interpretation for the weaker level of old/new confidence for the high-confidence hits with correct rearranged pair-type judgments is that these decisions were based on some Hippocampus
combination of weak recollection for the paired association, which had been violated in the rearranged pair, with weak familiarity for the target words (Kelley and Wixted, 2001).
fMRI Results The main objective for the current study was to compare activity in the MTL in conditions when recognition was based on associative memory (recollection-based decisions) with equally strong recognition when associative memory was not available. An ANOVA comparing the MTL activity associated with the three categories of high-confidence hits and the forgotten pairs identified a region in the left hippocampus where activity associated with high-confidence hits with intact judgments and with rearranged judgments was greater than activity associated with the forgotten pairs (voxel-wise threshold, P < 0.005; cluster size 53 voxels, P-corrected < 0.05). This result shows that the left hippocampus supported strong recollection of the paired-associations and that activity in this region increased as recollection strength increased (i.e., accuracy was 74% for rearranged judgments and increased to 88% for intact judgments). Activity in this region associated with the maybe judgments was not different than the level associated with the forgotten pairs and indicated that the left hippocampus did not support weak recollection (i.e., accuracy was 57% for maybe judgments). As noted earlier, however, four participants made few high-confidence hits with maybe judgments, and an interpretation of the null result related to hippocampal activity associated with the maybe category should be taken with caution considering the low power available for this particular analysis across the 18 participants. As shown in Table 1(pair-type judgements for highconfidence hits) and Figure 2b, the accuracy of the ‘‘maybe’’ pair-type judgments was at chance for 14 participants, and this category is referred to as guess in the analyses. An ANOVA comparing MTL activity associated with the recollection-based judgments (intact and rearranged), the judgments when associative memory was not available (guesses), and the forgotten pairs identified regions in the hippocampus, bilaterally, where activity increased in conjunction with increasing recollective strength (voxel-wise threshold, P < 0.005; Fig. 3). Significantly, in the right hippocampus, activity associated with the intact, rearranged, and guess judgments was greater than for the forgotten pairs (cluster size of 21 voxels, P-corrected < 0.05; pairwise comparisons of activity associated with guesses > forgotten, P < 0.01; activity associated with intact > guesses, P < 0.001). In the left hippocampus, increased activity for the guesses relative to the forgotten pairs approached significance (cluster size of 51 voxels, P-corrected < 0.05; pairwise comparisons of activity associated with guesses > forgotten, P < 0.06). These results show that the hippocampus supported highly confident recognition both when associative memory was available and when it was not. Although the basic analysis found no other MTL regions that showed selectively increased activity for any of the three
STRONG MEMORY IN THE HIPPOCAMPUS
15
FIGURE 3. Bilateral activity in the hippocampus is shown for 14 participants. ANOVA comparing the BOLD signal for the three categories of high-confidence hits and the forgotten pairs revealed that activity associated with intact (INT), with rearranged (RAR), and with guess (GUE) pair-type decisions is greater than for the forgotten pairs (P-corrected < 0.05, error bars represent the meansquare error); and the time courses for the hemodynamic response
for each condition (next to each region of interest with s.e.m.). The clusters of activity are localized on coronal views from the averaged anatomical scans of the MTL (y 5 216) for the left hippocampus (cluster of 51 voxels) and the right hippocampus (cluster of 21 voxels). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
associative memory categories, using a relaxed threshold that does not account for multiple comparisons (i.e., the clusterextent was smaller than the P-corrected < 0.05 threshold calculated with AlphaSim), an area in left perirhinal cortex showed
activity in association with all high-confidence hits that was greater than for the forgotten pairs, and the levels of activity across the associative memory categories were not different from each other. Hippocampus
16
WAIS
Next, as cortical regions have also been associated with recollection of context but not item familiarity in recent fMRI studies (Law et al., 2005; Staresina and Davachi, 2006; Wais et al., in press), the whole-brain data were examined to determine whether regions outside of the MTL showed greater activity for the recollection-based decisions (intact and rearranged) than the recollection-absent decisions (guesses). In relevant reviews, Badre and Wagner (2007) have identified functional regions in the ventrolateral prefrontal cortex (VLPFC) and Ciaramelli et al. (2008) have identified functional regions in the inferior parietal lobule (IPL) where activity is increased when recollection is available in comparison with when memory is based on item recognition. An ANOVA comparing whole-brain activity associated with high-confidence hits with intact, rearranged, and guess pairtype decisions did not identify activity in these cortical regions of interest (voxel-wise threshold, P < 0.005). Taking into account that this test may have lacked power because it included only the data from the participants (n 5 14) in the preceding MTL results, a direct comparison of the observations based on strong recollection was made with the observations based on recognition when associative memory was not available. A voxel-wise t-test (threshold at P < 0.005) that contrasted activity associated with high-confidence hits with intact judgments vs. activity associated with high-confidence hits with guess judgments identified a region in the left midVLPFC (pars triangularis, approximately BA45, cluster size of 15 voxels, p-corrected < 0.05; Fig. 4a) and a region in the inferior parietal cortex (inferior parietal lobule, approximately BA 40, cluster size of 64 voxels, P-corrected < 0.05; Fig. 4b) where activity was greater for intact than guess judgments. These results are convergent with research suggesting that cortical regions mediate the cognitive control processes that are recruited during successful associative memory (Ranganath et al., 2004b; Dobbins and Wagner, 2005; Vincent et al., 2006). A remaining question that arises from the fMRI results is about why hippocampal activity increased with decisions based on associative memory relative to decisions based on similarly strong recognition but when associative memory was not available. In other words, did the experience of recollection depend on a selective signal supported by the hippocampus or on other signals from cortical regions that conjoined with a strong hippocampal recognition signal? It was not possible to equate strength across all three categories of pair-type decisions, and, to examine this question with the data available, an additional analysis was performed in which memory confidence (indexed by ratings for the old/new decision) was equated for intact and guess decisions (Table 2). An ANOVA that compared activity associated with intact judgments, guess judgments, and the forgotten pairs identified a region in the right hippocampus where signals for the two categories of high-confidence hits were different from each other and different from the forgotten pairs (voxel-wise threshold, P < 0.005; Fig. 5). Right hippocampal activity associated with intact judgments was increased relative to the forgotten pairs, whereas activity associated with guess Hippocampus
judgments was decreased relative to the forgotten pairs (cluster size of 19 voxels, P-corrected < 0.05). These data show that the right hippocampus signaled high-confidence memory both when associative memory was available and when it was not. Although this pattern of activity at first appears unusual, given that hippocampal activity associated with recognition is typically greater than activity associated with missing studied items (i.e., when memory is very weak or absent), Kirwan and Stark (2004) found similar results that right hippocampal activity was greater when intact pairs were recognized than when pairs were missed but, also in that same region, activity for missed pairs was greater than recognizing rearranged pairs. The results in the additional analysis here also identified a region in the left hippocampus where activity was increased for strong recollection (cluster size of 51 voxels, P-corrected < 0.05). It is important to note that considering the level of spatial resolution in these fMRI data, a definitive conclusion cannot be made whether the same, different or a combination of neurons included in the right hippocampal region of interest in Figure 5 participated in the opposing activation signals associated with Intact judgments on the one hand and with Guess judgments on the other hand. Other research has been interpreted, however, to show that hippocampal activity associated with strong familiarity is decreased relative to both recollectionbased decisions and forgotten items (Yonelinas et al., 2005; see Fig. 1e). Notably, according to the interpretation by Yonelinas et al. (2005), activity in those same bilateral regions associated with strong recollection-based responses was increased relative to the forgotten items. Although the Yonelinas et al. (2005) study did not equate memory strength for recollection-based decisions with familiarity-based decisions, at least not in a fashion agreeable with the interpretation here, the pattern in their result is similar to that in the current study. Moreover, the difference in the results here between the comparison across associative memory categories (Fig. 3) and the comparison between the intact and guess judgments equated by old/new confidence (Fig. 5) suggests more than a selective role for recollection in right hippocampal function. As the level of activity associated with the guesses is different between the two functionally defined regions of interest, so is the location of their respective peak-voxels of activity. The potentially distorting effects of normalization and spatial smoothing are reasons to use caution in an interpretation about whether these regions of interest are juxtaposed or essentially overlap, but their functional activity shows that the right hippocampus signals high-confidence memory based on old/new confidence differently than high-confidence memory based on associative memory judgments.
DISCUSSION The current study used a new approach with a paired-associate task and removed two substantial problems that compromise the interpretations from prior research that suggested that the
STRONG MEMORY IN THE HIPPOCAMPUS
17
FIGURE 4. Cortical activity of interest is shown for 14 participants. T-tests comparing the BOLD signal for two categories of high-confidence hits in the left mid-VLPFC (approximately pars triangularis) and in the left inferior parietal lobe (approximately the angular gyrus) show that activity associated with intact (INT) pair-type decisions is greater than with guess (GUE) pair-type deci-
sions (P-corrected < 0.05, error bars represent the mean-square error). The clusters of activity are localized on coronal views from the averaged anatomical scans for the left mid-VLPFC (y 5 19, cluster of 15 voxels) and the left inferior parietal lobule (y 5 266, cluster of 64 voxels). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
hippocampus is recruited selectively during recollection (Giovanello et al., 2003; Jackson and Schacter, 2004; Staresina and Davachi, 2006; Chua et al., 2007; Vilberg and Rugg, 2007). The confound of memory confidence with memory processes was avoided in the analysis here by comparing the highconfidence hits from the old/new task across three categories of pair-type judgments. The approach also showed that responses in one of the three categories of high-confidence hits were based on low confidence and accuracy at chance for associative
memory judgments. Therefore, the associative memory analysis separated recollection-based responses from familiarity-based responses notwithstanding the ambiguity about the contribution of recollection in rearranged pair-type judgments. The fMRI results show that the hippocampus was engaged both when recollection of the word-pair associations was available and when it was not. Different interpretations about the basis of the pairtype guesses, however, lead to different conclusions about the roles of the hippocampus subserving recollection and familiarity. Hippocampus
18
WAIS
FIGURE 5. Bilateral activity in the hippocampus is shown for 14 participants (left hippocampus cluster of 51 voxels and right hippocampus cluster of 19 voxels). ANOVA comparing the BOLD signal for two categories of high-confidence hits with equated memory confidence and the forgotten pairs revealed that activity associated with intact (INT) and with guess (GUE) pair-type decisions is different than for the forgotten pairs in the right hippo-
campus (P-corrected < 0.05, error bars represent the mean-square error). The time courses for the hemodynamic response for each condition are shown next to each region of interest (s.e.m.). A comparison of the time courses extracted from the regions of interest in the right hippocampus in Figures 3 and 5 suggest that these clusters of activity do not overlap. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
Removing the Confound of Memory Strength With Memory Processes
strong recollection and familiarity because familiarity-based judgments are, on average, made with less confidence than recollection-based judgments (Slotnick and Dodson, 2005; Wixted, 2007; Wais et al., 2008). In the fMRI studies that did not separate the highly confident familiarity-based responses as the current study did, the mean level of activity associated with all familiarity-based responses included judgments made with
In the results from prior paired-associate studies, the comparisons of recollection-based responses with familiarity-based responses are generalized across a range from weak memory to strong memory (i.e., confidence in an old/new decision). Such an approach is not likely to result in an analysis of comparably Hippocampus
STRONG MEMORY IN THE HIPPOCAMPUS weak to high confidence. This approach has typically found a mean level of activity in the hippocampus associated with familiar items that is not different than forgotten items (Jackson and Schacter, 2004; Kirwan and Stark, 2004; Staresina and Davachi, 2006; Vilberg and Rugg, 2007). When confidence about old/new judgments is equated, as intended in the analysis here, the problem of averaging MTL activity associated with strong confidence and weak confidence in old/new responses is avoided. Activity can be contrasted between conditions of comparable old/new confidence when recollection is available and when it is not, and the view that the hippocampus subserves a selective role for recollection can be tested.
Interpretation 1: Pair-Type Guesses Indicate Responses Based on Familiarity The design in the current study offered the participants two ways to indicate recollection: first, that the Intact pairs were in the intact form in which they were studied; and second, that the rearranged pairs were rearranged from the study association. Participants gave a confidence rating with each recollection response. Critically, this approach tested associative memory with both intact pairs and rearranged pairs to give participants the option to indicate that associative memory was not available by responding that test pairs were ‘‘maybe’’ intact or ‘‘maybe’’ rearranged. In the behavioral results, the associative memory decisions given with high confidence (i.e., definitely or probably intact/rearranged) were made with high accuracy, yet the accuracy for associative memory decisions given with low confidence (i.e., ‘‘maybe intact’’ or ‘‘maybe rearranged’’) was at chance. In other words, the ‘‘maybe’’ pair-type responses amounted to guesses because studied associations for either of the words could not be recalled. The participants’ responses, therefore, indicated that the pair-type guesses were based on familiarity. According to the interpretation that the pair-type guesses were based on familiarity, the fMRI results show that the hippocampal signal increased for familiarity-based judgments and for recollection-based judgments, relative to the forgotten pairs. This appears to be the first finding for a gradient in hippocampal activity that signals high-confidence familiarity and then increases further in association with high-confidence recollection. Another test of the interpretation that the pair-type guesses denoted decisions based on familiarity is to examine whether the selection processes necessary for recollection were recruited. Associative memory depends on selection of relational and contextual information that is relevant to the retrieval goal. Selection processes involve top-down cognitive control that recruits regions in the PFC (Ranganath et al., 2004b; Badre and Wagner, 2007) and bottom-up allocation of attention that recruits posterior parietal regions (Cabeza, 2008; Ciaramelli et al., 2008). Goal-directed, top-down processes that guide the selection of information relevant to the recollection of an association are
19
subserved by a region in the left VLPFC (Dobbins and Wagner, 2005). Increased activity in this center of cognitive control has also been associated with recollection, but not with familiarity, in studies that examined recognition memory and the functional organization of the MTL (Jackson and Schacter, 2004; Law et al., 2005; Staresina and Davachi, 2006; Habib and Nyberg, 2008; Kirwan et al., 2008). Stimulus-driven, bottom-up processes that direct attention to task-relevant information have been associated with a region in the left inferior parietal lobule (IPL). Activity in this region increases in conjunction with activity in the hippocampal formation during recollection (Vincent et al., 2006), and this pattern is hypothesized as the neural network that evokes the ‘‘pop out’’ experience when a match is perceived between studied and retrieved associations (Ciaramelli et al., 2008). In the fMRI results here, activity in the left mid-VLPFC (Fig. 4a) was greater in association with high-confidence hits with intact judgments, which indicated highly accurate associative memory, than with high-confidence hits with guess judgments, which indicated when associative memory was not available. This result supports the interpretation that the pairtype guesses were based on familiarity by showing that cognitive control involved in the selection of associative information was recruited during retrieval that led to intact judgments, but not during retrieval that led to guess judgments. Additionally, if reading the two words at test retrieved a salient association formed during the study session, then participants’ attention to the stimuli should have increased because the pairing was highly relevant to their task. Therefore, during the associative memory probe, word pairs recollected to be intact would be expected to ‘‘pop out’’ to the participants as obvious matches with the study session. In the fMRI results, activity in a region of the left IPL (Fig. 4b) was greater in association with the high-confidence hits given intact judgments than given guess judgments. This pattern of IPL activity has been interpreted in other research as evidence that the allocation of attention to memory contents increased during recollection, but not during recognition based on simple familiarity for items (Cabeza, 2008; Ciaramelli et al., 2008).
Interpretation 2: Pair-Type Guesses Indicate Responses Based on Noncriterial Recollection An alternative interpretation for the high-confidence hits with pair-type guesses is that although participants indicated that they were guessing when they responded ‘‘maybe’’ about the pair-associations, they had actually recollected some other detail associated with one of the test words. In other words, the high confidence that participants had expressed in the old/new task was based on recollection of an association that was irrelevant to the pairing studied in the experiment, or about which the participant was not aware when giving a ‘‘maybe’’ response. This alternative view draws on the interpretation that highly confident recognition, such as the high-confidence hits in the data here, is based almost exclusively on recollection (Yonelinas, 1999; Parks and Yonelinas, 2007). According to this view, Hippocampus
20
WAIS
recollection produces the strongest memories and is evident when recognition confidence is given a rating that exceeds a high threshold, even when recollection can be analyzed by additional thresholds above the ‘‘high threshold’’ (Parks and Yonelinas, 2009). As a consequence, the target pairs rated ‘‘6’’ and given incorrect or ‘‘maybe’’ pair-type responses should be taken as noncriterial recollection, or as decisions based on taskirrelevant recollection about which the participant was unaware. Prior behavioral studies examining associative memory developed the hypothesis for noncriterial recollection as a third, independent process supporting recognition (Yonelinas and Jacoby, 1996). According to this account, noncriterial recollection is variable in strength and subject to false alarms, yet it exceeds the high-confidence threshold that is the telltale sign for recollection-based responses. Yonelinas and Jacoby (1996) reported that when participants were compelled to give an associative memory response during difficult discrimination conditions, they sometimes defaulted to the ‘‘automatic process’’ of noncriterial recollection that functioned as familiarity. Applied to the data from the current study, this would mean that the challenge of the 2.5 s response interval for each test trial in the fMRI scanner caused some of the responses to be based on ‘‘automatic,’’ noncriterial recollection and to be given as pairtype guesses.
CONCLUSIONS Using methods such as paired-associate or source memory tasks, retrieval based on recollection can be determined without question. Demonstration that memory retrieval is based solely on familiarity, however, is debatable because of the possibility that recognition could be based on recollection for some association about which the participant is not aware. Three points from the current behavioral data, prior examinations of noncriterial recollection, and the current fMRI data helped form the conclusion here that the pair-type guesses indicated high-confidence memory based on familiarity, albeit the interpretation based on noncriterial recollection cannot be ruled in or ruled out. First, participants were informed that each test pair was either ‘‘old’’ and intact from study, ‘‘old’’ and rearranged from study, or completely new. Having given a highly confident ‘‘old’’ response, the associative memory judgment was then between intact and rearranged, and a correct pair-type decision indicated recollection from the study session. A guess about the pair-type, therefore, indicated that associative memory was not available for either of the target words. This approach is straightforward and supports a conclusion about the guesses that is free from the ambiguity about the contribution of recollection which compromises the interpretation of the Rearranged pair-type decisions (Reinitz et al., 1992; Kelley and Wixted, 2001). Second, the principle distinction that arises for any dual-process theory of recognition is that recollection is measured in Hippocampus
terms of the specific demands of the memory task. A strict definition of recollection, such as the test instructions in the current study, makes clear that recollection is the retrieval of associative information that supports the required discrimination (i.e., between intact and rearrange pairs), and that familiarity reflects anything else that is retrieved. When Yonelinas and Jacoby (1996) directly examined the differences between recollection, noncriterial recollection and familiarity, they concluded that noncriterial recollection is not functionally separable from familiarity and that ‘‘noncriterial recollection is most appropriately treated as being familiarity.’’ Third, in the fMRI data, increased activity for the intact, relative to the guess, pair-type judgments in regions of the VLPFC and the IPL follows the same pattern reported by other studies that have examined task-specific recollection effects (Rugg et al., 2003; Habib and Nyberg, 2008). The results here show task-specific effects in cortical regions subserving cognitive control in association with responses based on recollection, but not in association with responses taken to be based on familiarity. Mandler (1980) originally proposed that familiarity is the type of recognition that succeeded when effortful recollection of specific details or associations fails. In other words, when you are sure that you know the man who you are sitting near on the bus, but cannot remember why, your recognition of his face is based on familiarity. According to this view, your recollection of the specific details that make the man memorable as the neighborhood baker is (perhaps temporarily) not available to you. It is this experience of familiarity that participants used in the study here to give their highly confident response that a pair was old when they could not recollect the association for either of the words. Their ‘‘maybe’’ response for the pair-type decision denoted that the words were familiar from the study session, but they could not recollect why. Results in the current study identified a region in the right hippocampus where the BOLD signal increased in association with high-confidence memory that was based on familiarity and on recollection (Fig. 3). In an adjacent region in the right hippocampus, activity associated with high-confidence recollection increased relative to the forgotten pairs, whereas activity associated with high-confidence familiarity decreased relative to the forgotten pairs (Fig. 5). The results here also suggest that prefrontal and parietal regions contribute together with the hippocampus in the neural network that subserves recollection. Further research is needed to examine how the hippocampus serves in conjunction with other MTL structures, as well as cortical networks, that have been shown in other work to support recognition based on familiarity.
Acknowledgments The author wishes to thank Larry Squire for his comments concerning the motivation and analysis of the study and John Wixted for his comments on an earlier draft of this manuscript.
STRONG MEMORY IN THE HIPPOCAMPUS
REFERENCES Badre D, Wagner A. 2007. Left ventrolateral prefrontal cortex and the cognitive control of memory. Neuropsychologia 45:2883–2901. Brown M, Aggleton J. 2001. Recognition memory: What are the roles of the perirhinal cortex and hippocampus? Nat Rev Neurosci 2:51– 61. Cabeza R. 2008. Role of parietal regions in episodic memory retrieval: The dual attentional process hypothesis. Neuropsychologia 46: 1813–1827. Chua E, Schacter D, Rand-Giovannetti E, Sperling R. 2007. Evidence for a specific role of the anterior hippocampal region in successful associative encoding. Hippocampus 17:1071–1080. Ciaramelli E, Grady C, Moscovitch M. 2008. Top-down and bottomup attention to memory: A hypothesis (AtoM) on the role of the posterior parietal cortex in memory retrieval. Neuropsychologia 46: 1828–1851. Curran T, Hintzman D. 1995. Violations of the independence assumption in process dissociation. J Exp Psychol Learn Mem Cogn 21:531–547. Daselaar S, Fleck M, Cabeza R. 2006. Triple dissociation in the medial temporal lobes: Recollection, familiarity, and novelty. J Neurophysiol 96:1902–1911. Dobbins I, Wagner A. 2005. Domain-general and domain-sensitive prefrontal mechanisms for recollecting events and detecting novelty. Cereb Cortex 15:1768–1778. Eichenbaum H. 2000. Cortical-hippocampal networks for declarative memory. Nat Rev Neurosci 1:51–61. Eichenbaum H, Yonelinas A, Ranganath C. 2007. The medial temporal lobe and recognition memory. Annu Rev Neurosci 30:123–152. Giovanello K, Verfaellie M, Keane M. 2003. Disproportionate deficit in associative recognition relative to item recognition in global amnesia. Cogn Aff Behav Neurosci 3:186–194. Giovanello K, Schnyer D, Verfaellie M. 2009. Distinct hippocampal regions make unique contributions to relational memory. Hippocampus 19:111–117. Habib R, Nyberg L. 2008. Neural correlates of availability and accessibility in memory. Cereb Cortex 18:1720–1726. Heathcote A. 2003. Item recognition memory and the receiver operating characteristic. J Exp Psychol Learn Mem Cogn 29:1210–1230. Jackson O, Schacter D. 2004. Encoding activity in anterior medial temporal lobe supports subsequent associative recognition. Neuroimage 21:456–462. Kan I, Giovanello K, Schnyer D, Makris N, Verfaellie M. 2007. Role of the medial temporal lobes in relational memory: Neuropsychological evidence from a cued-recognition paradigm. Neuropsychologia 45:2589–2597. Kelley R, Wixted J. 2001. On the nature of associative information in recognition memory. J Exp Psychol Learn Mem Cogn 27:701– 722. Kirwan C, Stark C. 2004. Medial temporal lobe activation during encoding and retrieval of novel face-name pairs. Hippocampus 14: 919–930. Kirwan C, Jones C, Miller M, Stark C. 2007. High-resolution investigation of the medial temporal lobe. Hum Brain Mapp 10:959– 966. Kroll N, Knight R, Metcalfe J, Wolf E, Tulving E. 1996. Cohesion failure as a source of memory illusions. J Mem Lang 35:176–196. Law J, Flanery M, Wirth S, Yanike M, Smith A, Frank L, Suzuki W, Brown E, Stark C. 2005. Functional magnetic resonance imaging activity during the gradual acquisition and expression of pairedassociate memory. J Neurosci 25:5720–5729.
21
Mandler G. 1980. Recognizing: The judgment of previous occurrence. Can Psychol Rev 87:252–271. Miller M, Beg M, Ceritoglu C, Stark C. 2005. Increasing the power of functional maps of the medial temporal lobe by using large deformation diffeomorphic metric mapping. Proc Natl Acad Sci USA 102:9685–9690. Parks C, Yonelinas A. 2007. Moving beyond pure signal-detection models: Comment on Wixted. Psychol Rev 114:188–202. Parks C, Yonelinas A. 2009. Evidence for a memory threshold in second-choice recognition memory responses. Proc Natl Acad Sci USA 106:11515–11519. Ranganath C, Yonelinas A, Cohen M, Dy C, Tom S, D’Esposito M. 2004a. Dissociable correlates of recollection and familiarity within the medial temporal lobes. Neuropsychologia 42:2–13. Ranganath C, Cohen M, Dam C, D’Esposito M. 2004b. Inferior temporal, prefrontal, and hippocampal contributions to visual working memory maintenance and associative memory retrieval. J Neurosci 24:3917–3925. Reinitz M, Lammers W, Cochran B. 1992. Memory-conjunction errors: Miscombination of stored stimulus features can produce illusions of memory. Mem Cognit 20:1–11. Rugg M, Henson R, Robb W. 2003. Neural correlates of retrieval processing in the prefrontal cortex during recognition and exclusion tasks. Neuropsychologia 41:40–52. Slotnick S, Dodson C. 2005. Support for a continuous (single-process) model of recognition memory and source memory. Mem Cognit 33:151–170. Squire L, Wixted J, Clark R. 2007. Recognition memory and the medial temporal lobe: A new perspective. Nat Rev Neurosci 8: 872–883. Staresina B, Davachi L. 2006. Differential encoding mechanisms for subsequent associative recognition and free recall. J Neurosci 26: 9162–9172. Stark C, Squire L. 2001. When zero is not zero: The problem of ambiguous baseline conditions in fMRI. Proc Natl Acad Sci USA 98: 12760–12766. Stark C, Squire L. 2003. Hippocampal damage equally impairs memory for single items and memory for conjunctions. Hippocampus 13:281–292. Villberg K, Rugg M. 2007. Dissociation of the neural correlates of recognition memory according to familiarity, recollection, and amount of recollected information. Neuropsychologia 45:2216–2225. Vincent J, Snyder A, Fox M, Shannon B, Andrews J, Raichle M. 2006. Coherent spontaneous activity identifies a hippocampal-parietal network. J Neurophysiol 96:3517–3531. Wais P. 2008. FMRI signals associated with memory strength in the medial temporal lobes: A meta-analysis. Neuropsychologia 46: 3185–3196. Wais P, Mickes L, Wixted J. 2008. Remember/know judgments probe degrees of recollection. J Cogn Neurosci 20:400–405. Wais P, Squire L, Wixted J. 2010. In search of recollection and familiarity signals in the hippocampus. J Cogn Neurosci 22:109–123. Wixted J. 2007. Dual-process theory and signal-detection theory of recognition memory. Psychol Rev 114:152–76. Yonelinas A. 1999. The contribution of recollection and familiarity to recognition and source-memory judgments: A formal dual-process model and an analysis of receiver operating characteristics. J Exp Psychol Learn Mem Cogn 25:1415–1434. Yonelinas A, Jacoby L. 1996. Noncriterial recollection: Familiarity as automatic, irrelevant recollection. Conscious Cogn 5:131–141. Yonelinas A, Otten L, Shaw K, Rugg M. 2005. Separating the brain regions involved in recollection and familiarity in recognition memory. J Neurosci 25:3002–3008.
Hippocampus
HIPPOCAMPUS 21:22–32 (2011)
The M-Current Inhibitor XE991 Decreases the Stimulation Threshold for Long-Term Synaptic Plasticity in Healthy Mice and in Models of Cognitive Disease ´ ngela Fonta´n-Lozano, Irene Sua´rez-Pereira, Jose´ Marı´a Delgado-Garcı´a, A ´ ngel Manuel Carrio´n* and A
ABSTRACT: Aging, mental retardation, number of psychiatric and neurological disorders are all associated with learning and memory impairments. As the underlying causes of such conditions are very heterogeneous, manipulations that can enhance learning and memory in mice under different circumstances might be able to overcome the cognitive deficits in patients. The M-current regulates neuronal excitability and action potential firing, suggesting that its inhibition may increase cognitive capacities. We demonstrate that XE991, a specific M-current blocker, enhances learning and memory in healthy mice. This effect may be achieved by altering basal hippocampal synaptic activity and by diminishing the stimulation threshold for long-term changes in synaptic efficacy and learning-related gene expression. We also show that training sessions regulate the M-current by transiently decreasing the levels of KCNQ/Kv7.3 protein, a pivotal subunit for the M-current. Furthermore, we found that XE991 can revert the cognitive impairment associated with acetylcholine depletion and the neurodegeneration induced by kainic acid. Together, these results show that inhibition of the M-current as a general strategy may be useful to enhance cognitive capacities in healthy and aging individuals, as well as in those with neurodegenerative diseases. V 2009 Wiley-Liss, Inc. C
KEY WORDS: synaptic plasticity; learning; memory; LTP; XE911; gene expression; kainate-induced neurodegeneration
INTRODUCTION Among voltage-gated K1 currents, the M-current is a primary transducer of changes in the chemical composition of the extracellular environment into modifications of intrinsic neuronal properties. The M-current was first identified in amphibian peripheral neurons (Brown and Adams, 1980) but later, it was also found in the mammalian peripheral and central nervous system (Halliwell and Adams, 1982; Brown, 1988). It is a low threshold, slowly activating and deactivating, and noninactivating voltage-dependent K1 current that limits repetitive firing and causes spike-frequency adaptation (Rowaski, 2000). Most of the bioDivisio´n de Neurociencias, Universidad Pablo de Olavide de Sevilla, Carretera de Utrera Km. 1, Sevilla, Spain Additional Supporting Information may be found in the online version of this article. Grant sponsor: Junta de Andalucı´a; Grant number: CVI-122; Grant sponsor: DGICYT; Grant number: BFI2002-00936. *Correspondence to: A´ngel M. Carrio´n, Divisio´n de Neurociencias, Universidad Pablo de Olavide, Carretera de Utrera Km. 1, Sevilla 41013, Spain. E-mail:
[email protected] Accepted for publication 4 September 2009 DOI 10.1002/hipo.20717 Published online 17 November 2009 in Wiley Online Library (wileyonlinelibrary.com). C 2009 V
WILEY-LISS, INC.
physical and pharmacological properties of the M-current are recapitulated upon heteromeric expression of K1 channel subunits encoded by members of the KCNQ/Kv7 gene family. Within that family, the KCNQ/Kv7.2 and three subunits playing a dominant role at most neuronal sites (Wang et al., 1998; Cooper et al., 2000), although KCNQ/Kv7.5 transcripts have also been ubiquitously detected in human brain (Lerche et al., 2000). The kinetic properties of the M-current are suggestive of a role in regulating neuronal excitability. Suppression of the M-current by muscarinic agonists or by selective blockers of the M-channels, such as linopirdine (Aiken et al., 1995; Costa and Brown, 1997; Lamas et al., 1997; Schnee and Brown, 1998) or XE991 (Zaczek et al., 1998), causes an increase in intrinsic excitability. Indeed, the suppression of the M-current shifts the firing mode of hippocampal CA1 pyramidal cells from regular firing to burst firing by augmenting the spike after-depolarization (Yue and Yaari, 2004), as well as by reducing the intrinsic subthreshold theta resonance (Hu et al., 2002; Peters et al., 2005). As spike bursts and subthreshold membrane potential oscillations are believed to be important for synaptic plasticity (Magee and Johnston, 1997; Thomas et al., 1998) and network oscillation (Buzsa´ki, 2002; Franse´n et al., 2004), the M-current may not only be important to control excitability but also for brain functions such as memory. On the basis of the physiological role of the M-current, it has been postulated that M-current suppressors may potentially enhance cognitive processes in certain circumstances (Brioni et al., 1993; Jentsch, 2000 for a review). Here we show that systemic administration of the M-current inhibitor, XE991, facilitates learning and memory in healthy mice by decreasing the stimulation threshold for certain cognitive processes and for long-term changes in synaptic efficacy. In addition, this inhibitor induces the expression of bdnf and arc. Furthermore, we show that synaptic activity induced by training provokes a transient decrease of KCNQ/ Kv7.3 in the hippocampus and perirhinal cortex. Finally, systemic administration of XE991 recovered cognitive deficits induced by transient cholinergic depletion and by the neurodegeneration induced by
ROLE OF KCNQ/Kv7 CHANNELS IN LEARNING AND MEMORY FACILITATION kainate acid administration. All these data suggest that XE991 may be a potential palliative treatment to augment the cognitive capacity of healthy and aging patients.
MATERIALS AND METHODS Animals Eight-week-old male Swiss mice, weighing between 25 and 30 g, were maintained on a 12/12 h light/dark cycle, under controlled environmental conditions. All behavioral and electrophysiological studies were carried out in accordance with European Union Council guidelines (86/609/EU) and following Spanish regulations for the use of laboratory animals in chronic experiments (BOE 67/8,509-12, 1988). Furthermore, all the experiments were approved by the local institutional animal care committee.
Drug Administration Fifteen minutes before to behavioral or electrophysiological testing, 10,10-bis(4-pyridinylmethyl)-9(10H)anthracenone dihydrochloride (XE991; 2.5 mg/kg i.p., Tocris, UK) or the saline alone was administered to the mice. This dose (2.5 mg/kg) of XE991 was judged to be optimal from dose–response curves in the object recognition test and from electrophysiological recordings. A double dose (5 mg/kg) induced status epilepticus (revealed by electrophysiological recording) and did not facilitate cognition in cognition tests in healthy mice. To generate the models of cognitive deficiency, 0.3 mg/kg of scopolamine (Sigma, Madrid, Spain) was administrated 20 min before training. For the model of neurodegenerative, kainate (Sigma, Madrid, Spain) was administrated (15 mg/kg) and after 21 days, learning and memory experiments were performed.
Behavioral Methods For the object recognition test, mice were tested in a rectangular arena (55 cm 3 40 cm 3 40 cm) located in a room with dim lighting and constant background noise. In the object recognition protocol, two different objects were placed in the arena during the training phase. After a delay time, one object was changed with a novel object. The aim was to test the animal’s memory of the original object by comparing the amount of time spent exploring the novel object against that for the familiar one. Selected objects consisted of plastic pieces with different forms and were thoroughly cleansed with 70% ethanol between trials to ensure the absence of olfactory cues. Before the experiment, the mice were habituated to the arena in the absence of objects for 20 min/day on 2 consecutive days. On the day of testing, the mice were left for 5, 10, or 15 min to explore the two objects. Retention tests were performed at the indicated times after the training session by placing the mice back in the arena for a 10-min session and by randomly exchanging one of the familiar objects with a novel one. The
23
time spent exploring each object was recorded, and the relative exploration of the novel object was expressed by a discrimination index (DI 5 (tnovel 2 tfamiliar)/(tnovel 1 tfamiliar)). DIs !0.2 were considered as efficient memory. The criteria for exploration were based strictly on active exploration. Exploration of an object was defined as directing the nose toward the object at a distance of (1.5 cm and/or touching it with the nose or its vibrissae. Going around or sitting on the object were not considered exploratory behavior. All trials were run with the experimenter blind to the drug treatment conditions. For all the behavior experiments, n was equal to six mice for each experimental group.
Electrophysiology Electrodes were implanted in the animals using stereotaxic coordinates (Paxinos and Franklin, 2001), as described in the study of Fonta´n-Lozano et al. (2007). Bipolar stimulating electrodes were implanted on the Schaffer’s collateral-commissural pathway of the dorsal hippocampus (from Bregma, AP: 1.5; L: 2.2 mm; depth from brain surface, 1.0–1.5 mm), and two recording electrodes were implanted in the ipsilateral stratum radiatum, underneath the CA1 area (from Bregma, AP: 2.2; L: 2.2 mm; depth from brain surface, 1.0–1.5 mm). All in-vivo recordings were performed at least 7 days after surgery. To evoke LTPs, each animal received five pulse trains (200 Hz, 100 ms) at a rate of 1/s. This protocol was administered either once or a total of six times at intervals of 1 min. The hippocampal activity recorded was stored digitally on a computer through an analog/digital converter (CED 1401 Plus, Cambridge, England) at a sampling frequency of 11–22 kHz and with an amplitude resolution of 12 bits. Computer programs (Spike 2 and SIGAVG from CED) were adapted to represent the extracellular synaptic field potential (fEPSP) recordings, and the slope of the evoked fEPSPs was collected as the first derivative (i.e., V/s) of the fEPSP records (V). Accordingly, five successive evoked field synaptic potentials at intervals of 5 min were averaged, and the mean value of the slope was determined for the rise-time period (i.e., the period of the slope between the initial 10% and the final 10% of the evoked field potential). The power spectrum of the hippocampal field activity was calculated using the fast Fourier transformation with a Hanning window. This parameter was expressed as the relative power and averaged across each session. The average was analyzed and compared using the wide-band model, considering the following bands: low theta (2–4 Hz) and theta (4–9 Hz). For paired-pulse facilitation, two stimuli of an intensity that evoked 35–40% of the maximum fEPSP response were delivered with an interstimulus interval of 50–200 ms. The percentage facilitation was calculated as (slope S2/slope S1) 3 100.
Mice Preparation for Immunohistochemistry and Gene Expression Studies Mice were habituated to the arena in the absence of objects for 20 min/day on 2 consecutive days. On the day of testing, Hippocampus
24
FONTA´N-LOZANO ET AL.
untrained and trained mice groups were left for 10 or 15 min to explore the arena without or with objects respectively. After that, animals were killed in each case at the indicated time.
Tissue Preparation and Immunohistochemistry To analyze the expression of the KCNQ/Kv7 subunits, five mice were killed by decapitation from each experimental group. The brain from each mouse was removed and the tissue was fixed by immersion in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 24 h at 48C. The tissue was cryoprotected in 30% sucrose-PBS for 2 days at 48C, embedded in 30% sucrose, and then maintained at 48C until it was sectioned on a cryotome. Coronal brain sections (50 lm) were processed for free-floating immunohistochemistry (de los Santos-Arteaga et al., 2003) using the specific antiserum raised against the different KCNQ/Kv7 subunits at a 1:500 dilution (antibodies supplied by Dr. Villarroel and characterized in Yus-Najera et al., 2003). To quantify the expression of this subunit, the optical density of different brain areas was measured using the image-J software. To minimize the variability of this technique, at least five sections from each animal were analyzed in at least two independently stained immunohistochemistry experiments.
FIGURE 1. XE991 administration enhanced learning and memory processes. Bar diagram illustrating the discrimination indices in the object recognition task obtained during training, short-term memory (STM, 1 h after training) and long-term memory (LTM, 24 h after training) sessions in mice trained for 5, 10, and 15 min (white, gray, and black bars, respectively) administered saline (left pot) or XE991 (right plot, n 5 8 for each group). *Statistical significance of the comparison between each test session and the training (OR memory) session in the same group. **P ! 0.01.
RESULTS XE991 an IM Antagonist that Improves Learning and Memory in Healthy Mice
mRNA Analysis by Reverse Transcription-PCR Hippocampus from mice killed 1 h after nontrained or training session were obtained. Total RNA from the hippocampus was extracted using Tripure reagent (Roche Products, Hertforshire, UK), and the RNA from a minimum of four animals per group was used for reverse transcription (RT)-PCR experiments. The primers used for PCR were as follows: for exon IV bdnf, 50 -CAGGAGTACATATCGGCCACCA-30 and 50 -GTAGGCCAAGTTGCCTTGTCCGT-30 ; and for arc, 50 CCAAGAAGTGGTGGGAGTTC-30 and 50 -AGTGTCTGG TACAGGTCCCG-30 . Arbitrary units were computed as the ratio between the optical density band corresponding to the gene studied in the 20–30th cycle and that of the gadph gene in the 15th amplification cycle. One unit was considered to be the ratio corresponding to the band with the lowest optical density of the genes studied in each experiment.
Statistical Analysis The statistical analysis of the results was performed with the SPSS package for Windows (SPSS, Chicago, IL). Unless otherwise indicated, data are represented as the mean 6 standard error of mean (SEM). The data collected were analyzed using a two-way analysis of variance (ANOVA) test with time or session as the repeated measure, and coupled to a contrast analysis where appropriate. One-way ANOVA allowed the statistical differences between the groups to be checked. Hippocampus
In the last two decades, it was suggested that inhibitors of the IM-potassium current might be useful tools to diminish the cognitive deficits caused by neurodegeneration. However, the best known IM-current inhibitor, linopirdine, has so far failed to show significant benefits in patients with Alzheimer’s disease (Rockwood et al., 1997; Borjesson et al., 1999; Cooper and Jan, 2003). XE991 is a new and more potent inhibitor of the IM-current (Zaczek et al., 1998) and it may have properties as a cognitive enhancer. To determine whether XE991 administration enhanced the cognitive abilities of healthy adult mice, we first studied training time dependence for the acquisition and storage of object recognition (OR). This capacity was tested in three groups of mice submitted to a 5, 10, or 15 min training session for OR memory, and short- or long-term memory (STM and LTM, respectively) was then tested 1 or 24 h after the training session (Fig. 1). Only the mice submitted to 15 min training displayed high values in the STM or LTM discrimination index (DIs) when compared to the remaining groups, as suggested by the session 3 training time interaction (F(8,63) 5 198.62, P < 10234). When we administrated XE991 (2.5 mg/kg, i.p.) 15 min before the 5 or 10 min OR training session, STM was similar to that observed in saline injected mice trained for 15 min. By contrast, only the XE991 injected mice subjected to a 10 min training session for OR displayed similar LTM, as indicated by the session 3 training time interaction (F(5,42) 5 15.82, P < 1025). Hence, the IM-current appears to influence learning and memory, and its inhibition by XE991 facilitates OR learning and memory in healthy mice.
ROLE OF KCNQ/Kv7 CHANNELS IN LEARNING AND MEMORY FACILITATION
25
Training Modulates KCNQ/Kv7.3 Channel Expression in the Hippocampus and Perirhinal Cortex
XE991 Enhances Long-Term Changes in the Synaptic Plasticity of Hippocampal CA3-CA1 Synapses
The KCNQ/Kv7 family of potassium channel is made up of five members (KCNQ/Kv7 1–5, reviewed in Jentsch, 2000), of which KCNQ/Kv7.2, 3, and 5 are widely expressed in the brain (Saganich et al., 2001; Jensen et al., 2005). To determine whether OR training influences the expression of KCNQ/Kv7.2, 3, and 5, we performed immunohistochemical analysis in brain tissue from untrained and trained mice sacrificed 1, 3, 6, and 24 h after the training session. We focused our analysis on the hippocampus and perirhinal cortex, two of the brain areas related with the acquisition and storage of OR memories (Brown and Aggleton, 2001; Dere et al., 2007; Fonta´n-Lozano et al., 2007). A rapid and transient decrease of around a 40% of the KCNQ/Kv7.3 subunit protein was evident in the perirhinal cortex between 1 and 6 h after training (Fig. 2A; F(4,10) 5 67.17, P < 0.01), although basal levels of KCNQ/Kv7.3 protein had been recovered 24 h after training. By contrast, we did not detect any changes in the expression of KCNQ/Kv7.2 and 5 after training. In the hippocampus, similar results were observed in the CA1 and CA3 fields, except that KCNQ/Kv7.3 remained below the basal levels of expression 24h after training (Fig. 2B, t(5) 5 8.81, P < 0.01 and t(5) 5 10.03, P < 0.01 for CA1 and CA3 fields, respectively). These results suggest that OR learning selectively regulated the protein levels of KCNQ/Kv7.3 in the hippocampus and perirhinal cortex.
Because XE991 administration seems to affect hippocampal functioning, we examined the effects of this drug on the basal synaptic state in mice and on LTP evoked in the Schaffer’s collateral-CA1 synapse in behaving mice. We used paired-pulse facilitation (PPF) to study the basal synaptic state with interpulse intervals ranging from 50 to 200 ms (Figs. 4A,B). In saline injected mice, the PPF was maximum at an interpulse interval of 50 ms and it diminished to basal levels at a 200 ms interpulse interval (F(2,6) 5 5.53, P 5 0.043). By contrast, the PPF was higher in XE991 injected mice at all the times tested when compared with saline injected mice (F(5,12) 5 20.44; P < 0.01), suggesting that the CA3-CA1 synapse was functionally affected in XE991 injected mice. This alteration might indicate that neuronal excitability to a second pulse is augmented in hippocampal neurons in XE991 injected mice. As long-term changes in synaptic efficacy in the hippocampus seem to be necessary to establish OR memory (Daoudal and Debanne, 2003; Gruart et al., 2006), we examined the effect of XE991 administration on the LTP of synaptic transmission evoked in the CA3-CA1 synapse in the mice. When we used high-frequency stimulation (HFS) protocols in saline injected mice, long-term changes in synaptic efficacy were only observed when six HFS were administrated. One or three HFS only provoked short-term changes in synaptic efficacy (Figs. 4C,D). However, in XE991 injected mice three HFS produced a sustained response similar to that provoked by the six HFS in saline injected mice (Figs. 4C,E). In addition, a single HFS protocol in XE991 injected mice provoked LTP but, with a weaker amplitude than that provoked by three HFS in XE991 injected mice or with six HFS in saline injected mice. These results indicate that long-term synaptic plasticity evoked by HFS trains is facilitated by XE991 administration.
XE991 Administration Regulates Basal Hippocampal Electrical Properties in Mice The hippocampus is a cortical region involved in information processing and memory consolidation (Squire and ZolaMorgan, 1991). To study the basal activity in the hippocampus, we implanted a bipolar stimulation electrode in the Schaffer’s collateral and a recording electrode among the apical dendrites of the CA1 field. The background hippocampal activity recorded in saline- and XE991 injected mice showed a transient change (over the 2 h after XE991 administration) in terms of amplitude and oscillatory properties during exploratory behavior (Fig. 3A). A Fourier analysis showed an increase in the amplitude of the power spectrum in the low-theta (124 Hz) and theta (529 Hz) oscillatory range (Fig. 3B). In addition, a relative spectrum analysis (Fig. 3C) detected an increase in the low-theta band oscillations in the hippocampus from XE991 injected mice (42.37% 6 2.18% and 19.45% 6 4.54% in XE991 and saline injected mice, respectively; t(5) 5 6.43, P < 0.05) and a decrease in theta oscillatory range (42.91% 6 1.89% and 57.72% 6 1.62% in XE991 and saline injected mice, respectively; t(5) 5 8.34, P < 0.01) when compared to saline injected mice. These data suggested that the M-current modulates basal hippocampal electrical oscillatory activity.
Hippocampal Training-Dependent Gene Expression Is Facilitate by XE991 Administration The IM-current inhibitor XE991 seems to cause an increase in neuronal excitability and facilitate the acquisition and storage of new information. Long-term memory stabilization requires new protein synthesis (Dudai, 2004; Inda et al., 2005; Alberini, 2005) and thus, we analyzed changes in gene expression during OR memory consolidation after 10 or 15 min training using semiquantitative RT-PCR (Fig. 5). We focused our attention on the bdnf gene, a trophic factor required for memory consolidation and associated to learning facilitation (Lee et al., 2004; Barco et al., 2005), and on the arc gene, a cytoskeletal-associated protein required for memory consolidation (Plath et al., 2006; Ploski et al., 2008). In saline-treated mice, 15 min OR memory training induced a consistent increase in bdnf and arc gene expression in the hippocampus with respect to untrained mice (t(7) 5 17.36, P < 0.01; and t(7) 5 17.7, P < 0.01; for bdnf and arc, respectively). AdminisHippocampus
26
FONTA´N-LOZANO ET AL.
FIGURE 2. OR memory training decreased the KCNQ/Kv7.3 levels in pivotal brain areas. Levels of the different KCNQ/Kv7 subunits were determined by immunohistochemistry in brains from mice killed 1, 3, 6, and 24 h after training session. Untrained mice were used as controls (five animals per each groups). (A) Bar diagrams illustrating the densitometric quantification of the immunoreactivity for different KCNQ/Kv7 subunits in the perirhinal cortex. Micrographs of representative KCNQ/Kv7 subunit immunohistochemistry in untrained and trained mice. Squares, circles,
and triangles represents KCNQ/Kv7.2, 3, and 5, respectively. (B) Bar diagrams illustrating the densitometric quantification of the immunoreactivity for each KCNQ/Kv7 subunit in the hippocampal CA1 and CA3 field. Representative micrographs of immunohistochemistry for each KCNQ/Kv7 subunit in untrained and trained mice are also shown. *Statistical significance of the densitometric difference between habituated and trained mice for each region. **P < 0.01.
tering XE991 to untrained mice did not significantly increase either bdnf or arc gene expression in the hippocampus with respect to saline injected untrained mice. Interestingly, XE991
treatment before a 10 min OR training session induced comparable levels of bdnf and arc mRNA expression in the hippocampus to that obtained in 15 min trained mice. These results
Hippocampus
ROLE OF KCNQ/Kv7 CHANNELS IN LEARNING AND MEMORY FACILITATION
FIGURE 3. XE991 administration induces alterations in the basal hippocampal electrophysiological properties. (A) Three seconds of a basal electrocorticogram recording from two selected saline- and XE991-injected mice are shown (saline- and XE991injested mice are represented in black and gray, respectively). The times in the XE991 administered mice are with respect to the injection. (B) Mean power spectra of hippocampal local field activity recorded from the CA1 pyramidal layer of saline or XE991 injected mice during exploratory behavior (n 5 5 mice per groups). (C) Relative spectrum quantification (mean 6 SEM) in the low theta, theta, alpha, beta, and gamma ranges for saline- and XE991 injected mice (n 5 5 for each group). *Statistical significance between saline and XE991 injected mice. *P ! 0.05; **P ! 0.01.
indicated that XE991 administration exerts a permissive role on gene expression related to cognitive processes. A consequence of this permissive effect may be the decrease in the stimulation threshold for long-term memory consolidation.
XE991 Improves Learning and Memory in Mouse Models of Cholinergic Depletion and Neurodegeneration As XE991 enhances cognitive abilities in healthy adult mice, we evaluated whether these agents were able to diminish
27
learning and memory impairment in mouse models related to Alzheimer disease. Scopolamine is a muscarinic cholinergic receptor antagonist that produces cognitive impairment by diminishing cholinergic neurotransmission (Lal et al., 1988). A single intraperitoneal dose of scopolamine (0.3 mg/kg) produced severe short-term (D.I.s: saline, 0.44 6 0.03; scopolamine, 0.041 6 0.074; t(15) 5 11.56, P < 0.01) and long-term (D.I.s: saline, 0.3 6 0.03; scopolamine, 0.016 6 0.031; t(15) 5 6.96, P < 0.01; Fig. 6A) memory defects when compared to saline injected mice. To test whether XE991 was able to restore scopolamine induce cognitive impairment, XE991 was coinjected together with scopolamine 15 min before training. Under these conditions, XE991 improved both short-term (D.I. for XE991, 0.31 6 0.035; t(15) 5 6.01, P < 0.01) as well as long-term (D.I. for XE991, 0.38 6 0.041; t(15) 5 6.09, P < 0.01) memory. Kainic acid (KA) is a neurotoxic agent that produces hippocampal neuronal death and cognitive impairment (Stajstrom et al., 1993). He et al. (2004) reported that 20 mg/kg of KA provokes epileptiform-like activity over several hours. In our hands, 15 mg/kg of KA provoked a seizure of stage 2–3 severity in the rating scale defined by Schauwecker and Steward (1997). This epileptic status disappeared 3–4 h after KA administration. As epileptiform-like activity may itself provoke cognitive deficiencies, we checked learning and memory deficiencies induced by KA 21 days of after KA injection. In this case, learning and memory deficiencies induced by KA are probably due to hippocampus neurodegeneration (see Fonta´nLozano et al., 2008) rather than to changes in electrical activity. When compared with saline injected mice, KA-treated mice displayed severe short-term (D.I.s: saline, 0.44 6 0.03; KA, 0.029 6 0.051; P < 0.001; Fig. 6B) and long-term (D.I.s: saline, 0.360.03; KA, 0.065 6 0.023; P < 0.001) memory deficits. However, when XE991 was injected 15 min before training, both short-term (D.I. for XE991 is 0.34 6 0.064; t(15) 5 10.6, P < 0.01) as well as long-term (D.I. for XE991 is 0.35 6 0.036; t(15) 5 7.76, P < 0.01) memory improved.
DISCUSSION A number of psychiatric and neurological disorders, as well as age related cognitive decline and mental retardation are associated with learning and memory impairments. The dominant paradigm for the development of treatments for these disorders is based on the specific mechanisms that underlie each of these diseases. However, this approach is hindered by the fact that there are a large number of different causes for such cognitive deficits. In addition, genetic heterogeneity has been associated with all the major causes of cognitive deficits, including Alzheimer’s disease, learning disabilities, mental retardation, and agerelated cognitive decline. Thus, developing targeted therapies for each of the many causes of this condition will be a formidable task. Therefore, in addition to the prevalent targeted Hippocampus
FIGURE 4. Early and late LTP is facilitated in mice administered XE991. (A) Basal neurotransmission was measured by paired pulses with interpulse intervals from 50 to 200 ls. The percentage of the paired-pulse facilitation in each interpulse interval in saline and XE991 injected mice is represented (black and gray symbols represent the saline and XE991 injected mice, respectively). (B) Representative paired-pulse recordings with 50–200 ls interpulse intervals [IPI] (black and gray recordings represent the saline and XE991 injected mice, respectively). (C) fEPSP slope during baseline and at different times after a single (*), three (&), or six (~) HFS (HFS: five trains of 200 Hz, 100 ls pulses at a rate of
1/s) evoked changes in synaptic efficacy of different durations in saline and XE991 injected mice. (D) Representative fEPSP recordings obtained during baseline (A), and 5, 60 and 120 min after HFS (B–D respectively) (black and gray recordings represent the saline and XE991 injected mice, respectively). (E), Summary of the changes in the fEPSP slope (mean 6 SEM) at different times after one (*), three (&), or six (~) HFS trains in vehicle- and XE991injected mice [black and gray, respectively] (n 5 6 for each group). *Statistical significance of the different groups in the same session. *P ! 0.05; **P ! 0.01.
ROLE OF KCNQ/Kv7 CHANNELS IN LEARNING AND MEMORY FACILITATION
29
inhibiting the M-current may be a strategy that generally improves the cognitive capacity of healthy mice, as well as that in mouse models of cognitive impairment. In 90’s, different studies demonstrated that linopirdine, the first M-current inhibitor available, was effective in selected memory tests (Fontana et al., 1994; Aiken et al., 1996) and ineffective in others (Flagmeyer and van der Staay, 1995). Also, linopirdine has so far failed to show significant benefits in patients with Alzheimer’s disease (Rockwood et al., 1997; Borjesson et al., 1999; Cooper and Jan, 2003), possibly because of the suboptimal pharmacokinetic properties of the drug and of the relatively low M-current-blocking potency. Here we showed that the XE991, a second generation of M-current inhibitors with greater potency relative to linopirdine (Zaczek et al., 1998) should be a useful therapeutic tool in cognitive diseases.
FIGURE 5. XE991 administration decreased the stimulation threshold for learning-induced gene expression. (A) Effect of saline or XE991 on bdnf and arc mRNA transcription in hippocampal tissue from untrained and 10 or 15 min OR trained mice. gadph mRNA served as an internal control (n 5 4 mice per group, obtained from two independent experiments). (B) Bar diagrams illustrating the densitometric quantification of bdnf and arc gene expression in each experimental group. Saline, black bars; XE991, gray bars. *Statistical significance of the difference with respect to untrained mice. **P ! 0.01.
therapy approach, there is a need to develop alternative strategies to ameliorate cognitive deficits irrespective of their specific genetic or environmental cause, approaches that might have a more general impact on cognition. Here, we demonstrate that
FIGURE 6. XE991 administration alleviated cognitive defects in mouse models of cognitive impairment. (A) Bar diagram illustrating the performance after 15 min OR training of scopolamine injected mice administered either with saline (white bars) or XE991 (gray square bars) during training, and STM or LTM sessions (n 5 8 for each group). (B) Bar diagram illustrating the performance after 15 min OR memory training of KA injected mice administered either with saline (white bars) or XE991 (gray square bars) during training, and STM or LTM sessions (n 5 8 for each group). Mice injected 21 days before training with saline served as controls (black bar). *Statistical significance of the difference with respect to training in the same group. **P ! 0.01. Hippocampus
30
FONTA´N-LOZANO ET AL.
The M-current is a voltage-gated potassium current that was first described in frog sympathetic neurons (Brown and Adams, 1980) but that has since been shown to be present in many other cell types, including neurons in the CNS (reviewed by Brown, 1988). Indeed, these channels are now known to be ubiquitous in the brain (Cooper et al., 2001). M-channels are encoded by members of the KCNQ/Kv7 gene family, and they are mainly composed of hetero-multimeric complexes of KCNQ/Kv7.2 and 3 subunits (Wang et al., 1998), with additional contributions from KCNQ/Kv7.2 homomers, and KCNQ/Kv7.5 and 3 heteromers (Shah et al., 2002; Hadley et al., 2003). We find that OR training provokes a transient and specific decrease of the KCNQ/Kv7.3 protein in hippocampus and perirhinal cortex, while the other KCNQ/Kv7 subunits remain unchanged. Recently, the KCNQ/Kv7.3 protein was described as a target of Nedd4–2 (Ekberg et al., 2007), a ubiquitin ligase that regulates the membrane density of many membrane proteins (Harvey and Kumar, 1999; Abriel and Stub, 2005). Hence, neuronal excitability, as well as learning and memory, may be regulated through ubiquitination and proteasome degradation of KCNQ/Kv7.3. Further experiments are ongoing to determine whether the proteasome pathway is involved in this process. Also, if decrease in KCNQ/Kv7.3 occurs in other learning and memory paradigms need to be elucidated. In the central nervous system, the particular properties of the M-current have led to the assumption that it serves to limit sustained neuronal excitation (Madison and Nicoll, 1984). The M-current is thought to control the frequency of action potentials and to generate a medium afterhyperpolarization (mAHP) after single or multiple action potentials (Gu et al., 2005), thereby preventing the development of a postspike after depolarization (Yue and Yaari, 2004). In addition, the suppression of the M-current shifts the firing mode of the hippocampal CA1 pyramidal cells from regular firing to burst firing (ADP, Yue and Yaari, 2004), as well as reducing intrinsic subthreshold theta resonance (Hu et al., 2002; Peters et al., 2005). Recently, it has been shown that the regulation of spike threshold in hippocampal neuron by KCNQ/Kv7 channels may be relevant to the changes in synaptic plasticity and by pairedpulse facilitation (Shah et al., 2008). Our in vivo electrophysiological experiments in free moving mice showed that M-current inhibition provokes alterations in the basal hippocampal activity. First, XE991 administration increases basal glutamatergic neurotransmission measured by a paired-pulse facilitation protocol. This effect may be mediated by presynaptic alterations leading to an increase in neurotransmitter release (Conradi, 1969), as well as changes in the postsynaptic compartment affecting the depolarization induced dendritic spike (Johnston et al., 1999). Second, M-current inhibition provoked an increase in the power spectrum of theta and low-theta frequency oscillations in the hippocampus. A more drastic effect on oscillatory cortical activity, electrical oscillation below 2Hz, has been described in transgenic mice overexpressing a human mutation in the KCNQ/Kv7.2 associated with epilepsy (Peters et al., 2005). This data suggest that the M-current is related to Hippocampus
the maintenance of theta oscillation in the hippocampus, a rhythmic activity that has been correlated with learning and memory. Indeed, the theta power and frequency, or the coherence between different brain regions, varies with performance for different cognitive tasks (Klimesch, 1999; Kahana et al., 2001), although we are far from understanding the role of the low theta rhythm in learning and memory. In humans, the role of electrical oscillations in the consolidation of new learning tasks during sleep has been studied (Plihal and Born, 1997, 1999). A positive influence on declarative memory consolidation and on a spatial learning task was identified for the first half of the night (slow-wave-sleep rich oscillations). Furthermore, the slow-wave-sleep period provokes the coupling of neocortical and hippocampal activities, an event required for the transfer of information between the hippocampus and the neocortex, as occurs during memory consolidation (Sirota et al., 2003). Our results indicate that the increase in low theta oscillation in XE991 injected mice could extend the temporal window for consolidation, possibly resulting in enhanced learning and memory. Similar results in hippocampal oscillation and learning consolidation facilitation have recently been described in adult mice submitted to an intermittent fasting diet (Fonta´n-Lozano et al., 2007). All these data suggest that low theta oscillation may be an important component for the facilitation of learning and memory processes. Learning and memory processes require long-lasting change in synaptic efficacy (Daoudal and Debanne, 2003; Gruart et al., 2006), as well as nuclear events (Abel and Lattal, 2001; Kandel, 2001). We show that inhibition of the M-current decreases the stimulation threshold for learning consolidation, LTP, and learning-induced gene expression. How the M-current inhibition may provoke all these alterations still remains unknown; however, the increase of excitability induced by Mcurrent inhibition could contribute to learning facilitation and learning-induced gene expression. Also, the shift in oscillatory rhythm to low theta frequencies could facilitate the induction and maintenance of LTP (Grover et al., 2009), facilitating the consolidation of memories (Burza´ki 1986; Pennartz et al., 2002; Fonta´n-Lozano et al., 2007). In summary, M-current inhibition may generate a permissive scenario whereby changes in hippocampal basal properties facilitate synaptic plasticity related with learning and memory. Similar general effects on learning and memory facilitation has been observed when mice were submitted to intermittent fasting or when histone deacetylase inhibitors are systemically administrated (Fonta´n-Lozano et al., 2007, 2008). The severity of cognitive disorders and the number of people affected by them (>5% of the population) makes efforts to develop adequate treatments all the more urgent (Tully et al., 2003; Melhman, 2004). Because of the limited efficacy of the current palliative treatments to decrease the cognitive symptoms associated with many diseases of the central nervous system, the development of general therapies for cognitive disorders is an emerging field of research. The M-current inhibitor XE991 facilitates such processes in healthy mice and in two mice models of cognitive impairment. Firstly, XE991 revert the amnesia pro-
ROLE OF KCNQ/Kv7 CHANNELS IN LEARNING AND MEMORY FACILITATION voked by the cholinergic antagonist scopolamine, a drug that simulates the cholinergic depletion associated with some neurodegenerative diseases, like Alzheimer’s disease. This palliative effect was previously described when linopirdine, a M-current antagonist that provokes acetylcholine release, was applied to animals with medial septum lesion (Brioni et al., 1993). Also, XE991 administration inhibits the cognitive impairment provoked by kainic acid induced neuronal death. In both cases the mechanism involved in this palliative action remain unknown, although XE991 probably causes a temporal increase in basal neuronal excitability that increases the efficacy of the damage circuit. On the basis of these results, we suggest XE991 may be a potential therapeutic agent to treat cognitive impairment associated with neurodegenerative diseases and aging.
Acknowledgments We thank Mrs. M.C. Sutil for her assistance with animal handling and Dr M.Sefton for editorial assistance. AF-L was supported by the FPU fellowship program from the Spanish Ministry of Science.
REFERENCES Abel T, Lattal KM. 2001. Molecular mechanisms of memory acquisition, consolidation and retrieval. Curr Opin Neurobiol 11:180– 187. Abriel H, Staub O. 2005. Ubiquitylation of ion channels. Physiology 20:398–407. Aiken SP, Lampe BJ, Murphy PA, Brown BS. 1995. Reduction of spike frequencyadaptation and blockade of M-current in rat CA1 pyramidal neurones bylinopirdine (DuP 996), a neurotransmitter release enhancer. Br J Pharmacol 115:1163–1168. Alberini CM. 2005. Mechanisms of memory stabilization: Are consolidation and reconsolidation similar or distinct processes? Trends Neurosci 28:51–56. Bo¨rjesson A, Karlsson T, Adolfsson R, Ro¨nnlund M, Nilsson L. 1999. Linopirdine (DUP 996): Cholinergic treatment of older adults using successive and non-successive tests. Neuropsychobiol 40:78– 85. Brioni JD, Curzon P, Buckley MJ, Arneric SP, Decker MW. 1993. Linopirdine (DuP996) facilitates the retention of avoidance training and improves performance of septal lesioned rats in the water maze. Pharmacol Biochem Behav 44:37–43. Brown D. 1988. M-currents: An update. Trends Neurosci 11:294– 299. Brown DA, Adams PR. 1980. Muscarinic suppression of a novel voltage sensitive K1 current in a vertebrate neuron. Nature 283: 673–676. Brown MW, Aggleton JP. 2001. Recognition memory: What are the roles of the perirhinal cortex and hippocampus? Nat Rev Neurosci 2:51–61. Buzsa´ki G. 1986. Hippocampal sharp waves: Their origin and significance. Brain Res 398:242–252. Buzsa´ki G. 2002. Theta oscillations in the hippocampus. Neuron 33:325–340. Cooper EC, Jan LY. 2003. M-channels: Neurological diseases, neuromodulation, and drug development. Arch Neurol 60:496–500. Cooper EC, Aldape KD, Abosch A, Barbaro NM, Berger MS, Peacock WS, Jan YN, Jan LY. 2000. Colocalization and coassembly of two
31
human brain M-type potassium channel subunits that are mutated in epilepsy. Proc Natl Acad Sci USA 97:4914–4919. Cooper EC, Harrington E, Jan YN, Jan LY. 2001. M channel KCNQ2 subunits are localized to key sites for control of neuronal network oscillations and synchronization in mouse brain. J Neurosci 21:9529–9540. Conradi S. 1969. Observations on the ultrastructure of the axon hillock and initial segment of lumbosacral motoneurons in the cat. Acta Physiol Scand 332:65–92. Costa AM, Brown BS. 1997. Inhibition of M-current in cultured rat superior cervical ganglia by linopirdine: Mechanism of action studies. Neuropharmacology 36:1747–1753. Daoudal G, Debanne D. 2003. Long-term plasticity of intrinsic excitability: Learning rules and mechanisms. Learn Mem 10:456–465. de los Santos-Arteaga M, Sierra-Domı´nguez SA, Fontanella GH, Delgado-Garcı´a JM, Carrion AM. 2003. Analgesia induced by dietary restriction is mediated by the kappa-opioid system. J Neurosci 23:11120–11126. Dere E, Huston JP, De Souza Silva MA. 2007. The pharmacology, neuroanatomy and neurogenetics of the one-trial object recognition in rodent. Neurosci Biobehav Rev 31:673–704. Dudai Y. 2004. The neurobiology of consolidations, or, how stable is the engram? Annu Rev Psychol 55:51–86. Ekberg J, Schuetz F, Boase NA, Conroy S-J, Manning J, Kumar S, Poronnik P, Adams DJ. 2007. Regulation of the voltage-gated Kchannels KCNQ2/3 and KCNQ3/5 by ubiquitination. J Biol Chem 282:12135–12142. Flagmeyer I, Van Der Staay FJ. 1995. Linopiridine (DUP 996; AVIVA): Its effects in the Morris water escape tank and on retention of an incompletely acquired bar-press response in rodents. Pharmacol Biochem Behav 51:111–117. Fonta´n-Lozano A, Sa´ez-Cassanelli JL, Inda MC, de los Santos-Arteaga M, Sierra-Domı´nguez SA, Lo´pez-Lluch G, Delgado-Garcı´a JM, Carrio´n AM. 2007. Caloric restriction increases learning consolidation and facilitates synaptic plasticity through mechanisms dependent on NR2B subunits of the NMDA receptor. J Neurosci 27:10185–10195. Fonta´n-Lozano A, Romero-Granados R, Troncoso J, Mu´nera A, Delgado-Garcı´a JM, Carrio´n AM. 2008. Histone deacetylase inhibitors improve learning consolidation in young and in KA-induced-neurodegeneration and SAMP-8-mutant mice. Mol Cell Neurosci 39:193–201. Fontana DJ, Inouye GT, Johnson RM. 1994. Linopirdine (DuP 996) improves performance in several tests of learning and memory by modulation of cholinergic neurotransmission. Pharmacol Biochem Behav 49:1075–1082. Franse´n E, Alonso AA, Dickson CT, Magistretti J, Hasselmo ME. 2004. Ionic mechanisms in the generation of subthreshold oscillations and action potential clustering in entorhinal layer II stellate neurons. Hippocampus 14:368–384. Grover LM, Kim E, Cooke JD, Holmes WR. 2009. LTP in hippocampal area CA1 is induced by burst stimulation over a broad frequency range centered around delta. Learn Mem 16:69–81. Gruart A, Mun˜oz MD Delgado-Garcı´a JM. 2006. Involvement of the CA3-CA1 synapse in the acquisition of associative learning in behaving mice. J Neurosci 26:1077–1087. Gu N, Vervaeke K, Hu H, Storm JF. 2005. Kv7/KCNQ/M and HCN/h, but not KCa2/SK channels, contribute to the somatic medium after-hyperpolarization and excitability control in CA1 hippocampal pyramidal cells. J Physiol 566:689–715. Hadley JK, Passmore GM, Tatulian L, Al-Qatari M, Ye F, Wickenden AD, Brown DA. 2003. Stoichiometry of expressed KCNQ2/ KCNQ3 potassium channels and subunit omposition of native ganglionic M channels deduced from block by etraethylammonium. J Neurosci 23:5012–5019. Halliwell JV, Adams PR. 1982. Voltage-clamp analysis of muscarinic excitation in hippocampal neurons. Brain Res 250:71–92. Hippocampus
32
FONTA´N-LOZANO ET AL.
Harvey KF, Kumar S. 1999. Nedd4-like proteins: An emerging family of ubiquitin-protein ligases implicated in diverse cellular functions. Trends Cell Biol 9:166–169. He XP, Kotloski R, Nef S, Luikart BW, Parada LF, McNamara JO. 2004. Conditional deletion of TrkB but not BDNF prevents epileptogenesis in the kindling model. Neuron 43:31–42. Hu H, Vervaeke K, Storm JF. 2002. Two forms of electrical resonance at theta frequencies, generated by M-current, h-current and persistent Na-current in rat hippocampal pyramidal cells. J Physiol 545:783–805. Inda MC, Delgado-Garcı´a JM, Carrio´n AM. 2005. Acquisition, consolidation, reconsolidation, and extinction of eyelid conditioning responses require de novo protein synthesis. J Neurosci 25:2070–2080. Jensen HS, Calløa K, Jespersena T, Jensen BS, Olesena S-P. 2005. The KCNQ5 potassium channel from mouse: A broadly expressed Mcurrent like potassium channel modulated by zinc, pH, and volume changes. Mol Brain Res 139:52–62. Jentsch TJ. 2000. Neuronal KCNQ potassium channels: Physiology and role in disease. Nat Rev Neurosci 1:21–30. Johnston D, Hoffman DA, Colbert CM, Magee JC. 1999. Regulation of backpropagating action potentials in hippocampal neurons. Curr Opin Neurobiol 9:288–292. Kandel ER. 2001. The molecular biology of memory storage: A dialogue between genes and synapses. Science 294:1030–1038. Kahana MJ, Seelig D, Madsen JR. 2001. Theta returns. Curr Opin Neurobiol 11:739–744. Klimesch W. 1999. EEG alpha and theta oscillations reflect cognitive and memory performance: A review and analysis. Brain Res Rev 29:169–195. Lal H, Kumar B, Forster J. 1988. Enhancement of learning and memory in mice by benzodiazepine antagonist. FASEB J 2:2707–2711. Lamas JA, Selyanko AA, Brown DA. 1997. Effects of a cognitionenhancer, linopirdine (DuP 996), on M-type potassium currents (IK(M)) and some other voltage- and ligand-gated membrane currents in rat sympathetic neurons. Eur J Neurosci 9:605–616. Lerche C, Scherer CR, Seebohm G, Derst C, Wei AD, Busch AE, Steinmeyer K. 2000. Molecular cloning and functional expression of KCNQ5, a potassium channel subunits that may contribute to neuronal M current diversity. J Biol Chem 275:22395–22400. Madison DV, Nicoll RA. 1984. Control of the repetitive discharge of rat CA 1 pyramidal neurones in vitro. J Physiol 354:319–331. Magee JC, Johnston D. 1997. A synaptically controlled, associative signal for Hebbian plasticity in hippocampal neurons. Science 275:209–213. Mehlman MJ. 2004. Cognition-enhancing drugs. Milbank Q 82:483– 506. Pennartz CMA, Uylings HBM, Barnes CA, McNaughton BI. 2002. Memory reactivation and consolidation during sleep: From cellular mechanisms to human performance. Prog Brain Res 138:143–166. Peters HC, Hu H, Pongs O, Storm JF, Isbrandt D. 2005. Conditional transgenic suppression of M channels in mouse brain reveals functions in neuronal excitability, resonance and behavior. Nat Neurosci 8:51–60. Plath N, Ohana O, Dammermann B, Errington ML, Schmitz D, Gross C, Mao X, Engelsberg A, Mahlke C, Welzl H, Kobalz U, Stawrakakis A, Fernandez E, Waltereit R, Bick-Sander A, Therstappen E, Cooke SF, Blanquet V, Wurst W, Salmen B, Bo¨sl MR, Lipp HP, Grant SG, Bliss TV, Wolfer DP, Kuhl D. 2006. Arc/Arg3.1 is essential for the consolidation of synaptic plasticity and memories. Neuron 52:437–444. Plihal W, Born J. 1997. Effects of early and late nocturnal sleep on declarative and procedural memory. J Cogn Neurosci 9: 534–547.
Hippocampus
Plihal W, Born J. 1999. Effects of early and late nocturnal sleep on priming and spatial memory. Psychophysiology 36:571–582. Ploski JE, Pierre VJ, Smucny J, Park K, Monsey MS, Overeem KA, Schafe GE. 2008. The activity-regulated cytoskeletal-associated protein (Arc/Arg3.1) is required for memory consolidation of pavlovian fear conditioning in the lateral amygdale. J Neurosci 28:12383–12395. Rockwood K, Beattie BL, Eastwood MR, Feldman H, Mohr E, PrysePhillips W, Gauthier S. 1997. A randomized, controlled trial of linopirdine in the treatment of Alzheimer’s disease. Can J Neurol Sci 24:140–145. Rogawski MA. 2000. KCNQ2/KCNQ3 K1 channels and the molecular pathogenesis of epilepsy: Implications for therapy. Trends Neurosci 23:393–398. Saganich MJ, Machado E, Rudy B. 2001. Differential expression of genes encoding subthreshold-operating voltage-gated K1 channels in brain. J Neurosci 21:4609–4624. Schauwecker PE, Steward O. 1997. Genetic determinants of susceptibility to excitotoxic cell death: Implications for gene targeting approaches. Proc Natl Acad Sci USA 94:4103–4108. Schnee ME, Brown BS. 1998. Selectivity of linopirdine (DuP 996), a neurotransmitter release enhancer, in blocking voltage-dependent and calcium-activated potassium currents in hippocampal neurons. J Pharmacol Exp Ther 286:709–717. Shah MM, Mistry M, Marsh SJ, Brown DA, Delmas P. 2002. Molecular correlates of the M-current in cultured rat hippocampal neurons. J Physiol 544:29–37. Shah MM, Migliore M, Valencia I, Cooper EC, Brown DA. 2008. Functional significance of axonal Kv7 channels in hippocampal pyramidal neurons. Proc Natl Acad Sci USA 105:7869–7874. Sirota A, Csicsvari J, Buhl D, Buzsa´ki G. 2003. Communication between neocortex and hippocampus during sleep in rodents. Proc Natl Acad Sci USA 100:2065–2069. Squire LR, Zola-Morgan S. 1991. The medial temporal lobe memory system. Science 253:1380–1386. Strajstrom CE, Chronopoulos A, Thurber S, Thompson JL, Holmes GL. 1993. Aged-dependent cognitive and behavioural deficits after kainate acid seizure. Epilepsia 34:420–432. Thomas MJ, Watabe AM, Moody TD, Makhinson M, O’Dell TJ. 1998. Postsynaptic complex spike bursting enables the induction of LTP by theta frequency synaptic stimulation. J Neurosci 18:7118– 7126. Tully T, Bourtchouladze R, Scott R, Tallman J. 2003. Targeting the CREB pathway for memory enhancers. Nature Rev Drug Discov 2:267–277. Wang HS, Pan Z, Shi W, Brown BS, Wymore RS, Cohen IS, Dixon JE, McKinnon D. 1998. KCNQ2 and KCNQ3 potassium channel subunits: Molecular correlates of the M-channel. Science 282:1890–1893. Yue C, Yaari Y. 2004. KCNQ/M channels control spike afterdepolarization and burst generation in hippocampal neurons. J Neurosci 24:4614–4624. Yus-Na´jera E, Mun˜oz A, Salvador N, Jensen BS, Rasmussen HB, Defelipe J, Villarroel A. 2003. Localization of KCNQ5 in the normal and epileptic human temporal neocortex and hippocampal formation. Neuroscience 120:353–364. Zaczek R, Chorvat RJ, Saye JA, Pierdomenico ME, Maciag CM, Logue AR, Fisher BN, Rominger DH, Earl RA. 1998. Two new potent neurotransmitter release enhancers, 10,10-bis(4-pyridinylmethyl)-9(10H)-anthracenone and 10,10-bis(2-fluoro-4-pyridinylmethyl)-9(10H)- anthracenone: Comparison to linopirdine. J Pharmacol Exp Ther 285:724–730.
HIPPOCAMPUS 21:33–47 (2011)
Developmental Profiling of Postnatal Dentate Gyrus Progenitors Provides Evidence For Dynamic Cell-Autonomous Regulation Jennifer A. Gilley,1,2 Cui-Ping Yang,1,2 and Steven G. Kernie1,2* ABSTRACT: The dentate gyrus of the hippocampus is one of the most prominent regions in the postnatal mammalian brain where neurogenesis continues throughout life. There is tremendous speculation regarding the potential implications of adult hippocampal neurogenesis, though it remains unclear to what extent this ability becomes attenuated during normal aging, and what genetic changes in the progenitor population ensue over time. Using defined elements of the nestin promoter, we developed a transgenic mouse that reliably labels neural stem and early progenitors with green fluorescent protein (GFP). Using a combination of immunohistochemical and flow cytometry techniques, we characterized the progenitor cells within the dentate gyrus and created a developmental profile from postnatal day 7 (P7) until 6 months of age. In addition, we demonstrate that the proliferative potential of these progenitors is controlled at least in part by cell-autonomous cues. Finally, to identify what may underlie these differences, we performed stem cell-specific microarrays on GFP-expressing sorted cells from isolated P7 and postnatal day 28 (P28) dentate gyrus. We identified several differentially expressed genes that may underlie the functional differences that we observe in neurosphere assays from sorted cells and differentiation assays at these different ages. These data suggest that neural progenitors from the dentate gyrus are differentially regulated by cell-autonomous factors that change over time. V 2009 Wiley-Liss, Inc. C
KEY WORDS: microarray
brain development; hippocampus; neural stem cell;
INTRODUCTION The subgranular zone (SGZ) in the dentate gyrus of the hippocampus is one of at least two neurogenic regions in the adult brain (Altman and Das, 1965; Luskin, 1993; Gage et al., 1995; Suhonen et al., 1996). It contains neural stem/progenitor cells that can give rise in vitro to various cell types including neurons, astrocytes and oligodendrocytes, though in vivo they are primarily precursors for dentate gyrus granular neurons (Frederiksen and McKay, 1988; Suhonen et al., 1996; Fricker et al., 1999; Seaberg and van der Kooy, 2002). During hippocampal neurogenesis progenitors self-renew and produce granular neurons that migrate into the granule cell layer of the dentate gyrus (Kempermann and Gage, 2000; Abrous et al., 2005). 1
Department of Pediatrics, University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, Texas; 2 Department of Developmental Biology, University of Texas Southwestern Medical Center, 6000 Harry Hines Blvd, Dallas, Texas Grant sponsor: NIH; Grant number: R01 NS048192. *Correspondence to: Steven G. Kernie, M.D., Departments of Pediatrics and Developmental Biology, UT Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd, Dallas, TX 75390. E-mail:
[email protected] Accepted for publication 11 September 2009 DOI 10.1002/hipo.20719 Published online 15 December 2009 in Wiley Online Library (wileyonlinelibrary.com). C 2009 V
WILEY-LISS, INC.
Several processes must occur in order for neurogenesis to take place. First, mitotic cells divide to give rise to progenitor cells that will eventually become new neurons (Cameron and McKay, 2001; Alvarez-Buylla et al., 2002). In the dentate gyrus these Type I ‘‘stem cells’’ express markers characteristic of both stem and astrocytic lineages and are able to self-renew and differentiate into various committed cell types (Cameron et al., 1998; Frederiksen and McKay, 1988; Cameron and McKay, 2001; Alvarez-Buylla et al., 2002; Edgar et al., 2002; Joels et al., 2004; Encinas et al., 2006). Type I cells are distinguished from other progenitors by expression of the astrocytic marker glial fibrillary acidic protein (GFAP) as well as GFP in the nestin transgenic animal (Miles and Kernie, 2008; Yu et al., 2008). Second, these newly formed progenitors (Type I and II) must commit and differentiate into a neuronal lineage (Cameron et al., 1998). During this process most mitotic progenitors divide and produce postmitotic cells that begin to express neuroblast-specific markers including doublecortin (DCX) and polysialic acid neural cell adhesion molecule (PSANCAM) (Seki and Arai, 1993; Francis et al., 1999; Gage, 2002; Seki, 2002; Joels et al., 2004; Kronenberg et al., 2006; Christie and Cameron, 2006; Duan et al., 2008; Peng et al., 2008). Next, these immature neurons (Type III) mature into granule cell neurons, which express neuronal nuclei (NeuN) (Mullen et al., 1992). The final process required for neurogenesis is the ability of these newly formed neurons to incorporate and function within their environment (Cameron and McKay, 2001; Goergen et al., 2002; Barkho et al., 2006). Although much is known about neurogenesis in the dentate gyrus, its postnatal development has not been as well characterized. Many studies describe characteristics of ‘‘adult’’ stem cells within the dentate gyrus, however, the age at which ‘‘adult’’ is achieved is variable and has been reported to be anywhere from four to 12 weeks postnatal (Kuhn et al., 1996; Tropepe et al., 1997; Wagner et al., 1999; Lemaire et al., 2000; Bailey et al., 2004; Enwere et al., 2004; Molofsky et al., 2006; Bastos et al., 2008; Bulloch et al., 2008; Goncalves et al., 2008; Yang et al., 2009). One purpose of this study was to developmentally profile these progenitors at different ages since the dentate gyrus itself develops almost exclusively during the postnatal period, though critical aspects of its timing beyond the first week remain unclear.
34
GILLEY ET AL.
Another important objective of this study was to identify key regulators that maintain neurogenesis in the adult brain. It is known that there are extrinsic factors that influence neurogenesis in the adult brain and participate in the neural stem cell niche (Doetsch, 2003; Riquelme et al., 2008; Tavazoie et al., 2008). These microenvironmental effectors include hormones, neurotransmitters, hypoxia, and trophic and morphogenic factors that are known to play roles in regulating neurogenesis (Cameron and Gould, 1994; Calof, 1995; Cameron et al., 1995; Ghosh and Greenberg, 1995; Muller et al., 1995; Kempermann et al., 1997; Cameron et al., 1998; Heller et al., 1996; Kernie et al., 2001; Goergen et al., 2002; Kempermann et al., 2003; Barkho et al., 2006; Kronenberg et al., 2006; Kaneko and Sawamoto, 2008). In addition, there are neural stem- and progenitor-specific factors that influence neurogenesis and include genes that regulate cell proliferation as well as ones that determine and regulate cell fate decisions. Although several factors have been shown to be involved in proliferation and cell fate, there are others that play mechanistic roles in regulating adult neurogenesis and remain unknown (Abrous et al., 2005). We hypothesized that specific genes influencing neurogenesis may change over time, and we therefore chose to characterize the postnatal dentate gyrus by creating a developmental profile of neurogenesis at various time points. Neurosphere assays reveal a functional advantage for the progenitors from younger mice, while those from older mice divide more slowly. Furthermore, differentiation assays show that P7 progenitors are more neurogenic than those from P28. We also quantified early hippocampal stem/progenitors expressing GFP in nestin-GFP transgenic mice and demonstrate a progressive decline in the percentage of GFP-expressing progenitors until two months of age. In addition we utilized stereological techniques to determine absolute numbers of different progenitor populations at relevant time points. We then identify with microarray analysis between two different progenitor populations (P7 and P28), candidate genes that may mediate the changes in neurogenic potential that we observe. These results suggest that declining stem/progenitor populations in the hippocampus are in part regulated and maintained by cell-autonomous genetic changes that occur over time.
MATERIALS AND METHODS Mice The Institutional Animal Use and Care Committee at UT Southwestern Medical Center approved all animal experiments. Animals were humanely housed and cared for at the Animal Resource Center within UT Southwestern, which is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. The transgenic nestin-eGFP mouse has been described previously and extensively characterized (Yu et al., 2008; Miles & Kernie, 2008; Koch et al., 2008; Shi et al., 2007). Briefly, a nesHippocampus
tin-rtTa-eGFP construct was used to create transgenic animals that express GFP exclusively within the neural tube and the proliferative zones of the dentate gyrus and subventricular zone (SVZ). The second intron of the nestin gene was also included as it contains the neural progenitor-specific enhancer element of the nestin promoter (Zimmerman et al., 1994).
Tissue Culture Whole hippocampus was dissected from GFP transgenic animals at the indicated time points. The hippocampus was coronally sectioned in 600 lm slices and the dentate gyrus was extracted from all slices. The digestion media consisted of activated papain: 42 ll of papain (Worthington), 27 ll of 100 mM Cystein-HCl and 6 ll of 100 mM EDTA in 425 ll of DMEM/F12. Serum-containing media consisted of 10% FBS in DMEM/F12 with 1% Penicillin/Streptavidin. Neural stem cell media included 1% N2 supplement (Gibco), B27 1:50 (Gibco), 10 lg/ll Heparin, 20 ng/ml bFGF (Sigma), and bEGF (Invitrogen) and 1% antibiotics in DMEM/F12. The semisolid media used in the neurosphere assays contains 60% neural stem cell media and 40% of 1.6% methylcellulose (Sigma). In the neurosphere assay, dentate gyri were isolated as explained above. Then they were incubated in activated papain for 20–30 min at 378. After digestion, the papain was inactivated by washing the tissue with DMEM/F12 containing 10% FBS. Then a flame-polished Pasteur pipette was used to dissociate the tissue into a single-cell suspension. This suspension was then passed through a 30 lm filter (Partec) and plated at a density of 20 cells/ll in semisolid media. The total number of cells/well was 20,000 in a 12-well plate. In neurosphere assays using fluorescent activated cell sorting (FACS) analysis, dentate gyrus cells were isolated in the same manner, however, the single-cell suspension was sorted using a MoFlo cell sorter (Dako). GFP-expressing cells were plated in the same conditions as the dentate gyrus cultures. All cultures were treated with 300 ll of semisolid medium every three to four days and analyzed 14 days after plating. To further determine whether neurosphere growth and proliferation was cell-autonomous, GFP-expressing progenitors from P7 and P28 dentate gyri were sorted and plated in 96-well plates with a final density of one cell/well. Neurospheres were cultured in neural stem cell media for 14 days and then quantified. For the differentiation assay dentate gyri were microdissected from P7 and P28 transgenic animals and plated in neural stem cell media. Once neurospheres began to form, they were collected, washed, and dissociated with acutase for 5 min at 378. Cells were then replated on coated chamber slides in differentiation media (1% serum in DMEM/F12) and allowed to differentiate for five days. The media was replaced on the third day. After five days in culture chamber slides were immunohistochemically labeled for analysis.
Immunohistochemistry Mice were perfused and their brains were removed and fixed overnight in 4% paraformaldehyde. After embedding in aga-
DEVELOPMENTAL REGULATION OF THE DENTATE GYRUS rose, brains were cut into 50 lm sections using a Leica vibratome machine and every sixth section was used for immunostaining. Sections were first washed three times in 0.3% TritonX in PBS and then blocked for 1 h at room temperature in 10% normal donkey serum (NDS) in 0.3% TritonX PBS. All fluorescent antibodies were mixed together and incubated for 2 h at room temperature in their appropriate dilutions in a total volume of 500 ll of 5% NDS in 0.3% TritonX PBS. Tissues were then washed three times with 0.3% TritonX PBS and incubated with secondary antibodies for 2 h. Slices were then washed two times with 0.3% TritonX PBS and two times with PBS. Tissue sections were placed onto glass slides, mounted with Immu-Mount and covered with plastic cover slips. Images were obtained using confocal microscopy (Zeiss, LSM 510). For the developmental profile, the following primary antibodies were used: rabbit anti-GFP 1:500 (Molecular Probes), mouse anti-NeuN 1:500 (Chemicon) and goat anti-DCX 1:200 (Santa Cruz). Staining of the SVZ and SGZ used chicken anti-GFP 1:500 (Aves Labs) and rat antiplatelet-derived growth factor receptor alpha (PDGFRa) 1:250 (BD Pharmingen) antibodies. Differentiated cells attached to the chamber slides were stained with mouse anti-neuron-specific Class III b-tubulin (TUJ-1) 1:2,000 (Covance) and rabbit anti-GFAP 1:2,000 (Invitrogen) to identify neurons and astrocytes, respectively. Cells were also stained with 40 ,6-diamidino-2-phenylindole (DAPI) 1:1,000 (FLUKA) to label the nucleus. Staining for design-based stereological counting was performed on 50-lm vibratome sections (described above). Sections were stained and incubated in primary antibodies overnight (to maximize antibody penetration) and were treated with biotinylated secondary antibodies followed by incubation with horseradish peroxidase-based Vectastain ABC Kit (Vector Laboratories). Incubation with 3,30 -diaminobenzidine (DAB) substrates was used to amplify and visualize the staining according to the manufacturer’s protocol (Vector Laboratories). Primary antibodies used include rabbit anti-GFP 1:500 (Invitrogen), goat anti-DCX 1:100, and rat anti-5-bromo-2-deoxyuridine (BrdU) 1:400. All secondary antibodies used a concentration of 1:350 and were purchased from Vector Laboratories. GltI immunostaining utilized the chicken anti-GFP antibody as well as mouse anti-GFAP 1:100 (BD Pharmingen), goat anti-DCX 1:200 (Santa Cruz), and guinea pig anti-glial high affinity glutamate transporter (GltI) 1:200 (Chemicon). All secondary antibodies used were in a final concentration of 1:200. GFP and BrdU colocalization studies used rabbit anti-GFP 1:500 and rat anti-BrdU 1:400 (Abcam) primary antibodies. Fluorescent secondary antibodies (1:200) were used to visualize the staining.
Western Blots Whole cell lysates were isolated from P14 dentate gyri for PDGFRa and GFP immunoblotting (Fig. 2L). An 8% polyacrylamide gel was used for PDGFRa blotting while a 12% gel
35
was run for GFP immunoblotting. Antibodies used include rabbit anti-GFP (Molecular Probes), rat anti-PDGFRa (BD Pharmingen), and rabbit anti-b-tubulin (Sigma). All primary antibodies were used at a concentration of 1:1,000. All secondary antibodies were purchased from Santa Cruz, raised in goat serum and used at a dilution of 1:5,000. For GltI immunoblotting lysates were extracted from P7 and P28 dentate gyri and from P14 cortex, hippocampus, dentate gyrus, and hippocampus without the dentate gyrus (Fig. 6B). Protein was loaded on a 10% polyacrylamide gel for both blots. Antibodies used include guinea pig anti-GltI (Chemicon) at 1:5,000 and rabbit anti-b-tubulin (Sigma) at 1:1000. Again, both secondary antibodies were from Santa Cruz and used in a dilution of 1:5,000. Protein concentrations were all measured using a Bradford Assay. For all gels 20 lg of protein were loaded per well. After the gels were run, protein was transferred to a PVDF membrane and blocked in 5% nonfat dry milk for one hour at room temperature. Primary and secondary antibodies were incubated for 2 h at room temperature. Membranes were washed with PBS with 1% Tween. Protein bands were visualized using a Lumi-Light Western Blotting Substrate Kit from Roche followed by development.
Flow Cytometry Dentate gyrus was microdissected from the hippocampus at the indicated time points. For quantification of GFP-expressing progenitors, fresh tissue was homogenized in DDP (20 ll DNase I, 2 ml 10x Dispase II, 10 ml 2x Papain in 20 ml F12/ DMEM) and incubated at 378C for 20 min. After incubation, tissue was triturated 100x with a pipette into a single-cell solution. Cells were spun down for 30 sec at 16,000 rpm and washed with F12/DMEM containing 10% FBS three times. Then cells were resuspended in the same media and passed through a 70-lm filter (BD Biosciences). Ten minutes before analysis propidium iodide (PI) was added to each sample to identify dead cells (1:1,000). Cell sorting was done using a MoFlo machine from Dako. GFP-expressing and GFP-negative cells were collected and the percent GFP-positive cells were calculated from the total number of collected cells. The negative control was nontransgenic hippocampus. Population purity was analyzed by sorting GFPexpressing and GFP-negative cells onto a glass slide and looking at their fluorescence using a fluorescent microscope (Data not shown). No less than six mice were used per sorting sample at each age. Experiments were performed in triplicate and error bars represent standard deviation. Significance was determined using one-way analysis of variance (ANOVA) and Bonferroni post hoc analysis for a P-value level less than 0.1% (***P < 0.001). Microarray analysis was performed using validated methods and protocols developed by Miltenyi Biotec (Auffray et al., 2007; Landgraf et al., 2007). Briefly, at least 500 P7 and P28 GFP-expressing cells were independently collected from FACsorting and lysed at 428 according to the manufacturer’s Hippocampus
36
GILLEY ET AL.
instructions (Miltenyi Biotec). Cells were shipped on dry ice to Miltenyi Biotec where the RNA was extracted and cDNA was amplified in a linear fashion using PCR. Equal cDNA amounts were spotted in quadruplicate on a Stem Cell-Specific Array consisting of 916 genes enriched in various stem cells. Sample collection and microarray analysis were completed in triplicate and candidate genes were significantly up or down regulated in all three trials. Genes were considered significantly up or down regulated if their expression levels were 1.7-fold higher or 0.58fold lower compared to the P7 control, respectively. Prior to analysis baseline expression levels were analyzed and normalized to account for initial differences in gene expression. Real-Time PCR was performed on FAC-sorted cells in order to determine relative GltI mRNA levels. Total RNA was extracted from P7 and P28 GFP-expressing cells and was reverse transcribed into cDNA with the SuperScript FirstStrand Synthesis System for RT-PCR (Invitrogen). All RealTime PCR reactions were performed in a 20 ll volume that included amplified cDNA, SYBR Green dye (Roche), 2.5 lM GAPDH (as an internal control), and GltI real-time PCR primers. The primer information is as follows: GAPDH forward: 50 -CTC AAC TAC ATG GTC TAC ATG TTC CA-30 ; GAPDH reverse: 50 -CCA TTC TCG GCC TTG ACT GT-30 ; GltI forward: 50 -GGA AGA TGG GTG AAC AGG C-30 ; GltI reverse: 50 -TTC CCA CAA ATC AAG CAG G-30 . Real-Time quantification was analyzed on the Applied Biosystems 7,500 Real-Time PCR System software.
Cell Quantification Differentiated cells were quantified based on TUJ-1 and GFAP expression. An Olympus BX50 microscope and Photometrics CoolSNAP camera were used. The software used for imaging was MetaVue by Molecular Devices. Images of DAPI, TUJ-1 and GFAP expression were captured and merged to determine colocalization. At least 200 cells were analyzed per sample and each time point was done in quadruplicate. Statistical significance was analyzed using an unpaired t-test (***P < 0.001). Stereological quantification was performed on an Olympus BX51 System Microscope with a MicroFIRE A/R camera (Optronics). The Optical Fractionator Probe within the Stereo Investigator sotftware (MBF Bioscience, MicroBrightField, Inc.) utilized an unbiased counting frame (Gundersen, 1980; West, 1993) and was used to quantify cell number. Counting was performed using a 100x oil immersion lens. At least 200 cells were counted (per animal) and the average number of counting fields examined was close to 300. The average number of sections counted was 10 while the average mounted thickness after processing was !35 lm. To reduce bias between samples, a number of measures were undertaken. All tissue was processed in the same manner (see Immunohistochemistry methods above). Furthermore, to decrease the effect of shrinkage on our tissues, we used a height sampling fraction of 30 lm to account for actual tissue thickness observed after processing. The area-sampling fraction for Hippocampus
DCX- and GFP-expressing cells was 1/8. BrdU quantification required an area-sampling fraction of one. Every sixth section (the section sampling fraction) was used to quantify cell populations within the SGZ and granular layer of the dentate gyrus. Furthermore, slides were only quantified if all sections were present and homogenously stained. The coefficient of variance for each animal quantified was always less than 15%. Confocal microscopy was used to quantify the number of BrdU and GFP double-positive cells within the SGZ and granular layer of the dentate gyrus. As described elsewhere (Miles, 2008 #87; Yu, 2008 #88), a Zeiss LSM 510 confocal microscopy utilizing Argon 488 and He 633 lasers was used to quantify double-labeled cells on a Zeiss Neofluar 40X/1.3 oil lens. Focusing through the z-axis of each cell was done to ensure only precisely colocalized signals were quantified. Slides were only analyzed if antibody penetration and signal intensity between sections was consistent. Every sixth section was stained as described and all BrdU-expressing cells within the SGZ and granular layer of the dentate gyrus were analyzed for their colocalization with GFP. Percentages were calculated based on the ratio of double-labeled cells to those only expressing BrdU. Statistics were calculated using one-way ANOVA for overall signficance followed by Bonferonni post hoc analysis to determine significance between multiple groups: ***P < 0.001, **P 5 0.001–0.01 and *P 5 0.01–0.05.
RESULTS Developmental Profile of the Postnatal Dentate Gyrus Dentate gyrus progenitors express a variety of well known markers throughout their ontogeny. We have previously characterized these subsets of progenitors using transgenic mice that we developed and have also used these mice to distinguish mature astrocytes from GFAP-expressing Type I dentate gyrus progenitors. Although both express GFAP, GFP-expressing progenitors lack expression of glutamine synthetase (GS), a mature astrocyte marker (Miles and Kernie, 2008). Furthermore, after hypoxic-ischemic and traumatic brain injury, reactive astrocytes lack expression of GFP, while GFP-expressing progenitors lack expression of GS (Miles and Kernie, 2008; Shi et al., 2007; Yu et al., 2008). We have therefore demonstrated that GFP is expressed in GFAP-expressing Type I progenitors but not in mature astrocytes. Here, we used this same transgenic to determine how other genetic characteristics change over time. We first performed immunostaining to examine the progenitor cell types present in mice at various ages. Anti-GFP antibody was utilized to identify Type I and II progenitor populations whereas anti-DCX was used for later neural progenitors (Type III) and anti-NeuN was used to identify mature neurons (Mullen et al., 1992; Limke and Rao, 2003; Joels et al., 2004). Mice at various ages were analyzed to encompass most stages of dentate gyrus development: postnatal days 7, 14, 21, 28, two
DEVELOPMENTAL REGULATION OF THE DENTATE GYRUS
37
FIGURE 1. Developmental profile of the postnatal dentate gyrus. GFP, DCX, and NeuN markers were used to represent various cell types within the dentate gyrus during the course of postnatal development. Staining with GFP identifies early Type I and II neural progenitors that have been labeled with the Nestin-eGFP transgenic mouse while DCX labels Type III later progenitors and NeuN marks mature neurons. Panels (A–D) are from P7 transgenic mice and (E–H) correspond to P28 mice.
Merged images are dentate gyri from P14, P21, two month, four month, and six-month-old animals can be seen in (I–M). An age-dependent decline in the number of GFP-expressing progenitors occurs during the course of development. However, this progenitor population appears to stabilize at two months of age. Scale bars in (A) and (I) are 50 lm. SGZ, subgranular zone; GL, granular layer; ML, molecular layer.
months, four months, and six months. Results demonstrate that the number of GFP- and DCX-expressing cells decrease as the mice age (Fig. 1). In particular we observed a steady decline in GFP-expressing cells until about two months of age at which time the population appeared to stabilize.
progenitors, then we would not expect to see PDGFRa expression in the dentate gyrus since GFP-expressing cells from the SVZ also coexpress PDGFRa (Jackson et al., 2006). Immunostaining of the SVZ and SGZ for GFP and PDGFRa was carried out to determine progenitor specificity. Merged confocal images show GFP and PDGFRa colocalization in the SVZ but not in the SGZ of the dentate gyrus (Fig. 2A–H). These results suggest that there are limited GFPexpressing SVZ neural progenitors mixed in with isolated dentate gyrus cells. Furthermore, hippocampal whole mounts were prepared and visualized using a fluorescent microscope and we demonstrate GFP-expressing cells in the SGZ of the dentate gyrus and along the CA1 region of the hippocampus (Fig. 2I). The dissected dentate gyrus, however, only consists of GFPexpressing cells from the SGZ and not from the SVZ (Fig. 2J). To further confirm these findings, whole cell lysates from P14 hippocampus and dentate gyrus were isolated and blotted for the presence of GFP and PDGFRa. As expected there was no detectable PDGFRa protein in the dentate gyrus lysates (Fig. 2L).
Dentate Gyrus Dissections Are Exclusive of GFP-Expressing Progenitors From the SVZ The hippocampus is contiguous with the subventricular zone (SVZ) and gross hippocampal dissection results in a mixed population of both dentate gyrus and SVZ progenitors (Seaberg and van der Kooy, 2002; Becq et al., 2005; Tonchev and Yamashima, 2006). Before the various progenitor populations could be quantified, we needed to ensure exclusive isolation of GFPexpressing cells from the SGZ of the dentate gyrus and not from the SVZ of the lateral ventricle. To do this, we took advantage of the differential expression patterns of plateletderived growth factor receptor alpha (PDGFRa) in the SVZ and the SGZ. If our dissection methods were exclusive of SVZ
Hippocampus
38
GILLEY ET AL.
FIGURE 2. Isolation of PDGFRa-negative neural progenitors. To distinguish between GFP-positive progenitors from the dentate gyrus and SVZ, PDGFRa expression was utilized to mark oligodendrocytes precursors in P7, P14, P21, and P28 mice. (A–D) shows a lack of GFP and PDGFRa coexpression within the SGZ. However, cells expressing both markers in (E– H) (yellow cells) are found within the SVZ of the lateral ventricle (arrows). In (I–K) transgenic whole mount tissue of the hippocampus, the dentate gyrus, and the hippocampus without the
dentate gyrus are pictured. The arrowhead in (I) highlights GFP-expressing neural progenitors from the lateral ventricle. Western blot analysis of GFP and PDGFRa protein expression in various parts of the hippocampus confirms that neural progenitors exclusively from the dentate gyrus were isolated. Scale bars in (A) and (B) are 50 lm each. ML, molecular layer; GL, granular layer; SGZ, subgranular zone; LV, lateral ventricle; H, hippocampus; DG, dentate gyrus; H-DG, hippocampus with dentate gyrus removed.
In Vitro Neurosphere Assays Suggest Cell-Autonomous Differences Between P7 and P28 Progenitors
We next wanted to test whether the proliferative differences observed were in fact cell-autonomous and not influenced by differing densities of progenitors at these ages or extracellular cues from adjacent cell types. We therefore used flow cytometry to sort GFP-expressing cells from microdissected P7 and P28 dentate gyrus and repeated the neurosphere assay while maintaining a density of 20 cells/ll. Again we observed a significant difference between neurospheres cultured from P7 and P28 GFP-expressing cells (Fig. 3D,E). Sorted GFP-expressing cells from P7 animals formed an average of 62.1 neurospheres per well that exceeded 50 lm while P28 progenitors produced 3.6 neurospheres per well (Fig. 3F). To further confirm that differences in growth and proliferation between P7 and P28 progenitors were due to cell-intrinsic factors, we performed a single-cell proliferation assay. Progenitors were sorted and plated at a density of one cell per well in a 96-well plate. After 14 days of culturing, no neurospheres of any size were derived from P28 progenitors in four different experiments. However, the P7 progenitors produced an average of 3 neurospheres per 96-well plate demonstrating that only
Due to the morphological differences between progenitor populations, we hypothesized that P7 and P28 progenitors would respond differently when cultured in vitro. To test this, cells were dissociated from P7 and P28 dentate gyrus and cultured at a density of 20 cells/ll and allowed to form neurospheres. To analyze the ability of the cultures to grow and proliferate, we counted the number of neurospheres present in each culture and measured their diameter 14 days after plating. We further assessed the number of neurospheres that were larger than 50 lm. The P7 progenitors were more proliferative and more easily formed neurospheres than the P28 progenitors (Fig. 3A,B). Furthermore, there were more neurospheres that were at least 50 lm in diameter. The P7 dentate gyrus cultures had an average of 260 neurospheres over 50 lm while the P28 cultures only saw an average of 35 neurospheres of at least 50 lm (Fig. 3C). These results suggest a functional difference between the P7 and P28 progenitor populations. Hippocampus
DEVELOPMENTAL REGULATION OF THE DENTATE GYRUS
39
FIGURE 3. Cell-autonomous factors affect progenitor growth and differentiation. To determine the growth phenotype of progenitors from P7 and P28 mice, neurosphere assays were used to quantify the number of neurospheres (over 50 lm) formed in culture. Representative pictures and quantification of neurospheres formed from the whole dentate gyrus (A–C) and GFP-positive
FAC-sorted cells (D–F) are depicted. P7-derived neurospheres seem to have a growth advantage compared to cultures from P28 progenitors. This phenotype is observed in both a cell-autonomous (D–F) and non cell-autonomous fashion (A–C). Scale bars in (A) represents 50 lm. Statistical analysis was performed using an unpaired t-test: ***P < 0.001 and **P 5 0.001–0.01.
single-cells sorted from the P7 hippocampus are capable of cell-autonomous self-renewal (not shown).
statistically much higher when compared to all other time points (P < 0.001) (Fig. 4D). Populations from older mice (two months, four months and six months) all exhibit similar percentages that were not statistically different from one another. Furthermore, there was about a 27-fold difference between the P7 and P28 progenitor populations. These quantification data support our observations from immunostaining and further demonstrate a steady decline in GFP-expressing early progenitors until about two months of age. These results suggest that adulthood senescence may occur around two months of age as indicated by the stabilized progenitor population preceding this developmental time point.
In Vitro Differentiation Suggests That P7 Progenitors Are More Neurogenic Than Those From P28 To determine the potential of progenitor cells to differentiate into neurons, primary neurospheres from P7 and P28 dentate gyrus were plated in serum for five days and then stained with 40 ,6-diamidino-2-phenylindole (DAPI), GFAP, and TUJ-1 to distinguish neurons (expressing TUJ-1) from astrocytes and undifferentiated progenitors (expressing GFAP). Progenitors from P7 mice preferentially differentiated into neurons (61.7%) while most P28 progenitors became astrocytes or remained undifferentiated (66.3%) (Fig. 4C). The percentage of GFAPexpressing cells and neurons observed were extremely significant when the two time points were compared (P < 0.001). More than 99% of cells observed expressed either GFAP or TUJ-1. This suggests that under these culture conditions, progenitors rarely differentiate into oligodendrocytes.
Quantification of GFP-Expressing Neural Progenitors Suggests Early Adulthood Senescence To further characterize the diminishing progenitor cell population, we used fluorescent activated cell sorting of the GFPexpressing cells to quantify the progenitor population at each postnatal time point. Interestingly, the P7 and P14 dentate gyrus had the highest percentage of progenitor cells and were
Design-Based Stereology of GFP-, DCX-, and 5-bromo-2-deoxyuridine (BrdU)-Expressing Cells Indicate Age-Dependent Changes in Cell Number To more closely analyze different progenitor populations within the dentate gyrus, design-based stereology and 3,30 -diaminobenzidine (DAB) immunohistochemistry was utilized to estimate the number of early progenitors (GFP-positive), late progenitors (DCX-positive), and BrdU-positive cells per dentate gyrus in P7, P28 and 2-month-old transgenic mice. The total number of GFP-expressing cells was significantly different for each age group, with the highest absolute number in P28 animals. P7 animals had an average of 27,000 positive cells per dentate gyrus, while P28 animals had nearly double this number at 47,000 positive cells and two-month-old animals had !16,000 (Fig. 5D). One-way ANOVA followed by a BonHippocampus
40
GILLEY ET AL. was performed. Confocal analysis revealed that P7 mice injected with a two-hour pulse of BrdU demonstrated colocalization between GFP and BrdU !50% of the time (Fig. 5P). This number increased in P28 animals with 64% colocalization and was lowest in two-month-old animals at 38% (Fig. 5P). This data can therefore be extrapolated to the number of GFPexpressing cells within the dentate gyrus (Fig. 5I–L). Although the total number of BrdU-expressing cells is highest at P7 (Fig. 5L), the percentage of GFP- and BrdU-labeled cells is highest at P28. The proportion of GFP-expressing early progenitors to DCX-expressing late progenitors is higher in P28 animals compared with P7 (Fig. 5D,H). High numbers of DCX-positive cells in P7 animals (Fig. 5F) may contribute to the decreased percentage of GFP- and BrdU-dual-labeled cells (Type I and II progenitors) compared with P28 animals. This suggests that the increased percentage of GFP- and BrdU-labeled cells observed at P28 (Fig. 5P) may be due to decreased numbers of DCX-positive cells at P28 and two months of age (Fig. 5H).
FIGURE 4. In vitro differentiation and quantification of GFPexpressing neural progenitors suggests early adulthood senescence. (A) and (B) illustrate representative images of differentiated cells derived from P7 and P28 progenitors, respectively. (C) Quantification of TUJ-1- and GFAP-expressing cells suggests that P7 progenitors preferentially mature into neurons while most P28 progenitors differentiate into astrocytes. The percentages of astrocytes and neurons are significant when P7 is compared with P28 (***P < 0.001) (D) The percent of GFP-expressing cells separated by flow cytometry declines over the course of development and begins to stabilize at two months of age. This suggests that adulthood senescence may occur around two months of age in mice as indicated by the stabilized progenitor population preceding this developmental time point. Statistical analysis utilized ANOVA followed by the Bonferroni correction. P7 and P14 are significantly different from all other samples but are not statistically different from one another (***P < 0.001). Error bars represent standard deviation and the scale bar (A) is 50 lm. Abbreviations: DAPI 5 40 ,6-diamidino-2-phenylindole; TUJ-1 5 neuron-specific class III b-tubulin.
feronni post hoc analysis determined that all three groups were significantly different from one another. The number of DCX-expressing cells, however, decrease over time where there were approximately 76,000 at P7, this decreased to 40,000 at P28 and 36,000 at 2 months of age (Fig. 5H). Unlike early Type I and II progenitors that express GFP the highest number of DCX-expressing cells is observed in younger (P7) mice while estimates at P28 and 2 months are not significantly different from one another. Similar to DCX staining, BrdU-positive cells were more numerous in P7 animals compared with P28 and two month olds. P7 animals had an average of 4,522 positive cells per dentate gyrus while P28 animals and two-month-old animals had 1,000 and 628 cells, respectively (Fig. 5L). Finally, to assess the percentage of GFP-expressing cells that are actually dividing in P7, P28, and two-month-old transgenic mice, dual-labeling immunofluorescence with GFP and BrdU Hippocampus
Microarray Analysis Reveals Differentially Expressed Genes Between P7 and P28 Progenitor Populations Results from the neurosphere assays suggest there are cell-autonomous factors that regulate a progenitor’s ability to proliferate and form neurospheres. Furthermore, the vast differences observed during the quantification of these progenitor populations at various developmental time points implies a role for intrinsic factors. We therefore hypothesized that differential gene expression may be involved in regulating and maintaining these hippocampal stem/progenitor cells. To identify these differences, we performed triplicate microarray analysis on P7 and P28 GFP-expressing dentate gyrus progenitors that were collected with FAC-sorting. Candidate genes were identified based on their significant up- or down-regulation (in all three trials) compared with the P7 controls. This stringent analysis revealed nine candidate genes, three down-regulated and six up-regulated in P28 progenitors compared to the P7 controls. The candidate genes and their known functions are shown in Table 1. These results suggest the presence of a few differentially expressed genes which may be involved in regulating or maintaining the neural progenitor cells at various developmental time points. The genes presented in Table 1 make for a small list, though this was generated using very stringent criteria. In fact, many more genes were identified that are also differentially regulated and these have been deposited in NCBI’s Gene Expression Omnibus (Edgar et al., 2002) and may be viewed using the following GEO Series accession number GSE15085 at the NCBI website: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc5GSE15085
GltI Expression Patterns Confirms Microarray Data The glutamate transporter GltI was found to be the most highly up-regulated gene in P28 progenitors. We validated this
DEVELOPMENTAL REGULATION OF THE DENTATE GYRUS
41
FIGURE 5. Significant age-dependent changes in progenitor cell number. To better assess different progenitor populations within the dentate gyrus, dual immunofluorescence and designbased stereological methods were utilized. Representative pictures and quantifications of DAB histology of GFP (A–D), DCX (E–H), and BrdU (I–L) are shown as well as GFP and BrdU immunofluorescence (M–P). GFP staining seems to decrease with age (A–D), although P28 animals have a significantly increased number of GFP-expressing cells compared to the other time points quantified (D). Unlike Type I and II progenitors, the number of DCXexpressing Type III progenitors remains highest in younger animals (E–H). BrdU expression (I–K) and cell number (L) are also
significantly increased in P7 animals compared to P28 and twomonth-old animals. Although GFP and BrdU immunofluorescence decrease with age (M–O), the percent of double-labeled cells is highest in P28 animals when compared with P7 and two-monthold transgenic animals (P). Significance was determined by ANOVA followed by Bonferonni post hoc analysis. P-values for GFP, DCX, BrdU, and double-labeling quantification are significant: ***P < 0.001, **P 5 0.001–0.01 and *P 5 0.01–0.05. Error bars represent standard deviation and the scale bar in (A) represents 50 lm. ML, molecular layer; GL, granular layer; SGZ, subgranular zone.
finding using a variety of methods. Relative GltI mRNA levels were detected in sorted GFP-expressing cells from P7 and P28 animals (Fig. 6A). We demonstrate that GltI mRNA levels are an average of 9.65-fold higher in the P28 sample when normalized to the P7 sample. This result is consistent with our microarray data in which GltI was up-regulated in P28 cells by a factor of 9.77 when compared with the P7 control (Table 1). To validate protein expression, Western blot analysis was carried out on whole cell lysates extracted from P7 and P28 dentate gyri. GltI protein levels are elevated in the P28 lysates compared to those from P7 (Fig. 6B). In addition, GltI expression patterns in P14 lysates derived from various parts of the brain demonstrate its relatively specificity to the dentate gyrus though it is also found in the cortex and other parts of the hippocampus (Fig. 6C).
GltI Is Expressed in Early But Not Late Progenitors Within the SGZ Since there are several progenitor subtypes within the dentate gyrus, we wanted to determine whether GltI expression was specific to the early stem/progenitor population. Therefore, the cell-specificity of GltI expression was ascertained by performing immunostaining on P7 and P28 dentate gyrus. Consistent with the microarray data, GltI protein becomes increasingly evident on P28 GFP-expressing progenitors and is barely expressed at P7 (Fig. 7). Furthermore, the majority of GFP- and GltIexpressing cells also colocalize with GFAP indicating that GltI is found in early type I progenitors (Fig. 7H). However, DCX did not colocalize with GFP- and GltI-expressing cells. These data confirm that GltI is not found on late (Type III) neural Hippocampus
42
GILLEY ET AL.
TABLE 1. Microarray Candidate Genes
Gene
Average ratio
Downregulated compared to P7 Cluster of Differentiation 47 (CD47) Chondroitin Sulfate Proteoglycan 2 (CSPG2) T-box 5 (Tbx5) Upregulated compared to P7 Apolipoprotein E (ApoE) Claudin 10 (Cldn10) Excitatory Amino Acid Transporter (Eaat2/GltI) Gap Junction Protein 4 (Gja4) Glypican 4 (Gpc4) Vitronectin (Vtn)
Known cellular functions
0.217
Cell adhesion
0.24
Cell adhesion and development Transcriptional regulator
0.49
4.35 5.75 9.77
Regulator of lipid metabolism Cell adhesion Glutamate transport
6.62
Gap junction protein
4.31
Cell proliferation and morphogenesis Cell adhesion and immune response
7.32
progenitors (Fig. 7P) and is specific to early stem/progenitor population of the P28 dentate gyrus.
DISCUSSION This study demonstrates intrinsic differences between young and old progenitors within the developing dentate gyrus and suggests that these differences may underlie their proliferative and differentiation potential. It is well known that the ability of neural progenitors in the dentate gyrus to self-renew and proliferate declines with age. Several groups have used BrdU incorporation and markers of newborn neurons to quantify
FIGURE 6. GltI mRNA and protein levels confirm microarray results. (A) Using qPCR, relative GltI mRNA levels are up-regulated in P28 GFP-expressing cells while (B) GltI protein levels are also elevated in lysates extracted from microdissected dentate gyrus. Widespread distribution of GltI protein throughout P14 Hippocampus
neurogenesis and progenitor proliferation in the dentate gyrus of rats (Kuhn et al., 1996; Heine et al., 2004; McDonald and Wojtowicz, 2005; Rao et al., 2006). Furthermore, expression patterns of various genes such as transforming growth factor beta (TGF-b), fibroblast growth factor 2 (FGF2), insulin growth factor1 (IGF1), brain-derived neurotrophic factor (BDNF) and vascular endothelial growth factor (VEGF) have been linked to age-related decreases in neurogenesis in the dentate gyrus (Hattiangady et al., 2005; Shetty et al., 2005; Buckwalter et al., 2006;). In addition, regulation of caspase activity in neurogenic regions of the brain also influences postnatal neurogenesis (Gemma et al., 2007; Tang et al., 2009). These studies, however, mainly demonstrate how the neurogenic niche changes over time and do not suggest cell-specific changes in gene expression that occur in the actual stem/progenitor population itself. Similar to embryonic neurogenesis, early postnatal neurogenesis in the dentate gyrus occurs in three stages (Nakai and Fujita, 1994; Li and Pleasure, 2007). During the development of the dentate gyrus, the primary dentate neuroepithelium located near the ventricular zone, is the site where the dentate precursor pool begins to proliferate, expand and form the first granular cells (Fujita, 1962; Fujita, 1963; Altman and Bayer, 1990; Frotscher et al., 2007; Li and Pleasure, 2007). Then the secondary dentate matrix is formed and serves as the scaffold for what is to become the granular layer of the dentate gyrus (Fujita, 1964; Altman and Bayer, 1990; Frotscher et al., 2007; Li and Pleasure, 2007). By E13.5, cells within the secondary matrix are very proliferative and migrate to the nascent dentate gyrus and by E17.5, the tertiary matrix forms within the future hilus and progenitors and granule cell populations begin to mix and migrate to the limbs of the dentate gyrus (Altman and Bayer, 1990; Li and Pleasure, 2007). Finally, the granule cell layers are condensed and the neural progenitors become aligned with the SGZ (Altman and Bayer, 1990; Li and Pleasure, 2007). The peak of granular cell neurogenesis occurs during the secondary and tertiary matrix at the end of the first postnatal week (Schlessinger et al., 1975). Our analysis begins at this time (P7), when these early developmental stages are complete and the dentate gyrus is discretely formed but is just beginning to mature into its neuronal layers.
lysates is observed in (C). C, Cortex; H, hippocampus; DG, dentate gyrus; H-DG, hippocampus with dentate gyrus removed. Statistical analysis in (A) was done using an unpaired t-test (**P 5 0.001–0.01).
DEVELOPMENTAL REGULATION OF THE DENTATE GYRUS
43
FIGURE 7. GltI colocalizes with GFAP-expressing Type I progenitors but is not found on DCX-expressing Type III neural precursors. In vivo immunofluorescence of GFP, GltI, and GFAP within the SGZ of P7 (A–D) and P28 (E–H) animals demonstrate that GltI is found on Type I neural progenitor cells. The high magnification image within (D) illustrates colocalization between GFP and GFAP-expressing cells (arrowhead) while only background levels of GltI are present. In (H) GltI levels are elevated and colocalize with both GFP and GFAP (arrow). However, not all
GFP and GltI double-positive cells express GFAP (H, arrowhead). DCX is not expressed on cells labeled with both GFP and GltI (P, arrowhead) and some GFP-expressing cells lack DCX expression altogether (L, arrowhead). Therefore, lack of colocalization between DCX and GltI in both P7 (I–L) and P28 (M–P) animals indicates GltI’s absence from later type III progenitors. Scale bar in (A) is 50 lm while scale bar within the inset picture (D) is 35 lm. SGZ, subgranular zone.
Our developmental profile (Fig. 1) broadly defines the different progenitor population present within the dentate gyrus while our quantification data more closely annotate Type I and II progenitors (Figs. 4D and 5). Consistent with our immunostaining results, the percentage of GFP-expressing cells within the dentate gyrus steadily declines over the course of the first several postnatal weeks. In addition to declining neurogenesis, progenitor cells from older animals experience age-related genetic changes as well. It is well known that the microenvironment affects the proliferative potential of a variety of organ-specific stem/progenitor cells (Fliedner et al., 1985; Ivasenko et al., 1990; Yanai et al., 1991; Jenkinson et al., 2003; Zhu et al., 2004; Boyle et al., 2007). Here, we demonstrate that declining neurogenesis observed in vivo is mimicked by our in vitro neurosphere cultures. Although the environment is likely relevant
in the SGZ during hippocampal development, our results indicate that decreased proliferation of P28 progenitors in culture might be due to intrinsic factors that are regulating their neurogenic potential. It is also important to point out that our neurosphere proliferation experiments were performed using neural progenitor cells derived exclusively from the dentate gyrus (Fig. 2). This is relevant due to the identification of several differences among various progenitor populations in vitro. For example, neurospheres derived from the SVZ contain more neurosphere-forming cells, are more proliferative and multipotent, and respond to FGF2 differently than SGZ-derived neurospheres (Becq et al., 2005). Furthermore, progenitors located throughout various regions of the SVZ and lateral ventricle display differential growth properties and suggest that the progenitor population Hippocampus
44
GILLEY ET AL.
from the SVZ is itself very heterogeneous (Golmohammadi et al., 2008). Other evidence suggests that neurospheres cultured from various regions of the brain retain a region-specific phenotype when cultured in vitro (Armando et al., 2007). We therefore needed to ensure that our neurosphere culture results were specific to progenitors derived from the SGZ of the dentate gyrus. There are two possibilities to explain the data revealed by the neurosphere assays. First, P7 GFP-expressing progenitors may be more proliferative than those from P28. Also, it might be that P7 brains contain an increased number of proliferative cells compared to P28. Although the results presented here do not definitively distinguish between these two possibilities, the underlying conclusion is the same, that the P7 dentate gyrus is more proliferative than at P28 and that this increase in proliferation is due at least in part to cell-autonomous effects. Results from the differentiation assay suggest that P7 and P28 progenitors preferentially mature into neurons and astrocytes, respectively (Fig. 4C). This suggests that early type I and type II progenitors from P28 transgenic mice possess less potential to become neurons when compared to P7. These agedependent changes in the ability to differentiate further suggest that these GFP-expressing progenitors are differentially and dynamically regulated. Furthermore, this finding might have several implications in regards to stem cell/progenitor therapies. For example, these results might help explain why certain populations of stem/progenitor cells are insufficient to stimulate neurogenesis in studies aimed at attenuating neurodegeneration. In addition to determining the percentage of GFP-expressing cells in the dentate gyrus (Fig. 4D), unbiased stereological quantifications were performed. Although the percentage of GFP-positive cells is highest in P7 brains (Fig. 4D), the actual cell number of GFP-positive cells within the SGZ and the granular layer of the dentate gyrus is highest at P28 (Fig. 5D). This can be explained by the morphological differences observed between the ages tested (P7, P28 and 2 months). The hippocampus in general is much smaller in early postnatal brains compared with those that are more mature. In addition, the development of the dentate gyrus is not complete until after the first week of life. As a result, the pattern of GFP-expressing cells is not structurally defined within the SGZ and granular layer at P7 (Fig. 5A). Therefore, any GFP-positive cells within the molecular layer of the dentate gyrus would not contribute to the overall cell number quantified with stereology. However, in P28 and 2-month old brains, the GFP-positive progenitors are restricted to the SGZ and granular layers (Fig. 5B,C). Using RNA extracted from the uniform GFP-expressing population of progenitors in vivo, we have identified several potential regulators that affect progenitor population in a cellautonomous manner. The most differentially expressed candidate we identified is GltI, which has well-defined roles in mature astrocytes (Tanaka, 2007). In the synaptic cleft, GltI acts to uptake excess glutamate between nerves in order to prevent neurotoxicity (Rothstein et al., 1994; Rothstein et al., 1996; Tanaka et al., 1997; Anderson and Swanson, 2000; Tanaka, 2007; Liang et al., 2008). However, little is known Hippocampus
about its role, if any, in regulating postnatal neurogenesis. Our expression data confirms reports of GltI within dentate gyrus progenitors in the SGZ (Fig. 6) (Bar-Peled et al., 1997). The different expression patterns of GltI on P7 and P28 progenitors suggest that it may somehow dynamically regulate early progenitors within the postnatal dentate gyrus. Because glutamate is the main excitatory amino acid neurotransmitter in the brain, glutamate receptors and transporters, including GltI, play key roles in maintaining brain homeostasis throughout development and life. However, functional studies are necessary to determine if and how GltI is a neurogenic regulator. We did identify other genes in our microarray study that are compelling targets and might explain some of these age-related genetic changes that we observe. For example wingless (Wnt1), which has been linked with stem cell maintenance in certain tissues (Sato et al., 2004; Kleber et al., 2005; Lowry et al., 2005), is down-regulated in P28 progenitors compared with P7 (ratio of 0.46). In addition, genes involved in positive regulation of cell proliferation such as Sox4 (Sinner et al., 2007) are also significantly down-regulated in P28 progenitors (ratio of 0.41). More importantly however, is the fact that VEGF, which has previously been associated with age-dependent decreases in neurogenesis (Hattiangady et al., 2005; Shetty et al., 2005; Buckwalter et al., 2006), is also variably expressed within our GFP-expressing progenitors. Although these variably expressed genes did not fulfill all our criteria for in depth analysis due to variance observed between replicates, they provide for validation of our techniques used as well as targets for further study. The data we present here provides evidence that the neurogenic niche is still undergoing dynamic transformational changes until two months of age when the neural progenitor population begins to stabilize. This observation therefore provides insight to studies that begin their analysis prior to P60 when the ‘‘adult’’ phenotype is not clearly established and the stem/progenitor population still retains characteristics of an earlier and more developmentally immature dentate gyrus. Age-dependent changes in progenitor proliferation, differentiation, and function have implications in stem cell therapy and our understanding of progenitor cell biology. Although it is not known what causes decreased neurogenesis in aging and diseased brains, cell-based therapies including ex vivo transplantation of stem cells and stimulation of endogenous progenitor proliferation are receiving much attention (Limke and Rao, 2002; Limke and Rao, 2003). However, the regulation of neurogenesis and its effect on the aging or diseased brain is both complex and not well understood. Understanding the mechanisms and relevance underlying these age-dependent changes is necessary before we will be able to utilize the therapeutic potential of neural stem/progenitors cells.
Acknowledgments The authors thank Gui Zhang and Ben Orr for their technical assistance. Special thanks also goes to Angela Mobley from the Flow Cytometry Core at UT Southwestern and Dr. Lawrence Weir of Miltenyi Biotec. Finally, we would like to thank
DEVELOPMENTAL REGULATION OF THE DENTATE GYRUS Dr. Neal Melvin experiments.
for
his
assistance
with
stereological
REFERENCES Abrous DN, Koehl M, Le Moal M. 2005. Adult neurogenesis: From precursors to network and physiology. Physiol Rev 85:523– 569. Altman J, Bayer SA. 1990. Prolonged sojourn of developing pyramidal cells in the intermediate zone of the hippocampus and their settling in the stratum pyramidale. J Comp Neurol 301:343–364. Altman J, Das GD. 1965. Autoradiographic and histological evidence of postnatal hippocampal neurogenesis in rats. J Comp Neurol 124:319–335. Alvarez-Buylla A, Seri B, Doetsch F. 2002. Identification of neural stem cells in the adult vertebrate brain. Brain Res Bull 57:751–758. Anderson CM, Swanson RA. 2000. Astrocyte glutamate transport: Review of properties, regulation, and physiological functions. Glia 32:1–14. Armando S, Lebrun A, Hugnot JP, Ripoll C, Saunier M, Simonneau L. 2007. Neurosphere-derived neural cells show region-specific behaviour in vitro. Neuroreport 18:1539–1542. Auffray C, Fogg D, Garfa M, Elain G, Join-Lambert O, Kayal S, Sarnacki S, Cumano A, Lauvau G, Geissmann F. 2007. Monitoring of blood vessels and tissues by a population of monocytes with patrolling behavior. Science 317:666–670. Bailey KJ, Maslov AY, Pruitt SC. 2004. Accumulation of mutations and somatic selection in aging neural stem/progenitor cells. Aging Cell 3:391–397. Bar-Peled O, Ben-Hur H, Biegon A, Groner Y, Dewhurst S, Furuta A, Rothstein JD. 1997. Distribution of glutamate transporter subtypes during human brain development. J Neurochem 69:2571–2580. Barkho BZ, Song H, Aimone JB, Smrt RD, Kuwabara T, Nakashima K, Gage FH, Zhao X. 2006. Identification of astrocyte-expressed factors that modulate neural stem/progenitor cell differentiation. Stem Cells Dev 15:407–421. Bastos GN, Moriya T, Inui F, Katura T, Nakahata N. 2008. Involvement of cyclooxygenase-2 in lipopolysaccharide-induced impairment of the newborn cell survival in the adult mouse dentate gyrus. Neuroscience 155:454–462. Becq H, Jorquera I, Ben-Ari Y, Weiss S, Represa A. 2005. Differential properties of dentate gyrus and CA1 neural precursors. J Neurobiol 62:243–261. Boyle M, Wong C, Rocha M, Jones DL. 2007. Decline in self-renewal factors contributes to aging of the stem cell niche in the Drosophila testis. Cell Stem Cell 1:470–478. Buckwalter MS, Yamane M, Coleman BS, Ormerod BK, Chin JT, Palmer T, Wyss-Coray T. 2006. Chronically increased transforming growth factor-beta1 strongly inhibits hippocampal neurogenesis in aged mice. Am J Pathol 169:154–164. Bulloch K, Miller MM, Gal-Toth J, Milner TA, Gottfried-Blackmore A, Waters EM, Kaunzner UW, Liu K, Lindquist R, Nussenzweig MC, Steinman RM, McEwen BS. 2008. CD11c/EYFP transgene illuminates a discrete network of dendritic cells within the embryonic, neonatal, adult, and injured mouse brain. J Comp Neurol 508:687–710. Calof AL. 1995. Intrinsic and extrinsic factors regulating vertebrate neurogenesis. Curr Opin Neurobiol 5:19–27. Cameron HA, Gould E. 1994. Adult neurogenesis is regulated by adrenal steroids in the dentate gyrus. Neuroscience 61:203–209. Cameron HA, Hazel TG, McKay RD. 1998. Regulation of neurogenesis by growth factors and neurotransmitters. J Neurobiol 36:287– 306.
45
Cameron HA, McEwen BS, Gould E. 1995. Regulation of adult neurogenesis by excitatory input and NMDA receptor activation in the dentate gyrus. J Neurosci 15:4687–4692. Cameron HA, McKay RD. 2001. Adult neurogenesis produces a large pool of new granule cells in the dentate gyrus. J Comp Neurol 435:406–417. Christie BR, Cameron HA. 2006. Neurogenesis in the adult hippocampus. Hippocampus 16:199–207. Doetsch F. 2003. A niche for adult neural stem cells. Curr Opin Genet Dev 13:543–550. Duan X, Kang E, Liu CY, Ming GL, Song H. 2008. Development of neural stem cell in the adult brain. Curr Opin Neurobiol 18:108– 115. Edgar R, Domrachev M, Lash AE. 2002. Gene expression omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res 30:207–210. Encinas JM, Vaahtokari A, Enikolopov G. 2006. Fluoxetine targets early progenitor cells in the adult brain. Proc Natl Acad Sci USA 103:8233–8238. Enwere E, Shingo T, Gregg C, Fujikawa H, Ohta S, Weiss S. 2004. Aging results in reduced epidermal growth factor receptor signaling, diminished olfactory neurogenesis, and deficits in fine olfactory discrimination. J Neurosci 24:8354–8365. Fliedner TM, Calvo W, Klinnert V, Nothdurft W, Prummer O, Raghavachar A. 1985. Bone marrow structure and its possible significance for hematopoietic cell renewal. Ann N Y Acad Sci 459:73–84. Francis F, Koulakoff A, Boucher D, Chafey P, Schaar B, Vinet MC, Friocourt G, McDonnell N, Reiner O, Kahn A, McConnell SK, Berwald, Netter Y, Denoulet P, Chelly J. 1999. Doublecortin is a developmentally regulated, microtubule-associated protein expressed in migrating and differentiating neurons. Neuron 23:247–256. Frederiksen K, McKay RD. 1988. Proliferation and differentiation of rat neuroepithelial precursor cells in vivo. J Neurosci 8:1144–1151. Fricker RA, Carpenter MK, Winkler C, Greco C, Gates MA, Bjorklund A. 1999. Site-specific migration and neuronal differentiation of human neural progenitor cells after transplantation in the adult rat brain. J Neurosci 19:5990–6005. Frotscher M, Zhao S, Forster E. 2007. Development of cell and fiber layers in the dentate gyrus. Prog Brain Res 163:133–142. Fujita S. 1962. Kinetics of cellular proliferation. Exp Cell Res 28:52– 60. Fujita S. 1963. The matrix cell and cytogenesis in the developing central nervous system. J Comp Neurol 120:37–42. Fujita S. 1964. Analysis of neuron differentiation in the central nervous system by tritiated thymidine autoradiography. J Comp Neurol 122:311–327. Gage FH. 2002. Neurogenesis in the adult brain. J Neurosci 22:612– 613. Gage FH, Coates PW, Palmer TD, Kuhn HG, Fisher LJ, Suhonen JO, Peterson DA, Suhr ST, Ray J. 1995. Survival and differentiation of adult neuronal progenitor cells transplanted to the adult brain. Proc Natl Acad Sci USA 92:11879–11883. Gemma C, Bachstetter AD, Cole MJ, Fister M, Hudson C, Bickford PC. 2007. Blockade of caspase-1 increases neurogenesis in the aged hippocampus. Eur J Neurosci 26:2795–2803. Ghosh A, Greenberg ME. 1995. Distinct roles for bFGF and NT-3 in the regulation of cortical neurogenesis. Neuron 15:89–103. Goergen EM, Bagay LA, Rehm K, Benton JL, Beltz BS. 2002. Circadian control of neurogenesis. J Neurobiol 53:90–95. Golmohammadi MG, Blackmore DG, Large B, Azari H, Esfandiary E, Paxinos G, Franklin KB, Reynolds BA, Rietze RL. 2008. Comparative analysis of the frequency and distribution of stem and progenitor cells in the adult mouse brain. Stem Cells 26:979–987. Goncalves MB, Suetterlin P, Yip P, Molina-Holgado F, Walker DJ, Oudin MJ, Zentar MP, Pollard S, Yanez-Munoz RJ, Williams G, Hippocampus
46
GILLEY ET AL.
Walsh FS, Pangalos MN, Doherty P. 2008. A diacylglycerol lipaseCB2 cannabinoid pathway regulates adult subventricular zone neurogenesis in an age-dependent manner. Mol Cell Neurosci 38:526– 536. Gundersen HJ. 1980. Stereology—Or how figures for spatial shape and content are obtained by observation of structures in sections. Microsci Acta 83:409–426. Guo Y, Shi D, Li W, Liang C, Wang H, Ye Z, Hu L, Wang HQ, Li Y. 2008. Proliferation and neurogenesis of neural stem cells enhanced by cerebral microvascular endothelial cells. Microsurgery 28:54–60. Hattiangady B, Rao MS, Shetty GA, Shetty AK. 2005. Brain-derived neurotrophic factor, phosphorylated cyclic AMP response element binding protein and neuropeptide Y decline as early as middle age in the dentate gyrus and CA1 and CA3 subfields of the hippocampus. Exp Neurol 195:353–371. Heine VM, Maslam S, Joels M, Lucassen PJ. 2004. Prominent decline of newborn cell proliferation, differentiation, and apoptosis in the aging dentate gyrus, in absence of an age-related hypothalamus- pituitary-adrenal axis activation. Neurobiol Aging 25:361– 375. Heller S, Ernsberger U, Rohrer H. 1996. Extrinsic signals in the developing nervous system: The role of neurokines during neurogenesis. Perspect Dev Neurobiol 4:19–34. Ivasenko IN, Klestova OV, Arkad’eva GE, Almazov VA. 1990. Role of stromal microenvironment in the regulation of bone marrow hemopoiesis after curantyl administration. Biull Eksp Biol Med 110:98–100. Jackson EL, Garcia-Verdugo JM, Gil-Perotin S, Roy M, QuinonesHinojosa A, VandenBerg S, Alvarez-Buylla A. 2006. PDGFR alpha-positive B cells are neural stem cells in the adult SVZ that form glioma-like growths in response to increased PDGF signaling. Neuron 51:187–199. Jenkinson WE, Jenkinson EJ, Anderson G. 2003. Differential requirement for mesenchyme in the proliferation and maturation of thymic epithelial progenitors. J Exp Med 198:325–332. Joels M, Karst H, Alfarez D, Heine VM, Qin Y, van Riel E, Verkuyl M, Lucassen PJ, Krugers HJ. 2004. Effects of chronic stress on structure and cell function in rat hippocampus and hypothalamus. Stress 7:221–231. Kaneko N, Sawamoto K. 2008. Adult neurogenesis in physiological and pathological conditions. Brain Nerve 60:319–328. Kempermann G, Gage FH. 2000. Neurogenesis in the adult hippocampus. Novartis Found Symp 231:220–235; discussion 235– 241:302–306. Kempermann G, Gast D, Kronenberg G, Yamaguchi M, Gage FH. 2003. Early determination and long-term persistence of adult-generated new neurons in the hippocampus of mice. Development 130:391–399. Kempermann G, Kuhn HG, Gage FH. 1997. More hippocampal neurons in adult mice living in an enriched environment. Nature 386:493–495. Kernie SG, Erwin TM, Parada LF. 2001. Brain remodeling due to neuronal and astrocytic proliferation after controlled cortical injury in mice. J Neurosci Res 66:317–326. Kleber M, Lee HY, Wurdak H, Buchstaller J, Riccomagno MM, Ittner LM, Suter U, Epstein DJ, Sommer L. 2005. Neural crest stem cell maintenance by combinatorial Wnt and BMP signaling. J Cell Biol 169:309–320. Koch JD, Miles DK, Gilley JA, Yang CP, Kernie SG. 2008. Brief exposure to hyperoxia depletes the glial progenitor pool and impairs functional recovery after hypoxic-ischemic brain injury. J Cereb Blood Flow Metab 28:1294–1306. Kronenberg G, Bick-Sander A, Bunk E, Wolf C, Ehninger D, Kempermann G. 2006. Physical exercise prevents age-related decline in precursor cell activity in the mouse dentate gyrus. Neurobiol Aging 27:1505–1513. Hippocampus
Kuhn HG, Dickinson-Anson H, Gage FH. 1996. Neurogenesis in the dentate gyrus of the adult rat: Age-related decrease of neuronal progenitor proliferation. J Neurosci 16:2027–2033. Landgraf P, Rusu M, Sheridan R, Sewer A, Iovino N, Aravin A, Pfeffer S, Rice A, Kamphorst AO, Landthaler M, Lin C, Socci ND, Hermida L, Fulci V, Chiaretti S, Foa R, Schliwka J, Fuchs U, Novosel A, Muller RU, Schermer B, Bissels U, Inman J, Phan Q, Chien M, Weir DB, Choksi R, De Vita G, Frezzetti D, Trompeter HI, Hornung V, Teng G, Hartmann G, Palkovits M, Di Lauro R, Wernet P, Macino G, Rogler CE, Nagle JW, Ju J, Papavasiliou FN, Benzing T, Lichter P, Tam W, Brownstein MJ, Bosio A, Borkhardt A, Russo JJ, Sander C, Zavolan M, Tuschi T. 2007. A mammalian microRNA expression atlas based on small RNA library sequencing. Cell 129:1401–1414. Lemaire V, Koehl M, Le Moal M, Abrous DN. 2000. Prenatal stress produces learning deficits associated with an inhibition of neurogenesis in the hippocampus. Proc Natl Acad Sci USA 97:11032– 11037. Li G, Pleasure SJ. 2007. Genetic regulation of dentate gyrus morphogenesis. Prog Brain Res 163:143–152. Liang J, Takeuchi H, Doi Y, Kawanokuchi J, Sonobe Y, Jin S, Yawata I, Li H, Yasuoka S, Mizuno T, Suzumura A. 2008. Excitatory amino acid transporter expression by astrocytes is neuroprotective against microglial excitotoxicity. Brain Res 1210:11–19. Limke TL, Rao MS. 2002. Neural stem cells in aging and disease. J Cell Mol Med 6:475–496. Limke TL, Rao MS. 2003. Neural stem cell therapy in the aging brain: Pitfalls and possibilities. J Hematother Stem Cell Res 12: 615–623. Lowry WE, Blanpain C, Nowak JA, Guasch G, Lewis L, Fuchs E. 2005. Defining the impact of beta-catenin/Tcf transactivation on epithelial stem cells. Genes Dev 19:1596–1611. Luskin MB. 1993. Restricted proliferation and migration of postnatally generated neurons derived from the forebrain subventricular zone. Neuron 11:173–189. McDonald HY, Wojtowicz JM. 2005. Dynamics of neurogenesis in the dentate gyrus of adult rats. Neurosci Lett 385:70–75. Miles DK, Kernie SG. 2008. Hypoxic-ischemic brain injury activates early hippocampal stem/progenitor cells to replace vulnerable neuroblasts. Hippocampus 18:793–806. Molofsky AV, Slutsky SG, Joseph NM, He S, Pardal R, Krishnamurthy J, Sharpless NE, Morrison SJ. 2006. Increasing p16INK4a expression decreases forebrain progenitors and neurogenesis during ageing. Nature 443:448–452. Mullen RJ, Buck CR, Smith AM. 1992. NeuN, a neuronal specific nuclear protein in vertebrates. Development 116:201–211. Muller HW, Junghans U, Kappler J. 1995. Astroglial neurotrophic and neurite-promoting factors. Pharmacol Ther 65:1–18. Nakai J, Fujita S. 1994. Early events in the histo- and cytogenesis of the vertebrate CNS. Int J Dev Biol 38:175–183. Peng Q, Masuda N, Jiang M, Li Q, Zhao M, Ross CA, Duan W. 2008. The antidepressant sertraline improves the phenotype, promotes neurogenesis and increases BDNF levels in the R6/2 Huntington’s disease mouse model. Exp Neurol 210:154–163. Rao MS, Hattiangady B, Shetty AK. 2006. The window and mechanisms of major age-related decline in the production of new neurons within the dentate gyrus of the hippocampus. Aging Cell 5:545–558. Riquelme PA, Drapeau E, Doetsch F. 2008. Brain micro-ecologies: neural stem cell niches in the adult mammalian brain. Philos Trans R Soc Lond B Biol Sci 363:123–137. Rothstein JD, Martin L, Levey AI, Dykes-Hoberg M, Jin L, Wu D, Nash N, Kuncl RW. 1994. Localization of neuronal and glial glutamate transporters. Neuron 13:713–725. Rothstein JD, Dykes-Hoberg M, Pardo CA, Bristol LA, Jin L, Kuncl RW, Kanai Y, Hediger MA, Wang Y, Schielke JP, Welty DF. 1996. Knockout of glutamate transporters reveals a major role for astro-
DEVELOPMENTAL REGULATION OF THE DENTATE GYRUS glial transport in excitotoxicity and clearance of glutamate. Neuron 16:675–686. Sato N, Meijer L, Skaltsounis L, Greengard P, Brivanlou AH. 2004. Maintenance of pluripotency in human and mouse embryonic stem cells through activation of Wnt signaling by a pharmacological GSK-3-specific inhibitor. Nat Med 10:55–63. Schlessinger AR, Cowan WM, Gottlieb DI. 1975. An autoradiographic study of the time of origin and the pattern of granule cell migration in the dentate gyrus of the rat. J Comp Neurol 159: 149–175. Seaberg RM, van der Kooy D. 2002. Adult rodent neurogenic regions: The ventricular subependyma contains neural stem cells, but the dentate gyrus contains restricted progenitors. J Neurosci 22:1784– 1793. Seki T. 2002. Hippocampal adult neurogenesis occurs in a microenvironment provided by PSA-NCAM-expressing immature neurons. J Neurosci Res 69:772–783. Seki T, Arai Y. 1993. Distribution and possible roles of the highly polysialylated neural cell adhesion molecule (NCAM-H) in the developing and adult central nervous system. Neurosci Res 17:265–290. Shetty AK, Hattiangady B, Shetty GA. 2005. Stem/progenitor cell proliferation factors FGF-2, IGF-1, and VEGF exhibit early decline during the course of aging in the hippocampus: Role of astrocytes Glia 51:173–186. Shi J, Miles DK, Orr BA, Massa SM, Kernie SG. 2007. Injury-induced neurogenesis in Bax-deficient mice: Evidence for regulation by voltage-gated potassium channels. Eur J Neurosci 25:3499–3512. Sinner D, Kordich JJ, Spence JR, Opoka R, Rankin S, Lin SC, Jonatan D, Zorn AM, Wells JM. 2007. Sox17 and Sox4 differentially regulate beta-catenin/T-cell factor activity and proliferation of colon carcinoma cells. Mol Cell Biol 27:7802–7815. Suhonen JO, Peterson DA, Ray J, Gage FH. 1996. Differentiation of adult hippocampus-derived progenitors into olfactory neurons in vivo. Nature 383:624–627. Tanaka K. 2007. Role of glutamate transporters in astrocytes. Brain Nerve 59:677–688. Tanaka K, Watase K, Manabe T, Yamada K, Watanabe M, Takahashi K, Iwama H, Nishikawa T, Ichihara N, Kikuchi T, Okuyama S, Kawashima N, Hori S, Takimoto M, Wada K. 1997. Epilepsy and
47
exacerbation of brain injury in mice lacking the glutamate transporter GLT-1. Science 276:1699–1702. Tang H, Wang Y, Xie L, Mao X, Won SJ, Galvan V, Jin K. 2009. Effect of neural precursor proliferation level on neurogenesis in rat brain during aging and after focal ischemia. Neurobiol Aging 30:299–308. Tavazoie M, Van der Veken L, Silva-Vargas V, Louissaint M, Colonna L, Zaidi B, Garcia-Verdugo JM, Doetsch F. 2008. A specialized vascular niche for adult neural stem cells. Cell Stem Cell 3:279–288. Tonchev AB, Yamashima T. 2006. Differential neurogenic potential of progenitor cells in dentate gyrus and CA1 sector of the postischemic adult monkey hippocampus. Exp Neurol 198:101–113. Tropepe V, Craig CG, Morshead CM, van der Kooy D. 1997. Transforming growth factor-alpha null and senescent mice show decreased neural progenitor cell proliferation in the forebrain subependyma. J Neurosci 17:7850–7859. Wagner JP, Black IB, DiCicco-Bloom E. 1999. Stimulation of neonatal and adult brain neurogenesis by subcutaneous injection of basic fibroblast growth factor. J Neurosci 19:6006–6016. West MJ. 1993. New stereological methods for counting neurons. Neurobiol Aging 14:275–285. Yanai N, Satoh T, Obinata M. 1991. Endothelial cells create a hematopoietic inductive microenvironment preferential to erythropoiesis in the mouse spleen. Cell Struct Funct 16:87–93. Yang Y, Takeuchi K, Rodenas-Ruano A, Takayasu Y, Bennett MV, Zukin RS. 2009. Developmental switch in requirement for PKA RIIbeta in NMDA-receptor-dependent synaptic plasticity at Schaffer collateral to CA1 pyramidal cell synapses. Neuropharmacology 56:56–65. Yu TS, Zhang G, Liebl DJ, Kernie SG. 2008. Traumatic brain injuryinduced hippocampal neurogenesis requires activation of early nestin-expressing progenitors. J Neurosci 28:12901–12912. Zhu JH, Tao QM, Chen JZ, Wang XX, Shang YP. 2004. Statins contribute to enhancement of the number and the function of endothelial progenitor cells from peripheral blood. Sheng Li Xue Bao 56:357–364. Zimmerman L, Parr B, Lendahl U, Cunningham M, McKay R, Gavin B, Mann J, Vassileva G, McMahon A. 1994. Independent regulatory elements in the nestin gene direct transgene expression to neural stem cells or muscle precursors. Neuron 12:11–24.
Hippocampus
HIPPOCAMPUS 21:48–55 (2011)
Interhippocampal Transfer in Passive Avoidance Task Modifies Metabolic Activity in Limbic Structures J.M. Cimadevilla,1 M. Me´ndez-Lo´pez,2 M. Me´ndez,2 and J.L. Arias2*
ABSTRACT: The hippocampus is probably the most studied brain structure regarding memory. Each brain hemisphere contains one hippocampus, and subjects with unilateral hippocampal lesions can perform adequately in several behavioral tasks. This property allows studying how both hippocampi interact. In this work, we show that the information acquired in a passive avoidance task with one hippocampus can be retrieved and used by the brain when the hippocampal side involved in the acquisition is blocked with TTX. The pre-exposition to the context is decisive. By combining behavioral tasks and cytochrome oxidase histochemistry we demonstrated that several brain structures, including the hippocampus, amygdale and other related regions, change their activity under the above-mentioned treatments. V 2009 Wiley-Liss, Inc. C
KEY WORDS: box; rat
memory; cytochrome oxidase; hippocampus; shuttle
INTRODUCTION The hippocampus is one of the most intriguing brain structures. Its participation in memory processes attracted neuroscientists’ attention (Scoville and Milner, 1957; Morris et al., 1982; Astur et al., 2002). It has been reported that permanent and reversible hippocampal lesions disrupt rodent’s behavior in spatial and non spatial memory tasks (Morris et al., 1982; Fenton and Bures, 1993; Lorenzini et al., 1996; Moser and Moser, 1998; Riedel et al. 1999; Cimadevilla et al., 2005; Cimadevilla et al., 2007). Unilateral interventions have been used on different occasions to study the hippocampal role in memory processes (Moser et al., 1998; Moser and Moser, 1998; Cimadevilla et al., 2005) or even to understand the communication between both hemispheres (Fenton et al., 1995). Hence, by temporally blocking the hippocampus it was demonstrated that information can be transferred between hippocampal sides. The experimental approach to this phenomenon is particularly important in the attempt to understand how memories are organized in the hippocampal system and the way of interaction between both hippocampi. 1
Department of Neuroscience, University of Almerı´a, Carretera de Sacramento s/n, 04120, Almerı´a, Spain; 2 Department of Psychology, University of Oviedo, Plaza de Feijoo s/n, 33003, Oviedo, Spain Grant sponsor: MEC, MICINN; Grant numbers: SEJ2005-05067/PSIC, PSI2008-02106, SEJ2007-63506. *Correspondence to: Jorge Luis Arias Perez, Department of Psychology, University of Oviedo, Plaza de Feijoo s/n, 33003, Oviedo, Spain. E-mail:
[email protected] Accepted for publication 11 September 2009 DOI 10.1002/hipo.20720 Published online 17 November 2009 in Wiley Online Library (wileyonlinelibrary.com). C 2009 V
WILEY-LISS, INC.
Moreover, a behavioral paradigm in which the training requirements are restricted in time can provide a good tool for studying brain changes during very well defined time periods. These features make the passive avoidance task a good candidate for studying how memories can be transferred between hemispheres, and which brain structures can be involved in this process. Nevertheless, transfer of information between both hippocampi in a passive avoidance task has so far not been reported, although it is well known that unilateral hippocampal inactivation impairs different stages of memory formation in the above mentioned task (Lorenzini et al. 1996; Cimadevilla et al. 2007). Conversely, our knowledge about the brain mechanisms involved during this process of transfer is very limited. There are several techniques than can be used for determining how the brain changes its activity in response to drugs or other interventions. Cytochrome oxidase (CO) histochemistry is used in neuroscience as a marker of neural functional activity. CO is a mitochondrial enzyme that catalyzes the transfer of electrons to oxygen generating ATP via the coupled process of oxidative phosphorylation (Wong-Riley, 1989). Sustained changes in synaptic activity are associated with altered ion pump activity, energy demand, and, ultimately, CO activity. CO activity is regulated by and closely correlated with neuronal functional activity (Wong-Riley, 1979). In this study we showed that hippocampal transfer can be obtained in a passive avoidance paradigm and unilateral hippocampal treatments with TTX caused COX changes not only in the hippocampus but also in other related structures.
MATERIALS AND METHODS Subjects We used 47 three-month-old male Wistar rats (300–350 g) from the breeding colony of the University of Oviedo (Spain). Subjects were housed in pairs on a 12-h light-dark cycle with food and water available ad libitum and a constant temperature of 20– 218C. Subjects were randomly distributed into the experimental groups and were tested in the light
INTERHIPPOCAMPAL TRANSFER AND METABOLISM
49
phase. The study was carried out in accordance with the European Communities Council Directive of 24 November 1986 (86/609/EEC) for the care and use of laboratory animals.
Surgery Subjects were anesthetized with ketamine (50 mg/kg i.p.) and xilazynum, (20 mg/kg i.m.). By using a Kopf (Tujunga, CA) sterotaxic frame they were implanted bilaterally with stainless-steel cannulas (25 ga) aimed at the dorsal hippocampus (3.5 mm behind Bregma, 2.5 mm lateral and 1 mm below dura according to Paxinos and Watson’s Atlas (2005). Cannulas and anchor screws were encased in dental acrylic.
Intrahippocampal Injections Tetrodotoxin (TTX) was used to block hippocampal activity (5 ng in one microliter of saline solution pH 7.4). TTX is a highly selective voltage-gated sodium channel blocker. The tissue inactivation lasts approximately 3 h (Zhuravin and Bures, 1991). The animal was gently restrained by hand and an internal cannula (32 ga) was inserted into the guide cannula so that it protruded 2 mm into the hippocampal target. The injection solution was delivered during 90 s using a Hamilton syringe connected to the internal cannula with a short piece of polyethylene tubing. Control injections consisted in introducing one microliter of a saline solution.
Apparatus The shuttle box (Ugo Basile 7552) was used to train the animals following a passive avoidance paradigm. This box was comprised of two Plexiglas compartments: an illuminated compartment (23 3 22 3 22 cm) lit by a 24V5W lamp, and a dark compartment (23 3 22 3 22 cm), connected via a sliding door (7 3 3.5 3 7 cm). The grid floor of the apparatus consisted of stainless-steel bars 0.3 cm in diameter at 1 cm intervals, connected to a shock scrambler (Controller 7551). The front panel of this generator displays the function of the latency time, door and shock indicators and the control door opening delay (10 s), duration and intensity of shock (3 s, 0.8 mA) and cut-off time (5 min).
Behavioral Training Training consisted of four phases: during habituation, rats were placed in the lit compartment and after 10 s the guillotine door was raised. The latency to cross to the dark compartment was measured. The rat was then returned to the vivarium. Animals that did not enter the dark chamber in the first five minutes were excluded from the task (n 5 1). One hour later, the acquisition phase began (second phase). The rat was placed in the illuminated chamber and the trial began after 10 s. The door was raised and the latency to cross the chamber was measured again. Immediately after, the door was closed and a three second unavoidable electric shock (0.8 mA) was delivered. The animal was kept in the dark chamber for 60 s and returned to the vivarium. Twenty-four hours later subjects were placed into
FIGURE 1. Schematic representation of the different phases included in the behavioral protocol and treatments received by each experimental group. The star indicates the injection side. L/R 5 left-right hippocampal sides.
the lit chamber for one min (third phase, re-exposition). The door was maintained closed. All the animals were taken again to the vivarium. Retrieval was tested 30 min later (fourth phase). The procedure was exactly the same as for the acquisition phase except that there was no shock. Three groups completed the behavioral protocol and received the following treatments (Fig. 1). Group One (n 5 9), called ‘‘transfer’’ received TTX into the right hippocampus 20 min before acquisition and TTX into the left hippocampus 15 min after the reexposition. So both hippocampi are active during the re-exposition to the context. Group Two (n 5 9), called ‘‘blocked-transfer control’’ received TTX into the right hippocampus 20 min before acquisition and TTX in the left hippocampus 20 min before the re-exposition to the context. According to this, one hippocampal side, the same one that was active during acquisition, was inactive during re-exposition to the context. Group Three (n 5 8), called ‘‘lateralized control’’ received TTX into the right hippocampus 20 min before acquisition and TTX into the right hippocampus 15 min after re-exposition to the context. The same hippocampus was blocked during acquisition and retrieval. This group was conceived as a control of transfer group, while studying CO activity in the next part of the experiment.
Histology Rats that completed the behavioral training were killed with an overdose of thiopental (100 mg/kg) and perfused transcardially with saline followed by formalin (10%) for 20 min. Brains were embedded in paraffin and 30 mm histological slices were extracted, stained with cresyl violet and the position occupied by the cannula was verified to correspond to the dorHippocampus
50
CIMADEVILLA ET AL.
sal hippocampus. Two animals were discarded. After the histology, the groups that completed the behavioral training were composed by 24 subjects: Transfer 5 8; blocked-transfer control 5 8; lateralized control 5 8.
CO Histochemistry Three additional groups of rats were used for CO estimation. The same procedure was followed but the behavioral protocol finished with the re-exposition to the environment (3rd phase) followed by inactivation. The idea was to capture the metabolic activity related to the transfer phenomenon. For CO estimation we used 20 subjects (transfer 5 5; blockedtransfer control 5 7; lateralized control 5 8). Ninety minutes following the reactivation phase in the apparatus or after the transfer and lateralized control groups received TTX, the animals were decapitated and changes in the metabolic activity of selected brain regions were analyzed using quantitative CO histochemistry, following the method described by Gonzalez-Lima and Cada (1994). Briefly, 30 lmthick brain sections were obtained using a cryostat microtome (Microm Heidelberg, Germany) and were incubated in 0.1 M phosphate buffer with 10% w/v sucrose and 0.5% v/v glutaraldehyde, pH 7.6. After this, the sections received baths of 0.1 M phosphate buffer with 10% w/v sucrose. Then 0.05 M Tris buffer, pH 7.6, with 275 mg/L cobalt chloride, 10% w/v sucrose, and 0.5% v/v dimethylsulfoxide. Sections were then incubated in a solution of 0.06 g cytochrome c (Sigma, St. Louis, MO), 0.016 g catalase, 40 g sucrose, 2 ml dimethyl-sulfoxide and 0.4 g diaminobenzidine tetrahydrochloride in 800 ml of 0.1 M phosphate buffer, at 378C for 1 h. The reaction was stopped by fixing the tissue in buffered formalin with 10% w/v sucrose and 4% v/v formalin. Finally the slides were dehydrated in ethanol and coverlipped with Entellan (Merck, Germany). To quantify enzymatic activity and to control staining variability across different baths of staining, sets of tissue homogenate standards obtained from Wistar rat brain were included with each bath. These standards were cut at different thicknesses (10, 30, 50, and 70 lm). Quantification of CO staining intensity was performed by densitometric analysis using a computer-controlled image analysis workstation (MCID, InterFocus Imaging Ltd., Linton, England). The relative optical density (OD) readings were obtained from the centromedial and basolateral amygdaloid nuclei, anteroventral, anteromedial and anterodorsal thalamic nuclei, medial and lateral mammillary nuclei, supramammillary nucleus, hippocampal areas (CA1 and CA3) and dentate gyrus. In each section, four nonoverlapping readings were taken bilaterally using a square-shaped sampling window that was adjusted for each region size. A total of 12 readings in each brain region (except for supramammillary and medial mammillary nuclei) were taken bilaterally per subject. These twelve measurements were averaged to obtain one mean per region for each subject. OD values were then converted to CO activity units, determined by the enzymatic activity of the brain standards measHippocampus
FIGURE 2. Latency of crossing to the dark chamber during acquisition (gray) and retrieval (black). Note that, although there were no differences between groups in crossing latencies during acquisition, blocked-transfer control displayed shorter crossing latencies than transfer group and lateralized control during retrieval (P < 0.001). Mean 1 standard error of mean (SEM).
ured spectrophotometrically. Measurements were performed by an investigator blind to the groups.
RESULTS Behavioral Study The three experimental groups did not differ in time crossing to the dark chamber during the habituation phase, as showed by a one-way analysis of variance (ANOVA) (F(2,21) 5 1.28, P 5 0.299). The same behavior was also evident during the acquisition phase. All groups were equivalent in crossing to the dark compartment at this phase (F(2,21) 5 2.44; P 5 0,12). However, after they experienced the shock, the mean crossing latency differed, as showed by a one way ANOVA (F(2,21) 5 16,77; P < 0.001). Post hoc Tukey test revealed that the blocked transfer group entered the dark chamber sooner than the other groups (P < 0.001) (see Fig. 2). A two way ANOVA (groups x trial—acquisition vs. retrieval), with repeated measures in the last factor, was applied to the crossing latencies in order to compare the different testing phases. The analysis discloses statistical main differences between groups (F(2,21) 5 16.9; P < 0.001), trial (F(1,21) 5 160.4; P < 0.001) and interaction (F(2,21) 5 7.44; P 5 0.004). Post hoc Tukey test showed that crossing latencies to the dark compartment during retention increased regarding acquisition (P < 0.001), and the blocked-transfer group showed shorter latencies than transfer and lateralized control groups (P < 0.001).
CO Estimation A one-way analysis of variance (ANOVA) was performed to compare latencies to enter the dark chamber in habituation and acquisition phases.
INTERHIPPOCAMPAL TRANSFER AND METABOLISM Analysis of the data showed that there was no difference between groups in the tendency to cross to the dark chamber during habituation phase (F(2,17) 5 1.834, P 5 0.19). Also, groups showed similar latencies during acquisition phase (F(2,17) 5 0.658, P 5 0.531). A two-way ANOVA was performed on brain CO activity data with transfer, blocked-transfer and lateralized control groups as between-subject factor and the right and left hemispheres as within-subject factor for each bilateral brain region measured. CA1 CO activity was different between groups (F(2,34) 5 3.899, P 5 0.03). Post hoc analysis revealed higher metabolic activity in the lateralized control group in comparison with transfer group (P 5 0.026). A group effect was found in CA3 (F(2,34) 5 3.501, P 5 0.041). Tukey test revealed a higher CO level in the lateralized control group in comparison with transfer group (P 5 0.033) and this difference was more evident within the right hemisphere (P 5 0.041). No differences between groups were found in DG CO activity (F(2,34) 5 0.423, P 5 0.658) and no side effect (see Fig. 3). Measures taken from anteroventral thalamic nuclei revealed a group effect (F(2,34) 5 8.667, P < 0.001). Tukey test revealed that transfer group showed lower CO activity than blockedtransfer and lateralized control groups (P < 0.001 and P 5 0.01, respectively). This pattern of difference was present in the left hemisphere (P < 0.05). Regarding the right side, transfer group showed lower activity than blocked-transfer group (P 5 0.028). CO activity was also different between groups in anteromedial thalamic nuclei (F(2,34) 5 12.371, P < 0.001). Transfer group showed lower metabolic activity than the other groups (P < 0.001 and P 5 0.002, respectively) and this pattern of differences was found in both left and right brain sides (P < 0.05). Moreover, CO activity of anterodorsal thalamic nuclei revealed significant group effect (F(2,34) 5 4.968, P 5 0.013). Tukey test showed that transfer group had lower activity than blocked-transfer group (P 5 0.013) and lateralized control group (P 5 0.044). A decreased CO level in transfer group compared with blocked-transfer group in the right side- (P 5 0.035) was also found (see Fig. 3). A one-way ANOVA of CO activity in medial mammillary nucleus showed no significant differences between groups (F(2,17) 5 2.93, P 5 0.081). The same analysis revealed differences in the supramammillary nucleus (F(2,17) 5 5.353, P 5 0.016). Transfer group showed lower CO activity than the other groups (P 5 0.046 and P 5 0.016, for blocked-transfer and lateralized control, respectively). Analysis of lateral mammillary nuclei CO activity showed differences between groups (F(2,34) 5 5.01, P 5 0.012). Post hoc Tukey test showed that transfer group had lower activity than blocked-transfer and lateralized control groups (P 5 0.038 and P 5 0.013, respectively) (see Fig. 3). Finally, regarding the amygdaloid nuclei, a two-way ANOVA applied to CO values of centromedial amygdaloid nuclei showed a significant effect of factor group (F(2,34) 5 7.694,
51
P 5 0.002), with blocked-transfer group showing higher CO activity than transfer and lateralized control groups (P 5 0.005 and P 5 0.006, respectively). Differences were found in the right centromedial amygdaloid nucleus where blocked transfer group showed higher CO activity than transfer and lateralized control groups (P < 0.05) but not in the left hemisphere. Analysis of basolateral amygdaloid nuclei revealed significant group effect (F(2,34) 5 4.782, P 5 0.015). Blocked-transfer group presented higher CO activity than lateralized control group only in the right side (P 5 0.022) (see Fig. 3).
DISCUSSION Reversible inactivation methods were used during the last decades to explore the neural basis of behavior (Ivanova and Bures, 1990; Fenton and Bures, 1993; Riedel et al., 1999). Unilateral hippocampal inactivation allows learning about the rules that govern memory processes. In this study using TTX as a blocker, it was demonstrated that with a nontrained hippocampus it is possible to retrieve the information previously acquired with the contralateral hippocampus, which was active during the acquisition phase. The phenomenon was described a few years ago in the Morris water maze (Fenton and Bures, 1994; Fenton et al., 1995). Since it is assumed that the information crosses from one hippocampus to the other, this process is called transfer. However, for the transfer to occur it requires a re-exposition to the context of training with both hippocampi active, since it is presumed that this experience will help the sharing of information between both hippocampal sides. In this study we showed that this phenomenon can also be reproduced in a passive avoidance task. After acquisition, subjects were introduced in the lit chamber of the shuttle-box for 60 s before retrieval (re-exposition phase). Transfer group and blocked-transfer group differed in their experience during the re-exposition phase. Hence, blocked-transfer group received TTX in the trained hippocampus before re-exposition, blocking a possible crossing of information between both hippocampi. On the contrary, transfer group received TTX after re-exposition, and according to this, both hippocampal sides were active at that phase. Results support a clear difference between both groups with transfer group mastering the task whereas blockedtransfer group was impaired. What happens during those 60 s of re-exposition is not clear. Once again in the training context, the brain will probably recover those memory traces associated with the previous experience. On transfer group, both hippocampi could share memories, since both sides are active at the same time. We speculate that this activation will be enough to support retrieval even when the trained hippocampus is inactivated. On the contrary, on blocked-transfer group only one hippocampus will retrieve memories during re-exposition to the context, and since the trained hippocampus is inactive at that phase, subjects can not retrieve memories accurately. Hippocampus
52
CIMADEVILLA ET AL.
FIGURE 3. Regional cytochrome c oxidase activity (units 5 lmol cytochrome c oxidized/(min g) wet tissue at 238C) in different limbic system regions after re-exposition to the lit chamber of the shuttle box. (a) CO activity in the dorsal hippocampus showed a general reduction in transfer group compared with lateralized control group in CA1 and CA3 subfields (#P < 0.05). This difference was also found in the right CA3 (*P < 0.05). In contrast, groups showed similar CO activity in the dentate gyrus. (b) CO activity in the anterior thalamic nuclei showed a general reduction in transfer group compared with blocked-transfer and lateralized control groups in all the nuclei studied (#P < 0.05). Regarding each side, transfer group showed lower CO activity than blockedtransfer group in the right anterodorsal and anteroventral thalamic nuclei (*P < 0.05). Transfer group also showed lower CO activity than the rest of groups in the left anteroventral nucleus (*P < 0.05). Similarly, transfer group presented a reduced CO activity
Hippocampus
compared with the other groups in both right and left anteromedial thalamic nuclei (*P < 0.05). (c) In the mammillary region, groups showed similar CO activity in the medial mammillary nucleus. On the contrary, CO activity of the lateral mammillary and supramammillary nuclei was lower in transfer group than in the rest of groups (#P < 0.05). (d) CO activity in the amygdaloid nuclei showed a significant increase in blocked-transfer group compared to transfer group and lateralized control group in the centromedial nuclei (#P < 0.05) and compared with lateralized control group in the basolateral nuclei (#P 5 0.018). A significant increase in the CO activity was observed in the blocked-transfer group as compared with the rest in the right centromedial amygdaloid nucleus (*P < 0.05). A similar phenomenon was noted in the right basolateral nucleus, which shows a significant difference with respect to lateralized control group (*P 5 0.022). Mean 1 SEM.
INTERHIPPOCAMPAL TRANSFER AND METABOLISM
Hippocampal Inactivation Effects Are Influenced by Several Variables It is important to point out that unilateral hippocampal treatments caused different effects according to several variables like task demands, cognitive processes blocked during the memory formation, age or gender. Hence, unilateral hippocampal inactivation blocked memory formation in an active place avoidance task (Cimadevilla et al., 2001), but their effects in the Morris water maze and passive avoidance tasks seem to be less pronounced (Fenton and Bures, 1993; Moser et al., 1998). Note that the difficulty as well as the amount of information required for a correct solution change from task to task. In this respect, subjects must find an unmarked place in the active place avoidance, by using uniquely distal cues, with no informative or even confusing intramaze cues. Intramaze and extramaze cues are available in the MWM; subjects can also develop different motor searching patterns. With all these strategies, subjects increase their chances of reaching the goal. Task demands are reduced in the PA task, where the experimental subjects must recognize a context with the consequences associated to it. In addition, retrieval is more susceptible to interference than acquisition under unilateral hippocampal intervention (Moser and Moser, 1998; Cimadevilla et al., 2005). Moser and Moser (1998) reported that the amount of hippocampal tissue required for retrieval is higher that that needed for acquisition. Therefore, subjects can master the MWM with one hippocampus (Fenton and Bures, 1993), but the same inactivation seriously impairs retrieval (Cimadevilla et al., 2005). With respect to consolidation processes, unilateral hippocampal inactivation blocked this phase in the active place avoidance and MWM. In both tasks, subjects displayed an impaired performance regarding controls. However, the unilateral hippocampal blockade did not affect consolidation processes in a PA task (Lorenzini et al. 1996; Cimadevilla et al., 2007). Finally, aged subjects and females seem to be more prone than males to memory disturbances under hippocampal inactivation (Poe et al., 2000; Cimadevilla and Arias, 2008). Hence, partial hippocampal inactivation disturbs spatial memory in aged subjects but not in young rats (Poe et al., 2000). Also, females under unilateral hippocampal inactivation showed inaccurate performance regarding males in the MWM not only during retrieval but also during reversal training, when the platform changes to a new position (Cimadevilla and Arias, 2008).
Transfer and Brain Activity What is happening in the brain during this process of transfer is not clear. The metabolism assessment can provide information about the energy consumption in the brain associated to that phenomenon. By using CO histochemistry, we tried to understand what happened during the transfer process. This technique is sensitive to brain function changes linked to learning processes. As reported before, CO histochemistry revealed brain changes in subjects that experienced classical conditioning
53
(Conejo et al., 2005) as well as in animals that performed a working memory version of the Morris water maze task (Conejo et al., 2004). In our study, CO activity levels were compared under the three different training conditions (transfer group, blockedtransfer group, and lateralized control group) and considering several brain structures directly or indirectly related with both the hippocampus and an appropriate performance in the task. Our analysis disclosed that several brain regions experienced metabolic changes: hippocampus, amygdale, anterior thalamic nucleus, and supramammillary area, whereas other brain regions like the mammillary bodies, and prelimbic and infralimbic areas displayed no change in their activity level (data not shown). CO reflected two different patterns of activity: Amygdale increased the level of activity in blocked-transfer group, which received TTX into the left hippocampus 15 min before re-exposure to the context. On the contrary, CA3, suprammamillary and anterior thalamic nuclei (ATN) reduced CO activity in transfer group, which received TTX into the left hippocampus 15 min after re-exposure to the context. A possible interpretation of this global analysis could indicate that CA3, the suprammamillary nucleus and the anterodorsal and anteroventral thalamic nuclei are related to the transfer process, since they are differentially activated in the transfer group. There is no change of activity in dentate gyrus and CA1, but there is in the CA3 region. Transfer group reduced CO activity in the right hippocampus in comparison to lateralized control group. It is important to consider at this point that CA3 sends the great majority of commissural projections of the Ammon’s Horn (Swanson et al., 1978), and according to this, if the transfer phenomenon affected the hippocampal areas, CA3 is a good candidate to participate in this process due to it’s central role in interhippocampal connections. This hypoactivity of CA3 could reflect the lack of necessity of this area for engaging new information, since the transfer of information would provide access to the previously formed memories. On the other hand, the suprammamillary nucleus is a part of the hypothalamic region that sends a large and direct projection to the hippocampal system (Vertes, 1992), playing an important role in the control of the rhythmic firing of hippocampal cells (theta rhythm). Yasoshima et al. (2005) showed that recall of inhibitory avoidance induced greater Fos-LI in the supramammillary nucleus which has close connection with the hippocampus. In our study, transfer group showed a reduced CO activity in comparison to blocked-transfer and lateralized control groups in the suprammamillary nucleus. This reduction was also seen in the CA3, as mentioned above, which agrees with the large amount of connections between both structures. Moreover, the anterior thalamic nuclei (ATN) are important components of memory circuits. They receive direct inputs from the subiculum and presubiculum (Meibach and Siegel, 1977) and there are also indirect routes linking the hippocampus with the anterior thalamic nuclei, like the Papez circuit, of which the projections via the mammillary bodies are probably the best known. Yasoshima et al. (2007) suggested that cells in Hippocampus
54
CIMADEVILLA ET AL.
the anterodorsal nucleus (AD) in the ATN play a role in the recall of somatically-based aversive learning. The AD might be one of the relevant structures for inhibitory avoidance memory formation and retrieval because the hippocampal-ATN circuitry plays crucial roles in spatial, contextual and episodic-like memory (Aggleton and Brown, 1999). Our experiments demonstrated that transfer group reduced its level of activity regarding the other groups in ATN. Since no differences were found in the mammillary bodies, we can assume that the subcortical brain circuit, which is formed by the projections from the hippocampal system to the hypothalamus (mammillary bodies), does not play an important role in this process. On the other hand, the amygdaloid complex suffers several metabolic changes in the right hemisphere. It is interesting to note that blocked-transfer group increased the level of activity in the intact hemisphere (right side) regarding transfer group. Note that in lateralized control group the right hemisphere was inactive, although no differences appeared between the right and left sides in this group (data not shown). A possible explanation of amygdale activation in blocked-transfer group could indicate that the subject cannot recognize the context, due to hippocampal inactivation, but they probably remember the consequences associated to the previous experience, which can activate the amygdaloidal nuclei. It is well known that the amygdale participates in passive avoidance tasks, since the emotional component is very important for an adequate performance. The increasing of activation in various nuclei of the amygdaloid complex can reflect their involvement in aversively motivated learning, which is indicated by the necessity of this structure for the full expression of an inhibitory performance. In fact, postacquisition TTX inactivation of both amygdalae is followed by severe impairment of passive avoidance retention in rats (Bucherelli et al., 1992). In conclusion, in this study we demonstrated that the transfer of information can be reproduced in this simple learning model and that CO activity is sensitive enough to disclose the involvement of different regions of the brain in learning and memory processes. These procedures can provide valuable information about the organization of memories in the hippocampal system and related structures.
Acknowledgments We thank Dr. Nobel Perdu Honeyman for help with English.
REFERENCES Aggleton JP, Brown MW. 1999. Episodic memory, amnesia, and the hippocampal-anterior thalamic axis. Behav Brain Sci 22:425–489. Astur RS, Taylor LB, Mamelak AN, Philpott L, Sutherland RJ. 2002. Humans with hippocampus damage display severe spatial memory impairments in a virtual Morris water task. Behav Brain Res 132:77–84. Hippocampus
Bucherelli C, Tassoni G, Bures J. 1992. Time-dependent disruption of passive avoidance acquisition by post-training intra-amygdala injection of tetrodotoxin in rats. Neuroscience Lett 140:231–234. Cimadevilla JM, Wesierska M, Fenton AA, Bures J. 2001. Inactivating one hippocampus impairs avoidance of a stable room-defined place during dissociation of arena cues from room cues by rotation of the arena. PNAS 98:3531–3536. Cimadevilla JM, Miranda R, Lopez L, Arias JL. 2005. Partial unilateral inactivation of the dorsal hippocampus impairs spatial memory in the MWM. Cogn Brain Res 25:741–746. Cimadevilla JM, Me´ndez M, Me´ndez-Lo´pez M, Arias JL. 2007. Unilateral hippocampal blockade reveals that one hippocampus is sufficient for learning a passive avoidance task. J Neurosci Res 85: 1138–1142. Cimadevilla JM, Arias JL. 2008. Different vulnerability in female’s spatial behavior after unilateral hippocampal inactivation. Neuroscience Lett 439:89–93. Conejo NM, Gonza´lez-Pardo H, Vallejo G, Arias JL. 2004. Involvement of the mammillary bodies in spatial working memory revealed by cytochrome oxidase activity. Brain Res 1011:107–114. Conejo NM, Lo´pez M, Cantora R, Gonza´lez-Pardo H, Lo´pez L, Begega A, Vallejo G, Arias JL. 2005. Brain metabolism after extended training in a fear conditioning task. Psicothema 17:563– 568. Fenton AA, Bures J. 1993. Place navigation in rats with unilateral tetrodotoxin inactivation of the dorsal hippocampus: Place but not procedural learning can be lateralized to one hippocampus. Behav Neurosci 107:552–564. Fenton AA, Bures J. 1994. Interhippocampal transfer of place navigation monocularly acquired by rats during unilateral functional ablation of the dorsal hippocampus and visual cortex with lidocaine. Neuroscience 58:481–491. Fenton AA, Arolfo MP, Nerad L, Bures J. 1995. Interhippocampal synthesis of lateralized place navigation engrams. Hippocampus 5:16–24. Gonza´lez-Lima F, Cada A. 1994. Cytochrome oxidase activity in the auditory system of the mouse: A qualitative and quantitative histochemical study. Neuroscience 63:559–578. Ivanova SF, Bures J. 1990. Conditioned taste aversion is disrupted by prolonged retrograde effect of intracerebral injection of tetrodotoxin in rats. Behav Neurosci 104:948–954. Lorenzini CA, Baldi E, Bucherelli C, Sacchetti B, Tassoni G. 1996. Role of dorsal hippocampus in acquisition, consolidation and retrieval of rat’s passive avoidance response: A tetrodotoxin functional inactivation study. Brain Res 730:32–39. Meibach RC, Siegel A. 1977. Thalamic projections of the hippocampal formation: Evidence for an alternate pathway involving the internal capsule. Brain Res 134:1–12. Morris RGM, Garrud P, Rawlins JNP, O’Keefe J. 1982. Place navigation impaired in rats with hippocampal lesions. Nature 297:681– 683. Moser EI, Krobert KA, Moser MB, Morris RGM. 1998. Impaired spatial learning after saturation of long-term potentiation. Science 281:2038–2042. Moser MB, Moser EI. 1998. Distributed encoding and retrieval of spatial memory in the hippocampus. J Neurosci 18:7535–7542. Paxinos G, Watson Ch. 2005. The Rat Brain in Stereotaxic Coordinates—The New Coronal Set, 5th ed. London: Elsevier Academic Press. Poe GR, Teed RG, Insel N, White R, McNaughton BL, Barnes CA. 2000. Partial hippocampal inactivation: Effects on spatial memory performance in aged and young rats. Behav Neurosci 114:940– 949. Riedel G, Micheau J, Lam AG, Roloff EL, Martin SJ, Bridge H, de Hoz L, Poeschel B, McCulloch J, Morris RGM. 1999. Reversible neural inactivation reveals hippocampal participation in several memory processes. Nature Neurosci 2:898–905.
INTERHIPPOCAMPAL TRANSFER AND METABOLISM Scoville W, Milner B. 1957. Loss of recent memory after bilateral hippocampal lesions. J Neurol Neurosurg Psychiatry 20:11–21. Swanson LW, Wyss JM, Cowan WM. 1978. An autoradiographic study of the organization of intrahippocampal association pathways in the rat. J Comp Neurol 181:681–716. Vertes RP. 1992. PHA-L analysis of projections from the suprammamillary nucleus in the rat. J Comp Neurol 326:595– 622. Wong-Riley M. 1979. Changes in the visual system of monocularly sutured or enucleated cats demonstrable with cytochrome oxidase histochemistry. Brain Res 171:11–28.
55
Wong-Riley M. 1989. Cytochrome oxidase: An endogenous metabolic marker for neuronal activity. Trends Neurosci 12:94–101. Yasoshima Y, Scott TR, Yamamoto T. 2005. Involvement of the supramammillary nucleus in aversive conditioning. Behav Neurosci 119:1290–1297. Yasoshima Y, Scott TR, Yamamoto T. 2007. Differential activation of anterior and midline thalamic nuclei following retrieval of aversively motivated learning tasks. Neuroscience 146:922–930. Zhuravin IA, Bures J. 1991. Extent of the tetrodotoxin induced blockade examined by pupillary paralysis elicited by intracerebral injection of the drug. Exp Brain Res 83:687–690.
Hippocampus
HIPPOCAMPUS 21:56–71 (2011)
Causal Evidence for the Involvement of the Neural Cell Adhesion Molecule, NCAM, in Chronic Stress-Induced Cognitive Impairments Reto Bisaz,1 Melitta Schachner,2,3,4 and Carmen Sandi1* ABSTRACT: In rodents, chronic stress induces long-lasting structural and functional alterations in the hippocampus, as well as learning and memory impairments. The neural cell adhesion molecule (NCAM) was previously hypothesized to be a key molecule in mediating the effects of stress due to its role in neuronal remodeling and since chronic stress diminishes hippocampal NCAM expression in rats. However, since most of the evidence for these effects is correlative or circumstantial, we tested the performance of conditional NCAM-deficient mice in the water maze task to obtain causal evidence for the role of NCAM. We first validated that exposure to chronic unpredictable stress decreased hippocampal NCAM expression in C57BL/6 wild-type mice, inducing deficits in reversal learning and mild deficits in spatial learning. Similar deficits in water maze performance were found in conditional NCAMdeficient mice that could not be attributed to increased anxiety or enhanced corticosterone responses. Importantly, the performance of both the conditional NCAM-deficient mice and chronically stressed wild-type mice in the water maze was improved by post-training injection of the NCAM mimetic peptide, FGLs. Thus, these findings support the functional involvement of NCAM in chronic stress-induced alterations and highlight this molecule as a potential target to treat stressrelated cognitive disturbances. V 2009 Wiley-Liss, Inc. C
KEY WORDS: memory; mice
chronic stress; NCAM; PSA-NCAM; spatial learning;
INTRODUCTION Many studies over recent decades have shown that exposure to excessive or long-lasting stress exerts deleterious effects on brain function and cognition, inducing and/or exacerbating neuropsychiatric conditions such as depression or bipolar disorders (Mazure et al., 1995; Heim and Nemeroff, 1999; Wiedenmayer, 2004; de Kloet et al., 2005; McEwen, 1
Brain Mind Institute, Ecole Polytechnique Federale de Lausanne, 1015 Lausanne, Switzerland; 2 Zentrum fu¨r Molekulare Neurobiologie, University of Hamburg, 20246 Hamburg, Germany; 3 Keck Center for Collaborative Neuroscience, Rutgers University, Piscataway, New Jersey; 4 Department of Cell Biology and Neuroscience, Rutgers University, Piscataway, New Jersey Additional Supporting Information may be found in the online version of this article. Grant sponsor: European Union; Grant numbers: FP6-2003-LIFESCIHEALTH-II-512012 (PROMEMORIA), FP7-HEALTH-F2M-2008-201600 (MemStick); Grant sponsor: Swiss National Science Foundation; Grant numbers: 3100A0-108102, 310000-120791; Grant sponsor: Intramural Funding from the Ecole Polytechnique Federale de Lausanne (EPFL). *Correspondence to: Prof. Carmen Sandi, Brain Mind Institute, Ecole Polytechnique Federale de Lausanne (EPFL), SV 2810, Station 19, 1015 Lausanne, Switzerland. E-mail:
[email protected] Accepted for publication 11 September 2009 DOI 10.1002/hipo.20723 Published online 17 November 2009 in Wiley Online Library (wileyonlinelibrary.com). C 2009 V
WILEY-LISS, INC.
2005; Sandi and Bisaz, 2007). Studies in animals have shown that chronic stress leads to a myriad of structural and functional alterations in several brain regions, with the hippocampus showing marked vulnerability and suffering from dendrite atrophy, impaired synaptic plasticity, and diminished neurogenesis (McEwen, 2000; Fuchs et al., 2001; McEwen and Lasley, 2003). Because of the role of the hippocampus in memory processes, these stress-induced alterations are often accompanied by learning and memory impairments (Squire et al., 2004; Kesner and Hopkins, 2006). In addition, recent work has shown that the medial prefrontal cortex (mPFC) is also highly sensitive, in terms of structural atrophy and behavioral output, to the impact of chronic stress (Cerqueira et al., 2007; Dias-Ferreira et al., 2009; Garrett and Wellman, 2009). Several studies in rodents have shown that the expression of the neural cell adhesion molecule (NCAM) is reduced in the hippocampus following chronic stress (Sandi et al., 2001; Touyarot and Sandi, 2002; Venero et al., 2002; Alfonso et al., 2006; Sandi and Touyarot, 2006). NCAM is an abundant cell adhesion macromolecule that exists in three main isoforms (NCAM-180, NCAM-140, and NCAM-120), each differing in their molecular weight, as well as in their distribution and function (Schuster et al., 2001; Kolkova, 2008). In addition to its Ca21-independent homophilic binding, NCAM also mediates heterophilic binding to tyrosine kinase receptors, such as fibroblast growth factor receptor (FGFR) and the glial cell line-derived brain derived neurotrophic factor family receptor a (GFRa), as well as to other cell adhesion molecules and various extracellular matrix components (Walmod et al., 2004; Kiselyov, 2008; Nielsen et al., 2008). NCAM participates in activitydependent synaptic rearrangements through: (i) the activation of intracellular signaling cascades (Buttner and Horstkorte, 2008; Ditlevsen and Kolkova, 2008); (ii) posttranslational modification that involves the attachment of extended chains of sialic acid (PSANCAM) and that provides NCAM with antiadhesive properties (Hildebrandt et al., 2008; Rutishauser, 2008); and (iii) altering its expression at the cell surface (Panicker et al., 2003; Sandi, 2004). Manipulations interfering with NCAM function (such as administration of NCAM antibodies or NCAM mimetic peptides) or its expression (through gene inactivation) impair long-term potentiation
NCAM MEDIATES STRESS-INDUCED COGNITIVE IMPAIRMENTS (LTP), leading to learning and memory deficits in a variety of cognitive tasks and to altered emotional behavior (for reviews, see Conboy et al., 2008; Hartz and Ronn, 2008). Furthermore, interventions that target the polysialylation of NCAM, either through the deletion of enzymes implicated in its polysialylation (ST8SiaII and ST8SiaIV) or by region-specific infusion of endoneuraminidase-N (Endo-N) that selectively cleaves the PSA moiety, have been shown to reduce synaptic plasticity in vitro and in vivo (Becker et al., 1996; Muller et al., 1996; Eckhardt et al., 2000). Indeed, such manipulations also produce learning and memory impairment in vivo (Angata et al., 2004; Venero et al., 2006; Lopez-Fernandez et al., 2007; Markram et al., 2007a,b). Interestingly, administration of the NCAM mimetic peptide FGL (a 15-amino-acid-long NCAMderived peptide, known to activate FGFR 1 and 2) potentiates NCAM function and it has been shown to: (i) induce neurite outgrowth and promote neuronal survival in vitro (Kiselyov et al., 2003; Neiiendam et al., 2004; Berezin and Bock, 2008); (ii) enhance cognitive functions under normal and neuropathological conditions in vivo (Cambon et al., 2004; Klementiev et al., 2007); and (iii) to have antidepressant-like properties in constitutional NCAM-deficient mice (Aonurm-Helm et al., 2008). NCAM also acts in neuronal remodeling and it can simultaneously interact with cytoskeletal elements, neurotrophic signals and intracellular signaling cascades (Kiss and Muller, 2001; Rougon and Hobert, 2003; Maness and Schachner, 2007). Since all of these elements have been implicated in the deleterious effects of stress (McEwen, 2002; Kuipers et al., 2003; McEwen, 2005; Duman and Monteggia, 2006), we hypothesized that NCAM might play a key role in stress-induced behavioral alterations (Sandi, 2004; Sandi and Bisaz, 2007; Bisaz et al., 2008). Since, most of the evidence available in support of this hypothesis is either correlative or circumstantial, we set out to provide causal evidence linking the stress-related decrease in NCAM expression with the cognitive impairments observed after stress. To address this issue, we compared learning and memory deficits in the Morris water maze in mice submitted to chronic stress and in conditional NCAM-deficient mice, in which the NCAM is ablated under the control of the aCaMKII promoter in hippocampal neurons postnatally (Bukalo et al., 2004). We also tested the ability of FGLs to counteract spatial learning deficits in conditional NCAM-deficient mice and in chronically stressed wild-type mice. Our results provide strong support for a key role of NCAM in chronic stress-induced cognitive alterations.
57
Charles River Laboratories (L’Arbresle Cedex, France). Mice were habituated to our animal facility for 18 days before beginning the experiments. Experiments on conditional NCAM-deficient mice were conducted in groups of age-matched adult conditional NCAM knockout male mice and their control littermates at 4 or 18 months of age. The generation of the conditional NCAMdeficient mice has been described previously (Bukalo et al., 2004). Briefly, homozygous NCAM-floxed mice were bred with homozygous NCAM-floxed mice that express the cre-recombinase under the control of the promoter of the a subunit of the calcium-calmodulin-dependent protein kinase II (aCaMKII). The progeny was homozygous for the NCAM-floxed alleles, half of them carried the aCaMKII-cre transgene (NCAMffcre) and the rest were control littermates (NCAMff ). These mice were backcrossed into the C57BL/6 background for more than 10 generations. All mice were housed in groups of two to five in standard plastic cages and they were kept under a 12 h light/dark cycle (lights on at 7.30 am) with ad libitum access to food and water. All the procedures carried out were performed in accordance with the Swiss National Institutional Guidelines on Animal Experimentation and they were approved by the Swiss Cantonal Veterinary Office Committee for Animal Experimentation. The number of animals used in the present study was kept to a minimum, as was animal suffering in all procedures.
Elevated Zero Maze Anxiety was measured in an elevated zero maze (EZM) as described previously (Madani et al., 2003). Briefly, mice were observed for 5 min in the EZM (a 5.5-cm-wide annular runway with a diameter of 46 cm and raised 46 cm above the ground) under dim and dispersed light conditions. Two opposing 908 sectors were protected by 13.5 cm high inner and outer walls. Thus, three zones were defined as follows: an intermediate zone comprising four 308 segments at the ends of the protection walls separated by the two 508 wide closed/protected and the two 708 wide open/unprotected exploration zones. With these boundaries, the entries into the open sectors were detected only when the animal entered into them with all four paws. The trajectories of each mouse were automatically recorded by video tracking (Ethovision 3.0, Noldus, Wageningen, Netherlands). The total number of entries into all the sectors served as an indicator of spontaneous locomotor activity, while differences in the number of entries and the time spent in the open sectors was taken as indicators of anxiety. Between sessions the maze was cleaned with 5% ethanol/water.
Chronic Unpredictable Stress MATERIALS AND METHODS Subjects Chronic unpredictable stress experiments were conducted on 3-month-old C57BL/6 wild-type male mice obtained from
After adaptation to the animal facility, all mice were characterized in terms of body weight, anxiety-like behavior in the EZM, locomotion and exploration in the open field (data not shown). The unpredictable chronic stress protocol involved exposing animals to a daily stressful situation at an unpredictable moment for 4 weeks (between 8 am and 4 pm, and Hippocampus
58
BISAZ ET AL.
randomly distributed over the 28 days). The stress stimuli used were either: 6 min tail suspension; 3 3 0.4 mA inescapable footshock; 4 h exposure to soiled, damp sawdust; 2 h exposure on an elevated platform; 1 h immobilization in a plastic tube; 30 min exposure to 168C; 2 days inversed light/dark cycle; 10 min exposure to an older, aggressive conspecific; 1 h exposure to a trimethylthiazoline (TMT); and 8 h with a 408 cage inclination. All animals were weighed and the state of their coat was evaluated on a regular basis (every 3–5 days). There were two chronic stress experiments. In the first one, the animals were divided into four equal groups of 8–9 mice, according to the initial behavioral characterization. Of these, two groups were left undisturbed and served as controls while the other two were exposed to chronic unpredictable stress for 28 days. One of the control groups and a group of stressed animals were used for behavioral studies, while the remaining two were used for biochemical analyses. The second chronic stress experiment was designed to test if the FGLs mimetic peptide would improve performance in the water maze. The protocol was the same as above, with one stress group injected with saline and the other stress group with FGLs.
Morris Water Maze The water maze apparatus consisted of a large white circular pool (140 cm diameter) filled with opaque colored water (268C 6 18C) and with a platform (10 3 10 cm2) submerged !1.3 cm under the water surface. The water maze was surrounded by gray curtains (25 cm from the pool periphery) containing several prominent visual cues. Data were collected using a video camera fixed to the ceiling that was connected to a video tracking system (Ethovision 3.0, Noldus, Wageningen, Netherlands). One day before training, all mice were habituated to the room, apparatus, and water by giving them a 2-min free swim trial with no platform present. Spatial learning sessions were conducted on three consecutive days (Days 1–3), performing four trials per day with an interval of 18 min between two consecutive trials. Each trial started by introducing the mouse into the maze with the aid of a cup, facing the pool wall, and at one of five possible positions that were randomly balanced between trials and days. The distance mice needed to swim to find the hidden platform were measured and if a mouse did not find the platform within 60 s, it was gently guided toward it. Each mouse had to remain on the platform for 15 s before it was returned into its waiting cage. Thigmotactic swimming (i.e., swimming close to the walls of the water maze) was also calculated as the percentage of time spent within 10 cm of the maze wall. During all spatial learning trials (Days 1–3), the platform remained in the same position. On Day 4, a 60 s probe test with no platform present was carried out, which was followed by a reversal learning session. At the end of the probe test the platform was reinserted into the pool in the opposite quadrant and the mice were guided to the new platform position where they stayed for 15 s. The platform remained in that opposite quadrant for all the reversal learning trials (Day 4). Hippocampus
For data analyses, trials were collapsed into blocks of two consecutive trials. For the probe test, the percentage time the mice swum in the target, opposite and adjacent quadrants, as well as the average distance to the target platform, were measured and served as an indication of spatial memory.
Tissue Sample Preparation Chronically stressed and control mice used to evaluate NCAM and PSA-NCAM expression in the hippocampus and the medial prefrontal cortex (n 5 8–9/group), were decapitated on the morning of Day 29 (24 h after the last applied stressor). Their brains were removed quickly and a sample of trunk blood was collected, for later quantification of basal corticosterone levels. The brains were frozen in isopentane and stored at 2808 until they were further processed. Hippocampal and medial prefrontal cortex (including the prelimbic, infralimbic, and cingulate cortex) fractions were obtained with tissue punches from 300 lm frozen brain slices, according to the protocol from Palkovits (1973). NCAM and PSA-NCAM levels in NCAMffcre and NCAMff mice (n 5 5–6/genotype) were quantified in naı¨ve animals, sacrificed in the morning and their brains were quickly removed. This tissue was kept on an ice-cold plate and the hippocampus and the medial prefrontal cortex were dissected out rapidly. Crude synaptosomal pellets were obtained according to a protocol modified from Lynch and Voss (1991) that has been previously shown to be effective to detect stress-induced changes in NCAM expression (Touyarot and Sandi, 2002; Touyarot et al., 2004; Venero et al., 2006). In brief, the tissue was homogenized with a plastic homogenizer in 10 vol. of ice cold HEPES (4 mM) buffered sucrose (0.32 M), containing a freshly added cocktail of protease inhibitors (Complete EDTAfree, Roche Diagnostics GmbH, Mannheim, Germany) and 1 mM EDTA. The solution was centrifuged at 1,000g for 5 min at 48C and the supernatant was centrifuged at 15,000g for 15 min, resuspending the resulting pellet in phosphate buffered saline (PBS) and 1% NP-40, containing protease inhibitors and 1 mM EDTA. The protein concentration for each sample was estimated by the method of Lowry et al. (1951).
Quantitative Immunoblotting of NCAM Expression levels of three main NCAM isoforms (NCAM180, NCAM-140, and NCAM-120) were measured in immunoblots of crude synaptosomal preparations of hippocampal and medial prefrontal cortex fractions. Synaptosomes from each mouse were incubated overnight at room temperature with endoneuraminidase-N (AbCys, Paris, France; final dilution 1:120) to selectively cleave the PSA moiety of NCAM. The reaction was stopped by boiling samples at 1008C for 5 min in 70 mM Tris-HCl (pH 6.8), 33 mM NaCl, 1 mM EDTA, 2% sodium dodecyl sulfate (SDS), 0.01% Bromophenol Blue, 10% glycerol and 3% dithiothreitol. 3.5 lg total protein from each sample was separated on 7.5% SDS-PAGE and transferred to a nitrocellulose membrane (Biotran BA85, Schleicher and Schuell). After saturation of nonspecific sites with 5% nonfat
NCAM MEDIATES STRESS-INDUCED COGNITIVE IMPAIRMENTS dry milk in 10 mM Tris-HCl (pH 7.4), containing 150 mM NaCl, 0.05% Tween-20 (TBST), the membranes were probed for 2 h at room temperature with primary antibodies against NCAM (1:5,000, Millipore) or actin (1:20,000, Sigma– Aldrich), washed with TBST, incubated for 2 h with the appropriate secondary horseradish peroxidase-linked antibodies, and finally developed using the SuperSignal West Dura Substrate (Pierce, Rockford, IL). Bands were detected using the ChemiDoc XRS system (Biorad, Hercules, CA) and densitometry analysis on band was calculated using Biorad Quantity One 4.2.3 software (Biorad Laboratories AG, Switzerland). Samples from individual subjects were subjected to duplicate immunoblots (Interblot coefficient of variability was 5.8%) and were the mean of both. Following normalization to within-lane actin (data not shown), the expression of all three main NCAM isoforms was expressed as the percentage of the control animals. The linear range of specific antibody signal detection was determined at the outset of these experiments and all experimental samples were loaded at a concentration within the linear range of the antibody signal detection.
PSA-NCAM ELISA PSA-NCAM levels were quantified in the same samples that were used for NCAM quantification by performing commercial PSA-NCAM ELISAs (AbCys Paris, France). A total volume of 100 ll from each sample was loaded at a concentration of 2 lg ml21 per well in duplicates and amounts of PSA were estimated according the manufacturer’s protocol. PSA-NCAM levels (ng PSA/lg of total protein) were calculated and normalized to total NCAM expression.
Corticosterone Assay Basal and stimulated corticosterone levels were measured in naive NCAMffcre and NCAMff mice (n 5 9/group). To assay the basal corticosterone levels, blood was collected by tail incision in the morning (8–10 am). Stimulated corticosterone levels were assayed after exposing the animals to a novel environment for 20 min (gray, round vertical plastic tube; 28 cm diameter, 37 cm high walls), also in the morning (8–10 am). Blood was collected immediately afterwards by tail incision. The blood was centrifuged (2,000g for 5 min at 48C), and the serum was extracted and stored at 2208C until the corticosterone levels were assayed by ELISA (Assay Design, Ann Arbor, MI). Under basal conditions, two animals (one of each genotype) displayed plasma corticosterone levels higher than the mean 6 2 standard deviations and they were excluded from the analysis.
FGLs Treatment FGLs, the dimeric form of the undecapeptide FGL (VAENQQGKSKA), was used in this study, which was kindly provided by ENKAM Pharmaceuticals A/S (Copenhagen, Denmark). FGLs were composed of two FGL monomers
59
linked through their N-terminal ends by iminodiacetic acid (Bachem AG, Bubendorf, Switzerland/NeoMPS, Strasbourg, France). FGLs purity was at least 96% when estimated by high performance liquid chromatography (HPLC). The effect of FGLs treatment was tested first in NCAMffcre mice and then in chronically stressed wild-type mice. In each case, mice were divided into two groups (n 5 5–11/group) and they were subcutaneously injected with either 10 mg kg21 FGLs in 0.9% NaCl (saline), or the corresponding volume of saline alone, on Days 1, 2, and 3, immediately after each learning session in the water maze (training Days 1–3).
Data Analysis All the results were expressed as the mean 6 standard error of the mean (SEM) and they were analyzed with the StatView version 5.0 package (SAS Institute, Cray, NC). The data was analyzed with the Student’s t test or by analysis of variance (ANOVA) with or without repeated measures, as appropriate. Post hoc tests (PLSD Fisher) were applied whenever ANOVA yielded significant interactions. In the water maze experiments, in addition to analyzing the global effects of the treatments over the whole learning procedure (with repeated measures ANOVA), data in the literature led us to set the a priori hypothesis that treatment and genotype effects might manifest as a sporadic impairment on individual training days (Venero et al., 2002; Sandi et al., 2003b; Bukalo et al., 2004; Cambon et al., 2004; Touyarot et al., 2004; Stewart et al., 2005; Wright and Conrad, 2008). Therefore, the Student’s t test was also applied to individual block trials in the water maze. Data regarding the NCAM isoforms was also analyzed with both repeated measures ANOVA and with Student’s t test for each isoform. The significance of the results was accepted at P ! 0.05.
RESULTS Physiological Effects of Chronic Stress: Change in Body Weight and Basal Plasma Corticosterone The stressful nature of the unpredictable chronic stress protocol used in this study was validated by evaluating body weight in control and stressed mice. The mice subjected to stress showed a significant reduction in body weight over the period analyzed (repeated measure ANOVAs for body weight change from Day 4 to 29: F7,224 5 12.563, P < 0.0001) when compared to the control mice (Fig. 1a; F1,32 5 38.267, P < 0.0001). In addition, ANOVA for repeated measurements showed a significant ‘‘stress 3 day’’ interaction throughout the chronic stress experiment (Fig. 1a; F7,224 5 11.277, P < 0.0001). The differences in body weight between stressed and control mice were particularly pronounced from Day 12 onward, and post hoc analyses identified significant differences between stressed and control mice on Days 12, 15, 19, 23, 26, and 29 (P ! 0.01). Hippocampus
60
BISAZ ET AL. NCAM expression when compared to controls (Fig. 2d; t15 5 3.306, P < 0.01).
Effect of Chronic Stress on Spatial and Reversal Learning in the Morris Water Maze
FIGURE 1. Reduced body weight and plasma corticosterone levels as a consequence of chronic stress in C57BL/6 wild-type male mice. (a) Stressed mice displayed a significant reduction in body weight throughout the chronic stress experiment when compared to control animals (n 5 16–18/group). (b) No significant difference in basal plasma corticosterone levels in the morning of Day 29 (24 h after the last applied stressor) was found when compared to control animals (n 5 8–9/group). Results are the mean 6 SEM (**P < 0.0001 vs. controls).
Trunk blood samples were taken on the morning of Day 29 to measure basal corticosterone levels (one day after the last stress stimulus had been applied). Chronically stressed animals displayed a tendency (although not significant) toward higher basal plasma corticosterone levels when compared to control mice (Fig. 1b; t15 5 21.401, P 5 0.09). This tendency was in agreement with a broad body of data showing similar results in mice and rats submitted to chronic stress (Touyarot and Sandi, 2002; Li et al., 2006, 2008; Sandi and Touyarot, 2006).
Effects of Chronic Stress on NCAM and PSA-NCAM Expression in the Hippocampus and the Medial Prefrontal Cortex ANOVA of repeated measures for the three major NCAM isoforms in hippocampal samples revealed a significant overall effect of chronic stress on the expression of these proteins (Fig. 2a; F1,15 5 4.962, P < 0.05). The Student’s t tests for each isoform revealed significant effects of stress on the expression of the NCAM-180 and NCAM-140 isoforms (NCAM180, t15 5 1.858, P < 0.05; NCAM-140, t15 5 2.291, P < 0.05), with lower levels of these isoforms detected in stressed animals than in controls. However, no significant differences were detected for the predominantly glial NCAM-120 isoform (t15 5 1.471, n.s.). Chronically stressed animals also displayed an overall increase of PSA-NCAM in the hippocampus when compared to controls (Fig. 2b; t15 5 21.819, P < 0.05). In the mPFC samples, repeated measures ANOVA for the three major NCAM isoforms revealed no significant overall effect of chronic stress (Fig. 2c; F1,15 5 0.75, n.s.). Additionally, Student’s t tests for each isoform revealed no significant effects of stress (NCA-180, t15 5 1.229, n.s.; NCAM-140, t15 5 0.849, n.s.; NCAM-120, t15 5 0.451, n.s.). However, in this brain region, chronically-stressed mice displayed an overall reduction of PSA-NCAM relative to Hippocampus
Spatial learning was assessed in a set of control and chronically stressed mice (n 5 8–9/group) starting one day after the end of the chronic stress procedure (i.e., on Day 29). The performance of the mice was analyzed in terms of the distance swum to find the platform instead of the latency, since overall the speed at which the chronically stressed mice swim was significantly higher than that of the controls (Supporting Information Fig. S1a; F1,15 5 7.910 P 5 0.01). Notably, this difference in speed was observed during the spatial learning days (Day 1–3: F1,15 5 9.321, P < 0.01) as well as in the probe test (t15 5 22.082, P 5 0.05). However, no significant difference in swim speed was found between stressed and control mice during the reversal learning session (Supporting Information Fig. S1a; F1,15 5 3.426, n.s.). Repeated measures ANOVA of the distances swum to find the hidden platform over the three training days revealed a significant effect for the blocks of trials factor (Fig. 3a; F5,75 5 5.445, P < 0.0003) indicating that overall, the distance swum diminished as the training proceeded. However, ANOVA indicated a lack of effect for the ‘‘stress’’ factor (F1,15 5 2.134, n.s.) and for the ‘‘block 3 stress’’ interaction (F5,75 5 0.596, n.s.). Since there is substantial data in the literature indicating that chronic stress leads to mild learning impairment in the water maze that only manifests sporadically in certain training trials (Venero et al., 2002; Sandi et al., 2003b; Bukalo et al., 2004; Cambon et al., 2004; Touyarot et al., 2004; Stewart et al., 2005; Wright and Conrad, 2008), we also performed Student’s t test analyses for each block of two trials to explore this possible effect. Indeed, we found that chronically stressed mice swam significantly longer distances to find the hidden platform specifically in the last block of trials (B6) on Day 3 when compared to control mice (Fig. 3a; t15 5 22.035, P < 0.03). Moreover, repeated measures ANOVA revealed no difference in thigmotactic behavior between control and stressed animals over the three spatial learning days (Supporting Information Fig. S1b; Day 1–3: F1,15 5 0.177, n.s.). On Day 4, the mice were submitted to a probe test for 60 s in which the platform had been removed. At the end of the probe test, the platform was set in the quadrant opposite to the training session and mice were allowed to remain on it for 15 s. Control mice exhibited strong quadrant preference during the probe test and spent significantly more time searching in the target quadrant than in the opposite and adjacent quadrants, indicating a focused navigation strategy toward the target quadrant (Fig. 3b; ANOVA; factor percentage time in quadrants: F2,21 5 14.196, P < 0.0001; post hoc analysis was P < 0.01 for the time spent in the target quadrant vs. opposite or adjacent quadrants). Conversely, chronically stressed mice performed at the level of chance and spent a similar amount of time swimming in the target, opposite and adjacent quadrants
NCAM MEDIATES STRESS-INDUCED COGNITIVE IMPAIRMENTS
61
FIGURE 2. Chronic stress led to reduced NCAM and increased PSA-NCAM expression in the hippocampus, as well as to reduced PSA-NCAM levels in the medial prefrontal cortex (mPFC), of C57BL/6 wild-type mice. (a) Stressed mice displayed a significant reduction in the NCAM-180 and NCAM-140 isoforms in the hippocampus when compared to controls. (b) PSA-NCAM
levels increased in the hippocampus of stressed mice when compared to controls. (c) No significant difference in NCAM expression levels in the mPFC were found between stressed and control mice. (d) PSA-NCAM levels were reduced in the mPFC of stressed animals when compared to controls. Results are the mean 6 SEM. (n 5 8–9/group, *P < 0.05 vs. controls, **P < 0.01 vs. controls).
(Fig. 3b; ANOVA; factor percentage time in the quadrants: F2,24 5 2.318, n.s.; post hoc analysis was n.s. for the time in the target quadrant vs. opposite or adjacent quadrants). Furthermore, when the average distance to the virtual target platform during the probe test was analyzed, chronically stressed mice displayed a significant higher average distance to the target platform than control mice (Fig. 3c; t15 5 21.705, P 5 0.05). Student’s t test revealed no difference in thigmotactic swimming between control and stressed mice during the probe test (Supporting Information Fig. S1b; t15 5 21.103, n.s.). Immediately following the probe test, mice were submitted to a reversal learning session on Day 4, with the escape platform positioned in the quadrant opposite to the training one (Fig. 3a). ANOVA for repeated measurements on the two blocks of two trials during the reversal learning session indicated a significant effect for the ‘‘stress’’ factor (F1,15 5 4.688, P < 0.05), but no effect for the ‘‘distance 3 stress’’ interaction (F1,15 5 0.52, n.s.). In addition, we found that stressed mice displayed significantly higher time performing thigmotactic swimming during reversal learning than controls (Supporting Information Fig. S1b; F1,15 5 5.063, P < 0.05). Taken together, these data indicate that 28 days of chronic unpredictable stress induced a mild impairment in spatial
learning and memory. In particular, it interfered with reversal learning, reducing cognitive flexibility when the task demanded the animals to ignore previously acquired information about the exact platform location and required them to relearn a new, relocated platform position. Increased swim speed accompanied learning deficits, while impaired reversal learning occurred along with enhanced thigmotaxis.
NCAM and PSA-NCAM Expression in the Hippocampus and the Medial Prefrontal Cortex of Conditional NCAM-Deficient Mice Hippocampal and mPFC NCAM and PSA-NCAM levels were evaluated in adult conditional NCAM-deficient (NCAMffcre) mice and their NCAMff littermates (n 5 5–6/group). To quantify NCAM expression, crude synaptosomal preparations were prepared from samples for each of these brain regions obtained from naive NCAMffcre and NCAMff animals. The expression of NCAM was defined in relation to that of the NCAMff animals (Figs. 4a,c) and when considering the three major NCAM isoforms together, the NCAMffcre mice showed significantly less NCAM than their NCAMff littermates in the hippocampus (Fig. 4a; F1,9 5 27.656, Hippocampus
62
BISAZ ET AL.
FIGURE 3. Chronic stress induced impairments in spatial and reversal learning, as well as in long-term spatial memory in the Morris water maze. (a) No difference between control and stressed animals was found during first two days of spatial learning (B1B4). However, stressed animals had to swim further to find the hidden platform in the second block of trials on Day 3 (B6). Moreover, stressed mice were significantly impaired in the reversal learning session on Day 4 when compared to control animals.
(b) During the probe test, stressed mice performed at chance level and they did not display a significant quadrant preference when compared to controls. (c) Additionally, stressed mice swam a farther average distance to the virtual target platform during the probe test when compared to control mice. Results are the mean 6 SEM. (n 5 8–9/group, *P ! 0.05 vs. controls, 1 P < 0.05 vs. all other quadrants of the same treatment group).
P < 0.001) as well as in the mPFC (Fig. 4c; F1,9 5 11.164, P 5 0.01). For NCAM expression levels in the hippocampus, ANOVA indicated a significant ‘‘genotype 3 NCAM isoform’’ interaction (Fig. 4a; F2,18 5 6.087, P < 0.01). Indeed, a post hoc PLSD Fisher analyses indicated that for each of the three NCAM isoforms, there was significantly less expression in NCAMffcre mice than in NCAMff mice (all P < 0.005; the same results were obtained when running independent Student’s t tests). Conversely, for NCAM expression in the mPFC, ANOVA revealed no significant ‘‘genotype 3 NCAM isoform’’ interaction (Fig. 4c; F2,18 5 0.729, n.s.). In fact, Student’s t tests confirmed that differences in expression levels between the two genotypes were significant for each of the three NCAM isoforms (Fig. 4c; NCAM-180: t9 5 22.999, P 5 0.02, NCAM-140: t9 5 23.843, P < 0.01, NCAM-120: t9 5 22.556, P 5 0.04). No differences in PSA-NCAM expression relative to NCAM were found between the two genotypes, neither in the hippocampus (Fig. 4b; t9 5 20.085, n.s.), nor in the mPFC (Fig. 4d; t9 5 20.437, n.s.) indicating that the relative polysialylation of NCAM molecules was not altered in conditional NCAM-deficient mice.
5a; F5,100 5 16.071, P < 0.0001), but no significant effects for either the ‘‘genotype’’ (F1,20 5 0.238, n.s.) or the ‘‘block 3 genotype’’ interaction (F5,100 5 0.617; n.s.). When the data from each of the block of trials was analyzed, NCAMffcre mice swam significantly longer to reach the platform on Block 4 (the second trial block of Day 2) than NCAMff mice (Fig. 5a; t20 5 1.782, P < 0.05), while no differences were observed in any other training blocks (all P > 0.1). Repeated measures ANOVA revealed no difference in the overall swim speed between animals of both genotypes (Supporting Information Fig. S2a; F1,20 5 1.671, n.s.). Moreover, NCAMffcre mice displayed no difference in swim speed during the 3 days of spatial learning (F1,20 5 1.351, n.s.), the probe test (t20 5 1.029, n.s.) or the reversal learning session on Day 4 (F1,20 5 3.039, n.s.). Moreover, no difference in thigmotactic swimming was found between animals of both genotypes during the 3 days of spatial learning (Supporting Information Fig S2b; F1,20 5 1.425, n.s.). Data from the probe test indicated a significant quadrant preference on Day 4 for both genotypes (Fig. 5b; ANOVA for percentage time in quadrants: NCAMffcre: F2,45 5 35.241, P < 0.0001; NCAMffcre: F2,15 5 10.575, P < 0.001; post hoc analysis were P < 0.01 for the percentage of time in the target quadrant vs. the opposite and adjacent quadrants for both genotypes), indicating that they spent most time swimming in the target quadrant rather than in the other quadrants. However, when both genotypes were compared, NCAMffcre mice spent significantly less time searching in the target quadrant (Fig. 5b; t20 5 21.704, P 5 0.05) than NCAMff mice. Likewise, NCAMffcre mice swam farther from the virtual target platform during the probe test than NCAMff mice (Fig. 5c; t20 5 1.733, P < 0.05). Student’s t test revealed no differ-
Spatial Learning and Reversal Learning of Conditional NCAM-Deficient Mice in the Morris Water Maze Spatial learning and memory was tested in NCAMffcre mice and their NCAMff littermates in the water maze following the same protocol used for the chronic stress experiment. Repeated measures ANOVA for the distance swum to the platform indicated a significant effect of the block of trials factor (Fig. Hippocampus
NCAM MEDIATES STRESS-INDUCED COGNITIVE IMPAIRMENTS
63
FIGURE 4. NCAM and PSA-NCAM expression levels in the hippocampus and the medial prefrontal cortex (mPFC) of conditional NCAM-deficient mice. (a,c) Conditional NCAM-deficient mice (NCAMffcre) express significantly less of all the major NCAM isoforms in the hippocampus and the mPFC when com-
pared to their NCAMff littermates. (b,d) No difference in PSANCAM in relation to NCAM expression was found in the hippocampus and the mPFC of NCAMffcre mice. Results are the means 6SEM. (n 5 5–6/group, *P < 0.05 vs. NCAMff, **P < 0.01 vs. NCAMff ).
ence in thigmotactic swimming between animals of both genotypes during the probe test (Supporting Information Fig. S2b; t20 5 0.809, n.s.). During reversal learning session on Day 4, repeated measures ANOVA indicated a significant effect of the training blocks (F1,20 5 6.346, P 5 0.02), as well as a significant effect of genotype since NCAMffcre mice swam further to find the hidden platform than NCAMff mice (F1,20 5 7.987, P 5 0.01: Fig. 5a). No effect of a ‘‘block 3 genotype’’ interaction was found (F1,20 5 0.004; n.s.). In fact, both PLSD Fisher post hoc analyses (P < 0.01) or the Student’s t tests for each block separately (Fig. 5a; Day 4; block 1: t20 5 2.335, P < 0.05; Day 4, block 2: t20 5 2.747, P < 0.05) indicated that there were differences between the two genotypes in each block. Repeated measures ANOVA for thigmotactic swimming during the reversal learning session revealed a significant difference between genotypes (Supporting Information Fig. S2b; F1,20 5 4.556, P < 0.05). Together, these data show that conditional NCAM-deficient (NCAMffcre) mice display mild impairment during spatial learning when compared with their NCAMff littermates, with weaker long-term spatial memory and severely impaired reversal learning. These differences were not accompanied by differences in swim speed, with thigmotactic behavior in the mutant
mice being only higher than in controls in the reversal learning session.
Anxiety-Like Behavior and Hormonal Stress Responses in Conditional NCAM-Deficient Mice Since we found that the performance of NCAMffcre mice in the water maze was impaired, we set up two experiments to explore the possibility that this genotype was associated with enhanced anxiety-like responses or reactions to stress that might have interfered with their behavior in the maze. A new set of animals was used to evaluate the anxiety-like behavior in the elevated zero maze (EZM) and NCAMffcre mice clearly explored the open, unprotected sectors of the EZM to a greater extent than NCAMff littermates. These behavioral differences were evident through the number of entries into the open sectors (Fig. 6a; t34 5 2.968, P < 0.01) and the time spent in the open sectors (Fig. 6b; t34 5 2.753, P < 0.01). Indeed, such differences are conventionally interpreted as an indication of lower levels of anxiety in the mutant mice, particularly since the total locomotor activity of both genotypes was comparable in this test (Fig. 6c; t34 5 1.735, n.s.). Hippocampus
64
BISAZ ET AL.
FIGURE 5. Spatial and reversal learning impairments, and reduced long-term spatial memory of conditional NCAM-deficient mice in the Morris water maze. (a) Conditional NCAM-deficient (NCAMffcre) mice displayed a mild impairment in spatial learning when compared to NCAMff littermates that was evident in the distance swum to find the hidden platform during the second block of trials on Day 2 (B4). Additionally, NCAMffcre mice performed significantly worse during the whole reversal learning session than
NCAMff. (b) During the probe test, NCAMffcre mice displayed significant target quadrant preference but they spent less time in the target quadrant than NCAMff animals. (c) Moreover, NCAMffcre mice swam farther to the virtual target platform during the probe test than NCAMff animals. The results are the means 6 SEM. (n 5 7–16/group, *P ! 0.05 vs. NCAMff, 1 P < 0.05 vs. all other quadrants of the same genotype).
Stress reactivity was analyzed by assessing plasma corticosterone levels in both genotypes, under basal conditions in the morning (when plasma corticosterone levels are low and stable) and after stimulation (i.e., immediately after exposing mice to a new, unfamiliar environment for 20 min). No difference in serum corticosterone levels was found between NCAMffcre and NCAMff mice, neither under basal conditions (t13 5 20.884, n.s.) nor after having been exposed to a new environment for 20 min (t15 5 1.199, n.s.: Fig. 6d).
Fig. S3a, F1,9 5 1.037, n.s.) and in thigmotactic swimming (Supporting Information Fig. S3b; F1,9 5 0.00003, n.s.) during Days 2 and 3 of spatial learning. In the probe test performed on Day 4, both FGLs- and saline-treated NCAMffcre mice showed significant quadrant preference (Fig. 7b; ANOVAs for the percent time in quadrants: NCAMffcre, saline: F2,15 5 29.991, P < 0.0001; NCAMffcre, FGL: F2,12 5 23.969, P < 0.0001) and they performed significantly above chance level (i.e., 25% in the target quadrant). Interestingly, post hoc analyses for the percentage time in the target quadrant vs. the opposite or adjacent quadrant was only significant for FGLs-treated (P < 0.01) but not for saline-treated animals (Fig. 7b). No difference between the groups was observed in the average distance to the target platform over the 60 s probe test (Fig. 7c; t9 5 20.662, n.s.). Moreover, no difference between both treatment groups was found in the swim speed (Supporting Information Fig. S3a; t9 5 1.421, n.s.) or in thigmotactic behavior (Supporting Information Fig. S3b; t9 5 0.403, n.s.) During the reversal learning session, FGLs-treated animals performed significantly better than saline-treated NCAMffcre mice (Fig. 7a; F1,9 5 4.988, P 5 0.05). While no significant effect of the ‘‘block 3 treatment’’ interaction was found (ANOVA for repeated measurements; factor ‘‘distance 3 treatment’’; Day 4: F1,9 5 0.096, n.s.), FGLs treated NCAMffcre mice displayed a significant higher swim speed than saline treated NCAMffcre mice (Supporting Information Fig. S3a; F1,9 5 5.112, P 5 0.05). No difference in thigmotactic behavior was found between both treatment groups (Supporting Information Fig. S3b; F1,9 5 0.003, n.s.).
Effect of FGLs Treatment on the Spatial Learning and Memory of NCAMffcre Mice in the Morris Water Maze To further verify the causal involvement of NCAM on the learning deficits observed in the water maze, a new experiment was designed to test whether post-training administration of the NCAM mimetic peptide FGLs improved the performance of NCAMffcre mice in the water maze (Fig. 7a). Repeated measures ANOVA for the blocks from Days 2 to 3 (note that the first FGLs injection was administered immediately after training on Day 1) indicated no overall effect of the ‘‘treatment’’ factor, (F1,9 5 3.085, n.s.), or for the ‘‘block 3 treatment’’ interaction, (F3,27 5 1.198, n.s.). When the data from each block of trials was analyzed separately, analyses indicated that FGLs-treated NCAMffcre performed significantly better (i.e., they swam less to find the platform) on the second block of Day 2 (B4) than saline treated mice of the same genotype (Fig. 7a; t9 5 2.294, P < 0.05). Repeated measures ANOVA revealed no difference between FGLs and untreated NCAMffcre mice in the swim speed (Supporting Information Hippocampus
NCAM MEDIATES STRESS-INDUCED COGNITIVE IMPAIRMENTS
65
FIGURE 6. Reduced anxiety-like behavior in the elevated zero maze (EZM) and normal hormonal stress response of conditional NCAM-deficient mice. Conditional NCAM-deficient mice (NCAMffcre) visited (a) and spent more time (b) in the open sectors of the EZM than their NCAMff littermates. (c) No difference
in the total zone entries was found between NCAMffcre and NCAMff mice. (d) Basal and stimulated corticosterone levels were similar in NCAMffcre and NCAMff mice. Results are the means 6 SEM. (n 5 16–20/group, **P < 0.01 vs. NCAMff ).
Therefore, administration of FGLs appeared to improve spatial learning (at a specific block trial during acquisition), memory, and reversal learning in conditional NCAM-deficient mice.
We then set an experiment to assess whether manipulating NCAM function through FGLs injections in a similar way as
in the previous experiment could improve chronic stressinduced deficits in the water maze (Fig. 8). Two groups of chronically stressed mice were included, one injected with FGLs and the other with saline. Repeated measures ANOVA for distance swum to find the hidden platform over Days 2 and 3 of spatial learning indicated no significant effect for the ‘‘treatment’’ factor (F1,19 5 0.687, n.s.) or for the ‘‘block 3 treatment’’ interaction (F3,57 5 0.538, n.s.). However, when data for each block was analyzed separately, FGLs treated mice swam significantly shorter to find the hidden platform on trial Block 4 than saline-treated mice (Fig. 8a; t19 5 21.699,
FIGURE 7. Effect of post-training FGLs administration (10 mg kg21) on spatial and reversal learning, and on long-term spatial memory of conditional NCAM-deficient mice in the Morris water maze. (a) In the second block of trials on Day 2 (B4) and during the reversal learning session, FGLs treated conditional NCAM-deficient mice (NCAMffcre) swam a shorter distance to find the platform when compared to saline treated NCAMffcre mice (arrows indicate post-training FGLs injections). (b) FGLs
treated NCAMffcre mice showed significant quadrant preference and they spent more time in the target quadrant than in the opposite or adjacent quadrants during the probe test. By contrast, saline treated NCAMffcre mice do not display target quadrant preference. (c) FGLs administration did not affect the average distance to the virtual target platform during the probe test. The results are the means 6 SEM. (n 5 5–6/group, *P ! 0.05 vs. saline treated, 1 P < 0.05 vs. all other quadrants of the same treatment group).
Effect of FGLs Treatment on the Spatial Learning and Memory of Chronically Stressed Wild-Type Mice in the Morris Water Maze
Hippocampus
66
BISAZ ET AL.
FIGURE 8. Effect of posttraining FGLs administration (10 mg kg21) on spatial and reversal learning, and on long-term spatial memory of chronically stressed C57BL/6 wild-type mice in the Morris water maze. (a) FGLs treated stressed wild-type mice swam a significantly shorter distance to find the hidden platform in the second block of trials on Day 2 (B4) and during the reversal learning session than saline treated stressed mice (arrows indicate posttraining FGLs injections). (b) Both FGLs and saline treated
stressed wild-type mice showed significant quadrant preference and they spent more time in the target quadrant than in the opposite or adjacent quadrants during the probe test. (c) FGLs administration did not affect the average distance to the virtual target platform during the probe test. The results are the means 6 SEM. (n 5 10–11/group, *P ! 0.05 vs. saline treated, 1 P < 0.05 vs. all other quadrants of the same treatment group).
P 5 0.05), while no difference was found in any other block (P > 0.1). Repeated measures ANOVAs revealed a higher overall swim speed of FGLs treated mice (Supporting Information Fig. S4a; F1,19 5 10.064, P < 0.01), but no differences in thigmotactic swimming (Fig. 4b; F1,19 5 0.224, n.s.) during Days 2 and 3 of spatial learning. In the probe test, both FGLs- and saline-treated stressed mice showed a significant quadrant preference (Fig. 8b; ANOVAs for the percent time in quadrants: stressed, saline: F2,27 5 11.093, P < 0.001; stressed, FGLs: F2,30 5 38.141, P < 0.0001) and both treatment groups performed significantly above the chance level. Post hoc analyses for the percentage time in the target quadrant vs. the opposite or adjacent quadrant were significant for both FGLs- and saline treated mice (P < 0.01). No difference between the groups was observed in the average distance to the target platform over the 60 s probe test (Fig. 8c; t19 5 20.928, n.s.). Swim speed was significantly higher in FGLs-treated than in salinetreated mice (Supporting Information Fig. S4a; t19 5 3.727, P < 0.01), while no difference was found in the time spent on thigmotactic swimming (Supporting Information Fig. S4b; t19 5 20.611, n.s.). During the reversal learning session on Day 4, FGLs-treated animals performed significantly better than saline-treated stressed mice (Fig. 8a; F1,19 5 4.295, P 5 0.05). The treatment did not induce significant differences in swim speed (Supporting Information Fig. S4a; F1,19 5 0.633, n.s.) or thigmotactic behavior (Supporting Information Fig. S4b; F1,19 5 4.025, n.s.). Therefore, similarly to NCAM-deficient mice, post-training administration of FGLs in chronically stressed wild-type mice appeared to improve spatial learning (at a specific block trial during acquisition) and, particularly, reversal learning.
DISCUSSION
Hippocampus
Previous work in rats identified reduced NCAM expression in the hippocampus and other forebrain areas after exposure to chronic stress (Sandi et al., 2001; Touyarot and Sandi, 2002; Venero et al., 2002; Touyarot et al., 2004; Alfonso et al., 2006; Sandi and Touyarot, 2006; Shin et al., 2009) in conjunction with deficits in spatial learning (Luine et al., 1994; Conrad et al., 1996; Park et al., 2001; Venero et al., 2002; Sandi et al., 2003a; Li et al., 2006). Using adult conditional NCAM-deficient mice in which the NCAM gene is ablated in glutamatergic neurons of the forebrain by the second postnatal week (Bukalo et al., 2004), we addressed whether the reduction in NCAM expression might have a causal effect on the learning deficits observed. Accordingly, we first validated that the effects of chronic stress described in rats can be generalized to mice and indeed, there is less hippocampal NCAM in C57BL/6 wild-type mice exposed to chronic unpredictable stress, as well as deficits in learning and reversal learning in the water maze. However, reduction in PFC levels in chronically stressed mice did not reach significance. In adulthood, conditional NCAM-deficient (NCAMffcre) mice show a marked reduction of NCAM in the hippocampus and prefrontal cortex and they displayed poor performance in the water maze that was, to some extent, comparable to that of stressed wild-type mice. These deficits could not be attributed to increased anxiety, which was less pronounced in these mice when compared to their NCAMff littermates, or to altered hormonal stress responses as their corticosterone levels were comparable to those of NCAMff mice. Interestingly, the performance of both NCAMffcre and wild-type chronically stressed mice in the water maze improved after post-training injections of the NCAM mimetic peptide FGLs. Together these findings provide strong support
NCAM MEDIATES STRESS-INDUCED COGNITIVE IMPAIRMENTS for a key role of NCAM in the stress-induced deficits in cognitive performance. To our knowledge, this is the first study showing that 28 days of chronic unpredictable stress in male mice leads to a reduction of hippocampal NCAM protein expression similar to that described in rats submitted to chronic restrain stress (Sandi et al., 2001; Touyarot and Sandi, 2002; Venero et al., 2002). Stress-induced elevations of glucocorticoids might play a role in the reduction of NCAM expression, as chronic treatment with corticosterone was shown to decrease NCAM expression in the frontal cortex (Sandi and Loscertales, 1999). Reduced transcription of NCAM may also contribute to this phenomenon as reduced NCAM mRNA levels were found in the hippocampus of rodents subjected to 3–4 weeks of chronic stress (Venero et al., 2002; Alfonso et al., 2006). Likewise, increased shedding of NCAM molecules from the cellular membrane has also been proposed as a response to stress-induced increases in extracellular ATP (Sandi, 2004). NCAM possesses intrinsic ecto-adenosyl triphosphate (ecto-ATP) activity, presumably associated with the extracellular fibronectin Type-III domain of the protein, which has also been implicated in the binding and concomitant activation of the FGFR (for a review see Hubschmann and Skladchikova, 2008). ATP is a neurotransmitter that is often released from neuronal cells as a cotransmitter in glutamatergic and noradrenergic synaptic vesicles (Burnstock, 1995; Mori et al., 2001) and it induces metalloprotease-mediated NCAM ectodomain release (Hubschmann and Skladchikova, 2008). Indeed, it is relevant to mention that increased levels of NCAM immunoreactive proteins were found in the CSF of patients suffering from mood disorders (Poltorak et al., 1996), which are frequently associated with increased stress and glucocorticoid levels (de Kloet et al., 2005; Pariante and Lightman, 2008). Interestingly, we also confirmed in wild-type mice earlier findings in rats, whereby hippocampal PSA-NCAM expression increases after chronic stress and/or chronic glucocorticoid treatment (Sandi et al., 2001; Touyarot and Sandi, 2002; Pham et al., 2003). However, while the reduction in NCAM was proposed to play a direct role in stress-induced cognitive impairment, the increase in PSA-NCAM was thought to be linked to compensatory mechanisms activated by the hippocampal remodeling induced by chronic stress (Sandi, 2004). In the PFC, previous work in rats had shown reduced NCAM expression after chronic exposure to glucocorticoids (Sandi and Loscertales, 1999) or stress (Huang et al., 2008). However, in our study, the slight reductions in NCAM levels observed in chronically-stressed mice did not reach significance, while PSA-NCAM expression was specifically inhibited by stress in this brain region. This is the first evidence that chronic stress can inhibit the levels of this neuroplasticity molecule (i.e., PSA-NCAM) in the PFC which, strikingly, it is the opposite effect to the one observed after chronic antidepressant treatments (i.e., enhancement of PSA-NCAM expression in the PFC) (Sairanen et al., 2007; Varea et al., 2007). We also showed that the stress-induced learning deficits previously described in chronically-stressed rats were similarly reproduced in the C57BL/6 mouse strain. A feature of the
67
effects produced by chronic stress in rats is the mild learning impairment during the acquisition phase in the water maze. This phenomenon normally manifests in one or only a few scattered trials in which the stressed animals perform worse than controls, even though stressed rats do acquire the spatial orientation learning in the overall training procedure (Venero et al., 2002; Sandi et al., 2003a; Touyarot et al., 2004; Stewart et al., 2005; Wright and Conrad, 2008). Likewise, the impairments observed in chronically-stressed mice during learning acquisition were observed on one of the training days. Moreover, we found that chronically-stressed mice performed poorly in the probe test and in particular, in the subsequent reversal learning session. A critical question that we addressed in this study was whether conditional NCAM deficiency would produce similar cognitive impairments as those seen in chronically stressed C57BL/6 wild-type mice. They did to a large extent under our experimental conditions, as NCAMffcre mice showed a mild learning impairment on one of the training days and reduced memory performance in the probe test. These learning deficits are in agreement with previous data indicating that these mice are less precise in spatial searching (Bukalo et al., 2004). Mice with constitutional loss of the NCAM gene were also found to exhibit profound learning and memory impairments in the water maze (Cremer et al., 1994; Stork et al., 2000), as well as in other hippocampus-dependent tasks such as contextual fear conditioning (Stork et al., 2000; Senkov et al., 2006). Importantly, as we found for stressed C57BL/6 wild-type mice, NCAMffcre mice were also markedly impaired in reversal learning, suggesting that NCAM deficiency induces a marked impairment in cognitive flexibility. Although difficult to compare since experiments were not run simultaneously, the magnitude of the reversal learning deficits displayed by NCAMffcre mice appeared to be greater than in chronically-stressed wildtype mice. Given that, in addition to the hippocampus, reversal learning in the water maze is thought to involve the prefrontal cortex (Cerqueira et al., 2005; de Bruin et al., 1994), differences in NCAM expression in the prefrontal cortex reported in this study between the two models (i.e., markedly lower in NCAMffcre mice, while not significantly reduced in chronically stressed wild-type mice) might be relevant for the behavioral differences. A note of caution should be drawn about the validity of the NCAMffcre mice model to assess for the implications of reduced NCAM expression induced by chronic stress. Although hippocampal NCAM levels were reduced in both models, the reduction was larger in the conditional knockout model, which was, in fact, the only one showing a significant reduction in the prefrontal cortex. While we focus our discussion here in the brain regions analyzed, differences in NCAM expression between the two models in other brain regions likely exist and, hence, could also contribute to the behavioral effects observed. Likewise, differences in brain-region specific content of other molecules induced by each of the two studied models might also be important. Note that, in our study, we detected important differences in the ratio of PSA-NCAM to NCAM expresHippocampus
68
BISAZ ET AL.
sion: while this was not changed in NCAMffcre mice, chronically stressed mice showed higher levels in the hippocampus but lower levels in the PFC. In addition, chronic stress is well known to affect a wide number of other molecules throughout the brain (de Kloet et al., 2005; McEwen, 2007). In fact, despite the resemblance of the behavioral alterations exhibited by the two models, there were also interesting differences. For example, chronically stressed mice showed increased swim speed at learning (not observed in NCAMffcre mice) that might indicate an anxious reaction to the task and, therefore, a nonfocused learning strategy. However, analysis of thigmotactic behavior failed to support this interpretation. In fact, enhancement of thigmotactic behavior was found for mice from both models in parallel with their impairment in reversal learning. It is important to note that the learning deficits observed in NCAMffcre mice are not due to a lack of NCAM expression during development, as NCAM expression in these mice is normal during development and the early postnatal period (Bukalo et al., 2004). Unlike constitutional NCAM knockout mice that were shown to display several changes in different brain structures, including the hippocampus (Cremer et al., 1994, 1997), no histological morphological abnormalities were observed by light microscopy in the NCAMffcre mice studied here (Bukalo et al., 2004). Likewise, these animals showed no alterations in hippocampal PSA-NCAM levels. Moreover, we can exclude the possibility that the spatial learning and memory deficits observed in NCAMffcre mice might be due to enhanced anxiety-like levels (Herrero et al., 2006). Indeed, the NCAMffcre mice show reduced anxiety-like behavior when naı¨ve animals are evaluated in the elevated zero mazes. Additionally, differences in stress responsiveness seem not to account for the behavioral changes since basal or stimulated (novelty exposure) corticosterone levels did not differ in NCAMffcre mice. It is important to note that the emotion- and stress-related responses in NCAMffcre mice differ from the impact of constitutive ablation of the NCAM gene. On the contrary, a mediating role of anxiety-like behavior and stress reactivity could not be excluded from behavioral deficits previously found in constitutive NCAM knockout mice, since they show enhanced anxiety-like behaviors (Cremer et al., 1994; Stork et al., 2000) and enhanced corticosterone responses (Stork et al., 1997). To further verify whether the spatial learning deficits associated with reduced NCAM expression were linked to alterations in NCAM function, immediately after each training session in the water maze we treated NCAMffcre and chronically stressed wild-type mice with the NCAM mimetic peptide FGLs. In both cases, we found that daily post-training FGLs treatment facilitated spatial learning performance on one of the training blocks (B4), just when these respective groups of mice demonstrated impaired learning. By contrast, memory retention during the probe tests was not affected. However, again, there was a marked improvement in reversal learning in both experiments. Systemic FGLs administration was effective in accordance with previous evidence showing that systemic injection of FGL rapidly penetrates into the cerebrospinal fluid (CSF) Hippocampus
(Secher et al., 2008). The FGL peptide corresponds to a part of NCAM that binds to and activates the FGFR, and it has been shown to promote synaptogenesis, enhance presynaptic functioning and to facilitate memory consolidation (Cambon et al., 2004; Secher et al., 2008). Given that former experiments in rats indicated a memory facilitating effect for FGL under normal conditions (i.e., in control animals not submitted to chronic stress or to any other treatment leading to a cognitive deficiency), the possibility exist that the effects reported here are not linked to the improvement of memory impairments linked to NCAM deficiency, but to a more general effect of FGL on cognitive enhancement. Importantly, administration of FGL to mid-age rats during a 4-week stress protocol, and subsequent boosting over the next 6 months of both stress and FGL treatment, prevented stress-induced spatial learning impairment in the water maze, while the same treatment did not enhance performance in control, unstressed, rats (Borcel et al., 2008). Therefore, although FGL seems to be able to induce cognitive enhancing properties both in normally performing animals and under conditions such as chronic stress, involving NCAM deficiency in hippocampus and PFC, the peptide’s effective dosage and administration regime might differ for the different conditions. Future studies should systematically address this important issue. Interestingly, these positive effects of FGL are not a general effect of NCAM mimetic peptides. Indeed, the NCAM mimetic peptide C3d, which interferes with NCAM-mediated cell–cell adhesion mediated by the first immunoglobulin domain of the molecule (Ronn et al., 1999), impairs memory formation in a variety of tasks including the water maze (Foley et al., 2000; Cambon et al., 2003; Hartz et al., 2003; Venero et al., 2006). The idea that interfering with NCAM adhesion impairs memory function is also supported by data from experiments involving intracerebroventricular infusion of NCAM antibodies (Doyle et al., 1992; Scholey et al., 1993; Mileusnic et al., 1995). Together, these findings highlight the functional significance of NCAM and particularly, that of the NCAM-FGFR interactions in learning and memory. In conclusion, the results of the present study support the view that alterations in NCAM expression in response to chronic stress are involved in stress-induced cognitive disturbances. They also indicate that treatments enhancing NCAM function may serve as a pharmacological intervention to prevent or overcome the cognitive impairments induced by stress. Although further work is needed to elucidate the implications of other CAMs in these regulatory changes, our findings highlight NCAM as important mediator of the negative effects of chronic stress on brain function and cognition.
Acknowledgments The authors thank ENKAM Pharmaceuticals (Copenhagen, Denmark) for providing the FGL peptide and Coralie Siegmund for her excellent technical assistance with the corticosterone assays.
NCAM MEDIATES STRESS-INDUCED COGNITIVE IMPAIRMENTS
REFERENCES Alfonso J, Frick LR, Silberman DM, Palumbo ML, Genaro AM, Frasch AC. 2006. Regulation of hippocampal gene expression is conserved in two species subjected to different stressors and antidepressant treatments. Biol Psychiatry 59:244–251. Angata K, Long JM, Bukalo O, Lee W, Dityatev A, Wynshaw-Boris A, Schachner M, Fukuda M, Marth JD. 2004. Sialyltransferase ST8Sia-II assembles a subset of polysialic acid that directs hippocampal axonal targeting and promotes fear behavior. J Biol Chem 279:32603–32613. Aonurm-Helm A, Jurgenson M, Zharkovsky T, Sonn K, Berezin V, Bock E, Zharkovsky A. 2008. Depression-like behaviour in neural cell adhesion molecule (NCAM)-deficient mice and its reversal by an NCAM-derived peptide, FGL. Eur J Neurosci 28:1618–1628. Becker CG, Artola A, Gerardy-Schahn R, Becker T, Welzl H, Schachner M. 1996. The polysialic acid modification of the neural cell adhesion molecule is involved in spatial learning and hippocampal long-term potentiation. J Neurosci Res 45:143–152. Berezin V, Bock E. 2008. NCAM mimetic peptides: An update. Neurochem Res 10.1007/s11064-008-9771-0. Bisaz R, Conboy L, Sandi C. 2009. Learning under stress: A role for the neural cell adhesion molecule NCAM. Neurobiol Learn Mem 91:333–342. Borcel E, Perez-Alvarez L, Herrero AI, Brionne T, Varea E, Berezin V, Bock E, Sandi C, Venero C. 2008. Chronic stress in adulthood followed by intermittent stress impairs spatial memory and the survival of newborn hippocampal cells in aging animals: Prevention by FGL, a peptide mimetic of neural cell adhesion molecule. Behav Pharmacol 19:41–49. Bukalo O, Fentrop N, Lee AY, Salmen B, Law JW, Wotjak CT, Schweizer M, Dityatev A, Schachner M. 2004. Conditional ablation of the neural cell adhesion molecule reduces precision of spatial learning, long-term potentiation, and depression in the CA1 subfield of mouse hippocampus. J Neurosci 24:1565–1577. Burnstock G. 1995. Noradrenaline and ATP: Cotransmitters and neuromodulators. J Physiol Pharmacol 46:365–384. Buttner B, Horstkorte R. 2008. Intracelluar ligands of NCAM. Neurochem Res 10.1007/s11064-008-9592-1. Cambon K, Venero C, Berezin V, Bock E, Sandi C. 2003. Post-training administration of a synthetic peptide ligand of the neural cell adhesion molecule, C3d, attenuates long-term expression of contextual fear conditioning. Neuroscience 122:183–191. Cambon K, Hansen SM, Venero C, Herrero AI, Skibo G, Berezin V, Bock E, Sandi C. 2004. A synthetic neural cell adhesion molecule mimetic peptide promotes synaptogenesis, enhances presynaptic function, and facilitates memory consolidation. J Neurosci 24: 4197–4204. Cerqueira JJ, Pego JM, Taipa R, Bessa JM, Almeida OF, Sousa N. 2005. Morphological correlates of corticosteroid-induced changes in prefrontal cortex-dependent behaviors. J Neurosci 25:7792– 7800. Cerqueira JJ, Mailliet F, Almeida OF, Jay TM, Sousa N. 2007. The prefrontal cortex as a key target of the maladaptive response to stress. J Neurosci 27:2781–2787. Conboy L, Bisaz R, Markram K, Sandi C. 2008. Role of NCAM in emotion and learning. Neurochem Res 10.1007/s11064-0089601-4. Conrad CD, Galea LA, Kuroda Y, McEwen BS. 1996. Chronic stress impairs rat spatial memory on the Y maze, and this effect is blocked by tianeptine pretreatment. Behav Neurosci 110:1321–1334. Cremer H, Lange R, Christoph A, Plomann M, Vopper G, Roes J, Brown R, Baldwin S, Kraemer P, Scheff S, Barthels D, Rajewsky K, Wille W. 1994. Inactivation of the N-CAM gene in mice results in size reduction of the olfactory bulb and deficits in spatial learning. Nature 367:455–459.
69
Cremer H, Chazal G, Goridis C, Represa A. 1997. NCAM is essential for axonal growth and fasciculation in the hippocampus. Mol Cell Neurosci 8:323–335. de Bruin JP, Sanchez-Santed F, Heinsbroek RP, Donker A, Postmes P. 1994. A behavioural analysis of rats with damage to the medial prefrontal cortex using the Morris water maze: Evidence for behavioural flexibility, but not for impaired spatial navigation. Brain Res 652:323–333. de Kloet ER, Joels M, Holsboer F. 2005. Stress and the brain: From adaptation to disease. Nat Rev Neurosci 6:463–475. Dias-Ferreira E, Sousa JC, Melo I, Morgado P, Mesquita AR, Cerqueira JJ, Costa RM, Sousa N. 2009. Chronic stress causes frontostriatal reorganization and affects decision-making. Science 325: 621–625. Ditlevsen DK, Kolkova K. 2008. Signaling pathways involved in NCAM-induced neurite outgrowth. Neurochem Res 10.1007/ s11064-008-9768-8. Doyle E, Nolan PM, Bell R, Regan CM. 1992. Intraventricular infusions of anti-neural cell adhesion molecules in a discrete posttraining period impair consolidation of a passive avoidance response in the rat. J Neurochem 59:1570–1573. Duman RS, Monteggia LM. 2006. A neurotrophic model for stressrelated mood disorders. Biol Psychiatry 59:1116–1127. Eckhardt M, Bukalo O, Chazal G, Wang L, Goridis C, Schachner M, Gerardy-Schahn R, Cremer H, Dityatev A. 2000. Mice deficient in the polysialyltransferase ST8SiaIV/PST-1 allow discrimination of the roles of neural cell adhesion molecule protein and polysialic acid in neural development and synaptic plasticity. J Neurosci 20: 5234–5244. Foley AG, Hartz BP, Gallagher HC, Ronn LC, Berezin V, Bock E, Regan CM. 2000. A synthetic peptide ligand of neural cell adhesion molecule (NCAM) IgI domain prevents NCAM internalization and disrupts passive avoidance learning. J Neurochem 74: 2607–2613. Fuchs E, Flugge G, Ohl F, Lucassen P, Vollmann-Honsdorf GK, Michaelis T. 2001. Psychosocial stress, glucocorticoids, and structural alterations in the tree shrew hippocampus. Physiol Behav 73: 285–291. Garrett JE, Wellman CL. 2009. Chronic stress effects on dendritic morphology in medial prefrontal cortex: Sex differences and estrogen dependence. Neuroscience 162:195–207. Hartz BP, Ronn LC. 2008. NCAM in long-term potentiation and learning. Neurochem Res 10.1007/s11064-008-9820-8. Hartz BP, Sohoel A, Berezin V, Bock E, Scheel-Kruger J. 2003. A synthetic peptide ligand of NCAM affects exploratory behavior and memory in rodents. Pharmacol Biochem Behav 75:861–867. Heim C, Nemeroff CB. 1999. The impact of early adverse experiences on brain systems involved in the pathophysiology of anxiety and affective disorders. Biol Psychiatry 46:1509–1522. Herrero AI, Sandi C, Venero C. 2006. Individual differences in anxiety trait are related to spatial learning abilities and hippocampal expression of mineralocorticoid receptors. Neurobiol Learn Mem 86:150–159. Hildebrandt H, Muhlenhoff M, Gerardy-Schahn R. 2008. Polysialylation of NCAM. Neurochem Res 10.1007/s11064-008-9724-7. Huang Q, Liu H, Zhu H, Zhou JN. 2008. Castration had no effect on decreased expression of the neural cell adhesion molecule in the prefrontal cortex of rats subjected to chronic mild stress. Int J Clin Exp Med 1:310–318. Hubschmann MV, Skladchikova G. 2008. The role of ATP in the regulation of NCAM function. Neurochem Res 10.1007/s11064008-9769-7. Kesner RP, Hopkins RO. 2006. Mnemonic functions of the hippocampus: A comparison between animals and humans. Biol Psychol 73:3–18. Kiselyov VV. 2008. NCAM and the FGF-receptor. Neurochem Res 10.1007/s11064-008-9666-0. Hippocampus
70
BISAZ ET AL.
Kiselyov VV, Skladchikova G, Hinsby AM, Jensen PH, Kulahin N, Soroka V, Pedersen N, Tsetlin V, Poulsen FM, Berezin V, Bock E. 2003. Structural basis for a direct interaction between FGFR1 and NCAM and evidence for a regulatory role of ATP. Structure 11: 691–701. Kiss JZ, Muller D. 2001. Contribution of the neural cell adhesion molecule to neuronal and synaptic plasticity. Rev Neurosci 12:297– 310. Klementiev B, Novikova T, Novitskaya V, Walmod PS, Dmytriyeva O, Pakkenberg B, Berezin V, Bock E. 2007. A neural cell adhesion molecule-derived peptide reduces neuropathological signs and cognitive impairment induced by Abeta25–35. Neuroscience 145:209– 224. Kolkova K. 2008. Biosynthesis of NCAM. Neurochem Res 10.1007/ s11064-008-9773-y. Kuipers SD, Trentani A, Den Boer JA, Ter Horst GJ. 2003. Molecular correlates of impaired prefrontal plasticity in response to chronic stress. J Neurochem 85:1312–1323. Li S, Wang C, Wang MW, Murakami Y, Matsumoto K. 2006. Impairment of the spatial learning and memory induced by learned helplessness and chronic mild stress. Pharmacol Biochem Behav 83: 186–193. Li S, Wang C, Wang W, Dong H, Hou P, Tang Y. 2008. Chronic mild stress impairs cognition in mice: From brain homeostasis to behavior. Life Sci 82:934–942. Lopez-Fernandez MA, Montaron MF, Varea E, Rougon G, Venero C, Abrous DN, Sandi C. 2007. Upregulation of polysialylated neural cell adhesion molecule in the dorsal hippocampus after contextual fear conditioning is involved in long-term memory formation. J Neurosci 27:4552–4561. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. 1951. Protein measurement with the Folin phenol reagent. J Biol Chem 193:265– 275. Luine V, Villegas M, Martinez C, McEwen BS. 1994. Repeated stress causes reversible impairments of spatial memory performance. Brain Res 639:167–170. Lynch MA, Voss KL. 1991. Presynaptic changes in long-term potentiation: Elevated synaptosomal calcium concentration and basal phosphoinositide turnover in dentate gyrus. J Neurochem 56:113–118. Madani R, Kozlov S, Akhmedov A, Cinelli P, Kinter J, Lipp HP, Sonderegger P, Wolfer DP. 2003. Impaired explorative behavior and neophobia in genetically modified mice lacking or overexpressing the extracellular serine protease inhibitor neuroserpin. Mol Cell Neurosci 23:473–494. Maness PF, Schachner M. 2007. Neural recognition molecules of the immunoglobulin superfamily: Signaling transducers of axon guidance and neuronal migration. Nat Neurosci 10:19–26. Markram K, Gerardy-Schahn R, Sandi C. 2007a. Selective learning and memory impairments in mice deficient for polysialylated NCAM in adulthood. Neuroscience 144:788–796. Markram K, Lopez Fernandez MA, Abrous DN, Sandi C. 2007b. Amygdala upregulation of NCAM polysialylation induced by auditory fear conditioning is not required for memory formation, but plays a role in fear extinction. Neurobiol Learn Mem 87: 573–582. Mazure CM, Kincare P, Schaffer CE. 1995. DSM-III-R Axis IV: Clinician reliability and comparability to patients’ reports of stressor severity. Psychiatry 58:56–64. McEwen BS. 2000. Effects of adverse experiences for brain structure and function. Biol Psychiatry 48:721–731. McEwen BS. 2002. Sex, stress and the hippocampus: Allostasis, allostatic load and the aging process. Neurobiol Aging 23:921–939. McEwen BS. 2005. Glucocorticoids, depression, and mood disorders: Structural remodeling in the brain. Metabolism 54(5, Suppl 1):20– 23. McEwen BS. 2007. Physiology and neurobiology of stress and adaptation: Central role of the brain. Physiol Rev 87:873–904. Hippocampus
McEwen BS, Lasley EN. 2003. Allostatic load: When protection gives way to damage. Adv Mind Body Med 19:28–33. Mileusnic R, Rose SP, Lancashire C, Bullock S. 1995. Characterisation of antibodies specific for chick brain neural cell adhesion molecules which cause amnesia for a passive avoidance task. J Neurochem 64: 2598–2606. Mori M, Heuss C, Gahwiler BH, Gerber U. 2001. Fast synaptic transmission mediated by P2X receptors in CA3 pyramidal cells of rat hippocampal slice cultures. J Physiol 535 (Part 1):115–123. Muller D, Wang C, Skibo G, Toni N, Cremer H, Calaora V, Rougon G, Kiss JZ. 1996. PSA-NCAM is required for activity-induced synaptic plasticity. Neuron 17:413–422. Neiiendam JL, Kohler LB, Christensen C, Li S, Pedersen MV, Ditlevsen DK, Kornum MK, Kiselyov VV, Berezin V, Bock E. 2004. An NCAM-derived FGF-receptor agonist, the FGL-peptide, induces neurite outgrowth and neuronal survival in primary rat neurons. J Neurochem 91:920–935. Nielsen J, Kulahin N, Walmod PS. 2008. Extracellular protein interactions mediated by the neural cell adhesion molecule, NCAM: Heterophilic interactions between NCAM and cell adhesion molecules, extracellular matrix proteins, and viruses. Neurochem Res 10.1007/ s11064-008-9761-2. Palkovits M. 1973. Isolated removal of hypothalamic or other brain nuclei of the rat. Brain Res 59:449–450. Panicker AK, Buhusi M, Thelen K, Maness PF. 2003. Cellular signalling mechanisms of neural cell adhesion molecules. Front Biosci 8: d900–d911. Pariante CM, Lightman SL. 2008. The HPA axis in major depression: Classical theories and new developments. Trends Neurosci 31:464–468. Park CR, Campbell AM, Diamond DM. 2001. Chronic psychosocial stress impairs learning and memory and increases sensitivity to yohimbine in adult rats. Biol Psychiatry 50:994–1004. Pham K, Nacher J, Hof PR, McEwen BS. 2003. Repeated restraint stress suppresses neurogenesis and induces biphasic PSA-NCAM expression in the adult rat dentate gyrus. Eur J Neurosci 17:879– 886. Poltorak M, Frye MA, Wright R, Hemperly JJ, George MS, Pazzaglia PJ, Jerrels SA, Post RM, Freed WJ. 1996. Increased neural cell adhesion molecule in the CSF of patients with mood disorder. J Neurochem 66:1532–1538. Ronn LC, Olsen M, Ostergaard S, Kiselyov V, Berezin V, Mortensen MT, Lerche MH, Jensen PH, Soroka V, Saffell JL, Doherty P, Poulsen FM, Bock E, Holm A. 1999. Identification of a neuritogenic ligand of the neural cell adhesion molecule using a combinatorial library of synthetic peptides. Nat Biotechnol 17:1000–1005. Rougon G, Hobert O. 2003. New insights into the diversity and function of neuronal immunoglobulin superfamily molecules. Annu Rev Neurosci 26:207–238. Rutishauser U. 2008. Polysialic acid in the plasticity of the developing and adult vertebrate nervous system. Nat Rev Neurosci 9:26–35. Sairanen M, O’Leary OF, Knuuttila JE, Castren E. 2007. Chronic antidepressant treatment selectively increases expression of plasticity-related proteins in the hippocampus and medial prefrontal cortex of the rat. Neuroscience 144:368–374. Sandi C. 2004. Stress, cognitive impairment and cell adhesion molecules. Nat Rev Neurosci 5:917–930. Sandi C, Bisaz R. 2007. A model for the involvement of neural cell adhesion molecules in stress-related mood disorders. Neuroendocrinology 85:158–176. Sandi C, Loscertales M. 1999. Opposite effects on NCAM expression in the rat frontal cortex induced by acute vs. chronic corticosterone treatments. Brain Res 828:127–134. Sandi C, Touyarot K. 2006. Mid-life stress and cognitive deficits during early aging in rats: Individual differences and hippocampal correlates. Neurobiol Aging 27:128–140. Sandi C, Merino JJ, Cordero MI, Touyarot K, Venero C. 2001. Effects of chronic stress on contextual fear conditioning and the
NCAM MEDIATES STRESS-INDUCED COGNITIVE IMPAIRMENTS hippocampal expression of the neural cell adhesion molecule, its polysialylation, and L1. Neuroscience 102:329–339. Sandi C, Davies HA, Cordero MI, Rodriguez JJ, Popov VI, Stewart MG. 2003a. Rapid reversal of stress induced loss of synapses in CA3 of rat hippocampus following water maze training. Eur J Neurosci 17:2447–2456. Sandi C, Merino JJ, Cordero MI, Kruyt ND, Murphy KJ, Regan CM. 2003b. Modulation of hippocampal NCAM polysialylation and spatial memory consolidation by fear conditioning. Biol Psychiatry 54:599–607. Scholey AB, Rose SP, Zamani MR, Bock E, Schachner M. 1993. A role for the neural cell adhesion molecule in a late, consolidating phase of glycoprotein synthesis six hours following passive avoidance training of the young chick. Neuroscience 55:499–509. Schuster T, Krug M, Stalder M, Hackel N, Gerardy-Schahn R, Schachner M. 2001. Immunoelectron microscopic localization of the neural recognition molecules L1, NCAM, and its isoform NCAM180, the NCAM-associated polysialic acid, beta1 integrin and the extracellular matrix molecule tenascin-R in synapses of the adult rat hippocampus J Neurobiol 49:142–158. Secher T, Berezin V, Bock E, Glenthoj B. 2008. Effect of an NCAM mimetic peptide FGL on impairment in spatial learning and memory after neonatal phencyclidine treatment in rats. Behav Brain Res 199:288–297. Senkov O, Sun M, Weinhold B, Gerardy-Schahn R, Schachner M, Dityatev A. 2006. Polysialylated neural cell adhesion molecule is involved in induction of long-term potentiation and memory acquisition and consolidation in a fear-conditioning paradigm. J Neurosci 26:10888–109898. Shin KY, Won BY, Heo C, Kim HJ, Jang DP, Park CH, Kim S, Kim HS, Kim YB, Lee HG, Lee SH, Cho ZH, Suh YH. 2009. BT-11 improves stress-induced memory impairments through increment of glucose utilization and total neural cell adhesion molecule levels in rat brains. J Neurosci Res 87:260–268. Squire LR, Stark CE, Clark RE. 2004. The medial temporal lobe. Annu Rev Neurosci 27:279–306. Stewart MG, Davies HA, Sandi C, Kraev IV, Rogachevsky VV, Peddie CJ, Rodriguez JJ, Cordero MI, Donohue HS, Gabbott PL, Popov VI. 2005. Stress suppresses and learning induces plasticity in CA3 of rat hippocampus: A three-dimensional ultrastructural study of
71
thorny excrescences and their postsynaptic densities. Neuroscience 131:43–54. Stork O, Welzl H, Cremer H, Schachner M. 1997. Increased intermale aggression and neuroendocrine response in mice deficient for the neural cell adhesion molecule (NCAM). Eur J Neurosci 9: 1117–1125. Stork O, Welzl H, Wolfer D, Schuster T, Mantei N, Stork S, Hoyer D, Lipp H, Obata K, Schachner M. 2000. Recovery of emotional behaviour in neural cell adhesion molecule (NCAM) null mutant mice through transgenic expression of NCAM180. Eur J Neurosci 12:3291–3306. Touyarot K, Sandi C. 2002. Chronic restraint stress induces an isoform-specific regulation on the neural cell adhesion molecule in the hippocampus. Neural Plast 9:147–159. Touyarot K, Venero C, Sandi C. 2004. Spatial learning impairment induced by chronic stress is related to individual differences in novelty reactivity: Search for neurobiological correlates. Psychoneuroendocrinology 29:290–305. Varea E, Blasco-Ibanez JM, Gomez-Climent MA, Castillo-Gomez E, Crespo C, Martinez-Guijarro FJ, Nacher J. 2007. Chronic fluoxetine treatment increases the expression of PSA-NCAM in the medial prefrontal cortex. Neuropsychopharmacology 32:803– 812. Venero C, Tilling T, Hermans-Borgmeyer I, Schmidt R, Schachner M, Sandi C. 2002. Chronic stress induces opposite changes in the mRNA expression of the cell adhesion molecules NCAM and L1. Neuroscience 115:1211–1219. Venero C, Herrero AI, Touyarot K, Cambon K, Lopez-Fernandez MA, Berezin V, Bock E, Sandi C. 2006. Hippocampal up-regulation of NCAM expression and polysialylation plays a key role on spatial memory. Eur J Neurosci 23:1585–1595. Walmod PS, Kolkova K, Berezin V, Bock E. 2004. Zippers make signals: NCAM-mediated molecular interactions and signal transduction. Neurochem Res 29:2015–2035. Wiedenmayer CP. 2004. Adaptations or pathologies? Long-term changes in brain and behavior after a single exposure to severe threat. Neurosci Biobehav Rev 28:1–12. Wright RL, Conrad CD. 2008. Enriched environment prevents chronic stress-induced spatial learning and memory deficits. Behav Brain Res 187:41–47.
Hippocampus
HIPPOCAMPUS 21:72–80 (2011)
Irradiation Enhances Hippocampus-Dependent Cognition in Mice Deficient in Extracellular Superoxide Dismutase Jacob Raber,1,2,3* Laura Villasana,1 Jenna Rosenberg,1 Yani Zou,4,5 Ting Ting Huang,4,5 and John R. Fike6,7,8
ABSTRACT: The effects of ionizing irradiation on the brain are associated with oxidative stress. While oxidative stress following irradiation is generally viewed as detrimental for hippocampal function, it might have beneficial effects as part of an adaptive or preconditioning response to a subsequent challenge. Here we show that in contrast to what is seen in wild-type mice, irradiation enhances hippocampusdependent cognitive measures in mice lacking extracellular superoxide dismutase. These outcomes were associated with genotype-dependent effects on measures of oxidative stress. When cortices and hippocampi were analyzed for nitrotyrosine formation as an index of oxidative stress, the levels were chronically elevated in mice lacking extracellular superoxide dismutase. However, irradiation caused a greater increase in nitrotyrosine levels in wild-type mice than mice lacking extracellular superoxide dismutase. These paradoxical genotype-dependent effects of irradiation on measures of oxidative stress and cognitive function underscore potential beneficial effects associated with chronic oxidative stress if it exists prior to a secondary insult such as irradiation. V 2009 C
Wiley-Liss, Inc.
KEY WORDS: water maze
hippocampus; EC-SOD; irradiation; fear conditioning;
INTRODUCTION Radiation-induced brain injury is a dose-limiting factor during therapeutic irradiation of the brain (Tofilon and Fike, 2000). While overt tissue injury generally occurs only after relatively high doses, there is a
1
Department of Behavioral Neuroscience, Oregon Health and Science University, Portland, Oregon; 2 Department of Neurology, Oregon Health and Science University, Portland, Oregon; 3 Division of Neuroscience ONPRC, Oregon Health and Science University, Portland, Oregon; 4 Department of Neurology and Neurological Sciences, Stanford University, Stanford, California; 5 GRECC, VA Palo Alto Health Care System, Palo Alto, California; 6 Brain and Spinal Injury Center, University of California, San Francisco, San Francisco, California; 7 Department of Neurological Surgery, University of California, San Francisco, San Francisco, California; 8 Department of Radiation Oncology, University of California, San Francisco, San Francisco, California Grant sponsor: NIH; Grant numbers: R01 NS46051, R01 AG24400; Grant sponsor: Alzheimer’s Association; Grant number: IIRG-05-14021; Grant sponsor: NASA; Grant number: NNJ05HE63G. *Correspondence to: Jacob Raber, Department of Behavioral Neuroscience, L470, Oregon Health and Science University, 3181SW Sam Jackson Park Road, Portland, OR 97239. E-mail:
[email protected] Accepted for publication 22 September 2009 DOI 10.1002/hipo.20724 Published online 17 December 2009 in Wiley Online Library (wileyonlinelibrary.com). C 2009 V
WILEY-LISS, INC.
strong likelihood of developing adverse reactions in terms of cognitive decline after relatively lower doses (Meyers and Brown, 2006). Such impairment has a diverse character, and in humans and animals often includes hippocampus-dependent functions involving learning, memory, and spatial information processing (Abayomi, 1996; Raber et al., 2004; Rola et al., 2004; Fan et al., 2007). The underlying mechanisms responsible for radiation-induced cognitive impairment have remained elusive, but important possibilities include alterations in the neurogenic cell populations in the dentate gyrus (DG) (Mizumatsu et al., 2003; Raber et al., 2004; Rola et al., 2004), loss of mature neurons in the DG (Fan et al., 2007), alterations in NMDA subunits (Shi et al., 2006), genetic risk factors (Villasana et al., 2006), and reductions in the immediate early gene Arc (Rosi et al., 2008). Additionally, radiation-induced reductions in cognitive performance may be associated with changes in the microenvironment (Rola et al., 2004; Fike et al., 2007), including oxidative stress, which can regulate the fate of neurogenic cells associated with cognitive function (Smith et al., 2000; Limoli et al., 2004; Fike et al., 2007). The redox environment is of particular importance in the central nervous system (CNS), where there is a relatively high rate of oxygen consumption and metabolic turnover (Lewen et al., 2000) and a relatively low level of endogenous antioxidants (Peuchen et al., 1997). There are several pathways that mitigate the physiological and pathological effects of reactive oxygen species (ROS) like superoxide, in mammalian cells (Riley, 1994). One of these pathways involves the superoxide dismutase (SOD) enzymes, which are critical elements of the cellular antioxidant defense mechanism (Muscoli et al., 2003). The SODs are oxidoreductases that remove superoxide by catalyzing the dismutation of the superoxide radical to hydrogen peroxide. Hydrogen peroxide is then metabolized to molecular oxygen and water by catalase or glutathione peroxidase. There are three different SOD isoforms that catalyze the same chemical reaction, but have different enzymatic properties and distinct subcellular localizations: CuZnSOD (SOD1) is localized in the cytoplasm, MnSOD (SOD2) is localized in the mitochondria, and extracellular SOD (EC-SOD, SOD3) is extracellular. Although the physiological roles of the
IRRADIATION ENHANCES HIPPOCAMPUS-DEPENDENT COGNITION IN MUTANT MICE SOD isoforms in mammalian cells are not completely understood, the extracellular isoform was shown to be associated with certain cognitive functions (Levin et al., 1998; Thiels et al., 2000), and alterations in EC-SOD expression using transgenic or mutant mice impaired learning (Levin et al., 1998). The mechanisms associated with this effect are not yet understood, but likely involve superoxide, which has been shown to have both positive and negative effects (Serrano and Klann, 2004; Hu et al., 2007; Kamsler et al., 2007; Kishida and Klann, 2007). In rodents, superoxide is a necessary signaling component of long-term potentiation (LTP), which is a widely studied form of synaptic plasticity (Hu et al., 2006, 2007). The source of superoxide is yet to be determined, but may include mitochondrial metabolism, monoamine oxidase and cyclo-oxygenase, nitric oxide synthase, and NADPH oxidase (Kishida and Klann, 2007). Regardless of the source and relevant targets of superoxide signaling (e.g., kinases, phosphatases), clearly ROS are important molecules involved in the regulation of learning and memory (Kishida and Klann, 2007). One way to characterize the extent to which ROS affects cognition, and if it is modulated by ionizing irradiation, is to use mutant mice that lack a specific antioxidant molecule, like an SOD isoform. A persistent level of oxidative stress in EC-SOD deficient mice (i.e., knock out, KO mice) is associated with a lower baseline level of neurogenesis relative to wild type (WT) mice (Rola et al., 2007). However, a modest dose of X-rays has no effect on neurogenesis in KO mice, but causes a highly significant reduction in WT animals (Rola et al., 2007). These data indicate that while oxidative stress can be generally viewed as detrimental for neurogenesis, it might also have a beneficial effect, at least in the context of an adaptive or preconditioning response to a subsequent challenge such as irradiation. The present study was performed to determine if such a beneficial effect would be seen in the context of cognitive performance and whether such an effect would be associated with genotypedependent effects on measures of oxidative stress.
73
by the OHSU IACUC and VA Palo Alto IACUC. Mice were maintained in a temperature- and light-controlled environment with a 12 h light/dark cycle and were provided food and water ad libitum.
Irradiation Mice were anesthetized (i.p., 80 mg kg21 ketamine (Sigma, St. Louis, MO) and 20 mg kg21 xylazine (Sigma)) and shamirradiated (controls) or irradiated at 2 months of age with a dose of 10 Gy using a Mark 1 Cesium Irradiator (J.L. Shepherd and Associates, San Fernando, CA). The cerebellum, eyes, and body were shielded with lead [see (Acevedo et al., 2008b) for details on the irradiation set up]. The 10 Gy dose used here was selected because it was shown to induce hippocampus-dependent cognitive impairments (Raber et al., 2004; Villasana et al., 2006). After irradiation, mice for the behavioral study were group-housed until 3 days prior to the behavioral testing. Mice for the western blot analyses were killed by cervical dislocation at 3 months of age.
Behavioral Testing Cognitive assessments began 3 months following irradiation or sham-irradiation, and consisted of 4 weeks of testing both hippocampus-dependent and hippocampus-independent cognitive performance. Three different hippocampus-dependent tests were included to increase the ability to detect potential genotype-dependent effects of irradiation on hippocampus-dependent cognitive performance. Because irradiation effects on the brain might not be limited to hippocampal function, nonhippocampus-dependent cognitive tests were also included. Mice were tested for hippocampus-dependent novel location and hippocampus-independent novel object recognition during the first week, for hippocampus-dependent spatial learning and memory in the water maze in the second week, for sensorimotor function on the rotorod in the third week, and for hippocampus-dependent contextual and hippocampus-independent cued fear conditioning in the last week. The individual researcher testing the mice was blinded to genotype and radiation treatment.
MATERIALS AND METHODS Novel location and novel object recognition tasks Mice Congenic EC-SOD KO mice (Carlsson et al., 1995) on the C57BL/6J (B6) background were initially obtained from Dr. James Crapo at the National Jewish Medical and Research Center, Denver, CO. The colony was maintained by backcrossing to B6 mice purchased from the Jackson Laboratory (Bar Harbor, ME). Two-month-old male homozygous KO mice (n 5 14 for the behavioral studies and n 5 4–5 for the western blot analysis) and their wild type (WT) littermate controls (n 5 16 for the behavioral study and n 5 3–4 for the western blot analysis), generated from the intercross of heterozygous KO mice, were used in the current study. All animal handling procedures were done in accordance with Federal guidelines and approved
To assess object recognition, mice were individually habituated for three consecutive days to a 16 3 16 in. open-field with clear plexiglass walls (Kinder Scientific, Poway, CA) for 5 min. On the fourth day, the mice were first given three 10min trials with three plastic objects in different corners of the open field. In subsequent trials, the familiar objects were exchanged with replicas. For the fourth 10-min trial, one of the familiar objects was moved from one corner of the field to another to evaluate hippocampus-dependent novel location recognition. For the fifth 10-min trail, a familiar object was replaced by a novel object to assess hippocampus-independent novel object recognition. There was a 3-min interval between each trial. During this time, the mice were placed back in their Hippocampus
74
RABER ET AL.
home cage and the open field and the objects were cleaned with 5% acetic acid to remove potential odors. The total time spent exploring all objects was compared between trials to assess the familiarization of each mouse with the objects (Acevedo et al., 2008b). The difference between the percent time spent exploring the object in the novel location (Trial 4) and the percent time spent exploring the same object in its original location (Trial 3) was calculated to assess novel location recognition. The percent time spent exploring the novel object during Trial 5 was calculated to assess novel object recognition. The percent time spent exploring the objects were then compared based on genotype and treatment.
Water maze The water maze test was used to assess hippocampus-dependent spatial learning and memory. A circular pool (diameter 140 cm) was filled with opaque water (248C) and mice were trained to locate a submerged platform (luminescence: 200 lux). To determine if irradiation affected the ability to swim or learn the water maze task, mice were first trained to locate a clearly marked platform (visible platform, Days 1 and 2). Mice were subsequently trained to locate the platform when it was hidden beneath the surface of opaque water (Days 3–5). Training during the hidden platform sessions (acquisition) required the mice to learn the location of the hidden platform based on extra-maze cues. For both visible and hidden sessions, there were two daily sessions, morning and afternoon, which were 2-h apart. Each session consisted of three trials (with 10-min intertrial intervals). A trial ended when the mice located the platform. Mice that failed to locate the platform within 60 s were led to the platform by placing a finger in front of their swim path. Mice were taken out of the pool after they were physically on the platform for a minimum of 3 s. During visible platform training, the platform was moved to a different quadrant of the pool for each session. For the hidden platform training, the platform location was kept constant. Mice were placed into the water facing the edge of the pool in one of nine randomized locations. The start location was changed for each trial. The swimming patterns of the mice were recorded with the Noldus Ethovision video tracking system (Ethovision XT, Noldus Information Technology, Wageningen, Netherlands) set at six samples/s. The time to locate the platform (latency) was used as a measure of performance for the visible and hidden sessions. Because swim speeds can influence the time it takes to reach the platform, they were also analyzed to assess if there were genotype or treatment differences in this measure. To measure spatial memory retention, probe trials (platform removed) were conducted 1 h after the last hidden trial of each mouse on each day of hidden platform training (i.e., a total of three probe trials). The time spent in the target quadrant, the quadrant where the platform was previously located during hidden platform training, was compared to the time spent in the three nontarget quadrants. For the probe trials, mice were Hippocampus
placed into the water in the quadrant opposite from the target quadrant.
Rotorod To exclude the possibility that radiation-induced alterations in motor function could negatively impact performance in the cognitive tests, all mice were tested on the rotorod (Rotamex-5, Columbus Instruments, Columbus, OH). Mice were placed on an elevated rod (3 cm 3 9.5 cm spindle 44.5-cm elevated) initially rotating at 5 rpm. The speed of the rotating rod was increased by 1 rpm every 3 s to a maximum of 24 rpm. Each trial ended when a fall was recorded by photo beams aligned with each individual mouse or if a mouse did not fall from the rod within 300 s. Mice received 3 trials each day, 30-min apart, for three consecutive days.
Conditioned fear Hippocampal function was also assessed using the contextual fear conditioning task. In this task, mice learned to associate the environmental context (fear conditioning chamber) with a mild foot shock (unconditioned stimulus, US). Because contextual fear conditioning is hippocampus- and amygdaladependent, the mild foot shock was also paired with a tone (conditioned stimulus, CS) to allow assessment of cued fear conditioning, which is amygdala- but not hippocampusdependent. When mice were re-exposed to the context or the tone, conditioned fear resulted in freezing behavior. Mice displayed this conditioned fear by ceasing all movement except for respiration (i.e., freezing). On Day 1, each mouse was placed in a fear conditioning chamber (Kinder Scientific, Poway, CA) and allowed to explore for 2 min before delivery of a 30-s tone (80 db) which was immediately followed by a 2-s foot shock (0.6 mA). Two minutes later, a second CS-US pair was delivered. On Day 2 each mouse was first placed in the fear conditioning chamber containing the exact same context, but there was no administration of a tone or foot shock. Freezing was analyzed for 3 min. One hour later, the mice were placed in a new context (containing a different odor, cleaning solution, floor texture, walls and shape) where they were allowed to explore for 3 min before being reexposed to the fear conditioning tone and freezing was assessed for an additional 3 min. Freezing was measured using a Noldus Ethovision video tracking system.
Nitrotyrosine Analysis To determine the level of oxidative stress in KO and WT mice at baseline and after irradiation, cortices and hippocampi were dissected on ice and processed as described (Zou et al., 2009). A mouse monoclonal antibody (clone 1A6, Upstate/ Millipore, Billerica, MA) was used as the primary antibody at a 1:2,000 dilution and HRP-tagged goat antimouse IgG (BioRad, Hercules, CA) was used as the secondary antibody at a 1:10,000 dilution. Chemiluminescence signals from western blots were captured either by X-ray film or by the Typhoon
IRRADIATION ENHANCES HIPPOCAMPUS-DEPENDENT COGNITION IN MUTANT MICE
75
system (GE Healthcare, Piscataway, NJ). Actin was used as an internal control for slight variations in protein loading and transferring. As tissues may contain low levels of nitrotyrosine, a nitrotyrosine immunoblotting control (Upstate/Millipore), which contains proteins that were nitrated with peroxynitrite, was used as a positive control for the primary antibody. Gel images were analyzed by Image J using the gel analysis function. Protein bands that belonged to the same group (the upper and lower nitrotyrosine bands or b-actin) were analyzed together using the Gels/Plot Lanes function. The area under each peak was subsequently calculated as a measure for the band intensity. Because the actual band intensities were different for different proteins and thresholds for the two groups of protein bands (the upper and lower nitrotyrosine bands or b-actin) were automatically adjusted by Image J, the generated measures reflect the relative differences in protein levels within the same group of samples. The band intensity obtained for each upper and lower nitrotyrosine band was then normalized to that of its corresponding b-actin to obtain relative expression levels.
Statistical Analyses Data were assessed for normality and homogeneity of variance to determine whether to use parametric or nonparametric statistical tests. All statistical analyses were performed using SPSS software (SPSS, Chicago, IL) or GraphPad Prism software (San Diego, CA). A two-way (2 3 2) AVOVA was used with genotype and treatment as between-subject factors to measure the effects of genotype and irradiation on novel location recognition, novel object recognition, and contextual and cued fear conditioning. To determine whether each group showed novel location recognition, a paired t-test was used to compare the percent time exploring the object in the new location (Trial 4) vs. the old location (Trial 3). To determine whether each group showed novel object recognition, a one-way ANOVA (within each group) was used to compare the percent time exploring the novel and familiar objects and, when appropriate, a Neuman-Keuls post hoc test was used. For the fear conditioning tasks, an unpaired t-test was used to compare between-group performances. For the water maze learning curves, visible and hidden platform sessions were analyzed separately. To compare water maze and rotorod learning curves in the different groups, a three-way repeated measures ANOVA was used with genotype and irradiation treatment as between-subject factors and session number as a within-subject factor. To compare mean water maze latency during the visible and hidden sessions in the different groups, a two-way ANOVA with genotype and irradiation treatment as between-subject factors was also used, and this analysis was followed using a Bonferroni post hoc test when appropriate. In the water maze probe trials, one-way ANOVAs were used for each group to assess spatial memory retention by comparing the percent time spent in each quadrant and when appropriate, a Neuman-Keuls post hoc test was used to compare the percent time spent in the target quadrant and the three nontarget quadrants. For an overall analysis of the western blot data, a three-way repeated measures ANOVA
FIGURE 1. Effects of 137Cs irradiation on object recognition of WT and KO mice. A. While sham-irradiated and irradiated WT mice and irradiated KO mice showed novel location recognition, sham-irradiated KO mice did not. *P < 0.05, paired t-test. B. Irradiation impaired novel object recognition in both WT and KO mice. *P < 0.05 novel object vs. familiar objects 1 and 2, one-way ANOVA, Dunnett’s post hoc test. WT, n 5 8 each and KO, n 5 7 each.
was used with treatment, genotype, and brain region as between-subjects factors and the upper and lower band as within-subject factor. T-tests were used to assess potential genotype differences in the upper band, lower band, or combined upper and low band for each treatment condition and brain region. Data were expressed as means 6 SEM unless otherwise noted. P < 0.05 was considered significant for all tests.
RESULTS There was no effect of irradiation on novel location recognition. Both sham-irradiated (t 5 2.3, P < 0.05) and irradiated WT (t 5 2.7, P < 0.05) mice showed novel location recognition (Fig. 1A). In contrast, while irradiated KO mice showed novel location recognition (t 5 2.1, P < 0.05), sham irradiated KO mice did not (t 5 0.7, P > 0.05) (Fig. 1A). In the hippocampus-independent novel object recognition task, sham-irradiated WT and sham-irradiated KO mice showed novel object recognition (sham-irradiated WT: ANOVA object preference, F(2,18) 5 6.9, P < 0.01; sham-irradiated KO: F(2,12) 5 6.3, P < 0.01). They showed greater preference for the novel object than the two familiar objects. In contrast, irradiated WT and KO mice showed no novel object recognition (Fig. 1B). Hippocampus
76
RABER ET AL.
In the water maze task, there were no significant group differences in swim speeds during the visible platform sessions (WT sham: 15.04 6 0.44 cm s21; WT irradiated: 15.81 6 0.46 cm s21; KO sham: 15.77 6 0.46 cm s21; KO irradiated: 16.03 6 0.35 cm s21). Therefore, time to reach the platform (latency) was used as performance measure. All groups showed improvement in the visible (F(3,78) 5 81.04, P < 0.001) and hidden platform sessions (F(5, 130) 5 7.87, P < 0.001) and there were no session by genotype or session by treatment interactions. During the visible, but not during the hidden, platform sessions, there was a genotype by treatment interaction for the average latency across the sessions (ANOVA genotype 3 treatment interaction, F(1, 26) 5 9.53, P < 0.01). Therefore, we assessed the effects of treatment in the two genotypes separately. There was an effect of irradiation in the WT mice (ANOVA effect of treatment, F(1,14) 5 9.6, P < 0.01) not seen in KO mice (sham-irradiated: 24.8 6 3.1 s; irradiated: 21.4 6 2.3 s). The average time to reach the platform across the visible platform sessions was lower in sham-irradiated (17.3 6 2.5 s) than irradiated (21.0 6 2.5 s) WT mice (Bonferroni test, P < 0.01), indicating that irradiation impaired task learning in WT mice. However, irradiation did not affect spatial learning to locate the hidden platform in either genotype (data not shown). In the first probe trial of the water maze test, sham-irradiated and irradiated WT mice and sham-irradiated KO mice did not show spatial memory retention (Fig. 2), as they did not spend more time searching in the target quadrant than in any other quadrant (ANOVA, quadrant within group analysis). In contrast, irradiated KO mice showed spatial memory retention (ANOVA quadrant preference, F(3,20) 5 13.65, P < 0.001) and spent most of their time searching in the target quadrant (Neuman-Keuls, P < 0.01 target vs. any other quadrant; Fig. 2). In the probe trial following an additional day of hidden platform training (Probe 2), shamirradiated and irradiated WT mice and irradiated KO mice showed spatial memory retention but sham-irradiated KO mice still did not (ANOVA quadrant preference, F(3, 16) 5 1.1; P 5 0.38); Fig. 2). Following a third day of hidden platform training (Probe 3), all groups showed spatial memory retention and spent more time searching in the target quadrant than in any other quadrant (Fig. 2). When sensorimotor function was assessed on the rotorod, all groups of mice improved their performance with training (ANOVA, effect of session: F(2,52) 5 24.56, P < 0.001) and there were no effects of genotype, treatment, or genotype by treatment interactions. Finally, hippocampal function was assessed using the contextual fear conditioning task. There were no significant effects of genotype or treatment on % baseline freezing (first 2 min prior to delivery of tone on Day 1) (WT sham: 10.6 6 0.9; WT irradiated: 11.6 6 1.6; KO sham: 9.7 6 1.7; KO irradiated: 10.6 6 1.3). However, there were opposing genotypedependent effects of irradiation on hippocampus-dependent contextual freezing (ANOVA, genotype 3 treatment interaction, F(1,22) 5 13.6, P < 0.01, Fig. 3A). Compared to shamirradiated WT mice, irradiated WT mice showed impairments in contextual fear conditioning (t 5 2.6, P < 0.05). In contrast, Hippocampus
FIGURE 2. Effects of 137Cs irradiation on water maze performance of WT and KO mice. C. Spatial memory retention of shamirradiated and irradiation WT and KO mice in the first 30 s of the water maze probe trials. In the probe trial following the first day of hidden platform training (Probe 1), only irradiated KO mice spent more time searching in the target quadrant than any other quadrant. In the probe trial following the second day of hidden platform training (Probe 2), sham-irradiated and irradiated WT mice and irradiated KO mice spent more time searching in the target quadrant than any other quadrant. In the probe trial following the third day of hidden platform training (Probe 3), all groups spent more time searching in the target quadrant than any other quadrant. *P < 0.05 target vs. any other quadrant, ANOVA and Newman-Keuls post hoc test. WT, n 5 8 each and KO, n 5 7 each.
compared to sham-irradiated KO mice, irradiation enhanced contextual fear conditioning in the KO mice (t 5 2.6, P < 0.05). As a result, irradiated KO mice showed significantly more contextual freezing than irradiated WT mice (t 5 3.9, P < 0.01). All groups of mice showed cued fear conditioning (paired t-test, Pre-CS vs. Post-CS within each group, P < 0.05) and there was no effect of genotype or treatment and no significant interaction (Fig. 3B). To determine whether these cognitive changes were associated with genotype-dependent effects on measures of oxidative stress, nitrotyrosine western blot analyses were used (Fig. 4).
IRRADIATION ENHANCES HIPPOCAMPUS-DEPENDENT COGNITION IN MUTANT MICE
FIGURE 3. Effects of 137Cs irradiation on fear conditioning of WT and KO mice. A. While irradiation impaired contextual fear conditioning of WT mice, it enhanced contextual fear conditioning of KO mice. *P < 0.05 vs. sham-irradiated genotype matched mice. **P < 0.01 between irradiated WT and irradiated KO mice. B. Sham-irradiated and irradiated WT and KO showed robust and comparable hippocampus-independent cued fear conditioning. WT, n 5 8 each and KO, n 5 7 each.
There were two main nitrotyrosine bands of !26 and 52 kDa (Fig. 4, top left panel). In cortex and hippocampus, the levels of both bands were higher in sham-irradiated KO than WT mice (Fig. 4, lower panels). There was an effect of treatment (F 5 6.822, P 5 0.015), a treatment 3genotype interaction (F 5 17.214, P < 0.0,001), and a treatment 3 genotype 3 brain region interaction (F(2,52) 5 6.909, P 5 0.014). In the cortex, irradiation increased the levels of the lower and the upper band and the combined levels in WT mice. In contrast, no increase in these levels was seen in KO mice. In the hippocampus, irradiation increased the combined levels in WT but not in KO mice. When the individual bands were analyzed, irradiation increased the levels of the upper band in both genotypes and the intensity of the lower band only in WT, but not in KO mice.
DISCUSSION The primary finding of the current study is that while EC-SOD deficiency is associated with some hippocampus-dependent impairments prior to irradiation, a finding also shown
77
by others (Levin et al., 1998), after irradiation, hippocampusdependent cognitive measures are enhanced in mice lacking EC-SOD. This paradoxical effect supports the idea that ROS can have both positive and negative effects, depending upon the circumstances (Levin et al., 1998; Kamsler and Segal, 2003; Valko et al., 2007). While the precise mechanism(s) responsible for the ‘‘protective’’ type of response seen here is (are) not yet known, in a general sense this effect resembles a preconditioning (Gori and Forconi, 2005; Pespeni et al., 2005), adaptive [reviewed in (Yu and Chung, 2006)], or inducible-like radioprotective response (Qutob et al., 2006), where a sublethal or potentially injurious stimulus (e.g., oxidative stress) induces tolerance to a subsequent and potentially more damaging insult (e.g., irradiation). For instance, ischemic/hypoxic preconditioning has been shown to protect the brain under some conditions (Murry et al., 1986; Kitagawa et al., 1991; Chen et al., 2003; for review Liu et al., 2009) but those effects can change with age (for review Schaller, 2007). Additionally, the same EC-SOD KO mice were shown to be more sensitive to focal cerebral ischemia injury (Sheng et al., 1999). These and other data highlight the complexity of preconditioning or adaptive responses and highlight the fact that such responses may be context-dependent. The relationship between EC-SOD and processes involved in learning and memory is not simple, with both over and under expression reported to impair learning, (Levin et al., 1998; Hu et al., 2006). Whether this represents differences in superoxide regulation, differing levels of hydrogen peroxide, components of the nitric oxide (NO) pathway and bio-availability of NO, or alterations in stress signaling pathways, needs to be determined. In the present study we used EC-SOD KO mice, which show indications of persistent oxidative stress but which do not display any compensatory changes in levels or activities of the other SOD isoforms, catalase or glutathione peroxidase (Rola et al., 2007). Prior to irradiation these mice showed some cognitive impairments, i.e., hippocampus-dependent novel location recognition, but for other cognitive measures there were no differences observed between sham-irradiated WT and sham-irradiated KO mice. This indicated that compared to WT mice, the EC-SOD phenotype was relatively subtle, which might relate to the fact that ROS might have both positive and negative effects on brain function. The nitrotyrosine analyses revealed two predominant nitrated proteins, of about 26 and 52 kDa, respectively. As increased levels of nitrated proteins have been shown in various neurodegenerative disorders (Castegna et al., 2003; Sacksteder et al., 2006), future efforts are warranted to identify these two proteins. In this study we used a battery of cognitive tests to address hippocampus-dependent function. Inclusion of multiple hippocampus-dependent cognitive tests is particularly important in comparative studies of wild-type and mutant mice, because the tests differ in the amount of training, complexity, and motivation, and might differ in their sensitivity for detecting potential detrimental or beneficial effects of irradiation. Additionally, the ability to detect potential effects of irradiation may depend Hippocampus
78
RABER ET AL.
FIGURE 4. Nitrotyrosine western blot analyses of hippocampal and cortical extracts of sham-irradiated and irradiated WT and KO mice. Upper left panel, a representative western blot image showing the upper and lower nitrotyrosine-positive bands at !52 and 26 kDa, respectively. Chemiluminescence signals from western blots were captured either by X-ray film or by the Typhoon system and b-actin was used as an internal control. A nitrotyrosine immunoblotting control (Lane C) containing proteins
nitrated with peroxynitrite was used as a positive control. The band intensity obtained for each upper (bottom left panel) and lower (bottom right panel) nitrotyrosine band was normalized to that of its corresponding b-actin to obtain relative expression levels. Combined relative expression levels of the upper and lower band (upper right panel) were also calculated. *P < 0.05 vs. treatment matched WT. NS, not significant. n 5 3–4 WT and n 5 4–5 KO mice.
upon the specific level of functioning under baseline conditions. That is, detrimental effects of irradiation may be more likely revealed when the cognitive function at baseline is intact while beneficial effects of irradiation may be more likely revealed when the cognitive function at baseline is impaired. In the present study, WT mice showed novel location recognition both before and after irradiation. This suggested that the novel location recognition test was not very sensitive in detecting detrimental effects of gamma irradiation in male mice. The situation is different for female mice, however, where it was shown that 10 Gy of 137Cs irradiation impaired novel location recognition in WT mice and those lacking apolipoprotein E (Acevedo et al., 2008a). On the other hand, the novel location recognition paradigm used here was sensitive enough to detect the beneficial effects of irradiation in mutant male mice not showing novel location recognition under baseline conditions
(Fig. 1A). That is, sham-irradiated KO mice showed impairments in this test, which is consistent with impairments in hippocampus-dependent learning and memory of EC-SOD KO mice reported by others (Levin et al., 1998), but KO mice did show novel location recognition after irradiation. Thus, this measure of hippocampal function was fully recovered following radiation exposure. These data indicate that the sensitivity of the novel location recognition test to detect effects of cranial irradiation is critically influenced by genetic and environmental factors. We also included a hippocampus-independent version of the object recognition test. In the present investigation, both shamirradiated WT and sham-irradiated KO mice showed hippocampus-independent novel object recognition (Fig. 1B), that is, they spent more time exploring the novel object. In both genotypes, this behavior was impaired by irradiation, indicating that
Hippocampus
IRRADIATION ENHANCES HIPPOCAMPUS-DEPENDENT COGNITION IN MUTANT MICE the effects of radiation on cognitive function were not limited to the hippocampus but also involved the cortex, and that preexisting and persistent oxidative stress did not impact this response. To the best of our knowledge, this is the first time that effects of gamma irradiation on hippocampus-independent novel object recognition have been reported. These data, together with the water maze data discussed below, also highlight the importance of including both hippocampus-dependent and nonhippocampus-dependent versions of cognitive tests in the evaluation of radiation effects on brain function. This study involved the assessment of spatial learning and memory using the Morris water maze, a test we have used previously after irradiation of adult WT (Raber et al., 2004) and mutant mice (Villasana et al., 2006). In our earlier study in WT male mice irradiated at 2 months of age, this test involved a single probe trial performed after 3 days of testing, and under that testing paradigm, radiation did not have any discernible effects (Raber et al., 2004). The current results confirmed this finding in that there were no differences between sham-irradiated and irradiated WT mice when the probe trial was done after the third day of hidden platform training (Fig. 2). Similar results were seen in EC-SOD KO mice. However, when probe trial performance was analyzed after the first and second day of testing, there were significant differences between sham-irradiated and irradiated KO mice, and in the probe trial following the first day of hidden platform training the irradiated KO mice clearly outperformed sham-irradiated WT mice (Fig. 2). These data, together with our studies in mutant and WT female mice (Villasana et al., 2006; Acevedo et al., 2008b), highlight the strength of including multiple probe trials in the design of the water maze to detect effects of irradiation, particularly in mutant mice. Thus, a water maze paradigm including only one probe trial at the end of 3 days of hidden platform training would not have revealed the genotype-dependent effects of irradiation in the current study. Given the complexities associated with cognitive function and the differential sensitivities of cognitive tests to detect effects of irradiation, we also used contextual fear conditioning to assess hippocampus-dependent emotional learning and memory. Sham-irradiated and irradiated mice showed genotypedependent contextual fear conditioning. While contextual fear conditioning was impaired in WT mice following irradiation, it was enhanced in KO mice (Fig. 3A). These data show that the contextual fear conditioning test is particularly sensitive to detect effects of cranial irradiation in WT and mutant male mice. Sham-irradiated and irradiated WT and KO both showed robust and comparable hippocampus-independent cued fear conditioning, suggesting that the effects of irradiation on contextual fear conditioning are hippocampus-dependent and not due to general impairments in fear conditioning, which is more amygdala-based. With regard to the overall radiation response, it was particularly striking that in contrast to what is seen in WT mice, performance in hippocampus-dependent novel location recognition, spatial memory retention in the first and second water maze probe trials, and contextual fear conditioning was
79
enhanced in animals deficient in EC-SOD. This improvement was associated with genotype-dependent effects on measures of oxidative stress. In cortex and hippocampus, nitrotyrosine levels were chronically elevated in KO mice but irradiation caused a greater increase in oxidative stress in wild-type mice than in mice lacking extracellular superoxide dismutase. In addition to measures of oxidative stress, genotype-dependent effects of hippocampal neurogenesis following irradiation might have contributed to the cognitive performance. In our earlier neurogenesis study that did not involve behavioral assessments, baseline neurogenesis was lower in KO than WT mice but following irradiation, neurogenesis was strongly reduced in WT, but not in KO mice (Rola et al., 2007). Thus the net results are that measures of oxidative stress are lower and hippocampal neurogenesis is higher in irradiated KO than WT mice. Whether or not measures of oxidative stress and/or neurogenesis play a causal or contributory role in the ‘‘protection’’ against radiation-induced cognitive impairments or the enhanced cognitive performance of irradiated mice when compared to shamirradiated mice needs to be established. A new animal model that involves the conditional expression of EC-SOD now exists that may provide a novel way to address this idea (Zou et al., 2009). Given that radiation affected the hippocampus-independent novel object recognition cognitive measure in WT as well as KO mice, it seems unlikely that neurogenesis alone is causally responsible for the improved cognitive performance seen here. Therefore, to understand the functional effects seen here, other avenues also need to be explored, including, perhaps, how specific molecular markers associated with learning and memory (Rosi et al., 2008) are affected in animals deficient in EC-SOD. While the molecular mechanisms underlying the opposing genotype-dependent effects of irradiation reported here are not yet known, increased efforts are warranted to study potential beneficial effects associated with chronic oxidative stress in the context of a secondary insult such as irradiation. Ultimately understanding how such an effect develops may provide new insight into the evolution of cognitive injury after irradiation and provide information useful for the development of new approaches for the management of radiation-induced brain injury.
REFERENCES Abayomi OK. 1996. Pathogenesis of irradiation-induced cognitive dysfunction. Acta Oncol 35:659–663. Acevedo SE, McGinnis G, Raber J. 2008a. Effects of 137Cs gamma irradiation on cognitive performance and measures of anxiety in Apoe-/- and wild-type female mice. Radiat Res 170:422–428. Acevedo SF, Tittle S, Raber J. 2008b. Transgenic expression of androgen receptors improves spatial memory retention in both sham-irradiated and 137Cs gamma-irradiated female mice. Radiat Res 170:572–578. Carlsson LM, Jonsson J, Edlund T, Marklund SL. 1995. Mice lacking extracellular superoxide dismutase are more sensitive to hyperoxia. Proc Natl Acad Sci USA 92:6264–6268. Hippocampus
80
RABER ET AL.
Castegna A, Thongboonkerd V, Klein JB, Lynn B, Markesberry WR, Butterfield A. 2003. Proteomic identification of nitrated proteins in Alzheimer’s disease brain. J Neurochem 85:1394–1401. Chen J, Graham SH, Zhu RL, Simon RP. 2003. Stress proteins and tolerance to focal cerebral ischemia. J Cereb Blood Flow Metab 16:566–577. Fan Y, Liu Z, Weinstein PR, Fike JR, Liu J. 2007. Environmental enrichment enhances neurogenesis and improves functional outcome after cranial irradiation. Eur J Neurosci 25:38–46. Fike JR, Rola R, Limoli CL. 2007. Radiation response of neural precursor cells. Neurosurg Clin N Am 18:115–27. Gori T, Forconi S. 2005. The role of reactive free radicals in ischemic preconditioning—Clinical and evolutionary implications. Clin Hemorheol Microcirc 33:19–28. Hu D, Serrano F, Oury TD, Klann E. 2006. Aging-dependent alterations in synaptic plasticity and memory in mice that overexpress extracellular superoxide dismutase. J Neurosci 26:3933–3941. Hu D, Cao P, Thiels E, Chu CT, Wu GY, Oury TD, Klann E. 2007. Hippocampal long-term potentiation, memory, and longevity in mice that overexpress mitochondrial superoxide dismutase. Neurobiol Learn Mem 87:372–384. Kamsler A, Segal M. 2003. Paradoxical actions of hydrogen peroxide on long-term potentiation in transgenic superoxide dismutase-1 mice. J Neurosci 23:10359–10367. Kamsler A, Avital A, Greenberger V, Segal M. 2007. Aged SOD overexpressing mice exhibit enhanced spatial memory while lacking hippocampal neurogenesis. Antioxid Redox Signal 9:181–189. Kishida KT, Klann E. 2007. Sources and targets of reactive oxygen species in synaptic plasticity and memory. Antioxid Redox Signal 9:233–244. Kitagawa K, Matsumoto M, Kuwabara K, Tagaya M, Ohtsuli T, Hata R, Ueda H, Handa N, Kimura K, Kamada T. 1991. Ischemic tolerance phenomenon detected in various brain regions. Brain Res 561:203–211. Levin ED, Brady TC, Hochrein EC, Oury TD, Jonsson LM, Marklund SL, Crapo JD. 1998. Molecular manipulations of extracellular superoxide dismutase: Functional importance for learning. Behav Genet 28:381–390. Lewen A, Matz P, Chan PH. 2000. Free radical pathways in CNS injury. J Neurotrauma 17:871–890. Limoli CL, Giedzinski E, Rola R, Otsuka S, Palmer TD, Fike JR. 2004. Radiation response of neural precursor cells: Linking cellular sensitivity to cell cycle checkpoints, apoptosis and oxidative stress. Radiat Res 161:17–27. Liu X-Q, Sheng R, Qin Z-H. 2009. The neuroprotective mechanism of brain ischemic preconditioning. Acta Pharmacol Sin (advanced online publication, July 20) 30:1071–1080. Meyers CA, Brown PD. 2006. Role and relevance of neurocognitive assessment in clinical trials of patients with CNS tumors. J Clin Oncol 24:1305–1309. Mizumatsu S, Monje ML, Morhardt DR, Rola R, Palmer TD, Fike JR. 2003. Extreme sensitivity of adult neurogenesis to low doses of x-irradiation. Can Res 63:4021–4027. Murry CE, Jennings RB, Reimer KA. 1986. Preconditioning with ischemia: A delay of lethal cell injury in ischemic myocardium. Circulation 74:1124–1136. Muscoli C, Cuzzocrea S, Riley DP, Zweier JL, Thiemermann C, Wang ZQ, Salvemini D. 2003. On the selectivity of superoxide dismutase mimetics and its importance in pharmacological studies. Br J Pharmacol 140:445–460. Pespeni M, Hodnett M, Pittet JF. 2005. In vivo stress preconditioning. Methods 35:158–164. Peuchen S, Bolanos JP, Heales SJ, Almeida A, Duchen MR, Clark JB. 1997. Interrelationships between astrocyte function, oxidative stress
Hippocampus
and antioxidant status within the central nervous system. Prog Neurobiol 52:261–281. Qutob SS, Multani AS, Pathak S, McNamee JP, Bellier PV, Liu QY, Ng CE. 2006. Fractionated X-radiation treatment can elicit an inducible-like radioprotective response that is not dependent on the intrinsic cellular X-radiation resistance/sensitivity. Radiat Res 166:590–599. Raber J, Rola R, LeFevour A, Morhardt DR, Curley J, Mizumatsu S, VandenBerg SR, Fike JR. 2004. Radiation-induced cognitive impairments are associated with changes in indicators of hippocampal neurogenesis. Radiat Res 162:39–47. Riley PA. 1994. Free radicals in biology: Oxidative stress and the effects of ionizing radiation. Int J Radiat Biol 65:27–33. Rola R, Raber J, Rizk A, Otsuka S, VandenBerg SR, Morhardt DR, Fike JR. 2004. Radiation-induced impairment of hippocampal neurogenesis is associated with cognitive deficits in young mice. Exp Neurol 188:316–330. Rola R, Zou Z, Huang T-T, Fishman K, Baure J, Rosi S, Milliken H, Limoli CL, Fike JR. 2007. Lack of EC-SOD in the microenvironment impacts radiation-induced changes in neurogenesis. Free Rad Biol Med 42:1133–1145. Rosi S, Andres-Mach M, Fishman KM, Levy W, Ferguson RA, Fike JR. 2008. Cranial irradiation alters the behaviorally induced immediate-early gene arc (activity-regulated cytoskeleton-associated protein). Cancer Res 68:9763–9770. Sacksteder CA, Qian W-J, Knyushko TV, Wang H, Chin MH, Lacan G, Melega WP, Camp IIDG, Smith RD, Smith DJ, Squier TC, Bigelow DJ. 2006. Endogenously nitrated proteins in mouse brain: links to neurodegenerative disease. Biochemistry 45:8009–8022. Schaller BJ. 2007. Influence of age on stroke and preconditioninginduced ischemic tolerance in the brain. Exp Neurol 205:9–19. Serrano F, Klann E. 2004. Reactive oxygen species and synaptic plasticity in the aging hippocampus. Ageing Res Rev 3:431–443. Sheng H, Brady TC, Pearlstein RD, Crapo JD, Warner DS. 1999. Extracellular superoxide dismutase deficiency worsens outcome from focal cerebral ischemia in the mouse. Neurosci Lett 267: 13–16. Shi L, Adams MM, Long A, Carter CC, Bennett C, Sonntag WE, Nicolle MM, Robbins M, D’Agostino R, Brunso-Bechtold JK. 2006. Spatial learning and memory deficits after whole-brain irradiation are associated with changes in NMDA receptor subunits in the hippocampus. Radiat Res 166:892–899. Smith J, Ladi E, Mayer-Proschel M, Noble M. 2000. Redox state is a central modulator of the balance between self-renewal and differentiation in a dividing glial precursor cell. Proc Natl Acad Sci USA 97:10032–10037. Thiels E, Urban NN, Gonzalez-Burgos GR, Kanterewicz BI, Barrionuevo G, Chu CT, Oury TD, Klann E. 2000. Impairment of long-term potentiation and associative memory in mice that overexpress extracellular superoxide dismutase. J Neurosci 20:7631– 7639. Tofilon PJ, Fike JR. 2000. The radioresponse of the central nervous system: A dynamic process. Radiat Res 153:357–370. Valko M, Leibfritz D, Moncol J, Cronin MT, Mazur M, Telser J. 2007. Free radicals and antioxidants in normal physiological functions and human disease. Int J Biochem Cell Biol 39:44–84. Villasana L, Acevedo S, Poage C, Raber J. 2006. Sex- and APOE isoform-dependent effects of radiation on cognitive function. Radiat Res 166:883–891. Yu BP, Chung HY. 2006. Adaptive mechanisms to oxidative stress during aging. Mech Ageing Dev 127:436–443. Zou Y, Chen CH, Fike JR, Huang TT. 2009. A new mouse model for temporal- and tissue-specific control of extracellular superoxide dismutase. Genesis 47:142–154.
HIPPOCAMPUS 21:81–92 (2011)
Phosphacan and Receptor Protein Tyrosine Phosphatase b Expression Mediates Deafferentation-Induced Synaptogenesis Janna L. Harris, Thomas M. Reeves, and Linda L. Phillips* ABSTRACT: This study documents the spatial and temporal expression of three structurally related chondroitin sulfated proteoglycans (CSPGs) during synaptic regeneration induced by brain injury. Using the unilateral entorhinal cortex (EC) lesion model of adaptive synaptogenesis, we documented mRNA and protein profiles of phosphacan and its two splice variants, full length receptor protein tyrosine phosphatase b (RPTPb) and the short transmembrane receptor form (sRPTPb), at 2, 7, and 15 days postlesion. We report that whole hippocampal sRPTPb protein and mRNA are persistently elevated over the first two weeks after UEC. As predicted, this transmembrane family member was localized adjacent to synaptic sites in the deafferented neuropil and showed increased distribution over that zone following lesion. By contrast, whole hippocampal phosphacan protein was not elevated with deafferentation; however, its mRNA was increased during the period of sprouting and synapse formation (7d). When the zone of synaptic reorganization was sampled using molecular layer/granule cell (ML/GCL) enriched dissections, we observed an increase in phosphacan protein at 7d, concurrent with the observed hippocampal mRNA elevation. Immunohistochemistry also showed a shift in phosphacan distribution from granule cell bodies to the deafferented ML at 2 and 7d postlesion. Phosphacan and sRPTPb were not colocalized with glial fibrillary acid protein (GFAP), suggesting that reactive astrocytes were not a major source of either proteoglycan. While transcript for the developmentally prominent full length RPTPb was also increased at 2 and 15d, its protein was not detected in our adult samples. These results indicate that phosphacan and RPTPb splice variants participate in both the acute degenerative and long-term regenerative phases of reactive synaptogenesis. These results suggest that increase in the transmembrane sRPTPb tyrosine phosphatase activity is critical to this plasticity, and that local elevation of extracellular phosphacan influences dendritic organization during synaptogenesis. V 2009 Wiley-Liss, Inc. C
KEY WORDS: entorhinal lesion; synaptic plasticity; proteoglycan; dentate gyrus; gene expression
INTRODUCTION Functional recovery after traumatic brain injury requires axonal sprouting and synaptic reorganization. There is increasing evidence that these processes are regulated by molecules within the extracellular enviDepartment of Anatomy and Neurobiology, School of Medicine, Virginia Commonwealth University Medical Center, Richmond, Virginia Grant sponsor: NIH, State of Virginia; Grant numbers: NS-044372, NS057758, NS-056247, CNI 07-302F. *Correspondence to: Dr. Linda L. Phillips, Department of Anatomy and Neurobiology, PO Box 980709, School of Medicine, Virginia Commonwealth University Medical Center, Richmond, VA 23298. E-mail:
[email protected] Accepted for publication 21 September 2009 DOI 10.1002/hipo.20725 Published online 15 December 2009 in Wiley Online Library (wileyonlinelibrary.com). C 2009 V
WILEY-LISS, INC.
ronment of the brain. The brain extracellular matrix (ECM) is enriched in proteoglycans, particularly CSPGs. As a group, CSPGs were initially thought to inhibit axon growth and plasticity both in vitro and in vivo (Snow et al., 1990; Oohira et al., 1991; Grumet et al., 1993; Davies et al., 1999). They are also major components of the inhibitory glial scar which forms after lesions to the brain or spinal cord (reviewed in Properzi et al., 2003). Published evidence shows that enzymes which degrade chondroitin sulfated-glycosaminoglycans (GAGs) may enhance axon regeneration and synaptic plasticity (Moon et al., 2001; Bradbury et al., 2002; Pizzorusso et al., 2002; Huang et al., 2006). More recently, focal cortical contusion was reported to differentially affect CSPG expression and reduce inhibitory proteoglycans in regions bordering the lesion core, potentially fostering local plasticity (Harris et al., 2009). By contrast, some CSPG family members are elevated after brain injury and appear to interact in a positive way with cell adhesion molecules and soluble growth factors to enhance axonal sprouting (Faissner et al., 1994; Sakurai et al., 1997; Bicknese et al., 1994; Schafer et al., 2008). Together, these data show that CSPG interactions may be complex, potentially both inhibitory and supportive of regenerative plasticity. The CSPG phosphacan and its related splice variants are one proteoglycan group whose members may play a supportive role in neuronal plasticity processes. Phosphacan (6B4 proteoglycan, or DSD-1 proteoglycan in the mouse) is a secreted alternative splice variant of the full length receptor protein tyrosine phosphatase b (RPTPb), a transmembrane receptor with intracellular tyrosine phosphatase activity (Maurel et al., 1994). A third form, the short receptor form sRPTPb, lacks the extracellular membrane-proximal sequence but retains the intracellular phosphatase activity. Secreted phosphacan and transmembrane sRPTPb are the two most prominent forms found in adult brain (Sakurai et al., 1996; Dobbertin et al., 2003). Produced by both neurons and glia (Snyder et al., 1996; Hayashi et al., 2005), phosphacan is a major component of brain matrix and can inhibit or promote axon growth, depending on the neuronal lineage (Garwood et al., 1999). This splice variant is secreted into the ECM and is distributed around synaptic junctions, but absent from the synaptic active zone (Miyata et al., 2004). Relative to other CSPGs (e.g., neurocan, versican, NG2) which may be up-
82
HARRIS ET AL.
regulated after CNS trauma (Asher et al., 2000; Asher et al., 2002; Morgenstern et al., 2002; Jones et al., 2003; Schafer et al., 2008), phosphacan may vary significantly after injury. For example, it is reduced following spinal cord injury (Jones et al., 2003; Tang et al., 2003), cortical stab injury (Dobbertin et al., 2003) and filter implant-induced glial scarring (McKeon et al., 1999). However, others report increased phosphacan after experimental stroke (Carmichael et al., 2005) and fimbria/ fornix lesion (Snyder et al., 1996). While one recent study (Schafer et al., 2008) reported no change in phosphacan mRNA expression after entorhinal deafferentation lesion, a systematic analysis of phosphacan splice variants during reactive synaptogenesis has not been made. Ultrastructural studies confirm that membrane-type full length RPTPb and sRPTPb expression are associated with postsynaptic dendrites and spines (Hayashi et al., 2005). Other reports suggest a role for RPTPb in the production of long term potentiation (LTP) within the mature brain, a function which is consistent with synaptic localization. RPTPb knockout mice display age-dependent abnormalities in hippocampal LTP, as well as impaired spatial learning and contextual fear conditioning (Niisato et al., 2005; Tamura et al., 2006). A number of intracellular targets for the RPTPb phosphatase have been identified (Kawachi et al., 1999; Meng et al., 2000; Pariser et al., 2005; Tamura et al., 2006), many of which direct neurite morphogenesis and synapse regeneration or stabilization (e.g., b-catenin, b-adducin, PSD-95, p190 RhoGAP). Together, these results suggest that RPTPb can influence synaptic plasticity, possibly through multiple molecular pathways. Clearly, expression of CSPGs like phosphacan and RPTPb is altered with CNS trauma and may act to regulate the subsequent axonal sprouting and synaptic recovery. In this study we document the spatial and temporal expression of phosphacan, RPTPb and sRPTPb during synaptic reorganization induced by unilateral EC lesion (UEC). UEC is a well-characterized rodent model of synaptic plasticity (Steward et al., 1988), generating robust synaptogenesis within the dentate gyrus during the first two weeks postlesion. This time course permits examination of mRNA and protein for the different phosphacan splice variants during periods of terminal degeneration (2d), afferent sprouting and synapse formation (7d) and synapse stabilization (15d). We report differential mRNA and protein expression for phosphacan and RPTPb splice variants over the first two weeks postinjury, supporting a role for these CSPGs in both the degenerative and regenerative phases of reactive synaptogenesis.
MATERIALS AND METHODS Experimental Animals Adult male Sprague-Dawley rats (300–390 g) were used in this study. Rats were randomly divided into three experimental Hippocampus
groups: 2d (n 5 17), 7d (n 5 23), and 15d (n 5 12) survival. Rats were housed in pairs within individual cages having food and water ad libitum, and subjected to a 12 h dark-light cycle at 228C. All protocols for injury and use of animals were approved by the Institutional Animal Care and Use Committee of Virginia Commonwealth University.
UEC Lesion Rats were subjected to UEC after the method of Loesche and Steward (1977). All animals were surgically prepared under isoflurane anesthesia delivered via nose cone (2% in carrier gas of 70% N2O, 30% O2). During all surgical procedures, body temperature was maintained at 378C via a thermostatically controlled heating pad (Harvard Apparatus). Rats were placed in a stereotaxic frame, and an area of skull was removed above the entorhinal cortex of the right hemisphere. Lesion current was passed through a Teflon-insulated wire electrode angled at 108 from vertical. Current was delivered (1.5 mA for 30 s) at a total of nine stereotaxic sites: 1.5 mm anterior to the transverse sinus, 3, 4, and 5 mm lateral to midline; and at 2, 4, and 6 mm ventral to the brain surface. After lesions were completed, the electrode was removed, the scalp was sutured closed over the surgical site and Bacitracin applied to the wound. Animals were monitored during full recovery from anesthesia and returned to their home cages.
Protein Extraction and Western Blotting Two rapid dissection procedures for ipsilateral and contralateral samples were performed. Whole hippocampi were removed from one subset of animals at 2d (n 5 5), 7d (n 5 9), or 15d (n 5 6) postlesion, and a dentate molecular layer enriched fraction was removed from a second subset of rats at 2d (n 5 3) and 7d (n 5 4) postlesion. For the whole hippocampal tissue, a serial extraction protocol was adapted from Dobbertin et al. (2003) in order to separate soluble extracellular phosphacan from transmembrane RPTPb and sRPTPb. Each hippocampus was homogenized in 1.75 ml detergent-free extraction buffer (50 mM Tris, 150 mM NaCl, 40 mM Na Acetate) containing protease inhibitors (Roche complete cocktail plus 2 lg/ml Pepstatin). Homogenates were centrifuged for 20 min at 100,000g and 48C. Supernatant was removed and stored at 2808C as the saline soluble fraction. Pellets were re-homogenized in 1.75 ml extraction buffer containing 1% Triton X-100, and agitated for 1 h at 48C prior to centrifugation for 20 min at 100,000g and 48C. Supernatant was removed and stored at 2808C as the detergent soluble fraction. In order to expose antigenic sites and maximize efficiency of antibody recognition, all protein samples were treated with chondroitinase ABC (chABC; Seikagaku America) to remove chondroitin sulfate GAG chains from proteoglycans. Aliquots of 540 ll for each sample were removed, buffered to pH 8 by addition of 400 mM Tris, and mixed with an additional protease inhibitor (Roche complete cocktail plus 2 lg/ml Pepstatin). The resulting preparation (600 ll volume) was incubated with 0.3 U chABC for 3 h, shaking at 378C. Reaction was stopped by returning samples to
PHOSPHACAN AND REACTIVE SYNAPTOGENESIS 2808C. Protein concentration was determined in aliquots from the treated samples using spectrophotometry (Shimadzu UV160; Shimadzu Scientific Instruments). For blot preparation, samples were heat denatured in XT sample buffer (Bio-Rad Laboratories), either 5 lg (phosphacan probe) or 18 lg (RPTPb probe) separated on 3–8% gels (Bio-Rad Laboratories, Hercules, CA) before transfer to PVDF membrane. Membranes were blocked in 5% milk-TBST (Tris buffered saline containing 0.05% Tween 20) for 1 h before being probed with either mouse phosphacan (3F8; Developmental Studies Hybridoma Bank, University of Iowa) or mouse RPTPb (BD Biosciences, San Jose, CA) antibody diluted in milk-TBST (1.5 lg/ml) overnight at 48C. This RPTPb antibody targets the c terminus common to both short and full length proteins (Levy et al., 1993), thereby recognizing each RPTPb isoform. Blots were subsequently washed with milk-TBST and then incubated for 1 h at room temperature in peroxidase conjugated goat antimouse secondary antibody (1:20,000; Rockland; Gilbertsville, PA) prior to a final washing in TBST and immunopositive signal visualization using Super Signal West Dura Extended Duration Substrate (Thermo Scientific; Rockford, IL). Parallel blots were incubated without primary antibody to confirm signal specificity. All blots were then imaged digitally with the G:Box ChemiHR system for densitometric analysis using GeneSnap software (SynGene; Frederick, MD). In each case, the ipsilateral densitometric measurements were expressed as a percent of contralateral value. The second subset of animals at 2d (n 5 3) and 7d (n 5 4) postlesion were prepared for enriched molecular layer extraction. Using coronal tissue blocks, the dentate molecular layer and a portion of the adjacent granule cell lamina were excised, homogenized in a 100 ll volume of TPER (Thermo Scientific) and centrifuged for 5 min at 8,000g and 48C. Supernatant was removed and stored at 2808C. As for whole hippocampal protein extracts, 5 lg of protein was separated on Criterion XT 3–8% gels and transferred to PVDF membrane prior to probing for phosphacan using the 3F8 antibody. Immunopositive bands were visualized, digitally imaged, and densitometrically analyzed as described above for whole hippocampal blots. After primary antibody binding data was captured, all blots were stripped and reprobed for b-actin (mouse monoclonal, Sigma, St. Louis, MO; 1:3,000) as a load control. Signal was visualized as described above for the 3F8 and RPTPb antibodies and blots imaged with the same G:Box ChemiHR system. No quantitative differences in load between the lanes were detected.
Confocal Immunohistochemistry At 2d (n 5 3) or 7d (n 5 3) postlesion, groups of rats were deeply anesthetized with sodium pentobarbital (60 mg/kg, i.p.) and sacrificed by transcardiac perfusion of 0.9% saline, followed by 4% formaldehyde in 0.1 M phosphate buffer (PB), pH 7.2. Brains were removed and postfixed overnight at 48C. Coronal sections (40 lm) were collected in 0.1 M PB, and
83
immunohistochemistry (IHC) was performed for phosphacan and RPTPb using the same antibodies described above for Western blotting. In some sections, 3F8 or anti-RPTPb was applied in combination with rabbit antibody to GFAP (Dako North America Inc., Westbury, NY) to determine if astrocytes were a possible source of phosphacan or RPTPb. Similarly, anti-RPTPb was used in colocalization experiments with rabbit postsynaptic density-95 (PSD-95) antibody (Zymed Laboratories, San Francisco, CA) to probe for association between the membrane receptor form and synaptic junctions. Since the binding site for 3F8 is partially masked by GAGs attached to the core protein (Dobbertin et al., 2003), we treated brain sections with chABC prior to phosphacan immunodetection. These sections were incubated for 1 h at 378C with chABC (0.1 U/ml in Tris Acetate buffer; 100 mM Tris-HCl, 30 mM Na-Acetate, pH 8.0), then washed 3 3 10 min in phosphate buffered saline (PBS) (PBS; Bio-Rad Laboratories, Hercules, CA) before subsequent immunodetection. Free-floating sections were then blocked for 30 min in peroxidase, washed 3 3 10 min in PBS and placed in blocking buffer (fish gelatin in PBS 1 0.05% Triton X-100) for 30 min. Next, the tissue sections were incubated in primary antibody (3F8, antiphosphacan 1:100, anti-RPTPb 1:500, anti-GFAP 1:5,000, anti-PSD-95 1:500), in paired combinations overnight at 48C. After primary antibody exposure, sections were washed 3 3 10 min in PBS, blocked again for 30 min and incubated with the appropriate fluorescent secondary antibodies (Alexa 488 goat-antimouse, or Alexa 594 donkey antirabbit; Invitrogen, Carlsbad, CA) for 2h. After a final series of PBS washes, sections were mounted onto Probe On Plus glass slides (Fisher Scientific, Pittsburgh, PA), and coverslipped with Vectashield (Vector Laboratories, Burlingame, CA). Minus primary controls were processed in parallel to confirm signal specificity. Images were captured with a Leica TCS-SP2 confocal microscope for qualitative analysis of protein distribution.
EM Ultrastructural Immunocytochemistry At 7d, a subset of animals (n 5 2) were deeply anesthetized with sodium pentobarbital (60 mg/kg, i.p.) and sacrificed by transcardiac perfusion of 0.9% saline, followed by mixed aldehyde fixative (4% paraformaldehyde and 0.2% glutaraldehyde) in 0.1 M PB, pH 7.2. Brains were removed and postfixed overnight at 48C. Coronal sections (40 lm) were collected in 0.1 M PB and processed for immunocytochemistry (ICC) with anti-RPTPb antibody (BD Biosciences, San Jose, CA) as indicated above, or with mouse MAB 5,210 antibody (Chemicon/ Millipore, Billerica, MA) to detect all phosphacan splice variants. Antibody binding was visualized with DAB as described previously (Phillips et al., 1994). Tissue was then placed in 1% osmium (0.1 M phosphate buffer, pH 7.2) and processed in resin prior to being flat-embedded on plastic slides. After the plastic had cured, sample regions of mid-dorsal hippocampus containing the CA1 and dentate gyrus were excised and a series of thick and thin sections cut on an Leica EM UC6i ultramicrotome (Leica Microsystems, Wetzlar, Germany). The thin Hippocampus
84
HARRIS ET AL.
TABLE 1. Primer Pairs Designed for qRT-PCR of RPTP Splice Variants and the Housekeeping Gene Cyclophilin A Gene Phosphacan forward Phosphacan reverse Phosphacan probe RPTPb forward RPTPb reverse RPTPb probe sRPTPb forward sRPTPb reverse sRPTPb probe Cyclophilin A forward Cyclophilin A reverse Cyclophilin A probe
sections were collected on membrane-coated slotted grids and observed on a Jeol JEM-1230 electron microscope equipped with a Gatan UltraScan 4000SP CCD camera. The granule cell and molecular layers of the dentate gyrus, both ipsilateral, and contralateral to the lesion, were systematically photographed at 5–10,0003 magnification for qualitative analysis.
RNA Isolation and qRT-PCR Whole hippocampi were rapidly dissected from a subset of animals at 2d (n 5 6), 7d (n 5 5), or 15d (n 5 6) postlesion and RNA extraction was performed under nuclease-free conditions. Each hippocampus (!100 mg) was homogenized in 1 ml Trizol Reagent (Invitrogen, Carlsbad, CA), mixed with 0.2 ml chloroform, and centrifuged for 15 min at 12,000g. RNA in the upper phase was removed and precipitated with 0.5 ml isopropyl alcohol. After centrifugation for 10 min at 12,000g, supernatant was removed and RNA pellets were washed in 75% ethanol. The pellets were dissolved in PCRgrade water (Ambion) and incubated at 558C for 10 min. All samples were then given two cycles of DNAse treatment to remove residual DNA contamination using the DNA-free DNase kit (Ambion; Austin, TX), according to the manufacturer’s protocol. Briefly, samples were incubated in a mixture of DNAse buffer and DNAse 1 for 20 min at 378C, followed by 10 min at room temperature with DNAse Inactivation Reagent. After centrifugation at 10,000g for 2 min to pellet inactivation reagent, supernatants were removed and stored at 2808C. RNA concentration and integrity were first assessed for all samples with the Experion automated electrophoresis system (Bio-Rad Laboratories; Hercules, CA), after which concentration was verified by NanoDrop spectrophotometry (ND-1000, NanoDrop Products; Wilmington, DE). Subsequent electropherogram analysis (Agilent Technologies; Santa Clara, CA) indicated high quality RNA (RIN " 7.0), clear peaks for 18S and 28S rRNA and minimal sample degradation. Total RNA was prepared for each sample in PCR-grade water (20 ng/ll) Hippocampus
Oligo sequence 50 -GGGCATTCAGGAGTATCCAACA-30 50 -TCCGTGACTCTTCTATTTTTACTTTCAT-30 50 -TCAGCACATCTCGTTCTATCCCTTTGCTCA-30 50 -GCAGAGGCCAGTAATAGTAGCCAT-30 50 -TAGATGAGAATACCAACAAGAACCACTAG-30 50 -ACACGATCACAAGGGGTATAACCGCCTT-30 50 -ACAATGAGGCCAGTAATAGTAGCCAT-30 50 -TAGATGAGAATACCAACAAGAACCACTAG-30 50 -AGACACGATCACAAGGGGTATAACCGCCT-30 50 -CTGTTTGCAGACAAAGTTCCAAA-30 50 -AGGAACCCTTATAGCCAAATCCTT-30 50 -CAGCAGAAAACTTTCGTGCTCTGAGCACT-30
and submitted for TaqMan One-Step RT-PCR on the ABI prism 7900 Sequence Detection system (Applied Biosystems; Foster City, CA). Specific TaqMan primers were designed to span exon–exon boundaries using Primer Express software (Applied Biosystems; Foster City, CA). Sequences for specific primers and TaqMan probes are listed in Table 1. A reference calibrator sample of total RNA was serially diluted and used to generate a standard Cycle Threshold (CT) vs. Quantity plot for each run. In all cases, the plot was linear and the correlation coefficients were greater than 0.98. PCR reactions were performed in triplicate and relative mRNA quantity was determined as a function of the CT for each test sample. By this standard curve method, change in target RNA was expressed in arbitrary units relative to the standard calibrator. These units were used for RNA data analysis and were plotted as relative RNA Quantity. Reactions had an overall efficiency between 90–100% as determined by the slope of the standard curves (range 23.1 to 24.2), calculated as E 5 10(-1/slope) 21. Negative controls on each plate included no-amplification and no-template conditions.
Statistical Analysis Changes in gene and protein expression following UEC were evaluated by comparison of ipsilateral deafferented samples to those from the uninjured contralateral hemisphere, providing within-subject matched controls. Results were expressed as percent of control value. The significance of UEC-induced changes in RNA levels was evaluated using mixed-design analysis of variance (ANOVA) (SPSS v11, MANOVA) with survival interval as between-subjects factor and hemisphere as within-subjects factor, and evaluations at single survival intervals implemented as a priori comparisons using simple main effects (Keppel, 1991; Levine, 1991). The normality of sample data distributions was confirmed with normal probability plots prior to ANOVA analyses. The significance of densitometric values from Western blots was analyzed using the Student’s t-test.
PHOSPHACAN AND REACTIVE SYNAPTOGENESIS
85
Phosphacan and RPTPb Immunohistochemistry
FIGURE 1. Time course of whole hippocampal phosphacan and RPTPb protein expression at 2, 7, and 15d after UEC. Western blot for phosphacan (A) within the deafferented hippocampus showed no difference from control at any of the three time points. 2d and 15d n 5 6; 7d n 5 9. In (B), Western blot probe for RPTPb (antibody recognizing both full length and short transmembrane phosphatase) showed a significant increase in protein within the deafferented hippocampus at all time points when compared with contralateral controls. Representative ipsilateral (left)/ contralateral (right) gel pairs are illustrated below each time point. 2d and 15d n 5 6; 7d n 5 9; *P < 0.02, **P < 0.001.
A probability of less than 0.05 was considered statistically significant for all tests.
RESULTS Expression of Phosphacan and RPTPb Protein After UEC Phosphacan protein expression was examined in the saline fraction of hippocampal homogenates extracted without detergent and treated with chABC. A 450 kD phosphacan band was resolved (Fig. 1A), corresponding to the profile of this splice variant reported by Dobbertin et al. (2003). Overall, we found no significant change in phosphacan protein at the three postinjury intervals. By contrast, when hippocampal homogenates extracted with detergent and chABC treated were probed for total RPTPb, we observed significant increase in the 250 kD band corresponding to sRPTPb (Dobbertin et al., 2003) at 2, 7, and 15d after UEC (Fig. 1B; 116.6 6 8.2%, P < 0.05; 118.4 6 9.5% and 120.2 6 10.4%, P < 0.001). By contrast, we failed to detect measurable full length RPTPb in our Western blot samples, consistent with the reported very low levels of this splice variant in adult tissue (Sakurai et al., 1996; Dobbertin et al., 2003). Reprobe of all blots for b-actin verified equal lane loading. These results show that UEC lesion differentially affects expression of the phosphacan and sRPTPb isoforms.
To confirm tissue distribution of phosphacan and sRPTPb, we examined the cellular localization of these proteoglycans using the same antibodies applied for the Western blot experiments. Confocal IHC analysis revealed prominent ipsilateral differences in expression at 2 and 7d postlesion for phosphacan. The greatest change occurred at 2d (Figs. 2A,B), with diffuse low level staining for phosphacan evident in the contralateral molecular layer (ML), and dense aggregates of the protein surrounding the cell bodies of dentate granule cells (arrows in Fig. 2A) and within the hilar subgranular zone (SGZ in Fig. 2A). By contrast, increased phosphacan immunoreactivity was seen over the outer ML ipsilateral to UEC lesion (arrows in Fig. 2B). Notably, the strong phosphacan staining within both granule cells and the subgranular zone was reduced on the deafferented side (Fig. 2B). This result would be consistent with the absence of change in total extracted phosphacan after injury but indicates a clear lesion-induced shift in the cellular distribution of the protein. Although Western blot results showed an increase of RPTPb protein throughout the first two weeks postlesion, IHC analysis of hippocampal RPTPb revealed the strongest ML labeling at 7d postlesion, the period of robust terminal sprouting, and synaptic reorganization (Fig. 2 C,D). Although the RPTPb antibody recognizes both full length and sRPTPb, the absence of full length RPTPb signal in our Western blots suggest that our IHC signal represents primarily sRPTPb. In this case, we found punctate RPTPb staining over the entire dentate ML (contralateral distribution shown in Fig. 2C), which was visibly increased in intensity on the ipsilateral deafferented side (Fig. 2D). When phosphacan was localized with ultrastructural ICC, we observed signal in both granule cell bodies and dendrites, consistent with the distribution of these isoforms seen by confocal IHC (data not shown). To further investigate whether reactive astrocytes express phosphacan or RPTPb after UEC, we also performed colocalization experiments where antibody for each splice variant was paired with GFAP antibody (Fig. 2). No phosphacan or RPTPb immunostaining was found within the cell bodies and major processes of ML astrocytes when examined with projected z stack images (arrows in Figs. 2E,F; astrocytes identified by yellow arrows shown at higher magnification within insets). These results suggest that neurons, not astrocytes, are the predominant source for phosphacan proteoglycans within the deafferented dentate gyrus. Because the punctate confocal distribution of RPTPb suggested synaptic localization, we also performed double label confocal IHC for RPTPb and the postsynaptic density marker PSD-95 at 7d after UEC. A subset of RPTPb positive puncta were found adjacent to sites stained with PSD-95 (arrows in Figs. 3A,B), suggesting contiguous distribution of the two proteins. Moreover, the relative density and size of RPTPb puncta appeared greater in the ipsilateral deafferented ML (Fig. 3B) when compared with the contralateral side (Fig. 3A). To further investigate whether this pattern might represent pre- or postsyHippocampus
86
HARRIS ET AL.
FIGURE 2. Localization of phosphacan and RPTPb in deafferented dentate gyrus 2 and 7d after UEC. Projected z stack images show phosphacan (3F8) signal at 2d (A, B) shows predominant localization over the granule cell body layer (GCL) and subgranular zone (SGZ) of contralateral control (arrows in A). After lesion (B), phosphacan is increased in the outer molecular layer (OML, arrows) and decreased in the GCL and SGZ. By contrast, the distribution of RPTPb at 7d postlesion (C, D) is punctate and visible throughout the ML at higher density (D) than in the con-
tralateral control (C). Confocal dual labeling of phosphacan at 2d (green in E) and RPTPb at 7d (green in F) with astrocyte marker (GFAP, red) in the deafferented ML suggests that reactive astrocytes are not the principal source of either splice variant. The two markers fail to show significant overlap of signal (white arrows in panels E, F). Individual examples of reactive astrocytes (yellow arrows) are show enlarged in each inset. Minus primary controls had only low background signal (G). Bar in A-D 5 20 lm; E-G 5 40 lm; insets 5 5 lm.
naptic localization of RPTPb, we performed ultrastructural ICC on parallel 7d cases. Here RPTPb was found predominantly within dendrites (asterisk in Fig. 3C), as well as in spines and adjacent to postsynaptic profiles (arrowheads, arrows in Fig. 3C inset). In some synaptic junctions, we also found evidence for RPTPb label near vesicles within presynaptic terminals (open arrow in Fig. 3C inset), a pattern similar to that observed by Hayashi et al. (2005) in embryonic cortical cells.
observed, normalization to total RNA mass may be the most accurate approach for standardization of gene expression (Tricarico et al., 2002; Meldgaard et al., 2006). Phosphacan mRNA was increased only at 7d (113.6 6 8.7%, P < 0.05). By contrast, elevations at both 2 and 15d postlesion for full length RPTPb mRNA (122.8 6 5.3% and 122.6 6 7.8%; P < 0.001) were found, while sRPTPb mRNA was elevated at all three time points (111.7 6 10.0% and 112.8 6 5.6% for 2 and 7d, P < 0.05; 118.8 6 6.9% for 15d, P < 0.01). Interestingly, the 2–15d increase in sRPTPb transcript was consistent with time course of injury-induced elevations sRPTPb protein (Fig. 1B). On the other hand, the 7d increase in hippocampal phosphacan mRNA was not correlated with increased phosphacan protein (Fig. 1A). While expressed protein ultimately affects phenotype, changes in mRNA may independently reflect a biological response to cell perturbation. Examples of mismatch between transcript and protein expression are well documented (Mehra et al., 2003). We next investigated whether the observed differences between hippocampal phosphacan transcript and protein might be explained by subregional differen-
Expression of Phosphacan and RPTPb mRNA After UEC Specific mRNA levels of phosphacan, full length RPTPb, and sRPTPb were measured in ipsilateral deafferented and contralateral control hippocampus at 2, 7, and 15d post lesion using qRT-PCR. We elected to express our qRT-PCR results relative total RNA input (Fig. 4) because of two prior observations. First, we have shown low sample variance when these proteoglycans are normalized to total RNA (Harris et al., 2009) and second, with the <2 fold changes in mRNA we Hippocampus
PHOSPHACAN AND REACTIVE SYNAPTOGENESIS
87
ined phosphacan protein level in Western blots at both 2 and 7d postlesion using ML/GCL enriched samples. At 7d, when qRT-PCR showed elevated transcript, but no increase in whole hippocampal phosphacan, tissue samples enriched in the deafferented ML revealed a significant increase in phosphacan protein (129.9 6 4.9%; P < 0.05; Fig. 5). Further, we also found that phosphacan was not increased in 2d ML enriched samples (Fig. 1A), consistent with the absence of significant rise in hippocampal phosphacan mRNA at this time. These results are supported by the fact that, despite a predominant shift in phosphacan from granule cell layer to ML at 2d, overall tissue signal for phosphacan was similar ipsilateral and contralateral to UEC lesion (Figs. 2A,B).
DISCUSSION This study provides new detail regarding the expression of extracellular phosphacan and the transmembrane tyrosine phosphatases RPTPb and sRPTPb during the time course of reactive synaptogenesis. Overall, whole hippocampal phosphacan showed increased mRNA and protein expression during presynaptic degeneration and terminal sprouting, while the membrane bound sRPTPb transcript and protein were elevated throughout the process of synaptogenesis. Lesion effects on full length RPTPb were limited to elevated mRNA at 2 and 15d. Notably, whole hippocampal sRPTPb had a correlated rise in both transcript and protein, while phosphacan and full length RPTPb did not. Samples enriched in the deafferented ML showed a
FIGURE 3. RPTPb distribution in dentate ML at 7d after UEC lesion. Confocal IHC shows RPTPb (green) at punctate sites in the outer contralateral ML (A). These sites are more numerous and of larger size in the deafferented ML (B), where RPTPb is found adjacent to postsynaptic sites (PSD-95 positive profiles, red; arrows in A, B). Ultrastructural ICC shows RPTPb localized in dendritic profiles (asterisk) in the outer ML of a contralateral hemisphere (C). Inset in C shows RPTPb present in a spine (arrowhead), postsynaptic (arrow) and presynaptic (open arrow) sites from a section without counter stain. Bar in A, B 5 25 lm; C 5 0.5 lm; inset 5 0.2 lm.
ces in protein distribution after lesion and our tissue sampling method.
Local Increase of Phosphacan in the Deafferented Molecular Layer The time course profiles for phosphacan and RPTPb were determined from whole hippocampal extracts, which could potentially dilute injury effects specific to the deafferented ML. In fact, our IHC results suggest that, with time after UEC, phosphacan was reduced over the granule cell layer and increased in the ipsilateral deafferented ML. Thus, we re-exam-
FIGURE 4. Quantitative RT-PCR analysis of hippocampal phosphacan, short and full length RPTPb transcripts at 2, 7, and 15 d after UEC lesion. Change in mRNA expression shown as percent of contralateral transcript level and normalized to total RNA input. Differences were detected between the three splice variants over time postlesion. Phosphacan mRNA increased only at 7d postlesion. The full length splice variant (RPTPb) was increased at 2 and 15d, while sRPTPb was elevated over all postlesion intervals. 2d and 15d n 5 6; 7d n 5 5; *P < 0.05, **P < 0.01. Hippocampus
88
HARRIS ET AL.
FIGURE 5. Phosphacan protein expression in the ML enriched samples from dentate gyrus at 2 and 7d following UEC lesion. Enriched ML extracts are compared with whole hippocampal data (replotted from Fig. 1). At 2d, when whole hippocampal phosphacan is elevated, ML phosphacan did not change relative to control values. By contrast, the 7d ML enriched fraction showed increased phosphacan, positively correlated with elevation of hippocampal phosphacan transcript at 7d. 2d n 5 4; 7d n 5 3; *P < 0.02.
local dendritic 7d elevation in phosphacan protein, not detectable in whole hippocampal extracts, and ultrastructural ICC confirmed RPTPb localization within ML dendrites and spines. Confocal dual label experiments indicated a close association between RPTPb and PSD-95, consistent with the ultrastructural RPTPb distribution, and showed that reactive astrocytes are not the primary source of phosphacan or RPTPb in the deafferented zone. These results suggest that expression of different phosphacan splice variants is correlated with different phases of reactive synaptogenesis and that postsynaptic proteins may be principal targets of the membrane bound tyrosine phosphatase during synapse reconstruction.
Phosphacan and Reactive Synaptogenesis Prior investigations have documented temporal shifts in the expression of phosphacan protein after UEC. Deller et al. (1997) first reported an increase in phosphacan immunolabeling within the deafferented zone 6d after UEC, the period where collateral sprouting and synapse formation is initiated. Our IHC results show a similar pattern of increased phosphacan expression; however, we also observed phosphacan staining in granule cell somata of the contralateral dentate gyrus at 2d postlesion, a pattern which was shifted to predominant distribution over the deafferented outer ML on the lesioned side. At Hippocampus
7d postlesion, the same regional shift in phosphacan protein was visible and granule cell signal intensity was reduced. Differences in phosphacan profile between the present study and that of Deller et al. may be due to antibody specificity. It is possible that 3F8 antibody recognition of the phosphacan protein core revealed patterns of tissue distribution not recognized with the DSD-1-PG IgM antibody used in the earlier report. Dual label confocal imaging failed to show phosphacan within ML reactive astrocytes, either at 2 or 7d after lesion. Differing levels of reactive gliosis across injury models could play a role since reactive astrocytes alter their phosphacan expression after certain types of experimental brain injury. For example, astrocytic expression of phosphacan is enhanced following spinal cord contusion (Vitellaro-Zuccarello et al., 2008), ischemic brain injury (Beck et al., 2008), and filter implant-induced glial scarring (McKeon et al., 1999). Notably, the attenuation of astroglial response to spinal cord contusion reversed the injury-induced rise in phosphacan, while elevated expression of other CSPGs (e.g., versican, neurocan, and brevican) remained unaltered (Vitellaro-Zuccarello et al., 2008). In contrast, our results suggest that UEC does not induce astrocyte production of phosphacan, but rather granule cells are a principal source of phosphacan after lesion and facilitate its redistribution to areas undergoing reactive synaptogenesis. This interpretation is consistent with prior observations by Okamoto et al. (2001) which show that phosphacan is not colocalized to astrocytes within adult hippocampus. We also found that significant increase in phosphacan transcript was correlated with the onset of sprouting and synaptogenesis at 7d postlesion. An earlier study employing UEC in combination with fimbria/fornix lesion reported increased phosphacan mRNA, however only the 20d postinjury interval was sampled (Snyder et al., 1996). Here we failed to observe a change in phosphacan transcript as late as 15d, but since we did not sample the 20d time point, it remains possible that UEC produces elevation of phosphacan message at both 7 and 20d. Alternatively, the more severe combined insult used by Snyder et al. may have produced a longer lasting effect on phosphacan transcription. A second, more recent report applied precise laser dissection methods to isolate the outer deafferented dentate ML following UEC, assessing transcript expression for a panel of CSPGs including phosphacan (Schafer et al., 2008). Using qRT-PCR, this group found no significant change in phosphacan mRNA between 6h and 14d postlesion, including the 7d time point where we report increase in whole hippocampal transcript for phosphacan. These differences could be explained in at least two ways. First, our whole hippocampal extracts used for qRT-PCR contained granule neurons which, based upon IHC results, contribute significantly to phosphacan transcript. The laser dissected samples did not include dentate granule cell laminae. Second, the Schafer et al. study used 18S rRNA as a qRT-PCR reference gene, which we recently reported to be highly variable in expression and altered over time following UEC (Harris JL et al., 2009). Normalization of transcript to 18S rRNA could eliminate detection of experimental effect for specific genes of interest.
PHOSPHACAN AND REACTIVE SYNAPTOGENESIS In contrast to IHC, Western blot analysis of whole hippocampal extracts failed to show a change in phosphacan protein 7d after UEC, the time point where whole hippocampal phosphacan transcript was elevated. This apparent asynchrony between mRNA and protein response to UEC reiterates the fact that transcript may not always predict protein content. Prior investigations have advised caution in interpreting proteoglycan expression after CNS trauma, particularly where mRNA quantification is the sole endpoint (Iaci et al., 2007). In the present study, several conditions might account for this asynchrony. Postinjury pathology at 7d could suppress hippocampal translation of phosphacan mRNA. Alternatively, newly translated phosphacan protein might be rapidly degraded through the activation of extracellular matrix metalloproteinases which cleave phosphacan (Muir et al., 2002) and are upregulated in the hippocampus after UEC (Falo et al., 2006). The tPA/plasmin proteolytic pathway could also affect phosphacan during reactive synaptogenesis since plasmin degrades hypothalamic phosphacan during physiologically induced synaptic plasticity (Miyata et al., 2005). Finally, our analysis of phosphacan in subregional tissue extracts suggests that sampling method may best explain the 7d transcript/protein mismatch. In contrast to the whole hippocampus, ML/GCL enriched tissue samples exhibited a 7d increase in phosphacan protein, consistent with the 7d elevation of phosphacan mRNA. This 7d increase in phosphacan protein represents a local dentate response related to the postinjury period when sprouting and synapse formation is initiated. It is well established that endogenous growth factors can promote such sprouting in the injured CNS (Nieto-Sampedro and Bovolenta, 1990; Cui, 2006; Deller et al., 2006). Phosphacan binds to a many of these growth factors, including basic fibroblast growth factor (FGF-2), pleiotrophin, amphoterin, and midkine (Milev et al., 1998a,b; Maeda et al., 1999). This binding may localize these molecules at sites of sprouting, or sequester them for later mobilization. Increased phosphacan within the ML/GCL enriched sample is also consistent with the shift in the protein to the outer ML seen with IHC. When these results are considered with those of the Schafer et al. study, a clearer picture of phosphacan response to UEC emerges. We posit that deafferentation increases granule cell transcription of the phosphacan splice variant, producing protein which is transported distally along dendrites to ML sites of synaptic reorganization. Future PCR studies which assess local ML/GCL changes in transcript, as well as in situ hybridization experiments to confirm cell sources of phosphacan mRNA will be necessary to test this possibility.
RPTPb and Reactive Synaptogenesis Relative to phosphacan and other extracellular proteoglycans, there are fewer studies describing RPTPb after CNS injury. Dobbertin et al. (2003) examined the expression of both receptor variants after cortical stab injury, where sRPTPb protein and transcript did not change, but full length RPTPb mRNA was reduced. A stab injury produces discrete tissue damage,
89
local blood-brain barrier disruption, and astroglial scarring, with notably less compensatory plasticity. If sRPTPb is more specifically supportive of synaptic remodeling, then no change in sRPTPb expression would be predicted after focal stab insult. By contrast, sclerotic hippocampi from epileptic patients exhibited increase in RPTPb immunoreactivity, primarily associated with gliosis and mossy fiber sprouting in the inner ML (Perosa et al., 2002). Similar to temporal lobe epilepsy (TLE), UEC induces axonal sprouting in the hippocampus. Here we also observed an elevation in RPTPb protein, which persisted throughout the 15d postinjury period. Since full length RPTPb is primarily a developmental isoform (Sakurai et al., 1996; Dobbertin et al., 2003), and we failed to detect measurable amounts of that form by Western blot analysis, we conclude that our anti-RPTPb IHC signal represents primarily sRPTPb. Thus, the punctate outer ML distribution of sRPTPb at 7d supports its interaction with the local environment to induce collateral sprouting and promote reorganization of postsynaptic dendrites. Indeed, our ultrastructural ICC studies confirmed localization of sRPTPb within ML postsynaptic spines and dendrites, a pattern also reported by others for pyramidal neurons in cortex and hippocampus (Miyata et al., 2004; Hayashi et al., 2005). In parallel confocal experiments, we found that ML sRPTPb was often localized adjacent to PSD-95 signal, suggesting that it is positioned near reorganizing postsynaptic sites to influence protein phosphorylation. A similar RPTPb distribution along Purkinje cell dendritic shafts and somatic membranes was reported by Fukazawa et al. (2008). This localization suggests sRPTPb effects on the distribution of synaptic proteins, possibly through their phosphorylation-directed positioning within the neuronal membrane (Kawachi et al., 1999; Fukazawa et al., 2008; Tezuka et al., 1999). Of these sRPTPb targets, at least two, ErbB4 (Erlich et al., 2000) and PSD-95 (Ansari et al., 2008), show a change in expression after traumatic brain injury. A similar role for RPTPb was described during morphogenesis of cerebellar Purkinje dendrites in vitro, where RPTPb inhibition produced aberrant dendritic structure (Tanaka et al., 2003). Again, it appears that the actions of this splice variant are mediated through neuronal populations, as little evidence of significant astrocytic sRPTPb was observed. Concurrent elevations in sRPTPb mRNA were also found throughout the time course of UEC induced synaptogenesis. Persistent elevation suggests that the protein product of this splice variant influences all phases of the synaptogenic process. Using the same UEC/FFX lesion model as for their phosphacan analysis, Snyder et al. (1996) reported an increase in RPTPb mRNA at 20d postinjury. In this study, we also found increased RPTPb transcript at 15d after UEC lesion alone, suggesting that a single deafferentation insult is sufficient to up-regulate transmembrane phosphacan genes during the later stages of reactive synaptogenesis. Interestingly, message for full length RPTPb did increase at 2d and 15d postlesion. However, since we could not detect this form with immunoblots, elevated transcript may represent the accumulation of stable, but untranslated mRNA. Using a cortical knife lesion model, Hippocampus
90
HARRIS ET AL.
Dobbertin et al. (2003) also report an acute postinjury shift in full length RPTPb transcript without detectable full length protein in their tissue extracts. Since expression of the full length isoform has been best described during embryological development (Sakurai et al., 1996) and UEC deafferentation induces synaptogenesis similar to that observed in the developing hippocampus, it is plausible that full length RPTPb also participates in synaptic reorganization after injury. Nevertheless, additional studies using more specific markers of RPTPb splice variants will be required to clarify the role of the full length isoform.
Complexity of Phosphacan and RPTPb Response After Brain Injury Although many CSPGs such as aggrecan (Schafer et al., 2008), neurocan (Asher et al., 2000; Schafer et al., 2008), brevican (Schafer et al., 2008), versican (Asher et al., 2002), and NG2 (Tang et al., 2003; Schafer et al., 2008) are up-regulated after CNS injury, phosphacan response may be mixed depending upon injury modality and survival interval sampled (Matsui et al., 2002; Dobbertin et al., 2003; Okamoto et al., 2003; Jones et al., 2003; Tang et al., 2003; Heck et al., 2004). In some cases, both increases and decreases in phosphacan occur concurrently within different subregions after injury. For example, phosphacan is reduced at the core but increased at the margin of spinal cord lesion (Tang et al., 2003), while it may be increased at the border of stroke infarct, but decreased in the peri-infarct tissue (Carmichael et al., 2005). This complex reaction pattern appears most related to the type of injury induced, extent of synaptic plasticity involved, and the postinjury time frame examined. When the injury site is physically separated from the deafferented region, as occurs with UEC lesion, necrotic cell death and glial scarring do not directly affect the adaptive synaptic plasticity processes (Deller et al., 2000). These differences would explain a differential phosphacan response relative to other published models, which may depend upon the extent of neuronal and glial cell loss in the areas examined and the survival intervals targeted (Jones et al., 2003; Tang et al., 2003). Alternatively, loss of neuronal activity by massive deafferentation in the UEC model could shift expression of proteins whose transcription is regulated by such activity. Here, the loss of perforant path input profoundly reduces synaptic drive, resulting in extensive plasticity and pronounced effects on phosphacan/RPTPb during each phase of reactive synaptogenesis. Interestingly, an inverse relationship between cell activity and phosphacan/RPTPb expression has been reported in supraoptic magnocellular neurons (Miyata et al., 2004), supporting the idea that change in neuronal activity may also affect proteoglycan expression following UEC deafferentation. Future studies employing pharmacological inactivation of the EC or septal formation may determine the extent to which aberrant neuronal activity regulates proteoglycan function during synaptic reorganization in the hippocampus. Hippocampus
CONCLUSIONS In summary, the data presented here provide new evidence that phosphacan and sRPTPb each play a role in the reactive synaptic plasticity that occurs after brain injury. These two splice variants are differentially elevated during reactive synaptogenesis induced by UEC lesion. Shifts in phosphacan expression appear more specifically associated with the period of sprouting and synapse reconstruction, while transmembrane RPTPb is activated throughout the period of synapse morphogenesis and stabilization. These changes appear to be primarily associated with neurons and likely contribute to adaptive recovery after brain insult. To promote better recovery after brain injury, we must continue to explore how ECM proteins affect the local environment during synaptogenesis. Such studies will help to define differences between injuries where recovery is successful and those where recovery fails.
Acknowledgments The authors gratefully acknowledge the expert technical assistance of Lesley Harris, Raiford Black and Nancy Lee.
REFERENCES Ansari MA, Roberts KN, Scheff SW. 2008. A time course of contusion-induced oxidative stress and synaptic proteins in cortex in a rat model of TBI. J Neurotrauma 25:513–526. Asher RA, Morgenstern DA, Fidler PS, Adcock KH, Oohira A, Braistead JE, Levine JM, Margolis RU, Rogers JH, Fawcett JW. 2000. Neurocan is upregulated in injured brain and in cytokine-treated astrocytes. J Neurosci 20:2427–2438. Asher RA, Morgenstern DA, Shearer MC, Adcock KH, Pesheva P, Fawcett JW. 2002. Versican is upregulated in CNS injury and is a product of oligodendrocyte lineage cells. J Neurosci 22:2225–2236. Beck H, Semisch M, Culmsee C, Plesnila N, Hatzopoulos AK. 2008. Egr-1 regulates expression of the glial scar component phosphacan in astrocytes after experimental stroke. Am J Pathol 173:77–92. Bicknese AR, Sheppard AM, O’Leary DD, Pearlman AL. 1994. Thalamocortical axons extend along a chondroitin sulfate proteoglycanenriched pathway coincident with the neocortical subplate and distinct from the efferent path. J Neurosci 14:3500–3510. Bradbury EJ, Moon LD, Popat RJ, King VR, Bennett GS, Patel PN, Fawcett JW, McMahon SB. 2002. Chondroitinase ABC promotes functional recovery after spinal cord injury. Nature 416:636–640. Carmichael ST, Archibeque I, Luke L, Nolan T, Momiy J, Li S. 2005. Growth-associated gene expression after stroke: Evidence for a growth-promoting region in peri-infarct cortex. Exp Neurol 193:291–311. Cui Q. 2006. Actions of neurotrophic factors and their signaling pathways in neuronal survival and axonal regeneration. Mol Neurobiol 33:155–179. Davies SJ, Goucher DR, Doller C, Silver J. 1999. Robust regeneration of adult sensory axons in degenerating white matter of the adult rat spinal cord. J Neurosci 19:5810–5822. Deller T, Haas CA, Naumann T, Joester A, Faissner A, Frotscher M. 1997. Up-regulation of astrocyte-derived tenascin-C correlates with neurite outgrowth in the rat dentate gyrus after unilateral entorhinal cortex lesion. Neuroscience 81:829–846.
PHOSPHACAN AND REACTIVE SYNAPTOGENESIS Deller T, Haas CA, Frotscher M. 2000. Reorganization of the rat fascia dentata after a unilateral entorhinal cortex lesion. Role of the extracellular matrix. Ann NY Acad Sci 911:207–220. Deller T, Haas CA, Freiman TM, Phinney A, Jucker M, Frotscher M. 2006. Lesion-induced axonal sprouting in the central nervous system. Adv Exp Med Biol 557:101–121. Dobbertin A, Rhodes KE, Garwood J, Properzi F, Heck N, Rogers JH, Fawcett JW, Faissner A. 2003. Regulation of RPTPbeta/phosphacan expression and glycosaminoglycan epitopes in injured brain and cytokine-treated glia. Mol Cell Neurosci 24:951–971. Erlich S, Shohami E, Pinkas-Kramarski R. 2000. Closed head injury induces up-regulation of ErbB-4 receptor at the site of injury. Mol Cell Neurosci 16:597–608. Faissner A, Clement A, Lochter A, Streit A, Mandl C, Schachner M. 1994. Isolation of a neural chondroitin sulfate proteoglycan with neurite outgrowth promoting properties. J Cell Biol 126:783– 799. Falo MC, Fillmore HL, Reeves TM, Phillips LL. 2006. Matrix metalloproteinase-3 expression profile differentiates adaptive and maladaptive synaptic plasticity induced by traumatic brain injury. J Neurosci Res 84:768–781. Fukazawa N, Yokoyama S, Eiraku M, Kengaku M, Maeda N. 2008. Receptor type protein tyrosine phosphatase zeta-pleiotrophin signaling controls endocytic trafficking of DNER that regulates neuritogenesis. Mol Cell Biol 28:4494–4506. Garwood J, Schnadelbach O, Clement A, Schutte K, Bach A, Faissner A. 1999. DSD-1-proteoglycan is the mouse homolog of phosphacan and displays opposing effects on neurite outgrowth dependent on neuronal lineage. J Neurosci 19:3888–3899. Grumet M, Flaccus A, Margolis RU. 1993. Functional characterization of chondroitin sulfate proteoglycans of brain: Interactions with neurons and neural cell adhesion molecules. J Cell Biol 120:815– 824. Harris JL, Reeves TM, Phillips LL. 2009. Injury modality, survival interval and sample region are critical determinants of qRT-PCR reference gene selection during long-term recovery from brain trauma. J Neurotrauma 26:1–13. Harris NG, Carmichael ST, Hovda DA, Sutton RL. 2009. Traumatic brain injury results in disparate regions of chondroitin sulfate proteoglycan expression that are temporally limited. J Neurosci Res 87:2937–2950. Hayashi N, Oohira A, Miyata S. 2005. Synaptic localization of receptor-type protein tyrosine phosphatase zeta/beta in the cerebral and hippocampal neurons of adult rats. Brain Res 1050:163–169. Heck N, Garwood J, Loeffler JP, Larmet Y, Faissner A. 2004. Differential upregulation of extracellular matrix molecules associated with the appearance of granule cell dispersion and mossy fiber sprouting during epileptogenesis in a murine model of temporal lobe epilepsy. Neuroscience 129:309–324. Huang WC, Kuo WC, Cherng JH, Hsu SH, Chen PR, Huang SH, Huang MC, Liu JC, Cheng H. 2006. Chondroitinase ABC promotes axonal re-growth and behavior recovery in spinal cord injury. Biochem Biophys Res Commun 349:963–968. Iaci JF, Vecchione AM, Zimber MP, Caggiano AO. 2007. Chondroitin sulfate proteoglycans in spinal cord contusion injury and the effects of chondroitinase treatment. J Neurotrauma 24:1743–1759. Jones LL, Margolis RU, Tuszynski MH. 2003. The chondroitin sulfate proteoglycans neurocan, brevican, phosphacan, and versican are differentially regulated following spinal cord injury. Exp Neurol 182:399–411. Kawachi H, Tamura H, Watakabe I, Shintani T, Maeda N, Noda M. 1999. Protein tyrosine phosphatase zeta/RPTPbeta interacts with PSD-95/SAP90 family. Brain Res Mol Brain Res 72:47–54. Keppel G. 1991. Design and Analysis: A Researcher’s Hand book. Englewood Cliffs, NJ: Prentice Hall. Levine G. 1991. A Guide to SPSS for Analysis of Variance. Hillsdale, NJ: Lawrence-Erlbaum, Associates.
91
Levy JB, Canoll PD, Silvennoinen O, Barnea G, Morse B, Honegger AM, Huang JT, Cannizzaro LA, Park SH, Druck T. 1993. The cloning of a receptor-type protein tyrosine phosphatase expressed in the central nervous system. J Biol Chem 268:10573–10581. Loesche J, Steward O. 1977. Behavioral correlates of denervation and reinnervation of the hippocampal formation of the rat: Recovery of alternation performance following unilateral entorhinal cortex lesions. Brain Res Bull 2:31–39. Maeda N, Ichihara-Tanaka K, Kimura T, Kadomatsu K, Muramatsu T, Noda M. 1999. A receptor-like protein-tyrosine phosphatase PTPzeta/RPTPbeta binds a heparin-binding growth factor midkine. Involvement of arginine 78 of midkine in the high affinity binding to PTPzeta. J Biol Chem 274:12474–12479. Matsui F, Kawashima S, Shuo T, Yamauchi S, Tokita Y, Aono S, Keino H, Oohira A. 2002. Transient expression of juvenile-type neurocan by reactive astrocytes in adult rat brains injured by kainate-induced seizures as well as surgical incision. Neuroscience 112:773–781. Maurel P, Rauch U, Flad M, Margolis RK, Margolis RU. 1994. Phosphacan, a chondroitin sulfate proteoglycan of brain that interacts with neurons and neural cell-adhesion molecules, is an extracellular variant of a receptor-type protein tyrosine phosphatase. Proc Natl Acad Sci USA 91:2512–2516. Mehra A, Lee KH, Hatzimanikatis V. 2003. Insights into the relation between mRNA and protein expression patterns I: Theoretical considerations. Biotechnol Bioeng 84:822–833. McKeon RJ, Jurynec MJ, Buck CR. 1999. The chondroitin sulfate proteoglycans neurocan and phosphacan are expressed by reactive astrocytes in the chronic CNS glial scar. J Neurosci 19:10778– 10788. Meldgaard M, Fenger C, Lambertsen KL, Pedersen MD, Ladeby R, Finsen B. 2006. Validation of two reference genes for mRNA level studies of murine disease models in neurobiology. J Neurosci Methods 156:101–110. Meng K, Rodriguez-Pena A, Dimitrov T, Chen W, Yamin M, Noda M, Deuel TF. 2000. Pleiotrophin signals increased tyrosine phosphorylation of beta beta-catenin through inactivation of the intrinsic catalytic activity of the receptor-type protein tyrosine phosphatase beta/zeta. Proc Natl Acad Sci USA 97:2603–2608. Milev P, Monnerie H, Popp S, Margolis RK, Margolis RU. 1998a. The core protein of the chondroitin sulfate proteoglycan phosphacan is a high-affinity ligand of fibroblast growth factor-2 and potentiates its mitogenic activity. J Biol Chem 273:21439–21442. Milev P, Chiba A, Haring M, Rauvala H, Schachner M, Ranscht B, Margolis RK, Margolis RU. 1998b. High affinity binding and overlapping localization of neurocan and phosphacan/protein-tyrosine phosphatase-zeta/beta with tenascin-R, amphoterin, and the heparin-binding growth-associated molecule. J Biol Chem 273: 6998–7005. Miyata S, Akagi A, Hayashi N, Watanabe K, Oohira A. 2004. Activity-dependent regulation of a chondroitin sulfate proteoglycan 6B4 phosphacan/RPTPbeta in the hypothalamic supraoptic nucleus. Brain Res 1017:163–171. Moon LD, Asher RA, Rhodes KE, Fawcett JW. 2001. Regeneration of CNS axons back to their target following treatment of adult rat brain with chondroitinase ABC. Nat Neurosci 4:465–466. Morgenstern DA, Asher RA, Fawcett JW. 2002. Chondroitin sulphate proteoglycans in the CNS injury response. Prog Brain Res 137:313–332. Muir EM, Adcock KH, Morgenstern DA, Clayton R, von Stillfried N, Rhodes K, Ellis C, Fawcett JW, Rogers JH. 2002. Matrix metalloproteases and their inhibitors are produced by overlapping populations of activated astrocytes. Brain Res Mol Brain Res 100:103– 117. Nieto-Sampedro M, Bovolenta P. 1990. Growth factors and growth factor receptors in the hippocampus. Role in plasticity and response to injury. Prog Brain Res 83:341–355. Hippocampus
92
HARRIS ET AL.
Niisato K, Fujikawa A, Komai S, Shintani T, Watanabe E, Sakaguchi G, Katsuura G, Manabe T, Noda M. 2005. Age-dependent enhancement of hippocampal long-term potentiation and impairment of spatial learning through the rho-associated kinase pathway in protein tyrosine phosphatase receptor type Z-deficient mice. J Neurosci 25:1081–1088. Okamoto M, Sakiyama J, Kurazono S, Mori S, Nakata Y, Nakaya N, Oohira A. 2001. Developmentally regulated expression of brain-specific chondroitin sulfated proteoglycans, neurocan and phosphacan, in the postnatal rat hippocampus. Cell Tissue Res 306:217–229. Okamoto M, Sakiyama J, Mori S, Kurazono S, Usui S, Hasegawa M, Oohira A. 2003. Kainic acid-induced convulsions cause prolonged changes in the chondroitin sulfate proteoglycans neurocan and phosphacan in the limbic structures. Exp Neurol 184:179–195. Oohira A, Matsui F, Katoh-Semba R. 1991. Inhibitory effects of brain chondroitin sulfate proteoglycans on neurite outgrowth from PC12D cells. J Neurosci 11:822–827. Pariser H, Ezquerra L, Herradon G, Perez-Pinera P, Deuel TF. 2005. Fyn is a downstream target of the pleiotrophin/receptor protein tyrosine phosphatase beta/zeta-signaling pathway: Regulation of tyrosine phosphorylation of fyn by pleiotrophin. Biochem Biophys Res Commun 332:664–669. Perosa SR, Porcionatto MA, Cukiert A, Martins JR, Passeroti CC, Amado D, Matas SL, Nader HB, Cavalheiro EA, Leite JP, NaffahMazzacoratti MG. 2002. Glycosaminoglycan levels and proteoglycan expression are altered in the hippocampus of patients with mesial temporal lobe epilepsy. Brain Res Bull 58:509–516. Phillips LL, Lyeth BG, Hamm RJ, Povlishock JT. 1994. Combined fluid percussion brain injury and entorhinal cortical lesion: A model for assessing the interaction between neuroexcitation and deafferentation. J Neurotrauma 11:641–656. Pizzorusso T, Medini P, Berardi N, Chierzi S, Fawcett JW, Maffei L. 2002. Reactivation of ocular dominance plasticity in the adult visual cortex. Science 298:1248–1251. Properzi F, Asher RA, Fawcett JW. 2003. Chondroitin sulphate proteoglycans in the central nervous system: Changes and synthesis after injury. Biochem Soc Trans 31:335–336. Sakurai T, Friedlander DR, Grumet M. 1996. Expression of polypeptide variants of receptor-type protein tyrosine phosphatase beta: The secreted form, phosphacan, increases dramatically during embryonic development and modulates glial cell behavior in vitro. J Neurosci Res 43:694–706. Sakurai T, Lustig M, Nativ M, Hemperly JJ, Schlessinger J, Peles E, Grumet M. 1997. Induction of neurite outgrowth through contac-
Hippocampus
tin and nr-CAM by extracellular regions of glial receptor tyrosine phosphatase beta. J Cell Biol 136:907–918. Schafer R, Dehn D, Burbach GJ, Deller T. 2008. Differential regulation of chondroitin sulfate proteoglycan mRNAs in the denervated rat fascia dentata after unilateral entorhinal cortex lesion. Neurosci Lett 439:61–65. Snow DM, Lemmon V, Carrino DA, Caplan AI, Silver J. 1990. Sulfated proteoglycans in astroglial barriers inhibit neurite outgrowth in vitro. Exp Neurol 109:111–130. Snyder SE, Li J, Schauwecker PE, McNeill TH, Salton SR. 1996. Comparison of RPTP zeta/beta, phosphacan, and trkB mRNA expression in the developing and adult rat nervous system and induction of RPTP zeta/beta and phosphacan mRNA following brain injury. Brain Res Mol Brain Res 40:79–96. Steward O, Vinsant SL, Davis L. 1988. The process of reinnervation in the dentate gyrus of adult rats: An ultrastructural study of changes in presynaptic terminals as a result of sprouting. J Comp Neurol 267:203–210. Tamura H, Fukada M, Fujikawa A, Noda M. 2006. Protein tyrosine phosphatase receptor type Z is involved in hippocampus-dependent memory formation through dephosphorylation at Y1105 on p190 RhoGAP. Neurosci Lett 399:33–38. Tanaka M, Maeda N, Noda M, Marunouchi T. 2003. A chondroitin sulfate proteoglycan PTPzeta /RPTPbeta regulates the morphogenesis of purkinje cell dendrites in the developing cerebellum. J Neurosci 23:2804–2814. Tang X, Davies JE, Davies SJ. 2003. Changes in distribution, cell associations, and protein expression levels of NG2, neurocan, phosphacan, brevican, versican V2, and tenascin-C during acute to chronic maturation of spinal cord scar tissue. J Neurosci Res 71:427–444. Tezuka T, Umemori H, Akiyama T, Nakanishi S, Yamamoto T. 1999. PSD-95 promotes fyn-mediated tyrosine phosphorylation of the Nmethyl-D-aspartate receptor subunit NR2A. Proc Natl Acad Sci USA 96:435–440. Tricarico C, Pinzani P, Bianchi S, Paglierani M, Distante V, Pazzagli M, Bustin SA, Orlando C. 2002. Quantitative real-time reverse transcription polymerase chain reaction: Normalization to rRNA or single housekeeping genes is inappropriate for human tissue biopsies. Anal Biochem 309:293–300. Vitellaro-Zuccarello L, Mazzetti S, Madaschi L, Bosisio P, Fontana E, Gorio A, De Biasi S. 2008. Chronic erythropoietin-mediated effects on the expression of astrocyte markers in a rat model of contusive spinal cord injury. Neuroscience 151:452–466.
HIPPOCAMPUS 21:93–107 (2011)
Altered Patterning of Dentate Granule Cell Mossy Fiber Inputs Onto CA3 Pyramidal Cells in Limbic Epilepsy John J. McAuliffe,1,2,3y Stefanie L. Bronson,1,4y Michael S. Hester,5 Brian L. Murphy,1,4 Rene´e Dahlquist-Topala´,1 David A. Richards,1,2,3,4 and Steve C. Danzer1,2,3,4,5* ABSTRACT: Impaired gating by hippocampal dentate granule cells may promote the development of limbic epilepsy by facilitating seizure spread through the hippocampal trisynaptic circuit. The second synapse in this circuit, the dentate granule cell!CA3 pyramidal cell connection, may be of particular importance because pathological changes occurring within the dentate likely exert their principal effect on downstream CA3 pyramids. Here, we utilized GFP-expressing mice and immunolabeling for the zinc transporter ZnT-3 to reveal the pre- and postsynaptic components of granule cell!CA3 pyramidal cell synapses following pilocarpine-epileptogenesis. Confocal analyses of these terminals revealed that while granule cell presynaptic giant boutons increased in size and complexity 1 month after status epilepticus, individual thorns making up the postsynaptic thorny excrescences of the CA3 pyramidal cells were reduced in number. This reduction, however, was transient, and 3 months after status, thorn density recovered. This recovery was accompanied by a significant change in the distribution of thorns along pyramidal cells dendrites. While thorns in control animals tended to be tightly clustered, thorns in epileptic animals were more evenly distributed. Computational modeling of thorn distributions predicted an increase in the number of boutons required to cover equivalent numbers of thorns in epileptic vs. control mice. Confirming this prediction, ZnT-3 labeling of presynaptic giant boutons apposed to GFP-expressing thorns revealed a near doubling in bouton density, while the number of individual thorns per bouton was reduced by half. Together, these data provide clear evidence of novel plastic changes occurring within the epileptic hippocampus. V 2009 Wiley-Liss, Inc. C
KEY WORDS: thorny excrescence; pilocarpine; epileptogenesis
ZnT-3;
synaptic
plasticity;
INTRODUCTION Dentate granule cells innervate CA3 pyramidal cells via presynaptic terminal expansions known as giant mossy fiber boutons. These terminals are
1
Department of Anesthesia, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio; 2 Department of Anesthesia, University of Cincinnati, Cincinnati, Ohio; 3 Department of Pediatrics, University of Cincinnati, Cincinnati, Ohio; 4 Program in Neuroscience, University of Cincinnati, Cincinnati, Ohio; 5 Molecular and Developmental Biology Graduate Program, University of Cincinnati, Cincinnati, Ohio Additional Supporting Information may be found in the online version of this article. y J. J. M. and S. L. B. contributed equally to this work. Grant sponsor: National Institute of Neurological Disorders and Stroke; Grant numbers: R01-NS-065020, R01-NS-062806; Grant sponsors: Cincinnati Children’s Hospital Medical Center; The Epilepsy Foundation of America. *Correspondence to: Dr. Steve C. Danzer, 3333 Burnet Avenue, ML 2001, Cincinnati, OH 45229-3039. E-mail:
[email protected] Accepted for publication 21 September 2009 DOI 10.1002/hipo.20726 Published online 15 December 2009 in Wiley Online Library (wileyonlinelibrary.com). C 2009 V
WILEY-LISS, INC.
among the largest and most potent in the entire brain (Henze et al., 2002; Lawrence and McBain, 2003), and the postsynaptic structures they innervate are also unique. Specifically, in contrast to the small head and short stalk typical of most dendritic spines, CA3 pyramidal spines are much larger, and most impressively, are frequently organized into elaborate compellations of spine heads or ‘‘thorns’’ all connected to the parent dendrite by a single stalk. These structures—termed thorny excrescences—are frequently enveloped by the apposed giant bouton, with the entire structure containing as many as 40 separate presynaptic release sites and corresponding postsynaptic densities (Chicurel and Harris, 1992). The efficacy of this structure in producing suprathreshold activation of targeted CA3 pyramidal cells has earned it the name ‘‘detonator synapse.’’ In the normal brain, the dentate limits the flow of information through the hippocampal trisynaptic circuit. Aberrant plastic changes of granule cell axons and dendrites occurring during the development of epilepsy, however, are hypothesized to impair this normal gating function, effectively destabilizing the circuit and promoting seizures (for review see Hsu, 2007). An important prediction of this hypothesis is that granule cell inputs to their target CA3 pyramidal cells will be enhanced in the epileptic brain, and indeed, recent studies have demonstrated that granule cell presynaptic terminals exhibit plastic changes indicative of increased potency (Goussakov et al., 2000; Pierce and Milner, 2001; Danzer et al., in press). The impact of these presynaptic changes, however, could be mitigated or exacerbated by changes among CA3 pyramidal cell postsynaptic thorny excrescences. Determining whether and how these structures are altered during the development of temporal lobe epilepsy (TLE) will provide key insights into the significance of granule cell!CA3 pyramidal cell transmission in epilepsy. Here, we take advantage of recent advances in transgenic labeling technologies and confocal microscopy to examine both pre- and postsynaptic components of the granule cell!CA3 pyramidal cell synapse in green fluorescent protein expressing mice, and develop and validate a novel immunostaining procedure for revealing mossy fiber terminals. These advances now make it possible to clearly resolve postsynaptic thorny excrescences and apposed presynaptic structures. Because of the complexity of these structures, it
94
MCAULIFFE ET AL.
has been difficult to resolve thorny excrescences with light microscopy in the past, which has limited quantification. The present study, therefore, not only examines plasticity of these structures during epileptogenesis, but provides new insights into their organization and distribution under control conditions.
METHODS Pilocarpine Model of Epilepsy Forty-two adult male Thy1-GFP-expressing mice (M line; Feng et al., 2000) on a C57BL/6 background were used for the present study. No female mice were used to avoid estrus cycleinduced changes in thorny excrescence density (Tsurugizawa et al., 2005). All procedures complied with institution and NIH guidelines for the care and use of animals. Two- to threemonth-old mice were injected subcutaneously (s.c.) with 1 mg kg21 methyl scopolamine nitrate in sterile saline, followed 15 min later with 380 mg kg21 pilocarpine in saline. Treatments were conducted between 10 AM and noon to control for diurnal variations. Mice were observed following the injections for the development of continuous seizure activity (status epilepticus or SE), defined behaviorally by continuous tonic/ clonic convulsions. Mice were given two doses of 10 mg kg21 diazepam at 15-min intervals 3 h after the onset of SE to control seizure activity and improve survival. With this protocol, mice experience a minimum of 3 h of continuous generalized seizure activity (unpublished observations based on EEG data). Control animals received all drugs and treatments, except they were given saline instead of pilocarpine. Control and pilocarpine-treated mice were sacrificed either 1 or 3 months after treatment, producing a total of four groups for analyses of GFP-expressing cells (1M controls, N 5 7; 1M SE, N 5 9; 3M controls, N 5 7; 3M SE, N 5 7). A second group of mice sacrificed 3 months after control treatment (N 5 5) or pilocarpine-status epilepticus (N 5 7) was generated for GFP1zinc transporter 3 (ZnT-3) double-labeling experiments. Mice were overdosed with pentobarbital (100 mg kg21) and perfused with phosphate buffered saline (PBS)11U ml21 heparin followed by 2.5% paraformaldehyde and 4% sucrose in PBS, pH 7.4. Brains were postfixed for 12 h, cryoprotected in sucrose (10, 20, 30%), snap-frozen in isopentane, and cryosectioned in the coronal plane at 60 lm. Slide mounted sections were stored at 2808C until use (Superfrost Plus slides; Fisher Scientific).
Immunohistochemistry As part of a separate study (Walter et al., 2007), mice used in the present study were injected with BrdU either 8 weeks before or 1 week after pilocarpine treatment (100 mg kg21 s.c. 3 3). BrdU treatment is not expected to impact any parameters examined in the present study, and both control and piloHippocampus
carpine-treated animals received identical treatments. Correspondingly, however, GFP immunohistochemistry was conducted using a combined BrdU/GFP immunostaining protocol, as described previously (Walter et al., 2007). Briefly, endogenous GFP expression was enhanced by incubating sections overnight at 48C in 5 lg ml21 rabbit polyclonal anti-GFP antibody (AB3080, Chemicon, Temecula, CA) in blocker followed by 1:750 Alexa Fluor 594 goat antirabbit antibodies (Molecular Probes, Eugene, OR). After rinsing in PBS, sections were dehydrated in alcohols, cleared in xylenes and mounted with Cytoseal. ZnT-3 immunolabeling was conducted on slide mounted sections. Sections were thawed in PBS, blocked for 1 h with 0.5% Igepal15% normal goat serum in PBS and incubated overnight in 1:10,000 polyclonal rabbit anti-ZnT-3 (Synaptic Systems, Go¨ttingen, Germany) in blocker at 48C. Sections were then rinsed in blocker, incubated for 2 h in 1:750 Alexa Fluor 594 goat antirabbit IgG in blocker, rinsed in PBS, and mounted with Gel Mount (Biomeda, Foster City, CA).
Microscopy and Data Collection GFP-expressing pyramidal cells were selected only from the CA3b or midportion region of CA3, being bounded by an imaginary line drawn between the tips of the blades of the dentate gyrus medially and the fimbria laterally (Lorente de No´, 1934; Ishizuka et al., 1995). CA3 pyramidal cells exhibit considerable regional heterogeneity (Buckmaster and Amaral, 2001; Gonzales et al., 2001), so the present study focused on CA3b to increase the likelihood of detecting differences among groups. Further selection criteria included: (1) bright GFPlabeling, so that thorny excrescences could be accurately counted; (2) Cell body located in the pyramidal cell layer; (3) apical dendritic projection through stratum lucidum; (4) presence of dendritic spines on process projecting into stratum radiatum, to distinguish pyramidal cells from aspiny interneurons (no labeled interneurons were observed, suggesting that the Thy1-GFP line specifically labels pyramidal cells in this region). Neuron selection was conducted with the experimenter blinded to treatment group, and since CA3 pyramidal cell labeling is relatively sparse in the Thy1-GFP line, all CA3 pyramidal cells meeting these selection criteria were imaged for analyses. Cell counts were conducted to determine whether pilocarpine-status epilepticus altered the number of GFP-expressing CA3 pyramidal cells. GFP-expressing pyramidal cells were counted in the CA3b region of hippocampus from randomly selected sections between the anterior–posterior coordinates Bregma 21.34 to 22.54. GFP-expressing cells were counted using a modified optical dissector method (Williams and Rakic, 1988; Peterson, 1999; Howell et al., 2002) and values expressed as the number of GFP-labeled CA3b pyramids per hippocampal section. Approximately eight 60-lm brain sections (two hippocampi/section) were scored for each animal. Pyramidal cells selected for analysis were imaged using a Leica SP5 confocal system set up on a DMI 6,000 inverted
ALTERED PATTERNING OF DENTATE GRANULE CELL MOSSY FIBER INPUTS
FIGURE 1. Confocal maximum projections from control animals showing branch order designations and thorny excrescence details for imaged pyramidal cells. (A) CA3b pyramidal cell with first (18), second (28), and third (38) order branches noted. The breadth of stratum lucidum (sl) is denoted by the bracket ({). For this particular cell, only second and third order branches would be quantified, as the first order dendritic segment is located outside of stratum lucidum. (B) CA3b pyramidal cell apical dendrite with three simple thorny excrescences possessing only a single thorn (asterisks) and two complex thorny excrescences with multiple thorns (circled region). Blue arrows denote the necks of the two complex thorny excrescences, while blue arrowheads denote some of the individual thorns making up these structures. (C) Pyramidal cell apical dendrite crossing the border between stratum lucidum (sl) and stratum radiatum (sr). Note the structural differences between thorns (blue arrowheads) in stratum lucidum and dendritic spines (filled arrowheads) in stratum radiatum. Scale bars 5 20 lm (A); 5 lm (B); 3 lm (C). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
microscope. Cells were first imaged under 103 (NA 0.3) magnification so that neuronal position relative to the dentate granule cell layer could be determined. Cell position was measured using Leica quantification software (1.3.1 build 525) by first drawing a straight line connecting the tips of the upper and lower blades of the dentate gyrus. The distance between the midpoint of this first line and the pyramidal cell body was scored as distance from the dentate gyrus. CA3 pyramidal cell apical dendrites projecting through stratum lucidum were then imaged using a 633 oil immersion objective (NA 1.4). The position of stratum lucidum was determined by the presence of GFP-labeled granule cell mossy fiber axons (Fig. 1A, sl). Pyramidal cell body depth in the pyramidal cell layer was determined by measuring the distance between the top of the soma and stratum lucidum and dividing by the total width of the pyramidal cell layer. Apical dendrites were imaged using 53 optical zoom with a 0.2-lm step through the z-depth of the tissue, producing three-dimensional confocal image ‘‘stacks.’’ Fluorochromes were excited using the 543-nm laser line and emission wavelengths between 610 and 650 nm were collected. Only pyramidal cell dendritic segments bounded by mossy fiber axons (i.e., within stratum lucidum) were used for analysis. The thorny excrescences of CA3 pyramidal cells possess a single, thin spine neck of variable length (Fig. 1B, blue arrows)
95
which emerges from the parent dendrite and gives rise to one or more bulbous ‘‘thorns’’ (Fig. 1B, asterisks and blue arrowheads) (Blackstad and Kjaerheim, 1961; Amaral and Dent, 1981; Chicurel and Harris, 1992; Stewart et al., 2005; Lauer and Senitz, 2006; Rollenhagen et al., 2007). The complexity of these structures has challenged researchers attempting to quantify them, not only because a single excrescence can have multiple thorns, but because excrescences are frequently organized into groups (Fig. 1B, circled region). Until recently, limits in light level microscopy have hampered clearly resolving these structures. Because of these limitations, previous light level studies have taken a number of quantitative approaches, including measuring the length of entire groups (Buckmaster and Amaral, 2001; Gonzales et al., 2001) or attempting to identify individual thorny excrescences (Jiang et al., 1998). Fortunately, advances in confocal imaging now make it possible to resolve the individual thorns making up a thorny excrescence (Danzer and McNamara, 2004; Tsurugizawa et al., 2005), and for the present study we have used this variable to quantify these structures. With the improved resolution, we found group measurements to be impractical, as it was difficult to account for the three-dimensional arrangement of groups around a dendrite with simple length measurements. Moreover, measurements of thorn density may have the most biological significance, as they are the targets for synaptic innervation (Chicurel and Harris, 1992). To quantify thorn density, three-dimensional confocal image stacks were imported into Neurolucida software for digital reconstruction (version 7.50.4; Microbrightfield). The branch order of dendritic segments contained within stratum lucidum was determined for each cell (see Fig. 1A, for an explanation of branch order) and the number of thorns along each segment was quantified. To be counted as a thorn, the structure had to possess a minimum diameter of 0.4 lm and be located in stratum lucidum. Notably, thorns are structurally distinct from simple spines located in adjacent stratum radiatum (Fig. 1C). Simple thorny excrescences with a single thorn were counted as one (Fig. 1B, asterisks), while each individual thorn was counted for complex thorny excrescences (Figs. 1B,C, arrowheads). All thorns meeting these criteria along segments of dendrite within stratum lucidum were counted. Thorn density was then defined as the number of thorns per micrometer of dendrite. Finally, while the larger size (relative to typical dendritic spines) and the use of three-dimensional confocal images stacks made the counting of thorns reliable and accurate, even in cases when the structure was located above or below the parent dendrite, the interconnecting branches making up a thorny excrescence were not consistently visible. Because of this limitation in light level resolution, and the frequent grouping of multiple thorny excrescences, we did not attempt to determine the average number of thorns making up individual thorny excrescences. To quantify dentate granule cell giant mossy fiber bouton area and complexity, image stacks used to assess thorn density were reexamined for giant mossy fiber boutons using Neurolucida software. All brightly-labeled giant mossy fiber boutons Hippocampus
96
MCAULIFFE ET AL.
contained within the image stack were identified. When five giant boutons or fewer were present, all were quantified. When more than five giant boutons were present, boutons were randomly selected for analysis. Additional selection criteria were used as described previously (Danzer et al., 2008, in press). Analysis consisted of (1) determining the maximum cross sectional area of each giant bouton (excluding filopodia) and (2) determining whether the giant bouton possessed satellite boutons. Maximum cross sectional area was determined as if the structure had been flattened into two-dimensions and then outlined (although this was actually done using three dimensional images stacks as this is more accurate than using maximum projections). Granule cell mossy fiber axons contact CA3 pyramidal cell dendrites at roughly 100–250-lm intervals in CA3 (Claiborne et al., 1986). At each of these contact points, either a single giant bouton, or a core giant bouton with satellite giant boutons is found (Galimberti et al., 2006). Satellite boutons met the same criteria used for giant mossy fiber boutons except they were connected to a core giant bouton by thin axonal process rather than being directly connected to the main mossy fiber axon (as the core bouton was).
Thorn Cover Set Analysis A set of 14 pyramidal cells was analyzed to determine the distribution of thorns along the length of each dendrite. Seven of these were from control animals and seven were from 3M SE mice. Dendrites were selected for analysis if thorn density over the entire region of dendrite examined exceeded 1.0 thorns/lm. This criterion was used to explore changes in thorn distribution among the subpopulation that accounted for the recovery in overall thorn density observed 3 months after status. Analyses were run on the entire portion of the dendritic tree reconstructed, rather than segments broken down by branch order, to insure that sufficient length was included for each cell so that values would be representative. All cells meeting the thorn density criterion for the two groups (control and 3M SE) were included in the set. Finally, we note that the thorn dataset includes thousands of data points arranged threedimensionally in space. Although the analytical approaches developed here are complex, we found this complexity to be necessary to appropriately characterize the results. A program was written in Mathematica1 6.1 (Wolfram Research, Champaign, IL) to compute the mean distance between thorns by sampling the entire dataset of almost 4,000 intervals. The program was designed to oversample the dataset so that low frequency events would be accurately represented. A second program was written to compute the minimum cover set for each dendrite, defined as the fewest number of spheres of a given radius r required to cover each thorn on the dendritic tree one time with no overlap. Analyses were run using the x-, y-, and z-coordinates for each thorn generated from Neurolucida reconstructions. To calculate the minimum cover set, the program first determined the number of near-neighbors for each thorn. A thornk was defined to be a near-neighbor of thorni if thornk was within a sphere centered on thorni with a radius of r. Hippocampus
The exclusion zone includes the surface of the sphere, thus the sets of near-neighbors are open sets. This process was then repeated until all the thorns in the set thornj 5 1 to N were assigned a value for number of near-neighbors, where N equals the total number of thorns. The program then scanned the near neighbor set to find the thorn with the largest number of near neighbors. This thorn and all of its near neighbors defined the first element of the minimum cover set. The thorns contained in this cover set element were removed from the parent dataset. Next, the thorn with the largest number of near neighbors among thorns remaining in the parent dataset was identified. This thorn and all its neighbors defined the second element of the minimum cover set. As before, the thorns in this second element of the minimum cover set were removed from the parent dataset. This process was reiterated until all thorns along the dendrite were accounted for, with the final number of elements being equal to the minimum cover set, or, restated, the minimum number of hypothetical giant mossy fiber boutons (hMFB) of a radius r required to cover every thorn on a segment of dendrite once and only once. The minimum cover set was then divided by the total number of thorns along the dendrite to give the cover fraction. Cover fraction is the number of hMFB’s required to cover one thorn, and is also a measure of thorn tendency to cluster. The more dispersed thorns are along the dendrite, the greater number of spheres will be required to cover all thorns on a dendrite, and the cover fraction will approach 1.0 (one sphere covers only one thorn). Conversely, dendrites with highly clustered thorns will require fewer spheres to reach full coverage (1 sphere covers many thorns), and cover fraction approaches 0. For these analyses, r was set at 1, 1.4, 1.6, 2.0, and 2.2 lm. Finally, near neighbor values for each dendrite were used to calculate the mean number of near neighbors. These values can be viewed as reflecting the mean number of thorns per hMFB.
Analysis of GFP1ZnT-3 Double-Labeling Endogenous GFP expression was used to identify CA3 pyramidal cell apical dendritic segments by an investigator blinded to treatment group. Analysis focused on cells with high thorn densities (thorn density > 1.0 lm). Cells meeting this criterion were randomly selected for further analysis. Endogenous GFP expression and ZnT-3 immunolabeling were imaged using confocal microscopy (633 magnification, 63 optical zoom, 0.2-lm step). Image stacks were imported into Neurolucida software and the association between individual thorns and ZnT-3 immunoreactive puncta (putative mossy fiber terminals) was determined. Thorns were assumed to be interacting if they were surrounded by or directly apposed to a ZnT-3 immunoreactive puncta. Once a single thorn was scored as interacting with a puncta, the borders of that puncta were defined within the x-, y-, and z-axis. Any additional thorns contacting this puncta were then counted as well. Thorns that were not in apposition to a ZnT-3 immunoreactive puncta were counted as ‘‘orphan’’ thorns, and ZnT-3 immunoreactive puncta that were adjacent to the dendritic shaft—but not apposed to a thorn— were not scored. ZnT-3 puncta density along the shaft was
ALTERED PATTERNING OF DENTATE GRANULE CELL MOSSY FIBER INPUTS
97
determined by dividing the number of puncta contacting at least one thorn by the length of dendrite examined, and the number of thorns per puncta was calculated by dividing the total number of thorns (excluding orphan thorns) by the number of interacting puncta. This value was used to derive the cover fraction (no. puncta to cover 100 thorns).
Statistics Statistical analyses were run using either Sigma Stat software (version 2.03) or STATA (version 10.1 for MAC, StataCorp, College Station, TX). Parametric tests were used for normally distributed data with equal variance and nonparametric tests were used for data that violated these assumptions. Individual tests used are noted in the text. P-values less than 0.05 were considered significant. In all cases, except for analyses of cells with thorn densities greater than 1.0/lm, values for individual neurons were averaged to provide a mean score for each animal and statistical analyses were conducted using these animal means.
Figure Preparation Images are confocal maximum projections and were prepared using Leica’s LAS-AF Confocal software (1.3.1 build 525) and Adobe Photoshop (version 7.0). Images were processed using an erosion filter run for one iteration with a three pixel radius (Leica software) to reduce background artifact. Contrast and brightness were adjusted identically for images meant for comparison. Neuronal reconstructions from confocal image stacks were generated as described previously (Walter et al., 2007). Briefly, pseudocolor three-dimensional image stacks were processed to remove all structures not belonging to the target cell. The processed image was then superimposed on the original maximum projection (in a contrasting color) to reveal the target cell in the context of its surroundings. This processing is necessary to effectively present three-dimensional data in a twodimensional format, and although composed of original confocal images, they should be viewed as neuronal reconstructions with the associated limitations. The 3D renderings for Figure 8 were produced by importing confocal image stacks into Slidebook 4.2 software (Intelligent Imaging Innovations, Boulder, CO.). Z-series image stacks were cropped to contain just the dendrite of interest together with a narrow region of surrounding ZnT-3 staining. They were Gaussian filtered with a 1.5 pixel kernel to improve image smoothness, and then volume rendered using Slidebook’s High Quality Dynamic Lighting module, with 75% opacity.
RESULTS Giant Mossy Fiber Bouton Area and Complexity are Increased 1 and 3 Months After Status Epilepticus Giant bouton area and complexity were assessed in 1- and 3-month control mice, and mice collected 1 (1M SE) and 3
FIGURE 2. Dentate granule cell giant mossy fiber bouton (MFB) area is significantly increased both 1 and 3 months after status epilepticus (SE) relative to controls. The percentage of giant boutons with satellites was also increased at both time points after status relative to controls. Confocal maximum projections depict a simple giant bouton from a control animal, and core boutons (core) with satellites (S) from animals 1 and 3 months after status. ***P < 0.001. Scale bar 5 5 lm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
(3M SE) months after pilocarpine-induced status epilepticus. Analysis of the 1 and 3 month control groups revealed that they were statistically equivalent for all parameters examined in the present study (t test), so the two groups were pooled, and from here on are referred to as controls. A total of 222 giant boutons from fourteen control mice, 209 giant boutons from eight 1M SE mice, and 135 boutons from seven 3M SE mice were examined. Mossy fiber bouton area was significantly increased in both groups of epileptic animals relative to controls (Fig. 2; P < 0.001, ANOVA with Tukey’s post test). The frequency of giant boutons connected to ‘‘satellite’’ boutons by short axonal processes was also increased at both poststatus time points relative to controls (Fig. 2; P < 0.001, KruskalWallis rank sum test with Dunn’s post test).
Thorny Excrescence Labeling in Thy1-GFP-Expressing Mice GFP-labeling of CA3b pyramidal cell thorny excrescences in the Thy1-GFP line was robust, allowing for easy quantification of their numbers and complexity. A total of 50 cells were imaged from control animals, 33 cells from 1M SE animals, and 26 cells from 3M SE animals. All CA3 pyramidal cells examined conformed to the morphology of ‘‘classical’’ CA3 pyramidal cells (Lorente de No, 1934; Amaral, 1978; Frotscher et al., 1988; Ishizuka et al., 1990, 1995; Li et al., 1994; Seress and Ribak, 1995; Buckmaster and Amaral, 2001). Cell bodies were located in the pyramidal cell layer and possessed one to three prominent apical dendrites (although a single cell with Hippocampus
98
MCAULIFFE ET AL.
four apical dendrites was observed in a control). Apical dendrites projected through stratum lucidum, stratum radiatum, and on into stratum lacunosum-moleculare. Basal dendrites projected into stratum oriens. These findings suggest that pyramidal cells labeled in the Thy1-GFP line are representative of the entire population; a conclusion consistent with previous work demonstrating that labeled dentate granule cells in this line are indistinguishable from granule cells labeled with other approaches (Vuksic et al., 2008). The present study focused on segments of CA3 pyramidal cell apical dendrites contained within stratum lucidum, the projection field of granule cell mossy fiber axons and the principal localization of pyramidal cell thorny excrescences. Stratum lucidum is relatively thin—50 lm or so—and correspondingly, only a small portion of a CA3 pyramidal cells’ apical dendritic tree is localized to this region. Pyramidal cell apical dendritic branches within this region were typically first, second, or third order (Fig. 1). Fourth and fifth order branches were occasionally observed, but were not further analyzed due to their low incidence. Total dendritic length examined, broken down by branch order, was 758 lm (1st order), 1,775 lm (2nd order) and 1,800 lm (3rd order) for control neurons. For animals examined 1 month after status, 616, 1,441, and 1,143 lm were scored for first, second, and third order branches, respectively. Finally, 512, 1,276, and 884 lm were scored from animals collected 3 months after status.
Thorny Excrescence Density Varies by Branch Order Intriguingly, although the localization of first, second, and third order branches to stratum lucidum renders them as potential targets for mossy fiber innervation, thorns were not distributed equally among segments. Specifically, comparisons of thorn density among branch orders for control animals revealed that density increased significantly with higher branch order (1st order, 0.370 6 0.237 thorns/lm; 2nd order, 1.000 6 0.394; 3rd order, 0.778 6 0.167; P 5 0.005 for 2nd and 3rd vs. 1st, Kruskal-Wallis rank sum test with Dunn’s post test). In light of these findings, comparisons between control and epileptic animals were made among equivalent branch orders.
Thorny Excrescence Density is Transiently Reduced Following Status Epilepticus One month after pilocarpine-induced status epilepticus, the density of thorns along 3rd order CA3 pyramidal cell dendritic segments was significantly reduced relative to pyramidal cells from control animals (Fig. 3; P 5 0.043, ANOVA with Tukey’s post test). Similar trends towards reduced thorn density were observed for 1st and 2nd order dendritic segments, although the effect did not reach significance (Figs. 3 and 4). Notably, the reduction in density along 3rd order segments was transient, and 3 months after status epilepticus thorn densities were statistically indistinguishable from controls (Fig. 3). Hippocampus
FIGURE 3. The density of thorns was significantly decreased along 3rd order dendritic segments 1 month after status epilepticus (SE) relative to control animals and animals examined 3 months after status. Similar reductions were observed for first and second order dendritic segments, although the effects did not reach significance. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
ALTERED PATTERNING OF DENTATE GRANULE CELL MOSSY FIBER INPUTS
FIGURE 4. Neuronal reconstructions showing apical dendritic segments from control animals and animals examined 1 (1M SE) and 3 (3M SE) months after status epilepticus (SE). Control animals, and animals examined 3 months after status occasionally exhibited dendrites with dramatic accumulations of thorny excrescences (B and F). In contrast, such large accumulations were rare 1 month after status, although occasional complex thorny excrescences were observed (D, arrow). Dendritic branch order (28 or 38) is noted adjacent to the relevant dendritic segment. Thorn density for each neuron shown is noted in the lower right corner of each image (thorns/lm; values are for all branch orders combined). Scale bar 5 10 lm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
99
The first approach used to examine thorn distributions patterns involved calculating the mean distance between thorns along individual dendrites. The mean value for the distance between adjacent spines (nearest neighbors) was 0.806 lm (95%CI 0.793–0.819) for cells from control animals and 0.880 lm (95%CI 0.860–0.900) for cells from animals examined 3 months after status epilepticus. The difference was significant (P < 0.001), confirming qualitative impressions that thorn distributions differed. To explore whether this difference might have biological significance, a second analytical approach was developed to predict that number of giant mossy fiber boutons that would be required to innervate the thorns in each dataset. Briefly, the three-dimensional coordinates for each thorn were used to determine the minimum number of spheres with a radius r required to cover every thorn along a dendritic tree one time. This value is defined as the minimum cover set, and was developed as a new means for quantifying thorn distribution to better assess the unique innervation patterns of CA3 pyramidal cells by granule cell giant boutons. Basically, the minimum cover set can be viewed as a measure of the smallest number of hypothetical giant mossy fiber boutons (hMFB) of a given radius r that would be required to cover every thorn along a length of dendrite without overlap. Here, the cover set was nor-
Analysis of pyramidal cell variability in epileptic animals revealed that the reduction in thorn density evident 1 month after status epilepticus was due largely to the disappearance of pyramidal cells with high densities. While 23.7% (9 of 38) of pyramidal cells from control animals exhibited densities greater than 1 thorn/lm on 3rd order branches, none of the cells examined 1 month after status exhibited such high densities on 3rd order branches (0 of 20). Three months after status, cells with dense accumulations of thorns reappeared, making up 27.7% (5 of 18) of the cell population examined.
Altered Thorn Distribution 3 Months After Status Epilepticus Intriguingly, although thorn density was similar between control pyramidal cells and cells examined 3 months after status, casual observation suggested that the distribution of thorns was altered after status—particularly for cells with higher thorn densities. Specifically, while thorns along control dendrites tended to be highly clustered, thorns in animals examined 3 months after status were more evenly distributed (Fig. 5). Simple density measurements, however, failed to capture this phenomenon (e.g., 10 clustered thorns produces the same overall density as 10 evenly distributed thorns for a given length of dendrite), so two alternate strategies were developed to analyze the data. Analyses focused on pyramidal cells with thorn densities greater than 1.0/lm—seven from control animals and seven from 3M SE animals.
FIGURE 5. (A) Neurolucida reconstructions showing thorn distributions along pyramidal cells from control animals and animals examined 3 months after status epilepticus (3M SE). Note the tendency for individual thorns to form clusters along the dendrites of cells from control animals (several examples are circled), while cells from epileptic animals tend to exhibit a more even thorn distribution. (B) Neurolucida reconstructions of representative giant mossy fiber boutons from control and 3M SE animals. Note that a single bouton is large enough to cover a cluster of thorns (red circles), but many boutons would be required to cover all of the thorns on a dendrite, particularly when individual thorns are separated by greater distances. Scale bar 5 10 lm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] Hippocampus
MCAULIFFE ET AL.
100
TABLE 1. Thorn Cover Set Analysis Cover fraction 3 100 (no. hMFB to cover 100 thorns) Median (range)
Mean number of thorns/hMFB 6 standard deviation
hMFB radius (lm)
hMFB area (lm)2
1.0 1.4 1.6 2.0 2.2
3.14 6.15 8.04 12.57 15.21
Control (N 5 7) 2.95 5.11 6.29 8.95 10.40
6 6 6 6 6
0.48 0.76 0.78 1.42 1.69
3M SE (N 5 7)
Kruskal-Wallis rank sum test P-value
6 6 6 6 6
0.159 0.025 0.029 0.047 0.064
2.65 4.27 5.24 7.50 8.66
0.42 0.64 0.80 1.20 1.48
Control (N 5 7) 43.2 28.4 24.1 17.2 16.1
(32–64) (25–45) (20–40) (15–32) (14–27)
3M SE (N 5 7) 51.4 36.0 30.8 22.5 18.7
(45–72) (31–53) (25–43) (18–35) (14–31)
Mann-Whitney rank sum test P-value 0.073 0.038 0.038 0.026 0.073
Thorn cover set analysis for CA3 pyramidal cells with thorn densities > 1/lm. Radii used to generate cover sets are given in Column 1. Corresponding hypothetical giant mossy fiber bouton (hMFB) areas are shown in Column 2. Significant values are in bold. Note that while cover fraction and mean number thorns/hMFB are correlated they cannot be directly converted from one to the other.
malized for each dendrite by dividing it by the total number of thorns to give the cover fraction; the number of hMFB’s required to cover one thorn. This analysis was also used to determine the mean number of thorns per hMFB of a given radius. Analyses were run with radii set to 1.0, 1.4, 1.6, 2.0, and 2.2. Middle values (1.4, 1.6, and 2.0) were selected to represent the range of giant bouton cross sectional areas observed in the present study (See Table 1 for hMFB values; measured MFB values ranged from 4.47 to 13.56 lm2 for the present study). Low and high end r values correspond to giant bouton areas that would be outside the biological range observed here (too small or excessively large, respectively). Minimum cover set analysis revealed a significant difference between control and 3M SE animals when r was set to biologically relevant values (Table 1; r 5 1.4, 1.6, or 2.0). Cover fraction was significantly higher in 3M SE animals relative to control animals, indicating that thorns were more dispersed, and suggesting that a greater number of giant boutons would be required to cover all thorns present. These differences were present even though overall thorn density was statistically identical between the two groups (control, 2.59 6 0.34; 3M SE, 2.50 6 0.41; P 5 0.867, t test), leading to a second finding: The mean number of thorns per hMFB significantly decreased in 3M SE animals for r 5 1.4, 1.6 or 2.0 (if cover fraction rises while thorn density remains the same, thorns/hMFB must drop; Table 1). By contrast, significant differences vanished when r was set at values that would reflect giant boutons either too large or too small to be biologically relevant (Table 1), indicating that the findings are not arbitrary in nature.
Pyramidal Cells From Epileptic Animals are Innervated by Greater Numbers of Giant Boutons To test the prediction that altered thorn distribution in animals exposed to status reflects input by larger numbers of giant mossy fiber boutons, a method for simultaneous labeling of pyHippocampus
ramidal cell thorny excrescences and apposed granule cell giant boutons was needed. Unfortunately, the Timm stain—a reliable histochemical method used to label mossy fiber axons based on their high zinc content—is not readily adaptable to dual labeling approaches, and antibodies for brain-derived neurotrophic factor (BDNF) and neuropeptide Y, albeit promising (Scharfman et al., 2002), only labeled a subset of GFP-expressing giant boutons (Danzer et al., 2004; Danzer, unpublished observations). A novel approach for revealing giant boutons was therefore developed using antibodies targeted against the zinc transporter, ZnT-3. Grossly, Znt-3 immunostaining produced a pattern of labeling virtually identical to the Timm stain (Fig. 6A). When the approach was tested by examining GFP-labeled giant boutons and ZnT-3 immunoreactive puncta, and almost perfect correspondence was observed (Figs. 6B,C). Specifically, of 147 GFP-expressing giant boutons examined (63 control, 84 3M SE), 145 were ZnT-3 positive (98.6%). Moreover, ZnT-3 labeling roughly corresponded to the borders of the giant boutons, as revealed by the GFP label. These data confirmed that Znt-3 immunostaining can be used as a reliable marker of granule cells giant mossy fiber boutons. Casual analysis of ZnT-3 immunolabeling in GFP expressing brain sections yielded several intriguing findings. First, a tight correspondence between GFP-labeled thorns and ZnT-3 immunoreactive puncta was readily apparent (Figs. 7 and 8). This was particularly true for the elaborate thorny excrescences, which were invariably associated with ZnT-3 immunoreactive puncta. By contrast, although most isolated thorns were also associated with immunoreactive puncta (Fig. 7, middle row), this was not always the case (not shown), and 16.0% 6 5.3% of thorns from controls, and 11.3% 6 2.1% of thorns from 3M SE animals (P 5 0.368, t test) were not apposed to immunoreactive puncta. Whether these thorns reflect silent synapses, input from cell populations other than granule cells, or other factors is not clear; however, given the nearly perfect correlation between ZnT-3 labeling and mossy fiber terminals, it seems unlikely that these thorns receive mossy fiber input.
ALTERED PATTERNING OF DENTATE GRANULE CELL MOSSY FIBER INPUTS
101
tions stained with ZnT-3 antibodies. GFP-labeled thorns were identified, and the number of thorns per ZnT-3 immunoreactive puncta, and the number of puncta per dendrite was determined. While thorn density did not differ between groups (control, 2.45 6 0.21 thorns/lm; 3M SE, 2.90 6 0.38; P 5 0.350, t test), the number of ZnT-3 puncta/10 lm of dendrite was significantly increased 3 months after status (control, 2.64 6 0.38; 3M SE, 4.92 6 0.33; P < 0.001, t test). Correspondingly, with more puncta, but relatively similar numbers of thorns, the number of thorns per puncta was significantly reduced (control, 10.74 6 1.95; 3M SE, 5.09 6 0.38; P 5 0.039, Mann-Whitney RST). Finally, the cover fraction (no. puncta to cover 100 thorns) increased significantly from 14.00 6 2.71 (control) to 21.28 6 1.52 (3M SE; P 5 0.02, t test). Together, these data provide independent support for the conclusion that in epileptic animals, although overall thorn density is preserved, the pattern of innervation shifts towards larger numbers of giant boutons contacting fewer thorns.
Evidence for Cell Intrinsic Rather Than Regional Regulation of CA3 Pyramidal Cell Thorn Density
FIGURE 6. (A) Confocal maximum projection showing GFPexpressing hippocampal granule cells and CA3 pyramidal cells (green) and ZnT-3 immunolabeling (blue). (B) Confocal images of GFP-expressing giant mossy fiber boutons from control animals. Maximum projections throughout the z-depth of the GFP-expressing boutons are shown on the left, while single optical sections are shown in the middle and right columns. Note the pronounced colocalization of ZnT-3 immunoreactivity to the GFP-expressing boutons. (C) Confocal maximum projection of a large giant bouton with satellites located in stratum oriens of a 3M SE animal (top). The core bouton (core) and satellites (S) are marked accordingly. ZnT-3 immunostaining alone and merged projections through the mid-region of this bouton are shown below. Scale bar for A 5 300 lm; B and C 5 5 lm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
To estimate the number of giant boutons innervating segments of CA3 pyramidal cell dendrites with thorn densities greater than 1.0/lm, 12 GFP-expressing cells from control animals and 16 from 3M SE animals were imaged from brain sec-
Finally, we sought to explore whether neuronal plasticity among cells with high thorn densities might be regulated by cell intrinsic or regional factors. To begin to address this issue, we first compared thorn density between different branches of the same cell. CA3 pyramidal cells frequently project several dendritic branches through stratum lucidum, and these branches are often separated by tens of microns. Apical dendritic branches from the same cell, therefore, may encounter different local environments. If local factors predominate in regulating the density of thorns, different branches belonging to the same neuron may vary considerably. Alternatively, if thorn density for a given neuron is regulated by intrinsic factors, different branches would be predicted to be similar despite physical separation. To explore these different scenarios, a subset of CA3 pyramidal cells for which multiple dendritic trees were present in stratum lucidum was analyzed. Despite striking variability among pyramidal cells present in the same tissue sections (Fig. 9), thorn densities on distinct dendritic trees belonging to the same cells were highly correlated (Fig. 10). Significant correlations were evident both for neurons from control animals (R 5 0.982, P < 0.0001, Pearson Product Moment Correlation, N 5 30) and epileptic animals (1M SE, N 5 22, R 5 0.932, P < 0.0001; 3M SE, N 5 9, R 5 0.988, P < 0.0001). By contrast, regional factors were not predictive of thorn density. Density was not significantly correlated with a cell’s bregma coordinates (Paxinos and Franklin, 2001), medial-lateral position within CA3b (measured relative to the dentate granule cell layer) or soma depth within the pyramidal cell layer for either control (Supporting Information Table 1) or epileptic groups (not shown). As a last note, no obvious associations between dendritic structure and thorn density were found (Supporting Information Table 1). Taken together, these finding suggest that pyramidal cell thorn density is regulated on a cell-by-cell, Hippocampus
102
MCAULIFFE ET AL.
FIGURE 7. Pseudocolored maximum projections of GFPexpressing CA3 pyramidal cells (red) and passing granule cell mossy fiber axons (green) are shown in the left column. Middle and right columns show ZnT-3 immunolabeling and merged ZnT31GFP labeling. The z-depth represented by these images is 3.0, 2.0, and 1.2 lm from top to bottom, respectively. Arrows denote thorns colocalized with ZnT-3 immunoreactive puncta. Asterisk denotes a GFP-expressing, ZnT-3 immunoreactive giant bouton. Scale bar 5 5 lm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
rather than a regional basis. That said, the present study was deliberately designed to select cells from relatively restricted anterior-posterior positions and pyramidal cell layer subregions (to reduce variability and increase the likelihood of detecting differences between control and epileptic animals). Whether comparisons between more disparate regions (e.g., CA3a vs. CA3c) would produce different results is not known.
Status Epilepticus Does Not Alter the Distribution of GFP Expressing Cells In the present study, CA3 pyramidal cells were labeled using the Thy1-GFP mouse line. Previous studies of hippocampal granule cells suggest that seizures do not alter the pattern of GFP labeling in these animals (Danzer and McNamara, 2004; Walter et al., 2007; Danzer et al., in press). To insure that this was also true for CA3 pyramidal cells, we examined the number and distribution of labeled cells in these animals. Counts of GFP-expressing CA3b pyramidal cells in dorsal hippocampus revealed a nonsignificant decrease in pilocarpine-treated animals relative to controls (control, 0.42 6 0.08 GFP-expressing pyramids/hippocampus; 1M SE, 0.34 6 0.13; 3M SE, 0.33 6 13; P 5 0.350, ANOVA on ranks), as would be expected given the well-established sensitivity of pyramidal cells to seizure-induced death (Shibley and Smith, 2002; Borges et al., 2003). The data suggests the perhaps as many as 20% of GFP expressing CA3 pyramidal cells are lost following pilocarpine treatment (note that due to the low density of GFP expressing pyramids—perhaps 1%—detecting such a small change would require a prohibitively large number of animals. Thus, the negative finding here should be interpreted cautiously). Most importantly for the present study, the predicted direction and small size of the change suggests that seizure activity is not dramatically altering GFP expression (e.g., a 2-fold increase would be readily detectable with >80% statistical power). Hippocampus
As an additional measure to insure against the possibility of bias introduced by changes in GFP expression, the regional distribution of pyramidal cells used for thorn density counts was examined. Pyramidal cells were found at equivalent bregma coordinates (control, 22.39 6 0.04; 1M SE, 22.28 6 0.08; 3M SE, 22.34 6 0.07; P 5 0.43, ANOVA), equivalent distances from the granule cell layer (control, 544 6 31 lm; 1M SE, 506 6 38 lm; 3M SE, 549 6 47 lm; P 5 0.695, ANOVA) and equivalent depths with the pyramidal cell layer (control, 45.0 6 4.6%; 1M SE, 42.2 6 5.0%; 3M SE, 43.5 6 6.0%; P 5 0.898, Kruskal-Wallis rank sum test; values were normalized by dividing the distance of the soma from stratum lucidum by the total thickness of the pyramidal cell layer, and are expressed as percentages). In summary, although selective cell loss could contribute to the current findings; similar number and distribution of GFP labeled cells in control and treated animals supports the conclusion that Thy1 driven GFP expression is a reliable and consistent marker of CA3 pyramidal cells.
DISCUSSION For the present study, dentate granule cell giant mossy fiber bouton structure and CA3 pyramidal cell thorn density were assessed in control animals and animals examined 1 and 3 months after status epilepticus. Both the pre- and postsynaptic components of the granule cell!pyramidal cell connection exhibited striking plastic changes after status. Granule cell presynaptic giant boutons were larger and possessed satellites with significantly greater frequency at both time points. CA3 pyramidal cell thorn density, on the other hand, was reduced 1 month after status and returned to control levels at 3 months. Further examination, however, revealed that while thorns tended to be highly clustered in control animals, 3 months after status thorns were more evenly distributed, possibly reflecting a shift towards greater numbers of giant boutons innervating a given pyramidal cell via fewer thorns at each contact point. Double-label analyses of GFP-expressing thorns and ZnT-3 immunoreactive giant boutons confirmed this prediction, demonstrating a persistent rearrangement of granule cell!CA3 pyramidal cell connectivity in the epileptic brain.
Mechanisms of Synaptic Plasticity Within Stratum Lucidum Altered connectivity in the epileptic brain was demonstrated using two independent approaches (statistical modeling of thorn distribution patterns and direct observation of giant bouton-thorn relationships). The similar results produced by both approaches, and the de facto replication of the finding in separate study groups demonstrates the robust nature of these changes. The underlying cellular mechanisms, however, have yet to be fully elucidated, although several possibilities seem likely. First, the increase in granule cell giant bouton satellite
ALTERED PATTERNING OF DENTATE GRANULE CELL MOSSY FIBER INPUTS
103
FIGURE 8. Three-dimensional renderings of portions of CA3 apical dendrites from control animals and animals examined 3 months after status (3M SE). GFP expression (green) reveals the apical dendrites and thorny excrescences, while ZnT-3 immunostaining (red) reveals mossy fiber terminals. Panels show 3D renderings of dendrites alone (left), followed by the same stretches of dendrite colabeled with anti-ZnT-3 antibodies. These merged images show
rotations around the Y axis at 458 intervals. Note the low density of ZnT-3 immunoreactive puncta associated with the control dendrite relative to dendrites from 3M SE animals. Also note the large thorny excrescence (>10 lm) in the middle series, and the multiple puncta associated with it. Grid sizes are 5 lm for the top and bottom series, and 1 lm for the middle series. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
frequency almost certainly contributes to the rise in the density of boutons innervating CA3 pyramids. The percentage of giant boutons with satellites increased from around 2% in control animals, to nearly 20% in animals examined 3 months after status (Fig. 2). Extending 10’s of microns from the core giant bouton (see Fig. 6C), these satellites could easily innervate either the same pyramidal cell as the core bouton, or neighboring pyramidal cells. Moreover, since the spacing between giant boutons/bouton with satellites is similar in control and epileptic animals in CA3b (roughly every 100 lm; Danzer et al., in press); an increase in the interval between these structures would not appear to offset the increase in satellites at each point. Giant mossy fiber boutons stimulate the formation of thorny excrescences (Amaral and Dent, 1981; Nowakowski and Davis, 1985; Represa et al., 1991; Gaiarsa et al., 1992; Robain et al., 1994), so satellite formation would be predicted to lead to de novo formation of thorns. Indeed, it is tempting to speculate that this process contributes to the recovery in thorn density evident between 1 and 3 months after status. A second contributing process may be granule cell neurogenesis. Status epilepticus is a potent neurogeneic stimuli, leading to the production of large numbers of new granule cells (Parent et al., 2006; Parent, 2007). Many of these new granule cells survive
and integrate into the hippocampus (van Praag et al., 2002; Espo´sito et al., 2005; Laplagne et al., 2006; Toni et al., 2008), forming giant boutons and innervating CA3 pyramids. Third, thorn retraction (and formation) may be important (Drakew et al., 1996; Jiang et al., 1998; Swann et al., 2000). Altered thorn distribution but preserved thorn densities may reflect thorn loss in some regions of the dendritic tree and new thorn formation in other regions. Finally, a role for selective death of CA3 pyramids cannot be excluded. The pilocarpine-status epilepticus model results in some CA3 pyramidal cell death (Mello et al., 1993; Shibley and Smith, 2002; Borges et al., 2003; Danzer et al., in press; Zhang et al., 2009), and it is conceivable that pyramidal cells with highly clustered thorns, for example, are more vulnerable to excitotoxic injury. Loss of cells with specific anatomical features could produce a shift in the observed properties of the surviving population.
Significance of Altered Patterns of Pyramidal Cell Innervation by Dentate Granule Cells The present study demonstrates an increase in the density of giant boutons contacting pyramidal cells dendrites, and a corresponding decrease in the number of thorns contacted by each Hippocampus
104
MCAULIFFE ET AL. inputs from granule cells that were already innervating the pyramid in question. In the former case, increasing the number of granule cells converging onto a given pyramidal cell might be expected to dramatically alter the gating and/or information processing capabilities of the hippocampus. A change of this nature could promote seizure spread through the hippocampus by creating high throughput pathways, or by recruiting pyrami-
FIGURE 9. Adjacent CA3b pyramidal cells can exhibit dramatically different thorn densities. (A) Pseudocolored confocal maximum projection showing GFP-labeling in the hippocampus of a control Thy1-GFP-expressing mouse. Dentate granule cells (dg) and CA1 pyramidal cells (CA1) are labeled in addition to CA3 pyramidal cells in these animals. Blue coloring highlights stratum lucidum, the projection pathway of granule cell mossy fiber axons and the region where CA3 pyramidal cells exhibit thorny excrescences. Arrows denote two adjacent CA3b pyramidal cells, which are shown at high resolution in the reconstruction presented in B. (B) Neuronal reconstructions of adjacent CA3b pyramidal cell apical dendrites with sparse (left) and dense (right) accumulations of thorns. Examples of simple thorny excrescences on both cells are denoted by arrowheads, while the bracket denotes a group of complex thorny excrescences (right). Scale bars: A 5 300 lm; B 5 10 lm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
bouton. The resulting preservation of overall thorn density is intriguing, and raises the possibility that the shift reflects homeostatic mechanisms aimed at maintaining normal activity levels. Notably, however, the shift implies a reduction in thorny excrescence complexity in these animals, and this complexity may have functional significance. Indeed, Reid et al., (2001) and Reid (2002) have suggested that the complexity of thorny excrescences gives the granule cell!CA3 pyramidal cell synapse a broad dynamic range relative to simple spine synapses, and the reduction in the number of thorns contacted by each bouton—by about 50% here—would presumably reduce this range, perhaps limiting normal plasticity at this synapse. Consistent with this idea, slices from epileptic animals exhibit a loss of paired-pulse facilitation, augmentation and long-term potentiation (LTP) (Goussakov et al., 2000). Another question raised by the present findings is whether increased density of innervating giant boutons reflects novel inputs from previously unconnected granule cells, or additional Hippocampus
FIGURE 10. The density of thorns along different branches of the same dendritic tree is highly correlated. Images show neuronal reconstructions of two different pyramidal cells, one with a high density of thorns (top, control) and one with low density (bottom, 3M SE). Arrows denote thorny excrescences with single thorns in the lower image. Note that even branches separated by 10’s of microns exhibit similar thorn densities. Average thorn density for each cell is given in the lower left corner of each image. Scale bar 5 10 lm. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
ALTERED PATTERNING OF DENTATE GRANULE CELL MOSSY FIBER INPUTS dal cells into an epileptic circuit. In the latter case, granule cells that form satellites would use these structures to innervate their target pyramidal cell at multiple points. This contrasts with the control situation, in which most granule cells innervate target CA3 pyramids one time, at one point (via their only giant bouton in the region; Acsa´dy et al., 1998). Whether single inputs to large numbers of thorns are more or less effective than multiple inputs to fewer thorns is currently unknown. Clearly, determining the functional impact of the plastic changes identified here, and whether they reflect homeostatic processes, pathological changes or some combination thereof will ultimately require direct physiological measurements. Nonetheless, it is worth pointing out that the epileptic brain is characterized by hyperexcitability and persistent deficits in hippocampal dependant learning tasks (Leite et al., 1990; Chauvie`re et al., 2009; Mu¨ller et al., 2009), which suggests that homeostatic processes, if operable, are only partially effective.
CA3 Pyramidal Cells Exhibit Striking Variability in Mossy Fiber Innervation An unexpected finding of the present study was the profound variability in the density of thorns exhibited even by adjacent pyramidal cells (Fig. 9). This variability is remarkable given the relative homogeneity of spine densities exhibited by other neuronal types. Dentate granule cell spine densities, for example, vary by a factor of less than 10 from neuron to neuron (Pun, Murphy and Danzer, unpublished observations). In contrast, thorn density varied by more than a factor of 300 for control neurons (range 0.013 to 4.668 thorns/lm; all branch orders combined). This variability has not been previously appreciated, perhaps due to the difficulty with older approaches in quantifying excrescences. It is also possible that pyramidal cells with particularly low densities of excrescences observed here are specific to mouse. Studies in primates (Frotscher et al., 1988; Buckmaster and Amaral, 2001) suggest that thorny excrescences in these higher mammals are more complicated than those described in rat (Fitch et al., 1989), and rat excrescences may similarly exceed mouse excrescences in complexity. Finally, although we were unable to identify any morphological or regional features that might account for the variability in thorn density among neurons, it is notable that (1) adjacent pyramidal cells can exhibit widely varying excrescence densities and (2) adjacent branches of the same pyramidal cell exhibit highly correlated densities. These observations imply that mossy fibers preferentially innervate certain pyramidal cells, while avoiding even closely adjacent cells. A further implication of this observation is that there are significant functional differences among pyramidal cells, although physiological studies will be required to fully explore this idea.
SUMMARY The cellular basis of temporal lobe epilepsy, and the nature of the disruptions leading to comorbid conditions in epilepsy,
105
such as cognitive impairment, have not been fully elucidated. Although changes occurring within the dentate gyrus (e.g., mossy fiber sprouting) have been known for decades (Tauck and Nadler, 1985; Sutula et al., 1989; Nadler, 2003), plasticity of granule cell!CA3 pyramidal cell connections has been less intensely explored. Here, we demonstrate a remarkable restructuring of granule cell!CA3 pyramidal cell contacts. Given that this is a pivotal synapse in the classic hippocampal ‘‘trisynaptic circuit’’ we predict that these changes may be indicative of important functional changes occurring within the hippocampus.
Acknowledgments The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Neurological Disorders and Stroke or the National Institutes of Health. Thy1-GFP mice were generously provided by Dr. Guoping Feng (Duke University, Durham, NC). The authors thank Keri Kaeding for useful comments on earlier versions of this manuscript.
REFERENCES Acsa´dy L, Kamondi A, Sı´k A, Freund T, Buzsa´ki G. 1998. GABAergic cells are the major postsynaptic targets of mossy fibers in the rat hippocampus. J Neurosci 18:3386–3403. Amaral DG. 1978. A Golgi study of cell types in the hilar region of the hippocampus in the rat. J Comp Neurol 182(4, Part 2):851–914. Amaral DG, Dent JA. 1981. Development of the mossy fibers of the dentate gyrus: I. A light and electron microscopic study of the mossy fibers and their expansions. J Comp Neurol 195:51–86. Blackstad TW, Kjaerheim A. 1961. Special axo-dendritic synapses in the hippocampal cortex: Electron and light microscopic studies on the layer of mossy fibers. J Comp Neurol 117:133–159. Borges K, Gearing M, McDermott DL, Smith AB, Almonte AG, Wainer BH, Dingledine R. 2003. Neuronal and glial pathological changes during epileptogenesis in the mouse pilocarpine model. Exp Neurol 182:21–34. Buckmaster PS, Amaral DG. 2001. Intracellular recording and labeling of mossy cells and proximal CA3 pyramidal cells in macaque monkeys. J Comp Neurol 430:264–281. Chauvie`re L, Rafrafi N, Thinus-Blanc C, Bartolomei F, Esclapez M, Bernard C. 2009. Early deficits in spatial memory and theta rhythm in experimental temporal lobe epilepsy. J Neurosci 29: 5402–5410. Chicurel ME, Harris KM. 1992. Three-dimensional analysis of the structure and composition of CA3 branched dendritic spines and their synaptic relationships with mossy fiber boutons in the rat hippocampus. J Comp Neurol 325:169–182. Claiborne BJ, Amaral DG, Cowan WM. 1986. A light and electron microscopic analysis of the mossy fibers of the rat dentate gyrus. J Comp Neurol 246:435–458. Danzer SC, McNamara JO. 2004. Localization of brain-derived neurotrophic factor to distinct terminals of mossy fiber axons implies regulation of both excitation and feedforward inhibition of CA3 pyramidal cells. J Neurosci 24:11346–11355. Danzer SC, Kotloski RJ, Walter C, Hughes M, McNamara JO. 2008. Altered morphology of hippocampal dentate granule cell presynaptic and postsynaptic terminals following conditional deletion of TrkB. Hippocampus 18:668–678. Hippocampus
106
MCAULIFFE ET AL.
Danzer SC, He XP, Loepke AW, McNamara JO. Structural plasticity of dentate granule cell presynaptic terminals during the development of limbic epilepsy. Hippocampus 10.1002/hipo.20589. Drakew A, Mu¨ller M, Ga¨hwiler BH, Thompson SM, Frotscher M. 1996. Spine loss in experimental epilepsy: Quantitative light and electron microscopic analysis of intracellularly stained CA3 pyramidal cells in hippocampal slice cultures. Neuroscience 70:31– 45. Espo´sito MS, Piatti VC, Laplagne DA, Morgenstern NA, Ferrari CC, Pitossi FJ, Schinder AF. 2005. Neuronal differentiation in the adult hippocampus recapitulates embryonic development. J Neurosci 25:10074–10086. Feng G, Mellor RH, Bernstein M, Keller-Peck C, Nguyen QT, Wallace M, Nerbonne JM, Lichtman JW, Sanes JR. 2000. Imaging neuronal subsets in transgenic mice expressing multiple spectral variants of GFP. Neuron 28:41–51. Fitch JM, Juraska JM, Washington LW. 1989. The dendritic morphology of pyramidal neurons in the rat hippocampal CA3 area. I. Cell types. Brain Res 479:105–114. Frotscher M, Kraft J, Zorn U. 1988. Fine structure of identified neurons in the primate hippocampus: A combined Golgi/EM study in the baboon. J Comp Neurol 275:254–270. Galimberti I, Gogolla N, Alberi S, Santos AF, Muller D, Caroni P. 2006. Long-term rearrangements of hippocampal mossy fiber terminal connectivity in the adult regulated by experience. Neuron 50:749–763. Gaiarsa JL, Beaudoin M, Ben-Ari Y. 1992. Effect of neonatal degranulation on the morphological development of rat CA3 pyramidal neurons: Inductive role of mossy fibers on the formation of thorny excrescences. J Comp Neurol 321:612–625. Goussakov IV, Fink K, Elger CE, Beck H. 2000. Metaplasticity of mossy fiber synaptic transmission involves altered release probability. J Neurosci 20:3434–3441. Gonzales RB, DeLeon Galvan CJ, Rangel YM, Claiborne BJ. 2001. Distribution of thorny excrescences on CA3 pyramidal neurons in the rat hippocampus. J Comp Neurol 430:357–368. Henze DA, Wittner L, Buzsa´ki G. 2002. Single granule cells reliably discharge targets in the hippocampal CA3 network in vivo. Nat Neurosci 5:790–795. Howell K, Hopkins N, Mcloughlin P. 2002. Combined confocal microscopy and stereology: A highly efficient and unbiased approach to quantitative structural measurement in tissues. Exp Physiol 87:747–756. Hsu. 2007. The dentate gyrus as a filter or gate: A look back and a look ahead. Prog Brain Res 163:601–613. Ishizuka N, Weber J, Amaral DG. 1990. Organization of intrahippocampal projections originating from CA3 pyramidal cells in the rat. J Comp Neurol 295:580–623. Ishizuka N, Cowan WM, Amaral DG. 1995. A quantitative analysis of the dendritic organization of pyramidal cells in the rat hippocampus. J Comp Neurol 362:17–45. Jiang M, Lee CL, Smith KL, Swann JW. 1998. Spine loss and other persistent alterations of hippocampal pyramidal cell dendrites in a model of early-onset epilepsy. J Neurosci 18:8356–8368. Laplagne DA, Espo´sito MS, Piatti VC, Morgenstern NA, Zhao C, van Praag H, Gage FH, Schinder AF. 2006. Functional convergence of neurons generated in the developing and adult hippocampus. PLoS Biol 4:e409. Lauer M, Senitz D. 2006. Dendritic excrescences seem to characterize hippocampal CA3 pyramidal neurons in humans. J Neural Transm 113:1469–1475. Lawrence JJ, McBain CJ. 2003. Interneuron diversity series: Containing the detonation—Feedforward inhibition in the CA3 hippocampus. Trends Neurosci 26:631–640. Leite JP, Nakamura EM, Lemos T, Masur J, Cavalheiro EA. 1990. Learning impairment in chronic epileptic rats following pilocarpine-induced status epilepticus. Braz J Med Biol Res 23:681–683. Hippocampus
Li XG, Somogyi P, Ylinen A, Buzsa´ki G. 1994. The hippocampal CA3 network: An in vivo intracellular labeling study. J Comp Neurol 339:181–208. Lorente de No´ R. 1934. Studies on the structure of the cerebral cortex. II. Continuation of the study of the ammonic system. J Psychol Neurol (Lpz) 46:113–177. Mello LE, Cavalheiro EA, Tan AM, Kupfer WR, Pretorius JK, Babb TL, Finch DM. 1993. Circuit mechanisms of seizures in the pilocarpine model of chronic epilepsy: Cell loss and mossy fiber sprouting. Epilepsia 34:985–995. Mu¨ller CJ, Gro¨ticke I, Bankstahl M, Lo¨scher W. 2009. Behavioral and cognitive alterations, spontaneous seizures, and neuropathology developing after a pilocarpine-induced status epilepticus in C57BL/ 6 mice. Exp Neurol 219:284–297. Nadler JV. 2003. The recurrent mossy fiber pathway of the epileptic brain. Neurochem Res 28:1649–1658. Nowakowski RS, Davis TL. 1985. Dendritic arbors and dendritic excrescences of abnormally positioned neurons in area CA3c of mice carrying the mutation ‘‘hippocampal lamination defect’’. J Comp Neurol 239:267–275. Parent JM. 2007. Adult neurogenesis in the intact and epileptic dentate gyrus. Prog Brain Res 163:529–540. Parent JM, Elliott RC, Pleasure SJ, Barbaro NM, Lowenstein DH. 2006. Aberrant seizure-induced neurogenesis in experimental temporal lobe epilepsy. Ann Neurol 59:81–91. Paxinos G, Franklin KB. 2001. The Mouse Brain in Stereotaxic Coordinates. London: Academic Press. Peterson DA. 1999. Quantitative histology using confocal microscopy: Implementation of unbiased stereology procedures. Methods 18:493–507. Pierce JP, Milner TA. 2001. Parallel increases in the synaptic and surface areas of mossy fiber terminals following seizure induction. Synapse 39:249–256. Reid CA. 2002. The role of dendritic spines: Comparing the complex with the simple. Eur J Pharmacol 447:173–176. Reid CA, Fabian-Fine R, Fine A. 2001. Postsynaptic calcium transients evoked by activation of individual hippocampal mossy fiber synapses. J Neurosci 21:2206–2214. Represa A, Dessi F, Beaudoin M, Ben-Ari Y. 1991. Effects of neonatal gamma-ray irradiation on rat hippocampus—I. Postnatal maturation of hippocampal cells. Neuroscience 42:137–150. Robain O, Barbin G, Billette de Villemeur T, Jardin L, Jahchan T, Ben-Ari Y. 1994. Development of mossy fiber synapses in hippocampal slice culture. Brain Res Dev Brain Res 80:244–250. Rollenhagen A, Sa¨tzler K, Rodrı´guez EP, Jonas P, Frotscher M, Lu¨bke JH. 2007. Structural determinants of transmission at large hippocampal mossy fiber synapses. J Neurosci 27:10434–10444. Scharfman HE, Sollas AL, Smith KL, Jackson MB, Goodman JH. 2002. Structural and functional asymmetry in the normal and epileptic rat dentate gyrus. J Comp Neurol 454:424–439. Seress L, Ribak CE. 1995. Postnatal development of CA3 pyramidal neurons and their afferents in the Ammon’s horn of rhesus monkeys. Hippocampus 5:217–231. Shibley H, Smith BN. 2002. Pilocarpine-induced status epilepticus results in mossy fiber sprouting and spontaneous seizures in C57BL/6 and CD-1 mice. Epilepsy Res 49:109–120. Stewart MG, Davies HA, Sandi C, Kraev IV, Rogachevsky VV, Peddie CJ, Rodriguez JJ, Cordero MI, Donohue HS, Gabbott PL, Popov VI. 2005. Stress suppresses and learning induces plasticity in CA3 of rat hippocampus: A three-dimensional ultrastructural study of thorny excrescences and their postsynaptic densities. Neuroscience 131:43–54. Sutula T, Cascino G, Cavazos J, Parada I, Ramirez L. 1989. Mossy fiber synaptic reorganization in the epileptic human temporal lobe. Ann Neurol 26:321–330. Swann JW, Al-Noori S, Jiang M, Lee CL. 2000. Spine loss and other dendritic abnormalities in epilepsy. Hippocampus 10:617–625.
ALTERED PATTERNING OF DENTATE GRANULE CELL MOSSY FIBER INPUTS Tauck DL, Nadler JV. 1985. Evidence of functional mossy fiber sprouting in hippocampal formation of kainic acid-treated rats. J Neurosci 5:1016–1022. Toni N, Laplagne DA, Zhao C, Lombardi G, Ribak CE, Gage FH, Schinder AF. 2008. Neurons born in the adult dentate gyrus form functional synapses with target cells. Nat Neurosci 11:901– 907. Tsurugizawa T, Mukai H, Tanabe N, Murakami G, Hojo Y, Kominami S, Mitsuhashi K, Komatsuzaki Y, Morrison JH, Janssen WG, Kimoto T, Kawato S. 2005. Estrogen induces rapid decrease in dendritic thorns of CA3 pyramidal neurons in adult male rat hippocampus. Biochem Biophys Res Commun 337: 1345–1352. van Praag H, Schinder AF, Christie BR, Toni N, Palmer TD, Gage FH. 2002. Functional neurogenesis in the adult hippocampus. Nature 415:1030–1034.
107
Vuksic M, Del Turco D, Bas Orth C, Burbach GJ, Feng G, Mu¨ller CM, Schwarzacher SW, Deller T. 2008. 3D-reconstruction and functional properties of GFP-positive and GFP-negative granule cells in the fascia dentata of the Thy1-GFP mouse. Hippocampus 18:364–375. Walter C, Murphy BL, Pun RY, Spieles-Engemann AL, Danzer SC. 2007. Pilocarpine-induced seizures cause selective time-dependent changes to adult-generated hippocampal dentate granule cells. J Neurosci 27:7541–7552. Williams RW, Rakic P. 1988. Three-dimensional counting: An accurate and direct method to estimate numbers of cells in sectioned material. J Comp Neurol 278:344–352. Zhang S, Khanna S, Tang FR. 2009. Patterns of hippocampal neuronal loss and axon reorganization of the dentate gyrus in the mouse pilocarpine model of temporal lobe epilepsy. J Neurosci Res 87: 1135–1149.
Hippocampus
HIPPOCAMPUS 21:108–119 (2011)
Proteomics Reveal Rat Hippocampal Lateral Asymmetry A. Samara,1,2* K. Vougas,3 A. Papadopoulou,3 E. Anastasiadou,4 N. Baloyanni,3 E. Paronis,5 G.P. Chrousos,1,2 and G.Th. Tsangaris3 ABSTRACT: Brain laterality has been observed in animals and humans structurally, functionally, and behaviorally. MRI and CT scans have revealed pathological and normal brain asymmetry. A coarse assessment of rat or human brain fails to expose profound left/right differences, while a finer examination of its structure reveals an array of asymmetric features. This lateralization may be derived from evolutionary, genetic, developmental, epigenetic, and pathologic factors. However, brain structure and function is complex and macroscopic or microscopic asymmetries may be hard to discern from random fluctuations. This study concentrated on the hippocampus and we explored laterization employing a molecular high-throughput approach. Using proteomic analysis based on a combined approach of 2-D PAGE and MS, we examined differential protein expression in the hippocampi (left vs. right) of young adult male rats. Initial proteomic analysis demonstrated quantitative differences of approximately eighty proteins between the right (RH) and left hippocampus (LH). These were primarily energy-, cell metabolism-, stress-inducible chaperone proteins and cytoskeleton- proteins. Analysis revealed higher abundance of metabolic enzymes related to cellular energy metabolism, in the RH than the LH. In contrast, higher concentrations of proteins which are located mainly in astrocytes were shown in the LH than the RH. Immunoblotting of brain-specific proteins, on single animal hippocampal lysates confirmed the expression of Dynamin-1, DRP2, synapsin-1 and others, to be higher in the RH than LH lysates. These findings demonstrate major laterality in the expression of protein molecules between the two hippocampi providing a fertile field for mapping studies relating molecular, neuroimaging and functional data. Undoubtedly, asymmetries found at the animal level are hard to extrapolate to humans; however, studies in animal models will increase our understanding of the developing and adult brain and the healthy and diseased brain. V 2009 Wiley-Liss, Inc. C
KEY WORDS: hippocampus; asymmetry; rat; lateralization; differential proteomics; DRP2; Rip; DHCR24; alpha internexin
1
Biomedical Research Foundation of the Academy of Athens (BRFAA), Clinical Research Centre, Laboratory of Endocrinology and Metabolism, 4, Soranou Efesiou Street, 11527 Athens, Greece; 2 Aghia Sofia Children’s Hospital, 1st Department of Pediatrics, University of Athens, Greece; Thivon & Livadias str, Goudi; 11528 Athens, Greece; 3 BRFAA, Proteomics Research Unit; 4, Soranou Efesiou Street, 11527 Athens, Greece; 4 BRFAA, Basic Research Centre II, Laboratory of Genetics; 4, Soranou Efesiou Street, 11527 Athens, Greece; 5 BRFAA, Experimental Surgery Centre, 4, Soranou Efesiou Street, 11527 Athens, Greece Additional Supporting Information may be found in the online version of this article. *Correspondence to: Athina Samara, PhD, BRFAA, Clinical Research Centre, 4, Soranou Efessiou Street, 11527, Athens, Greece. E-mail:
[email protected] Accepted for publication 21 September 2009 DOI 10.1002/hipo.20727 Published online 17 December 2009 in Wiley Online Library (wileyonlinelibrary.com). C 2009 V
WILEY-LISS, INC.
INTRODUCTION Structural brain asymmetry has been extensively documented since 1968, when a postmortem human brain study revealed that the left planum temporale was larger than the right one (Geschwind and Levitsky, 1968). Brain functional asymmetry has also been well known mostly for handedness and hemispheric language specializations (Toga and Thompson, 2003). Imaging technology has further enabled in vivo studies to measure asymmetry. Thus, pathologic and normal brain asymmetry may be determined and quantified by magnetic resonance imaging (MRI), positron emission tomography (PET), and computer assisted tomography (CT) scans. The results of this type of morphometric imagery suggest volumetric, cross sectional, surface areas, and length interhemispheric differences pointing to brain lateral asymmetry. Moreover, asymmetry in the healthy brain needs to be fully comprehended to understand developmental changes, and pathologic findings, such as those seen in bipolar disorder, epilepsy, schizophrenia, posttraumatic stress disorder, Parkinson’s disease, Alzheimer’s disease, and other brain disorders. Brain asymmetry and lateralization has also been observed at the biochemical level. Examples include the asymmetric distribution of GABA binding sites in the cerebral cortex, hippocampus, cerebellar hemispheres, striatum, and thalamus (Oke et al., 1980), the hippocampal nitric oxide system with its right/left lateralization (Kristofikova et al., 2008), the dopaminergic enrichment of the right brain (Afonso et al., 1993) and the increased expression of D2 receptors on the left striatum (Schneider et al., 1982). In addition, there is the asymmetric distribution of endogenous diacylglycerol in rat cerebral hemispheres (Ginobili de Martinez et al., 1986) with deacylation and reacylation of complex lipids being more active in the right than the left hemisphere (Ginobili de Martinez et al., 1985). The hippocampus is involved in many neural disorders. Studies of humans with unilateral hippocampal damage propose fine functional differences, as lesions in the left hippocampus (LH) affect verbal memory, whereas lesions in the right hippocampus (RH) result in deficits of nonverbal memory tasks (Milner, 1972). A study of healthy individuals pointed out that regarding hippocampal volume asymmetry the RH was larger than the LH (Pruessner et al., 2000).
PROTEOMICS REVEAL RAT HIPPOCAMPAL LATERAL ASYMMETRY Another analysis of hippocampal asymmetry based on visualization and statistical characterization techniques, revealed that the RH was wider than the LH along its lateral side (Wang et al., 2001). Hence, the bulk of the literature supports a rightward hippocampal asymmetry. Regarding biochemical findings, in the case of the hippocampus, a high-affinity choline uptake (HACU) system directly associated with a synthesis of acetylcholine (ACh) in the hippocampus of rats was found in the LH compared to the RH of adult male (but not female) animals (Kristofikova et al., 2004). In a study of hippocampal corticoid receptors, and regardless of behavioral lateralization, there was also a tendency for a right dominance in mineralocorticoid receptor binding capacity (Neveu et al., 1998). Although an early study in the micro-anatomy of the rat hippocampus by the use of combined microdensitometric and quantitative image analysis techniques, revealed lack of morphological asymmetry (Niglio et al., 1990), there is accumulating data on neurochemical and anatomic brain asymmetries in rats, mice, and other noprimate studies (Denenberg et al., 1978; Kolb et al., 1982; Dewberry et al., 1986; Molodtsova, 1999; Kawakami et al., 2003; Halpern et al., 2005). The developmental origins of brain lateralization remain unknown. Early studies of asymmetry in the cross-sectional width of the hippocampus reported that the RH was significantly thicker than the LH in young male rats by about 8% (Diamond et al., 1982, 1983); a difference decreased with age until the asymmetry was no longer statistically significant (at postnatal Day 90). Using the rat as a model system, we isolated left and right hippocampi from normal healthy young adult rats (8–10 weeks old) for highthroughput proteomic analysis and identified the proteins differentially expressed between the two hippocampi.
MATERIAL METHODS Animals and Hippocampal Dissection Young adult male Sprague Dawley rats used for all experiments were age-matched (8–10 weeks old). We used both pooled hippocampi from littermates (8), and unrelated animals; one pool of 14 animals composed by two different litters (fathered by the same male), and two other groups adding up to a total of 16 unrelated animals. Immediately after decapitation, the rat brains were removed and dissected, followed by bilateral removal of both left and right hippocampus. Dissected hippocampi were washed with normal saline solution and stored at 2808C till further processing. When tissues were used for 2DE the hippocampi were (each) suspended in 50-ll sample buffer containing 20 mM Tris, 7 M urea, 2 M thiourea, 4% CHAPS, 10 mM 1,4-dithioerythritol, 1 mM EDTA, and a mixture of protease inhibitors (1 mM PMSF and 1 tablet CompleteTM (Roche Diagnostics, Basel, Swiss) per 50 ml of suspension buffer) and phosphatase inhibitors (0.2 mM Na2VO3 and 1 mM NaF). The suspension was sonicated for !40 s and centrifuged at 150,000g for 20 min. The protein content in the supernatant was determined using the Bradford method. The protein concentration was !5 mg per hippocampus.
109
Two-Dimensional Gel Electrophoresis (2DE) The 2D-E was performed as previously described (Fountoulakis et al., 2005). Briefly 1 mg total protein was applied by the noncup technique on 17 cm immobilized pH 3–10 and pH 4–7 nonlinear gradient IPG strips (Bio-Rad, Hercules, CA) and electrophoresed for 100 kVh. For the two-dimensional electrophoresis 12% SDS polyacrylamide gels were used in a Proteiner apparatus (Bio-Rad) and the gels were stained with colloidal Coomassie Blue (Novex, San Diego, CA). Alternatively, 400 lg of total protein were electrophoresed at the aforementioned conditions, but the gels were finally Silver stained (Invitrogen Carlsbad, CA). After homogenization, the protein content was determined using the EXPERION automated electrophoresis station in combination with the Protein 260 Analysis Kit (Bio-Rad Laboratories, Hercules, CA) as previously described. The 2DE was performed on 17-cm immobilized pH 3–10 or pH 4–7 nonlinear gradient IPG strips (BioRad). For the two-dimensional electrophoresis 12% SDS polyacrylamide gels were used in a Proteiner apparatus (Bio-Rad). Gels were stained with colloidal Coomassie Blue (Novex, San Diego, CA) or Silver Stain (Invitrogen).
Gel Image Analysis Coomassie stained gels were scanned on a GS-800 BioRad calibrated densitometer at 400 DPI resolution. The resulting TIFF images were analyzed using the PDQuest Software v 8.0. Protein spots presenting differences in their expression level were outlined and matched, empirically (automatically and then manually) after careful examination and background subtraction. Only these selected spots where evaluated for statistically significant differentiation in expression levels using the Student’s t-test. For each spot relative spot volume was calculated dividing spot volume value by the sum of total spot volume values. Total spot volume was calculated by the software, and this referred to the sum volume of all the spots on the gel. Alpha-synuclein levels remain constant throughout the independent experiments (n 5 4). We analyzed the relative spot volumes for each protein, and we used alpha-synuclein for expression normalization purposes. In brief, we calculated the ratio of the relative spot volume for each protein of interest, to the alpha-synuclein relative spot volume. To perform the necessary calculations for that step, individual alpha synuclein relative spot volume for each gel was taken into consideration. The values of each protein’s mean density ratio given in Tables 1 and 2 represent the mean density ratio of each protein vs. the alpha synuclein mean density ratio, where the alpha-synuclein spot is considered 1.
Statistical Analysis Statistical analysis was carried out by the R-language for statistical computing. The statistical significance of each protein spot’s expression level difference across the two hippocampi (left and right)—for both relative spot volumes and relative spot volume ratios to alpha-synuclein—was calculated using Hippocampus
110
SAMARA ET AL.
TABLE 1. The Hippocampal Proteins (28) Whose Expression Differs and are More Abundantly Expressed in the Right Rat Hippocampus (RH), With Accession Number, Protein ID From SwissProt and Name, Mascot Score, Their Alpha-Synuclein Normalized Values (and Standard Deviation) as Described in the Materials and Methods Section
Accession number
Protein ID from Swiss prot
Q5XHZ0
TRAP1_RAT
Q9Z2L0
VDAC1_RAT
Q68FS2 Q6AY84 Q62952
CSN4_RAT SCRN1_RAT DPYL3_RAT
P54311
GBB1_RAT
P04764 P50516 P09951 P47819 P42669 P68370 P19527 O35179 P47942
ENOA_RAT VATA_MOUSE SYN1_RAT GFAP_RAT PURA_MOUSE TBA1A_RAT NFL_RAT SH3G2_RAT DPYL2_RAT
P12075
COX5B_RAT
Q8CAQ8 P48500 P63245 Q8CHH9 P08082 P19804 Q05982 Q3T1K5 P07335 P97532 P50398 O35331 Q07266 Q6P9V9 P62870 P02688 P21575 P49432
IMMT_MOUSE TPIS_RAT GBLP_RAT SEPT8_MOUSE CLCB_RAT NDKB_RAT NDKA_RAT CAZA2_RAT KCRB_RAT THTM_RAT GDIA_RAT PDXK_RAT DREB_RAT TBA1B_RAT ELOB_RAT MBP_RAT DYN1_RAT ODPB_RAT
Hippocampus
Mean Mean spot spot volume volume Expression level P-value
FDR correction Status
Protein name
Score
RH
LH
Heat shock protein 75 kDa, mitochondrial [Tumor necrosis factor type 1 receptorassociated protein, TRAP-1,TNFRassociated protein 1] Voltage-dependent anion-selective channel protein 1 COP9 signalosome complex subunit 4 Secernin-1 Dihydropyrimidinase-related protein 3 [Collapsin response mediator protein 4, CRMP-4] Guanine nucleotide-binding protein G(I)/ G(S)/G(T) subunit beta-1 Alpha-enolase Vacuolar ATP synthase catalytic subunit A Synapsin-1 Glial fibrillary acidic protein Transcriptional activator protein Pur-alpha Tubulin alpha-1A chain Neurofilament light polypeptide SH3-containing GRB2-like protein 2 Dihydropyrimidinase-related protein 2, [Collapsin response mediator protein 2, CRMP-2] Cytochrome c oxidase subunit 5B, mitochondrial Mitochondrial inner membrane protein [Mitofilin] Triosephosphate isomerase Guanine nucleotide-binding protein subunit beta-2-like 1 Septin-8 Clathrin light chain B Nucleoside diphosphate kinase B Nucleoside diphosphate kinase A F-actin-capping protein subunit alpha-2 Creatine kinase B-type 3-mercaptopyruvate sulfurtransferase Rab GDP dissociation inhibitor alpha Pyridoxal kinase Drebrin Tubulin alpha-1B chain Transcription elongation factor B polypeptide 2 Myelin basic protein S Dynamin-1 Pyruvate dehydrogenase E1 component subunit beta, mitochondrial precursor
60
687.84
200.02
3.44
0.000025
1
0.0009
Pass
168
570.92
171.40
3.33
0.000046
2
0.0019
Pass
159 116 88
1382.20 2200.20 593.42
397.13 590.51 129.44
3.48 3.73 4.58
0.000067 0.000141 0.000143
3 4 5
0.0028 0.0037 0.0046
Pass Pass Pass
80
1059.63
347.08
3.05
0.000332
6
0.0056
Pass
151 174 152 273 87 70 197 101 182
1806.54 942.06 477.36 2239.19 2626.30 506.24 2899.34 2817.11 1125.30
581.67 265.02 127.82 872.76 813.68 174.82 965.33 753.21 327.16
3.11 3.55 3.73 2.57 3.23 2.90 3.00 3.74 3.44
0.000386 0.000469 0.000474 0.000523 0.000679 0.000716 0.000722 0.000822 0.001101
7 8 9 10 11 12 13 14 15
0.0065 0.0074 0.0083 0.0093 0.0102 0.0111 0.0120 0.0130 0.0139
Pass Pass Pass Pass Pass Pass Pass Pass Pass
64
640.20
161.49
3.96
0.001170 16
0.0148
Pass
117
2674.08
561.41
4.76
0.001774 17
0.0157
Pass
166 58
1201.85 830.17 2174.81 1098.29
1.45 1.98
0.002533 18 0.002707 19
0.0167 0.0176
Pass Pass
153 75 125 91 110 71 86 64 147 162 137 97
1653.90 755.74 1971.28 1261.45 730.65 2243.34 268.30 383.20 1410.80 2263.80 697.40 627.37
580.11 229.56 810.37 405.32 245.49 981.66 74.46 104.92 448.16 685.51 188.93 267.29
2.85 3.29 2.43 3.11 2.98 2.29 3.60 3.65 3.15 3.30 3.69 2.35
0.002938 0.003363 0.003778 0.003781 0.003932 0.004203 0.004256 0.004497 0.004715 0.004959 0.005191 0.005639
20 21 22 23 24 25 26 27 28 29 30 31
0.0185 0.0194 0.0204 0.0213 0.0222 0.0231 0.0241 0.0250 0.0259 0.0269 0.0278 0.0287
Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass
61 93 62
967.49 2491.50 279.98
279.07 633.91 87.52
3.47 3.93 3.20
0.005966 32 0.006614 33 0.006985 34
0.0296 0.0306 0.0315
Pass Pass Pass
PROTEOMICS REVEAL RAT HIPPOCAMPAL LATERAL ASYMMETRY TABLE 1
Accession number P48721 P16617 P62828 P62815 P60711 P16446 Q5XI73 P51635 P11980 P25113 Q5XI32 P23565 P28480 P05197 P31977 P50137 Q6PCU2 P81155 Q9Z2I9 Q6P9T8
111
(Continued)
Protein ID from Swiss prot
Mean Mean spot spot volume volume Protein name
GRP75_RAT PGK1_RAT RAN_RAT VATB2_RAT
Stress-70 protein, mitochondrial Phosphoglycerate kinase 1 GTP-binding nuclear protein Ran Vacuolar ATP synthase subunit B, brain isoform ACTB_RAT Actin, cytoplasmic 1 PIPNA_RAT Phosphatidylinositol transfer protein alpha isoform GDIR1_RAT Rho GDP-dissociation inhibitor 1 AK1A1_RAT Alcohol dehydrogenase [NADP1] KPYM_RAT Pyruvate kinase isozymes M1/M2 PGAM1_RAT Phosphoglycerate mutase 1 CAPZB_RAT F-actin-capping protein subunit beta AINX_RAT Alpha-internexin TCPA_RAT T-complex protein 1 subunit alpha EF2_RAT Elongation factor 2 EZRI_RAT Ezrin TKT_RAT Transketolase VATE1_RAT Vacuolar ATP synthase subunit E 1 VDAC2_RAT Voltage-dependent anion-selective channel protein 2 SUCB1_MOUSE Succinyl-CoA ligase [ADP-forming] betachain, mitochondrial TBB2C_RAT Tubulin beta-2C chain
Expression level P-value
FDR correction Status
Score
RH
LH
158 56 94 217
1410.98 1866.36 2553.33 1631.69
673.05 932.52 803.89 593.16
2.10 2.00 3.18 2.75
0.007008 0.007078 0.007315 0.008803
35 36 37 38
0.0324 0.0333 0.0343 0.0352
Pass Pass Pass Pass
54 59
786.74 563.95
372.11 322.48
2.11 1.75
0.010689 39 0.012987 40
0.0361 0.0370
Pass Pass
567.91 137.72 297.52 110.35 1264.95 643.95 1279.62 655.59 438.79 194.25 1841.29 422.16 821.39 360.57 496.11 262.08 1956.64 1052.64 1578.53 725.98 1181.49 332.89 1017.64 515.74
4.12 2.70 1.96 1.95 2.26 4.36 2.28 1.89 1.86 2.17 3.55 1.97
0.013712 0.013785 0.014449 0.017525 0.017785 0.017878 0.018484 0.020077 0.021262 0.021365 0.023750 0.032120
41 42 43 44 45 46 47 48 49 50 51 52
0.0380 0.0389 0.0398 0.0407 0.0417 0.0426 0.0435 0.0444 0.0454 0.0463 0.0472 0.0481
Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass Pass
72
1686.15
769.32
2.19
0.034685 53
0.0491
Pass
123
907.59
351.58
2.58
0.043091 54
0.0500
Pass
206 177 130 161 164 227 109 145 78 108 117 74
Mascot score >49 means significance P < 0.05.
Student’s t-test. The a-level was set at 0.05 hence 95% confidence. (R Development Core Team, 2008). As mentioned previously only the selected spots where evaluated for statistically significant difference in expression levels using the Student’s t test. The FDR (false discovery rate) was controlled to compensate for multiple comparisons.
Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF-MS) Peptide analysis and protein identification were performed as previously described (Fountoulakis et al., 2005). The selected spots were manually detected by Melanie 4.02 (GeneBio, Geneve Bioinformatics S.A., Geneva, Swiss) software on the gels they excised by the Proteiner SPII (Bruker Daltonics, Bremen, Germany). In the case of Coomassie stained spots they destained with 30% acetonitrile in 50 mM ammonium bicarbonate, while in the case of silver stained spots they destained with an appropriate destaining kit (Invitrogen). Destained spots were dried in a speed vacuum concentrator (MaxiDry Plus; Heto, Allered, Denmark). Each dried gel piece from the spot was rehydrated
with 5 ll of 1 mM ammonium bicarbonate containing 50 ng trypsin (Roche Diagnostics) and left in the dark overnight at room temperature. Twenty microlitre of 50% acetonitrile, containing 0.1% trifluoroacetic acid were added to each gel piece and incubated for 15 min with constant shaking. The resulting peptide mixture (1 ll) was simultaneously applied with 1 ll of matrix solution, consisting of 0.8% (Sigma-Aldrich), a-cyano-4hydroxycinnamic acid standard peptides des-Arg-bradykinin (904.4,681 Da; r-Aldrich), and adrenocorticotropic hormone fragment 18–39, (2465.1989 Da; Sigma-Aldrich) in 50% acetonitrile and 0.1% trifluoroacetic acid. Samples were analyzed for PMF with matrix-assisted laser desorption-mass spectrometry (MALDI-MS) in a time-of-flight mass spectrometer (Ultraflex II, Bruker Daltonics, Bremen, Germany). Laser shots (n 5 1,000) at intensity between 40 and 60% were collected and summarized using the FlexControl v2.2 software by Bruker. Peak list was created with Flexanalysis v2.2 software, by Bruker. Smoothing was applied with the Savitzky-Golay algorithm (width 0.2 mz, cycle number 1). S/N threshold ratio of 2.5 was allowed. SNAP algorithm was used for peak picking. Matching peptides and protein searches were performed automatically. Each spectrum was interpreted with Mascot Software v2.0 (Matrix Sciences, London, Hippocampus
112
SAMARA ET AL.
TABLE 2. The Hippocampal Proteins (13) Whose Expression Differs and are More Abundantly Expressed in the Right Rat Hippocampus (LH), With Accession Number, Protein ID From SwissProt and Name, Mascot Score, Their Alpha-Synuclein Normalized Values (and Standard Deviation) as Described in the Materials and Methods Section
Accession number
Protein ID from Swiss prot
Q13813 O88600
SPTA2_RAT HSP74_RAT
P53534 P08461
PYGB_RAT ODP2_RAT
P08081 P63018 O88989 P46462
CLCA_RAT HSP7C_RAT MDHC_RAT TERA_RAT
P35704 P34058 P82995 Q3KR86 P85108 P07323
PRDX2_RAT HS90B_RAT HS90A_RAT Q3KR86_RAT TBB2A_RAT ENOG_RAT
Mean spot volume
Mean spot volume
Protein name
Score
RH
LH
Expression level
P-value
FDR correction
Status
Spectrin alpha chain, brain Heat shock 70 kDa protein 4 [Ischemia responsive 94 kDa protein] Glycogen phosphorylase, brain form Dihydrolipoyllysine-residue acetyltransferase component of pyruvate dehydrogenase complex, mitochondial Clathrin light chain A Heat shock cognate 71 kDa protein Malate dehydrogenase, cytoplasmic Transitional endoplasmic reticulum ATPase Peroxiredoxin-2 Heat shock protein HSP 90-beta Heat shock protein HSP 90-alpha Inner membrane protein, mitochondrial Tubulin beta-2A chain Gamma-enolase (Neural enolase, Neuron-specific enolase, NSE)
463 73
127.07 123.90
502.80 357.78
3.96 2.89
0.000005 0.000010
1 2
0.0036 0.0071
Pass Pass
151 116
942.09 708.58
2558.76 2229.82
2.72 3.15
0.000051 0.000201
3 4
0.0107 0.0143
Pass Pass
77 247 81 267
530.51 223.30 117.77 352.47
1458.72 697.85 328.35 1234.65
2.75 3.13 2.79 3.50
0.000460 0.000981 0.001478 0.001556
5 6 7 8
0.0179 0.0214 0.0250 0.0286
Pass Pass Pass Pass
59 172 156 148 207 192
769.89 369.75 1088.38 286.50 496.91 171.12
2078.47 1764.90 2005.20 967.18 1271.03 290.44
2.70 4.77 1.84 3.38 2.56 1.70
0.005940 0.008730 0.014527 0.016556 0.019263 0.030378
9 10 11 12 13 14
0.0321 0.0357 0.0393 0.0429 0.0464 0.0500
Pass Pass Pass Pass Pass Pass
Mascot score >49 means significance P < 0.05.
UK). For peptide identification, the monoisotopic masses were used and a mass tolerance of 0.0025% (25ppm) was allowed. All extraneous peaks, such as trypsin autodigests, matrix, and keratin peaks, were not considered for the protein search. Cysteine carbamidomethylation and methionine oxidation were set as fixed and variable modifications, respectively. One miscleavage was allowed. The peptide masses were then compared with the theoretical peptide masses of all proteins from rodents using the SWISS-PROT and TREMBL databases. The probability score identified by the software was used as the criterion of the identification. Probability score with P < 0.05 identified by the software was used as the criterion for affirmative protein identification. The obtained mass spectra and the detected peak-lists led to protein identification are provided as supplementary material (Supporting Information Table 1).
IMMUNOBLOTTING For one dimensional gel electrophoresis, equal amounts of rat hippocampal protein lysates were resolved on 12% SDSHippocampus
polyacrylamide gels, subsequently blotted to Hybond-ECL nitrocellulose membranes (Amersham Biosciences) and blocked with Tris-buffered saline containing 5% (w/v) nonfat dry milk and 0.05% Tween-20 (TBST buffer). The membranes were then probed with primary antibodies and HRP-conjugated secondary antibodies (Chemicon) in TBST. The antibodies used were anti-beta-III-tubulin (G7121, Promega), anti- acetylated alpha-tubulin (T6793, Sigma-Aldrich), anti-Seladin-1/ DHCR24 (S4697, Sigma-Aldrich), anti-beta-actin (AC74, Sigma-Aldrich), anti-alpha-internexin (Ab7259, Abcam), antiDRP2 (Ab36201, Abcam), anti-PCNA (ab29, Abcam), antineurofilament heavy chain (ab7795, Abcam), antisynapsin-1 (clone 3C11, DSHB, IA, IA), anti-Rip (Rip, DSHB, IA, IA), antineurofilament associated protein (E/C8, DSHB, IA, IA), anti- HSP70 (sc1060, Santa Cruz), anti-ABCA2 (SC48440, Santa Cruz), anti-GFAP (Z334, DAKO), antidynamin-1 (D25520, Transduction Laboratories), and anti-alpha-synuclein (610787, BD Biosciences). Immunoreactivity was visualized by ECL chemiluminescence detection system (Santa Cruz) and exposure to X-ray film (Fuji). The films were scanned and images analyzed using quantity one image processing software (Bio-Rad).
PROTEOMICS REVEAL RAT HIPPOCAMPAL LATERAL ASYMMETRY
RESULTS We constructed high-resolution maps from the hippocampal lysates of the right and left hippocampi. The PD Quest 2D analysis software demonstrated that using a broad range pH 3–10 IPG strip for the first dimension, there were !920 spots on each RH and LH gel (Fig. 1A). We narrowed the range of IPG strips for the first dimension (pH 4–7) with equal protein loading, to increase resolution. In this pH range, PD quest identified !840 and 640 spots on the RH and LH, respectively (Fig. 1B). We thus identified and picked the differentially expressed spots (a mean of 65 spots per compared pair of gels), which were then gel-excised and trypsin-digested. We analyzed the trypsin digest extracts with mass spectrometry and we identified 154 proteins that correspond to 68 unique gene products (all identified proteins, their spectra, and molecular weights can be found in the Supporting Information Table 1). The proteins that were more abundant in the RH are illustrated in Table 1, while Table 2 shows the proteins that we found to be overexpressed in the LH. Furthermore, we used immunoblotting to confirm some of the protein expression differences analyzed with the differential proteomics approach. Protein lysates from left and right hippocampal protein lysates were resolved by gel electrophoresis and blotted with the appropriate dilution of antibody (as described in the Materials and Methods Section). The Figure 2A shows representative western blot analyses of some of the proteins revealed by the 2D gel electrophoresis. These were the neurofilaments alpha-internexin and NFL, dynamin-1, DRP2, GFAP, HSP70, and synapsin-1. The bars (on the right) correspond to the relative quantification of the proteins (left), using alpha-synuclein as the gel loading control. Other than the hsp70 protein, the rest of the proteins showed higher levels at the RH, at steady state, in the two healthy young adult hippocampi. We normalized the values of the expression levels to alpha-synuclein because the proteomic analysis identified this protein’s expression levels to be equally expressed in both hippocampi. Moreover, we performed more immunoblottings for some proteins we chose to substantiate our rightward protein asymmetry hypothesis. These proteins were alpha- and beta-IIItubulin, neurofilament heavy chain, beta-actin, ABCA2 and DHCR24, and PCNA. Interestingly, in the case of acetylated alpha- and beta-III-tubulin more than one bands appeared. There were bands of lower molecular weight (Bands 2, 3 in both cases, and Band 4 in the case of acetylated alpha-tubulin) that were either absent or differentially expressed after overloading of sample for the immunoblotting (50 lg). In both alphaand beta-III-tubulin blots we quantified the bands annotated 1 and 3 (see Fig. 2B). Unlike the other cytoskeletal proteins, beta-actin was more abundant in the left hippocampus.
DISCUSSION This proteomic analysis study targeted the hippocampus, a region important for intermediate memory and control of the
113
stress response. We explored the hypothesis of laterization employing a detailed molecular high-throughput approach. We employed proteomic analysis based on a combined approach of 2D PAGE and MS, to examine differential protein expression in the hippocampi (left vs. right) of young adult male rats. This proteomic analysis demonstrated quantitative differences of 41 proteins between the RH and LH. These are proteins involved mainly in energy and cell metabolism pathways, transport and vesicle trafficking, cytoskeletal structure and changes, and protein processing, such as translation, folding, and degradation. In agreement with the rightward trend of hippocampal asymmetry in the literature, the results confirmed that of the 41 proteins differentially expressed between RH and LH, !68% were more abundant in the RH. The proteomic analysis identified several energy and metabolism-related enzymes that were more abundant in the right hippocampus, such as pyruvate dehydrogenase, 3-mercaptopyruvate sulfurtransferase and vacuolar ATP synthases. Glycogen production in the brain is almost exclusively localized in the astrocytes (Peters et al., 1991), and we found higher abundance of glycogen phosphorylase and malate dehydrogenase in the LH. Glycogen phorphorylase is located in astrocytes (Wiesinger et al., 1997) and influences energy transfer to neurons, and malate dehydrogenase has been reported to be more expressed in astrocytes than neurons (Lovatt et al., 2007). The higher expression of the transcriptional activator protein Pur-alpha (Gallia et al., 2000; Fountoulakis et al., 2005) in the RH, might justify a normal upregulation in transcriptional and translational activity in the RH, accounting for different cellular homeostatic dynamics between the two hippocampi. The glycolytic enzymes alpha- and gamma-enolase both appeared differentially expressed. The expression of these genes is regulated both developmentally and tissue-specifically. The embryonic and ubiquitous alpha–alpha isoform is distributed in most adult cell populations. During brain ontogenesis, a transition occurs in two tissues with high and fluctuating energy requirements: from the alpha–alpha isoform toward the specific isoforms, alpha–gamma and gamma–gamma in neurons (Zomzely-Neurath, 1982). In our study, alpha enolase was more abundant in the RH, while the gamma isoform expression was higher in the LH. The respective mean relative spot volumes were higher in the RH. On the other hand, clathrin light chain A and B were more abundant in the LH and RH, respectively, whereas, in this case, the respective mean relative spot volumes were higher in the LH. Clathrin light chains are randomly distributed in triskelions (Kirchhausen et al., 1983; Kirchhausen and Toyoda, 1993). However, a higher proportion of clathrin light chain B has been observed in cells and tissues that require and maintain a highly regulated pathway of secretion (Acton and Brodsky, 1990). It is interesting to note that neuronal activity-dependent changes in alternative splicing of clathrin light chain B has also been observed, which may contribute to neuronal plasticity (Daoud et al., 1999). Apart from plasticity, fine-tuned vesicle transport and trafficking are vital for synaptic formation. A study of cultured Hippocampus
114
SAMARA ET AL.
FIGURE 1. About 1 mg of protein lysate from left (LH) and right (RH) hippocampi of young adult rats was run in 2D electrophoresis. Bidimensional maps were constructed from the 2D gels, with the differentially expressed proteins indicated by arrows. The PD quest 2D analysis software demonstrated that using a broad range pH 3–10 IPG strip for the first dimension, there were !920 spots on both the RH and LH gels (A). Narrowing down the range of the IPG strips for the first dimension (pH 4–7), and using equal protein loading, we further increased resolution. In this pH range, PD Quest identified approximately 840 and 640 spots on the RH and LH, respectively (B). We thus identified and picked the differentially expressed spots (a mean of 65 spots per compared pair of gels), which were then gel-excised and trypsin-digested. We
Hippocampus
analyzed the trypsin digest extracts with a mass spectrometer and we identified 154 proteins that correspond to 68 unique gene products (all identified proteins, their spectra, and molecular weights can be found in the Supporting Information Table 3). A. Illustrating the proteins found to be overexpressed in the right hippocampus (RH) vs. the left hippocampus (LH), as shown in Table 1. Accordingly, B shows the proteins that were found to be overexpressed in the left hippocampus vs. the right (Table 2). The alpha-synuclein normalized values of the proteins that were found to be overexpressed in the right hippocampus vs. the left hippocampus, as shown in Table 1. Consequently, Table 2 shows the proteins that were overexpressed in the left hippocampus vs. the right.
PROTEOMICS REVEAL RAT HIPPOCAMPAL LATERAL ASYMMETRY
FIGURE 2. (A) Equal protein amounts of total protein lysates from left and right hippocampal protein lysates of young adult rats were resolved by gel electrophoresis and blotted with the appropriate dilution of antibody. The figure shows representative western blot analyses of some of the proteins revealed by the 2D gel electrophoresis. These were the neurofilaments alpha internexin and NFL, dynamin-1, DRP2, GFAP, HSP70 and synapsin-1. The bars (on the right) correspond to the relative quantification of the proteins (left), using alpha-synuclein as the gel loading control. We used alpha-synuclein because the proteomic analysis identified its expression levels to be equally expressed in the both hippocampi. (B) Equal protein amounts of total protein lysates from left and right hippocampal protein lysates of young adult rats were resolved by gel electrophoresis and blotted with the appropriate dilution of antibodies. The figure shows representative western blots of some proteins we chose to substantiate our rightward protein asymmetry hypothesis. The proteins were alpha- and beta-III-tubulin, neurofilament heavy chain, beta actin, ABCA2 and DHCR24, and PCNA. In the case of acetylated alpha- and beta-III-tubulin more than one bands appeared. There were bands of lower molecular weight (bands 2, 3 in both cases, and band 4 in the case of acetylated alpha-tubulin) that were either absent or differentially expressed after ‘‘overloading’’ of sample for the immunoblotting. In both alpha- and beta-III-tubulin blots we quantified the bands annotated 1 and 3. Unlike tubulin, the beta actin expression levels were higher on the left hippocampal lysates. The blots shown are representative of a series of experiments performed in single hippocampal lysates or lysates from pooled hippocampi of littermates or irrelevant but same-aged animals (LH and RH stand for left and right hippocampus respectively).
115
hippocampal neurons demonstrated that there were !200 vesicles per synaptic bouton (Micheva et al., 2006). Our hippocampal laterization results further suggest a rightward synaptic activity asymmetry. This was supported by the overexpression of the developmentally regulated brain protein aka drebrin (Harigaya et al., 1996), dynamin-1, clathrin light chain B, endophilin-1 (SH3-containing GRB2-like protein 2) synapsinI, secernin, and VDAC1 (Morciano et al., 2009). Septin-8, also overexpressed in the RH, is a protein enriched in the presynapses, and controls the binding of synaptobrevin VAMP2 to synaptophysin (Fukata et al., 2002). The Rab GDP dissociation inhibitor alpha, which also regulates presynaptic plasticity (Ishizaki et al., 2000) was more abundant in RH. To validate our proteomic analysis we performed immunoblotting for some of the proteins derived from the differential analysis. We normalized their expression levels to alpha-synuclein, whose expression levels were equal between the two hippocampi. Additionally, we used immunoblotting to validate that the expression of other proteins was different. These proteins were known partners or downstream activated proteins of those that were derived from this proteomic analysis. Although our results demonstrated a rightward asymmetry, we found that Hsp70 was more abundant in the LH. The higher expression of hsp70 may indicate faster response in mediating neuroprotection at steady state in the healthy left rat hippocampus. One of the proteins found more abundantly expressed on the RH, was dynamin-1. Dynamin-1 is member of the dynamin subfamily of GTP-binding proteins. It possesses unique mechanochemical properties used to tubulate and sever membranes. Dynamin-1 is neuron specific and participates in clathrin-dependent independent synaptic vesicle recycling, internalization and/or signal transduction of a variety of G-protein coupled receptors and growth factor receptors and is also involved in neuronal growth (Torre et al., 1994; McClure and Robinson, 1996; Smillie and Cousin, 2005). Increased vesicle trafficking could reflect increased synaptic activity and plasticity. Another protein that is important for synaptic plasticity is GFAP. GFAP is a marker of mature astrocytes surrounding the synapses and some of their processes interacting with synapses (Fields and Stevens-Graham, 2002). Its overexpression in the RH (demonstrated both by proteomics and western blot) might be either a sign of increased proliferation in the RH, or a change in astrocytic motility. Alternatively, it might simply be an indication of a posttranslational modification suggesting changes in synaptic plasticity. Besides, the higher expression of the neurofilament alpha-internexin in the RH (confirmed by western blot), might again be due to posttranslational modifications, or neuronal morphological changes in response to synaptic potentiation (Tononi and Cirelli, 2003). These findings agree with our observations on increased synaptic vesicle trafficking and may underline a rightward shift of synaptic plasticity in the hippocampus. Our proteomics analysis corroborated this notion, having identified drebrin being abundant in the RH. Drebrin is one of the most abundant neuron-specific F-actin-binding proteins, found exclusively in dendrites and particularly concentrated in dendritic spines Hippocampus
116
SAMARA ET AL.
receiving excitatory inputs. The immunoblots also established that the expression profile of synapsin-1, DRP2, and NFL was higher in the RH than the LH lysates. The neuronal specific protein synapsin-1 coats synaptic vesicles and binds to the cytoskeleton. A presynaptic protein that, when phosphorylated, enhances the availability of synaptic vesicles for release contributing, functionally to the hippocampal synaptic plasticity. DRP2 is a protein enriched in the postsynaptic density. It interacts with tubulin dimers and promotes microtubule assembly for axon outgrowth (Fukata et al., 2002). The higher expression of DRP2 in the RH, where synaptic formation might be enhanced, may delineate its role, its overexpression during development, and its implication in the formation of multiple axons (Goshima et al., 1995; Minturn et al., 1995; Byk et al., 1998; Inagaki et al., 2001; Kawano et al., 2005). Although we only saw significant changes in the neurofilament light polypeptide (NFL), we also performed western blots for neurofilament-heavy chain which confirmed the rightward hippocampal shift. The neurofilament proteins are key players in the maintenance and remodeling of the neuronal cytoskeleton, and it has been suggested that their changes correspond to alterations in neuronal architecture. Furthermore, we used an anti- beta-III-tubulin antibody on immunoblotting, and illustrated its quantitative overexpression on the RH. Moreover, as seen in Figure 2B, in both cases of beta-III and alpha-tubulin, we found bands of lower molecular weight in the blots that are also differentially expressed between the hippocampi. Beta-III-Tubulin is expressed in neurons and sertoli cells and constitutes !25% of brain tubulin. There is considerable tubulin heterogeneity resulting from a large tubulin gene family encoding numerous isotypic forms and also from numerous post-translational modifications (MacRae, 1997). These post-translational modifications of tubulin include acetylation, phosphorylation, tyrosination, polyglutamylation, and polyglycylation (Diaz-Nido et al., 1990; Panda et al., 1994). Further work is currently in progress, to identify these different bands appearing in immunoblots, of left and right hippocampal lysates. Besides, using the Rip oligodendrocyte marker we indirectly demonstrated the MBP abundance in the RH (Friedman et al., 1989; Berger and Frotscher, 1994). Nonetheless, stripping and reprobing the same membranes with a beta-actin specific antibody, showed LH beta-actin overexpression, while the proteomics had revealed that brain-spectrin was also overexpressed in the LH. Spectrin is another protein that interacts with actin and is highly expressed in the dentate gyrus of the hippocampus (Riederer et al., 1987). The position of the protein spot on the gel indicated that we identified a cleaved isoform of spectrin; nevertheless, our mass spectroscopy data revealed that the digests were of the full length protein. In any case, this form is not due to caspase-3 digestions since there were no other apoptosis-related proteins identified in the LH lysates. Spectrin digestion by calpain-I is an irreversible, normal brain phenomenon underlying neurite extension and synaptogenesis [review (Czogalla and Sikorski, 2005)]. Hippocampus
Additionally, we performed immunoblotting against PCNA, following the accumulating evidence on adult hippocampal neurogenesis, and showed that its expression was slightly higher in the RH. The results of immunoblotting for ABCA2 also showed that its expression was slightly higher in the RH. ABCA2 is a marker of neural progenitors as it is expressed in the subventricular zone of the lateral ventricle and the dentate gyrus of the hippocampal formation (Broccardo et al., 2006). It is also expressed in the sites of continual neurogenesis in the adult brain, and in nestin immunopositive cells differentiated in vitro from embryonic stem cells (Broccardo et al., 2006). ABCA2 is absent from astrocytes while it is expressed by oligodendrocytes (Zhou et al., 2001; it has been proposed to be a ‘‘guardian’’ of membrane composition and fluidity at the level of vesicles (Broccardo et al., 2006). It shows an intracellular localization in lysosomal-related vesicles (Broccardo et al., 2006) and the turnover of sterols relies on the endolysosomal compartment. Neuronal progenitors are largely dependent on sterols for growth and synaptogenesis (Mauch et al., 2001; Canolle et al., 2004). Taking into consideration the effect of steroids on neuronal plasticity (Ishii et al., 2007) and that cholesterol depletion in the rat hippocampus inhibits synaptic transmission and synaptic plasticity (Frank et al., 2008), we performed western blots for DHCR24/Seladin-1; the final reductase of the postsqualenic cholesterol biosynthesis pathway. DHCR24 is highly expressed in the hippocampus and associated with both neuroprotection and neurodegeneration (Greeve et al., 2000; Waterham et al., 2001; Iivonen et al., 2002; Wechsler et al., 2003; Benvenuti et al., 2005; Di Stasi et al., 2005; Lu et al., 2008; Luciani et al., 2008). Though the expression of DHCR24 was high, there was no significant variation in the expression profile of the protein between the two hippocampi of the healthy rat brains. This comparative approach of RH vs. LH analysis of healthy rats did not contain all the proteins expressed, but only those differentially expressed between the two hippocampi and those that 2D analysis detected. The 2D electrophoresis technology limitations favor results toward abundant soluble cytosolic proteins, and low-abundance protein subtle differences might have escaped detection. Low- and high-molecular mass proteins were also underrepresented in 2-D gels. Moreover, the differences detected could be due both to modifications in the total amount of proteins, and post-translational modifications, which would result in separate spots on the 2D maps. However, it is worth mentioning that 70% of the brain proteins identified from 2D gels have theoretical pI values between 5 and 8, and 15% between 4 and 5 (Fountoulakis, 2004). Thus, due to the limitations of proteomic analysis we only identified proteins mainly at the acidic part of the proteome. Our findings substantiated the significance of high throughput brain analysis approaches, to decipher the role of differential protein expression levels in brain asymmetry. Some known candidates are DRP2 in schizophrenia, dynamin-1 in amyloid plaques, drebrin (Harigaya et al., 1996) in Alzheimer’s disease. Also, heat shock protein misfolding is thought to be linked to the pathogenesis of many age-associated, neurodegenerative dis-
PROTEOMICS REVEAL RAT HIPPOCAMPAL LATERAL ASYMMETRY orders, including Parkinson’s disease, Alzheimer’s disease, Huntington’s disease, and amyotrophic lateral sclerosis (Mouradian, 2002; Rakhit et al., 2002). Changes in VDAC1 have also been documented in Alzheimer’s disease and Down syndrome patients (Yoo et al., 2001). In addition, most of the identified and brain-disorder-related proteins have been cited in recent proteomic analyses (Fountoulakis et al., 2005; Chen et al., 2006). We propose that the expression levels of a group of specific proteins, already implicated in neurodegeneration and/or other brain disorders, could be assessed as a group taking into account their lateral asymetry. Indisputably, the brain, and especially the neurons, are never at a steady state. Their biochemical and synaptic activities mirror acute and chronic changes in the brain metabolism. Our findings demonstrated a major laterality in the expression of protein molecules between the hippocampi of healthy rat brains. Further work is in progress in our laboratory to identify this steady-state molecular asymmetry in healthy young animals. We suggest a rightward synaptic asymmetry, and a neuroprotective leftward role, based on the abundance of specific proteins at steady state. We further suggest that hippocampal asymmetry is a normal variant, rather than a pathologic finding. We hope that these data will provide a novel perspective for mapping studies relating molecular, neuroimaging, and functional data. Undoubtedly, asymmetries found at the animal level are hard to extrapolate to humans; however, studies in the animals will increase our understanding of the developing, adult, and aging brain, and the healthy and diseased brain.
Acknowledgments Athina Samara would like to acknowledge the NARSAD YI2007 award, for partial financial support. The Rip, E/C8 & 3C11 (anti SYNORF1), developed by Susan Hockfield & Russell Matthews, Gary Ciment, and Erich Buchner, respectively, were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, IA, IA 52242.
REFERENCES Acton SL, Brodsky FM. 1990. Predominance of clathrin light chain LCb correlates with the presence of a regulated secretory pathway. J Cell Biol 111:1419–1426. Afonso D, Santana C, Rodriguez M. 1993. Neonatal lateralization of behavior and brain dopaminergic asymmetry. Brain Res Bull 32:11–16. Benvenuti S, Luciani P, Vannelli GB, Gelmini S, Franceschi E, Serio M, Peri A. 2005. Estrogen and selective estrogen receptor modulators exert neuroprotective effects and stimulate the expression of selective Alzheimer’s disease indicator-1, a recently discovered antiapoptotic gene, in human neuroblast long-term cell cultures. J Clin Endocrinol Metab 90:1775–1782. Berger T, Frotscher M. 1994. Distribution and morphological characteristics of oligodendrocytes in the rat hippocampus in situ and in
117
vitro: An immunocytochemical study with the monoclonal Rip antibody. J Neurocytol 23:61–74. Broccardo C, Nieoullon V, Amin R, Masmejean F, Carta S, Tassi S, Pophillat M, Rubartelli A, Pierres M, Rougon G, Nieoullon A, Chazal G, Chimini G. 2006. ABCA2 is a marker of neural progenitors and neuronal subsets in the adult rodent brain. J Neurochem 97:345–355. Byk T, Ozon S, Sobel A. 1998. The Ulip family phosphoproteins— Common and specific properties. Eur J Biochem 254:14–24. Canolle B, Masmejean F, Melon C, Nieoullon A, Pisano P, Lortet S. 2004. Glial soluble factors regulate the activity and expression of the neuronal glutamate transporter EAAC1: Implication of cholesterol. J Neurochem 88:1521–1532. Chen WQ, Kang SU, Lubec G. 2006. Protein profiling by the combination of two independent mass spectrometry techniques. Nat Protoc 1:1446–1452. Czogalla A, Sikorski AF. 2005. Spectrin and calpain: A ‘‘target’’ and a ‘‘sniper’’ in the pathology of neuronal cells. Cell Mol Life Sci 62:1913–1924. Daoud R, Da Penha Berzaghi M, Siedler F, Hubener M, Stamm S. 1999. Activity-dependent regulation of alternative splicing patterns in the rat brain. Eur J Neurosci 11:788–802. Denenberg VH, Garbanati J, Sherman DA, Yutzey DA, Kaplan R. 1978. Infantile stimulation induces brain lateralization in rats. Science 201:1150–1152. Dewberry RG, Lipsey JR, Saad K, Moran TH, Robinson RG. 1986. Lateralized response to cortical injury in the rat: Interhemispheric interaction. Behav Neurosci 100:556–562. Di Stasi D, Vallacchi V, Campi V, Ranzani T, Daniotti M, Chiodini E, Fiorentini S, Greeve I, Prinetti A, Rivoltini L, Pierotti MA, Rodolfo M. 2005. DHCR24 gene expression is upregulated in melanoma metastases and associated to resistance to oxidative stress-induced apoptosis. Int J Cancer 115:224–230. Diamond MC, Murphy GM Jr, Akiyama K, Johnson RE. 1982. Morphologic hippocampal asymmetry in male and female rats. Exp Neurol 76:553–565. Diamond MC, Johnson RE, Young D, Singh SS. 1983. Age-related morphologic differences in the rat cerebral cortex and hippocampus: Male-female; right-left. Exp Neurol 81:1–13. Diaz-Nido J, Serrano L, Lopez-Otin C, Vandekerckhove J, Avila J. 1990. Phosphorylation of a neuronal-specific beta-tubulin isotype. J Biol Chem 265:13949–13954. Fields RD, Stevens-Graham B. 2002. New insights into neuron-glia communication. Science 298:556–562. Fountoulakis M. 2004. Application of proteomics technologies in the investigation of the brain. Mass Spectrom Rev 23:231–258. Fountoulakis M, Tsangaris GT, Maris A, Lubec G. 2005. The rat brain hippocampus proteome. J Chromatogr B Analyt Technol Biomed Life Sci 819:115–129. Frank C, Rufini S, Tancredi V, Forcina R, Grossi D, D’Arcangelo G. 2008. Cholesterol depletion inhibits synaptic transmission and synaptic plasticity in rat hippocampus. Exp Neurol 212:407– 414. Friedman B, Hockfield S, Black JA, Woodruff KA, Waxman SG. 1989. In situ demonstration of mature oligodendrocytes and their processes: An immunocytochemical study with a new monoclonal antibody, rip. Glia 2:380–390. Fukata Y, Itoh TJ, Kimura T, Menager C, Nishimura T, Shiromizu T, Watanabe H, Inagaki N, Iwamatsu A, Hotani H, Kaibuchi K. 2002. CRMP-2 binds to tubulin heterodimers to promote microtubule assembly. Nat Cell Biol 4:583–591. Gallia GL, Johnson EM, Khalili K. 2000. Puralpha: A multifunctional single-stranded DNA- and RNA-binding protein. Nucleic Acids Res 28:3197–3205. Geschwind N, Levitsky W. 1968. Human brain: Left-right asymmetries in temporal speech region. Science 161:186–187. Ginobili de Martinez MS, Rodriguez de Turco EB, Barrantes FJ. 1985. Endogenous asymmetry of rat brain lipids and dominance of Hippocampus
118
SAMARA ET AL.
the right cerebral hemisphere in free fatty acid response to electroconvulsive shock. Brain Res 339:315–321. Ginobili de Martinez MS, Rodriguez de Turco EB, Barrantes FJ. 1986. Asymmetry of diacylglycerol metabolism in rat cerebral hemispheres. J Neurochem 46:1382–1386. Goshima Y, Nakamura F, Strittmatter P, Strittmatter SM. 1995. Collapsin-induced growth cone collapse mediated by an intracellular protein related to UNC-33. Nature 376:509–514. Greeve I, Hermans-Borgmeyer I, Brellinger C, Kasper D, Gomez-Isla T, Behl C, Levkau B, Nitsch RM. 2000. The human DIMINUTO/DWARF1 homolog seladin-1 confers resistance to Alzheimer’s disease-associated neurodegeneration and oxidative stress. J Neurosci 20:7345–7352. Halpern ME, Gunturkun O, Hopkins WD, Rogers LJ. 2005. Lateralization of the vertebrate brain: Taking the side of model systems. J Neurosci 25:10351–10357. Harigaya Y, Shoji M, Shirao T, Hirai S. 1996. Disappearance of actinbinding protein, drebrin, from hippocampal synapses in Alzheimer’s disease. J Neurosci Res 43:87–92. Iivonen S, Hiltunen M, Alafuzoff I, Mannermaa A, Kerokoski P, Puolivali J, Salminen A, Helisalmi S, Soininen H. 2002. Seladin-1 transcription is linked to neuronal degeneration in Alzheimer’s disease. Neuroscience 113:301–310. Inagaki N, Chihara K, Arimura N, Menager C, Kawano Y, Matsuo N, Nishimura T, Amano M, Kaibuchi K. 2001. CRMP-2 induces axons in cultured hippocampal neurons. Nat Neurosci 4:781–782. Ishii H, Tsurugizawa T, Ogiue-Ikeda M, Asashima M, Mukai H, Murakami G, Hojo Y, Kimoto T, Kawato S. 2007. Local production of sex hormones and their modulation of hippocampal synaptic plasticity. Neuroscientist 13:323–334. Ishizaki H, Miyoshi J, Kamiya H, Togawa A, Tanaka M, Sasaki T, Endo K, Mizoguchi A, Ozawa S, Takai Y. 2000. Role of rab GDP dissociation inhibitor alpha in regulating plasticity of hippocampal neurotransmission. Proc Natl Acad Sci USA 97:11587–11592. Kawakami R, Shinohara Y, Kato Y, Sugiyama H, Shigemoto R, Ito I. 2003. Asymmetrical allocation of NMDA receptor epsilon2 subunits in hippocampal circuitry. Science 300:990–994. Kawano Y, Yoshimura T, Tsuboi D, Kawabata S, Kaneko-Kawano T, Shirataki H, Takenawa T, Kaibuchi K. 2005. CRMP-2 is involved in kinesin-1-dependent transport of the Sra-1/WAVE1 complex and axon formation. Mol Cell Biol 25:9920–9935. Kirchhausen T, Toyoda T. 1993. Immunoelectron microscopic evidence for the extended conformation of light chains in clathrin trimers. J Biol Chem 268:10268–10273. Kirchhausen T, Harrison SC, Parham P, Brodsky FM. 1983. Location and distribution of the light chains in clathrin trimers. Proc Natl Acad Sci USA 80:2481–2485. Kolb B, Sutherland RJ, Nonneman AJ, Whishaw IQ. 1982. Asymmetry in the cerebral hemispheres of the rat, mouse, rabbit, and cat: The right hemisphere is larger. Exp Neurol 78:348–359. Kristofikova Z, Stastny F, Bubenikova V, Druga R, Klaschka J, Spaniel F. 2004. Age- and sex-dependent laterality of rat hippocampal cholinergic system in relation to animal models of neurodevelopmental and neurodegenerative disorders. Neurochem Res 29:671–680. Kristofikova Z, Kozmikova I, Hovorkova P, Ricny J, Zach P, Majer E, Klaschka J, Ripova D. 2008. Lateralization of hippocampal nitric oxide mediator system in people with Alzheimer disease, multiinfarct dementia and schizophrenia. Neurochem Int 53:118–125. Lovatt D, Sonnewald U, Waagepetersen HS, Schousboe A, He W, Lin JH, Han X, Takano T, Wang S, Sim FJ, Goldman SA, Nedergaard M. 2007. The transcriptome and metabolic gene signature of protoplasmic astrocytes in the adult murine cortex. J Neurosci 27:12255–12266. Lu X, Kambe F, Cao X, Kozaki Y, Kaji T, Ishii T, Seo H. 2008. 3beta-Hydroxysteroid-delta24 reductase is a hydrogen peroxide scavenger, protecting cells from oxidative stress-induced apoptosis. Endocrinology 149:3267–3273. Hippocampus
Luciani P, Deledda C, Rosati F, Benvenuti S, Cellai I, Dichiara F, Morello M, Vannelli GB, Danza G, Serio M, Peri A. 2008. Seladin-1 is a fundamental mediator of the neuroprotective effects of estrogen in human neuroblast long-term cell cultures. Endocrinology 149:4256–4266. MacRae TH. 1997. Tubulin post-translational modifications—Enzymes and their mechanisms of action. Eur J Biochem 244:265–278. Mauch DH, Nagler K, Schumacher S, Goritz C, Muller EC, Otto A, Pfrieger FW. 2001. CNS synaptogenesis promoted by glia-derived cholesterol. Science 294:1354–1357. McClure SJ, Robinson PJ. 1996. Dynamin, endocytosis and intracellular signalling (review). Mol Membr Biol 13:189–215. Micheva KD, Taylor CP, Smith SJ. 2006. Pregabalin reduces the release of synaptic vesicles from cultured hippocampal neurons. Mol Pharmacol 70:467–476. Milner B. 1972. Disorders of learning and memory after temporal lobe lesions in man. Clin Neurosurg 19:421–446. Minturn JE, Fryer HJ, Geschwind DH, Hockfield S. 1995. TOAD64, a gene expressed early in neuronal differentiation in the rat, is related to unc-33, a C. elegans gene involved in axon outgrowth. J Neurosci 15:6757–6766. Molodtsova GF. 1999. Sexual and interhemispheric differences in the involvement of serotonin from the hippocampus and amygdaloid body in the processing of new and repeatedly presented information in rats. Zh Vyssh Nerv Deiat Im I P Pavlova 49:408–415. Morciano M, Beckhaus T, Karas M, Zimmermann H, Volknandt W. 2009. The proteome of the presynaptic active zone: From docked synaptic vesicles to adhesion molecules and maxi-channels. J Neurochem 108:662–675. Mouradian MM. 2002. Recent advances in the genetics and pathogenesis of Parkinson disease. Neurology 58:179–185. Neveu PJ, Liege S, Sarrieau A. 1998. Asymmetrical distribution of hippocampal mineralocorticoid receptors depends on lateralization in mice. Neuroimmunomodulation 5:16–21. Niglio T, Caporali MG, Scotti de Carolis A, Ricci A, Amenta F. 1990. Absence of right-left asymmetries in the rat hippocampus as demonstrated by Timm staining. Acta Anat (Basel) 139:283–286. Oke A, Lewis R, Adams RN. 1980. Hemispheric asymmetry of norepinephrine distribution in rat thalamus. Brain Res 188:269– 272. Panda D, Miller HP, Banerjee A, Luduena RF, Wilson L. 1994. Microtubule dynamics in vitro are regulated by the tubulin isotype composition. Proc Natl Acad Sci USA 91:11358–11362. Peters A, Palay SL, de Webster H. 1991. The Fine Structure of the Nervous System: Neurons and Their Supporting Cells. New York: Oxford University Press. Pruessner JC, Li LM, Serles W, Pruessner M, Collins DL, Kabani N, Lupien S, Evans AC. 2000. Volumetry of hippocampus and amygdala with high-resolution MRI, three-dimensional analysis software: Minimizing the discrepancies between laboratories. Cereb Cortex 10:433–442. R Development Core Team. 2008.R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria. Rakhit R, Cunningham P, Furtos-Matei A, Dahan S, Qi XF, Crow JP, Cashman NR, Kondejewski LH, Chakrabartty A. 2002. Oxidationinduced misfolding and aggregation of superoxide dismutase and its implications for amyotrophic lateral sclerosis. J Biol Chem 277:47551–47556. Riederer BM, Zagon IS, Goodman SR. 1987. Brain spectrin(240/235) and brain spectrin(240/235E): Differential expression during mouse brain development. J Neurosci 7:864–874. Schneider LH, Murphy RB, Coons EE. 1982. Lateralization of striatal dopamine (D2) receptors in normal rats. Neurosci Lett 33:281– 284. Smillie KJ, Cousin MA. 2005. Dynamin I phosphorylation and the control of synaptic vesicle endocytosis. Biochem Soc Symp 72:87–97.
PROTEOMICS REVEAL RAT HIPPOCAMPAL LATERAL ASYMMETRY Tanaka Y, Yamada K, Zhou CJ, Ban N, Shioda S, Inagaki N. 2003. Temporal and spatial profiles of ABCA2-expressing oligodendrocytes in the developing rat brain. J Comp Neurol 455:353–367. Toga AW, Thompson PM. 2003. Mapping brain asymmetry. Nat Rev Neurosci 4:37–48. Tononi G, Cirelli C. 2003. Sleep and synaptic homeostasis: A hypothesis. Brain Res Bull 62:143–150. Torre E, McNiven MA, Urrutia R. 1994. Dynamin 1 antisense oligonucleotide treatment prevents neurite formation in cultured hippocampal neurons. J Biol Chem 269:32411–32417. Wang L, Joshi SC, Miller MI, Csernansky JG. 2001. Statistical analysis of hippocampal asymmetry in schizophrenia. Neuroimage 14:531–545. Waterham HR, Koster J, Romeijn GJ, Hennekam RC, Vreken P, Andersson HC, FitzPatrick DR, Kelley RI, Wanders RJ. 2001. Mutations in the 3beta-hydroxysterol Delta24-reductase gene cause desmosterolosis, an autosomal recessive disorder of cholesterol biosynthesis. Am J Hum Genet 69:685–694.
119
Wechsler A, Brafman A, Shafir M, Heverin M, Gottlieb H, Damari G, Gozlan-Kelner S, Spivak I, Moshkin O, Fridman E, Becker Y, Skaliter R, Einat P, Faerman A, Bjo¨rkhem I, Feinstein E. 2003. Generation of viable cholesterol-free mice. Science 302:2087. Wiesinger H, Hamprecht B, Dringen R. 1997. Metabolic pathways for glucose in astrocytes. Glia 21:22–34. Yoo BC, Fountoulakis M, Cairns N, Lubec G. 2001. Changes of voltage-dependent anion-selective channel proteins VDAC1 and VDAC2 brain levels in patients with Alzheimer’s disease and down syndrome. Electrophoresis 22:172–179. Zhou C, Zhao L, Inagaki N, Guan J, Nakajo S, Hirabayashi T, Kikuyama S, Shioda S. 2001. Atp-binding cassette transporter ABC2/ABCA2 in the rat brain: A novel mammalian lysosome-associated membrane protein and a specific marker for oligodendrocytes but not for myelin sheaths. J Neurosci 21:849–857. Zomzely-Neurath CE. 1982. Nervous-system-specific proteins: 14–3-2 protein, antigen alpha and neuron-specific enolase. Scand J Immunol Suppl 9:1–40.
Hippocampus
