Mechanisms underlying rule learning-induced enhancement of excitatory and inhibitory synaptic transmission

Transcription

1 J Neurophysiol 107: , First published November 30, 2011; doi: /jn Mechanisms underlying rule learning-induced enhancement of excitatory and inhibitory synaptic transmission Drorit Saar, Iris Reuveni, and Edi Barkai Department of Neurobiology, University of Haifa, Haifa, Israel Submitted 18 April 2011; accepted in final form 22 November 2011 Saar D, Reuveni I, Barkai E. Mechanisms underlying rule learning-induced enhancement of excitatory and inhibitory synaptic transmission. J Neurophysiol 107: , First published November 30, 2011; doi: /jn Training rats to perform rapidly and efficiently in an olfactory discrimination task results in robust enhancement of excitatory and inhibitory synaptic connectivity in the rat piriform cortex, which is maintained for days after training. To explore the mechanisms by which such synaptic enhancement occurs, we recorded spontaneous miniature excitatory and inhibitory synaptic events in identified piriform cortex neurons from odor-trained, pseudo-trained, and naive rats. We show that olfactory discrimination learning induces profound enhancement in the averaged amplitude of AMPA receptormediated miniature synaptic events in piriform cortex pyramidal neurons. Such physiological modifications are apparent at least 4 days after learning completion and outlast learning-induced modifications in the number of spines on these neurons. Also, the averaged amplitude of GABA A receptor-mediated miniature inhibitory synaptic events was significantly enhanced following odor discrimination training. For both excitatory and inhibitory transmission, an increase in miniature postsynaptic current amplitude was evident in most of the recorded neurons; however, some neurons showed an exceptionally great increase in the amplitude of miniature events. For both excitatory and inhibitory transmission, the frequency of spontaneous synaptic events was not modified after learning. These results suggest that olfactory discrimination learning-induced enhancement of synaptic transmission in cortical neurons is mediated by postsynaptic modulation of AMPA receptor-dependent currents and balanced by long-lasting modulation of postsynaptic GABA A receptor-mediated currents. olfactory learning; piriform cortex; miniature synaptic currents RATS THAT ARE TRAINED in an olfactory discrimination task demonstrate a dramatic increase in their capability to acquire memories of new odors, once they have learned the first discrimination task. Acquisition of such skill is termed rule learning (Saar et al. 1998, 1999). A line of evidence shows that olfactory discrimination rule learning results with widespread enhancement of excitatory and inhibitory synaptic transmission in the piriform cortex, which lasts for days (Brosh and Barkai 2009; Knafo et al. 2005; Saar et al. 1999, 2002). Modulation of postsynaptic AMPA receptors (AMPAR) has been long suggested to have a key role in the early cellular events leading to learning and memory formation (Kessels and Malinow 2009; Malinow and Malenka 2002). Enhanced AMPAR expression was shown in lateral amygdala neurons after fear conditioning (Rumpel et al. 2005) and cue-reward learning (Tye et al. 2008) and in hippocampal neurons after Address for reprint requests and other correspondence: E. Barkai, Dept. of Neurobiology, Faculty of Sciences, Haifa Univ., Haifa 31905, Israel ( ebarkai@research.haifa.ac.il). inhibitory avoidance training (Whitlock et al. 2006). In these forms of learning, postsynaptic AMPAR trafficking occurs within less than an hour after training and lasts for several hours (Whitlock et al. 2006). This time frame is consistent with the notion that such modulation in AMPAR expression is only the first step in long-term memory formation, which is then maintained by hardwired modifications in synaptic connectivity and single spine morphology (Bourne et al. 2007; Kasai et al. 2003). In a complex form of learning such as odor discrimination, which requires several days of training, enhancement in excitatory synaptic transmission in the piriform cortex was only observed 3 days after rule learning and lasted 5 days, outlasting morphological changes (Knafo et al. 2005; Saar et al. 1999). Activity-induced enhancement in inhibitory synaptic transmission has been shown in several regions of the mammalian brain (Grunze et al. 1996; Holmgren and Zilberter 2001; Komatsu and Yoshimura 2000; Scelfo et al. 2008). Both preand postsynaptic mechanisms are implicated in the process. An activity-induced increase in the number of GABA A receptors was shown in the hippocampus after kindling (Nusser et al. 1998), and enhanced GABA release was shown after brainderived neurotrophic factor application (Ohba et al. 2005). OD learning results with long-lasting enhancement of inhibitory synaptic transmission onto proximal dendrites of piriform cortex pyramidal neurons (Brosh and Barkai 2009). The mechanism underlying the olfactory discrimination learning-induced long-lasting modifications in synaptic transmission is yet to be described. Indirect evidence implicates both pre- and postsynaptic processes: reduced paired-pulse facilitation of excitatory and inhibitory synaptic potentials (Brosh and Barkai 2009; Saar et al. 1999) implies that synaptic release is enhanced after learning but does not rule out postsynaptic modifications. Also, enhanced rate of rise of evoked excitatory postsynaptic potentials (Saar et al. 2002) and hyperpolarization of the GABA A reversal potential (Brosh and Barkai 2009) suggest the involvement of postsynaptic processes. In the present study, we used whole cell voltage-clamp recordings of spontaneous miniature postsynaptic events to further examine the mechanisms underlying the enhancement of synaptic transmission in the mammalian cortex after complex learning. MATERIALS AND METHODS Animal Training Subjects and apparatus. Age-matched young adult (8 wk old at the beginning of training) Sprague-Dawley male rats were used. Before being trained, they were maintained on a 23.5-h water-deprivation /12 Copyright 2012 the American Physiological Society

2 LEARNING-INDUCED INCREASE IN MINIATURE SYNAPTIC EVENTS 1223 schedule, with food available ad libitum. Olfactory discrimination training protocol was performed daily on each trained and pseudotrained rat in a four-arm radial maze (Fig. 1A), as previously described (Saar et al. 1999, 2001), with commercial odors that are regularly used in the cosmetics and food industry, such as peach, lemon, orange, cinnamon, pine, rose, and strawberry, diluted 1:1,000 from commercial concentrated liquid. All animal experiments were done according to National Institutes of Health guidelines and were approved by the University of Haifa animal use committee. Training. Olfactory training consisted of 20 trials per day for each rat as previously described (Saar et al. 2001). In short, in each trial the rat had to choose between two odors (positive and negative cue) presented simultaneously. Rats designated to the trained group were rewarded with drinking water on choosing the positive cue. Rats in the pseudo-trained group were rewarded in a random fashion on choosing any odor. The criterion for learning was at least 80% positive-cue choices in the last 10 trials of a training day (Saar et al. 1999, 2001; Staubli et al. 1987). Rats in the naive group were water deprived, but not exposed to the maze. Our previous studies show that two learning phases can be clearly distinguished (Saar et al. 1998, 1999; Zelcer et al. 2006): the first phase of rule learning, which usually requires 7 8 days, in which rats develop a strategy for performing the odor discrimination task, and the second phase of enhanced learning capability, in which rats can learn new odors within 1 2 training days (see Fig. 1B). In this study, rats were trained with one pair of odors until they reached the criterion for rule learning. They were subsequently allowed to rest for 3 4 days, after which brain slices were prepared. This time window was selected because most olfactory learning-induced physiological and morphological synaptic modifications first appear 3 days after rule learning (Knafo et al. 2005; Saar et al. 1999, 2002). Slice Preparation and Recordings Slice preparation. Coronal brain slices of 300 m were cut from the posterior pirifom cortex as previously described (Saar et al. 1998) and kept for 1hinoxygenated (95% O 2 5% CO 2 ) normal saline Ringer solution (in mm: 124 NaCl, 3 KCl, 2 MgSO 4, 1.25 NaH 2 PO 4, 26 NaHCO 3, 2 CaCl 2, and 10 glucose). Slices were then placed in a recording chamber under an infrared differential interference contrast microscope and perfused with Ringer at 30 C. Whole cell voltageclamp recordings were obtained from visually identified pyramidal neurons in layer II of the piriform cortex. All electrophysiological recordings were performed using Axopatch 1D (Molecular Devices), and the data were acquired using pclamp 9 (Molecular Devices). All experiments were blind; the identity of the rat from which neurons were recorded (naive, trained, or pseudo-trained) was not known to the person conducting the experiments and measurements. Recordings of spontaneous events were performed min after membrane rupture and lasted for up to 20 min. One or two neurons were recorded from each rat. Miniature excitatory postsynaptic current recordings. To compare between populations of cells from different rat groups, we aimed to maintain the cells at physiological conditions and avoided the use of any artificial blockers that might bias the data by variation in blocker effect. Therefore, cells were voltage-clamped at a membrane potential (V m )of 80 mv, which is the resting potential in these cells, as previously measured with intracellular current-clamp recordings (Saar et al. 1998). At this voltage, most voltage-dependent channels are closed and NMDA receptors are rarely activated. The recording electrode was filled with a solution containing (in mm) 140 K-gluconate, 1 EGTA, 6 KCl, 4 NaCl, 2 MgCl 2, and 10 HEPES, ph 7.25, 280 mosm. Miniature excitatory postsynaptic currents (mepscs) were recorded in the presence of 1 M tetrodotoxin (TTX; Fig. 2A). At the end of each experiment, additional recording was performed with 20 M 6,7-dinitroquinoxaline-2,3-dione (DNQX) in the perfusing solution, for at least 10 min, to ensure that there was no contamination of the data with non-ampar-mediated events. Given the low concentration of chloride in the patch pipette solution (10 mm), the reversal potential of GABA-mediated inhibitory postsynaptic currents (IPSCs) would be around 65 mv, and thus miniature IPSCs (mipscs) at V m 80 mv should be very small. Miniature inhibitory postsynaptic current recordings. To record GABA A -mediated mipscs, the recording electrode was filled with a solution containing (in mm) 140 CsCl, 1 EGTA, 6 KCl, 4 NaCl, 2 MgCl 2, and 10 HEPES, ph 7.25, 280 mosm. In these conditions, the reversal potential of chloride has a value of about 0 mv, and thus strong GABA A -mediated currents can be studied at holding potential of 60 mv (Fig. 4A). The perfusion solution contained TTX (1 M), and also DNQX (20 M) and D-2-amino-5-phosphonovaleric acid (50 M), to block glutamatergic synaptic transmission via AMPARs, to allow recording of pure IPSCs. Indeed, the mpscs were abolished by bicuculline methiodide (20 M) application (Fig. 4A). During the continuous recording, the current response to a 200-ms voltage step of 5 mv was monitored at 1 Hz. A change in the response caused exclusion of the data. For data analysis, each event was detected by eye and measured using the Mini Analysis software by Synaptosoft. For both mepscs and mipscs, the averaged amplitude and the median of amplitudes of all spontaneous events were Fig. 1. Training apparatus and rule learning. A: schematic representation of the olfactory maze. Protocols for trained and pseudo-trained rats were similar: an electronic start command randomly opens 2 of 8 valves (V), releasing pressured airstreams with positive-cue odor (P) into one of the arms and negative-cue odor (N) into another. Eight seconds later, the 2 corresponding guillotine doors (D) are lifted to allow the rat to enter the selected arms. On reaching the far end of an arm (90 cm long), the rat s body interrupts an infrared beam (I, arrow) and a drop of drinking water is released from a water hose (W) into a small drinking well (for a trained rat, only if the arm contains the positive-cue odor; for pseudo-trained rats, randomly). A trial ends when the rat interrupts a beam or after 10 s, if no beam is interrupted. A fan is operated for 15 s between trials, to remove odors. Each rat had 20 trials per day. B: trained rats demonstrated acquisition of rule learning. Seven consecutive days of training were required for this group to reach criterion for discriminating between the first pair of odors (80% positive-cue choices). Discrimination between any new pair of odors, starting from the third pair and onward, could be reached within 1 day. Values are means SE; n 11 rats.

3 1224 LEARNING-INDUCED INCREASE IN MINIATURE SYNAPTIC EVENTS Fig. 2. Learning-induced enhancement of the AMPA receptor (AMPAR)-mediated synaptic currents A: miniature synaptic events recorded in a neuron from a trained rat at a holding potential of 80 mv in the presence of 1 M tetrodotoxin (TTX). An example of a minimal-amplitude current appears in the second trace from top. Note that most events are larger than the minimal-amplitude event. B: the averaged median amplitude of miniature events in neurons from trained rats was markedly higher than in neurons from the 2 control groups, which did not differ from each other. The median was determined for each neuron from all spontaneous events. Values (means SE) represent the average of all cells in each group. **P C: cumulative frequency distributions of all medians in the 3 rat groups. Each point represents the median in 1 neuron. Note that whereas the median miniature excitatory postsynaptic current (mepsc) amplitude appears to have increased in most neurons in the trained group, a subpopulation of 6 neurons of the 22 recorded show particularly strong enhancement. D: the frequency of the miniature spontaneous events did not differ between rat groups. calculated for each neuron, and means were then calculated for each group. Statistical Analysis Between-group comparisons were done using one-way ANOVA, and post hoc multiple nondirectional t-tests were then applied to compare between two groups. Values reported in the text are means SD. Data presented in graphs are means SE. RESULTS Learning-Induced Enhancement of AMPAR-Mediated Currents Piriform cortex brain slices were prepared 3 4 days after rule learning (learning of a novel pair of odors to criterion within 1 training session). TTX-insensitive mepscs were recorded from layer II pyramidal neurons (Fig. 2A). Olfactorydiscrimination learning induced a dramatic increase in the averaged amplitude of miniature currents. The averaged mepsc amplitude in neurons from trained rats ( pa, n 22 neurons taken from 20 rats) was significantly (P 0.01) higher compared with the averaged value in neurons from naive ( pa, n 14 neurons from 13 rats) and pseudo-trained rats ( pa, n 14 neurons from 14 rats), which did not differ from each other. Similar differences were observed when comparing the median of amplitudes in the three groups; the averaged mepsc median in neurons from trained rats ( pa, n 22) was significantly (P 0.01) higher compared with the averaged value in neurons from naive ( pa, n 14) and pseudo-trained rats ( pa, n 14), which did not differ from each other (Fig. 2B). At the end of each recording, DNQX (20 M) was applied via the perfusing solution. In all recordings, no spontaneous events were observed in the presence of DNQX, indicating that only AMPAR-mediated currents were measured. The averaged rise time and the averaged decay time did not differ among the three groups (Table 1). Also, cells from the three groups did not differ in their response to standard voltage steps (Table 1). The root mean square (RMS) of noise was similar in all the recordings (Table 1). No relation was found between the averaged rise time, decay time, and voltagestep response in a cell, the RMS noise in the recording, and the averaged EPSC amplitude measured in the same cell (Table 2). The increase in mepsc averaged amplitude was apparent throughout the sampled neuronal population (Fig. 2C). To further explore the nature of this change, amplitude distribution histograms of miniature events were constructed for each neuron. In neurons from the control groups, a sharp peak was always apparent around the value of 5 pa, with steep decay of the curve, showing just a few events with amplitudes higher than 12.5 pa (more than 2 SD above the average in controls) (Fig. 3A). In neurons from trained rats, amplitude distribution curves showed a first peak similar to that in controls, but the curve decayed at a much lower rate (Fig. Table 1. EPSC kinetics and basic membrane properties are not modified by learning Naive Trained Pseudo-Trained n EPSC rise time, ms EPSC decay time, ms Response to 5-mV step, pa RMS noise, pa Values are means SD; n no. of neurons. Rise time was measured for each detected event from baseline to peak. Decay time was measured for each detected event from peak to one-third the amplitude of the event. The basic membrane properties of neurons in which excitatory postsynaptic currents (EPSCs) were recorded are shown: the current response to a 200-ms voltage step of 5 mv was evoked at 0.16 Hz, and root mean square (RMS) noise was measured from 750-ms baseline recordings in which no spontaneous events were detected. No significant difference was found between groups in all the above parameters.

4 LEARNING-INDUCED INCREASE IN MINIATURE SYNAPTIC EVENTS 1225 Table 2. Relation between membrane properties or recording conditions and averaged EPSC amplitude in a cell Goodness of Linear Regression Fit, R 2 Averaged event rise time, ms Event decay time, ms Response to 5-mV step, pa RMS noise, pa Graphs were created for all cells (n 40) of averaged event amplitude (pa) vs. each parameter listed. All linear regressions that were applied resulted in poor fit (R ). 3B). High-amplitude events were frequently apparent, and a significant amount of events had amplitudes higher than 12.5 pa. Notably, in about one-quarter of the neurons recorded from trained rats (6 of 22 neurons), the majority of synaptic events had amplitudes higher than the average in controls (8.5 pa, see example in Fig. 3B, histogram at right, and 3C), indicating that in these particular cells, most recorded synapses are strengthened after learning. Notably, although the averaged amplitude of spontaneous events in neurons from this group ( pa) was significantly higher (P 0.001) than in other neurons from the trained group ( pa), the averaged rise time ( ms in neurons with high EPSC amplitudes vs ms in other neurons) and decay times ( ms in neurons with high EPSC amplitudes vs ms in other neurons) did not differ between these subgroups of trained neurons. These data indicate that the difference in averaged spontaneous EPSC (sepsc) amplitude between the neurons with exceptionally high amplitudes and others is not a result of the distance of EPSC generation from the soma. The increase in the averaged amplitude of sepsc was not accompanied by modification in the frequency of spontaneous events (Fig. 2D). Such a dramatic increase in the averaged amplitude of sepsc, without any apparent modification in the frequency of these spontaneous events, suggests that olfactory discrimination learning is accompanied by long-lasting modulation of postsynaptic AMPAR-mediated currents, rather than by an increase in the probability of presynaptic release. In addition, the averaged frequency of mepscs recorded in the neurons that had the highest averaged amplitudes of spontaneous events ( events/min, n 5 neurons) was similar to the averaged frequency recorded in other neurons taken from the trained group ( events/min, n 17 neurons), further indicating that the learning-induced enhancement of spontaneous events is not mediated by a presynaptic process. Learning-Induced Postsynaptic Enhancement of GABA A - Mediated Unitary Synaptic Events GABA A -mediated sipscs were recorded from layer II pyramidal neurons 3 4 days after rule learning (Fig. 4A). Olfactory discrimination learning induced a dramatic increase in the averaged amplitude of miniature events. The averaged mipsc amplitude in neurons from trained rats ( pa, n 18 neurons from 14 rats) was significantly (P 0.03) higher Fig. 3. Large events appear after learning. A: amplitude histograms of 3 representative neurons from the pseudo-trained group. Neurons are presented according to increasing values of their averaged mepsc amplitudes. The left histogram represents the neuron positioned at point 0.25 along the cumulative frequency graph (shown in Fig. 2C), the middle histogram at point 0.5, and the right histogram at point Although differences in histograms are apparent, most events in the 3 neurons did not exceed the averaged amplitude value of the pseudo-trained group (8.5 pa). B: amplitude histograms of 3 representative neurons from the trained group. Neurons are presented according to increasing values of their averaged mepsc amplitudes. The left histogram represents the neuron positioned at point 0.25 along the cumulative frequency graph (Fig. 2C), the middle histogram at point 0.5, and the right histogram at point One-half of events in the left histogram exceed the averaged value observed for the pseudo-trained group, and most events in the right histogram have an amplitude of 8.5 pa or more. C: in all neurons, the fraction of high-amplitude events is correlated with the averaged amplitude. Note that in several neurons from trained rats, more than 70% of the events were larger than 8.5 pa.

5 1226 LEARNING-INDUCED INCREASE IN MINIATURE SYNAPTIC EVENTS Fig. 4. Learning-induced enhancement of the GABA A -mediated synaptic currents. A: miniature inhibitory synaptic events recorded in a neuron from a trained rat at a holding potential of 60 mv in the presence of 1 M TTX, 20 M 6,7-dinitroquinoxaline-2,3-dione (DNQX), and 50 M D-2-amino-5-phosphonovaleric acid (D-APV). Spontaneous activity was totally abolished in the presence of the GABA A blocker bicuculline methiodide (BIM; 20 M). B: the averaged median of miniature events in neurons from trained rats was significantly higher than in neurons from the pseudo-trained groups. The median was determined for each neuron from all spontaneous events. Values (means SE) represent the average of all cells in each group. *P C: cumulative frequency distributions of event medians of the 3 groups. Each point represents the median in 1 neuron. Note that whereas the averaged spontaneous inhibitory postsynaptic current (sipsc) median appears to have increased in most neurons in the trained group, a subpopulation of 6 neurons of the 20 recorded show particularly strong enhancement. D: the frequency of the inhibitory spontaneous events did not differ between groups. compared with the averaged value in neurons from pseudotrained ( pa, n 10 neurons from 8 rats) and naive rats ( pa, n 9 neurons from 7 rats). Similar differences were observed when comparing the averaged median of amplitudes in the three groups; the averaged mipsc median in neurons from trained rats ( pa, n 18) was significantly (P 0.01) higher compared with the averaged value in neurons from naive ( pa, n 9) and pseudo-trained rats ( pa, n 10), which did not differ from each other (Fig. 4B). The averaged rise time and decay time did not differ between the trained and pseudotrained groups (Table 3). Also, cells from the two groups did not differ in their response to standard voltage steps (Table 3). RMS noise was similar in all the recordings (Table 3). Similar to miniature excitatory synaptic events, the increase in mipscs averaged amplitude was apparent throughout the neuronal cell population (Fig. 4C). Amplitude distribution histograms of miniature IPSCs were constructed for each neuron. In neurons from the pseudotrained group, a sharp peak was always apparent around the value of 10 pa, with steep decay of the curve, showing just a few events with amplitudes higher than 22 pa (2 SD above the average in control) (Fig. 5A). Amplitude distribution curves of Table 3. IPSC kinetics and basic membrane properties are not modified by learning Trained Pseudo-Trained n IPSC rise time, ms IPSC decay time, ms Response to 5-mV step, pa RMS noise, pa Values are means SD; n no. of neurons. Rise time was measured for each detected event from baseline to peak. Decay time was measured for each detected event from peak to one-third the amplitude of the event. The basic membrane properties of neurons in which inhibitory postsynaptic currents (IPSCs) were recorded are shown: the current response to a 200-ms voltage step of 5 mv was evoked at 0.16 Hz, and RMS noise was measured from 750-ms baseline recordings in which no spontaneous events were detected. No significant difference was found between groups in all the above parameters. neurons from trained rats peaked at higher amplitudes (14 pa) and decayed at lower rates (Fig. 5B). As observed for synaptic excitation, the widespread learning-induced enhancement of synaptic inhibition resulted with a significant increase in the appearance of high-amplitude miniature synaptic events in the trained neuronal population. In about 25% of neurons recorded from trained rats (5 of 18 neurons recorded) the majority of synaptic events had amplitudes higher than the average in control (18.2 pa, see example in Fig. 5B, histogram at right, and 5C), indicating that in these cells most recorded synapses are strengthened after learning. The notable (30%) increase in the averaged amplitude of mipsc was not accompanied by modification in the frequency of spontaneous events (Fig. 4D). Here, too, the averaged frequency of mipscs recorded in the neurons that had the highest averaged amplitudes of spontaneous events ( events/min, n 5 neurons) was similar to the averaged frequency recorded in other neurons taken from the trained group ( events/min, n 13 neurons), suggesting that olfactory discrimination learning is accompanied by long-lasting modulation of postsynaptic GABA A -mediated currents, and not by an increase in the probability of presynaptic GABA release. DISCUSSION We have previously shown, using in vivo and in vitro field potential recordings, as well as intracellular current-clamp recordings, that complex odor learning is accompanied by pronounced enhancement in excitatory synaptic connectivity into and within the piriform cortex, balanced by enhancement in inhibitory synaptic transmission (Brosh and Barkai 2009; Cohen et al. 2008; Saar et al. 1999, 2002). In the present study, using recordings and analysis of spontaneous synaptic events, we show that olfactory learning-induced synaptic enhancement is due to an increase in postsynaptic AMPAR- and GABA A receptor-dependent currents.

6 LEARNING-INDUCED INCREASE IN MINIATURE SYNAPTIC EVENTS 1227 Differences in Postynaptic Current Amplitudes Are Not a Result of Differences in Recording Conditions Inadequate clamping of the membrane potential would result in distortion of both the amplitude and the rise time of the events (Williams and Mitchell 2008). We observed no difference in rise time between the groups, indicating that clamping conditions were similar in all groups and could not underlie the observed differences in event amplitude. Also, RMS noise was similar in all groups so that the differences in averaged event amplitude could not be due to small events being hidden in the noise in one group only. No relation was found between the size of RMS noise in each cell and the size of averaged EPSC amplitude measured in this cell. Differences in Postsynaptic Current Amplitudes Are Not Due to Differences in Basic Neuronal Properties The substantial increase in the amplitude of mpscs in the piriform cortex is not a result of changes in passive membrane properties of these cells. In our previous studies (see for example, Saar et al. 1998), we used intracellular current-clamp recordings to examine the passive properties of neurons from trained and control rats. We found no significant difference in input resistance, resting potential, or membrane time constant between groups. Also, in the present voltage-clamp experiments, the cells responses to standard voltage steps did not differ among the three groups. A change in the electrotonic distance of the dendritic tree could affect the observed change in miniature EPSC and IPSC Fig. 5. The proportion of large inhibitory synaptic events is markedly increased after learning. A: amplitude histograms of 3 neurons from the pseudo-trained group. Neurons are presented according to increasing values of their averaged sipsc amplitudes. The left histogram represents the neuron positioned at point 0.25 along the cumulative frequency graph (shown in Fig. 4C), the middle histogram at point 0.5, and the right histogram at point Although differences in histograms are apparent, most events in the 3 neurons did not exceed the averaged amplitude value of the pseudo-trained group. B: amplitude histograms of 3 neurons from the trained group. Neurons are presented according to increasing values of their averaged sipsc amplitudes. The left histogram represents the neuron positioned at point 0.25 along the cumulative frequency graph (Fig. 4C), the middle histogram at point 0.5, and the right histogram at point More than onehalf of events in the left histogram do not exceed the averaged value observed for the pseudo-trained group, whereas most events in right histogram have amplitude of 18.2 pa or more. C: in all neurons, the fraction of highamplitude events is correlated with the averaged amplitude. Note that in several neurons from trained rats, more than 70% of the events were larger than 18.2 pa. amplitudes. However, we have previously shown, using microscopic techniques, that there is no odor training-related change in the actual size of the dendritic tree in piriform cortex pyramidal neurons (Knafo et al. 2001). Using intracellular recordings at current-clamp configuration, we did observe a 10% decrease in electrotonic distance of evoked postsynaptic potentials in neurons from trained rats (Saar et al. 2001). However, this 10% reduction can hardly underlie the 60% increase in the EPSC amplitudes and the 30% increase in the IPSC amplitudes observed here. Learning-Induced Synaptic Enhancement is Mediated by Postsynaptic Mechanisms Our data suggest that odor learning-induced increases in excitatory and inhibitory synaptic transmission are both mediated by postsynaptic mechanisms. An increase in the probability of synaptic release, or in the number of presynaptic release sites, would result with an increase in the frequency of miniature events. Our data show that the increase in events amplitude is not accompanied by changes in the frequency of spontaneous synaptic events. Indeed, only a mild (10%) increase in the number of synaptic spines was observed 3 days after odor learning (Knafo et al. 2005), when the increase in synaptic transmission strength is most apparent. These data indicate that most of the observed increase in synaptic strength does not originate from presynaptic modulations. Our data do not rule out an increase in vesicle content, which may contribute to an increase in quantal size (see Edwards

7 1228 LEARNING-INDUCED INCREASE IN MINIATURE SYNAPTIC EVENTS 2007 for review). We did not observe changes in the size of the smallest events, and the first peak of smallest events was similar in all groups, at around 5 pa. However, there is still a possibility that even smaller events are hidden in the noise, and these minimal events may be altered in trained rats. Interestingly, learning-induced enhancement in inhibitory synaptic transmission has been shown in the barrel cortex following classical aversive conditioning (Gierdalski et al. 2001; Jasinska et al. 2010; Siucinska 2006). Unlike our present findings, such changes are mediated by presynaptic modifications, namely, an increase in inhibitory synaptic connectivity. Notably, these presynaptic changes in synaptic inhibition appear early after learning (Gierdalski et al. 2001; Jasinska et al. 2010; Siucinska 2006) and are short lasting (Gierdalski et al. 2001). Together, these data indicate that learning-induced enhancement in synaptic inhibition may be initially mediated by presynaptic morphological modifications and subsequently by postsynaptic physiological modifications. Indications for Increase in Postsynaptic Conductances Enhanced excitatory transmission was observed in vivo after learning in the ascending (from the olfactory bulb) and descending (from the orbitofrontal cortex) fibers terminating on layer II pyramidal neurons (Cohen et al. 2008) and also in the intrinsic fibers interconnecting these neurons in brain slices (Saar et al. 1999, 2002). These increases had similar magnitudes ( 60%) in all pathways (Cohen et al. 2008; Saar et al. 1999, 2002). The extent of increase observed in these pathways is similar to the 64% increase in the amplitude of mpscs observed here. One possible explanation for such consistency in the extent of excitatory synaptic strengthening is that it results from postsynaptic modifications in neurons in piriform cortex such that they respond strongly, regardless of the source of input. A growing body of evidence demonstrates a similar large overall increase in synaptic strength ( 50%) in different brain structures following various training paradigms (McKernan and Shinnick-Gallagher 1997; Sacchetti et al. 2001, 2004; Tye et al. 2008; Yin et al. 2009). Although learning-induced synaptic enhancement is apparent in most recorded neurons, prominent extensive changes are present in a subgroup of neurons that entails only about one-quarter of the pyramidal cell population. Notably, this finding applies to both excitatory and inhibitory synaptic transmission. Whether a key role in learning-induced, long-term enhancement of synaptic transmission can be attributed to this particular group of neurons remains to be explored. Apparent Discrepancy With Learning-Induced Reduction in Paired-Pulse Facilitation Because learning-induced synaptic modifications are also expressed by reduced paired-pulse facilitation (PPF) of the excitatory synaptic responses evoked by stimulating glutamatergic neurons, it was previously interpreted that learning results with enhanced synaptic release from these excitatory neurons (Brosh and Barkai 2009; McKernan and Shinnick- Gallagher 1997; Saar et al. 1999). However, enhanced excitatory synaptic release is only one way to interpret PPF modulation. An alternative explanation might be, for example, enhanced GABA B -mediated suppression of such excitatory synaptic release. It has been previously shown that GABA B - receptors mediate inhibition of synaptic release through modulation of presynaptic Ca 2 channels (Chalifoux and Carter 2010, 2011; Marshall 2008; Otmakhova and Lisman 2004; Tang and Hasslemo 1994). In particular, activity-induced GABA B -dependent plasticity was attributed to activation of presynaptic receptors of inhibitory neurons (Ivenshitz and Segal 2006). According to this scenario, the first stimulus in a paired-pulse protocol causes a GABA transient that reduces release during the second pulse, through a di-synaptic mechanism. Indeed, our recent findings (unpublished data) indicate that learning-induced reduction in PPF is abolished in the presence of a GABA B -receptor antagonist, supporting our conclusion that learning-induced enhancement of excitatory synaptic transmission is mediated by postsynaptic mechanisms, whereas suppression of excitatory transmission from the same neurons is mediated by presynaptic GABA B receptor activation. Relation to Previously Described Forms of Synaptic Plasticity The profound increase in excitatory synaptic strength observed after rule learning is not the result of classical homeostatic synaptic scaling (Nelson and Turrigiano 2008; Turrigiano et al. 1998). Learning-induced increase in synaptic strength occurs when the spike firing rate is enhanced (Calu et al. 2007; Roesch et al. 2007; Saar and Barkai 2003) rather than decreased (Nelson and Turrigiano 2008; Turrigiano et al. 1998). Thus, whereas homeostatic synaptic scaling is a negative feedback control mechanism, learning-induced whole cell synaptic strengthening appears to be a positive feedback mechanism, balanced by a parallel increase in inhibitory conductance. Finally, our data show that olfactory discrimination learninginduced physiological modifications not only overlap morphological changes but also outlast them. Physiological modifications in synaptic transmission are maintained for at least 4 days after learning compared with modifications in spine density, which are only apparent up to 3 days after learning (Knafo et al. 2005). Thus we suggest that olfactory discrimination learning is not mediated by initial physiological modifications that are than replaced by more permanent, morphological changes. Rather, such rule learning involves a more complex interaction between these two manifestations of synaptic plasticity. Functional Significance of Enhancement in Synaptic Excitation and Inhibition Because excitatory synaptic transmission and neuronal excitation are both profoundly enhanced by learning (Saar et al. 1999, 2001), the cortex may enter an overexcited state that may prevent any efficient ability to store memories (see for example, Barkai et al. 1994; Hasselmo and Barkai 1995). Strengthening inhibition is a possible homeostatic mechanism for restoring the balance between excitation and inhibition, and hence for preventing the epileptic-like activity. A short activation of a specific subset of excitatory neurons would allow efficient strengthening of connection between these neurons, whereas subsequent activation of widespread synaptic inhibition would prevent uncontrolled spread of activation. Thus we suggest that synaptic excitation and inhibition are combined together to enable long-term memory of olfactory discrimina-

8 LEARNING-INDUCED INCREASE IN MINIATURE SYNAPTIC EVENTS 1229 tion learning capabilities while maintaining a stability of activation in the piriform cortex. To conclude, our data show that olfactory discrimination rule learning results with long-lasting physiological modification in both excitatory and inhibitory synaptic transmission in piriform cortex neurons. Such modifications are mediated by postsynaptic processes and occur in most recorded neurons but are pronounced to a greater extent in a subgroup of cells. GRANTS This research was supported by a grant from the Israel Science Foundation (to E. Barkai). DISCLOSURES No conflicts of interest, financial or otherwise, are declared by the author(s). AUTHOR CONTRIBUTIONS Author contributions: D.S., I.R., and E.B. conception and design of research; D.S. performed experiments; D.S. and I.R. analyzed data; D.S., I.R., and E.B. interpreted results of experiments; D.S., I.R., and E.B. drafted manuscript; D.S., I.R., and E.B. edited and revised manuscript; D.S., I.R., and E.B. approved final version of manuscript; I.R. and E.B. prepared figures. REFERENCES Barkai E, Bergman RE, Horwitz G, Hasselmo ME. Modulation of associative memory function in a biophysical simulation of rat piriform cortex. J Neurophysiol 72: , Bourne J, Harris KM. Do thin spines learn to be mushroom spines that remember? Curr Opin Neurobiol 17: , Brosh I, Barkai E. Learning-induced enhancement of feedback inhibitory synaptic transmission. Learn Mem 16: , Calu DJ, Roesch MR, Stalnaker TA, Schoenbaum G. Associative encoding in posterior piriform cortex during odor discrimination, and reversal learning. Cereb Cortex 17: , Chalifoux JR, Carter AG. GABA B receptors modulate NMDA receptor calcium signals in dendritic spines. Neuron 66: , Chalifoux JR, Carter AG. GABA B receptor modulation of synaptic function. Curr Opin Neurobiol 21: , Cohen Y, Reuveni I, Barkai E, Maroun M. Olfactory learning-induced long-lasting enhancement of descending, and ascending synaptic transmission to the piriform cortex. J Neurosci 28: , Edwards RH. The neurotransmitter cycle, and quantal size. Neuron 55: , Gierdalski M, Jablonska B, Siucinska E, Lech M, Skibinska A, Kossut M. Rapid regulation of GAD67 mrna, and protein level in cortical neurons after sensory learning. Cereb Cortex 11: , Grunze HC, Rainnie DG, Hasselmo ME, Barkai E, Hearn EF, McCarley RW, Greene RW. NMDA dependent modulation of CA1 local circuit inhibition. J Neurosci 16: , Hasselmo ME, Barkai E. Cholinergic modulation of activity-dependent synaptic plasticity in the piriform cortex. J Neurosci 15: , Holmgren CD, Zilberter Y. Coincident spiking activity induces long-term changes in inhibition of neocortical pyramidal cells. J Neurosci 21: , Ivenshitz M, Segal M. Simultaneous NMDA-dependent long-term potentiation of EPSCs and long- term depression of IPSCs in cultured rat hippocampal neurons. J Neurosci 26: , Jasinska M, Siucinska E, Cybulska-Klosowicz A, Pyza E, Furness DN, Kossut M, Glazewski S. Rapid, learning-induced inhibitory synaptogenesis in murine barrel field. J Neurosci 30: , Kasai H, Matsuzaki M, Noguchi J, Yasumatsu N, Nakahara H. Structurestability-function relationships of dendritic spines. Trends Neurosci 26: , Kessels HW, Malinow R. Synaptic AMPA receptor plasticity behavior. Neuron 61: , Knafo S, Grossman Y, Barkai E, Benshalom G. Olfactory learning is associated with increased spine density along apical dendrites of pyramidal neurons in the rat piriform cortex. Eur J Neurosci 13: , Knafo S, Libersat F, Barkai E. Dynamics of learning-induced spine redistribution along dendrites of pyramidal neurons in rats. Eur J Neurosci 21: , Komatsu Y, Yoshimura Y. Activity-dependent maintenance of long-term potentiation at visual cortical inhibitory synapses. J Neurosci 20: , Malinow R, Malenka RC. AMPA receptor trafficking, and synaptic plasticity. Annu Rev Neurosci 25: , Marshall FH. The role of GABA B receptors in the regulation of excitatory neurotransmission. Results Probl Cell Differ 44: 87 98, McKernan MG, Shinnick-Gallagher P. Fear conditioning induces a lasting potentiation of synaptic currents in vitro. Nature 390: , Nelson SB, Turrigiano GG. Strength through diversity. Neuron 60: , Nusser Z, Hajos N, Somogyi P, Mody I. Increased number of synaptic GABA A receptors underlies potentiation at hippocampal inhibitory synapses. Nature 395: , Ohba S, Ikeda T, Ikegaya Y, Nishiyama N, Matsuki N, Yamada MK. BDNF locally potentiates GABAergic presynaptic machineries: target-selective circuit inhibition. Cereb Cortex 15: , Otmakhova NA, Lisman JE. Contribution of I h and GABA B to synaptically induced afterhyperpolarizations in CA1: a brake on the NMDA. response. J Neurophysiol 92: , Roesch MR, Stalnaker TA, Schoenbaum G. Associative encoding in posterior piriform cortex during odor discrimination and reversal learning. Cereb Cortex 17: , Rumpel S, LeDoux J, Zador A, Malinow R. Postsynaptic receptor trafficking underlying a form of associative learning. Science 310: , Saar D, Barkai E. Long-term modifications in intrinsic neuronal properties, and rule learning in rats. Eur J Neurosci 17: , Saar D, Grossman Y, Barkai E. Reduced after-hyperpolarization in rat piriform cortex pyramidal neurons is associated with increased learning capability during operant conditioning. Eur J Neurosci 10: , Saar D, Grossman Y, Barkai E. Reduced synaptic facilitation between pyramidal neurons in the piriform cortex after odor learning. J Neurosci 19: , Saar D, Grossman Y, Barkai E. Cellular correlates of olfactory learning in the rat piriform cortex. J Neurosci 12: , Saar D, Grossman Y, Barkai E. Learning-induced enhancement of postsynaptic potentials in pyramidal neurons. J Neurophysiol 87: , Sacchetti B, Lorenzini CA, Baldi E, Bucherelli C, Roberto M, Tassoni G, Brunelli M. Long-lasting hippocampal potentiation, and contextual memory consolidation. Eur J Neurosci 13: , Sacchetti B, Scelfo B, Tempia F, Strata P. Long-term synaptic changes induced in the cerebellar cortex by fear conditioning. Neuron 42: , Scelfo B, Sacchetti B, Strata P. Learning-related long-term potentiation of inhibitory synapses in the cerebellar cortex. Proc Natl Acad Sci USA 105: , Siucinska E. GAD67-positive puncta: contributors to learning-dependent plasticity in the barrel cortex of adult mice. Brain Res 1106: 52 62, Staubli U, Fraser D, Faraday R, Lynch G. Olfaction and the data memory system in rats. Behav Neurosci 101: , Tang AC, Hasselmo ME. Selective suppression of intrinsic but not afferent fiber synaptic transmission by baclofen in the piriform (olfactory) cortex. Brain Res 659: 75 81, Turrigiano GG, Leslie KR, Desai NS, Rutherford LC, Nelson SB. Activitydependent scaling of quantal amplitude in neocortical neurons. Nature 391: , Tye KM, Stuber GD, de Ridder B, Bonci A, Janak PH. Rapid strengthening of thalamo-amygdala synapses mediates cue-reward learning. Nature 453: , Whitlock JR, Heynen AJ, Shuler MG, Bear MF. Learning induces longterm potentiation in the hippocampus. Science 313: , Williams SR, Mitchell SJ. Direct measurement of somatic voltage clamp errors in central neurons. Nat Neurosci 11: , Yin HH, Mulcare SP, Hilário MR, Clouse E, Holloway T, Davis MI, Hansson AC, Lovinger DM, Costa RM. Dynamic reorganization of striatal circuits during the acquisition, and consolidation of a skill. Nat Neurosci 12: , Zelcer I, Cohen H, Richter-Levin G, Lebiosn T, Grossberger T, Barkai E. A cellular correlate of Learning-induced metaplasticity in the hippocampus. Cereb Cortex 16: , 2006.