Learning Promotes Subfield-Specific Synaptic Diversity in Hippocampal CA1 Neurons

Abstract The hippocampus is functionally heterogeneous between the dorsal and ventral subfields with left–right asymmetry. To determine the possible location of contextual memory, we performed an inhibitory avoidance task to analyze synaptic plasticity using slice patch-clamp technique. The training bilaterally increased the AMPA/NMDA ratio at dorsal CA3–CA1 synapses, whereas the training did not affect the ratio at ventral CA3–CA1 synapses regardless of the hemisphere. Moreover, sequential recording of miniature excitatory postsynaptic currents and miniature inhibitory postsynaptic currents from the same CA1 neuron clearly showed learning-induced synaptic plasticity. In dorsal CA1 neurons, the training dramatically strengthened both excitatory and inhibitory postsynaptic responses in both hemispheres, whereas the training did not promote the plasticity in either hemisphere in ventral CA1 neurons. Nonstationary fluctuation analysis further revealed that the training bilaterally increased the number of AMPA or GABAA receptor channels at dorsal CA1 synapses, but not at ventral CA1 synapses, suggesting functional heterogeneity of learning-induced receptor mobility. Finally, the performance clearly impaired by the bilateral microinjection of plasticity blockers in dorsal, but not ventral CA1 subfields, suggesting a crucial role for contextual learning. The quantification of synaptic diversity in specified CA1 subfields may help us to diagnose and evaluate cognitive disorders at the information level.


Introduction
The hippocampus is functionally heterogeneous between the dorsal and ventral subfields (Fanselow and Dong 2010), with left-right asymmetry (Shinohara et al. 2013). Dorsal subfields seem to serve cognitive functions, whereas ventral subfields correspond to the affective hippocampus (Moser and Moser excitatory and inhibitory inputs at CA1 synapses (Mitsushima et al. 2011(Mitsushima et al. , 2013. However, there is no synaptic evidence to prove the location of encoded memory within a broad CA1 area. First, we analyzed learning-induced synaptic plasticity in 4 CA1 subfields to analyze the learning-induced synaptic plasticity. Second, we sequentially recorded miniature EPSCs (mEPSCs) and miniature IPSCs (mIPSCs) in the same neuron to specify the CA1 subfield of learning-created synaptic diversity. Considering that each presynaptic vesicle contains approximately 2000 glutamate (Ryan et al. 1993;Hori and Takahashi 2012) or 2500 GABA molecules (Telgkamp et al. 2004;Pugh and Raman 2005), the mE(I)PSC analysis allows for the quantification of postsynaptic currents and plasticity. Nonstationary fluctuation analysis further revealed subfield-specific evidence of learning at a single-channel level. Moreover, by analyzing the appearance probability of the synaptic strength in each neuron, we proposed a new approach to quantify learning-induced synaptic diversity as self-entropy increases after the training. The learning clearly increased the cell-specific self-entropy levels in dorsal, but not ventral CA1 subfields, and local blockade of the synaptic plasticity blocked the learning in dorsal but not ventral CA1 subfields, suggesting contributory CA1 subfields at the information level. Since learning is known to modulate both excitatory and inhibitory synaptic plasticity in key brain areas such as hippocampus (Mitsushima et al. 2013), amygdala (Lin et al. 2011;Ganea et al. 2015), or cortical areas (Ghosh et al. 2015(Ghosh et al. , 2016Kida et al. 2016), this approach may help us to diagnose and evaluate cognitive disorders in multiple brain regions.

Animals
Young male Sprague-Dawley rats (postnatal 28-31 days of age) were used. After weaning, same sex groups of 2-3 rats were housed in plastic cages (length 25 cm, width 40 cm, height 25 cm) at a constant temperature of 23 ± 1°C under a constant cycle of light and dark (light on: 8:00 A.M. to 8:00 P.M.). But, the rats were individually housed at least 24 h prior to the experiment to avoid any episodic experience. Food (MF, Oriental Yeast Co. Ltd, Tokyo Japan) and tap water were available ad libitum in all experimental periods. All animal housing and surgical procedures followed the guidelines of the Institutional Animal Care and Use Committee of Kanagawa Dental University and Yamaguchi University. The guidelines comply with the Guide for the Care and Use of Laboratory Animals published by the National Institute of Health (NIH Publication No. 85-23, revised 1996).

Inhibitory Avoidance Task
The IA training apparatus (length: 33 cm, width: 58 cm, height: 33 cm) was a 2-chambered box consisting of a lighted safe side and a dark shock side separated by a trap door ( Fig. 1A; Mitsushima et al. 2011Mitsushima et al. , 2013. For training, rats were placed in the light side of the box facing a corner opposite the door. After the trap door was opened, the rats could enter the dark box at will. The latency before entering the novel dark box was measured as a behavioral parameter (latency before IA learning). Soon after the animals entered the dark side, we closed the door and applied a scrambled electrical foot-shock (2 s, 1.6 mA) via electrified steel rods in the floor of the box. The rats were kept in the dark compartment for 10 s before being returned to their home cage. Untrained control rats were not moved from their home cages.
Thirty minutes after the procedure described above, the rats were placed in the light side. The latency before entering the dark box was measured as an indicator of learning performance (latency after IA learning).

Drug Injection
Under sodium pentobarbital anesthesia (30-50 mg/kg, i.p.), a stainless-steel guide cannula (outer diameter, 0.51 mm) was implanted stereotaxically into the just above the target region of the dorsal or ventral hippocampus. The experiment was performed 1-3 days after the implantation. After cannula implantation, a stylet was inserted into the guide until drug injection.
On the day of the experiment, the stylet was replaced with 1.0 mm longer injector without restraint animals in their home cage (outer diameter 0.31 mm). The coordinates of dorsal CA1 were 3.0 mm posterior to bregma, 2.0 mm lateral to the midline, and 3.8 mm below the surface of the skull. The coordinates of ventral CA1 were 4.2 mm posterior to bregma, 5.5 mm lateral to the midline, and 5.5 mm below the surface of the skull.

Electrophysiological Recordings
We recently published detailed technical protocol of slice-patch clamp technique for analyzing learning-induced synaptic plasticity with a short demonstration movie . Using the technique, we examined learning-induced synaptic plasticity in dorsal or ventral CA1 neurons.

The AMPA/NMDA Ratio
The AMPA/NMDA ratio is conventional way to evaluate postsynaptic plasticity at glutamatergic excitatory synapses. Since concomitant increases in both components may not change the ratio, further analysis of AMPA evoked responses is necessary to elucidate the receptor-specific plasticity (i.e., I/O curve for evoked EPSCs, amplitude of miniature AMPA receptormediated current, or further fluctuation analysis of the current). The recording chamber was perfused with physiological solution bubbled with the gas mixture and maintain the temperature at 22-25°C. Then, we added 0.1 mM picrotoxin to the solution to block the GABA A -mediated response. We also added 4 μM 2-chloroadenosine to stabilize the evoked neural response. The patch recording pipettes were filled with the intracellular solution for voltage-clamp recordings. The resistance of the recording pipette in the aCSF was between 4 and 7 MΩ.
To analyze the function of CA3-CA1 synapses, bipolar tungsten stimulating electrodes (Unique Medical Co., Ltd., Tokyo, Japan) were placed in CA1~200-300 μm lateral from recorded cells. The electrically evoked EPSC amplitudes were measured from the peak of the postsynaptic current to the basal current level immediately before electrical stimulation. The stimulus intensity was increased until a synaptic response with an amplitude >−10 pA was recorded. AMPA/NMDA ratios were calculated as the ratio of the peak current at −60 mV to the current at +40 mV 150 ms after stimulus onset (40-60 traces averaged for each holding potential).

Miniature Postsynaptic Current Recordings
Miniature excitatory postsynaptic currents (mEPSCs) are thought to correspond to the responses elicited by the presynaptic release of a single vesicle of glutamate. In contrast, miniature inhibitory postsynaptic currents (mIPSCs) are thought to correspond of GABA. Increase in the amplitudes of mEPSCs and mIPSCs reflect postsynaptic transmission strengthening, while increase in the event frequency reflects increases in the number of functional synapses or the presynaptic release probability.
For the miniature recordings, the mEPSCs (−60 mV holding potential) and mIPSCs (0 mV holding potential) were recorded sequentially for 5 min in the same CA1 neuron. The miniature events were detected using the software, and the events above 10 pA were used for the analysis. We recorded at least for 5 min to determine the events frequency of mEPSCs or mIPSCs. The amplitudes of the events were averaged to obtain the mean amplitude. Bath application of an AMPA receptor blocker (CNQX, 10 μM) or GABA A receptor blocker (bicuculline methiodide, 10 μM) consistently blocked the mEPSC or mIPSC events, respectively.

Nonstationary Fluctuation Analysis
AMPA receptor-mediated evoked EPSCs and GABA A receptormediated mIPSCs were analyzed by nonstationary fluctuation analysis (Ghosh et al. 2015;Ono et al. 2016). To isolate fluctuations in current decay due to stochastic channel gating, the mean waveform was scaled to the peak of individual E(I)PSCs. The requirements for such analysis include a stable current decay time course throughout the recording and an absence of any correlation between the decay time course and peak amplitude. The relationship between the peak-scaled variance and the mean current is given by the following equation: where σ 2 is the variance, I is the mean current, N is the number of channels activated at the peak of the mean current, i is the unitary conductance, and b l is the background variance. In our experiments, 31-69 EPSCs and 14-133 IPSCs were analyzed from selected epochs in each of the cells in which there was no correlation between current decay (63% decay time) and peak amplitude (P > 0.05, Spearman's rank-order correlation test).
The weighted mean channel current can be estimated by fitting the full parabola or initial slope of the relationship. All the analysis was done using MATLAB software (MathWorks, MA, USA). The number of channels was further divided by the corresponding value of mean E(I)PSC amplitude to obtain the single channel current.

Self-Entropy Analysis
We used standard spreadsheet software (Excel 2010, Microsoft Co., Redmond, WA, USA) to calculate the self-entropy per neuron. First, we obtained 4 miniature parameters (mean mEPSC amplitude, mean mIPSC amplitude, mean mEPSC frequency, and mean mIPSC frequency) in individual CA1 pyramidal neurons. Then, we determined the distribution of appearance probability of 4 miniature parameters separately using 1dimensional Kernel density analysis. Geometric/topographic feature of the appearance probability was calculated using Kernel density analysis. Let X 1 , X 2 ,…, X n denote a sample of size n from real observations. The Kernel density estimate of P at the point x is given by the following equations: where K is a smooth function called the Gaussian kernel function and h > 0 is the smoothing bandwidth that controls the amount of smoothing. We chose Silverman's reference bandwidth or Silverman's rule of thumb (Silverman 1986;Sheather 2004). It is given by the following equation: where A = min (standard deviation, interquartile range/1.34). By normalizing integral value in untrained controls, we found the distribution of appearance probability at any point. Then, we calculated the appearance probability at selected points. All data points for probability in untrained and trained rats were converted to self-entropy (bit) using the Shannon entropy concept, defined from the Information Theory (Shannon 1948).
Based on the probability distribution, we calculated individual appearance probability of all recorded neurons. Then, the appearance probability of the neuron was converted to the self-entropy using Shannon's formula (= −LOG [appearance probability of the neuron, 2]) (Fig. 3F). For graphic expression, the distribution was visualized 2-dimensionally in the R software environment (R Foundation for Statistical Computing, Vienna, Austria) (Figs 3C, G,L,P and 4B,F,J,N).

Statistical Analysis
We used the paired t test to analyze IA latency and unpaired t test to analyze estimated open channel numbers. The AMPA/ NMDA ratio, mEPSC, mIPSC, and self-entropy were analyzed using 2-way factorial ANOVA in which the between-group factors were laterality and training. We used one-way factorial ANOVA to evaluate the difference in miniature responses between dorsal and ventral synapses. The Shapiro-Wilk test and F-test were used for normality and equality of variance, respectively. Because the self-entropy data had large variations within a group, we performed log (1 + x) transformation prior to the analysis (Mitsushima et al. 1994). P < 0.05 was considered significant.

Inhibitory Avoidance Task
To investigate a possible location of the contextual memory in 4 CA1 subfields, rats were subjected to an IA task ( Fig. 1A; Izquierdo et al. 1998;Mitsushima et al. 2011Mitsushima et al. , 2013. In this learning paradigm, rats were allowed to cross from a light box to a dark box, where an electric foot-shock (1.6 mA, 2 s) was delivered. Half an hour after the IA task, we measured the latency in the illuminated box as contextual learning performance. With paired foot-shock, the latency was much longer after training than before the training (t 11 = 14.0, P < 0.0001).

Miniature Postsynaptic Currents in Dorsal CA1 Neurons
To further analyze the learning-dependent synaptic plasticity, we recorded mEPSC or mIPSC in the presence of 0.5 μM TTX on both sides of the dorsal hippocampus (Fig. 3A). By changing the membrane potential, we sequentially recorded mEPSCs (at −60 mV) and mIPSCs (at 0 mV) from the same neuron as reported previously (Mitsushima et al. 2013). The postsynaptic currents are thought to correspond to the response elicited by a single vesicle of glutamate or GABA. In contrast, the number of synapses is known to affect the frequency of the events (Pinheiro and Mulle 2008).
At dorsal CA1 synapses, the strength of AMPA receptormediated excitatory inputs versus GABA A receptor-mediated inhibitory inputs was measured in each neuron and plotted 2-dimensionally (Fig. 3B). The Kernel analysis revealed the distribution of appearance probability (Fig. 3C). Although untrained rats exhibited low and narrow distribution range, IA-trained rats had a broad distribution suggesting a diversity of synaptic input onto CA1 neurons. For mEPSCs, 2-way ANOVA revealed a significant main effect of training (F 1,104 = 18.780, P < 0.0001), but the main effect of laterality (F 1,104 = 2.237, P = 0.14) or interaction (F 1,104 = 0.998, P = 0.32) was not significant (Fig. 3D). For mIPSCs, 2-way ANOVA revealed a significant main effect of training (F 1,104 = 44.627, P < 0.0001), but the main effect of laterality (F 1,104 = 0.724, P = 0.40) or interaction (F 1,104 = 0.299, P = 0.59) was not significant (Fig. 3E). These results suggest that the training bilaterally strengthened both excitatory and inhibitory synapses onto dorsal CA1 neurons, regardless of the hemisphere.
The balance of excitatory/inhibitory (E/I) inputs was obtained by dividing the mean mEPSC amplitude by the mean mIPSC amplitude of the same neuron. For the E/I balance of miniature amplitudes, the main effect of training (F 1,104 = 0.203, P = 0.65), laterality (F 1,104 = 2.469, P = 0.12), or interaction (F 1,104 = 0.002, P = 0.96) was not significant (Fig. 3H). Thus, the training did not affect the balance of mEPSC versus mIPSC amplitudes, suggesting the balance of excitatory versus inhibitory input strength onto dorsal CA1 neurons.

Self-Entropy in Dorsal CA1 Neurons
Based on the information theory of Shannon (1948), we calculated appearance probability of the mean amplitudes of mEPSCs and mIPSCs. First, we found the distribution of appearance probability in untrained controls (Fig 3C, left), and then we analyzed the appearance probability of all recorded neurons one-by-one. Each probability of single neuron was calculated as the self-entropy and plotted 2-dimensionally (Fig. 3F). For example, a point with high appearance probability (around the mean level of mE(I)PSC amplitude) indicates low self-entropy, whereas a point with very rare probability (a deviated point of mE(I)PSC amplitude) indicates high self-entropy.
We found that all recorded neurons exhibited different selfentropy each other (Fig. 3F). In the dorsal CA1, self-entropy in the excitatory synapse exhibited a significant main effect of training (F 1,104 = 9.322, P = 0.0029), but the main effect of laterality (F 1,104 = 1.229, P = 0.27) or interaction (F 1,104 = 0.879, P = 0.35) was not significant (Fig. 3F). Similarly, self-entropy in the inhibitory synapse exhibited a significant main effect of training (F 1,104 = 21.393, P < 0.0001), but the main effect of laterality (F 1,104 = 1.205, P = 0.27) or interaction (F 1,104 = 0.007, P = 0.93) was not significant (Fig. 3F). The Kernel analysis further visualized the density distribution (Fig. 3G). Thus, the training clearly increased the self-entropy of dorsal CA1 neurons in both hemispheres. The average level was 13.4 ± 0.2 bits in untrained rats, whereas IA-trained rats showed 30.3 ± 8.0 bits per single CA1 neuron (Fig. 3I).
For the E/I balance of miniature amplitudes, the main effects of training (F 1,103 = 0.244, P = 0.62), laterality (F 1,103 = 0.069, P = 0.79), and interaction (F 1,103 = 2.287, P = 0.13) were not significant (Fig. 3Q). The training did not affect the balance of mEPSC versus mIPSC amplitudes, suggesting the balance of excitatory versus inhibitory input strength onto ventral CA1 neurons.

Self-Entropy in Ventral CA1 Neurons
Using the distribution of appearance probability in untrained controls (Fig 3L, left), we calculated the self-entropy of all recorded neurons one-by-one (Fig. 3O). We found all recorded neurons exhibited different self-entropy each other. In ventral CA1 neurons, self-entropy in the excitatory synapse did not exhibit a significant main effect of training (F 1,103 = 0.001, P = 0.97), laterality (F 1,103 = 0.356, P = 0.55), or interaction (F 1,103 = 0.930, P = 0.34). Similarly, self-entropy in the inhibitory synapse did not exhibit a significant main effect of training (F 1,103 = 1.284, P = 0.26), laterality (F 1,103 = 1.158, P = 0.28), or interaction (F 1,103 = 2.023, P = 0.16). Thus, the training did not affect the self-entropy in either hemisphere, and the visualized density distribution was shown in Figure 3P. The average levels of selfentropy were 15.0 ± 0.2 bits (untrained) and 14.9 ± 0.3 bits (IAtrained) per single CA1 neuron (Fig. 3R).

Frequencies of the mE(I)PSC Events in Dorsal CA1 Neurons
The number of functional synapses is known to affect the frequency of the mEPSC/mIPSC events. At dorsal CA1 synapses, the frequency of mEPSC versus mIPSC events was measured in each neuron and plotted 2-dimensionally (Fig. 4A). The Kernel analysis revealed the distribution of appearance probability (Fig. 4B). Although untrained rats exhibited low and narrow distribution range, IA-trained rats had a broad distribution suggesting a diversity of the number of functional synapses onto a single CA1 neuron. For mEPSCs, 2-way ANOVA revealed a significant main effect of training (F 1,104 = 6.942, P = 0.0097), but the main effect of laterality (F 1,104 = 0.023, P = 0.88) or interaction (F 1,104 = 0.035, P = 0.85) was not significant (Fig. 4C). For mIPSCs, 2-way ANOVA revealed a significant main effect of training (F 1,104 = 13.893, P = 0.0003), but the main effect of laterality (F 1,104 = 1.760, P = 0.19) or interaction (F 1,104 = 0.054, P = 0.82) was not significant (Fig. 4D). These results suggest that the training increased in the number of excitatory and inhibitory synapses onto dorsal CA1 neurons in both hemispheres.
The balance of excitatory/inhibitory (E/I) frequency was obtained by dividing the mean mEPSC frequency by the mean mIPSC frequency of the same neuron. For the E/I balance of miniature frequency, the main effect of training (F 1,104 = 0.198, P = 0.66), laterality (F 1,104 = 0.149, P = 0.70), or interaction (F 1,104 = 0.78, P = 0.38) was not significant (Fig. 4G). Thus, the training did not affect the balance of mEPSC versus mIPSC frequency, suggesting the balance of the number of excitatory versus inhibitory synapses onto dorsal CA1 neurons.

Self-Entropy of the Frequency in Dorsal CA1 Neurons
Using the distribution appearance probability in untrained controls (Fig. 4B, left), we analyzed the appearance probability at selected points. The probabilities at all data points were calculated as the self-entropy and plotted 2-dimensionally (Fig. 4E). Although all recorded neurons exhibited different self-entropy each other, the Kernel analysis further visualized the density distribution (Fig. 4F). IA training dramatically increased the amount of information per dorsal CA1 neurons (Fig. 4H).

Frequencies of the mE(I)PSC Events in Ventral CA1 Neurons
IA training did not affect the frequency in ventral CA1 neurons. The frequency of mEPSC versus mIPSC events was measured in each neuron and plotted 2-dimensionally (Fig. 4I). Ventral CA1 neurons exhibited relatively wide distribution range in both untrained and IA-trained rats, and the Kernel analysis showed the distribution of appearance probability (Fig. 4J). For mEPSCs, the main effects of training (F 1,103 = 0.017, P = 0.90), laterality (F 1,103 = 2.144, P = 0.15), and interaction (F 1,103 = 0.22, P = 0.64) were not significant (Fig. 4K). For mIPSCs, the main effects of training (F 1,103 = 0.866, P = 0.35), laterality (F 1,103 = 0.922, P = 0.34), and interaction (F 1,103 = 0.273, P = 0.60) were not significant (Fig. 4L). Thus, the training affected frequency of neither mEPSC nor mIPSC regardless of the hemispheres. These results suggest that the training did not affect the number of excitatory and inhibitory synapses onto ventral CA1 neurons, regardless of the hemisphere.
For the E/I balance of miniature frequency, the main effects of training (F 1,103 = 0.75, P = 0.39) and interaction (F 1,103 = 0.086, P = 0.77) were not significant (Fig. 4O), but right side of CA1 neurons exhibited higher E/I balance of the frequency than left side (F 1,103 = 4.865, P = 0.03). The results suggest that CA1 neurons receive more excitatory inputs in the right hemisphere than in the left hemisphere, providing a synaptic evidence of laterality. Meanwhile, the training did not affect the balance of mEPSC versus mIPSC frequency, suggesting the balance of the number of excitatory versus inhibitory synapses onto ventral CA1 neurons.

The Number of Postsynaptic AMPA Receptor Channels
To examine whether IA alters the number of AMPA receptors, we used evoked EPSC responses to calculate the number of opening AMPA receptors at dorsal or ventral CA3-CA1 synapses using nonstationary fluctuation analysis (Fig. 5A). The number of open AMPA receptors was significantly larger in IA-trained rats than untrained rats at dorsal CA3-CA1 synapses (t 33 = 2.28, P = 0.029), but not at ventral CA3-CA1 synapses (t 18 = 0.32, P = 0.75, Fig. 5B). Although the single-channel current was significantly greater at ventral synapses than dorsal synapses (F 1,53 = 6.02, P = 0.017), the training did not affect the single-channel current (Fig. 5C). These results suggest that the training promotes postsynaptic plasticity by increasing the number of AMPA receptor-channels at dorsal but not ventral CA3-CA1 synapses.

The Number of Postsynaptic GABA A Receptor Channels
To examine whether IA alters the number of GABA A receptors, we used mIPSC responses to calculate the number of opening GABA A receptors at dorsal or ventral CA1 synapses (Fig. 5D).
The number of open GABA A receptors was larger in IA-trained rats than untrained rats at dorsal CA1 synapses (t 96 = 2.30, P = 0.024), but not at ventral CA1 synapses (t 81 = 0.44, P = 0.66, Fig. 5E). Although ventral synapses possessed more Cl − channels than dorsal synapses (F 1,179 = 6.532, P = 0.011), neither area nor the training affected the single-channel current (Fig. 5F). These results suggest that the training promotes postsynaptic plasticity by increasing the number of GABA A receptorchannels at dorsal but not ventral CA1 synapses.

Discussion
Rat hippocampus is known to contain approximately 311 500 CA1 pyramidal neurons, receiving 13 059-28 697 CA3-CA1 synapses and up to 1 742 temporoammonic synapses from entorhinal cortex per single neuron (Bezaire and Soltesz 2013). Although both structural and functional heterogeneity are known in dorso/ventral or left /right CA1 neurons, the location of learning-induced synaptic plasticity has not been specified in the broad CA1 area. Here we found that the training increased AMPA receptor-mediated responses at dorsal CA3-CA1 synapses in both hemispheres, whereas ventral CA3-CA1 synapses did not show the plasticity in either hemisphere. The specified CA1 subfields of learning-induced plasticity provide a synaptic evidence of dorso/ventral heterogeneity at the synapse level. Nonstationary fluctuation analysis further revealed evidence at a single channel level. We found the training significantly increased the postsynaptic number of open AMPA receptors at dorsal CA1 synapses, whereas the training did not affect the ventral CA1 synapses (Fig. 5B). By combining in vivo gene delivery and in vitro patch-clamp recordings, we previously demonstrated that contextual learning depends on synaptic delivery of GluA1-containing AMPA receptors at dorsal CA1 synapses at the molecular level (Mitsushima et al. 2011). The increase in the number of open channels without changes in the single-channel current further revealed learning-induced current changes at the AMPA receptor-mediated CA1 synapses.
Although laterality was not clear in our laboratory conditions, right side of CA1 exhibited more power of gamma oscillation and spine density than left side in rats reared in the spatially enriched conditions (Shinohara et al. 2013). In humans, patients with unilateral damage to the right hippocampus exhibit spatial memory deficits (Abrahams et al. 1996), whereas damage to the left hippocampus impairs verbal semantic representation (Richardson et al. 2004). In the present study, the right sides of synapses tended to have greater AMPA/NMDA ratios, mEPSC amplitudes, and self-entropy than the left side after training, though the laterality was not significant. Fibers through the ventral hippocampal commissure are known to connect bilateral CA1 (Amaral and Witter 1995;Kawakami et al. 2003), inducing high coherence of CA1 theta waves during running or REM sleep in the freely moving state (Patel et al. 2012). As unilateral CA1 blockade of AMPA receptor delivery (Mitsushima et al. 2011(Mitsushima et al. , 2013 fail to impair the IA learning, bilateral CA1 neurons may work together to compensate for impairment of the other in rats in normal laboratory conditions. The question arises as to whether the synaptic strength contributes to memory. In regards to the excitatory synapses, contextual learning requires AMPA receptor delivery, as bilat-