Optogenetic inhibition-mediated activity-dependent modification of CA1 pyramidal-interneuron connections during behavior
Abstract
In vitro work revealed that excitatory synaptic inputs to hippocampal inhibitory interneurons could undergo Hebbian, associative, or non-associative plasticity. Both behavioral and learning-dependent reorganization of these connections has also been demonstrated by measuring spike transmission probabilities in pyramidal cell-interneuron spike cross-correlations that indicate monosynaptic connections. Here we investigated the activity-dependent modification of these connections during exploratory behavior in rats by optogenetically inhibiting pyramidal cell and interneuron subpopulations. Light application and associated firing alteration of pyramidal and interneuron populations led to lasting changes in pyramidal-interneuron connection weights as indicated by spike transmission changes. Spike transmission alterations were predicted by the light-mediated changes in the number of pre- and postsynaptic spike pairing events and by firing rate changes of interneurons but not pyramidal cells. This work demonstrates the presence of activity-dependent associative and non-associative reorganization of pyramidal-interneuron connections triggered by the optogenetic modification of the firing rate and spike synchrony of cells.
Introduction
It is increasingly recognized that plastic reorganization of brain circuits needed for learning and other cognitive functions involves not only principal cells but also local inhibitory interneurons (Buzsáki, 2010). A large body of work has demonstrated that excitatory connections onto inhibitory interneurons, as well as inhibitory connections on principal cells, are often plastic (Kullmann and Lamsa, 2007; McBain and Kauer, 2009). However, the rules governing the plasticity of excitatory synapses on inhibitory interneurons are not always similar to that targeting other principal cells (Bartos et al., 2011; Lamsa et al., 2007; Pelkey et al., 2017). Even in the CA1 region of the hippocampus, plasticity rules can be different, depending on the experimental conditions and fibers stimulated. Although in the majority of cases some pyramidal-interneuron cell connections show anti-Hebbian non-associative plasticity, others show weight changes that are governed by Hebbian rules (Le Roux et al., 2013; Nissen et al., 2010; Topolnik et al., 2009). In addition, some inhibitory interneuron types in the hippocampus do not seem to possess plastic synapses with their excitatory inputs, such as the CCK cells (Nissen et al., 2010).
In vivo work, primarily during anesthesia, has also demonstrated that plastic alterations can occur between afferent excitatory fibers and CA1 interneurons (Buzsáki and Eidelberg, 1982; Lau et al., 2017). This work showed that stimulation of these fibers could either up- or downregulate evoked spike responses, depending on the interneuron subtype. However, our knowledge about the precise reorganization of pyramidal-interneuron connections during behavior is limited because of the technical challenge of directly performing patch-clamp recordings from monosynaptically connected pyramidal cell-interneuron pairs during such conditions. It is, however, possible to study these connections indirectly by identifying monosynaptically connected pyramidal cell-interneuron pairs by using cross-correlation analysis of the spike timing and measuring spike transmission probability between them (Csicsvari et al., 1998; Csicsvari et al., 2003; Marshall et al., 2002). Early work demonstrated the behavioral state-dependent modulation of such connections (Csicsvari et al., 1998). Moreover, changes in spike transmission probability have been seen in the prefrontal cortex during behavioral tasks (Fujisawa et al., 2008). In the hippocampus, spatial learning can cause lasting changes in these connections (Dupret et al., 2013). While such studies provide strong evidence of plasticity at excitatory-interneuron synapses, so far, no data has established a causal link between pre- and postsynaptic firing to changes in connection strength during behavior.
In this study, we optogenetically interfered with the circuit function by activating Halorhodopsin or Archaerhodopsin in a subpopulation of pyramidal cells and interneurons. This manipulation led to the inhibition of a subset of pyramidal cells and interneurons and also light-triggered disinhibition of many pyramidal cells (Gridchyn et al., 2020; Schoenenberger et al., 2016). Here, we examined whether these light-induced rate changes and the associated network effects could lead to the lasting reorganization of pyramidal-interneuron connection weights, as assessed by monosynaptic spike transmission probabilities.
Results
We recorded multiple unit and field potential activities from the dorsal hippocampus in five rats, during exploration and quiet immobility sessions. In these rats, Halorhodopsin (NpHR-YFP, n = 4 rats) or Archaerhodopsin (ArchT, n = 1 rat) was expressed in the dorsal CA1 region of the hippocampus under the control of the CaMKIIα promoter using an adeno-associated virus (AAV2/1). In four rats, fifteen independently-movable tetrodes and one 200 µm optic fiber centered in the middle of the tetrode bundle targeted the dorsal CA1 region while, in the remaining animal, 24 tetrodes and four optic fibers were used (see Materials and methods). We recorded during four 25 min exploration sessions in which first a familiar environment (FAM1) was explored followed by a novel environment (NOV), and finally, the animals were returned to the familiar environment for the remaining two sessions. During the second familiar exploration session laser stimulation was applied (FAML) in a fixed part of the environment but not in the last exploration (FAM2, see Figure 1A). We tested whether the light application affected the behavior of the animals (Figure 1—figure supplement 1). In all sessions, neither the average speed nor the occupancy within the light stimulation sector were significantly different, compared to the part of the environment where no light was triggered (all p>0.5607). We identified monosynaptically-connected pyramidal cell-interneuron pairs by calculating the cross-correlation of their spike firing times and testing for the presence of a short-latency (1–2 ms) sharp (1–2 ms wide) peak. This peak indicates the presence of a monosynaptic connection in which the presynaptic pyramidal cell can discharge the postsynaptic interneuron within a short latency. In turn, the magnitude of the peak, that is, its transmission probability, reflects the connection weight between a given cell pair. Changes in firing rate across sessions by either or both cells in the pair would result in an alteration of the magnitude of the peak that does not reflect a change in this connection weight, but rather a general change in the probability of joint firing. To account for this, throughout all the analysis, we measured the chance probability that the pair fires together, by averaging the correlation probabilities over the 10–50 ms time bins and subtracting this from the peak. Altogether, we identified 78 pyramidal cell-interneuron pairs in these recordings (see Materials and methods).

Light-induced firing changes in the CA1 region by halorhodopsin-mediated inhibition of a subgroup of pyramidal cells and interneurons.
(A) Experimental paradigm: on each recording day the animal was exposed three times to the same familiar environment (FAM1, FAML, and FAM2), including one session in which light stimulation was triggered (FAML) as the animal explored a defined sector of the environment (1/3 – ½ of the arena). In addition, the animal also explored a novel environment (NOV). Each behavioral session was flanked by sleep, with 500 ms light pulses given in the last. (B) examples of cells in which light application suppressed activity and triggered an elevated rate, through disinhibition. int, interneuron; pyr, pyramidal cells. Light responses were measured during the last rest session by applying 500 ms test light pulses. The histograms show the probability of spiking within the 20 ms time bins. (C) The mean firing rate of the postsynaptic interneurons (left) and presynaptic pyramidal cells (right) that were part of a detected monosynaptic cell pair were plotted during FAM1 vs. FAML sessions. Lines represent the x = y line. Note that the majority of interneurons were inhibited by the light, whereas several pyramidal cells exhibited either prominent suppression or excitation of their rate. (D) Monosynaptic spike transmission probabilities also exhibit alterations during the FAML session with more cell pairs showing a reduction of spike transmission probabilities relative to FAM1.
The light stimulation inhibited not only a selected population of pyramidal cells but also many interneurons, while a further group of pyramidal cells increased their firing due to disinhibition, as assessed directly by their light responses to brief light pulses in the rest session at the end of the recordings (Figure 1B). In our previous work, we showed that these disinhibited pyramidal cells only increased their firing after the maximum light-mediated suppression on interneurons (Gridchyn et al., 2020; Schoenenberger et al., 2016). We also showed before (Schoenenberger et al., 2016) that both somatostatin- and parvalbumin immunopositive interneurons can express transgenes following AAV-mediated transduction in agreement with earlier work (Nathanson et al., 2009). It is possible, however, that other adeno-associated virus serotypes or the usage of the same virus in other brain regions may yield principal cell-specific expression. As a result, many pyramidal cell - interneuron pairs with monosynaptic connections showed changes in firing rate during the FAML session, when compared to FAM1 (Figure 1C). The altered network activity during the light session is demonstrated by significantly lower correlation of the FAM1 vs FAML rates as compared to rates measured in alternating 5 s time windows within FAM1, both for pyramidal cells and interneurons (all p<0.0001, Z-test). However, no significant differences were found in the median of FAM1 and FAML firing rates (interneuron p=0.3213 pyramidal cell p=0.1448, Mann-Whitney test). The spike transmission probability of these cell pairs was also changed (Figure 1D). As for firing rates, the correlation FAM1 vs FAML spike transmission values was lower than that measured in alternating 5 s time windows within FAM1 (p<0.0001, Z-test). Moreover, there was a significant reduction in the median of the spike transmission probabilities from FAM1 to FAML (p<0.01. Mann-Whitney test). Because in the spike transmission measurements the chance probability that cells randomly fire together was compensated for, light-induced network modifications altered the ability of the pyramidal cell to drive the postsynaptic interneuron, beyond that of the firing rate alterations-mediated changes. Changes in connection weight between cell pairs during the FAML session could be either transient or reflect longer-term plasticity that outlasts optogenetic stimulation. Moreover, connection strength could change as place cells remap their place fields during exploration of a different environment (Wilson and McNaughton, 1993). We, therefore, tested whether significant changes in the spike transmission of monosynaptic pairs can be seen across sessions, relative to the baseline identified in FAM1 (Figure 2A–B). To do this, we generated a score that represented the absolute value of normalized spike transmission differences between sessions (difference divided by the sum, see Materials and methods). Overall, this score was the largest between NOV-FAM1 and FAML-FAM1 sessions. However, while the changes between FAM2-FAM1 sessions were about 30% weaker relative to the others, they were still significantly larger than zero (all p<0.001; ANOVA). In addition, changes between FAM2-FAM1 sessions were significantly larger than changes within FAM1 sessions as assessed by correlations measured in alternating 5 s time windows (p<0.0268, F-test), independent of the variability across animals (p=0. 0562, Likelihood-ratio test).

Light application triggered lasting changes in spike transmission probabilities in the same environmental context.
(A) Representative examples of monosynaptic cross-correlations demonstrating altered spike transmission probabilities across different sessions. Left histograms show the cross-correlations calculated during the entire recording session (all), which were used to detect the monosynaptic pairs. The remaining histograms show the cross-correlations at different sessions. Chance joint firing probability was estimated by the average cross-correlation values in the ±10–50 ms bins and subtracted. Each bin represents a 1 ms time windows in [−50 ms, +50 ms] intervals. (B) Mean (± SEM) absolute difference of spike transmission probabilities, measured as relative change (difference/sum) between the odd- and even-numbered 5 s intervals of the FAM1 session and between FAM1 and other sessions. Mean (± SEM) absolute difference of spike transmission probabilities, measured as relative change (difference/sum), relative to the first FAM1 session. Note the significant reorganization of the spike transmission probabilities across all sessions, with FAM2-FAM1 being the weakest. *p=0.0268, ***p<0.0001. (C) Prediction (i.e., correlation) of transmission probabilities in FAM2 with those in the previous exploration sessions ***p<0.0001. (D) Partial correlations to illustrate the influence of each session on FAM2, while removing the effect of other behavioral sessions. Note that the linear mixed model comparison analysis showed that the NOV session did not influence FAM2 spike transmission when the FAM1 spike transmissions were taken into consideration, whereas FAM1 did influence FAM2. Significance for linear mixed model comparison is indicated. *p=0.0105, ***p<0.0001, ns not significant. (E) Spike transmission values plotted in the FAM1 and FAM2 sessions. Cell pairs that increased (red) and decreased (blue) their spike transmission in the FAML relative to FAM1 are displayed separately. Diagonal line: x = y. (F) Relative (difference/sum) changes of spike transmission probabilities between FAML-FAM1 predict those of FAM2-FAM1 changes. The relative FAM2-FAM1 changes of cell pairs with reduction (blue) and increase (red) in spike transmission from FAM1 to FAML are also displayed along a single line on the right to illustrate the negative and positive bias of these groups. The solid diagonal line represents the regression line for the data. Horizontal line: median.
These population changes suggest that pyramidal cell-interneuron synaptic weights were, to some degree, reorganized by both exposure to a new environment and artificially by light stimulation. However, connection strengths were different between the first and last exposure to the same familiar environment (FAM1 and FAM2), indicating that a more lasting change had also occurred. We set out to determine the factors that accounted for the connection strength observed in FAM2 by testing at the population level whether spike transmission in FAM2 was predicted by that observed in the previous sessions. In addition, we controlled for the possible variability across animals by including animal identity as a variable into the analysis (Figure 2C–D). Spike transmissions in FAM2 were strongly predicted by the observed connection weights in all the previous sessions (all p<0.0001, F-test, Figure 2C), indicating that connection strength between the pairs was only partially reorganized. The variability across animals did not significantly account for variability in the spike transmission (all p>0.5656, Likelihood-ratio test). We then used a linear model comparison to reveal which behavioral sessions best explained the monosynaptic connection strengths in FAM2 (Figure 2D). We found that both FAM1 and FAML predicted FAM2 spike transmission, independent of the other (all p<0.0105, F-test). However, the model comparison showed that the NOV no longer predicted FAM2 when the effect of FAM1 was taken into consideration (p=0.4453, F-test) nor the variation across animals contributed (all p>0.5656, Likelihood-ratio test). Thus, the changes in connection weights during NOV did not influence the subsequent weights observed in FAM2, which instead was explained by weights in both FAM1 and FAML, independently. Therefore, while spike transmission values were more similar within the same environmental context, light application significantly biased spike transmission and caused lasting changes in FAM2 relative to FAM1.
The significant influence of the FAML session on FAM2 suggests that the light-induced alterations of the network activity led to lasting changes in the spike transmission probabilities even in the absence of light in the same environmental context. Direct optogenetic inhibition of some cells and the indirect firing increase of others led to either an increase or a decrease of spike transmission probabilities during the light application. Therefore, next, we tested whether the direction of change in spike transmission in FAML relative to FAM1 predicted similar change from FAM1 to FAM2 (Figure 2E–F). Indeed, those cell pairs in which light application led to a decrease in spike transmission relative to FAM1 also maintained weaker spike transmission values in FAM2, while cell pairs with light-enhanced transmission showed a persistent increase in transmission probability in FAM2 (all p<0.0001, F-test). Moreover, the relative change of spike transmission (difference divided by the sum) between FAML-FAM1 predicted the score change from FAM1 to FAM2 (p<0.0001, F-test), indicating that larger relative changes in spike transmission between FAML-FAM1 were accompanied by similarly larger changes between FAM2-FAM1. Thus, light-induced changes in neuronal firing during FAML may lead to lasting modifications of spike transmission. The variability across animals did not significantly account for variability in the transmission probability change (p>0.3046, Likelihood-ratio test).
The light application can change the firing rate of the presynaptic pyramidal cell or the postsynaptic interneuron, which may, in turn, account for the observed plastic changes in transmission probability. Therefore, to address whether our effects reflected a pre- or postsynaptic mechanism, we assessed the relationship between firing rate changes within connected pairs and the modification of their monosynaptic connection. To do this, we calculated the relative firing rate changes between FAML-FAM1 for both pyramidal cells and interneurons, a measure that reflects the influence of light on the baseline firing of these cells. We then analyzed whether this measure predicted changes in transmission probabilities between FAM1 and FAM2, which reflects the longer-lasting change in synaptic strength. As a control, we also measured relative firing rate changes between FAM2-FAM1 because rate alterations reflecting the alterations in the average excitatory inputs cells received. Such changes in excitability may have influenced spike transmission beyond the rate alteration-mediated changes of the chance joint firing probability, which later were already compensated for by normalizing the histograms (Figure 3A–B). We found that interneuron rate changes of both FAML-FAM1 and FAM2-FAM1 influenced FAM2-FAM1 spike transmission changes, even when each other’s influence was taken into account (all p<0.0064, Likelihood-ratio test). However, pyramidal cell rate changes during FAML no longer significantly influenced FAM2-FAM1 spike transmission changes when FAM2-FAM1 rate change was taken into account (p=0.0999, Likelihood-ratio test). This suggests that changes in the excitability of interneurons during the light application, as assessed by rate changes, influenced the spike transmission strength subsequently in FAM2 even when the FAM2-FAM1 rate alterations of the interneuron were compensated for. In these models, the variability across animals did not significantly account for variability in the transmission probability change (all p>0.0907, Likelihood-ratio test). In addition to the changing firing rate, the light application can cause remapping in a subpopulation of cells (Schoenenberger et al., 2016). However, a change in spike transmission in FAM2 did not predict the degree of pyramidal place field remapping (p=0.1612, F-test).

Light-induced firing rate changes of interneurons but not pyramidal cells influenced lasting familiar environment-associated spike transmission alterations between before and after the light application session.
The relative changes in rate and transmission probabilities are expressed as a score throughout (difference/sum). The influence of light-induced firing rate changes on spike transmission alterations between before and after the light application session. The relative changes in rate and transmissionprobabilities are expressed as a score throughout (difference/sum). (A) The correlation predicts relative FAM2-FAM1 spike transmission changes based on relative rate changes of FAML-FAM1 and FAM2-FAM1 sessions. Both correlations (left) and partial correlations (right) are shown. The comparisons of linear mixed models with one or both rate change variables show that interneuron rates in both FAML and FAM2 independently influence FAM2-FAM1 spike transmission changes. (B) same as (A) but for pyramidal cells. In this case, FAML but not FAM2 rates independently predict FAM2-FAM1 spike transmission changes. (C) Relative FAML-FAM1 rate change of interneurons versus the relative spike transmission probability changes (FAM2-FAM1) with their presynaptic pyramidal partner. Right plots spike transmission changes were plotted for the rate decrease (blue) and increase (red) pairs. Horizontal lines: median. Note that almost all pairs that exhibited an interneuron rate increase during FAML also increased their spike transmission in FAM2 and the rate decrease group exhibited a significantly smaller spike transmission change than the rate increase group. (D) Same as (C) but for pyramidal rate changes. Here the direction of rate change does not predict spike transmission changes. *p<0.0256, **p<0.0068, ***p<0.0001, ns not significant.
The finding that interneuron rate change between FAM2-FAM1 itself independently influenced FAM2-FAM1 spike transmission changes suggests that the excitability of the postsynaptic interneuron has a strong influence on the strength of spike transmission. Spike transmission changes we detected in the NOV session may, in part, have been caused by the excitability change of the interneurons. To test whether rate changes of pyramidal cells or interneurons in NOV influenced spike transmission in FAM2, we calculated normalized (difference divided by the sum) spike transmission and rate changes, in order to predict weight change separately for both cell types (Figure 4). As before, only interneuron rate changes predicted spike transmission changes, even when the pyramidal rate change was taken into consideration (interneuron: p<0.0001; pyramidal: p=0.1672, Likelihood-ratio test). A similar analysis was performed for the FAML session itself, where again, we found that only interneuron rate alterations predicted spike transmission alterations during the light application, even when each-others’ contribution was considered (interneuron: p<0.0001; pyramidal: p=0.5885, Likelihood-ratio test). In these models, the variability across animals did not significantly account for variability in the transmission probability changes in NOV-FAM1 and FAML-FAM1 (all p>0.382, Likelihood-ratio test).

The influence of interneuron rate change on spike transmission changes in the NOV and FAML sessions, relative to FAM1.
(A) Left: correlation of pyramidal and interneuron relative rate changes (NOV-FAM1) and the corresponding relative spike transmission changes. Partial correlations are also shown on the right. In both cases, interneuron rate changes predict the corresponding spike transmission changes but not pyramidal cells according to linear mixed model comparison. (B) same as (A) but comparing FAML-FAM1. Interneuron rate changes had a strong influence on spike transmission changes. All relative rates and transmission changes are measured as difference/sum. ***p<0.0001, ns not significant.
Next, we examined whether the direction of firing rate change during light application influenced the spike transmission change between FAM1 and FAM2 (Figure 3C–D). As before, we considered FAM1 as a baseline and generated a normalized score for the rate and weight change. Interneurons that exhibited elevated or reduced firing rate during FAML session exhibited significantly different changes in monosynaptic spike transmission probability across FAM2 and FAM1 (p<0.0001, F-test) independent of the variability across animals (p=0.3372, Likelihood-ratio test). Pyramidal cells did not show such a relationship (p=0.4499, F-test) and the variability across animals did not influence this result (p=0.1359, Likelihood-ratio test). Therefore, suppressed interneurons tended to weaken their monosynaptic weights with presynaptic pyramidal cells, whereas those that increased their rate exhibited increased weights. These effects lasted after the light application in the same environmental context. To confirm that the observed changes in monosynaptic weight were indeed mediated by a postsynaptic change in the interneuron firing rate, we differentiated monosynaptic cell pairs into four groups, according to whether the pair exhibited a rate increase or decrease, of both pyramidal cells and interneurons (Figure 5). A two-way ANOVA analysis showed that the interneuron increase and decrease groups were significantly different, independent of the pyramidal increase and decrease as a factor (p<0.0004, F-test), but not the pyramidal group (p=0.4067, F-test), and the variability across animals did not influence the result (p=0.2749, Likelihood-ratio test) and no significant interactions were seen between cell pair groups (p=0.6109, F-test).

Frequency distribution of relative FAM2-FAM1 spike transmission changes for monosynaptic cell pairs according to the direction of change of the pre- and postsynaptic cell partner.
We found that the modulation of interneuron activity during light stimulation or exposure to a novel environment directly influenced changes in transmission probabilities between putative monosynaptic connections of pyramidal-interneuron cell pairs. This raises the possibility that such changes are activity-dependent. To examine this, we measured the number of instances in which pyramidal cell firing occurred in 10 ms, 20 ms, 50 ms, and 100 ms time windows before or after the interneuron spike as well as their sum for all spike-pairing events (Figure 6). The relative change (difference/sum) of spike pairing events was calculated between FAML and FAM1 and these spike pairing changes predicted significantly FAM2-FAM1 spike transmission change in all three cases for 50 ms window (p<0.00099, F-test) even when multiple testing correction was performed while the variability across animals did not influence the result (all p>0.3913, Likelihood-ratio test). In the other tested time intervals, the correlations were not significant (all p>0.0517, F-test). Moreover, spike pairing event number no longer predicted the FAM2-FAM1 spike transmission changes when interneuron and pyramidal changes were together taken into account, (all p>0.417, F-test). However, spike pairing itself was strongly predicted (all R2>0.859) by pyramidal and interneuron firing rates, explaining why spike transmission changes could not be predicted independently from the combined interneuron and pyramidal rates by spike pairing.

The influence of pyramidal cell-interneuron pairing on spike transmission changes.
(A) The number of pairing events in FAML predicted FAM2-FAM1 spike transmission changes. The number of spike pairing events were measured in cases when interneuron spike followed by pyramidal spike within 50 ms (+50 ms) and those where it preceded that (−50 ms) and the sum of both events (all). The relative difference (difference divided by the sum) of pairing event numbers between FAML and FAM1 was calculated. Pairing change predicted with relative spike transmission change (difference divided by the sum, unfilled white histograms on the left panels). We also examined whether the light-induced interneuron and pyramidal firing rate changes that itself altered the number of pairing events alone can explain this prediction. The number of pairing events no longer predicted spike transmission changes when both pyramidal and interneuron rate changes were taken into account according to the linear mixed model comparison. (B) Change in the number of spike pairing events strongly predicted the change in interneuron and pyramidal firing rates. R2 values and their 95% confidence intervals are plotted. ***p<0.00099, ns not significant.
Discussion
Here we showed that light-induced, optogenetic alterations of the CA1 network activity can trigger lasting alterations of the monosynaptic spike transmission probability of pyramidal-cell interneuron pairs. During the light session, both changes in the postsynaptic interneuron rates and the pairing probability of cell firing predicted the changes in monosynaptic spike transmission within the same familiar environment, when sessions before and after the light interference were compared. In addition, we observed spike transmission changes in the novel environment relative to the familiar environment, which took place before the light application. These changes were specific to the novel environment, however, and were not maintained during the subsequent familiar sessions. This suggests that altered interneuron firing rate and the activity-dependent alterations of the pyramidal-interneuron spike pairing can modify pyramidal-interneuron connection weights during exploratory behavior. Our study did not use control animals in which only YFP was expressed. Therefore, we cannot exclude the possibility that optogenetic channel expression, or, perhaps, light application enhanced the plasticity on pyramidal-interneuron synapses. Yet, we observed similar activity-dependent changes during spatial learning before (Dupret et al., 2013). So, it is likely that the optogenetic, light-mediated rate alteration was a primary driver of the activity-dependent, lasting connection weight changes.
We used a measure of spike transmission probability that compensated for the changes in the chance probability that the two cells fire together as a result of firing rate alterations. However, the average depolarization level of the cell, as reflected by its mean firing rate, can lead to more efficient spike transmission, even without changes in the synaptic weight. Indeed, the spike transmission changes from FAM1 to FAM2 were influenced by the rate changes of the postsynaptic interneuron between these sessions. In turn, this suggests that the postsynaptic interneuron’s general level of depolarization can influence spike transmission. However, in a similar manner, changes in FAML-FAM1 interneuron rate also predicted FAM2-FAM1 spike transmission changes. Indeed, when FAML interneuron rate increased, a stronger spike transmission was seen subsequently in FAM2, while reduced spike transmission was associated with reduced interneuron rate. Considering that the interneuron rate increase in FAML was not directly mediated by the light, we cannot exclude the possibility that stronger pyramidal inputs caused the interneuron rate increase in FAML that is caused by the plastic strengthening of these connections. Nevertheless, the excitability of interneurons in FAM2 alone, which may be indicative of plasticity-mediated input to interneurons, did not explain the spike transmission changes. Indeed, rate changes in FAML could predict spike transmission changes in FAM2-FAM1 independent of FAM2-FAM1 rate changes. That is, FAML-FAM1 rate changes predicted FAM2-FAM1 spike transmission changes even when the FAM2-FAM1 interneuron rate changes were accounted for. Therefore, FAML-FAM1 interneuron rate changes had further predictive value beyond those observed by FAM2-FAM1 rates changes and consequently excitability/depolarization alterations in FAML that were no longer present in FAM2 still predicted FAM2-FAM1 spike transmission changes. This finding indicates that interneuronal excitability changes during light application session caused lasting changes of the pyramidal interneuron connections, beyond any lasting nonspecific excitability changes that occurred between FAM2-FAM1.
We observed changes in spike transmission between FAM1 and the novel environment, which were larger in amplitude than those across the familiar environment before and after the light application. Exposure to the novel environment nevertheless did not influence changes in the familiar environment. Can we expect that pyramidal-interneuron weights change from one environment to another but they revert to the previous configuration when the animal is returned to the first environment? Although we cannot exclude this possibility, it is more likely that the average depolarization level of each interneuron is different across different environments, which in turn reveals different monosynaptic connections and connection strengths. Interneuron rates reorganize across the familiar and novel environment, which may reflect the influence of different nonspecific neurotransmitters (Nitz and McNaughton, 2004; Wilson and McNaughton, 1993). Moreover, spike transmission changes across the familiar and novel environments were predicted by the rate changes of interneurons but not pyramidal cells. This suggests that postsynaptic effects such as differences in interneuron depolarization levels contributed to the changes in spike transmission between different environments. In addition to nonspecific neuromodulator transmitters, presynaptic place cells that were specifically active in the novel environment may have caused changes in the interneuron depolarization as well. Although novel environments may not entirely reorganize pyramidal connections, spatial learning is able to do so (Dupret et al., 2013). During the course of spatial learning, some interneurons increase their rates while others decrease, which are accompanied by changes in spike transmission probabilities of monosynaptic pairs. However, these spike transmission changes depended on both pyramidal and interneuron rate changes. One common aspect of the light and the spatial learning-mediated monosynaptic spike transmission changes is that, in both cases, it took place in a familiar environment in which some of the place cells remapped their place fields. In Dupret et al., 2013 paradigm, some cells altered their place fields to represent the changed goal locations while in our paradigm, some of the place cells whose in-field firing was inhibited by the light remapped their place fields (Schoenenberger et al., 2016).
Can rules derived from in vitro observations, or those seen in vivo during anesthesia through afferent stimulations explain our findings during behavior? Our interneurons were recorded in the CA1 pyramidal layer where Ca++ permeable AMPA receptors mediate the primary form of LTP and LTD. This form of plasticity requires the stimulation of their synaptic inputs and the hyperpolarization or a non-depolarized state of the interneuron (Le Roux et al., 2013; Nissen et al., 2010). In addition, metabotropic glutamate receptors can further regulate dendritic Ca++ levels and the direction of synaptic plasticity (Camiré and Topolnik, 2014). In our case, the firing rate alteration of the interneurons was the strongest predictor of spike transmission change. The firing rate of the interneuron during FAML may reflect both the afferent excitation level of the cell as well as the light-induced inhibition; both of which could contribute to plastic changes. Excitability changes caused by the light stimulation are also expected to contribute to our observed results because both FAML-FAM1 and FAM2-FAM1 rate changes independently predicted FAM2-FAM1 spike transmission changes. It has been shown that Schaffer collateral stimulation enhances the excitability of CA1 PV cells following the stimulation, mediated via mGluR5 receptors (Campanac et al., 2013). In our dataset, the majority of interneurons showed a reduction in firing rate. Therefore, interneurons may be able to undergo both up- and downregulation of their excitation levels, which are not exclusively controlled by Schaffer collateral inputs.
We also saw that spike pairing of the pyramidal cells and interneuron in 50 ms time windows weakly influenced spike transmission, independent of the light-induced rate changes of these cells. Indeed, parvalbumin-positive CA1 interneurons exhibit NMDA-dependent associative plasticity as well (Le Roux et al., 2013) on their feedback connections from CA1 pyramidal cells. This may explain our spike pairing results. In vivo work suggested that theta-frequency afferent stimulation was optimal to induce LTP or LTD-like changes (Lau et al., 2017). Our light application occurred during theta oscillations, where such afferents would indeed provide theta-rhythmic stimulation of the CA1 interneurons. However, such a pairing relationship was observed only for a 50 ms time window. Moreover, it was no longer significant when pyramidal and interneuron rate changes were together taken into account. Because our optogenetic manipulation altered rates of individual cells without specifically influencing pyramidal-interneuron spike pairings, the combination of pyramidal and interneuron rates strongly predicted spike pairing probability. This can explain why spike pairing no longer predicted spike transmission changes when these rates were taken into account. After all, this result shows that rate-predicted spike pairing numbers are as good as the real ones to predict spike transmission changes. Nevertheless, we cannot exclude the possibility that the independent rate alterations of pyramidal cells and interneurons in FAML governed spike transmission probability changes, without spike pairing itself directly influencing it. Future work in which interneuron (or a certain genetic type) firing rate is selectively altered by optogenetics may provide further evidence for the independent contribution of postsynaptic interneuron depolarization in plasticity. Nevertheless, even in such manipulations, indirect alteration of pyramidal rates (e.g., because of disinhibition) is expected to occur.
Overall, our data indicate that during active behavior, changes in interneuron excitability that is coupled with spike pairing or altered presynaptic pyramidal spiking during theta epochs may trigger plasticity at the excitatory inputs to CA1 interneurons. Learning and the associated reorganization of the CA1 network may be a condition where such changes occur naturally.
Materials and methods
Reagent type (species) or resource | Designation | Source or reference | Identifiers | Additional information |
---|---|---|---|---|
Strain, strain background (Rattus norvegicus) | Long-Evans Rats | Janvier, France | RRID:RGD-631593 | |
Recombinant DNA reagent (Rattus norvegicus) | AAV2/1.CAMKII.ArchT.GFP.WPRE.SV40 | Penn Vector Core | RRID:Addgene: 26971-AAV1 | |
Transfected construct | AAV2/1.CaMKIIα::eNpHR3.0-YFP | Penn Vector Core | RRID:Addgene: 99039-AAV1 | |
Software, algorithm | Python | Python | RRID:SCR_008394 https://www.python.org | |
Software, algorithm | LFP Online | GtiHub | https://github.com/igridchyn/lfp_online | |
Other | 12 um tungsten wires | California Fine Wire | M294520 | |
Other | Headstage amplifier | Axona, St. Albans, UK | http://www.axona.com |
This study used previously published electrophysiological spike data (Schoenenberger et al., 2016). Accordingly, the experimental and spike clustering work has been described in this previous work in detail. Data from one additional rat recorded in the same paradigm and analyzed using the same methods was included in the data set.
Surgery for virus injection and microdrive implantation
Request a detailed protocolFour male adult rats (Long Evans, 300–500 g) were injected with a recombinant adeno-associated virus to express Halorhodopsin-YFP in the dorsal CA1 area (AAV2/1.CaMKIIα::eNpHR3.0-YFP Zhang et al., 2007, obtained from the Penn Vector Core facility, 1.6 × 1013 genome copies/mL; Addgene 26971) and the remaining animal was injected with a recombinant adeno-associated virus to express Archaerhodopsin (ArchT) in the dorsal CA1 area (AAV2/1.CaMKII::ArchT.GFP.WPRE.SV40 (Boyden et al., 2005), obtained from the Penn Vector Core facility, 6.41*1012 genome copies per mL). The virus was injected at four sites into dorsal CA1 of the right hemisphere in four rats and bilaterally in one rat: site 1: −3.0 AP, ±2.2 L, 2.1 DV; site 2: −3.7 AP, ±2.9 L, 2.0 DV; site 3: −4.3 AP, ±3.5 L, 2.0 DV; site 4: −5.0 AP, ±4.2 L, 2.2 DV. 3.5 weeks after virus injection, animals were implanted with 15 (28 in one rat) independently movable wire-tetrodes under deep anesthesia using isoflurane (0.5–2%), oxygen (1–2 L/min) and an initial dose of buprenorphine (0.1 mg/kg). Tetrodes were attached to the 15-tetrode (24-tetrode and 4-octrode in one rat) microdrive assemblies, enabling their independent movement. The tetrodes were constructed from four individual tungsten wires, 12 µm in diameter (H-Formvar insulation with Butyral bond coat, California Fine Wire, Grover Beach CA), twisted and then heated to bind them into a single bundle. The tips were then gold plated to reduce their impedance to 200–300 kΩ.
A 200 μm/0.48 NA optic fiber stub (Doric Lenses) located in the center of the tetrode array was used to apply laser light directly to the dorsal CA1 area. During surgery, a craniotomy was prepared above the dorsal hippocampus centered at AP = −4.0; ML = ± 3.0. Two stainless steel screws inserted through the skull above the cerebellum served as ground and reference electrodes, and six additional screws were used to permanently attach the microdrive assembly to the skull. Implantation was performed such as to position the tip of the optic fiber at a depth of 1.7 mm. The paraffin-wax coated electrodes and microdrives were then daubed with bone cement to encase the electrode-microdrive assembly and anchor it to the screws in the skull. Following a recovery period of 7 days, the tetrodes were lowered to their target locations over a further period of around 7 d. Tetrode locations were identified by electrophysiological markers such as theta band power, sharp wave polarity, and the presence of ripple oscillations, and by extrapolating the location of the electrodes by tracing the distances back along each electrode tract according to the daily advancement of the recorded electrodes. Implanted animals were housed individually in a separate room under a 12 hr light/12 hr dark cycle with ad libitum access to water, and they were maintained in a food-deprived state between 85–90% (plus an incremental 5 g per week) of their post-operative weight. Experiments were performed during the light phase. All rats used in this study were naïve and not used for additional procedures before surgery.
All procedures involving experimental animals were carried out in accordance with Austrian (Austrian Federal Law for experiments with live animals) animal law under a project license (BMBWF-66.018/0015-WF/V/3b/2014, BMBWF-66.018/0018-WF/V/3b/2019) approved by the Austrian Federal Science Ministry (BMWFW).
Data acquisition
Request a detailed protocol32-channel unity-gain preamplifier panels (Axona Ltd, St Albans, Hertfordshire, UK) were used to reduce cable movement artifacts. Wide-band (0.1/1 Hz – 5 kHz) recordings were taken, and the amplified local field potential and multiple-unit activity were continuously digitized at 24 kHz using a 128-channel data acquisition system (Axona Ltd, St Albans, Hertfordshire, UK). Two red LEDs mounted on the preamplifier headstage were used to track the location of the animal.
Green/yellow laser light for NpHR activation was provided by a 561 nm DPSS laser system equipped with an acousto-optic modulator (Omicron). The light was coupled into an optic fiber (four optic fibers in one rat) connected to a fiber-optic rotary joint (Doric lenses) from where a 200 μm/0.48 NA patch cord transmitted the light to the microdrive. Laser intensity was set to reach 25 mW total power at the tip of every implanted fiber stub. Data were recorded 6–7 weeks after AAV injection to ensure sufficient NpHR-YFP/ArchT GFP expression levels.
Behavioral paradigms
Request a detailed protocolData was recorded while the animals explored different arenas or rested in a sleep box. The sleep box was small (20 cm × 27 cm) with 60 cm high walls and cushioned with a terry towel for the animal to sleep/rest comfortably. During training and electrode positioning, the animals were familiarized with a 120 cm circular environment with 20 cm high walls (minimum of 60 min of exposure per day for at least seven days) that served as the familiar arena in all experiments. Curtains were used to enclose this arena and provide a stable set of external cues. In all exploration sessions, small food pellets were dropped at random from an automated overhead system (2–3/min) to motivate the animals to explore the entire arena. For recordings in a novel environment, several other arenas with different sizes, shapes, and textures were used. In addition, curtains were opened to provide novel distal room cues.
Typical recording days consisted of 10 sessions: four exploration sessions flanked by five sleep sessions and a final test session where brief laser pulses were applied while the animal still rested in the sleep box. Typically, sleep and exploration sessions lasted 25 min, whereas the laser test session lasted 18 min. The animals first explored the familiar arena. After visiting a different novel arena, the familiar arena was explored again, but laser illumination was automatically triggered when the animal entered a specific part of the arena (light zone). Finally, the same arena was explored again. All exploration sessions were flanked by sleep. The light zone was defined by a center position and an angle between 120° and 180° such that it covered one-third to half of the arena. The initial angle defining the illumination zone was random and thus random also with respect to the hippocampal place fields. Every day, a novel illumination zone that had about 50% overlap with the previous day’s zone was defined. During the course of the project and also within individual animals, the angle defining the size of the illumination zone was increased to include more place fields in the light zone. After completion of the experiments, the rats were deeply anesthetized and perfused through the heart first with PBS followed by a 4% buffered formalin phosphate solution for the histological verification of electrode tracks and optic fiber position. Furthermore, NpHR-YFP/ArchT-GFP expression in dorsal CA1 was verified in each animal by checking the fluorescence of the YFP/GFP tag.
Spike sorting and unit classification
Request a detailed protocolUnit isolation and clustering procedures have been described before (Csicsvari et al., 1998). Briefly, after resampling of the raw data to 20 kHz, action potentials were extracted from the digitally high-pass filtered (0.8–5 kHz) signal. The power computed in a sliding window (12.8 ms) and action potentials with a power of >5 SD from the baseline mean were selected. The spike features were then extracted using principal components analyses. The detected action potentials were then segregated into putative multiple single units using an automatic clustering software (Harris et al., 2000) (http://klustakwik.sourceforge.net/). Finally, the generated clusters were manually refined by a graphical cluster cutting program. Only units with clear refractory periods in their autocorrelation and well-defined cluster boundaries were used for further analysis. Periods of waking spatial exploration, immobility, and sleep were clustered together. The stability of the cells was verified by plotting spike features over time. In addition, an isolation distance (based on Mahalanobis distance, [Harris et al., 2000]) was calculated to ensure the spike clusters did not overlap during the recordings. CA1 pyramidal cells and interneurons were discriminated by their autocorrelations, firing rate, and waveforms (Csicsvari et al., 1999; Henze et al., 2000). In total, we recorded 1842 pyramidal cells and 91 interneurons.
Pyramidal cell-interneuron coupling
Request a detailed protocolIsolation of monosynaptically-connected pyramidal cell-interneuron pairs was performed as described previously by identifying cross-correlograms between pyramidal cells and interneurons that exhibited a large, sharp peak in the 0.5–2.5 ms bins (after the discharge of the reference pyramidal cells) (Csicsvari et al., 1998). Because the number of action potentials used for the construction of these cross-correlograms varied from cell to cell, the histograms were normalized by dividing each bin by the number of reference pyramidal spike events (Csicsvari et al., 1998). The connection strength was thus accessed by measuring the spike transmission probability at the monosynaptic peak indicating the probability that the pyramidal cell would discharge its postsynaptic interneuron partner. However, the chance probability of the two cells firing together was subtracted in order to account for firing rate change-related fluctuations in the correlation strength. The chance firing probability was estimated by averaging the 10–50 ms bins on both sides of the histogram. The significance level for the monosynaptic peak was set at three standard deviations from the baseline (p<0.000001) (Abeles, 1982; Csicsvari et al., 1998). In addition, to filter out sparse histograms, we only considered monosynaptic pairs in which either FAM1 or FAM2 contained at least 1000 spike coincidence counts with the −50 to 50 ms intervals, and the SD of the bin values were less than one-third of the mean bin value.
Comparison of firing rate and firing field analysis
Request a detailed protocolTo compare firing rates between two sessions, we calculated the relative firing rate change by dividing the signed difference between the mean firing rates by the sum of the mean rates (i.e. c=(r2-r1)/(r2+r1)), where r1 and r2 denote the mean firing rates in the two sessions that are compared (Leutgeb et al., 2004; Leutgeb et al., 2005). This score is always between −1 and 1, and the extreme values −1 and 1 mean that a neuron is firing exclusively in one of the two sessions. A similar measure was used to measure the relative change of spike transmission and spike pairing events across different sessions.
Statistical analyses
Request a detailed protocolWe used linear mixed models and ANOVA to determine the significance of variables in predicting spike transmission probabilities and their changes. We used a mixed model comparison to test whether predictions were independent of other variables and displayed corresponding partial correlations in the figures. We added the animal variable as a random effect to all linear mixed models to account for variability across animals and used the comparison of linear mixed model and linear model without an animal variable to test whether an animal variable contributed significantly to the prediction of spike probability and their changes. We used Holm-Bonferroni multiple testing correction to account for comparisons in multiple time windows in the analysis with prediction through pairing.
Data availability
Original data and programs are available in the scientific repository of the Institute of Science and Technology Austria (https://doi.org/10.15479/AT:ISTA:8563).
-
IST AustriaOptogenetic Alteration of Hippocampal Network Activity.https://doi.org/10.15479/AT:ISTA:8563
References
-
Role of the cortical neuron: integrator or coincidence detector?Israel Journal of Medical Sciences 18:83–92.
-
Millisecond-timescale, genetically targeted optical control of neural activityNature Neuroscience 8:1263–1268.https://doi.org/10.1038/nn1525
-
Direct afferent excitation and long-term potentiation of hippocampal interneuronsJournal of Neurophysiology 48:597–607.https://doi.org/10.1152/jn.1982.48.3.597
-
Oscillatory coupling of hippocampal pyramidal cells and interneurons in the behaving ratThe Journal of Neuroscience 19:274–287.https://doi.org/10.1523/JNEUROSCI.19-01-00274.1999
-
Behavior-dependent short-term assembly dynamics in the medial prefrontal cortexNature Neuroscience 11:823–833.https://doi.org/10.1038/nn.2134
-
Intracellular features predicted by extracellular recordings in the Hippocampus in vivoJournal of Neurophysiology 84:390–400.https://doi.org/10.1152/jn.2000.84.1.390
-
Long-term synaptic plasticity in hippocampal interneuronsNature Reviews Neuroscience 8:687–699.https://doi.org/10.1038/nrn2207
-
Long-term plasticity in identified hippocampal GABAergic interneurons in the CA1 area in vivoBrain Structure & Function 222:1809–1827.https://doi.org/10.1007/s00429-016-1309-7
-
Input-specific learning rules at excitatory synapses onto hippocampal parvalbumin-expressing interneuronsThe Journal of Physiology 591:1809–1822.https://doi.org/10.1113/jphysiol.2012.245852
-
Presynaptic plasticity: targeted control of inhibitory networksCurrent Opinion in Neurobiology 19:254–262.https://doi.org/10.1016/j.conb.2009.05.008
-
Differential modulation of CA1 and dentate gyrus interneurons during exploration of novel environmentsJournal of Neurophysiology 91:863–872.https://doi.org/10.1152/jn.00614.2003
-
Hippocampal GABAergic inhibitory interneuronsPhysiological Reviews 97:1619–1747.https://doi.org/10.1152/physrev.00007.2017
-
Activity-dependent plasticity of hippocampal place mapsNature Communications 7:11824.https://doi.org/10.1038/ncomms11824
Decision letter
-
Laura L ColginSenior and Reviewing Editor; University of Texas at Austin, United States
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
Acceptance summary:
This study investigates into changes in pyramidal cell and interneuron spike transmission probability in pyramidal cell-interneuron monosynaptic pairs in hippocampal area CA1 following optogenetic inhibition and disinhibition of some of these cells in behaving rats exploring novel and familiar environments. Insights regarding how cell-cell correlations change with experience in freely behaving animals add to our understanding of how hippocampal circuits support learning and memory. The work also changes the current thinking about processes of plasticity in hippocampal networks and has relevance for interpretation of datasets in which optical inhibition is applied.
Decision letter after peer review:
Thank you for sending your article entitled "Optogenetic inhibition-mediated activity-dependent modification of CA1 pyramidal-interneuron connections during behavior" for peer review at eLife. Your article is being evaluated by three peer reviewers, and the evaluation is being overseen by a Reviewing Editor and Michael Frank as the Senior Editor.
The reviewers were all generally positive about the manuscript but also felt that it has a number of major issues in its current form. The lack of a YFP only control group was a major concern. Without an appropriate control group, the authors are unable to definitively state that their effects are due to optogenetic manipulations (e.g., such changes may just occur across the normal passage of time). However, the reviewers agreed in the consultation and discussion that they would not require the authors to include this optimal control group, assuming that the authors are able to demonstrate in some other way that the reported effects are actually due to light stimulation. One suggestion was offered by reviewer 2 (point 4). Another suggestion that came up in the Discussion was to compare the first and second halves of light-free sessions (i.e., FAM1 and FAM2). Another major concern was the lack of cell-type specificity of the CaMKII promoter (see reviewer 1's point 1 and reviewer 3's major point 2). Another essential revision is to demonstrate that the optogenetic manipulations did not cause behavioral changes that could potentially explain the differences in neuronal activity and co-activity patterns. Issues were also raised regarding statistical analyses (see reviewer 1's point 4 and other statistics-related points from reviewer 3). Lastly, the authors have previously shown the effects of these optogenetic manipulations on place cell responses. However, this was not entirely clear, and the authors should not assume that all readers have read their earlier paper (see points about place cells from reviewer 2 below). The authors should address the points raised by reviewer 2 by explicitly stating the effects that they have already shown and citing their earlier work. In addition, the authors should also discuss how the current plasticity results may be involved in the place cell changes reported in their earlier paper. The reviews are printed in their entirety below.
Reviewer #1:
The manuscript by Schoenenberger et al. all investigates into changes in pyramidal cell interneuron spike transmission probability in pyramidal cell-interneuron monosynaptic pairs in the rat hippocampal area CA1 following direct optogenetic inhibition and indirect disinhibition of some of these cells. The manuscript is a clever follow-up and re-analysis of the dataset collected for a paper that has already been published (Schoenenberger et al., 2016). The current analysis is focused at lasting changes of CA1 pyramidal cell-interneuron monosynaptic pairs following optogenetic inhibition. The authors report that these lasting changes are primarily predicted by changes in interneuron excitability during optogenetic modifications of firing rate and synchrony of hippocampal CA1 neurons.
This is an interesting study from one of the leading labs in the field and the reported changes in spike transmission probability between pyramidal cells and interneurons following optogenetic circuit manipulation are of potential importance for our understanding of activity-dependent reorganization of hippocampal circuits during learning.
There are some questions related to the analysis and to the interpretation of the data that should be addressed by the authors at this stage.
1) Similar to their previous paper (Schoenenberger et al., 2016), the authors' efforts here are somewhat also stymied by the apparent lack of cell-type specificity of the CaMKII viral promoter. This limitation should be explicitly stated and discussed in the current manuscript, especially since there are still conflicting results out there suggesting that these promoters could indeed be reliably used for cell-type specific targeting of principal cells, which is clearly not the case.
2) Relatedly, it is somewhat unfortunate that the optogenetic manipulation approach and the resulting dataset did not really allow the authors to directly test their main finding in the manuscript – that changes in interneuron excitably primarily dictates plasticity in their afferent inputs. This would have ideally required to selectively excite and/or inhibit interneurons in CA1. Of note, specific rAAV promoters actually do exist for selective manipulations of GABAergic interneurons (i.e., Dimidschstein et al., 2016).
3) The authors should analyze and report if the optogenetic manipulation has caused any acute (during FAML) or chronic (during FAM2) effects on the animals' behavior and to what extent these behavioral changes may contribute to the observed changes in pyramidal cell-interneuron spike transmission probability. On another note, the proper control for optogenetic experiments should have been the use of rAAV with a static fluorophore (i.e., YFP alone).
4) Statistical analysis should also be performed and reported on animal as a unit in order to better account for inter-animal variance.
Reviewer #2:
Schoenenberger et al. examine spike timing dynamics of pyramidal cell interneuron pairs in familiar and novel environments and following perturbation of pyramidal cell sub-populations in a familiar environment. The main findings address spike transmission probabilities for pyramidal cell interneuron pairs that are likely to have synaptic connections (based on short-latency cross-correlation peaks). Optogenetic inhibition of sub-populations of pyramidal neurons decreases activity in some neurons (as expected) and increases activity in others. Interneurons are largely inhibited in their activity during inhibition of pyramidal cells. Presumably, this is a result of reduced pyramidal cell excitation of feedback interneurons. Alterations in firing appear to produce longer-lasting changes in spike transmission probabilities. This suggests that spike timing relationships between pyramidal cells and interneurons are significant features of network dynamics that impact representation in hippocampus. In general, the findings are solidly backed up by the analytical approach and statistics and are clearly presented. There are some aspects of the data that seem absent from the report that might well make well place the results in a larger context and might make for greater impact.
1) The authors show that altered spike transmission probabilities in a novel environment are observed, but that these alterations do not impact dynamics in subsequent visits to the familiar environment. The authors do not reach far enough to attempt to explain this. Alterations induced by inhibition in the familiar environment do persist. These differences are interpreted as reflecting learning in the form of changes in pyramidal cell-interneuron synapses. This seems inconsistent with the lack of effect of novel environment dynamics. What might constitute an explanation for this? Perhaps there is some interaction with other inputs to interneurons that is critical? Furthermore, it would be of interest to determine whether it is the novelty of the environment that precludes persistence of spike transmission probabilities or simply the fact that the animal is in a different environment with a largely different set of pyramidal cell ensemble activity patterns.
2) The optogenetic inhibition was applied only in certain regions of the environment, yet the authors make no use of this design feature. Are the observed effects limited to pyramidal cell-interneuron pairs for which the pyramidal cell has a place field in the region of the environment where inhibition was applied? In general, a major limitation of the work is that it does not consider the effects of alterations in pyramidal cell-interneuron spike transmission on representation of place. Do the observed changes yield rate-remapping, global remapping, partial remapping, etc.? In the absence of such analyses, it is unclear whether one should consider the observed changes in interaction to be impactful on network function or not.
3) When considering the effects of co-activity, the authors should expand beyond the somewhat arbitrary time window of 20ms. It would be more informative to test a range of intervals and determine at what point paired spikes have no impact on subsequent transmission.
4) To place the results in context, the authors might include an analysis of odd versus even minutes of one or all of the inhibition-free sessions. This will provide somewhat of a baseline for spike transmission changes.
5) Do pyramidal cells released with increased firing in response to optogenetic inhibition of other pyramidal cells exhibit place-specific firing?
Reviewer #3:
In this manuscript, the authors intended to demonstrate that plastic changes occurred at the hippocampal pyramidal to interneuron connections in response to optogenetic manipulation during a spatial task in rats. To do so, the authors applied light stimulations as rats explored a familiar open arena (FAML session), after and before the rats explored the same arena without light in two sessions (FAM1, FAM2). The authors then compared spike transmission probability between identified putative pyramidal-interneuron pairs, as a measure of their connections, among these sessions. The authors made two key conclusions: (i) Optogenetic light stimulations led to lasting plasticity in spike transmission between connected pyramidal neurons and interneurons. (ii) The plastic changes were caused by firing rate changes in the postsynaptic interneurons. What and where synaptic plasticity occurs during behavior and how it leads to a particular learning behavior in vivo are important questions. However, identifying and manipulating synaptically connected neuron pairs is difficult in behaving animals. I applaud the authors' effort in this study and its outcome will be valuable to our understanding of learning and memory. However, as much as I like the study, I also have major concerns, which need to be resolved to make sure the key conclusions in the manuscript are valid.
1) Regarding the first conclusion, a major concern is that it is difficult to tell whether the observed changes had anything to do with the light stimulations. Judging from Figure 1C, D and Figure 2E, the changes in spike transmission and firing rate were small. One possibility is that these changes could arise just passively with time or other unrelated experience. What is lacking is a control experiment that includes the same recording procedure, but without the light stimulation or even better, with a control light stimulation session when neurons are designed not to respond (like a different color of light). I understand that this takes a lot of effort. However, at least in the existing data, the authors should analyze how spike transmissions within FAM1 or FAM2 fluctuate and how the fluctuation level was compared to the changes between FAM1 and FAM2.
2) I am confused by the authors' interpretation of the direct effect of light on firing rates of interneurons. The authors used halorhodopsin to inhibit neurons under the control of CaMKIIα. First, isn't it true that the promotor would restrict the halorhodopsin expression to pyramidal neurons, not much in interneurons? If this is not true, the authors need to provide references or histological evidence for this. Second, even if this is not true, how can the authors make sure the inhibition is caused directly by light, but not by the inhibition of pyramidal neurons? The evidence for a direct inhibition of interneurons (Figure 1B, for a light in sleep session) seems not strong, because the light was on for a long time. Third, some interneurons even increased their firing rates during FAML, which cannot be a direct effect of light. The picture of the light's direct effect is not clear in my mind, given a potential mixture of potential direct inhibition from light, if true, and the indirect inhibition from pyramidal neurons. The authors should clearly describe and discuss this issue, since this is important to their other key conclusion (see below).
3) The authors concluded that the plasticity in spike transmission was mainly caused by the light-induced changes in the firing rates of postsynaptic interneurons. First, how can the authors be sure that this is not the other way around? That is, plasticity could occur first, then led to reduced firing rates in interneurons? Second, the authors need to clearly state the direction of spike transmission plasticity caused by the interneuron rate change: lower/higher firing rates of interneuron lead to lower/higher spike transmissions. The authors touched the issue in the Discussion but seemed vague about this. One concern is that the enhanced firing rates in some interneurons were clearly not a direct effect of light stimulations, but they were important to the correlations the authors used to make the conclusion. I believe the authors should be straightforward about this and modify the conclusion accordingly.
4) One key result is Figure 3A. Here the FAML/FAM2 partial correlation was only reduced slightly from FAML alone, even though FAM2 alone was quite high. Can the authors verify and explain this?
5) The authors described the number of animals, the number of cells, and the number of pairs analyzed. It is unclear how many days were recorded, how many cells or pairs were obtained from each animal per day, whether at least the key results can be seen in multiple animals, and whether the same pairs were repeatedly used in the analysis.
6) What are the N's in Figure 2B-D, Figure 3A-B, Figure 4? What does each sample mean in these plots? If each sample was a day and there were 63 pairs in a number of days/animals, were there sufficient number of pairs for computing correlation or regression on a given day? What is the prediction in the labels? There is no description about this key analysis in the method.
7) What is the y-axis in Figure 1B?
[Editors' note: further revisions were suggested prior to acceptance, as described below.]
Thank you for submitting your article "Optogenetic inhibition-mediated activity-dependent modification of CA1 pyramidal-interneuron connections during behavior" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Laura Colgin as the Senior Editor and Reviewing Editor. The reviewers have opted to remain anonymous.
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions would benefit from additional new data or new experiments, as reviewers feel is the case with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions would benefit from additional supporting data.
In the latter case, our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.
Summary:
This study investigates into changes in pyramidal cell interneuron spike transmission probability in pyramidal cell-interneuron monosynaptic pairs in hippocampal area CA1 following optogenetic inhibition and disinhibition of some of these cells in behaving rats exploring novel and familiar environments. Insights regarding how cell-cell correlations change with experience in freely behaving animals add to our understanding of how hippocampal circuits support learning and memory. The work also changes the current thinking about processes of plasticity in hippocampal networks and has relevance for interpretation of datasets in which optical inhibition is applied. However, the paper does include technical confounds, as noted below.
Revisions for this paper:
Figure 1C, D: the description of these two figure panels was not quantitative in both the main text and the figure legend. The authors stated that the plots "showed changes" and "altered the ability…", although the data points were highly correlated. It is better to quantify whether the changes were more than those in a control condition (like odd vs even minutes or two halves within FAM1). The authors' statements could be supported by the lower FAML-FAM1 correlations than those of the control condition. Also, it may be worth quantifying the overall changes (mean or median differences) in rate and transmission probability to understand whether a significant net effect of the stimulation occurred in this experiment.
Revisions expected in follow-up work:
1) Ideally, the proper controls for optogenetic experiments (i.e., YFP-only control) should have been included.
2) Follow-up experiments will include specific rAAV promoters for selective manipulations of GABAergic interneurons.
https://doi.org/10.7554/eLife.61106.sa1Author response
Reviewer #1:
[…]
There are some questions related to the analysis and to the interpretation of the data that should be addressed by the authors at this stage.
1) Similar to their previous paper (Schoenenberger et al., 2016), the authors' efforts here are somewhat also stymied by the apparent lack of cell-type specificity of the CaMKII viral promoter. This limitation should be explicitly stated and discussed in the current manuscript, especially since there are still conflicting results out there suggesting that these promoters could indeed be reliably used for cell-type specific targeting of principal cells, which is clearly not the case.
Now we discuss the lack of cell specificity in the manuscript. Cell specificity of viruses using CaMKII promoter may depend on the region where the virus is expressed, and type of virus used. An earlier study has observed similar effects as well, e.g., Nathanson et al., 2009. In the Schoenenberger et al., 2016 paper we presented immunolabeling results showing that, in addition to pyramidal cells, our virus indeed expressed in both parvalbumin and somatostatin immunopositive cells and the light response delay of 1-2ms was similar for both the inhibited pyramidal cells and interneurons.
2) Relatedly, it is somewhat unfortunate that the optogenetic manipulation approach and the resulting dataset did not really allow the authors to directly test their main finding in the manuscript – that changes in interneuron excitably primarily dictates plasticity in their afferent inputs. This would have ideally required to selectively excite and/or inhibit interneurons in CA1. Of note, specific rAAV promoters actually do exist for selective manipulations of GABAergic interneurons (i.e., Dimidschstein et al., 2016).
Yes, unfortunately, in this manuscript, we were not able to selectively manipulate interneuron activity using new tools that have recently emerged. This finding was unexpected and selectively manipulating interneurons would require a collection of a dataset in size and effort similar to the one we use in this study. Note also that even in cases in which interneuron firing is specifically altered using optogenetics, such manipulations will indirectly influence the firing of pyramidal cells. For example, suppressing interneuron activity will lead to the disinhibition of pyramidal cells. We see such disinhibitory effect even when we selectively inhibit a smaller subset of CCK interneurons in transgenic mice.
3) The authors should analyze and report if the optogenetic manipulation has caused any acute (during FAML) or chronic (during FAM2) effects on the animals' behavior and to what extent these behavioral changes may contribute to the observed changes in pyramidal cell-interneuron spike transmission probability. On another note, the proper control for optogenetic experiments should have been the use of rAAV with a static fluorophore (i.e., YFP alone).
In the Schoenenberger et al., 2016 paper, we compared the speed across different sessions of the animal and possible occupancy differences in the light zone and outside. Now, we performed the same analysis in our extended data as well (Figure 1—figure supplement 1). No significant differences were seen either in occupancy or speed, inside and outside the light zone across FAM1, FAML and FAM2.
In relation to the control experiments, as suggested by the Editor’s letter, we assessed the within-session variability of spike transmission probabilities in FAM1 and showed that these are smaller than those seen across FAM1-FAM2 (Figure 2B).
4) Statistical analysis should also be performed and reported on animal as a unit in order to better account for inter-animal variance.
As suggested, in all analyses, we incorporated animal as a random effect in our linear mixed models and ANOVA and show that this did not influence our results.
Reviewer #2:
[…]
1) The authors show that altered spike transmission probabilities in a novel environment are observed, but that these alterations do not impact dynamics in subsequent visits to the familiar environment. The authors do not reach far enough to attempt to explain this. Alterations induced by inhibition in the familiar environment do persist. These differences are interpreted as reflecting learning in the form of changes in pyramidal cell-interneuron synapses. This seems inconsistent with the lack of effect of novel environment dynamics. What might constitute an explanation for this? Perhaps there is some interaction with other inputs to interneurons that is critical? Furthermore, it would be of interest to determine whether it is the novelty of the environment that precludes persistence of spike transmission probabilities or simply the fact that the animal is in a different environment with a largely different set of pyramidal cell ensemble activity patterns.
In the revision, we discuss further the possible mechanism behind the altered spike transmission in the novel environment. We show in Figure 4A that spike transmission changes from familiar to novel environments were predicted by the change of the interneuron rates but not by the pyramidal rates. This suggests a postsynaptic effect, which, at least in part, may be related to the change of the depolarization/excitability of interneurons. We agree with the reviewer that other inputs to the interneurons may be a cause for the spike transmission changes from the familiar to the novel environment. Such inputs may include other place cell or non-specific neuromodulation, e.g., acetylcholine levels are higher in a novel environment. Unfortunately, we do not have data in which two different familiar environments and a novel environment are all simultaneously recorded, which would be required if pyramidal-interneuron spike transmissions reorganize across familiar environments.
2) The optogenetic inhibition was applied only in certain regions of the environment, yet the authors make no use of this design feature. Are the observed effects limited to pyramidal cell-interneuron pairs for which the pyramidal cell has a place field in the region of the environment where inhibition was applied? In general, a major limitation of the work is that it does not consider the effects of alterations in pyramidal cell-interneuron spike transmission on representation of place. Do the observed changes yield rate-remapping, global remapping, partial remapping, etc.? In the absence of such analyses, it is unclear whether one should consider the observed changes in interaction to be impactful on network function or not.
In the previous Schoenenberger et al., 2016 paper, we showed that a fraction of place cells that were inhibited by the light remapped their place fields as a result of light-mediated inhibition. In a new analysis, we attempted to relate place field remapping to the changes of monosynaptic connections, but we did not see a relationship.
We report this in the revision by saying that “In addition to the changing the firing rate, light application can cause remapping in a subpopulation of cells (Schoenenberger et al., 2016). However, a change in spike transmission in FAM2 did not predict the degree of pyramidal place field remapping (P=0.1612, F-test)”.
We may not have been able to see such a relationship because only a subpopulation of light-inhibited cells exhibited place field remapping.
3) When considering the effects of co-activity, the authors should expand beyond the somewhat arbitrary time window of 20ms. It would be more informative to test a range of intervals and determine at what point paired spikes have no impact on subsequent transmission.
We checked the effect in additional time windows of 10ms, 20ms, 50ms and 100ms. When we included animal and used different intervals (leading to multiple comparisons) as additional factors, only the 50ms time window was significant. Moreover, spike paring no longer predicted the spike transmission changes when pyramidal and interneuron rate changes together were accounted for. This is possibly due to the fact that spike pairing alone during FAML is very strongly (R2=0.849) predicted by the combined rate changes (Figure 6B). We comment on these results in the Discussion now:
“However, such a pairing relationship was observed only for 50 ms time window. […] Nevertheless, we cannot exclude the possibility that the independent rate alterations of pyramidal cells and interneurons in FAML governed spike transmission probability changes, without spike pairing itself directly influencing it.”
4) To place the results in context, the authors might include an analysis of odd versus even minutes of one or all of the inhibition-free sessions. This will provide somewhat of a baseline for spike transmission changes.
We thank the reviewer for this suggestion, and we perform such an analysis as well to show stability. Indeed, in the difference in the spike transmission measured in alternating time windows within FAM1 was significantly less than across FAM1-FAM2 (Figure 2B).
5) Do pyramidal cells released with increased firing in response to optogenetic inhibition of other pyramidal cells exhibit place-specific firing?
Yes, in the Schoenenberger et al., 2016 paper, we showed that both disinhibited and inhibited pyramidal cells exhibited place-related firing. Interestingly, disinhibition did not alter the place fields, whereas inhibition triggered place field remapping.
Reviewer #3:
[…]
1) Regarding the first conclusion, a major concern is that it is difficult to tell whether the observed changes had anything to do with the light stimulations. Judging from Figure 1C, D and Figure 2E, the changes in spike transmission and firing rate were small. One possibility is that these changes could arise just passively with time or other unrelated experience. What is lacking is a control experiment that includes the same recording procedure, but without the light stimulation or even better, with a control light stimulation session when neurons are designed not to respond (like a different color of light). I understand that this takes a lot of effort. However, at least in the existing data, the authors should analyze how spike transmissions within FAM1 or FAM2 fluctuate and how the fluctuation level was compared to the changes between FAM1 and FAM2.
In the revision, as suggested by reviewer 2 as well, we compare spike transmissions within FAM1 using alternating time windows and compare it to the changes occurring across FAM1-FAM2 sessions. The within-session variability of spike transmission was significantly less than those across the FAM1-FAM2 sessions (Figure 2B). Note, however, that some additional findings of the manuscript also argue against the possibility that our effects are simply due to random changes that may occur over time passed. First, we showed that changes in the light session but not those in the novel session predict the changes across FAM1-to-FAM2 sessions. Second, we show that other factors such interneuron rate changes will also predict spike transmission changes. We do not see how random fluctuations of spike transmission could still lead to predictions that involve factors of the light session but not the other “control” intervening session of the novel environment.
2) I am confused by the authors' interpretation of the direct effect of light on firing rates of interneurons. The authors used halorhodopsin to inhibit neurons under the control of CaMKIIα. First, isn't it true that the promotor would restrict the halorhodopsin expression to pyramidal neurons, not much in interneurons? If this is not true, the authors need to provide references or histological evidence for this. Second, even if this is not true, how can the authors make sure the inhibition is caused directly by light, but not by the inhibition of pyramidal neurons? The evidence for a direct inhibition of interneurons (Figure 1B, for a light in sleep session) seems not strong, because the light was on for a long time. Third, some interneurons even increased their firing rates during FAML, which cannot be a direct effect of light. The picture of the light's direct effect is not clear in my mind, given a potential mixture of potential direct inhibition from light, if true, and the indirect inhibition from pyramidal neurons. The authors should clearly describe and discuss this issue, since this is important to their other key conclusion (see below).
We further discussed these issues in the revised manuscript. In the Schoenenberger et al., 2016 paper, we quantified these effects. We showed that both the light-inhibited pyramidal cells and interneurons suppressed their firing within a short 1-2ms time delay relative to the light onset. The light responses were tested using 500ms light pulses in the end of the recordings while the animal was rested. We also performed immunolabeling in the Schoenenberger et al., 2016 study and showed that both parvalbumin and somatostatin immunopositive cells expressed halorhodopsin. Note also that an earlier paper reported a similar effect Nathanson et al., 2009. Of course, we cannot exclude that this virus is more specific in other brain regions or other virus serotypes or constructs may be more specific.
3) The authors concluded that the plasticity in spike transmission was mainly caused by the light-induced changes in the firing rates of postsynaptic interneurons. First, how can the authors be sure that this is not the other way around? That is, plasticity could occur first, then led to reduced firing rates in interneurons? Second, the authors need to clearly state the direction of spike transmission plasticity caused by the interneuron rate change: lower/higher firing rates of interneuron lead to lower/higher spike transmissions. The authors touched the issue in the Discussion but seemed vague about this. One concern is that the enhanced firing rates in some interneurons were clearly not a direct effect of light stimulations, but they were important to the correlations the authors used to make the conclusion. I believe the authors should be straightforward about this and modify the conclusion accordingly.
Indeed, Figure 3C shows that the interneuron rate increase in FAML is associated with a stronger spike transmission, whereas in cases of reduced spike transmission, FAM2 interneuron rate is weaker. As asked, in the revision, we spelled out this relationship in the Discussion. We agree that the rate increase of some interneurons may be caused by strengthened pyramidal connections triggered by complex network effects mediated by light-induced inhibition of a subgroup of pyramidal cells and interneurons. We discussed this scenario in revision, as suggested by the reviewer.
4) One key result is Figure 3A. Here the FAML/FAM2 partial correlation was only reduced slightly from FAML alone, even though FAM2 alone was quite high. Can the authors verify and explain this?
This finding showed that, although FAM2-FAM1 interneuron rate changes predicted the corresponding spike transmission changes, similar rate changes between FAML-FAM1 have further predictive value. The relationship between FAM2-FAM1 rate and spike transmission changes may solely suggest that excitability changes of the interneuron led to the spike transmission changes. However, with the partial correlation (and the associated ANOVA model comparisons), we were able to show that rate (i.e., excitability) alterations in the light session were able to independently influence the FAM1-FAM2 spike transmission changes. That is that excitability/depolarization alterations in FAML that were no longer present in FAM2 still influenced FAM1-FAM2 spike transmission changes. We explain this better in the revision.
5) The authors described the number of animals, the number of cells, and the number of pairs analyzed. It is unclear how many days were recorded, how many cells or pairs were obtained from each animal per day, whether at least the key results can be seen in multiple animals, and whether the same pairs were repeatedly used in the analysis.
In the original work in four animals, we recorded from n=31, 23, 5, 4 detected cell pairs. In the revision, we added one additional animal with n=16 cell pairs to ensure that all the results could be replicated independently of the critical contribution of a single animal. In all analyses, we included animal as a random effect and showed that this did not influence our results. In three animals we recorded in two recording days, while four recording days were used in the remaining two animals. The electrodes were moved between recording days to ensure that a different set of cells are recorded across days. Therefore, we think that only very few cell pairs may have recorded across the two recordings days. Yet, even if we did record from the same cell pairs across days, the light zone location was changed daily; therefore, the cells experienced an entirely different network activity background in our manipulations with different potential outcomes.
6) What are the N's in Figure 2B-D, Figure 3A-B, Figure 4? What does each sample mean in these plots? If each sample was a day and there were 63 pairs in a number of days/animals, were there sufficient number of pairs for computing correlation or regression on a given day? What is the prediction in the labels? There is no description about this key analysis in the method.
In each plot, each dot represents the value related to a single cell pair so n=78 (extended dataset) always. With a few sessions, we were able to record >10 pairs yet the yield of detecting these monosynaptic pairs was relatively low. Prediction refers to the linear regression and the correlation coefficient (r). We added correlation to the legend. We thank the reviewer for pointing out the confusion. However, we kept prediction on the axes because prediction is easier to understand (according to our opinion) than using “correlation with” on these labels.
7) What is the y-axis in Figure 1B?
It shows the probability of a spike occurring within the 20ms time bins. Now we also explain it in the legend.
[Editors' note: further revisions were suggested prior to acceptance, as described below.]
Revisions for this paper:
Figure 1C, D: the description of these two figure panels was not quantitative in both the main text and the figure legend. The authors stated that the plots "showed changes" and "altered the ability…", although the data points were highly correlated. It is better to quantify whether the changes were more than those in a control condition (like odd vs even minutes or two halves within FAM1). The authors' statements could be supported by the lower FAML-FAM1 correlations than those of the control condition. Also, it may be worth quantifying the overall changes (mean or median differences) in rate and transmission probability to understand whether a significant net effect of the stimulation occurred in this experiment.
We performed the requested quantification and indeed the within session (FAM1) correlations we higher that across session (FAM1-FAML) ones, both for firing rate and spike transmission. Median rates were not significantly different however because cells either reduced or increased their rate during the light application. But there was a significant reduction in the spike transmission probabilities from FAM1 to FAML.
Revisions expected in follow-up work:
1) Ideally, the proper controls for optogenetic experiments (i.e., YFP-only control) should have been included.
We acknowledge this in the beginning of the Discussion saying that “Or study did not use control animals in which only YFP was expressed. Therefore, we cannot exclude the possibility that optogenetic channel expression, or, perhaps, light application enhanced the plasticity on pyramidal-interneuron synapses. Yet, we observed similar activity-dependent changes during spatial learning before (Dupret et al., 2013). So, it is likely that the optogenetic, light-mediated rate alteration was a primary driver of the activity-dependent, lasting connection-weight changes. “
2) Follow-up experiments will include specific rAAV promoters for selective manipulations of GABAergic interneurons.
We state these follow-up experiments in the Discussion but we also point out that the disinhibition of pyramidal cells might hinder these experiments: “Future work in which interneuron (or a certain genetic type) firing rate is selectively altered by optogenetics may provide further evidence for the independent contribution of postsynaptic interneuron depolarization in plasticity. Nevertheless, even in such manipulations, indirect alteration of pyramidal rates (e.g., because of disinhibition) is expected to occur.”
https://doi.org/10.7554/eLife.61106.sa2Article and author information
Author details
Funding
Austrian Science Fund (I02072)
- Jozsef Csicsvari
Swiss National Science Foundation
- Philipp Schoenenberger
Austrian Science Fund (I03713)
- Jozsef Csicsvari
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Acknowledgements
We thank Michele Nardin and Federico Stella for comments on an earlier version of the manuscript. K Deisseroth for providing the pAAV-CaMKIIα::eNpHR3.0-YFP plasmid through Addgene. E Boyden for providing AAV2/1.CaMKII::ArchT.GFP.WPRE.SV40 plasmid through Penn Vector Core. This work was supported by the Austrian Science Fund (I02072 and I03713) and a Swiss National Science Foundation grant to PS. The authors declare no conflicts of interest.
Ethics
Animal experimentation: All procedures involving experimental animals were carried out in accordance with Austrian (Austrian federal Law for experiments with live animals) animal law under a project license (BMBWF-66.018/0015-WF/V/3b/2014, BMBWF-66.018/0018-WF/V/3b/2019) approved by the Austrian Federal Science Ministry (BMWFW).
Senior and Reviewing Editor
- Laura L Colgin, University of Texas at Austin, United States
Version history
- Received: July 15, 2020
- Accepted: October 3, 2020
- Accepted Manuscript published: October 5, 2020 (version 1)
- Version of Record published: October 20, 2020 (version 2)
Copyright
© 2020, Gridchyn et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
-
- 1,838
- Page views
-
- 258
- Downloads
-
- 3
- Citations
Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.
Download links
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Further reading
-
- Chromosomes and Gene Expression
- Neuroscience
Mathys et al. conducted the first single-nucleus RNA-seq (snRNA-seq) study of Alzheimer’s disease (AD) (Mathys et al., 2019). With bulk RNA-seq, changes in gene expression across cell types can be lost, potentially masking the differentially expressed genes (DEGs) across different cell types. Through the use of single-cell techniques, the authors benefitted from increased resolution with the potential to uncover cell type-specific DEGs in AD for the first time. However, there were limitations in both their data processing and quality control and their differential expression analysis. Here, we correct these issues and use best-practice approaches to snRNA-seq differential expression, resulting in 549 times fewer DEGs at a false discovery rate of 0.05. Thus, this study highlights the impact of quality control and differential analysis methods on the discovery of disease-associated genes and aims to refocus the AD research field away from spuriously identified genes.
-
- Neuroscience
The strength of a fear memory significantly influences whether it drives adaptive or maladaptive behavior in the future. Yet, how mild and strong fear memories differ in underlying biology is not well understood. We hypothesized that this distinction may not be exclusively the result of changes within specific brain regions, but rather the outcome of collective changes in connectivity across multiple regions within the neural network. To test this, rats were fear conditioned in protocols of varying intensities to generate mild or strong memories. Neuronal activation driven by recall was measured using c-fos immunohistochemistry in 12 brain regions implicated in fear learning and memory. The interregional coordinated brain activity was computed and graph-based functional networks were generated to compare how mild and strong fear memories differ at the systems level. Our results show that mild fear recall is supported by a well-connected brain network with small-world properties in which the amygdala is well-positioned to be modulated by other regions. In contrast, this connectivity is disrupted in strong fear memories and the amygdala is isolated from other regions. These findings indicate that the neural systems underlying mild and strong fear memories differ, with implications for understanding and treating disorders of fear dysregulation.