1. Neuroscience
Download icon

Local circuit allowing hypothalamic control of hippocampal area CA2 activity and consequences for CA1

  1. Vincent Robert
  2. Ludivine Therreau
  3. Vivien Chevaleyre
  4. Eude Lepicard
  5. Cécile Viollet
  6. Julie Cognet
  7. Arthur JY Huang
  8. Roman Boehringer
  9. Denis Polygalov
  10. Thomas J McHugh
  11. Rebecca Ann Piskorowski  Is a corresponding author
  1. INSERM UMR1266, Institute of Psychiatry and Neuroscience of Paris, Team Synaptic Plasticity and Neural Networks, Université de Paris, France
  2. GHU Paris Psychiatrie and Neurosciences, France
  3. Laboratory for Circuit and Behavioral Physiology, RIKEN Center for Brain Science, Japan
Research Article
  • Cited 0
  • Views 831
  • Annotations
Cite this article as: eLife 2021;10:e63352 doi: 10.7554/eLife.63352

Abstract

The hippocampus is critical for memory formation. The hypothalamic supramammillary nucleus (SuM) sends long-range projections to hippocampal area CA2. While the SuM-CA2 connection is critical for social memory, how this input acts on the local circuit is unknown. Using transgenic mice, we found that SuM axon stimulation elicited mixed excitatory and inhibitory responses in area CA2 pyramidal neurons (PNs). Parvalbumin-expressing basket cells were largely responsible for the feedforward inhibitory drive of SuM over area CA2. Inhibition recruited by the SuM input onto CA2 PNs increased the precision of action potential firing both in conditions of low and high cholinergic tone. Furthermore, SuM stimulation in area CA2 modulated CA1 activity, indicating that synchronized CA2 output drives a pulsed inhibition in area CA1. Hence, the network revealed here lays basis for understanding how SuM activity directly acts on the local hippocampal circuit to allow social memory encoding.

Introduction

The hippocampus is critical for memory formation and spatial navigation (Buzsáki and Moser, 2013; Eichenbaum and Cohen, 2014), yet basic questions persist regarding the underlying circuitry and cellular components. While area CA2 has been shown to play a significant role in several hippocampal processes including social memory formation (Hitti and Siegelbaum, 2014; Stevenson and Caldwell, 2014) sharp-wave ripple generation (Oliva et al., 2016a) and spatial encoding (Kay et al., 2016), information about the local circuitry and cellular mechanisms allowing these functions is lacking. There is mounting evidence that generalizations cannot be made from the rich understanding of areas CA1 and CA3, as neurons in area CA2 have been shown to have unique molecular expression profiles (Cembrowski et al., 2016; Lein et al., 2004), morphology (Bartesaghi and Ravasi, 1999; Nò, 1934) and cellular properties (Robert et al., 2020; Srinivas et al., 2017; Sun et al., 2014). Notably, and in contrast to area CA1, CA2 pyramidal neurons do not undergo high-frequency stimulation-induced synaptic plasticity (Dasgupta et al., 2020; Zhao et al., 2007). Rather, the excitability of this region is tightly controlled by a highly plastic network of inhibitory neurons (Leroy et al., 2017; Nasrallah et al., 2015; Piskorowski and Chevaleyre, 2013). When active, CA2 pyramidal neurons (PNs) can strongly drive area CA1 (Chevaleyre and Siegelbaum, 2010; Kohara et al., 2014; Nasrallah et al., 2019), thereby influencing hippocampal output. Furthermore, CA2 neurons also project to area CA3, where they recruit inhibition (Boehringer et al., 2017; Kohara et al., 2014) and act to control hippocampal excitability. Thus, CA2 neurons are poised to have long-reaching effects in the hippocampus, and a better understanding of the regulation of neuronal activity in this region is needed.

The hypothalamic supramammillary (SuM) nucleus sends projections to both area CA2 and the dentate gyrus (DG) (Haglund et al., 1984; Vertes, 1992). These long-range connections have been shown in several species including rodents, primates, and humans (Berger et al., 2001; Haglund et al., 1984; Wyss et al., 1979) where they are present in early hippocampal development. The SuM has been found to be active during a wide variety of conditions including novel environment exposure (Ito et al., 2009), reinforcement learning (Ikemoto, 2005; Ikemoto et al., 2004), food anticipation (Le May et al., 2019), and during REM sleep and arousal (Pedersen et al., 2017; Renouard et al., 2015). This nucleus is also known for participating in hippocampal theta rhythm (Pan and McNaughton, 2002; Pan and McNaughton, 1997), possibly by its direct projection to the hippocampus or by modulation of the medial septum (Borhegyi et al., 1998; Vertes and Kocsis, 1997), and regulating spike-timing between hippocampus and the cortex (Ito et al., 2018). Disruption of SuM neuron activity with pharmacological methods (Aranda et al., 2008; Shahidi et al., 2004) or lesions (Aranda et al., 2006) has been reported to disrupt hippocampal memory. Serotonin depletion of the SuM leads to deficiencies in spatial learning in the Morris water maze, and results in altered hippocampal theta activity (Gutiérrez-Guzmán et al., 2012; Hernández-Pérez et al., 2015). Salient rewarding experiences also activate the SuM, as evidenced by cFos expression in monoaminergic SuM neurons by consumption of rewarding food (Plaisier et al., 2020). Furthermore, the rewarding aspects of social aggression have been shown to involve an excitatory circuit between the hypothalamic ventral premammillary nucleus and the SuM (Stagkourakis et al., 2018). It has recently been shown that there are two separate populations of cells in the SuM that target either CA2 or the DG (Chen et al., 2020). In the DG, the SuM terminals release both glutamate and GABA (Boulland et al., 2009; Chen et al., 2020; Hashimotodani et al., 2018; Pedersen et al., 2017; Soussi et al., 2010). The SuM-DG projection has been recently shown to play a role in modulating DG activity in response to contextual novelty (Chen et al., 2020) and spatial memory retrieval (Li et al., 2020). In contrast, functional studies of the SuM-CA2 projection have found that this connection is entirely glutamatergic (Chen et al., 2020). It was recently discovered that the CA2-projecting SuM neurons are active during social novelty exposure, and their selective stimulation prevents expression of a memory of a familiar conspecific (Chen et al., 2020). These findings strongly suggest that the SuM-CA2 connection conveys a social novelty signal to the hippocampus. Furthermore, recent in vivo recordings from the SuM in anesthetized rats reported that a subset of SuM neurons were active earlier than CA2 and other hippocampal cells during SWR (Vicente et al., 2020), indicating a possible role for the SuM-CA2 projection in shaping area CA2 activity prior to SWR onset.

Even with the anatomical and in vivo data, the properties and consequences of SuM activation on area CA2 activity remain unexplored. In this study, we use a combination of approaches to specifically examine the effects of SuM input stimulation on neuronal activity in hippocampal area CA2. Here, we show that the SuM-evoked post-synaptic excitation of CA2 PN is controlled by SuM-driven inhibition. We identified PV-expressing basket cells as the neuronal population most strongly excited by SuM input in area CA2, and thus likely responsible for the feedforward inhibition evoked by SuM in CA2 PNs. We found that recruitment of this inhibition enhances the precision of AP firing by area CA2 PNs in conditions of low and high cholinergic tone. Finally, we observed that the resulting synchronized CA2 PN activity drives inhibition in area CA1, thereby providing a circuit mechanism through which SuM can modulate hippocampal excitability by controlling area CA2 output.

Results

SuM axons provide excitatory glutamatergic input to pyramidal neurons in area CA2 and CA3a

Its small size and cellular heterogeneity have made the SuM a difficult region to study. It has been shown that the source of vesicular glutamate transporter 2 (VGluT2)-immunopositive boutons in area CA2 originate from the SuM (Halasy et al., 2004). In order to more closely examine the SuM-CA2 long-range connection, we injected a retrograde canine adenovirus type 2 (CAV-2) into area CA2 of the hippocampus to permit the expression of Cre-recombinase (Cre) in hippocampal-projecting SuM neurons, and an adeno-associated virus (AAV) was injected into the SuM to allow the expression of EGFP under the control of Cre (Figure 1—figure supplement 1A). In five animals the injection of retrograde CAV-2 was sufficiently targeted to area CA2, as indicated by the presence of EGFP-expressing SuM axonal fibers primarily in this hippocampal area (Figure 1—figure supplement 1B). We stained for calretinin to define the boundaries of the SuM nucleus (Pan and McNaughton, 2004). Consistent with recent findings using retrograde AAV vectors (Chen et al., 2020), we observed that CA2-projecting cells express calretinin and are located in the medial SuM (Figure 1—figure supplement 1C–D). These cells were located bilaterally, ventral to the fiber bundles that traverse the SuM (Figure 1—figure supplement 1C). Furthermore, we confirmed that these cells also stain for VGluT2 (Figure 1—figure supplement 1E).

In order to better understand the cellular targets and consequences of SuM input activity in area CA2, we injected an AAV to express channelrhodopsin(H143R)-YFP (ChR2-EYFP) under the control of Cre into the SuM of a transgenic mouse line with Cre expression controlled by the VGluT2 promoter, the Tg(Slc17a-icre)10Ki line (Borgius et al., 2010; Figure 1—figure supplement 1F). In parallel, we used the Csf2rb2-Cre mouse line that selectively expresses Cre in the SuM (Chen et al., 2020; Figure 1A). We found that with both transgenic mouse lines we could reproducibly restrict expression of ChR2-EYFP in the SuM and avoid infecting nearby hypothalamic regions that also project to the hippocampus (Figure 1A, Figure 1—figure supplement 1F). Furthermore, with both lines of transgenic mice, we observed identical patterns of SuM fiber localization in the hippocampus. EYFP-containing SuM axons were found throughout the granule cell layer of the DG and in area CA2 (Figure 1B) where they clustered around the pyramidal layer (stratum pyramidale, SP). The SuM fiber projection area was clearly restricted to area CA2, as defined by expression of the CA2-specific markers PCP4 (Figure 1—figure supplement 1B) and RGS14 (Figure 1B), and did not spread to neighboring areas CA3 and CA1. In order to maximize the precision of our experiments, we frequently only achieved partial infection of the SuM, as indicated by the sparseness of ChR2-EYFP-containing fibers in comparison to the number of VGluT2-stained boutons in this region (Figure 1—figure supplement 1G–H).

Figure 1 with 2 supplements see all
Selective functional mapping of SuM neurons that project to hippocampal area CA2.

(A) Left, diagram illustrating the injection of AAVs into the SuM. Middle, sagittal image indicating the infected SuM area expressing hCHR2(H134R)-EYFP (green). Right, expanded view of injection site in the Csf2rbr-Cre mouse line. (B) Left, hCHR2(H134R)-EYFP -expressing SuM fibers (green) and nissl staining (blue) in the hippocampus. Right, higher magnification image of area CA2 with hCHR2(H134R)-EYFP -expressing SuM fibers (green) and nissl staining (blue) and RGS14 staining (magenta) to label area CA2. (C) CA2 pyramidal neurons in the SuM-innervated region receive excitatory transmission. (C1) Example CA2 PN reconstruction (dendrites in black, axons in gray, hippocampal stratum borders shown in dotted line, area demarcated in blue corresponds to the expanded image in C2). (C2) Biocytin labeling of the recorded cell proximal dendrites, scale bar represents 10 µm. (C3) AP firing and repolarizing sag potential in response to steps of +800 and −400 pA current injection. (C4) Light-evoked EPSPs (top traces, individual traces shown in gray, average trace shown in black) and EPSCs (bottom traces, individual traces shown in gray, average trace shown in black). (D) CA3a pyramidal neurons in the SuM-innervated region receive excitatory transmission. (D1) Example CA3 PN reconstruction (dendrites in brown, axons in light brown, hippocampal stratum borders shown in dotted line, area demarcated in blue corresponds to the expanded image in D2). (D2) Biocytin labeling of the recorded cell proximal dendrites, note the presence of thorny excrescences, as indicated by the red arrows; scale bar represents 10 µm. (D3) AP firing and repolarizing sag potential in response to steps of +800 and −400 pA current injection. (D4) Light-evoked EPSPs (top traces, individual traces shown in gray, average trace shown in black) and EPSCs (bottom traces, individual traces shown in gray, average trace shown in black).

We performed whole-cell current and voltage clamp recordings of PNs across the hippocampal CA regions and activated projecting axons with pulses of 488 nm light in acute hippocampal slices. Following all recordings, we performed post-hoc anatomical reconstructions of recorded cells and axonal fibers, as well as immunohistochemical staining for CA2-area markers. Additionally, injection sites were examined post hoc to ensure correct targeting of the SuM.

We observed that photostimulation of SuM axons elicited excitatory post-synaptic responses in 63% of PNs (n = 166 of 263 cells) located in area CA2. PNs in this region shared similar overall dendritic morphologies and electrophysiological properties (Table 1) but differed along two criteria. First, in stratum lucidum where the DG mossy fibers (MF) project, some PNs clearly had thorny excrescences (TE) while others had very smooth apical dendrites (Figure 1C–D). Based on the presence of TEs, we classified cells as CA2 or CA3a PNs (unequivocal distinction was possible for 148 neurons). Second, the distribution of the locations of PN soma along the radial axis of the hippocampus allowed us to cluster them as deep (closer to stratum oriens, SO) or superficial (closer to stratum radiatum, SR) subpopulations (unequivocal distinction was possible for 157 neurons). We found that the SuM-PN connectivity was not different between CA2 and CA3 PNs (Table 2, χ² test for CA2 and CA3 PNs, p=0.572) or between deep and superficial PNs (Table 2, χ² test for deep and superficial PNs, p=0.946). Light-evoked excitatory post-synaptic potentials (EPSPs) and excitatory post-synaptic currents (EPSCs) recorded at −70 mV were of fairly small amplitude (Figure 1C–D) that were similar regardless of the PN type or somatic location (Table 2, Mann-Whitney U test for CA2 and CA3 PNs, p=0.409; Mann-Whitney U test for deep and superficial PNs, p=0.306). Because no significant differences in post-synaptic responses to SuM input stimulation were observed between CA2 and CA3 PNs as well as between deep and superficial PNs, data from all PNs was pooled for the rest of the study. The small amplitude of SuM input-evoked post-synaptic responses in PNs was not due to suboptimal stimulation of SuM axons as EPSC amplitudes rapidly reached a plateau when increasing light intensity (Figure 1—figure supplement 2A–B). We are confident that this transmission is due to action potential-generated vesicle release because all transmission was blocked following application of the sodium channel blocker tetrodotoxin (TTX) (Figure 1—figure supplement 2B). The pure glutamatergic nature of the SuM input was confirmed by the complete block of light-evoked synaptic transmission following the application of the AMPA and NMDA receptors antagonists NBQX and D-APV (Figure 1—figure supplement 2C; amplitudes were −16 ± 4.8 pA in control and −1.8 ± 0.3 pA in NBQX and D-APV, n = 6; Wilcoxon signed-rank test, p=0.03). These data confirm that SuM inputs provide long-range glutamatergic excitation to CA2 and CA3 PNs in area CA2.

Table 1
Electrophysiological properties of pyramidal neurons in SuM-innervated area.
VM (mV)RM (MOhm)CM (pF)
CA2 PN (n = 81)−69.8 ± 0.7059.2 ± 2.65209 ± 11.4
CA3 PN (n = 31)−70.3 ± 1.0672.4 ± 4.82211 ± 15.7
StatisticsMann-Whitney U test
p=0.997
Student T test
p=0.020*
Mann-Whitney U test
p=0.625
PN deep (n = 57)−71.1 ± 0.7664.0 ± 3.94200 ± 12.3
PN superficial (n = 76)−69.3 ± 0.6764.9 ± 3.19196 ± 11.8
StatisticsStudent T test
p=0.077
Mann-Whitney U test
p=0.777
Mann-Whitney U test
p=0.588
Table 2
Characteristics of SuM light-evoked transmission onto pyramidal neurons.
EPSC
Cell typeConnectivity (%)Amplitude (pA)Rise time (ms)Decay time (ms)Latency (ms)Success rate
CA2 PN56 (n = 58 of 103)−16 ± 1.92.9 ± 0.114 ± 0.82.4 ± 0.20.44 ± 0.03
CA3 PN49 (n = 22 of 45)−23 ± 5.93.0 ± 0.214 ± 0.92.7 ± 0.30.56 ± 0.06
Statisticsχ² test
p=0.572
Mann-Whitney U test
p=0.409
Mann-Whitney U test
p=0.391
Mann-Whitney U test
p=0.797
Mann-Whitney U test
p=0.156
Student T test
p=0.074
PN deep56 (n = 35 of 63)−15 ± 2.03.5 ± 0.216 ± 1.03.5 ± 0.40.39 ± 0.03
PN superficial56 (n = 53 of 94)−20 ± 3.03.1 ± 0.215 ± 0.92.7 ± 0.30.51 ± 0.04
Statisticsχ² test
p=0.946
Mann-Whitney U test
p=0.306
Mann-Whitney U test
p=0.051
Mann-Whitney U test
p=0.314
Mann-Whitney U test
p=0.083
Mann-Whitney U test
p=0.072
IPSC
Cell typeConnectivity (%)Amplitude (pA)Rise time (ms)Decay time (ms)Latency (ms)Success rate
CA2 PN35 (n = 19 of 55)197 ± 41.33.8 ± 0.425 ± 1.26.3 ± 0.70.55 ± 0.06
CA3 PN57 (n = 16 of 28)145 ± 23.44.5 ± 0.425 ± 1.27.5 ± 0.90.54 ± 0.05
Statisticsχ² test
p=0.134
Mann-Whitney U test
p=0.870
Student T test
p=0.203
Mann-Whitney U test
p=0.896
Mann-Whitney U test
p=0.303
Student T test
p=0.893
PN deep47 (n = 16 of 34)199 ± 40.63.8 ± 0.425 ± 1.47.2 ± 0.80.52 ± 0.07
PN superficial47 (n = 26 of 55)167 ± 27.54.9 ± 0.426 ± 1.26.8 ± 0.70.50 ± 0.05
Statisticsχ² test
p=0.987
Mann-Whitney U test
p=0.258
Student T test
p=0.047*
Student T test
p=0.564
Student T test
p=0.706
Student T test
p=0.796

PNs in area CA2 receive mixed excitatory and inhibitory responses from the SuM input

Using whole-cell voltage clamp recordings in area CA2 and the dentate gyrus (DG), we have previously shown that the CA2-targeting and DG-targeting SuM neurons have contrasting neurotransmitter modalities (Chen et al., 2020). Our results and other have demonstrated that glutamate and GABA are co-released at SuM-DG synapses (Boulland et al., 2009; Chen et al., 2020; Hashimotodani et al., 2018; Pedersen et al., 2017; Soussi et al., 2010), but that the SuM-CA2 synapses are exclusively glutamatergic (Chen et al., 2020). We have previously shown that SuM input stimulation in area CA2 evokes a very large inhibitory post-synaptic current (IPSC) that is entirely due to feedforward inhibition based on the delayed response latencies of IPSCs as compared to EPSCs, the complete block of IPSCs by NBQX and APV, and the complete abolition of IPSCs but sparing of EPSCs in the presence of TTX and 4-amino pyridine (Chen et al., 2020). Because photostimulation of SuM input elicited excitatory post-synaptic potentials (PSPs) of fairly small amplitude in area CA2 PNs held at −70 mV (Figure 1C 4 and D4), we asked if the amplitude of SuM input stimulation-evoked PSPs in PNs could be controlled by feedforward inhibition. Interestingly, blocking inhibitory transmission with the GABAA and GABAB receptor antagonists SR95531 and CGP55845A led to a significant increase of light-evoked PSP amplitude recorded in area CA2 PNs (Figure 2A–C; amplitudes of the first response were 0.18 ± 0.05 mV in control and 0.24 ± 0.05 mV in SR95531 and CGP55845A, n = 14; Wilcoxon signed-rank tests, p=0.004 for the first PSP, p=0.013 for the second PSP, p<0.001 for the third PSP). Thus, this result demonstrates a negative control of SuM-driven excitation by feedforward inhibition.

SuM input drives inhibition that controls excitation in CA2 PNs.

Whole-cell current clamp recordings of light-evoked post-synaptic potentials (PSPs) from SuM input stimulation onto CA2 PNs reveal contribution of feedforward inhibition in dampening excitatory input at −70 mV. (A) Diagram illustrating whole-cell recording configuration in acute hippocampal slices. During these experiments, direct current was injected as necessary to maintain a membrane potential of −70 mV. (B) Sample traces of three 10 Hz SuM light-evoked PSPs in a CA2 PN before and after blocking inhibitory transmission (control shown in black, SR95531 and CGP55845A in gray). (C) Summary graph of light-evoked PSP amplitudes recorded in PNs before and after application of 1 µM SR95531 and 2 µM CGP55845A (individual cells shown as thin lines, population average shown as thick line, error bars represent SEM, n = 14; Wilcoxon signed-rank tests, p=0.004 for the first PSP, p=0.013 for the second PSP, p<0.001 for the third PSP). (D-H) Summation of SuM synaptic potentials with SR and SLM electrical input stimulation. (D) Diagram illustrating the recording configuration similar to panel A but with stimulating electrodes positioned in stratum lacunosum moleculare (SLM) and stratum radiatum (SR). (E) Left, example traces of PSPs evoked by SuM fiber light stimulation alone (blue trace). Center, PSPs evoked by electrical stimulation of SR inputs alone (black) or paired with simultaneous SuM stimulation (orange). Right, PSPs evoked by electrical stimulation of SLM inputs alone (black) or paired with SuM stimulation (green). (F) Plots of the difference between the mathematical summation of the amplitudes of the SuM PSP amplitude and electrical stimulation (linear summation) and the measured SuM +electrical PSP. Left, SR inputs are not significantly different from zero, indicating that SuM and SR inputs linearly summate. Right, for the first pulse, the measured SLM + SuM amplitude is significantly smaller (n = 10; T test, p=0.014) than the expected linear summation. (G) Left, example traces of 10 Hz trains of PSPs of either electrical stimulation (black traces) or trains of paired electrical and light stimulation of SuM fibers (SR + SuM in orange or SLM + SuM in green). Right, traces with amplitudes normalized to the first PSP for both the electrical and simultaneous light and electrical PSPs. The amplitudes for all PSPs are measured from the potential immediately before each stimulus. (H) Summary plots of the summation ratio of the 2nd, 3rd, and 4th PSP for electrical stimulation (black symbols) or paired stimulation of SR + SuM (left, orange) or SLM + SuM (right, green).

Given the combination of direct excitation and feedforward inhibition from SuM inputs onto CA2 pyramidal cells, we asked how this input would summate with other synaptic inputs in the CA2 dendritic arbor. Hippocampal area CA2 receives synaptic input from CA3 in stratum radiatum (SR). Stimulation of CA3 inputs evokes a very strong feedforward inhibition, such that it is exceptionally difficult to evoke action potential firing in CA2 pyramidal neurons when inhibitory transmission is intact (Chevaleyre and Siegelbaum, 2010; Nasrallah et al., 2015; Piskorowski and Chevaleyre, 2013). Additionally, CA2 PNs receive synaptic input from the entorhinal cortex in stratum lacunosum molecular (SLM). These inputs are very distal but relatively less attenuated in CA2 PNs in comparison to distal inputs in CA1 (Chevaleyre and Siegelbaum, 2010; Srinivas et al., 2017). In order to answer how the SuM input interacts with the CA3 and entorhinal inputs in area CA2, we electrically stimulated synaptic inputs in SR and SLM in the presence and absence of simultaneous SuM fiber stimulation (Figure 2D). In summary, we found that when the CA2 PNs were kept at −70 mV, SuM input stimulation paired with SR or SLM input had a net depolarizing effect. We measured the amplitudes of the light-evoked SuM PSP, the electrically evoked PSP of either SR or SLM stimulation and the paired SuM and electrical PSP (Figure 2E). For SR input stimulation, we found no significant difference between the observed paired SR+SuM amplitude and the calculated linear summated amplitude (SR alone + SuM alone) (Figure 2F). This was observed for all four pulses of input summations delivered at 10 Hz. However, for the SLM input stimulation, the observed paired amplitude was significantly smaller than the linear summation of the two inputs (SLM alone + SuM alone) for the first stimulus (n = 10; T test, p=0.014) (Figure 2F). This observation is expected, as the attenuation of distal dendritic SLM inputs causes the peak of the PSP to be delayed relative to the more somatic SuM input. Thus, the SuM input paired with either SR or SLM input stimulation has minor depolarizing effect on the PSP in CA2 PNs. However, the SuM input might have different effect on the SR and SLM inputs depending on the precise timing of their activation.

We also examine the summation ratio for a train of 4 PSPs at 10 Hz from SR and SLM synaptic inputs stimulation with and without simultaneous SuM input stimulation (Figure 2G–H). We observed a significant reduction of the summation ratio as measured by the ratio of the n-th pulse to the first (Pn/P1) for both SR (n = 10; repeated-measures ANOVA, p=2.3×10−4) and SLM (n = 10; repeated-measures ANOVA, p=8.5×10−4). This observation that concomitant SuM activity is reducing the level of facilitation of several pulses in a train indicates that the short-term dynamics of the SuM-driven excitation and feedforward inhibition are playing a role to prevent cellular excitation from other inputs.

Basket cells are strongly recruited by the SuM input

Because the hippocampus hosts a variety of interneurons (INs) that are involved in controlling specific aspects of PN excitability, we wished to establish which kind of IN was targeted by the SuM input to area CA2. We performed whole-cell recordings from INs in this area and assessed post-synaptic excitatory responses to SuM axons stimulation in these cells (Figure 3). In contrast with previous reports of an exclusive innervation of PNs by SuM (Maglóczky et al., 1994), we observed robust light-evoked excitatory transmission from SuM axons in 35 out of 62 interneurons (INs) with soma located in SP. Following biocytin-streptavidin staining and anatomical reconstructions of recorded INs (allowing unequivocal identification in 48 neurons), we were able to classify INs based on their physiological properties, somatic location and axonal arborization location. We classified 22 cells as basket cells (BCs) because their axonal arborizations were restricted to SP (Figure 3A). BCs fired APs at high frequency either in bursts or continuously upon depolarizing current injection and showed substantial repolarizing sag current when hyperpolarized (Figure 4A, Table 3). Light-evoked EPSCs and PSPs were readily observed in the vast majority of BCs (Figure 3A, C and D, Table 4) and reached large amplitudes in some instances. An additional 26 INs with soma in SP were classified as non-BCs because their axon did not target SP (Figure 3B). In our recordings, these cells fired in bursts and showed little sag during hyperpolarizing current injection steps (Table 3). We consistently observed no or very minor light-evoked excitatory transmission onto non-BCs (Figure 3B–C, Table 4). Furthermore, we recorded from 17 INs that had soma in stratum oriens (SO) and nine in stratum radiatum (SR). Like non-BCs, these INs did not receive strong excitation from SuM fibers (Table 4). This data is consistent with the conclusion that SuM input preferentially forms excitatory synapses onto basket cells in area CA2.

SuM input provides strong excitatory glutamatergic transmission to basket cells (BCs) in area CA2.

(A-B) Left, diagrams illustrating whole-cell recordings in area CA2 and SuM fiber stimulation in acute slice preparation. Middle, example reconstruction of different cell types (soma and dendrites in thick lines, axon in thin lines, hippocampal strata in dotted gray lines). Right, sample traces of light-evoked EPSPs (top, individual traces in gray, average trace in black) and EPSCs (bottom, individual traces in gray, average trace in black). (A) Basket cell in area CA2. (B) Non-basket cell in area CA2. (C) Summary graph of light-evoked EPSC potencies in PNs, BCs, and non-BCs in area CA2 (individual cells shown as dots, population average shown as thick line, error bars represent SEM, PNs: n = 166; BC INs: n = 18; non-BCs: n = 13; Kruskal-Wallis test with Dunn-Holland-Wolfe post hoc test, p=0.022). (D) Summary graph of light-evoked PSP amplitudes in PNs, BCs and non-BCs (individual cells shown as dots, population average shown as thick line, error bars represent SEM, PNs: n = 20; BCs: n = 10; non-BCs: n = 4; Kruskal-Wallis test with Dunn-Holland-Wolfe post hoc test, p<0.001). (E) Left, proportion of post-synaptic CA2 PNs, BCs and non-BCs firing action potentials time-locked to light stimulation of SuM input. Right, sample traces of light-evoked action potentials in a BC recorded in current-clamp at resting membrane potential (top) and in cell-attached (bottom) configurations.

SuM input provides excitation to Parvalbumin-expressing BCs.

(A) Three biocytin reconstructions of BC INs with dendrites in red and axons in light red. Inset, current clamp steps to −400 pA and +400 pA display high-frequency AP firing and repolarizing sag current. (B) Corresponding light-evoked EPSCs and EPSPs for the three reconstructed neurons (individual traces in gray, average trace in black). (C) Corresponding PV immunostaining of the three interneurons: parvalbumin staining, biocytin labeling of the recorded cell, and merge (PV in magenta and biocytin in green).

Table 3
Electrophysiological properties of interneurons in SuM-innervated area.
VM (mV)RM (MOhm)CM (pF)Firing adaptation indexSag (mV)
Basket cell (n = 16)−57.3 ± 1.38144 ± 28.164.0 ± 8.700.74 ± 0.059.4 ± 1.0
Non-Basket Cell (n = 12)−55.6 ± 1.84224 ± 46.852.0 ± 5.900.57 ± 0.065.9 ± 1.4
Interneuron SO (n = 6)−57.0 ± 3.16201 ± 21.044.7 ± 5.310.61 ± 0.117.6 ± 1.9
Interneuron SR (n = 8)−60.1 ± 2.89282 ± 49.839.6 ± 3.180.65 ± 0.098.1 ± 2.1
StatisticsOne-way
ANOVA test
p=0.527
One-way
ANOVA test
p=0.100
Kruskal-Wallis test
p=0.354
One-way
ANOVA test
p=0.238
One-way
ANOVA test
p=0.292
Table 4
Characteristics of excitatory SuM light-evoked transmission onto interneurons and pyramidal cells.
Cell typeConnectivity (%)Amplitude (pA)Rise time (ms)Decay time (ms)Latency (ms)Success rate
Pyramidal cell63 (n = 166 of 263)−19 ± 1.6*3.4 ± 0.1*15 ± 0.5*2.9 ± 0.10.46 ± 0.02
Basket cell82 (n = 18 of 22)−43 ± 8.7*1.7 ± 0.3*8.4 ± 1.3*3.1 ± 0.40.59 ± 0.07
Non-Basket Cell
Interneuron SO
Interneuron SR
39 (n = 10 of 26)
12 (n = 2 of 17)
11 (n = 1 of 9)
−16 ± 2.82.6 ± 0.512 ± 1.43.4 ± 0.70.36 ± 0.06
Statisticsχ² test
p=0.006*
Kruskal-Wallis test
p=0.016
Dunn-Holland-Wolfe post hoc
p<0.05*
One-way
ANOVA test
p<0.001
Tukey post hoc
p<0.001*
One-way
ANOVA test
p<0.001
Tukey post hoc
p<0.001*
One-way
ANOVA test
p=0.580
One-way
ANOVA test
p=0.066

To fully assess the strength of SuM inputs onto the different cell types, we examined the following parameters for each population: the connectivity, success rate, amplitude, potency, kinetics, and latencies of EPSCs as well as the resulting depolarization of the membrane potential. First, SuM inputs preferentially innervated BCs as evidenced by a higher connectivity of EPSCs in BCs than in PNs or other INs (Table 4). Importantly, excitatory responses had short latencies with limited jitter (Table 4) indicating that the connection was monosynaptic in all cell types. When voltage-clamping cells at −70 mV, light-evoked EPSCs could be compared between different cell populations. However, not every photostimulation gave rise to an EPSC leading to an average success rate that tended to be highest in BCs (Table 4). In addition, BCs appeared to receive more excitation from the SuM input than other cells types, as the amplitude of EPSCs was larger in BCs than in PNs (Table 4). EPSCs recorded in BCs also had faster kinetics than in PNs (Table 4). Interestingly, combining the success rate of EPSCs with their respective amplitudes to compute the potency of the SuM synapses revealed that it was significantly larger in BCs than in PNs and non-BCs (Figure 3C; potencies were −12 ± 1.6 pA for PNs, n = 166; −29 ± 7.8 pA for BCs, n = 18; −5.9 ± 1.5 pA for non-BCs, n = 13; Kruskal-Wallis test with Dunn-Holland-Wolfe post hoc test, p=0.022). Consequently, EPSPs recorded at −70 mV were of larger amplitude in BCs than in PNs and non-BCs (Figure 3D; amplitudes were 0.44 ± 0.06 mV for PNs, n = 20; 1.71 ± 0.57 mV for BCs, n = 10; 0.53 ± 0.07 mV for non-BCs, n = 4; Kruskal-Wallis test with Dunn-Holland-Wolfe post hoc test, p<0.001). When recording cell-attached or current-clamping BCs at their resting membrane potential (VM), photostimulation of SuM axons was able to evoke AP firing (Figure 3E) in multiple instances (n = 7 of 13). However, this was never observed in PNs (n = 0 of 78), non-BCs (n = 0 of 16), SR INs (n = 0 of 9) or SO INs (n = 0 of 8). These results show that SuM projections to area CA2 preferentially provide excitation to BCs that are likely responsible of the feedforward inhibition observed in PNs. This is in accordance with an efficient control of area CA2 PNs excitation by the SuM inhibitory drive as axons from BCs deliver the feedforward inhibition to the peri-somatic region of PNs, effectively shunting incoming PSPs from both the SuM and from dendritic-targeted inputs in SR and SLM.

Parvalbumin-expressing basket cells mediate the feedforward inhibition driven by SuM

In the hippocampus, BCs express either cholecystokinin (CCK) or parvalbumin (PV) (Klausberger and Somogyi, 2008). We found that in response to a 1 s depolarizing pulse, most BCs that received strong SuM excitatory input displayed very fast AP firing with little accommodation in the AP firing frequency (Figure 4A–B, Table 3). This firing behavior is similar to what has been reported for fast spiking PV-expressing BCs in CA1 (Pawelzik et al., 2002). In contrast, CCK-expressing BCs show a lower firing frequency and more accommodation during the train (Pawelzik et al., 2002). This result suggests that BCs connected by the SuM may be expressing PV. To directly confirm this hypothesis, we performed post hoc immunostaining of recorded interneurons that received strong excitation from SuM input. Because of the dialysis inherent to the whole-cell recording conditions, we encountered difficulty staining for multiple cells. However, PV-immunoreactivity could unequivocally be detected in either the soma or dendrites of seven connected BCs (Figure 4C). Therefore, this data demonstrates that at least a fraction of the recorded BCs connected by the SuM is expressing PV.

Hence, to address whether the lack of PV staining in some cells was a consequence of dialysis or resulted from the fact that non-PV +BCs are also connected, we made use of different strategies to differentiate PV+ and CCK+ INs. First, we wished to genetically confirm that PV+ INs are involved in the SuM-driven feedforward inhibition of area CA2 PNs. We used inhibitory Gi-DREADD to selectively inhibit PV+ INs in area CA2 while monitoring feedforward IPSCs from area CA2 PNs in response to SuM stimulation. To achieve that, we injected AAVs expressing a Cre-dependent hM4D(Gi) inhibitory Gi-DREADD in area CA2 of PV-Cre mice together with AAVs expressing ChR2 with a pan-neuronal promoter in the SuM (Figure 5A). While we were able to obtain very specific expression of DREADD in PV+ INs, only a fraction of PV+ INs had detectable DREADD expression as quantified by immunohistochemistry (Figure 5B; fraction of PV+ INs expressing DREADDs in CA2 = 75 ± 3.5%, n = 13). We observed a substantial reduction of SuM-evoked IPSC amplitude recorded in area CA2 PNs upon application of 10 µM of the Gi-DREADD ligand CNO (Figure 5C; amplitudes were 847 ± 122 pA in control and 498 ± 87 pA in CNO hence a 42 ± 6.0% block, n = 13; paired-T test, p<0.001). Although we never measured a complete block of inhibitory responses, this result unequivocally places PV+ INs as mediators of the SuM feedforward inhibition of area CA2 PNs. The incomplete block of IPSCs observed in these experiments indicates that either additional non-PV+ INs are recruited by SuM input or that our silencing of PV-mediated feedforward inhibition is incomplete. This could be a consequence of partial infection of PV +INs in area CA2 by AAVs carrying DREADDs and partial silencing of DREADD-expressing PV+ INs by CNO. To address the latter, we performed whole-cell recordings from Gi-DREADD-expressing CA2 PV+ INs labeled with mCherry and monitored the variations in VM level and action potential firing to SuM input stimulation before and after CNO application (Figure 5—figure supplement 1A). We found that CNO application caused a significant hyperpolarization of Gi-DREADD-expressing CA2 PV+ INs, albeit modest in magnitude (Figure 5—figure supplement 1B–D; VM were −55.3 ± 2.3 mV in ACSF and −61.8 ± 2.7 mV in CNO hence a −6.5 ± 2.4 mV hyperpolarization, n = 6; Wilcoxon signed-rank test, p=0.031). While this confirmed the relevance of our silencing strategy, it highlighted the possibility that Gi-DREADD-expressing CA2 PV+ INs may not be fully silenced by CNO. Indeed, we observed residual SuM-evoked AP firing in these cells after CNO application (Figure 5—figure supplement 1D–E). These data indicate that synaptically evoked somatic AP firing is not fully blocked by CNO in Gi-DREADD-expressing CA2 PV+ INs. Because it is difficult to distinguish between partial silencing of PV INs by Gi-DREADDs or recruitment of other types of INs in the SuM-driven feedforwards inhibition, we adopted other complementary strategies to answer this question.

Figure 5 with 1 supplement see all
Parvalbumin-expressing BCs mediate the feedforward inhibition recruited by photostimulation of SuM fibers.

(A-C) Silencing of PV+ INs by inhibitory DREADDs reduces SuM feedforward inhibition onto area CA2 PNs. (A) Diagram illustrating the method to infect SuM neurons and selectively inhibit PV+ INs in area CA2. An AAV allowing the Cre-dependent expression of inhibitory DREADD was injected bilaterally into area CA2 of the dorsal hippocampus and another AAV allowing the expression of ChR2 was injected into the SuM of PV-Cre mice, allowing optogenetic stimulation of SuM inputs and pharmacogenetic inhibition of PV+ INs by application of the DREADD agonist CNO at 10 µM. (B) Example immunostaining against PV, DREADD and biocytin labeling in area CA2 from a slice used in these experiments. (C) Left, diagram of the recording configuration in hippocampal slices. Center, sample traces (control in red, CNO in gray). Right, summary graph of light-evoked IPSC amplitudes recorded in CA2 PNs before and after application of 10 µM CNO (n = 13, error bars represent SEM). (D) Application of the P/Q-Type voltage-activated calcium channel blocker ω-agatoxin TK results in nearly complete loss of feedforward inhibition recruited by light activation of SuM inputs in area CA2. D1, sample traces, (top, control in red, ω-agatoxin TK in gray) and summary graph of light-evoked IPSC amplitudes recorded in CA2 PNs before and after application of 200 nM ω-agatoxin TK (bottom, n = 5, error bars represent SEM). D2, sample traces (top, SR95531 and CGP55845A in black, ω-agatoxin TK in gray) and summary graph of light-evoked EPSC amplitudes before and after application of 200 nM ω-agatoxin TK (bottom, n = 6, error bars represent SEM). (E) Application of the mu-opioid receptor agonist, DAMGO, results in the complete abolition of light-evoked SuM inhibitory transmission. E1, sample traces (top, control in red, DAMGO in gray) and summary graph of light-evoked IPSC amplitudes recorded in CA2 PNs before and after application of 1 µM DAMGO (bottom, n = 6, error bars represent SEM). E2, sample traces (top, SR95531 and CGP55845A in black, DAMGO in gray) and summary graph of light-evoked EPSC amplitudes before and after application of 1 µM DAMGO (bottom, n = 17, error bars represent SEM). (F) Application of the delta-opioid receptor agonist, DPDPE, results in the long-term depression of light-evoked SuM inhibitory transmission. F1, sample traces (top, control in red, DPDPE in gray) and summary graph of light-evoked IPSC amplitudes before and after application of 0.5 µM DPDPE (bottom, n = 7, error bars represent SEM). F2, sample traces (top, SR95531 and CGP55845A in black, DAMGO in gray) and summary graph of light-evoked EPSC amplitudes before and after application of 0.5 µM DPDPE (bottom, n = 7, error bars represent SEM).

We used a pharmacological strategy to selectively manipulate PV+ INs by targeting their GABA release machinery. In the neocortex, P/Q-type voltage-gated calcium channels are necessary for GABA release from PV+ fast spiking INs onto PNs (Zaitsev et al., 2007). In contrast, N-type calcium channels are primarily involved in GABA release from CCK+ INs (Hefft and Jonas, 2005). Thus, we recorded SuM input-evoked EPSCs and IPSCs in CA2 PNs before and after application of the P/Q-type voltage-gated calcium channels specific blocker ω-agatoxin TK (200 nM) (Figure 5D). We observed a near-complete block of IPSCs upon ω-agatoxin TK application (Figure 5D 1, IPSC amplitudes were 245.5 ± 92.6 pA in control and 35.0 ± 15.4 pA in ω-agatoxin TK hence a 81.8 ± 3.9% block, n = 5; paired-T test, p<0.001), suggesting a major contribution from PV+ INs to SuM-driven feedforward inhibition consistent with our previous results. However, we observed that excitatory transmission from SuM axons was also partially blocked by ω-agatoxin TK application, as SuM input-evoked EPSCs were significantly reduced although not abolished (Figure 5D 2, EPSC amplitudes were −51.8 ± 5.9 pA in SR95531 and CGP55845A and −26.5 ± 5.4 pA after ω-agatoxin TK hence a 49.6 ± 5.6% block, n = 6; paired-T test, p<0.001). This observation indicates that glutamate release from SuM axons relies on P/Q-type voltage-gated calcium channels to some degree, thereby complicating the interpretation of the reduction of IPSC amplitude in CA2 PNs.

It has previously been demonstrated that PV+ BC transmission can be strongly attenuated by mu opioid receptor activation (MOR) while CCK+ BC transmission is insensitive to MOR activation (Glickfeld et al., 2008). Thus, we recorded from PNs in area CA2 and examined the sensitivity of light-evoked IPSCs to the application of the MOR agonist DAMGO (Figure 5E). We found that there was a near complete block of the light-evoked IPSC amplitude following 1 μM DAMGO application (Figure 5E 1; IPSC amplitudes were 343 ± 123 pA in control and 31 ± 12.4 pA in DAMGO hence a 88 ± 5.0% block, n = 6 PNs; Wilcoxon signed-rank test, p=0.031), while direct excitatory transmission remained unaffected (Figure 5E 2; EPSC amplitudes were −6.7 ± 1.1 pA in SR95531 and CGP55845A and −5.6 ± 0.9 pA after DAMGO, n = 17 PNs; Wilcoxon signed-rank test, p=0.19). While this result is in agreement with our DREADD and ω-agatoxin TK results showing a major contribution of PV +INs to the SuM-driven feedforward inhibition, it should be noted that the dichotomy between PV+ versus CCK+ INs sensitivity to DAMGO has not been directly verified in area CA2.

It has recently been shown that delta opioid receptors (DORs) are specifically expressed in a fraction of PV+ cells in the hippocampus (Erbs et al., 2012). Furthermore, PV+ INs in area CA2 are the substrate of an iLTD of feedforward inhibition from CA3 mediated by delta opioid receptor (DOR) activation (Nasrallah et al., 2019; Piskorowski and Chevaleyre, 2013). Therefore, we sought to further refine our characterization of the SuM feedforward inhibition by assessing its sensitivity to DOR activation (Figure 5F). Application of 0.5 µM of the DOR agonist DPDPE led to a long-term reduction of light-evoked IPSCs recorded in area CA2 PNs, similar to the iLTD seen by CA3 input stimulation (Figure 5F 1; IPSC amplitudes were 168 ± 28 pA in control and 64 ± 22 pA in DPDPE hence a 61 ± 14% block by DPDPE, n = 7; paired-T test, p=0.015), while leaving direct EPSCs unaffected (Figure 5F 2; EPSC amplitudes were −4.0 ± 1.6 pA in SR95531 and CGP55845A and −3.1 ± 1.1 pA after DPDPE, n = 7; Wilcoxon signed-rank test, p=0.22). Further confirming the PV+ nature of INs responsible for the SuM feedforward inhibition, this result reveals that both the local CA3 and long-range SuM inputs converge onto an overlapping population of INs to inhibit area CA2 PNs, thus enabling cross-talk between these routes through synaptic plasticity of PV+ INs. However, since DORs are only expressed in a fraction of PV+ INs and therefore only reduces but does not fully block PV+ IN-mediated GABA release (Nasrallah et al., 2019; Piskorowski and Chevaleyre, 2013), it is difficult to know if the remaining SuM-evoked IPSCs are from PV+ INs not expressing DOR or from other INs recruited by the SuM input.

Altogether, these four methods strongly suggest that SuM inputs selectively recruit PV +interneurons to inhibit CA2 PNs. Although individually each method does not conclusively demonstrate that SuM input exclusively targets PV+ INs, the consistent reduction of SuM-driven feedforward inhibition of CA2 PNs observed with every approach allows us to conclude that PV + cells are predominantly targeted by SuM inputs in area CA2.

The feedforward inhibitory drive from SuM controls pyramidal neuron excitability

Given SuM axonal stimulation triggers an excitatory-inhibitory sequence in post-synaptic PNs, we asked which effect would prevail on PN excitability. In order to assess this, we mimicked an active state in PNs by injecting constant depolarizing current steps sufficient to sustain AP firing during 1 s while photostimulating SuM axons at 10 Hz (Figure 6A–B). We observed that recruitment of SuM inputs significantly delayed the onset of the first AP (Figure 6C; latency to the first AP were 221 ± 19.9 ms in control and 233 ± 19.1 ms with photostimulation, hence a 12.1 ± 4.3 ms increase upon photostimulation, n = 12; paired-T test, p=0.016). In addition, given SuM neurons display theta-locked firing in vivo, we asked if rhythmic inhibition driven by SuM inputs in area CA2 could pace AP firing in PNs by defining windows of excitability. Indeed, photostimulation of SuM axons at 10 Hz led to a significant decrease of variability in the timing of AP firing by PNs (Figure 6D–E; standard deviations of the first AP timing were 36.9 ± 11 ms in control and 24.7 ± 7.4 ms with photostimulation, hence a 12.3 ± 5.3 ms decrease upon photostimulation, n = 12; Wilcoxon signed-rank tests, p<0.001 for the first AP, p=0.008 for the second AP, p=0.004 for the third AP). Both the delay of AP onset and the reduction of AP jitter stemmed from the feedforward inhibition recruited by SuM inputs as application of GABAA and GABAB receptor antagonists abolished these effects of SuM stimulation (Figure 6C–E; latency to the first AP were 232 ± 19.8 ms in SR95531 and CGP55845A and 235 ± 18.0 ms with photostimulation, n = 6; Wilcoxon signed-rank test, p=0.44; standard deviations of the first AP timing were 11.9 ± 2.0 ms in SR95531 and CGP55845A and 7.1 ± 1.5 ms with photostimulation, n = 6; Wilcoxon signed-rank tests, p=0.22 for the first AP, p=0.16 for the second AP, p=0.09 for the third AP). These results reveal that the purely glutamatergic SuM input, by recruiting feedforward inhibition, has an overall inhibitory effect on PN excitability and can influence the timing and jitter of area CA2 PN action potential firing.

Area CA2 PNs receive a net inhibitory drive from SuM that controls AP firing properties.

(A) Diagram illustrating whole-cell recordings of area CA2 PNs and SuM fiber light stimulation in acute slice preparation. (B) Example traces of a CA2 PN action potential firing in response to current injection in the absence (black traces) or presence of 10 Hz photostimulation of SuM inputs (red traces). (C) Action potential onset latency is increased with 10 Hz SuM input photostimulation. Left, sample traces of the first AP in control and with inhibition blocked by 1 µM SR95531 and 2 µM CGP55845A application (light-off in black, light-on in red, light-off in SR95531 and CGP55845A in gray, light-on in SR95531 and CGP55845A in purple). Right, summary graph of photostimulation-induced delay of AP firing in area CA2 PNs before and after application of SR95531 and CGP55845A (control shown in red, n = 12, paired-T test, p=0.016; SR95531 and CGP55845A shown in purple, n = 6; Wilcoxon signed-rank test, p=0.44; individual cells shown with dots, boxplot represents median, quartiles, 10th and 90th percentiles). (D) Sample traces of AP firing in repeated trials (light-off in black, light-on in red, light-on in SR95531 and CGP55845A in purple; during experiment photostimulation was interleaved with control but traces are grouped here for demonstration purposes). (E) AP jitter in CA2 PNs is reduced by activation of SuM inputs. Left, summary graph of the standard deviation of AP firing with or without 10 Hz photostimulation (n = 12; Wilcoxon signed-rank test, p<0.001 for the first AP, p=0.008 for the second AP, p=0.004 for the third AP; individual cells shown with thin lines, population average shown as thick line, error bars represent SEM). Right, photostimulation-induced reduction of AP firing standard deviation in control and in SR95531 and CGP55845A (control, n = 12; Wilcoxon signed-rank tests, p<0.001 for the first AP, p=0.008 for the second AP, p=0.004 for the third AP; SR95531 and CGP55845A, n = 6; Wilcoxon signed-rank tests, p=0.22 for the first AP, p=0.16 for the second AP, p=0.09 for the third AP; individual cells shown with dots, boxplot represents median, quartiles, 10th and 90th percentiles).

One drawback of these results is that the injection of current steps to evoke action potential firing is not entirely representative of CA2 PN activity, as there is no synaptic input leading to AP firing. It has been reported that the AP discharge of SuM neurons in vivo is phase-locked to the hippocampal theta rhythm (Kocsis and Vertes, 1994). Because theta rhythm is a brain state characterized by elevated levels of acetylcholine, we approximately mimicked these conditions in the hippocampal slice preparation by bath application of 10 µM of the cholinergic agonist carbachol (CCh). Under these conditions, CA2 PNs depolarize and spontaneously fire rhythmic bursts of APs, and the properties of these AP bursts are tightly controlled by excitatory and inhibitory synaptic transmission (Robert et al., 2020). Thus, we decided to examine how SuM input stimulation influenced CA2 PN firing under these conditions.

Because SuM neurons fire in bursts at theta frequency in vivo (Kirk et al., 1996), and because the elevated cholinergic tone accompanying theta can activate muscarinic receptors that alter the synaptic release properties of many synapses in the brain, we examined how synaptic transmission from the SuM input to area CA2 was affected by the application of 10 µM carbachol (CCh) (Figure 7—figure supplement 1A; Kirk et al., 1996; Kocsis and Vertes, 1994). With GABA receptors blocked to first assess the SuM excitatory transmission only, we observed that CCh decreased the amplitude and increased the PPR of SuM-evoked EPSCs in CA2 PNs (Figure 7—figure supplement 1B). This suggests a decrease of glutamate release by SuM axons induced by CCh. We found similar results for SuM-evoked feedforward inhibitory transmission to CA2 PNs as IPSC amplitude was decreased and PPR increased with CCh application (Figure 7—figure supplement 1C). Next, we examined the relative short-term dynamics of SuM-evoked excitatory and inhibitory transmission to CA2 PNs. For this, both EPSCs and IPSCs were recorded from the same individual CA2 PNs upon repeated SuM input stimulation with five pulses delivered at 10 Hz before and after CCh application (Figure 7—figure supplement 1D–G). We observed that both SuM-evoked EPSCs and IPSCs underwent short-term depression, as evidenced by a decrease in amplitude along the pulse train as well as amplitude ratios between subsequent pulses over the first pulse (Pn/P1) (Figure 7—figure supplement 1D–F, Figure 7—figure supplement 1—source data 1). It is worth noting that the Pn/P1 ratio was similar for EPSCs and IPSCs and that the E/I ratio did not significantly change with repeated SuM input stimulation (Figure 7—figure supplement 1F–G, Figure 7—figure supplement 1—source data 1). This indicates that the SuM influence over CA2 PN may remain overall inhibitory during prolonged SuM input activation. Similarly influencing both EPSCs and IPSCs, application of 10 µM CCh affected these short-term dynamics of the SuM-CA2 PN transmission by decreasing the amplitude of the initial response (Figure 7—figure supplement 1D–E, Figure 7—figure supplement 1—source data 1) but limiting the subsequent short-term depression of SuM-evoked PSCs amplitude (Figure 7—figure supplement 1D–F, Figure 7—figure supplement 1—source data 1). Interestingly, the overall effect of repeated SuM input stimulation on post-synaptic responses in area CA2 PNs was even more biased toward inhibition after CCh application as the E/I ratio of PSCs during the pulse train was lower in CCh as compared to control (Figure 7—figure supplement 1G, Figure 7—figure supplement 1—source data 1), possibly because of a lesser depression of IPSCs as compared to EPSCs (Figure 7—figure supplement 1D–F, Figure 7—figure supplement 1—source data 1) which could be due to a CCh-induced depolarization of INs mediating SuM-evoked feedforward inhibition. Altogether, these observations match with our findings of the SuM input having an overall inhibitory influence over area CA2, and suggest that this effect might be more gradual over time but even stronger in conditions of elevated cholinergic tone.

Under these conditions of elevated cholinergic tone, we asked how the spontaneous AP bursting activity of CA2 PNs would be affected by activation of the SuM input by triggering 10 s-long trains of 0.5 ms light pulses delivered at 10 Hz to stimulate SuM axons at the onset of bursts (Figure 7A). Because of the intrinsic cell-to-cell variability of bursting kinetics, we photostimulated SuM inputs only during interleaved bursts in the same cells. To do this, bursts were detected automatically with an online threshold detection system that started the photostimulation pulse train after the first AP of every alternating burst, starting with the second burst (Figure 7A–B). For analysis, the number of APs and bursting kinetics could be compared within the same cell. We observed a significant decrease in the number of APs fired during a burst when SuM inputs were photostimulated as compared to interleaved control bursts (Figure 7C–D; numbers of APs per burst were 15.2 ± 2.3 in control and 6.9 ± 1.3 with photostimulation, n = 7; paired-T test, p=0.031). In control bursts, the AP firing rate of CA2 PNs initially increases, and then progressively decreases. In the photostimulation bursts, the initial increase of AP firing frequency was absent, and the subsequent AP firing frequency was reduced (Figure 7E; two-way ANOVA on firing rate over time in light-on vs light-off conditions; light factor, p<0.001; time factor, p<0.001; light x time factor, p=0.052).

Figure 7 with 1 supplement see all
SuM input shapes CA2 PN AP bursts in conditions of elevated cholinergic tone.

(A) Diagram illustrating whole-cell recordings of area CA2 PNs with light stimulation of SuM fibers in an acute slice preparation. (B) Sample trace of spontaneous AP bursting activity recorded from a CA2 PN during bath application of 10 µM CCh. For every even-numbered burst, a 10 Hz photostimulation (blue bars) was delivered to excite SuM inputs in area CA2 allowing a comparison of burst AP firing in the same cell. (C) Sample traces of AP firing during bursts for light-off (left, black) and light-on (right, red) epochs. (D) Comparison of AP number per burst for light-off (black) and light-on (red) events (n = 7; individual cells shown as thin lines, population average shown as thick line, error bars represent SEM; paired-T test, p=0.031). (E) Average firing rate during spontaneous burst events with SuM photostimulation (red, light-on) and controlled interleaved burst events (black, light-off). Shaded area represents SEM for seven cells each with between 3 and 13 bursts analyzed in light-on and light-off conditions (two-way ANOVA, light factor: p<0.001, time factor: p<0.001, light x time factor: p=0.052). (F) Example burst events with (red) and without (black) SuM photostimulation overlayed and on a scale that shows the rapidly hyperpolarizing membrane potential that occurs with SuM input stimulation. (G) Comparison of bursts duration for events with (red) and without (black) photostimulation (n = 7; individual cells shown as thin lines, population average shown as thick line, error bars represent SEM; paired-T test, p=0.037). (H) Comparison of time elapsed to next burst onset following bursts with (red) or without (black) photostimulation (n = 7; individual cells shown as thin lines, population average shown as thick line, error bars represent SEM; paired-T test, p=0.001).

In the presence of CCh, CA2 PNs undergo a depolarization of the membrane potential that is followed by a period of AP firing as the membrane potential remains depolarized for several seconds, and then slowly hyperpolarizes until the next bursting event (Robert et al., 2020). We observed that photostimulation of SuM inputs resulted in a striking reduction in the amount of time the membrane potential remained depolarized, and this is likely why the burst duration was significantly shorter in bursts with SuM photo-stimulation (Figure 7F–G; burst duration was 4.0 ± 1.1 s in control and 1.6 ± 0.5 s with photostimulation, n = 7; paired-T test, p=0.037). The rate and level of VM repolarization following bursts were not significantly changed by SuM input photostimulation (VM repolarization rate was −3.3 ± 0.6 mV/s in control and −3.6 ± 0.7 mV/s with photostimulation, n = 7; paired-T test, p=0.601; post-burst VM was −62.8 ± 1.7 mV in control and −62.0 ± 2.0 mV with photostimulation, n = 7; paired-T test, p=0.173); however, the inter-burst time interval was reduced. Indeed, AP bursts with SuM input activation were followed more rapidly by another burst of APs than the ones without SuM input activation (Figure 7B and H; time until next burst was 93 ± 14 s in control and 59 ± 17 s with photostimulation, n = 7; paired-T test, p=0.001), which could be due to both short-term depression of inhibitory transmission after repeated activation during the SuM input photostimulation train and reduced activation of hyperpolarizing and shunting conductances during bursts shortened by SuM input photostimulation. Thus, in our preparation, SuM input activation is able to modify the spontaneous bursting activity of CA2 PNs under conditions of high cholinergic tone.

As SuM input controls burst firing of action potentials and likely paces activity in area CA2, we wondered how the subsequent output of CA2 PNs would affect their post-synaptic targets. Because CA2 PNs strongly project to CA1 PNs, this activity is likely to influence CA1 encoding and hippocampal output. Thus, we examined the consequences of SuM-CA2 input stimulation on area CA1 both in vivo and in acute slices treated with CCh to induce spontaneous activity (Figure 8).

Consequences of SuM input on area CA2 output to CA1.

(A) Diagram illustrating in vivo recording in CA1 with tetrodes and SuM axon terminals stimulation over CA2 with an implanted optical fiber. (B) Representative data from four multi-unit recordings. Raster plot (top) showing CA1 AP firing activity before and during photostimulation of SuM fibers in area CA2. The corresponding firing rate histogram (middle) of four tetrodes placed in the CA1 pyramidal cell layers, as well as plots of standard deviation (SD; bottom). Red lines indicate ±3 SD. (C) Individual (gray) and average (red) normalized firing rates from 34 multiunit recordings, three consecutive light stimulation epochs are displayed to help visualizing the consistency of the effect of SuM input light stimulation over area CA2 on CA1 multi-unit firing; the shaded area represents the SEM. (D) Diagram illustrating whole-cell recordings of area CA1 PNs and SuM fiber light stimulation over area CA2 in acute slice preparation. (E-H) Example waterfall plots (E, G) and corresponding peri-stimulus time histogram (F, H) (population average shown as thick line, shaded area represents SEM) of EPSCs (black) and IPSCs (red) recorded from a CA1 PN ex vivo during photostimulation of SuM input over area CA2 with bath application of 10 µM CCh.

ChR2-EYFP was expressed in the SuM of Csf2rb2-Cre mice in a Cre-dependent manner and the mice were implanted with a microdrive targeting tetrodes to region CA1 and an optical fiber to the SuM terminals in CA2 (Figure 8A). Mice were placed in a small box (familiar context) and left free to explore as blue (473 nm) laser light pulses (50 ms pulse width) were applied to the SuM terminals at 10 Hz. Across 23 recording sessions in five mice, we found that the activation of SuM terminals in CA2 resulted in a significant and reproducible change in the multiunit spiking activity recorded in the pyramidal cell layer of CA1 on 34 of 55 tetrodes. The firing rate change was similar across individual tetrodes (Figure 8B–C), with a decrease in the normalized firing rate starting shortly after laser onset and continuing for about 10 ms, followed immediately by a rebound-like increase to about 20% greater than baseline firing rate (Figure 8B–C).

In order to get a better mechanistic understanding of this observation, we set out to decipher how SuM activity in area CA2 influences CA1 in the hippocampal slice preparation. To this end, we used the same photostimulation protocol used in vivo that consisted of light stimulation trains of 50 ms-long pulses delivered at 10 Hz for 1 s, repeated every 10 s for 2 min and interleaved with light-off sweeps of the same duration, with the microscope objective centered on area CA2. Whole-cell patch-clamp recordings of CA1 PNs were obtained in acute hippocampal slices superfused with CCh and subjected to this light stimulation protocol (Figure 8D). We asked what synaptic events may be responsible for the decreased firing of CA1 units observed 10–20 ms after light onset in vivo (Figure 8A–C). Whole-cell recordings of CA1 PNs showed an absence of EPSCs time-locked to the photostimulation in all but one case (n = 11/12) (Figure 8E–F). In contrast, we often (n = 7/12) observed light-evoked IPSCs in CA1 PNs occurring 10–20 ms after light onset (Figure 8G–H). Therefore, the reduction in firing of CA1 units in vivo is likely caused by increased inhibitory inputs onto CA1 PNs within 10–20 ms of SuM fiber stimulation over area CA2. This result highlights a contribution of SuM input to controlling CA2 output that regulate CA1 activity in vivo and provides a mechanistic interpretation of this observation at the circuit level.

Discussion

In this study, we provide direct evidence for a functional connection between the hypothalamus and the hippocampus. Using stereotaxic injection of viral vectors in combination with transgenic mouse lines to express channelrhodopsin in a projection-specific manner, we have been able to selectively stimulate SuM axons in area CA2 of the hippocampus, allowing for the direct examination of synaptic transmission. This approach yielded novel functional information about the SuM input post-synaptic targets and the overall physiological consequences of its activation. We found that, in contrast to previous anatomical reports, SuM input forms synapses onto both PNs and INs in area CA2. The excitatory drive evoked by light-stimulation of SuM input was significantly larger for BC INs, which we demonstrate are likely PV+. The resulting feedforward inhibition recruited by SuM input stimulation enhanced the precision of AP timing of CA2 PNs in conditions of low and high cholinergic tone relevant to different brain states. The modified CA2 output evoked poly-synaptic inhibition in area CA1, likely responsible for a decrease in firing rate of CA1 units in vivo. Overall, we demonstrate that SuM input controls CA2 output to area CA1 by recruiting feedforward inhibition.

SuM input to area CA2 forms a microcircuit where PV+ basket cells strongly inhibit pyramidal neurons

Glutamatergic innervation of area CA2 by the SuM has been previously described by tracing studies (Kiss et al., 2000; Soussi et al., 2010) and presumed to form synapses exclusively onto PNs (Maglóczky et al., 1994). Our experimental strategy allowed for the direct examination of the post-synaptic targets of SuM glutamatergic axons. Our results confirm that PNs in area CA2 indeed receive excitatory synapses from SuM axons. However, in contrast to what had been proposed in previous studies, we observed that SuM inputs target not only PNs but also INs in area CA2. Importantly, we identified a specific subpopulation of INs as BCs which were the cell type most potently excited by SuM. These BCs could fire action potentials upon SuM input photostimulation leading to a substantial feedforward inhibition of neighboring PNs. We found that at resting membrane potentials, the mixed excitatory/inhibitory SuM input resulted in a net depolarization of the membrane potential in CA2 PNs. However, when the SuM input was paired with either inputs in SR or SLM, we observed a decrease in the summation ratios of trains of synaptic inputs consistent with a perisomatic shunting inhibition driven by SuM in area CA2. Furthermore, we found that with elevated cholinergic tone, recruitment of BCs by SuM controlled PNs excitability and shaped spontaneous burst firing. This finding demonstrates that SuM activity can pace action potential firing in PNs through recruitment of feedforward inhibition.

The population of INs potently excited by SuM transmission display many features that motivate us to classify them as PV+ BCs. They have somas located in the somatic layer, have densely packed perisomatic-targeted axons, are fast spiking and show PV immuno-reactivity. The selective expression of GiDREADD in PV+ cells allows for selective silencing that reduces SuM-driven feedforward inhibition of area CA2 PNs. With these techniques, however, we were unable to sufficiently silent PV+ cells in area CA2, leaving open the possibility that another population of basket cell is targeted by SuM input. The feedforward inhibitory transmission recruited by SuM stimulation is highly sensitive to MOR activation. While this supports our hypothesis that PV+ cells are targeted by SuM input, MORs are not entirely exclusive to PV+ cells (Stumm et al., 2004). We also show that the SuM-recruited feedforward inhibition is sensitive to DOR activation. Unlike MORs, DORs have been shown to be specific to PV+ cells in area CA2, however, only a sub-population of PV +INs express this receptor (Nasrallah et al., 2019; Piskorowski and Chevaleyre, 2013) leaving open the possibility that the remaining IPSCs evoked by SuM stimulation are not from PV+ cells. We also show in this work that SuM-evoked inhibitory currents are blocked by the application of ω-agatoxin TK, indicating that these recruited INs express P/Q-type CaV channels, consistent with PV+ BCs (Zaitsev et al., 2007). However, we also saw that ω-agatoxin TK also blocked gluatamatergic transmission from SuM inputs, preventing a simple interpretation of these results. Thus, while there is ample evidence that SuM inputs target PV+ BCs in area CA2, we cannot exclude the possibility that other populations of BCs, such as CCK+ INs are also targeted by these inputs. PV+ BCs in the hippocampus have been shown to be modulated by CCK (Lee et al., 2011) which would have very interesting implications for the effect of SuM activity in area CA2. Furthermore, it was recently shown that PV+ BCs actively inhibit CCK+ BCs, enabling a complementary perisomatic inhibitory system that allows for brain-state-dependent activity during behavior (Dudok et al., 2021).

Recent studies have indicated that the SuM input to CA2 plays a key role in social novelty discrimination (Chen et al., 2020). Our findings are very consistent with the finding that DOR-mediated inhibitory synaptic plasticity of PV+ INs in area CA2 is required for social recognition memory (Domínguez et al., 2019). Furthermore, exposure to a novel conspecific induces a DOR-mediated plasticity in this same inhibitory network in area CA2 (Leroy et al., 2017). Thus, our finding that SuM input acts via PV+ interneurons fits with previous studies and provides a link between social novelty detection and local CA2 hippocampal inhibitory plasticity.

Overall, the local circuitry and consequences of SuM input to area CA2 contrasts with the SuM-DG path (Hashimotodani et al., 2018; Li et al., 2020; Mizumori et al., 1989; Nakanishi et al., 2001). Previously, we have shown that unlike the SuM-DG synapse, the SuM-CA2 synapse is entirely glutamatergic (Chen et al., 2020). In this study, we use both a VGluT2-Cre and SuM-Cre mouse lines to demonstrate how the combination of direct excitation and feedforward inhibition regulates CA2 PN AP firing. Our data shows that SuM activity results in synchronized feedforward inhibition from CA2 to CA1 which decreases CA1 PN firing. While our results are very intriguing given the importance of area CA2 in propagation of hippocampal network activity (Oliva et al., 2016a), further questions remain. CA2 PNs also receive excitatory input from DG cells via the mossy fibers (Kohara et al., 2014; Llorens-Martín et al., 2015). It has been postulated that by increasing DG excitability, the SuM may also be indirectly acting on CA2 (Silkis and Markevich, 2020). These circuits merit further exploration.

Consequences of SuM input on area CA2 output

Recent work has demonstrated a strong excitatory drive from area CA2 to CA1 (Chevaleyre and Siegelbaum, 2010; Kohara et al., 2014; Nasrallah et al., 2019). Consequently, modification of CA2 output through synaptic plasticity (Nasrallah et al., 2019) or neuromodulation (Tirko et al., 2018) affects CA1 activity. This observation is critical when considering social memory formation, which is known to depend on CA2 output (Hitti and Siegelbaum, 2014; Stevenson and Caldwell, 2014) and is likely encoded in downstream ventral CA1 (Okuyama et al., 2016). CA2-targeting cells in the SuM have recently been shown to be highly active during novel social exploration (Chen et al., 2020). From our results, we hypothesize that this novel social signal from the SuM, acts via the PV +inhibitory network in area CA2 to control the timing of CA2 output onto area CA1.

By recruiting feedforward inhibition, SuM activity paces and temporally constrains AP firing from CA2 PNs undergoing depolarization. More critically, in conditions of elevated cholinergic tone relevant to SuM activity in vivo, CA2 PNs depolarize and fire bursts of APs that can be shaped by SuM input both by controlling AP firing as well as membrane depolarization. While this result was obtained by triggering SuM input stimulation to the onset of burst firing by CA2 PNs, in vivo and acute slice experiments revealed a consistent influence of CA1 PN AP firing by SuM input to area CA2 regardless of the timing of SuM input stimulation relative to CA2 PN AP burst firing. These results demonstrate a powerful control of SuM input over CA2 output when PNs are spontaneously firing bursts of APs, a firing mode that is most efficient at influencing CA1 activity (Tirko et al., 2018). Optogenetic experiments have recently shown that CA2 PNs can drive a strong feedforward inhibition in area CA1 (Kohara et al., 2014; Nasrallah et al., 2019). Although SuM input likely does not directly drive feedforward inhibition in area CA1 (Chen et al., 2020), the recruitment of feedforward inhibition in area CA2 by SuM input activation could curtail the time window of spontaneous firing in CA2 PNs and effectively lead to a synchronized drive of feedforward inhibition by area CA2 over area CA1. We postulate that the concerted IPSC that we detect in area CA1 with SuM fiber photostimulation in area CA2 corresponds to the large decrease in firing that is observed in CA1 multi-unit recordings in vivo. Thus, these data provide evidence for a long-range control of CA2 bursting activity and the consequences in downstream area CA1 in conditions of high cholinergic tone that accompanies theta oscillations in vivo during which SuM is active.

Relevance of the SuM input to area CA2 for hippocampal oscillations

The activity of hippocampal neurons is orchestrated by brain rhythms, notably theta and gamma oscillations that are prominent during exploration and linked to the learning and memory functions of the hippocampus (Buzsáki, 2002; Buzsáki and Wang, 2012; Colgin, 2016). The SuM is active during these brain states and contributes to theta oscillations in the hippocampus (Kirk et al., 1996; Kirk and McNaughton, 1993; Kocsis and Vertes, 1994; McNaughton et al., 1995; Pan and McNaughton, 2002; Pan and McNaughton, 1997; Thinschmidt et al., 1995). Here, we show that the SuM controls area CA2 output to CA1 by recruiting PV +BCs, which are important for both theta and gamma oscillations (Fuchs et al., 2007; Gulyás et al., 2010; Korotkova et al., 2010; Mann and Mody, 2010). Through its perisomatic mono-synaptic excitation and PV +BC-mediated di-synaptic inhibition of CA2 PNs, the SuM likely contributes to enforcing theta-locked windows of excitability shaping CA2 PNs output. Area CA2 can influence CA1 activity not only by direct projections but also through its interactions with both CA3 (Boehringer et al., 2017) and EC (Chevaleyre and Siegelbaum, 2010; Rowland et al., 2013) which are major contributors to CA1 theta and gamma oscillations (Buzsáki, 2002; Colgin, 2016). CA2 axons target both CA1 stratum oriens and radiatum (Nasrallah et al., 2019), thus the CA2 projections to CA1 likely contributes to the theta and slow gamma oscillations observed in these strata in CA1 (Belluscio et al., 2012; Colgin et al., 2009; Schomburg et al., 2014). Indeed, CA2 PNs show theta- and gamma-modulation of their activity (Fernandez-Lamo et al., 2019; Oliva et al., 2016b), and chemogenetic manipulations of their excitability bidirectionally influences hippocampal low gamma power (Alexander et al., 2018). Further, chronic block of CA2 output transmission leads to hippocampal hyperexcitability and disrupts CA1 theta phase preference and spatial coding (Boehringer et al., 2017). Therefore, by providing a theta-locked input shaping CA2 PN activity, the SuM is poised to contribute to oscillatory activity in downstream brain regions receiving CA2 input. Indeed, chemogenetic activation or silencing of SuM glutamatergic neurons respectively increases or decreases theta and gamma power in the EEG (Pedersen et al., 2017). Further, the SuM is involved in coordinating activity between the prefrontal cortex, the thalamus and area CA1 as evidenced by a loss of theta coherence amongst these regions upon SuM optogenetic silencing during a spatial task requiring action planning (Ito et al., 2018). Altogether, these studies point to the SuM as a crucial component in the regulation of hippocampal oscillations and our findings shed light on an aspect of this circuit.

Gating of area CA2 activity by PV +INs and significance for pathologies

The density of PV +INs in area CA2 is strikingly higher than in neighboring areas CA3 and CA1 (Botcher et al., 2014; Piskorowski and Chevaleyre, 2013). This population of INs has been shown to play a powerful role in controlling the activation of CA2 PNs by CA3 inputs (Nasrallah et al., 2015). We show in this study that long-range inputs from the SuM can strongly recruit PV +BCs, which in turn inhibit PNs in this area. Hence, both intra-hippocampal inputs from CA3 and long-range inputs from the SuM converge onto PV +INs to control CA2 PN excitability and output.

Postmortem studies have reported losses of PV +INs in area CA2 in pathological contexts including bipolar disorder (Benes et al., 1998), Alzheimer’s disease (Brady and Mufson, 1997), and schizophrenia (Benes et al., 1998; Knable et al., 2004). Consistent with these reports, in a mouse model of the 22q11.2 deletion syndrome, a major risk factor for schizophrenia in humans, we found a loss of PV staining and deficit of inhibitory transmission in area CA2 that were accompanied by impairments in social memory (Piskorowski et al., 2016). We postulate that the PV +INs altered during pathological conditions may be the same population of PV +BCs recruited by long-range SuM inputs. Indeed, the DOR-mediated plasticity onto PV +INs is altered in the 22q11.2 deletion syndrome mouse model, and we show here that a fraction of the PV +INs targeted by the SuM also express DOR. Thus, the loss of function of PV +INs in area CA2 could disrupt proper long-range connection between the hippocampus and the hypothalamus and possibly contribute to some of the cognitive impairments observed in schizophrenic patients and animal models. Further, pharmacological mouse models of schizophrenia have reported increased c-fos immunoreactivity in the SuM as well as memory impairments (Castañé et al., 2015). Although several alterations in these models of schizophrenia could lead to deficits of hippocampal-dependent behavior, abnormalities of the SuM projection onto area CA2 appear as a potential mechanism that warrants further investigation.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or
reference
IdentifiersAdditional
information
Genetic reagent (Mus. musculus)Tg(Slc17a-icre)10KiBorgius et al., 2010Tg(Slc17a-icre)10Ki; VGluT2-cre
Genetic reagent (Mus. musculus)Csf2rb2-CreChen et al., 2020Csf2rb2-Cre; SuM-cre
Genetic reagent (Mus. musculus)Pvalbtm1(cre)Arbr/J (PV-Cre)JacksonRRID:IMSR_JAX:017320
Genetic reagent (adeno-associated virus)AAV9.EF1a.DIO.hChR2(H134R).EYFPAddgeneRRID:Addgene 20298
Genetic reagent (adeno-associated virus)AAV9.hSynapsin.EGFP.WPRE.bGHAddgeneRRID:Addgene 51502
Genetic reagent (adeno-associated virus)AAV.Synapsin.DIO.hM4D(Gi).mCherryMcHugh Laboratory, Riken
Genetic reagent (adeno-associated virus)AAV2/9.hSyn.hChR2(H134R).EYFP.WPRE.hGHAddgeneRRID:Addgene 26973
Genetic reagent (Canine adeno virus)CAV2-crePlatforme de Vectorologie de MontpellierCAV Cre
AntibodyAnti-RGS14
(mouse monoclonal)
NeuroMab73–422
RRID:AB_2877596
(1:300)
AntibodyAnti-GFP
(chicken polyclonal)
Abcamab13970
RRID:AB_300798
(1:10,000)
AntibodyAnti-VGluT2 (guinea pig polyclonal)MilliporeAB2251
RRID:AB_1587626
(1:10000)
AntibodyAnti-parvalbumin (rabbit polyclonal)SwantPV27
RRID:AB_2631173
(1:2000)
AntibodyAnti- PCP4 (rabbit polyclonal)SigmaHPA005792
RRID:AB_1855086
(1:600)
AntibodyAnti-Calretinin
(mouse monoclonal)
MilliporeMAB1568
RRID:AB_94259
(1:500)
AntibodyAnti-mCherry
(rat monoclonal)
Life technologiesM11217
RRID:AB_2536611
(1:5000)
OtherFar-red neurotraceLife technologiesN21483
RRID:AB_2572212
(1:300)
Peptide, recombinant proteinAlexa-546-conjugated streptavidinLife TechnologiesS11225
RRID:AB_2532130
(1:500)
Peptide, recombinant proteinBiocytinHelloBioHB5035(4mg/mL)
Chemical compound, drugNBQXHelloBioHB0443(10 µM)
Chemical compound, drugD-APVHelloBioHB0225(50 µM)
Chemical compound, drugSR95531Tocris1262(1 µM)
Chemical compound, drugCGP55845ATocris1248(2 µM)
Chemical compound, drugDPDPEAlfa AesarJ66293(0.5 µM)
Chemical compound, drugDAMGOTocris1171(1 µM)
Chemical compound, drugClozapine N-oxide (CNO)HelloBioHB1807(10 µM)
Chemical compound, drugTetrodotoxin (TTX)Tocris1078(0.2 µM)
Chemical compound, drugCarbamoylcholine chloride (CCh)Tocris2810(10 µM)
Chemical compound, drugω-agatoxin TKAlomone labsSTA-530(200 nM)
Software, algorithmMatlabMathworkshttp://www.mathworks.com
RRID:SCR_001622
Software, algorithmIgor ProWavemetricshttp://www.wavemetrics.com
RRID:SCR_000325
RRID:SCR_004186 (neuromatic)
Software, algorithmOriginProOriginLab Corporationhttp://www.originlab.com
RRID:SCR_014212
Software, algorithmpClampMolecular Deviceshttp://www.moleculardevices.com
RRID:SCR_011323
Software, algorithmAxographAxographhttp://www.axograph.com
RRID:SCR_014284

All procedures involving animals were performed in accordance with institutional regulations (French Ministry of Research and Education protocol #12406–2016040417305913). Animal sample sizes were estimated using power tests with standard deviations and ANOVA values from pilot experiments. A 15% failure rate was assumed to account for stereotaxic injection errors and slice preparation complications. Every effort was made to reduce animal suffering.

Use of the Tg(Slc17a-icre)10Ki mouse line

Request a detailed protocol

We used the Tg(Slc17a-icre)10Ki mouse line that was previously generated (Borgius et al., 2010) and expresses the Cre recombinase under control of the slc17a6 gene coding the vesicular glutamate transporter isoform 2 (VGluT2).

Use of the Csf2rb2-Cre mouse line

Request a detailed protocol

 We used the Csf2rb2-Cre mouse line that was recently generated (Chen et al., 2020) and expresses the Cre recombinase under control of the Csf2rb2 gene that shows selective expression in the SuM.

Use of the Pvalbtm1(cre)Arbr/J mouse line

Request a detailed protocol

We used the Pvalbtm1(cre)Arbr/J mouse line that was previously generated (Hippenmeyer et al., 2005) and expresses the Cre recombinase under control of the Pvalbm gene coding parvalbumin (PV).

Stereotaxic viral injection

Request a detailed protocol

Animals were anaesthetized with ketamine (100 mg/kg) and xylazine (7 mg/kg). The adeno-associated viruses AAV9.EF1a.DIO.hChR2(H134R).EYFP and AAV9.hSynapsin.EGFP.WPRE.bGH were used at 3 × 108 vg, the AAV.Synapsin.DIO.hM4D(Gi).mCherry was used at 3.6 × 109 vg and the AAV2/9.hSyn.hChR2(H134R).EYFP.WPRE.hGH was used at 3.7 × 1013 vg. The retrograde tracer CAV2-cre virus was used at 2.5 × 1012 vg. 500 nL of virus was unilaterally injected into the brain of 4-week-old male wild type C57BL/6, Tg(Slc17ab-icre)10Ki (VGluT2-Cre), csf2rb2-cre (SuM-Cre) or Pvalbtm1(cre)Arbr/J (PV-Cre) mice at 100 nL/min and the injection cannula was left at the injection site for 10 min following infusion. In the case of AAV.Synapsin.DIO.hM4D(Gi)-mCherry injection in PV-Cre mice, bilateral injections were performed in dorsal CA2. The loci of the injection sites were as follows: anterior–posterior relative to bregma: −2.8 mm for SuM, −1.6 mm for CA2; medial-lateral relative to midline: 0 mm for SuM, 1.9 mm for CA2; dorsal-ventral relative to surface of the brain: 4.75 mm for SuM, 1.4 mm for CA2.

Electrophysiological recordings

Request a detailed protocol

Transverse hippocampal slices were prepared at least 3 weeks after viral injection and whole-cell patch-clamp recordings were performed from PNs and INs across the hippocampal CA regions. In the case of PV-Cre mice injected with AAV.Synapsin.DIO.hM4D(Gi)-mCherry, slices were prepared 6 weeks after viral injection. Animals were deeply anesthetized with ketamine (100 mg/kg) and xylazine (7 mg/kg), and perfused transcardially with a N-methyl-D-glucamin-based (NMDG) cutting solution containing the following (in mM): NMDG 93, KCl 2.5, NaH2PO4 1.25, NaHCO3 30, HEPES 20, glucose 25, thiourea 2, Na-ascorbate 5, Na-pyruvate 3, CaCl2 0.5, MgCl2 10. Brains were then rapidly removed, hippocampi were dissected out and placed upright into an agar mold and cut into 400-µm-thick transverse slices (Leica VT1200S) in the same cutting solution at 4°C. Slices were transferred to an immersed-type chamber and maintained in artificial cerebro-spinal fluid (ACSF) containing the following (in mM): NaCl 125, KCl 2.5, NaH2PO4 1.25, NaHCO3 26, glucose 10, Na-pyruvate 2, CaCl2 2, MgCl2 1. Slices were incubated at 32°C for approximately 20 min then maintained at room temperature for at least 45 min prior to patch-clamp recordings performed with either potassium- or cesium-based intracellular solutions containing the following (in mM): K- or Cs-methyl sulfonate 135, KCl 5, EGTA-KOH 0.1, HEPES 10, NaCl 2, MgATP 5, Na2GTP 0.4, Na2-phosphocreatine 10 and biocytin (4 mg/mL).

ChR2 was excited by 488 nm light delivered by a LED attached to the epifluorescence port of the microscope. Light stimulations trains consisted of 2–10 pulses, 0.5 ms long, delivered at 10 Hz, repeated every 20 s for at least 20 sweeps. Stimulating pipettes filled with ACSF were placed in stratum radiatum (SR) of CA1 to antidromically excite CA3-CA2 synapses and in stratum lacunosum moleculare (SLM) to stimulate distal dendritic inputs in area CA2. Synaptic currents were evoked with a constant voltage stimulating unit (Digitimer Ltd.) set at 0.1 ms at a voltage range of 5–10 V. For the patch-clamp recordings in area CA1 with stimulation of SuM axons in area CA2, 50 ms long light stimulation pulses were delivered every 10 s. We used a light intensity of 25 mW/mm² which was experimentally determined as the lowest irradiance allowing TTX-sensitive maximal responses in all cell types and conditions. Data were obtained using a Multiclamp 700B amplifier, sampled at 10 kHz and digitized using a Digidata. The pClamp10 software was used for data acquisition. Series resistance were <20 MOhm and were not compensated in voltage-clamp, bridge balance was applied in current-clamp. An experimentally determined liquid junction potential of approximately 9 mV was not corrected for. Pharmacological agents were added to ACSF at the following concentrations (in µM): 10 NBQX and 50 D-APV to block AMPA and NMDA receptors, 1 SR95531 and 2 CGP55845A to block GABAA and GABAB receptors, 1 DAMGO to activate µ-opioid receptors (MOR), 0.5 DPDPE to activate δ-opioid receptors (DOR), 10 clozapine N-oxide (CNO) to activate hM4D(Gi) DREADDs, 10 CCh to activate cholinergic receptors, 0.2 tetrodotoxin (TTX) to prevent sodic action potential generation, 200 nM ω-agatoxin TK to block P/Q-type voltage-gated calcium channels.

Surgery for in vivo recordings

Request a detailed protocol

All surgeries were performed in a stereotaxic frame (Narishige). Csf2rb2-cre male mice from 3 to 6 months of age were anaesthetized using 500 mg/kg Avertin. pAAV.DIO.hChR2(H134R).EYFP was injected into the SuM (−2.7 mm AP, +0.4 mm ML, −5.0 mm DV) using a 10 μL Hamilton microsyringe (701LT, Hamilton) with a beveled 33 gauge needle (NF33BL, World Precision Instruments (WPI)). A microsyringe pump (UMP3, WPI) with controller (Micro4, WPI) were used to set the speed of the injection (100 nl/min). The needle was slowly lowered to the target site and remained in place for 5 min prior to start of the injection and the needle was removed 10 min after infusion was complete. Following virus injection, a custom-built screw-driven microdrive containing six independently adjustable nichrome tetrodes (14 μm diameter), gold-plated to an impedance of 200–250 kΩ was implanted, with a subset of tetrodes targeting CA1, and an optic fiber (200 μm core diameter, NA = 0.22) targeting CA2 (−1.9 mm AP, ±2.2 mm ML, −1.6 mm DV). Following recovery, the tetrodes were slowly lowered over several days to CA1 pyramidal cell layer, identified by characteristic local field potential patterns (theta and sharp-wave ripples) and high amplitude multiunit activity. During the adjustment period the animal was habituated every day to a small box in which recording and stimulation were performed.

In vivo recording protocol

Request a detailed protocol

Recording was commenced following tetrodes reaching CA1. To examine the impact of SuM terminal stimulation in CA2 the mice were returned to the small familiar box and trains of 10 light pulses (473 nm, 10 mW/mm2 and pulse width 50 ms) were delivered to the CA2 at 10 Hz. The pulse train was repeated every 10 s for at least 20 times as the animals freely explored the box. Multiunit activity was recorded using a DigitalLynx 4SX recording system running Cheetah v.5.6.0 acquisition software (Neuralynx). Broadband signals from each tetrode were filtered between 600 and 6000 Hz and recorded continuously at 32 kHz. Recording sites were later verified histologically with electrolytic lesions as described above and the position of the optic fiber was also verified from the track.

In vivo data analysis

Request a detailed protocol

Spike and event timestamps corresponding to onset of each laser pulse were imported into Matlab (MathWorks) and spikes which occurred 50 ms before and 100 ms after each laser pulse were extracted. Raster plots were generated using a 1 ms bin size. Similar results were obtained using 5 ms and 10 ms bin size (data not shown). Firing rate histograms were calculated by dividing total number of spikes in each time bin by that bin’s duration. Each firing rate histogram was normalized by converting it into z-score values. Mean standard deviation values for the z-score calculation were taken from pre-laser pulse time period. To average the response across all mice, for each tetrode the firing rate in each bin was normalized to the average rate in the pre-laser period.

Immunochemistry and cell identification

Request a detailed protocol

Midbrains containing the injection site were examined post-hoc to ensure that infection was restricted to the SuM.

Post-hoc reconstruction of neuronal morphology and SuM axonal projections were performed on slices and midbrain tissue following overnight incubation in 4% paraformaldehyde in phosphate buffered saline (PBS). Midbrain sections were re-sliced sagittally to 100 µm thick sections. Slices were permeabilized with 0.2% triton in PBS and blocked overnight with 3% goat serum in PBS with 0.2% triton. Primary antibody (Life technologies) incubation was carried out in 3% goat serum in PBS overnight at 4°C. Channelrhodopsin-2 was detected by chicken primary antibody to GFP (Life technologies) (1:10,000 dilution) and an alexa488-conjugated goat-anti chick secondary. Other primary antibodies used were mouse anti-RGS14 (Neuromab) (1:300 dilution), rabbit anti- PCP4 (Sigma) (1:600 dilution), guinea pig anti-VGluT2 antibody (Milipore) (1:10,000 dilution), rabbit anti-parvalbumin antibody (Swant) (1:2000 dilution). Alexa-546-conjugated streptavidin (Life technologies), secondary antibodies and far-red neurotrace (Life technologies) incubations were carried out in block solution for 4 hr at room temperature. Images were collected with a Zeiss 710 laser-scanning confocal microscope.

Reconstructed neurons were classified as either PNs or INs based on the extension and localization of their dendrites and axons. PNs were classified as deep (closest to stratum oriens) or superficial (closest to stratum radiatum) based on the radial position of their soma in the pyramidal layer. CA1, CA2, and CA3 PNs were identified based on their somatic localization, dendritic arborization and presence of thorny excrescences (TE). Among INs with somas located in the pyramidal layer (stratum pyramidale, SP), discrimination between BCs and non-BCs was achieved based on the restriction of their axons to SP or not, respectively. When available, firing patterns upon injection of depolarizing current step injection, action potential (AP) half-width, amount of repolarizing sag current upon hyperpolarization from −70 mV to −100 mV by current step injection, membrane resistance (RM) and capacitance (CM) were additionally used for cell identification. CA2 and CA3a PNs as well as superficial and deep PNs displayed similar firing patterns, AP width, sag current, RM and CM; the only statistically difference being a larger RM of CA3a compared to CA2 PNs which is consistent with previous studies (Chevaleyre and Siegelbaum, 2010; Sun et al., 2017). In contrast, INs had faster firing rates, shorter AP width, higher RM and lower CM than PNs. BCs further differed from non-BCs by the presence of a larger sag potential. All recorded neurons that could not be unequivocally identified as PNs or INs were excluded from analysis. SuM connectivity to each neuronal population was quantified by dividing the number of cells that displayed a post-synaptic response to SuM input stimulation by the total number of cells sampled for each neuronal population across all recording sessions with successful SuM-CA2 transmission.

Data analysis and statistics

Request a detailed protocol

 Electrophysiological recordings were analyzed using IGORpro (Wavemetrics) and Clampfit (Molecular devices) software. For accurate measurements of the kinetics and latencies of post-synaptic responses, the following detection process was used. For each cell, average traces were used to create a template waveform that was then fitted to individual traces and measurements were performed on the fitted traces. When only amplitudes of responses were needed, traces were baselined and amplitudes were simply measured at the peak of the responses. Firing adaptation indexes were computed as the ratio of the initial firing rate (instantaneous firing frequency of the first spikes) over the steady-state firing rate (average instantaneous firing frequency for spikes occurring after 500 ms in the 1 second-long current step). Results are reported ± SEM. Statistical significance was assessed using χ² test, Student’s T test, Mann-Whitney U test, Wilcoxon signed-rank test, Kruskal-Wallis test, one-way or two-way ANOVA where appropriate.

Data availability

The data analysed for this study are included in the manuscript and supporting files. These are included in tables, and data file for Figure 7.

References

    1. Kocsis B
    2. Vertes RP
    (1994)
    Characterization of neurons of the supramammillary nucleus and mammillary body that discharge rhythmically with the hippocampal theta rhythm in the rat
    Journal of Neuroscience 14:7040–7052.
    1. Nò L
    (1934)
    Studies on the structure of the cerebral cortex. II. Continuation of the study of the ammonic system
    Journal für Psychologie und Neurologie 46:113–175.

Decision letter

  1. Katalin Toth
    Reviewing Editor; University of Ottawa, Canada
  2. Laura L Colgin
    Senior Editor; University of Texas at Austin, United States
  3. Sreedharan Sajikumar
    Reviewer; Singapore, National University of Singapore

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The study describes the properties of inputs from the supramammillary nucleus (SuM) to the CA2 area of the hippocampus. This is an important area of study given the increasing importance of SuM inputs and area CA2 in social memory encoding.

Novel information is presented in this paper about the influence of the SuM input on the local hippocampal network in CA2 and the effects of this input on network activity in CA1. The authors use complimentary methods to address this question including patch-clamp recordings and optogenetics. The study will be of interest to readers who study hippocampal function, supramammillary nucleus function, social memory, or neuropsychiatric disorders in which CA2 circuits are affected.

Decision letter after peer review:

Thank you for submitting your article "Local circuit allowing hypothalamic control of hippocampal area CA2 1 activity and consequences for CA1" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Laura Colgin as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Sreedharan Sajikumar (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As the editors have judged that your manuscript is of interest, but as described below that additional experiments are required before it is published, 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). First, because many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)

The study describes the properties of inputs from the supramammillary nucleus (SuM) to the CA2 area of the hippocampus. Novel information is presented by the influence of the SuM input on the local hippocampal network in the CA2 and what the effect of this input is on network activity in the CA1. The authors use complimentary methods to address this question including patch-clamp recordings and optogenetics. Overall the reviewer found this study important, the experiments well-designed and the data of high quality. However, there are several key points raised by the reviewers to strengthen the data in order to fully support the authors' conclusions, addressing these will require additional experimental work. The list below summarizes the list of required experiments reviewers agreed would be necessary for having full confidence in the authors' conclusions. Reviewers' individual comments contain additional issues that would require the authors to make changes to the discussion, figure legends etc.

1. The authors would need to show the effect of SuM stimulation on on synaptically triggered APs and not only on Aps evoked with a current step.

2. The change in the balance of EPCs and IPSCs in a train should be demonstrated in a single cell.

3. The properties of monosynaptic/disynaptic events should be compared and the lack of direct GABAergic input from the SuM demonstrated. The authors should quantify the delay time to light-evoked IPSCs to address whether the SuM-CA2 inputs are forming monosynaptic or disynaptic GABAergic connections to pyramidal neurons, as it is possible SuM neurons co-release glutamate and GABA to CA2. Given the importance of the mono vs. disynaptic innervation of different types of cells, the authors should go beyond the TTX experiments (as TTX would block a disynaptic EPSC) and also use 4-AP to recover the TTX blocked current to unequivocally prove that they inputs are monosynaptic.

4. The preferential role of PV+ cells should be shown with a more selective pharmacological approach.

5. The authors should elaborate on how SuM stimulation influences theta/γ rhythms in the CA1 area.

Reviewer #1:

In this study Robert et al. describes the properties of long-range projections from the SuM to the CA2 area of the hippocampus. The authors identified direct excitatory and indirect inhibitory drive from SuM inputs on CA2 pyramidal neurons and showed that direct excitatory drive impinges on PV-positive basket cells. The overall effect of the input on CA2 activity was an increased precision of APs. The study also suggests that the input from the CA2 drives inhibition in the CA1 area. The study provides very interesting and new information about the cellular properties of SuM input in the CA2 area. This is an important question given the increasing importance of SuM inputs in social memory encoding. The study is timely, currently we have very limited data about the features and exact cellular profile of this input. The study is using elegant technical approaches o answer the central question of the study. While the study is addressing an important question and provides novel data, the authors central claim about the role of feedforward inhibition would need to be strengthened by the addition of experiments addressing how E-I balance changes in trains in individual neurons and how this can be linked to changes in the temporal precision of synaptically evoked APs.

Action potentials are evoked with a current step. Since the study is focused on the network effects of feedforward inhibition, it would be useful to see how the properties of synaptically evoked action potentials change. In the cortex and in the CA1 feed forward inhibition was shown to limit the temporal summation of excitatory inputs which lead to decrease in AP jitter (Gabernet et al., 2005, Pouille and Scanziani, 2001). In order to map these dynamics APs should be evoked via synaptic stimulation and not through current injection.

The authors show recordings of monosynaptic EPSCs in pyramidal cells and interneurons. It would be important to know how inhibitory and excitatory PSCs change in a train. Recordings from single cells held at E-GLUT ad E-GABA would allow the authors to monitor excitatory and inhibitory events in a train ad map how their balance changes. Can the change in E-I balance explain the change in AP jitter?

What are the characteristics of the SuM-driven inhibitory currents? Does the latency and jitter of monosynaptic EPSCs and disynaptic IPSCs differ? If one is monosynaptic and the other is disynaptic one would expect significant differences in both of these parameters.

How do the authors exclude the contribution of feed-back inhibition? feedforward and feed-back inhibition both could have an impact on the temporal precision of APs.

Reviewer #2:

The article brings to light the functional consequences of the activity of SuM afferents terminating at CA2 neurons in the hippocampus using a combination of a variety of methods like whole-cell voltage clamp and optogenetics. In addition, the authors provide evidence that modulation of the CA2 neurons by SuM afferents affects the activity pattern of CA1 neurons. Specifically, the study reveals that the 'functional' connectivity between SuM and CA2 is mainly mediated by the regulation of PV+ basket cells that are involved in the feed forward inhibition of CA2 principal neurons. This study is also relevant in the context of neuropsychiatric disorders where PV+ IN density in the CA2 area is preferentially reduced.

It would be good if some results and implications are further clarified for better understanding in the Discussion section:

1. The results indicate that SuM recruits a feed forward inhibition onto CA2 PNs, which contributes to the shaping of CA2 AP firing. However, it is not entirely intuitive how the feed forward inhibition of CA2 PNs by SuM also reduces CA1 activity, as CA2 has also been known to recruit strong feed forward inhibition onto CA1. This would intuitively suggest that decrease in CA2 activity by photostimulation of SuM afferents will in turn decrease the feed forward inhibition by CA2 onto CA1, and thereby increase CA1 activity. However, the results suggest otherwise. Would this be suggestive of a stronger direct excitatory projection from CA2 to CA1 PNs that is more dominant than the feed forward inhibition of CA1 PNs by CA2? This may be a good point to further elaborate on in the Discussion section, so that the effect of SuM-CA2 connectivity on CA1 output becomes clearer.

2. In the introduction section line 44, it is written that 'CA2 neurons do not undergo NMDA-mediated synaptic plasticity'. This may not be always the case; rather it may be better to rephrase 'NMDA-mediated' as 'high frequency stimulation-induced'. It has been shown previously that NK1 receptor activation by pharmacological application of substance P in hippocampal slices triggers a slow onset NMDA-dependent LTP in CA2 neurons by high frequency stimulation of CA3 afferents to CA2 (Dasgupta et al., 2017).

3. Line 250: "BC transmission is insensitive to MOR activation (Glickfeld et al., 2008)."

Was the Glickfeld study done in CA2 neurons? If not, it would be good to show that PV+ CA2 BCs are also sensitive to DAMGO and to what degree?

The experiment shows that IPSC in PNs are inhibited by DAMGO that should have enhanced light induced EPSCs if PV+ BCs are responsible for feed forward inhibition. But it seems that has not been observed. What are direct EPSCs – electrical stimulation of CA3-CA2 synapses?

4. Overall, the results seem to suggest that SuM stimulation would induce a net inhibition (?) of CA2 PNs by recruiting interneurons (INs). However, the role played by the direct glutamatergic connections from SuM to CA2 PNs is not entirely clear. Is it less prominent due to sparse SuM-PN projections compared to SuM-IN connections in the CA2 area? It may be good to elaborate on this a bit in the discussion.

Reviewer #3:

In this manuscript, Robert et al., demonstrated that medial SuM send glutamatergic projections to the hippocampal CA2 region, and stimulation of these projections exert mixed excitatory and inhibitory responses in CA2 pyramidal neurons. Furthermore, they showed that SuM-CA2 circuits recruit local PV basket cells to provide feedforward inhibition to CA2 pyramidal cells, which increases the precision of action potential firing in conditions of low and high cholinergic tone. Finally, they performed in vivo electrophysiology recording to show that stimulation of SuM-CA2 projections can influence CA1 activity. Overall, this is a well-designed study, and the quality of the data is high. The authors performed impressive amount of electrophysiology recording in acute slices and provided detailed information on how long-distance SuM projection neurons regulate CA2 pyramidal cell activity. These findings provide insights into how SuM activity directly acts on the local hippocampal circuit to modulate social memory encoding. However, there are some concerns that need to be addressed.

1. The authors performed CAV-based retrograde tracing and demonstrated that medial SuM sends glutamatergic projections to CA2. These results are in contrast to a recent study (Li et al., eLife 2020) showing that lateral SuM neurons send dense projections to both CA2 and DG, and the SuM-DG projections release both glutamate and GABA to dentate granule cells. Based on the results from this study and the study from Li et al., does that mean medial SuM neurons are different from lateral SuM neurons in terms of the neurotransmitters they release? The authors need to clarify this point and provide additional ephys data to show that pyramidal cells do not receive direct GABAergic inputs upon stimulation of SuM-CA2 projections using high-chloride internal solution to reveal the IPSCs.

2. The authors claim that SuM-CA2 circuits recruit local PV basket cells to provide feedforward inhibition to CA2 pyramidal cells. While the data presented are supportive, they are not entirely convincing. Specifically, MOR agonist DAMGO is not specific to PV BCs. Though DAMGO has a preferential effect on PV cells over CCK cells, other interneuron types have been shown to be sensitive to DAMGO manipulation. Therefore, these results are subject to alternative interpretation that other types of CA2 local interneurons maybe involved. To show whether PV BCs is the sole interneuron subtype involved, the authors may use a P/Q type calcium channel blocker, ω-agatoxin-TK, as P/Q Ca2+ channels are unique to PV BCs. In addition, chemogenetic inhibition of PV BCs was used, but light-evoked IPSCs are not completely blocked. The authors claimed this could be due to partial silencing of PV BCs. However, there is no evidence showing the efficacy of 10µM CNO application in suppressing CA2 PV basket cell activity. These data should be provided in order to draw such conclusion.

3. CCK basket cells are known to excite PV basket cells (Lee et al., 2011) via a pertussin-toxin sensitive pathway. Is it possible that SuM-CA2 mediated excitation of PV basket cells includes a CCK intermediary? This point should be discussed.

4. The in vivo recording data showed that SuM-CA2 circuit stimulation decreases the firing rate of CA1 pyramidal cells followed by increased firing rate in these cells. Then the authors performed slice recording and showed that the reduced firing rate of CA1 neurons in vivo is likely caused by increased inhibitory inputs onto CA1 pyramidal cells. Figure 7G-H seems to explain the reduced events in the first phase of the tetrode recordings, but not the rebound part. Is there some circuit component that is lost when making slices? Furthermore, what does SuM-CA2 circuit stimulation do to theta/γ rhythms in CA1? These data should be available in the tetrode recordings.

https://doi.org/10.7554/eLife.63352.sa1

Author response

The study describes the properties of inputs from the supramammillary nucleus (SuM) to the CA2 area of the hippocampus. Novel information is presented by the influence of the SuM input on the local hippocampal network in the CA2 and what the effect of this input is on network activity in the CA1. The authors use complimentary methods to address this question including patch-clamp recordings and optogenetics. Overall the reviewer found this study important, the experiments well-designed and the data of high quality. However, there are several key points raised by the reviewers to strengthen the data in order to fully support the authors' conclusions, addressing these will require additional experimental work. The list below summarizes the list of required experiments reviewers agreed would be necessary for having full confidence in the authors' conclusions. Reviewers' individual comments contain additional issues that would require the authors to make changes to the discussion, figure legends etc.

1. The authors would need to show the effect of SuM stimulation on on synaptically triggered APs and not only on Aps evoked with a current step.

We agree with the reviewers that establishing the role of SuM input in controlling synaptically-driven AP firing is most relevant. However, unlike other hippocampal pyramidal neurons, studying synaptically-evoked action potential firing in CA2 PNs is very difficult with optimal slice health and without blocking inhibitory transmission since CA2 PNs have low intrinsic excitability and receive strong inhibition as detailed below. Because of their low membrane resistance and very hyperpolarized membrane potential, CA2 PNs have an intrinsically low excitability which makes it difficult for any given input to cause AP firing in basal conditions. Furthermore, stimulation of CA3 axons evokes a very large feedforward inhibition that strongly prevents action potential firing. Likewise, distal dendritic inputs are extremely attenuated and also rarely lead to AP firing under physiological conditions. For these reasons we used current step injection-induced AP firing as a mean to induce spiking and test the effect of SuM input stimulation. Admittedly, this experimental design is somewhat artificial. We have previously discovered that CA2 PNs spontaneously depolarize and fire burst of action potentials under conditions of elevated cholinergic tone. It is known that the properties of these burst are strongly dependent on both excitatory and inhibitory synaptic transmission (Robert et al., 2020), and also the SuM is most likely to be active during theta rhythm, when cholinergic tone is elevated in the hippocampus. Therefore, we examined how the carbachol-induced bursting of CA2 PNs was altered by SuM input stimulation. We think that the results presented in Figure 7 are directly relevant to the reviewers’ request.

We acknowledge that this experimental design does not allow for the contribution of specific inputs to AP bursting to be established. Thus, we performed additional experiments using electrical stimulation of axonal inputs in SR (CA3 axons) or SLM (EC axons) paired with SuM inputs stimulation. This new data is presented in Figure 2 and described in the text (pages 7-8). Unfortunately, for the vast majority of our slices, we were unable to evoke AP firing with reasonable electrical stimulation intensities, as is very typical for CA2 PNs. However, we were able to determine how SuM input summates with SR and SLM synaptic inputs, as well as determine how the summation ratio of these separate inputs is altered by SuM activity. Notably, we observed a reduction in PSP summation from both SR and SLM repeated stimulation when paired with SuM input activation, which is consistent with a shunting effect of the perisomatic-targeting basket cells recruited by SuM to drive inhibition in CA2 PNs.

2. The change in the balance of EPCs and IPSCs in a train should be demonstrated in a single cell.

We thank the reviewers for raising this point which is very important for the interpretation of the data with repeated SuM input stimulation presented in Figures 6, 7 and 8. In supplemental figure 4 we present IPSCs and EPSCs all recorded in the same CA2 PNs in the presence and absence of carbachol (described in pages 14-15). In short, we observed a consistent short-term depression of both SuM-driven monosynaptic EPSCs and di-synaptic IPSCs in CA2 PNs under basal conditions, with the resulting E/I ratio being stable over the train of SuM stimuli. Application of carbachol reduced initial amplitudes of both EPSCs and IPSCs while decreasing their short-term depression, resulting in a decrease of E/I ratio suggesting that the SuM inhibitory influence over CA2 PNs is even stronger and more sustained in conditions of elevated cholinergic tone.

3. The properties of monosynaptic/disynaptic events should be compared and the lack of direct GABAergic input from the SuM demonstrated. The authors should quantify the delay time to light-evoked IPSCs to address whether the SuM-CA2 inputs are forming monosynaptic or disynaptic GABAergic connections to pyramidal neurons, as it is possible SuM neurons co-release glutamate and GABA to CA2. Given the importance of the mono vs. disynaptic innervation of different types of cells, the authors should go beyond the TTX experiments (as TTX would block a disynaptic EPSC) and also use 4-AP to recover the TTX blocked current to unequivocally prove that they inputs are monosynaptic.

The reviewers are completely correct in asking for these experiments in order to confirm that indeed the SuM-CA2 inputs are entirely glutamatergic. All of these points are crucial to our investigation. In fact, we have already performed all of these requested experiments and analysis. These data were published as an extended data figure in Chen et al., Nature 2020. The exclusive mono-synaptic glutamatergic nature of the SuM-CA2 transmission was demonstrated and directly compared to the SuM-DG synapse with measurements of EPSC vs IPSC onset latency and application of TTX with 4AP or NBQX + APV. We regret not summarizing more clearly in the text of this manuscript our previous findings and have added a description of these results to our manuscript (page 7).

4. The preferential role of PV+ cells should be shown with a more selective pharmacological approach.

We have taken the advice from reviewer #3 and performed a series of experiments using the P/Q voltage-gated calcium channel blocker ω-agatoxin TK. Block of these channels is predicted to only inhibit Gaba release from PV+ interneurons. These results of these experiments are in figure 5 and described page 12. Our experiments showed a near-complete block of SuM-driven IPSCs in CA2 PNs consistent with PV+ INs mediating this feedforward inhibition. However, the interpretation of these results is complicated by the fact that application of ω-agatoxin TK also reduced glutamatergic transmission at the SuM-CA2 synapse. If we assume that the SuM-IN synapse has a similar population of channels, then part of the reduction in inhibitory transmission may not be due to PV+ IN specificity, but by the reduction in glutamate release. While these results are not fully conclusive proof, we have still included them, as they further characterize the SuM input.

As for the chemogenetic inhibition of PV+ INs, we took the reviewers advice and provided more direct evidence that 10 µM CNO is not sufficiently suppressing CA2 PV basket cell activity. We performed recordings of PV+ INs in area CA2 and measured changes in the intrinsic properties and action potential firing of these cells before and following CNO application. We did observe a small hyperpolarization of the membrane potential, however we still observed AP firing in several of the cells, indicating that this hyperpolarization is insufficient to fully silence PV+ INs. This is shown in supplemental figure 3 and described on page 11. GiDREADD can also act on pre-synaptic terminals, so it is not entirely conclusive if the PV+ cells that fire action potentials still release GABA. However, we feel that in combination with the incomplete expression of the GiDREADD in all PV+ interneurons, and the residual AP firing, that the 10 µM CNO is not entirely effective at suppressing PV+ IN activity in our experiments.

We would like to mention that the expression of DOR receptors has been shown to be expressed in 35% of PV+ BCs in the hippocampus (Erbs, 2012) and furthermore, we have shown that transmission exclusively from PV+ cells in area CA2 undergoes a DOR-dependent depression (Piskorowski and Chevaleyre, 2013) that can be induced by application of DPDPE. We conclude that the effect of DPDPE on SuM transmission is compelling evidence that these inputs target PV+ cells.

5. The authors should elaborate on how SuM stimulation influences theta/γ rhythms in the CA1 area.

We have added a section to the discussion (pages 19-21) about theta/γ rhythms and the potential contribution of SuM input. We agree whole-heartedly with the reviewer about the fundamental importance of examining how SuM contributes to these brain activities in vivo. Please see our extended answer to reviewer 3 below.

Reviewer #1:

In this study Robert et al. describes the properties of long-range projections from the SuM to the CA2 area of the hippocampus. The authors identified direct excitatory and indirect inhibitory drive from SuM inputs on CA2 pyramidal neurons and showed that direct excitatory drive impinges on PV-positive basket cells. The overall effect of the input on CA2 activity was an increased precision of APs. The study also suggests that the input from the CA2 drives inhibition in the CA1 area. The study provides very interesting and new information about the cellular properties of SuM input in the CA2 area. This is an important question given the increasing importance of SuM inputs in social memory encoding. The study is timely, currently we have very limited data about the features and exact cellular profile of this input. The study is using elegant technical approaches o answer the central question of the study. While the study is addressing an important question and provides novel data, the authors central claim about the role of feedforward inhibition would need to be strengthened by the addition of experiments addressing how E-I balance changes in trains in individual neurons and how this can be linked to changes in the temporal precision of synaptically evoked APs.

Action potentials are evoked with a current step. Since the study is focused on the network effects of feedforward inhibition, it would be useful to see how the properties of synaptically evoked action potentials change. In the cortex and in the CA1 feed forward inhibition was shown to limit the temporal summation of excitatory inputs which lead to decrease in AP jitter (Gabernet et al., 2005, Pouille and Scanziani, 2001). In order to map these dynamics APs should be evoked via synaptic stimulation and not through current injection.

Please see our response to point #1 in the above section.

The authors show recordings of monosynaptic EPSCs in pyramidal cells and interneurons. It would be important to know how inhibitory and excitatory PSCs change in a train. Recordings from single cells held at E-GLUT ad E-GABA would allow the authors to monitor excitatory and inhibitory events in a train ad map how their balance changes. Can the change in E-I balance explain the change in AP jitter?

These experiments examining the changes in E/I balance with a train of inputs has been added to the manuscript (Supplemental figure 4, pages 14-15). We thank the reviewer for this insight. In fact, the E-I balance does not change during a train of stimulation, staying between 0.5 and 0.6 (supplemental figure 4). However, we do see a large change in the E/I ratio in high cholinergic tone (that also does not change during the train), indicating that the inhibitory drive of the SuM input is increased under these conditions.

What are the characteristics of the SuM-driven inhibitory currents? Does the latency and jitter of monosynaptic EPSCs and disynaptic IPSCs differ? If one is monosynaptic and the other is disynaptic one would expect significant differences in both of these parameters.

These points are very critical to understanding the local SuM-CA2 circuitry. These experiments have been published as supplemental data for Chen et al., Nature 2020. We have also added a description of these results to our current manuscript (page 7).

How do the authors exclude the contribution of feed-back inhibition? feedforward and feed-back inhibition both could have an impact on the temporal precision of APs.

We thank the reviewer for raising this important point which we failed to address in our initial submission. In basal conditions, feedback inhibition is not at play because stimulation of the SuM input evokes only small amplitude EPSPs that remain subthreshold in CA2 PNs, as documented in Figures 1 and 2. Indeed, we never observed SuM-evoked AP firing from CA2 PNs in our recordings (n = 0 out of 78, Figure 3). However, when inducing action potential firing in CA2 PNs either by current injection (Figure 6) or CCh application (Figure 7), the contribution of feedback inhibition becomes very relevant. Indeed, it is likely that CA2 PNs recruit not only feedback inhibition (Mercer, 2012a, Mercer, 2012b) but also recurrent excitation (Cui, 2013, Hitti, 2014, Okamoto, 2019), both of which could impact the temporal precision of APs. In the case of current step injection-induced firing, the contribution of feedback inhibition and recurrent excitation to reducing AP jitter can be examined with our pharmacology experiments: in the GABA receptors blocked condition (Figure 6), the effect of SuM input stimulation on both the AP latency and jitter is abolished, thereby indicating that the main factor imposing the initial AP delay and constraining AP jitter is SuM-driven feedforward inhibition. This is because, at least for the first AP evoked by current step injection in the single CA2 PN recorded from, it is reasonable to assume that other CA2 PNs were not actively firing thus removing the possibility of feedback inhibition and recurrent excitation from the equation. Furthermore, the small probability of PNs to be connected (1.4%) and the small size of CA2-CA2 EPSP (~0.5mV) (Okamoto, 2019) makes it very unlikely that recurrent excitation plays any role when a single PN is depolarized. The problem becomes much more complicated when considering the CCh-induced AP burst firing conditions, as many CA2 PNs likely burst fire somewhat independently of the single CA2 PN recorded in any given experiment. Therefore, feedback inhibition and recurrent excitation are to be expected in these conditions, thereby potentially contributing to the temporal precision of AP firing in the recorded CA2 PN. Unfortunately, typical pharmacology blocking GABA or glutamate receptors is not suitable here as the CCh-induced AP burst firing pattern of CA2 PNs relies on intact synaptic transmission (Robert, 2020), thus making the examination of the respective roles of feedforward versus feedback inhibition in these conditions difficult to probe experimentally. Nevertheless, we deem reasonable to interpret our data with SuM input stimulation during CCh-induced spontaneous AP burst firing as follows: by recruiting perisomatic-targeted feedforward inhibition on most CA2 PNs, SuM input stimulation might effectively enforce a restricted time window during which CA2 PNs can undergo or sustain AP burst firing, thereby limiting independent burst firing from individual CA2 PNs and instead shaping a synchronized output from many CA2 PNs subsequently driving feedforward inhibition over area CA1 as shown in Figures 7 and 8.

Reviewer #2:

The article brings to light the functional consequences of the activity of SuM afferents terminating at CA2 neurons in the hippocampus using a combination of a variety of methods like whole-cell voltage clamp and optogenetics. In addition, the authors provide evidence that modulation of the CA2 neurons by SuM afferents affects the activity pattern of CA1 neurons. Specifically, the study reveals that the 'functional' connectivity between SuM and CA2 is mainly mediated by the regulation of PV+ basket cells that are involved in the feed forward inhibition of CA2 principal neurons. This study is also relevant in the context of neuropsychiatric disorders where PV+ IN density in the CA2 area is preferentially reduced.

It would be good if some results and implications are further clarified for better understanding in the Discussion section:

1. The results indicate that SuM recruits a feed forward inhibition onto CA2 PNs, which contributes to the shaping of CA2 AP firing. However, it is not entirely intuitive how the feed forward inhibition of CA2 PNs by SuM also reduces CA1 activity, as CA2 has also been known to recruit strong feed forward inhibition onto CA1. This would intuitively suggest that decrease in CA2 activity by photostimulation of SuM afferents will in turn decrease the feed forward inhibition by CA2 onto CA1, and thereby increase CA1 activity. However, the results suggest otherwise. Would this be suggestive of a stronger direct excitatory projection from CA2 to CA1 PNs that is more dominant than the feed forward inhibition of CA1 PNs by CA2? This may be a good point to further elaborate on in the Discussion section, so that the effect of SuM-CA2 connectivity on CA1 output becomes clearer.

The referee is correct in that the connection between CA2 and CA1 is predominantly excitatory. CA2 PNs make a powerful excitatory connection with CA1 PNs (Chevaleyre and Siegelbaum, 2010, Kohara et al., 2014). CA2 PNs also drive feedforward inhibition onto CA1 PNs (Boehringer et al., 2017, Nasrallah et al., 2019). Here we show that under conditions that resemble active exploration (high cholinergic tone) the SuM-CA2-CA1 elicits a reliable poly-synaptic IPSC in CA1 PNs (Figure 8), as well as an occasional poly-synaptic CA1 EPSC. Our data does not directly show that SuM feedforward inhibition of CA2 causes a drop in CA2-CA1 excitation underlying the decreased CA1 activity following SuM-CA2 stimulation. Instead, our data shows that SuM inhibition of CA2 yields synchronized feedforward inhibition from CA2 to CA1 which decreases CA1 activity. We appreciate the complexity of this idea and have done our best to elaborate in the discussion (page 19).

2. In the introduction section line 44, it is written that 'CA2 neurons do not undergo NMDA-mediated synaptic plasticity'. This may not be always the case; rather it may be better to rephrase 'NMDA-mediated' as 'high frequency stimulation-induced'. It has been shown previously that NK1 receptor activation by pharmacological application of substance P in hippocampal slices triggers a slow onset NMDA-dependent LTP in CA2 neurons by high frequency stimulation of CA3 afferents to CA2 (Dasgupta et al., 2017).

We thank the reviewer for this clarification. In the text, “NMDA-mediated” has been replaced by “high frequency stimulation-induced” as suggested.

3. Line 250: "BC transmission is insensitive to MOR activation (Glickfeld et al., 2008)."

Was the Glickfeld study done in CA2 neurons? If not, it would be good to show that PV+ CA2 BCs are also sensitive to DAMGO and to what degree?

The experiment shows that IPSC in PNs are inhibited by DAMGO that should have enhanced light induced EPSCs if PV+ BCs are responsible for feed forward inhibition. But it seems that has not been observed. What are direct EPSCs – electrical stimulation of CA3-CA2 synapses?

In the experiments where we applied DAMGO, the EPSCs were recorded in presence of GABAR blockers. Therefore, any change in GABA transmission would not affect the light induced EPSC.

The Glickfeld study was performed in CA1 only. However, expression of MOR mRNA is very strong in area CA2 and a large fraction of PV+ INs are also positive for MOR throughout the hippocampus (Ralf et al., Journal of Comparative Neurology, 2004). We haven’t directly monitored the effect of DAMGO on PV+ transmission in area CA2. However, in an ongoing study, we stimulated CA3 inputs and monitors the electrically evoked IPSCs in CA2 pyramidal cells (Author response image 1). We found that DAMGO application strongly reduces IPSCs (by ~75%) and completely occludes subsequent effect of the DOR agonist DPDPE. We are confident that DPDPE is acting solely on PV+ cells in area CA2. Therefore, from these results we can conclude that PV+ INs in area CA2 are also strongly sensitive to DAMGO as they are known to be in area CA1.

Author response image 1

4. Overall, the results seem to suggest that SuM stimulation would induce a net inhibition (?) of CA2 PNs by recruiting interneurons (INs). However, the role played by the direct glutamatergic connections from SuM to CA2 PNs is not entirely clear. Is it less prominent due to sparse SuM-PN projections compared to SuM-IN connections in the CA2 area? It may be good to elaborate on this a bit in the discussion.

We feel that the combined excitation and feedforward inhibition from the SuM to area CA2 is acting together to coordinate the timing of CA2 PN output. We have expanded this explanation in the discussion (page 18).

Reviewer #3:

In this manuscript, Robert et al. demonstrated that medial SuM send glutamatergic projections to the hippocampal CA2 region, and stimulation of these projections exert mixed excitatory and inhibitory responses in CA2 pyramidal neurons. Furthermore, they showed that SuM-CA2 circuits recruit local PV basket cells to provide feedforward inhibition to CA2 pyramidal cells, which increases the precision of action potential firing in conditions of low and high cholinergic tone. Finally, they performed in vivo electrophysiology recording to show that stimulation of SuM-CA2 projections can influence CA1 activity. Overall, this is a well-designed study, and the quality of the data is high. The authors performed impressive amount of electrophysiology recording in acute slices and provided detailed information on how long-distance SuM projection neurons regulate CA2 pyramidal cell activity. These findings provide insights into how SuM activity directly acts on the local hippocampal circuit to modulate social memory encoding. However, there are some concerns that need to be addressed.

1. The authors performed CAV-based retrograde tracing and demonstrated that medial SuM sends glutamatergic projections to CA2. These results are in contrast to a recent study (Li et al., eLife 2020) showing that lateral SuM neurons send dense projections to both CA2 and DG, and the SuM-DG projections release both glutamate and GABA to dentate granule cells. Based on the results from this study and the study from Li et al., does that mean medial SuM neurons are different from lateral SuM neurons in terms of the neurotransmitters they release? The authors need to clarify this point and provide additional ephys data to show that pyramidal cells do not receive direct GABAergic inputs upon stimulation of SuM-CA2 projections using high-chloride internal solution to reveal the IPSCs.

This is a very valid point, and as we have detailed above, the experiments directly comparing the SuM transmission into the DG and area CA2 have been published as a supplemental figure in Chen et al., Nature 2020. As the reviewer mentions, in that publication we present multiple lines of evidence supporting the premise that the CA2-projecting and DG-projecting cells in the SuM are from different populations and use different neurotransmitter modalities. These results are supported by prior histological results by Soussi et al., EJN, 2010 that show the CA2-projecting SuMM fibers do not co-express GAD65 unlike the DG-projecting SUML fibers that express both VGLUT2 and GAD65.

In the Li et al., paper (eLife, 2020) the authors show that AAV5-DiO.eYFP injected into the SuM in a Vgat-cre mouse line results in fibers in area CA2. These results lack control data showing the specificity of the AAV5 viral vector in the vgat-cre line (co-labelling with Gad65/67…). In our experience, it is very easy to get non-specific expression of AA5 vectors in area CA2. In fact, we use this problem to our advantage in Dominguez et al., Cell Reports 2019.

2. The authors claim that SuM-CA2 circuits recruit local PV basket cells to provide feedforward inhibition to CA2 pyramidal cells. While the data presented are supportive, they are not entirely convincing. Specifically, MOR agonist DAMGO is not specific to PV BCs. Though DAMGO has a preferential effect on PV cells over CCK cells, other interneuron types have been shown to be sensitive to DAMGO manipulation. Therefore, these results are subject to alternative interpretation that other types of CA2 local interneurons maybe involved. To show whether PV BCs is the sole interneuron subtype involved, the authors may use a P/Q type calcium channel blocker, ω-agatoxin-TK, as P/Q Ca2+ channels are unique to PV BCs. In addition, chemogenetic inhibition of PV BCs was used, but light-evoked IPSCs are not completely blocked. The authors claimed this could be due to partial silencing of PV BCs. However, there is no evidence showing the efficacy of 10µM CNO application in suppressing CA2 PV basket cell activity. These data should be provided in order to draw such conclusion.

As the reviewer suggested, we used ω-agatoxin-TK to explore whether PV+ BCs are the sole population of interneurons. Consistent with PV+ cells being the primary population underlying the feedforward inhibition onto CA2 PNs, we saw a nearly complete block of light-evoked IPSCs. However, we also examined the direct excitatory SuM-CA2 glutamatergic synapse and found that application of ω-agatoxin-TK resulted in a 50% block of glutamatergic transmission. Thus, if we assume that the glutamatergic SuM-IN synapse is also reduced then the reduction of feedforward IPSC may not be because of the specificity of P/Q Ca2+ channel expression of PV+ cells, but because the SuM glutamatergic transmission is reduced. While these results do not provide the definitive result we were hoping for, we have still included them in Figure 5 to be considered in parallel with the other experiments (pages 11-12).

As for the chemogenetic inhibition of PV BCs, we took the reviewers advice and provided more direct evidence that 10 µM CNO is not sufficiently suppressing CA2 PV basket cell activity. We performed recordings of PV+ INs in area CA2 and measured changes in the intrinsic properties and action potential firing of these cells before and following CNO application. We did observe a small hyperpolarization of the membrane potential, however in we still observed AP firing in several of the cells, indicating that this hyperpolarization is insufficient to fully silence PV+ INs. This is shown in supplemental figure 3 and described page 11. GiDREADD can also act on pre-synaptic terminals, so it is not entirely conclusive if the PV cells that fire action potentials still release GABA. However, we feel that in combination with the incomplete expression of the GiDREADD in all PV+ interneurons, and the residual AP firing, that the 10 µM CNO is not entirely effective at suppressing PV-BC activity in our experiments.

3. CCK basket cells are known to excite PV basket cells (Lee et al., 2011) via a pertussin-toxin sensitive pathway. Is it possible that SuM-CA2 mediated excitation of PV basket cells includes a CCK intermediary? This point should be discussed.

This is a very intriguing point, and we thank the reviewer for this suggestion. Lee et al. 2011 demonstrates that CCK application results in a depolarization of PV+ cells via a nonselective calcium conductance linked to intracellular calcium store release. The idea that CCK release would potentially enhance ability of SuM inputs to recruit PV+ cells is quite interesting and have added this to the discussion (page 19). Similarly, we have also mentioned the other intriguing finding that CCK cells actively inhibit PV+ basket cells (Dodek 2021), presenting two ways in which CCK cells may be acting indirectly to modulate SuM input to area CA2 (page 19).

4. The in vivo recording data showed that SuM-CA2 circuit stimulation decreases the firing rate of CA1 pyramidal cells followed by increased firing rate in these cells. Then the authors performed slice recording and showed that the reduced firing rate of CA1 neurons in vivo is likely caused by increased inhibitory inputs onto CA1 pyramidal cells. Figure 7G-H seems to explain the reduced events in the first phase of the tetrode recordings, but not the rebound part. Is there some circuit component that is lost when making slices? Furthermore, what does SuM-CA2 circuit stimulation do to theta/γ rhythms in CA1? These data should be available in the tetrode recordings.

This is a very interesting point, and a central flaw in acute slice physiology. There are definitely circuit components that are lost when making transverse hippocampal slices. We have added two paragraphs to the Discussion section detailing the potential role of SuM-CA2 input to theta and γ oscillations (pages 19-21).

The in vivo data in the paper was based on short stimulation pulses in resting animals where theta and γ are minimal. Thus, further examination of theta and γ requires further distinct experiments. As this manuscript elucidates the local circuitry underlying the SuM input in area CA2, and the overall effect of SuM activity on action potential firing in this region, further examination of network oscillations is outside the scope of this work.

However, we still went through all of our data to try to determine if there is a large apparent effect of SuM activity in theta / γ. Because of the current Covid-19 crisis, we have not been able to continue with the majority of our experimental work for this project, so we can only present preliminary data from two animals with acceptable trials, tetrode locations (in CA1), fiber placement over the area CA2, and properly targeted virus in the SuM. We present in Author response image 2 our very preliminary results of CA1 LFP with and without optical stimulation on a circular linear track. As the mice ran on the track, we applied a protocol with 1 minute ON, 1 minute OFF of 10 Hz light stim in CA2. For analysis, we calculated the PSD from the CA1 LFP with the middle 40” of all laser ON periods and the middle 40” of all laser OFF periods, then averaged. Power across the theta (4-12Hz) and γ (30-90 Hz) bands was normalized to the total power in the 0-4 Hz band for each trial, then averaged across the two animals.

Author response image 2

As visible in the wavelet spectragrams across the whole trial, we do not see any obvious change in theta or γ power when the laser is on. We also present (Author response image 3) the normalized power in the theta and γ bands in OFF/ON conditions and averaged across the 2 mice. Currently we do not see any obvious changes with this preliminary experimental design. These results have convinced us that this question is not easily answered in these challenging times and is needs to be explored in a later study.

Author response image 3
https://doi.org/10.7554/eLife.63352.sa2

Article and author information

Author details

  1. Vincent Robert

    INSERM UMR1266, Institute of Psychiatry and Neuroscience of Paris, Team Synaptic Plasticity and Neural Networks, Université de Paris, Paris, France
    Contribution
    Conceptualization, Resources, Formal analysis, Investigation, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared
  2. Ludivine Therreau

    INSERM UMR1266, Institute of Psychiatry and Neuroscience of Paris, Team Synaptic Plasticity and Neural Networks, Université de Paris, Paris, France
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
  3. Vivien Chevaleyre

    1. INSERM UMR1266, Institute of Psychiatry and Neuroscience of Paris, Team Synaptic Plasticity and Neural Networks, Université de Paris, Paris, France
    2. GHU Paris Psychiatrie and Neurosciences, Paris, France
    Contribution
    Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
  4. Eude Lepicard

    INSERM UMR1266, Institute of Psychiatry and Neuroscience of Paris, Team Synaptic Plasticity and Neural Networks, Université de Paris, Paris, France
    Contribution
    Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
  5. Cécile Viollet

    INSERM UMR1266, Institute of Psychiatry and Neuroscience of Paris, Team Synaptic Plasticity and Neural Networks, Université de Paris, Paris, France
    Contribution
    Project administration, Writing - review and editing
    Competing interests
    No competing interests declared
  6. Julie Cognet

    INSERM UMR1266, Institute of Psychiatry and Neuroscience of Paris, Team Synaptic Plasticity and Neural Networks, Université de Paris, Paris, France
    Contribution
    Resources, Investigation, Project administration
    Competing interests
    No competing interests declared
  7. Arthur JY Huang

    Laboratory for Circuit and Behavioral Physiology, RIKEN Center for Brain Science, Saitama, Japan
    Contribution
    Resources, Methodology
    Competing interests
    No competing interests declared
  8. Roman Boehringer

    Laboratory for Circuit and Behavioral Physiology, RIKEN Center for Brain Science, Saitama, Japan
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2856-3262
  9. Denis Polygalov

    Laboratory for Circuit and Behavioral Physiology, RIKEN Center for Brain Science, Saitama, Japan
    Contribution
    Formal analysis, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8165-5257
  10. Thomas J McHugh

    Laboratory for Circuit and Behavioral Physiology, RIKEN Center for Brain Science, Saitama, Japan
    Contribution
    Supervision, Funding acquisition, Investigation, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1243-5189
  11. Rebecca Ann Piskorowski

    1. INSERM UMR1266, Institute of Psychiatry and Neuroscience of Paris, Team Synaptic Plasticity and Neural Networks, Université de Paris, Paris, France
    2. GHU Paris Psychiatrie and Neurosciences, Paris, France
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Validation, Methodology, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    rebecca.piskorowski@inserm.fr
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0120-2360

Funding

RIKEN Brain Science Institute

  • Thomas J McHugh

Ministry of Education, Culture, Sports, Science and Technology (19H05646)

  • Thomas J McHugh

Ministry of Education, Culture, Sports, Science and Technology (19H05233)

  • Thomas J McHugh

Agence Nationale de la Recherche (ANR-13-JSV4-0002-01)

  • Rebecca Ann Piskorowski

Agence Nationale de la Recherche (ANR-18-CE37-0020-01)

  • Rebecca Ann Piskorowski

Ville de Paris (Programme Emergences)

  • Rebecca Ann Piskorowski

Brain and Behavior Research Foundation (NARSAD Young INvestigator Award)

  • Rebecca Ann Piskorowski

Fondation pour la Recherche Médicale (FTD20170437387)

  • Vincent Robert

Schizo-Oui

  • Vincent Robert

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

Equipment for the IPNP mouse husbandry facility was funded by the Région Ile de France. This work was supported by the RIKEN Center for Brain Science (TJM), Grant-in-Aid for Scientific Research from MEXT (19H05646; TJM), Grant-in-Aid for Scientific Research on Innovative Areas from MEXT (19H05233; TJM), ANR-13-JSV4-0002-01 (RAP), ANR-18-CE37-0020-01 (RAP), the Ville de Paris Programme Emergences (RAP), and the Brain and Behavioral Research Foundation NARSAD Young Investigator Grant (RAP) and the Foundation Recherche Médicale, FRM:FTD20170437387 and a gift from Schizo-Oui (VR).

Ethics

Animal experimentation: All procedures involving animals were performed in accordance with institutional regulations (French Ministry of Research and Education protocol #12406-2016040417305913). Animal sample sizes were estimated using power tests with standard deviations and ANOVA values from pilot experiments. A 15% failure rate was assumed to account for stereotaxic injection errors and slice preparation complications. Every effort was made to reduce animal suffering.

Senior Editor

  1. Laura L Colgin, University of Texas at Austin, United States

Reviewing Editor

  1. Katalin Toth, University of Ottawa, Canada

Reviewer

  1. Sreedharan Sajikumar, Singapore, National University of Singapore

Publication history

  1. Received: September 22, 2020
  2. Accepted: May 17, 2021
  3. Accepted Manuscript published: May 18, 2021 (version 1)
  4. Version of Record published: May 26, 2021 (version 2)

Copyright

© 2021, Robert 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

  • 831
    Page views
  • 160
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Neuroscience
    Xiaoxuan Jia et al.
    Research Article

    Temporal continuity of object identity is a feature of natural visual input, and is potentially exploited -- in an unsupervised manner -- by the ventral visual stream to build the neural representation in inferior temporal (IT) cortex. Here we investigated whether plasticity of individual IT neurons underlies human core-object-recognition behavioral changes induced with unsupervised visual experience. We built a single-neuron plasticity model combined with a previously established IT population-to-recognition-behavior linking model to predict human learning effects. We found that our model, after constrained by neurophysiological data, largely predicted the mean direction, magnitude and time course of human performance changes. We also found a previously unreported dependency of the observed human performance change on the initial task difficulty. This result adds support to the hypothesis that tolerant core object recognition in human and non-human primates is instructed -- at least in part -- by naturally occurring unsupervised temporal contiguity experience.

    1. Neuroscience
    Nick Taubert et al.
    Research Article

    Dynamic facial expressions are crucial for communication in primates. Due to the difficulty to control shape and dynamics of facial expressions across species, it is unknown how species-specific facial expressions are perceptually encoded and interact with the representation of facial shape. While popular neural network models predict a joint encoding of facial shape and dynamics, the neuromuscular control of faces evolved more slowly than facial shape, suggesting a separate encoding. To investigate these alternative hypotheses, we developed photo-realistic human and monkey heads that were animated with motion capture data from monkeys and humans. Exact control of expression dynamics was accomplished by a Bayesian machine-learning technique. Consistent with our hypothesis, we found that human observers learned cross-species expressions very quickly, where face dynamics was represented largely independently of facial shape. This result supports the co-evolution of the visual processing and motor control of facial expressions, while it challenges appearance-based neural network theories of dynamic expression recognition.