Recruitment of parvalbumin and somatostatin interneuron inputs to adult born dentate granule neurons

GABA is a key regulator of adult-born dentate granule cell (abDGC) maturation so mapping the functional connectivity between abDGCs and local interneurons is required to understand their development and integration into the hippocampal circuit. We recorded from birthdated abDGCs in mice and photoactivated parvalbumin (PV) and somatostatin (SST) interneurons to map the timing and strength of inputs to abDGCs during the first 4 weeks after differentiation. abDGCs received input from PV interneurons in the first week, but SST inputs were not detected until the second week. Analysis of desynchronized quantal events established that the number of GABAergic synapses onto abDGCs increased with maturation, whereas individual synaptic strength was constant. Voluntary wheel running in mice scaled the GABAergic input to abDGCs by increasing the number of synaptic contacts from both interneuron types. This demonstrates that GABAergic innervation to abDGCs develops during a prolonged post-mitotic period and running scales both SST and PV synaptic afferents.


Retroviral birthdating. A replication incompetent retrovirus based on MMLV (Moloney Murine Leu-
kemia Virus) expressing RFP was prepared as described 21,22 . Briefly, GP2-293 packaging cells (Clontech) were co-transfected with p-CAG-RFP and p-CMV-vsv-g (plasmids were a gift from Fred H. Gage at the Salk Institute) using Lipofectamine 2000 (Invitrogen) 21,23,24 . The media was collected from transfected cells 3-and 6-days posttransfection, filtered, and centrifuged at 25,000 rpm to precipitate the virus 21,23,24 . 6-8 week old PV-ChR2 and SST-ChR2 mice of either sex were anesthetized using ketamine/xylazine and 1 μl of virus was injected bilaterally into the hilar region of the dentate gyrus (from Bregma: 0.24 mm posterior, 0.16 mm lateral, 0.24 mm ventral) at a rate of ~ 0.3 μl/min 23 . Slice preparation and electrophysiology. Slicing and electrophysiology were carried out as previously described 23 . Briefly, 250 μm coronal slices were prepared at 7, 14, 21, and 28 (± 1) days post retrovirus injection. Fluorescence targeted recordings were made from RFP-expressing neurons in the inner granule cell layer. The composition of the intracellular solution used for voltage clamp recording was (in mM): 95 CsF, 25 CsCl, 10 Cs-HEPES, 10 Cs-EGTA, 2 NaCl, 2 Mg-ATP, 10 QX-314, 5 TEA-Cl, and 5 4-AP, pH adjusted to 7.3 with CsOH. For asynchronous release, slices were perfused with oxygenated artificial cerebrospinal fluid (ACSF) containing 6 mM SrCl 2 , 1 mM MgCl 2 and 0.5 mM CaCl 2 . Data were collected and analyzed using pClamp 10 software (Molecular Devices). Neurons were voltage-clamped at − 70 mV to record IPSCs. Inclusion of D-APV (50 μM) and CNQX (10 μM) ensured isolation of GABAergic events. Bicuculine (10 μM) was added after recording of optoIPSCs to confirm that events were GABAergic. MiniAnalysis (Synaptosoft) was used to analyze asynchronous events (aIPSCs). ChR2-expressing interneurons were photoactivated using a 5 ms pulse of LED light (470 nm) to the entire visual field. Input-output curves indicate that 500 mV into the LED driver (~ 350 μW of light measured at the slice) consistently evoked the maximal IPSC amplitude in postsynaptic mature and birthdated DGCs and that the response plateaus at higher stimulation intensities ( Supplementary Fig. 1). Mature, unlabeled cells were recorded from the outer third of the GCL as most abDGCs have been demonstrated to migrate only to the inner third with only ~ 10% migrating to the outer third of the GCL 25 . Running wheels. 4-5-week-old PV-ChR2 and SST-ChR2 mice of either sex were single housed with running wheels (MedAssociates) for 3 weeks prior to retroviral injection and were returned to the cage with a running wheel until they were sacrificed for experiments [7,14,21 or 28 (± 1) days]. Total distance run was quantified for each day apart from during postoperative quarantine, however animals continued to have access to wheels during this period ( Supplementary Fig. 3). Animals ran an average of 7.8 km per day.
Immunohistochemistry. [6][7][8] week old PV-ChR2 and SST-ChR2 mice of either sex were anesthetized using ketamine/xylazine and perfused with PBS containing 0.02% sodium nitrite and 4 mM MgSO 4 followed by 2% paraformaldehyde in 0.1 M sodium acetate buffer (pH 6.5) for 6 min, and 2% paraformaldehyde in 0.1 M sodium borate buffer (pH 8.5) for 12-18 min 26,27 . The brains were removed from the skull, post-fixed overnight in 2% paraformaldehyde in 0.1 M sodium borate buffer (pH 8.5), then sectioned coronally at 50 μm on a Leica Vibratome VT1000s in PBS. The sections were collected and stored at 4 °C in PBS containing 0.02% sodium azide. Free-floating sections were rinsed for 30 min in PBS followed by four 15-min washes in PBS containing 0.1% Triton X-100 (Sigma, T8787) and 0.1% bovine serum albumin (BSA; Sigma, A4503). The sections were then blocked with 3% normal donkey serum (NDS; Jackson ImmunoResearch, 017-000-121) for  Data analysis. OptoIPSCs were analyzed using Clampfit (Molecular Devices) and aIPSCs were analyzed using MiniAnalysis (Synaptosoft). Additional data analysis was conducted with Microsoft Excel and OriginPro software. Comparisons were made with a Mann-Whitney U test using OriginPro. Differences were considered significant when p < 0.05. All data are reported as mean ± SEM.

Results
Development of PV and SST inputs to abDGCs. Using PV-ChR2 and SST-ChR2 mice we studied the functional development of GABAergic synapses from these two interneuron types to abDGCs. We first confirmed that PV Cre and SST Cre mice had Cre recombinase expression that reflected endogenous expression of PV and SST by crossing these mice to Ai9 (RCL-tdT) mice to label Cre expressing interneurons. Immunolabeling of PV-tdTom mice for SST and immunolabeling of SST-tdTom mice for PV demonstrated that Cre expression was primarily distinct in the two interneuron populations (Figs. 1b, 2a) as has been previously reported 28 . In addition, expression of YFP in PV Cre mice crossed with Ai32 (RCL-ChR2(H134R)/EYFP)(PV-ChR2) mice clearly demonstrated a pattern of axonal and dendritic labelling of PV interneurons enriched in the GCL consistent with their known patterns of innervation ( Fig. 1c) 7,8 . In SST Cre mice crossed with Ai32 mice (SST-ChR2), YFP expression showed a distinct pattern in which the fluorescence was observed in the hilar and molecular regions consistent with the known axonal plexus of SST interneurons ( Fig. 2b) 7,8 . Thus, the two mouse lines express Cre recombinase consistent with the expression of neurochemically defined populations of interneurons.
To identify and birthdate newborn DGCs, we injected a modified retrovirus expressing RFP (RV-RFP) into the dentate gyrus of 6-8 week old PV-ChR2 or SST-ChR2 mice. Representative labeled neurons at 21 dpi are visible in the dentate GCL and have the morphology of GCs (Fig. 1a). Notably there was not significant co-labeling with doublecortin (DCX) that primarily labels abDGCs less than 21 days that are localized to the subgranular zone (SGZ) (Fig. 1a) 22 . Full field LED illumination (5 ms) of the slice with the abDGC centered in the field of view illuminated an area approximately 500 μm in diameter. We generated input-output curves by increasing the power of illumination and in all successive experiments the LED power was set to ensure photoactivation of GABA release was maximal ( Supplementary Fig. 1). By recording from RFP labeled abDGCs at different timepoints (days post injection, dpi) we measured the macroscopic GABA current in neurons at 7, 14, 21, and 28 dpi and in mature, unlabeled DGCs from the outer granule cell layer 29 . Optogenetic activation of PV interneurons with a single pulse of light elicited an inhibitory postsynaptic current (optoIPSC) in 69.2% of 7 dpi abDGCs (n = 13, 6 cells, animals respectively), and 100% of cells at later timepoints (Fig. 1d,e). The optoIPSC amplitude in PV-ChR2 mice increased over the first 3 weeks following differentiation but plateaued between 21 and 28 dpi before increasing further in mature, unlabeled DGCs ( Fig. 1d,f, Table 1). No postsynaptic response to activation of SST interneurons was detected in any of the abDGCs we recorded at 7 dpi (n = 9, 4), but the same stimulation elicited an optoIPSC in 92.9% (n = 14, 6) of 14 dpi abDGCs and 100% of cells recorded at later timepoints ( Fig. 2c,d). The optoIPSC amplitude increased as abDGCs matured, indicating an increase in input from SST interneurons during the first 4 weeks of development of abDGCs ( Fig. 2c,e, Table 1).
A comparison of the IPSC responses from PV and SST interneurons provided some interesting contrasts ( Supplementary Fig. 2). That we were able to evoke an IPSC in 7 dpi abDGCs in PV-ChR2 mice, but not SST-ChR2 mice indicates abDGCs receive input from PV interneurons prior to SST interneurons (Supplementary Fig. 2a,b). This is consistent with a previous study that found that optogenetic activation of SST interneurons did not evoke an IPSC in 4 dpi abDGCs, but activation of PV interneurons did elicit an IPSC 5 . At 14 dpi, the optoIPSC amplitude was larger in response to activation of PV than SST interneurons ( Supplementary Fig. 2c, Table 1). But at 28 dpi the optoIPSC amplitude was larger in response to activation of SST interneurons, indicating that developing abDGCs receive more input from PV interneurons early in development, and that SST input grows rapidly around 3 weeks after differentiation.

Quantal aIPSCs evoked from SST and PV interneurons.
To determine whether the average size of individual synaptic connections from the two interneuron types was also scaling with development we analyzed optogenetically-evoked asynchronous IPSCs (aIPSCs) by adding strontium chloride to the extracellular solution to desynchronize GABA release (Fig. 3a) 30 . We recorded aIPSCs at 21 and 28 dpi and in mature DGCs, aIPSC frequency was too low for reliable quantification of events at 14 dpi. We did not observe a significant change in aIPSC amplitude across ages of abDGCs or between recordings from PV-ChR2 and SST-ChR2 mice, indicating that the potency of individual synapses is relatively constant during this time ( Fig. 3a,b, Table 2). This also indicates that the increase in amplitude of optoIPSCs across these same time points results from an increasing number of synaptic inputs from both PV and SST interneurons as abDGCs mature. Using these results we were able to calculate an estimate of the number of functional synapses formed from each interneuron type onto developing abDGCs (synapse number ~ optoIPSC amplitude/aIPSC amplitude) (   www.nature.com/scientificreports/ to running wheels. PV-ChR2 or SST-ChR2 mice were single housed with running wheels beginning 3 weeks prior to surgery and were then injected with retrovirus to label newborn neurons when mice were 6-8 weeks old (Fig. 4a). We then performed targeted patch-clamp recordings from RFP-expressing cells at 7, 14, 21, and 28 dpi and recorded optogenetically-evoked IPSCs as described above. Mice ran an average of 7.8 km/day, consistent with what has been previously reported in C57Bl/6 mice 36 ( Supplementary Fig. 3). A greater number of recorded neurons at 7 dpi had functional inputs from PV interneurons in mice that had undergone voluntary wheel running (7 dpi control: 69.2%, n = 13, 6 to PV runner: 90%, n = 12, 3) (Fig. 4b,c). Running scaled the amplitude of optoIPSCs in abDGCs at all dpi recorded, but interestingly had no effect on PV input to mature DGCs (Fig. 4d, Table 1).
Wheel running failed to accelerate the formation of SST inputs to abDGCs at the earliest timepoints and there were no SST responses in any recorded cells at 7 dpi as with the non-runner group (7 dpi control: n = 9, 4 and 7 dpi runner: n = 12, 2). At 14 dpi responses were found in nearly all recorded cells [92.9% of controls (n = 14, 6) and 100% of runners (n = 16, 3)] (Fig. 5b). As with the PV inputs, running scaled the SST optoIPSC amplitude at all developmental timepoints in abDGCs. Recordings from abDGCs from runner mice demonstrated a larger SST optoIPSC amplitude than recordings from controls (Fig. 5a,c, Table 1). Interestingly, running normalizes the age-dependent differences in inputs from specific interneuron types observed at 14 and 28 dpi in controls. These findings indicate that running scales the GABAergic input that developing abDGCs receive from both PV and SST interneurons, but voluntary wheel running had no detectable effect on interneuron inputs to mature DGCs.
To determine whether running also affected the quantal amplitude of PV and SST synapses we again used strontium replacement to record desynchonized optogenetically-evoked aIPSCs from control mice and runners. We found that access to running wheels had no effect on the average quantal amplitude of IPSC from either PV or SST interneurons in 21 dpi, 28 dpi and mature GCs (Figs. 4e,f, 5d,e, Table 2). These results indicate that the potency of individual inhibitory synapses is not affected by running and that the scaling of the macroscopic GABA currents from SST and PV neurons to abDGCs reflects an increase in the number of synaptic contacts in mice that undergo voluntary wheel running which is evident from calculating the ratio of the average macroscopic current to quantal amplitude for each dpi (PV 21 dpi control: 25

Discussion
Adult born granule cells have intrinsic properties that make them more excitable than mature GCs, yet they remain only sparsely active because of their reduced excitatory synaptic innervation 37 . GABA synapses are the first to be established on post-mitotic DGCs as early as a few days after differentiation 5 . GABA is important for survival, dendritic development, glutamatergic synaptogenesis, and due to the relatively depolarized reversal potential, GABA can induce firing of young abDGCs 3,38 . www.nature.com/scientificreports/ While the importance of GABAergic input to the development of very young abDGCs is clear, a better understanding of how abDGCs are innervated during the first 4 weeks after differentiation will help us get a better idea of how these cells integrate into the mature hippocampal circuitry and how they contribute to hippocampal function. Here, we focused on determining the synaptic inputs to abDGCs from two prevalent interneuron types in the dentate gyrus and determined how running activity of the animals modified these connections. As expected, the optoIPSC amplitude from stimulation of both PV and SST interneurons increased across the first 4 weeks of maturation and was even larger in mature DGCs. Confirming the early input from PV interneurons 5 , we found PV mediated responses in a majority of abDGCs at the earliest post-mitotic time we tested (7 dpi), at which point there was no detectable contribution of SST interneurons. Comparison of the amplitudes of the optoIPSC from PV and SST interneurons demonstrated that PV interneurons made the largest contribution at 14 dpi, but at 28 dpi the maximal input from SST interneurons is larger than that from the PV cells. We found that the strength of individual synapses was relatively constant across development, and this allowed us to estimate the number of synaptic inputs from both SST and PV interneurons as abDGCs mature. While our study was ongoing, a study from Groisman and colleagues used a similar approach to map GABA input from PV and SST interneurons onto abDGCs 12 . That study focused on later timepoints after differentiation and concluded that full maturation does not occur until 8 weeks post-differentiation. Moreover, that study did not measure the true quantal size of GABAergic afferents and concluded that the increase in macroscopic currents over abDGC development resulted from scaling of individual synapses rather than synapse number. Our focus on earlier postmitotic timepoints compared both the macroscopic and true quantal stimulated events from PV and SST inputs and demonstrates that there is an increase in the number of synaptic inputs from each of these interneurons but the relative contribution changes as abDGCs mature.
Adult neurogenesis is a highly regulated process that can be influenced by genetic factors, age, environment, and activity. Voluntary wheel running in mice increases both proliferation and survival of abDGCs, leading to an overall increase in the number of newborn neurons in the dentate gyrus 34 . Running also improves performance on pattern separation and spatial learning tasks, indicating a functional change in the hippocampal circuitry following running [31][32][33] . However, it is worth noting that there may be no causal relationship between neurogenesis and the running mediated effects on anxiety and learning 39,40 . For instance, one recent study found environmental enrichment (including running wheels) affected spatial learning even after irradiation and ablation of adult hippocampal neurogenesis 41 . So, while the behavioral effects of running may be due to an increased number of newborn neurons, they could also reflect additional alterations in the maturation and integration of abDGCs in mice that have been housed with a running wheel. Voluntary wheel running leads to increased dendritic length and branching in 7 and 16-17 day-old abDGCs and increased density of dendritic spines in 21-day-old abDGCs [42][43][44] . Running also increases seizure-induced expression of the immediate early gene Arc in 21-day-old abDGCs, indicating that running promotes the integration of newborn neurons into the existing hippocampal circuit 45 . Functionally, running reduces the input resistance and increases the percentage of abDGCs with GABAergic input in 7 dpi abDGCs in rats 46 and increased and the number of NMDA only synapses in 7 dpi abDGCs in mice 43 . Interestingly, the number of anatomically traced presynaptic neurons to abDGCs was not affected by running, which together with our finding that running increases the optoIPSC amplitude suggests that the primary effect of running is to increase the number of synaptic contacts per presynaptic neuron onto very young abDGCs 43 . On the other hand, in 5-week-old abDGCs running decreased connectivity between local hippocampal neurons without affecting mIPSC frequency or amplitude 47 . Thus, there may be differential agedependent effects on connectivity between abDGCs and local circuit interneurons that are caused by running. The present study included animals of both sexes, however it is worth noting that proliferation and survival of abDGCs may be differentially modulated by stress and learning in males and females 48 .
Our approach enabled us to focus on the inputs to abDGCs from two defined interneuron types and perform a comprehensive analysis of the effects of running on functional connectivity of abDGCs during the first 4 weeks after differentiation. We found that there was a scaling of all GABA inputs to abDGCS from both SST and PV interneurons, but quantal size did not scale in parallel. The most parsimonious explanation for these results would be that the number of synaptic inputs to abDGCs (rather than the potency of individual synapses) scales after voluntary wheel running. Given that previous studies found no effect of running on the numbers of connected local interneurons this might suggest that individual axons form more synaptic connections with abDGCs in mice that are runners. Future tracing studies with high resolution neuroanatomy would be required to confirm this conclusion. Interestingly we found no effect of running on GABA input to mature cells. A recent study used a Fos-Trap approach to identify mature granule cells that are active during a single 2-h bout of running and found considerable plasticity of their excitatory inputs from the entorhinal cortex 49 . In our analysis we were not able to specifically identify DGCs that might have been active during long term running, but it is possible that PV and SST inputs might be differentially modulated in active DGCs compared to the total population. PV interneurons are the most well-characterized source of GABA inputs to abDGCs. GABA release from PV interneurons maintains quiescence of adult neural stem cells (aNSCs) while in vivo optogenetic or chemogenetic activation of PV interneurons increases the survival and promotes dendritic growth in abDGCs 5,11,50 . PV interneurons in the dentate are primarily basket cells with somata in the granule cell layer and axons forming perisomatic synapses onto mature DGCs 7,8 . Much less is known about SST inputs to abDGCs, although Song and colleagues reported that they were unable to detect SST inputs to less than 1-week-old abDGCs using optogenetic activation 5 . SST interneurons in the dentate are primarily HIPP cells that form synapses on the distal dendrites of mature DGCs 7,8 . Therefore, it is likely that the temporal sequence of innervation from these distinct populations of interneurons is determined in part by the laminar organization of interneuron axons in the dentate gyrus. At the earliest timepoint, when dendrites have not extended beyond the granule cell layer, abDGCs receive perisomatic input from PV interneurons with axons in the granule cell layer. As the abDGCs mature and extend their dendrites into the molecular layer, they encounter the axons of SST interneurons and form functional synapses.
Scientific Reports | (2020) 10:17522 | https://doi.org/10.1038/s41598-020-74385-2 www.nature.com/scientificreports/ In summary, we characterized the temporal sequence of development of inputs from SST and PV interneurons onto abDGCs during the first 4 weeks after differentiation and determined that voluntary wheel running scales these GABAergic connections. Our data are consistent with an age and activity dependent increase in the number of functional synapses from both PV and SST interneurons which could reflect an increase in the number of synapses formed per axon with an individual abDGC. Voluntary running does not accelerate the process of connectivity as there was no shift in when abDGCs receive SST input. Together these studies further our understanding of the how abDGCs mature and are integrated into the hippocampal network.