Activation of 5-HT7 receptors in the mouse dentate gyrus selectively enhances GABAergic inhibition of hilar mossy cells without affecting plasticity at the perforant path synapse

Background The study examined the effects of 5-HT7 receptor activation on GABAergic transmission within the dentate gyrus and plasticity at the glutamatergic perforant path input. Methods Immunofluorescence imaging was performed using transverse hippocampal slices from transgenic mice expressing GFP under the Htr7 promoter. This was followed by whole-cell patch clamp electrophysiological recordings of spontaneous inhibitory postsynaptic currents recorded from dentate granule cells and hilar mossy cells — two glutamatergic neuron types present in the dentate gyrus. Extracellular recordings of field excitatory postsynaptic potentials were then performed to assess whether 5-HT7 receptor activation influenced theta burst stimulation-evoked plasticity of the perforant path synaptic input. Results It was found that parvalbumin and somatostatin interneurons in the dentate gyrus expressed GFP. Activation of 5-HT7 receptors increased GABAergic transmission targeting mossy cells but not granule cells. However, there was no effect of 5-HT7 receptor activation on perforant path plasticity either with intact or blocked GABAA receptor signaling. Conclusion The presence of 5-HT7 receptors in a subset of parvalbumin and somatostatin interneurons in the mouse dentate gyrus could mean that they are involved in the inhibitory control of dentate gyrus activity, although the effect seems to be confined to mossy cells and did not translate to changes in perforant path plasticity. Further experiments are needed to fully elucidate the functional role of these receptors in the dentate gyrus.


Introduction
GABAergic transmission is a potent regulator of the plasticity of excitatory synapses [1,2].Although the history of the research on long-term synaptic plasticity of perforant path synapses in the dentate gyrus (DG) goes back more than 50 years [3], its modulation by local DG interneurons remains incompletely understood.Perforant path synapses in the DG are capable of bidirectional plasticity (long-term potentiation, LTP, and long-term depression, LTD [4]).The primary neuronal type of the DG are small, glutamatergic granule cells forming the granular layer but in addition to several types of GABAergic interneurons [5], that are located mainly along the base of the granular layer and in the hilus, the DG contains also less numerous, large, glutamatergic mossy cells, residing in the hilus [6].
All these different neuronal types form within the DG a complex network of synaptic connections that functions as a gate for further information flow into the hippocampus proper [7,8].
An important factor involved in LTP modulation in the DG is the serotonergic innervation [9], acting via 5-HT 1A [10], 5-HT 2C [11] and 5-HT 4 [12] receptors.The 5-HT 7 receptor (5-HT 7 R), the last member of the 5-HT receptor family to be identified [13][14][15], is abundant in the DG as well [16], however, its involvement in the induction of synaptic plasticity in the DG has not been investigated.Activation of 5-HT 7 Rs enhances neuronal membrane excitability and/or spiking behavior in CA1 pyramidal neurons [17,18] so it is conceivable that 5-HT 7 R activation might alter the conditions for the induction of synaptic plasticity in the DG.Thus, to examine the pattern of 5-HT 7 R expression in the DG, in the present study we used a transgenic mouse line expressing GFP under the Htr7 promoter.Following up on immunofluorescence results showing that GFP related to 5-HT 7 R expression is present predominantly in DG inhibitory interneurons, further experiments focused on elucidating the functional consequences of 5-HT 7 R activation on GABAergic control of the two excitatory cell types of the DG -granule cells and mossy cells.The final set of experiments aimed to characterize the potential effects of 5-HT 7 R activation on the plasticity of the medial perforant path input to the DG.

Animals
Male wild-type and transgenic C57BL/6J mice were maintained on a 12/12 h light/dark schedule with standard food and tap water available ad libitum.The 5-HT 7 -GFP strain used in the study -Tg(Htr7-EGFP)ST29Gsat/Mmucd -was generated as part of the GENSAT project [19].
The proportion of GFP-expressing parvalbumin (PV + ) and somatostatin (SOM + ) interneurons was assessed using microphotographs from 3 mice (5-6 sections per animal).Images were automatically processed with FIJI/ImageJ scripts [20].The workflow consisted of per-channel background subtraction (rolling ball method), followed by automated local thresholding.Bernsen, IsoData and moments segmentation algorithms were used for PV, SOM and GFP signals, respectively, based on empirical testing in example images.Thresholded images were size-filtered to retain objects corresponding to intact and fully stained neuronal somata.Colocalization was assessed separately in each optical slice using the pixel value product method, followed by manual verification and, finally, z-projection and cell counting.

Whole-cell patch clamp recordings of spontaneous inhibitory postsynaptic currents
Slices were placed in the recording chamber and superfused at 4-5 ml/min with warm (32 ± 0.5°C) recording ACSF composed of (in mM): 132 NaCl, 2 KCl, 2.5 CaCl 2 , 1.3 MgSO 4 , 1.2 NaH 2 PO 4 , 24 NaHCO 3 and 10 glucose, continuously bubbled with a mixture of 95% O 2 and 5% CO 2 .Neurons were visualized on a Nikon Eclipse Fn1 upright microscope (Nikon Europe B.V.) equipped with 900 nm IR DIC optics, a 40×0.8NA water immersion objective and sCMOS camera (CellCam Kikker, Cairn Research), along with a computer-controlled piezo xy stage and micromanipulators (Sensapex).Patch pipettes were pulled from 1.5 mm OD/0.86 mm ID borosilicate glass capillaries (Sutter Instruments) using a P87 puller (Sutter Instruments).Spontaneous inhibitory postsynaptic currents (sIPSCs) were recorded from dentate granule cells as well as hilar mossy cells in whole-cell voltage clamp mode using a Multiclamp 700B amplifier (Molecular Devices).Signals were filtered at 2 kHz and digitized at 20 kHz with a National Instruments PCIe-6353 data acquisition card controlled by the open-source ACQ4 software package [22].Biocytin (0.3%, HelloBio) was included in the recording pipette solution for later verification of cell morphology.A high-chloride recording pipette solution was used for recording sIPSCs and contained (in mM): 130 KCl, 5 NaCl, 0. dentate granule cells had a small ovoid cell body and were characterized by a relatively hyperpolarized resting membrane potential recorded right after break-in (typically ~−80 mV).They also displayed a typical membrane resistance of 200-300 MΩ and low intrinsic excitability [23].Hilar mossy cells, on the other hand, typically had large, triangular or trapezoidal cell bodies.Upon break-in, they were characterized by a large cell capacitance, membrane resistance of 150-300 MΩ, relatively higher excitability and high magnitude, frequent spontaneous synaptic activity [24].After recording cells were further confirmed to belong to these two neuronal populations on the basis of dendritic morphology revealed by biocytin labeling.

Morphological verification of recorded cells
Slices were fixed in 4% PFA in PBS for 1 h.After several washing steps, they were incubated in PBS containing 0.1% TX-100 and Alexa Fluor 647-conjugated streptavidin (1:500, Invitrogen) for 2 h at RT on an orbital shaker.After washing in PBS and counterstaining with DAPI (Invitrogen) slices were mounted on glass slides and coverslipped using the Vectashield mounting medium (Vector Laboratories).Verification of individual neuron morphology was performed after acquiring z-stacks taken with a 20×0.8NA air apochromatic objective on a Zeiss Axio Imager Z2 fluorescence microscope using the Apotome 2 module (Carl Zeiss).

Extracellular recordings of medial perforant path-dentate gyrus plasticity
Extracellular recordings were performed using the same ACSF and hardware as the patch clamp recordings.The WinLTP software package was used for recording of plasticity experiments [25].In a subset of recordings 100 µM picrotoxin (HelloBio) was added to the ACSF to block inhibitory transmission via GABA A receptors.Field extracellular excitatory postsynaptic potentials (fEPSPs) were elicited using a concentric bipolar stimulation electrode (FHC) placed in the middle of the dentate gyrus molecular layer of the suprapyramidal blade, corresponding to the medial perforant path.The recording electrode (3-5 MΩ patch clamp recording pipette filled with recording ACSF) was placed approximately 200-400 µm away from the stimulating electrode (Fig. 4).Monosynaptic fEPSPs were elicited every 20 s with 200 µs current pulses using a constant-current stimulus isolation unit (WPI).After a 10-15 min stabilization period, the stimulation intensity was adjusted to evoke fEPSPs with a rising slope approximately 50% of the maximum value (typically 20-30 µA).A theta-burst stimulation (TBS) protocol was used to induce plasticity after recording a 20 min baseline period.TBS consisted of a series of 5 high-frequency 100 Hz 200 µs pulses repeated five times every 200 ms.This was repeated 5 times every 10 s.The TBS protocol was chosen for a physiologically plausible LTP induction mechanism [26].After TBS the recordings continued for 1 h with baseline stimulation intensity.

Pharmacological activation of 5-HT 7 receptors
Drugs were bath-applied in final concentrations for both patch clamp and extracellular recordings.
LP-211 has been extensively used in functional studies of 5-HT 7 Rs [27][28][29][30][31].Control and/or baseline recordings were performed in 0.1% DMSO alone.The highly selective 5-HT 7 R antagonist SB 269970 (2 µM, HelloBio) was used in a subset of recordings to confirm the selectivity and specificity of any LP-211-evoked effects.The slice was perfused with the antagonist for at least 15 min before commencing recordings.
In patch clamp experiments the effect of LP-211 on GABAergic transmission was evaluated relative to baseline levels in each cell.After obtaining the whole-cell configuration and verifying the electrophysiological phenotype of the recorded cell in current clamp mode using a series of hyper-and depolarizing current pulses, the recording was switched to voltage clamp at −76 mV and after several minutes of stabilization a 4 min baseline period of sIPSC activity was recorded in ACSF containing 10 µM NBQX and 0.1% DMSO.Next, the perfusion line was switched to ACSF containing 10 µM NBQX with 1 µM LP-211 in 0.1% DMSO and the slice was perfused for 12 min to allow the compound to activate the receptors and equilibrate before recording another 4 min period.Access resistance had to stay below 20 MΩ and not change by more than 20%, otherwise the recording was discarded.For extracellular experiments, 1 µM LP-211 was added to the ACSF and allowed to perfuse the slice approximately 1 h before the planned induction of plasticity, i.e. 30 min before starting the recording.In control recordings 0.1% DMSO alone was used within the same time frame.

Electrophysiology data analysis and statistics
For patch clamp experiments and sIPSC detection, the 4 min "baseline" and "LP-211" segments were analyzed to detect sIPSCs using a deconvolution algorithm [32] implemented in the Eventer software package [33].Extracellular recordings were analyzed using WinLTP software [25], where three consecutive fEPSP traces (stimulation every 20 s) were averaged to obtain one low-noise reading of fEPSP slope for every minute of the recording.The maximum slope was calculated from the rising slope of the fEPSP in a sliding 1.5 ms time window.The data were then normalized to the mean slope of the 20 min baseline period for each recording separately.The final values for plasticity magnitude were thus expressed as percentages of the mean baseline slope level.Statistical data analysis and graphing of the results was performed using the R programming language and RStudio IDE [34,35].R packages used were the tidyverse collection [36] and patchwork [37].Statistics on sIPSC frequency and amplitude before and after LP-211 administration were performed using a two-tailed paired samples t-test.Due to the bidirectionality of plasticity results and resulting lack of normality, extracellular recordings were analyzed by comparing whole distributions of plasticity magnitudes between the control and experimental groups using the nonparametric two-sample Kolmogorov-Smirnov test.The resulting distributions were plotted as cumulative probability histograms using the ecdf() R function.For all statistical procedures, inference was two-tailed and performed with the arbitrary significance threshold set at p < 0.05.Extracellular summary fEPSP data were plotted as mean ± SEM.

Htr7 promoter-dependent GFP is expressed in PV and SOM interneurons in the mouse dentate gyrus
The observed GFP immunostaining pattern in 5-HT 7 R-GFP mice consisted of sparse, but distinct, labeling of neuronal cell bodies, which, in transverse sections of the dentate gyrus, were primarily located in the hilus, with only single immunopositive cells present in the granule and molecular layers.These GFP + cells were presumptive inhibitory interneurons predominantly co-expressing SOM or PV (Fig. 1), but typically not calretinin or pro-cholecystokinin.For the latter two markers, only a handful of cells with weak fluorescent signal corresponding to putative GFP expression were found (not shown).Quantification of PV and SOM cell bodies revealed GFP co-expression was widespread in these populations, as double-immunopositive cells constituted 59% (223/379) of all PV + and 70% (271/382) of all SOM + neurons.

Activation of 5-HT 7 receptors does not affect the theta-burst-evoked plasticity of the medial perforant path-granule cell synapse
To characterize potential effects of 5-HT 7 R activation on the plasticity of the medial perforant path input to the DG, fEPSPs evoked by stimulation of the medial perforant path were recorded.In conditions with intact GABAergic signaling TBS failed to induce LTP with the majority of recordings resulting in no LTP or even LTD and the mean plasticity magnitude of 85.3 ± 8.32% relative to baseline (i.e.100%; Fig. 4 A1-A2).Activation of 5-HT 7 Rs by preincubating slices with ACSF containing 1 µM LP-211 did not change this distribution of plasticity magnitudes (D 17,17 = 0.23529, p = 0.75061, exact two-sample Kolmogorov-Smirnov test), with the mean plasticity magnitude in the LP-211 group of 87.5 ± 5.13%.The exact same plasticity induction protocol resulted in a robust LTP when 100 µM picrotoxin (PTX) was added to the ACSF to block fast GABAergic inhibition (Fig. 4 B1-B2), with the mean plasticity magnitude of 153 ± 24%.This time the distribution of plasticity magnitudes strongly favored LTP over LTD, however, when LP-211 was included in the recording ACSF the distribution of plasticity magnitudes remained unchanged (D 19,13 = 0.27126, p = 0.51867, exact two-sample Kolmogorov-Smirnov test), with the mean plasticity magnitude of 166 ± 28% (Fig.

Discussion
The results of this study indicate that in the dentate gyrus 5-HT 7 Rs are expressed predominantly on inhibitory interneurons of the hilar region.However, activation of these receptors selectively enhances inhibitory synaptic input to mossy cells but not granule cells.
It has been shown that mossy cells function as local circuit integrators and exert modulatory influence on dentate granule cells as well as the CA3 through back-projecting pathways [38,39].They receive considerable excitatory synaptic input from granule cells' massive specialized synaptic boutons (mossy terminals) apposed to thorny excrescences present on their cell bodies and proximal dendrites giving them their "mossy" appearance [40].In turn, mossy cells are monosynaptically coupled to granule cells both ipsi-and contralaterally.However, this connection is only weakly excitatory, prone to synaptic failures and strongly masked by GABAergic inhibition.Therefore it is believed that the primary effect of mossy cell activation on granule cells is in fact non-direct and inhibitory, via the excitation of local GABAergic interneurons [41], with some studies suggesting that this interaction can become reversed in pathological states of hippocampal circuit activity such as those resulting from acute trauma and epileptogenesis [42,43].Furthermore, direct excitatory connections have also been discovered between individual mossy cells, which could in theory further consolidate their influence on granule cell activity [44].
Given the considerable glutamatergic drive via mossy terminals, the activity of mossy cells has to be tightly controlled by various subtypes of dentate and hilar interneurons [45][46][47].We have not identified the interneurons that form the inhibitory input to mossy cells, however, it is likely that this input is at least partially provided by hilar somatostatin-positive cells [40,48].Activation of 5-HT 7 Rs by LP-211 and the subsequent increase in GABAergic transmission recorded from mossy cells is most probably a direct result of increased spiking activity of these interneurons, as 5-HT 7 Rs are known enhancers of neuronal excitability [17,18,49,50].However, it is important to note that somatostatin interneurons also regulate granule cell activity and the question remains as to why we did not record increased GABAergic transmission in those cells.One possible explanation is that the majority of somatostatin interneurons targeting dentate granule cells are the so-called hilar perforant path-associated (HIPP) interneurons, which are characterized by weak, slow and unreliable inhibition of granule cell distal dendrites in the outer molecular layer, in contrast to strong, fast and reliable perisomatic inhibition coming from parvalbumin basket cells [47].Any increase in somatostatin interneuron spiking activity would have to be strong enough to overcome the weak synaptic coupling in order to be reliably detected by somatic granule cell recordings.In the case of mossy cells, however, their thick proximal dendrites [51] are presumably characterized by lower axial resistances compared to thin granule cell dendrites and thus are more conductive to postsynaptic currents.
Moreover, recently, a second subpopulation of dentate somatostatin interneurons was characterized in addition to HIPP cells -the so-called hilar interneurons (HILs), whose axon collaterals extensively innervate the dentate hilar region where mossy cell somata are located [48].Further research is needed to characterize the degree of connectivity between these two neuron types, however activation of 5-HT 7 Rs on HIL cells could in theory explain the increased sIPSC frequency recorded in mossy cells.
There are limited sources in the literature concerning if and how parvalbumin-positive interneurons innervate hilar mossy cells.However, the majority of parvalbumin interneurons in the dentate gyrus are basket cells characterized by large triangular cell bodies in the subgranular layer and extensive, basket-like branching of axon collaterals around granule cell somata, with few axons present in the hilus [47].As for the rest of the parvalbumin interneuron cell population, not much is known about their postsynaptic targets and these may in fact include mossy cells.However, it is important to note that in general, hippocampal parvalbumin interneurons have different electrophysiological characteristics compared to their somatostatin counterparts.Although they are usually fast-spiking neurons, they have a lower input resistance and a more hyperpolarized resting membrane potential [47], and are often quiescent in standard ex vivo slice recording conditions.Therefore activation of 5-HT 7 Rs located on these cells might not provide a depolarization strong enough for spontaneous spiking to occur, at least not in standard, baseline conditions such as those employed in this ex vivo study.Given that we demonstrate robust 5-HT 7 R-GFP labeling of dentate parvalbumin interneurons, future experiments would need to study its functional implication in vivo or during in vivo-like activity patterns evoked pharmacologically or electrically, where different classes of GABAergic interneurons would be driven to spiking.
One such attempt was made in this study, where we performed extracellular recordings measuring the synaptic plasticity of the medial perforant path input to the dentate gyrus.A theta-burst stimulation protocol was chosen (Fig. 4), as it reliably evokes hippocampal LTP and is physiological in nature (theta frequency hippocampal oscillations) in contrast to the more commonly used high frequency stimulation paradigm [26,52,53].The dentate gyrus perforant path input is under strong GABAergic control and without blocking GABA A receptors it is highly resistant to LTP and even prone to LTD, as shown in our experiments.However, it was in principle possible for the 5-HT 7 R-evoked increase in GABAergic inhibition of mossy cells to shift the distribution of plasticity magnitudes.Somewhat surprisingly, preincubation of slices in ACSF containing LP-211 did not affect the distribution of plasticity magnitudes either with intact or blocked fast GABAergic transmission.One possible explanation for this lack of effect is that, as we demonstrated, only GABAergic transmission to mossy cells was affected by 5-HT 7 R activation, with no changes in sIPSCs recorded from granule cells.Mossy cells are potent regulators of granule cell activity, however they usually form translaminar or even contralateral connections, which are inevitably severed in transverse slice preparations typically employed to study hippocampal plasticity [40].Another possibility is that 5-HT 7 R-expressing interneurons are simply not recruited strongly enough by the TBS protocol for the 5-HT 7 R activation to be able to reliably influence plasticity.Especially since GABAergic transmission targeting granule cells -whose dendrites are the source of fEPSP signal in extracellular recordings of perforant path plasticity -did not change following 5-HT 7 R activation.
In spite of decades of research, serotonergic control of hippocampal activity remains an important research topic, perhaps because of the multitude of serotonin receptors and their associated molecular pathways, not to mention their role in various brain circuit pathologies.To date, many serotonin receptor subtypes have been identified in the dentate gyrus including, but not limited to 5-HT 1A , 5-HT 1B , 5-HT 1D , 5-HT 2A , 5-HT 2C , 5-HT 3 , 5-HT 4 , 5-HT 5A , 5-HT 6 and 5-HT 7 [12,[54][55][56][57][58][59][60][61][62].This study, while providing preliminary insights into the function of 5-HT 7 Rs in the dentate gyrus, in fact resulted in more questions than answers, for example about the functional consequences of 5-HT 7 R activation on dentate gyrus circuit activity and synaptic plasticity in vivo.Or the precise somatostatin and parvalbumin interneuron subtypes which express these receptors.Successfully answering these and other questions in future studies could lead the scientific community to a better understanding of the role of the serotonin system in health and disease.

3
CaCl 2 , 2 MgCl 2 , 10 HEPES, 10 phosphocreatine-Na 2 , 5 Na 2 -ATP, 0.4 Na-GTP and 0.3 EGTA.Osmolarity and pH were adjusted to 290 mOsm and 7.2 respectively.Recording pipettes had open tip resistances of approximately 4-6 MΩ when filled with this solution.The AMPA/kainate receptor blocker NBQX (10 µM, HelloBio) was included in the recording ACSF to isolate GABAergic synaptic activity.A holding potential of −76 mV was used to reliably record sIPSCs with few/no action potential currents present.Cell identification was based on anatomical location (mossy cells -dentate gyrus hilar region; granule cells -the outer third part of the granule cell layer in the supra-pyramidal blade of the ventral/intermediate dentate gyrus), electrophysiological properties and cell morphology.Typical

Figure 1 .
Figure 1.Representative composite frame image (A) and corresponding single channel images of a mouse dentate gyrus section stained for GFP (green; B), SOM (yellow; C) and PV (red; D).Cell bodies immunopositive for SOM-GFP and PV-GFP indicated with solid arrowheads and contour arrowheads, respectively.

Figure 4 .
Figure 4. Effects of LP-211 on the magnitude of medial perforant path-dentate gyrus plasticity.(A) Anatomical diagram showing the placement of recording and stimulation electrodes.Abbreviations: EC -entorhinal cortex; CA1 -cornu Ammonis 1; DG -dentate gyrus; MPP -medial perforant path.(B1) Effects of LP-211 preincubation on fEPSP slope (%) before and after TBS.Inset: raw fEPSP traces of a representative control recording before (lighter green) and after (darker green) TBS.(B2) Cumulative probability histograms showing that preincubation of slices with LP-211 did not affect the distribution of plasticity magnitudes (p = 0.75061, exact two-sample Kolmogorov-Smirnov test).(C1) Effects of LP-211 preincubation on fEPSP slope (%) before and after TBS with 100 µM picrotoxin (PTX) present in the ACSF to block GABAA receptors.Inset: raw fEPSP traces of a representative control recording before (lighter green) and after (darker green) TBS.(C2) Cumulative probability histograms showing that preincubation of slices with LP-211 with PTX present in the ACSF did not affect the distribution of plasticity magnitudes (p = 0.51867, exact two-sample Kolmogorov-Smirnov test).