Selective Overexpression of Collybistin in Mouse Hippocampal Pyramidal Cells Enhances GABAergic Neurotransmission and Protects against PTZ-Induced Seizures

Abstract Collybistin (CB) is a rho guanine exchange factor found at GABAergic and glycinergic postsynapses that interacts with the inhibitory scaffold protein, gephyrin, and induces accumulation of gephyrin and GABA type-A receptors (GABAARs) to the postsynapse. We have previously reported that the isoform without the src homology 3 (SH3) domain, CBSH3–, is particularly active in enhancing the GABAergic postsynapse in both cultured hippocampal neurons as well as in cortical pyramidal neurons after chronic in vivo expression in in utero electroporated (IUE) rats. Deficiency of CB in knock-out (KO) mice results in absence of gephyrin and gephyrin-dependent GABAARs at postsynaptic sites in several brain regions, including hippocampus. In the present study, we have generated an adeno-associated virus (AAV) that expresses CBSH3– in a cre-dependent manner. Using male and female VGLUT1-IRES-cre or VGAT-IRES-cre mice, we explore the effect of overexpression of CBSH3– in hippocampal pyramidal cells or hippocampal interneurons. The results show that: (1) the accumulation of gephyrin and GABAARs at inhibitory postsynapses in hippocampal pyramidal neurons or interneurons can be enhanced by CBSH3– overexpression; (2) overexpression of CBSH3– in hippocampal pyramidal cells can enhance the strength of inhibitory neurotransmission; and (3) these enhanced inhibitory synapses provide protection against pentylenetetrazole (PTZ)-induced seizures. The results indicate that this AAV vector carrying CBSH3– can be used for in vivo enhancement of GABAergic synaptic transmission in selected target neurons in the brain.

Studies of CB knock-out (KO) mice reveal that loss of CB results in severe disruption to clustering of gephyrin and GABA A Rs in certain brain regions, including the hippocampus (Papadopoulos et al., 2007(Papadopoulos et al., , 2008. This results in decreases in GABAergic neurotransmission, deficiencies in synaptic plasticity, increased levels of anxiety, and impaired spatial learning in CB KO mice. In humans, a mutation in ARHGEF9, the gene that encodes CB, was first identified in a patient with epileptic encephalopathy (Harvey et al., 2004) but additional pathogenic mutations have since been identified in patients, contributing to a spectrum of ailments that include epilepsy, hyperekplexia, intellectual disability, anxiety, autism, and schizophrenia (Marco et al., 2008;Kalscheuer et al., 2009;Lesca et al., 2011;Shimojima et al., 2011;Lionel et al., 2013;Long et al., 2015;Bhat et al., 2016;Klein et al., 2017;Wang et al., 2018;Aarabi et al., 2019;Chiou et al., 2019;Yao et al., 2020).
Overexpression of gephyrin or GABA A Rs does not significantly affect synaptic GABA A R or gephyrin clustering in cultured hippocampal neurons (Chiou et al., 2011). However, it has been demonstrated that overexpression of CBSH3-in cultured hippocampal neurons induces the formation of large synaptic gephyrin and GABA A R clusters (Kalscheuer et al., 2009;Chiou et al., 2011;Tyagarajan et al., 2011;Soykan et al., 2014). In the present study, we examined the effect of overexpressing a constitutively active isoform of CB, CB2SH3-, in selected cells of the dorsal hippocampus in adult mice. By overexpressing CB2SH3-in specific neuron-types, and in adulthood, we aimed to address whether the increase in CB2SH3-expression enhances the postsynaptic accumulation of gephyrin and GABA A Rs, GABAergic synaptic transmission in the developed hippocampus, and whether it can effectively protect against pentylenetetrazole (PTZ)induced acute seizures.

Animals
All animals included in this study were treated following protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Connecticut and follow National Institute of Health guidelines. Both male and female mice of VGAT-IRES-Cre (JAX: Slc32a1tm2(cre)Lowl/J; stock #016962) or VGLUT1-IRES2-Cre (JAX: B6;129S-Slc17a7tm1.1(cre) Hze/J; stock #023527) background were used.

Tissue preparation, immunohistology, and image acquisition
Three weeks after unilateral viral injection, mice were deeply anesthetized with an overdose of ketamine/xylazine (60/8 mg/kg) and cardiac perfusion was performed with 0.12 M phosphate buffer (PB; pH 7.4) followed by 4% PLP fixative in 0.12 M PB (4% paraformaldehyde, 1.37% lysine, and 0.21% sodium periodate). Brains were cryoprotected with 30% sucrose, frozen, and sectioned with a freezing microtome (25 mm thick) and stored in 0.02% sodium azide in 0.1 M PB, pH 7.4 at 4°C. Immunofluorescence was performed as described elsewhere (Li et al., 2005a;Fekete et al., 2017;Miralles et al., 2020). Briefly, free-floating brain sections were incubated with 5% normal goat serum, 0.3% Triton X-100 in 0.1 M PB for 1 h at room temperature. Sections were then incubated for 2 d in primary antibody cocktails raised in different species in 2% NGS/0.3% Triton X-100 in 0.1 M PB at 4°C. Sections were washed and incubated in a mixture of fluorophore-conjugated secondary antibodies in 2% NGS/0.3% Triton X-100 in 0.1 M PB for 1 h at room temperature. When necessary, a DAPI counterstain (200 nM) was applied following a wash after secondary incubation. Sections were then washed again and mounted onto gelatin-coated glass slides with Prolong Gold Antifade mounting solution (Invitrogen). Confocal images of brain sections were acquired using an A1R laser scanning confocal microscope (Nikon Instruments) with a Plan Apo VC 60Â/1.4 oil immersion objective. For the quantifications presented in Figure 1. The images were collected with a LSM 800 confocal laser-scanning microscope (Zeiss) with a 63Â/ 1.4 oil-immersion objective. The pinhole was set at 1.2 Airy units. Images were collected as single optical sections. The low-magnification images of brain sections for EGFP/ mCitrine/DAPI were collected by epifluorescence with a Plan Fluor 10Â/0.3 objective.
Individual slices containing the CA1 hippocampus were transferred to a recording chamber mounted on a fixed-stage microscope (Olympus BX5.1WI) and perfused continuously (;2 ml/min) with normal Ringer's solution (equilibrated with 5% CO 2 ; pH 7.3). Transfected CA1 hippocampal neurons were visually identified by fluorescent signal from EGFP or mCitrine. All recordings were made with an Axopatch 200B patchclamp amplifier, digitized with a Digidata 1322A A/D converter, and recorded using pCLAMP 10.0 software (Molecular Devices). Recordings were obtained at room temperature (;22°C) with patch electrodes pulled from borosilicate glass capillaries (Harvard Apparatus) on a twostage puller (P-97; Sutter Instrument) to a DC resistance of 5-7 MV when filled with a pipette solution containing the following: 130 mM CsCl, 8.5 mM NaCl, 5 mM HEPES, 4 mM MgCl 2 , 4 mM Mg2-ATP, 0.3 mM Na-GTP, and 0.6 mM EGTA. Electrode tips were coated with Sylgard 184 (Dow Corning). A gigaohm seal (!1 GV) was achieved using a cell-attached voltage-clamp configuration with holding potential matched to the resting membrane potential (V hold = À70 mV) and with no current generated by the amplifier (I amp = 0 pA). Whole-cell access ( 20 MV) was obtained to characterize spontaneous synaptic input. Spontaneous mIPSCs were characterized in the presence of bath applied CNQX (10 mM) and TTX (0.5 mM), and events were confirmed with bath application of picrotoxin (100 mM). mIPSCs were analyzed using Clampfit 10.7 and detected with characteristic kinetics (fast rising phase followed by a slow decay). Each automatically detected event was also visually inspected to exclude false positives. Recordings were discarded if Ra varied 10% were injected into CA1 of VGLUT1-IRES-cre mice, resulting in expression of mCitrine or EGFP in the glutamatergic pyramidal cell, and images were taken with epifluorescence microscopy (A-E) or confocal microscopy (F, G). In mCitrine-HACB2-AAV-injected mice, the transduced pyramidal cells expressing during an experiment, and capacitance and Ra compensation (70%) were used to minimize voltage errors.

Behavioral study
Preliminary trials were performed on naive mice to determine dose of PTZ that would reliably induce acute seizures while minimizing mortality. Three weeks after bilateral viral infusion, mice were subjected to behavioral study. Mice were given a single injection of PTZ (40 mg/ kg, i.p.) to induce acute seizures. The mice were then placed into a cage and behavior was observed. Seizures were scored according to a modified Racine scale (Racine, 1972;Lüttjohann et al., 2009;Singh et al., 2016) with the following scores: 1, sudden behavioral arrest; 2, facial jerking; 3, myoclonic jerks; 4, clonic seizure, sitting; 5, clonic, tonic-clonic seizure, lying on belly; 6, clonic, tonic-clonic seizure with loss of posture and wild jumping; with notation of time of onset of each intensity of seizure and tallying of maximal seizure intensity in 2-min bins. Seizure behavior typically subsides by 30 min after PTZ administration, so animals were then returned to their home cages. Animals were killed, brains collected, and sections examined by epifluorescence microscopy to confirm appropriate transduction of hippocampal CA1.

Experimental design and statistical analysis
All image analysis was performed using ImageJ (RRID: SCR_003070). For analysis of immunofluorescence, brain sections were analyzed from mice injected with HACB2-AAV (VGLUT1-cre: n = 6, VGAT-cre: n = 4) or EGFP-AAV (VGLUT1-cre: n = 6, VGAT-cre: n = 4). For each animal, each quantified value is the average of 10-15 measurements [regions of interest (ROIs) or individual interneurons], as explained below. Because of the densely packed nature of transduced CA1 pyramidal cells in VGLUT1-cre mice, ROIs were used to measure impact of viral transduction. In each section, five images were taken per hemisphere, and three ROIs with areas of 500 mm 2 were created within stratum pyramidale (SP) of CA1 in each to determine changes in perisomatic synapses. These values were averaged to present a single value for each animal. This was repeated within stratum radiatum (SR) and stratum oriens (SO) to measure changes in dendritic synapses. To determine changes in synaptic proteins in transduced interneurons of CA1 (of VGAT-cre mice), cell bodies within SP were outlined and then expanded inward/outward to create a 15-pixel thick ring-shaped ROI around transduced neurons. VGAT, gephyrin, and GABA A R-g 2-subunit immunoreactivities were then measured in these ROIs. Five to seven cells were measured per brain section, two sections per animal, to calculate an average value for each animal. Since only one hemisphere was injected, the other hemisphere served as an internal control to ensure there was no significant variability from mouse to mouse. Additional injected mice were excluded from study in cases of off-target injection or lack of transduction (VGAT-cre: four mice, VGLUT1-cre: three mice). For electrophysiological and behavioral study in VGLUT1cre mice, mice were injected bilaterally. For electrophysiological experiments, data are presented from transduced cells (HACB2-AAV: n = 5, EGFP-AAV: n = 6; 1 cell per mouse) from injected mice. The final 100 events before bath application of picrotoxin was used for analysis. For behavioral experiments, data are presented for injected mice (HACB2-AAV: n = 15, EGFP-AAV: n = 12), and the observer was blinded to the condition of the mouse during seizure scoring. Additional injected mice were excluded from study if they died after PTZ administration (HACB2-AAV: one mouse, EGFP-AAV: two mice) or if brain examination determined off-target injection or lack of transduction. No additional mortality was observed during or following procedures in adeno-associated virus (AAV)-injected mice. Statistical analyses were performed using Prism 8 (GraphPad). Statistical tests include Mann-Whitney U tests for two-group comparisons when sample size was below 10, and unpaired t tests when sample size was .10, Kolmogorov-Smirnov tests, and Fisher's exact test. Statistical data can be found in Table 1. Summary data are presented as mean 6 SEM from n mice or neurons, as indicated.

Mice injected with AAV-HA-CB2SH3-express both mCitrine and HA-CB
In order to validate that our cre-dependent viral construct was capable of transduction of appropriate cells in CA1, we examined the pattern of mCitrine and HA expression in injected mice.
In VGLUT1-IRES-cre mice injected with the HACB2-AAV, large swathes of the pyramidal cell layer in CA1 expressed mCitrine (Fig. 1A), interspersed with the occasional non-transduced cell, likely indicating an interneuron. As expected, none of the cells in SO were transduced, since cell bodies there belong to GABAergic cells. The transduced cells showing mCitrine fluorescence also continued mCitrine also overexpressed HA-CB2SH3-, as detected by Ms anti-HA (B, F) and Rb anti-CB (D, F). In CA1 of EGFP-AAV-injected mice, HA was not detected (C, G), and only a low endogenous level of CB (E, G) was visible. Quantification from confocal images for mean intensity of HA immunofluorescence in SP is presented in H, and integrated intensity of CB immunofluorescent clusters is presented in I. J-R, The mcitrine-HACB2-AAV (J, K, M, O) or a control EGFP-AAV (L, N, P) were injected into CA1 of VGAT-IRES-cre mice, resulting in overexpression of mCitrine or EGFP in GABAergic interneurons and images were taken with epifluorescence microscopy (J-N) or confocal microscopy (O-P). In mCitrine-HACB2-AAV-injected mice, the transduced interneurons expressed mCitrine and overexpressed HA-CB2SH3-, as detected by Ms anti-HA (K, O) and Rb anti-CB (M, O). In CA1 of EGFP-AAV-injected mice, HA was not detected (L, P), and only a low endogenous level of CB (N, P) was visible. Quantification from confocal images for mean intensity of HA immunofluorescence in interneurons is presented in Q, and integrated intensity of perisomatic CB immunofluorescent clusters is presented in R. Scale bars:25 mm (A, J) and 10 mm (B-G, K-P). Asterisks in the quantification graphs indicate significance using Mann-Whitney U test, *p,0.05, **p,0.01.   1B) and increased immunoreactivity with Rb anti-CB (Fig. 1D). In CA1 of the VGLUT1-IRES-cre mice injected with the EGFP-AAV, we did not detect any HA (Fig. 1C), nor any increased levels of CB (Fig. 1E). To quantify the extent to which CB was expressed over endogenous levels, confocal images were taken (Fig. 1F Fig. 1I).
In injected VGAT-IRES-cre mice, mCitrine expression in dorsal hippocampal CA1 matched the expected distribution of interneurons (Fig. 1J), with transduced cells scattered through pyramidal cell layer, as well as cell bodies in the SO. Since our viral construct was inducing bicistronic expression of mCitrine and HA-CB2SH3-, we confirmed that transduced cells were also expressing HA-CB2SH3-. We were able to detect both HA (Fig. 1K) and an increased level of CB (Fig. 1M) in transduced cells of CA1 in mice injected with the HACB2-AAV compared with EGFP-AAV-injected mice (Fig. 1L,N, respectively). As expected, all neurons in the SP of EGFP-AAV-injected mice showed endogenous CB expression. However, in HACB2-AAV transduced interneurons, CB expression was noticeably stronger than the endogenous CB expression of either the surrounding non-transduced pyramidal neurons (Fig. 1M) or interneurons transduced with EGFP-AAV ( Fig. 1M vs N, O vs P). Analysis of confocal images (Fig. 1O,P) revealed that HA MFI is significantly increased in HACB2-AAV transduced interneurons (CB: 133.3 6 11.7 A.U.; EGFP: 25.3 6 1.1 A.U.; Mann-Whitney U test, p = 0.029; Fig. 1Q) as is integrated fluorescence intensity of CB puncta (CB: 617.6 6 59.5; EGFP: 428.3 6 42.6 A.U.; Mann-Whitney U test, p = 0.029; Fig. 1R). While there was no change in the density of the clusters (Mann-Whitney U test, p = 0.69), there was a moderate but not significant increase in size of clusters (CB: 0.211 6 0.019 mm 2 ; EGFP: 0.161 6 0.008 mm 2 Mann-Whitney U test, p = 0.057).
Conditional overexpression of HA-CB2SH3-in CA1 pyramidal neurons results in enhanced gephyrin and perisomatic GABAAR-c2 clusters in VGLUT1-cre mice To investigate the extent to which overexpression of HA-CB2SH3-affects hippocampal GABAergic synapses, we examined presynaptic and postsynaptic protein expression in HACB2-AAV transduced glutamatergic pyramidal cells of CA1 ( Fig. 2A), as compared with those transduced with EGFP-AAV (Fig. 2B). Because of the packed nature of the transduced hippocampal pyramidal cells, regions within CA1 were examined rather than individual transduced cells: ROIs within the SP that contain the cell bodies of the pyramidal cells, or ROIs within SO or SR populated by processes of the pyramidal cells. In VGLUT1-cre mice injected with the HACB2-AAV, there were significant increases in average size of gephyrin clusters in each region of the transduced CA1 (SO: CB: 0.343 6 0.015 mm 2 , EGFP: 0.253 6 0.008 mm 2 ; SP: CB: 0.247 6 0.011 mm 2 , EGFP: 0.195 6 0.008 mm 2 ; SR: CB: 0.288 6 0.014 mm 2 , EGFP: 0.229 6 0.008 mm 2 ; Mann-Whitney U test, p = 0.002, p = 0.009, p = 0.026; respectively; Fig. 2C). The gephyrin clusters in these regions of HACB2-AAV-transduced CA1 also had increased MFI in the SO (32.4% increase), SP (61.5% increase), and SR (43.7% increase) compared with EGFP-AAV-transduced  Mann-Whitney U test, p = 0.004, p = 0.015, and p = 0.015; respectively; Fig. 2D). Size, density and MFI of gephyrin clusters were also increased compared with non-transduced contralateral hemisphere in the HACB2-AAV-injected mouse brains. There were no significant differences in any of the measured parameters between contralateral hemispheres of HACB2-AAV-injected mice and EGFP-AAV-injected mice.
In the SP of the HACB2-AAV-injected mice, there was also an increase of size of g 2 GABA A R subunit clusters compared with the EGFP-AAV-injected mice (CB: 0.197 6 0.009 mm 2 vs EGFP: 0.160 6 0.008 mm 2 ; Mann-Whitney U test, p = 0.032; Fig. 2E-H). No significant differences in GABA A R g 2 density or MFI were detected in the SP. Moreover, no difference in cluster size, density or MFI were found in the SO or SR of HACB2-AAV versus EGFP-AAV-injected mice.
Regarding presynaptic VGAT expression, there was no significant change to puncta size, density, or MFI of the presynaptic VGAT in SO, SP or SR in the HACB2-AAV versus EGFP-AAV-injected mice (Fig. 2I,J).

Conditional overexpression of HA-CB2SH3-enhances gephyrin clusters in CA1 interneurons in VGAT-cre mice
We examined the changes to presynaptic and postsynaptic proteins at the GABAergic synapse as a result of overexpression of HA-CB2SH3-in CA1 interneurons (Fig. 3A). There was a marked increase in the size of perisomatic gephyrin clusters on neurons transduced with the HACB2-AAV (0.350 6 0.017 mm 2 ) compared with those transduced with the EGFP-AAV (0.179 6 Rb anti-g 2 (blue; ii) and GP anti-VGAT (magenta; iii). G, H, The size (G) and density (H) of the GABA A R g 2 clusters on the perimeter of the transduced cells were quantified. The size (I) and density (J) of the VGAT puncta surrounding the transduced cell have been quantified. Each dot represents one animal and error bars indicate SEM. Two sections per animal were processed and at least five cells per brain section were quantified. Scale bars: 20 mm (A, Bi) and 10 mm (A, Bii, Biii, E, F). Asterisks in the quantification graphs indicate significance using Mann-Whitney U test; *p , 0.05, **p , 0.01. 0.015 mm 2 ; Mann-Whitney U test, p = 0.029; Fig. 3A-C) and a 140% increase in MFI of clusters (Mann-Whitney U test, p = 0.029). This was accompanied by an increase in the density of gephyrin clusters in neurons transduced with HACB2-AAV compared with EGFP-AAV (0.436 6 0.047 and 0.207 6 0.034 clusters/mm 2 , respectively; Mann-Whitney U test, p = 0.029; Fig. 3D). Despite gephyrin clusters in HACB2-AAV transduced interneurons being nearly twice the size, and twice as numerous as in EGFP-AAV transduced interneurons, there was only a modest, but not significant, increase in the size of GABA A R g 2 clusters (CB: 0.214 6 0.016 vs EGFP: 0.169 6 0.008 mm 2 ; Mann-Whitney U test, p = 0.057; Fig. 3E-G) and no significant change in density of clusters (CB: 0.139 6 0.005 vs EGFP: 0.138 6 0.011 puncta/mm 2 ; Mann-Whitney U test, p . 0.99; Fig. 3H).

Synaptic inhibitory neurotransmission is strengthened in HA-CB2SH3-overexpressing CA1 pyramidal neurons
To determine whether the enlarged gephyrin and GABA A R clusters in VGLUT1-cre mice transduced with HACB2-AAV corresponded to stronger synaptic inhibitory neurotransmission, we measured spontaneous mIPSCs in transduced CA1 pyramidal neurons in brain slices from mice bilaterally injected with either HACB2-AAV or control EGFP-AAV. We found that in HACB2-AAV transduced CA1 pyramidal cells, there were more inhibitory events, many with higher amplitudes than in the control EGFP-AAV-transduced cells (Fig. 4A). These mIPSCs were eliminated in the presence of picrotoxin, ascertaining that they correspond to GABA A R currents.
A cumulative relative frequency plot indicates a right shift in the amplitudes of mIPSCs in HACB2-AAV-transduced neurons (Kolmogorov-Smirnov p , 0.001; Fig. 4B) and greater amplitudes (CB: À40.91 6 4.28 pA; EGFP: À27.88 6 2.71 pA; Mann-Whitney U test, p = 0.046; Fig. 4C), suggesting an increase in the number of postsynaptic GABA A Rs. The mIPSCs in HACB2-AAV-transduced cells shifted to lower interevent intervals (Kolmogorov-Smirnov, p , 0.001; Fig. 4D) and these transduced neurons demonstrated higher frequencies (CB: 0.537 6 0.095 Hz; EGFP: 0.181 6 0.036 Hz; Mann-Whitney U test, p = 0.004; Fig. 4E) than control EGFP-AAV cells. This finding when taken together with our results demonstrating no change in size or density of presynaptic VGAT, likely indicates a higher probability of release from presynaptic GABAergic terminals. Finally, we analyzed the rise time and decay tau of mIPSCs, to determine whether there might be any changes in receptor kinetics which might speak to alterations in distribution of receptors or receptor subunit composition. We found that while there was a significant increase in 10-90% rise times of mIPSCs in HACB2-AAV neurons (CB: 1.132 6 0.040 ms; EGFP: 0.941 6 0.013 ms; Mann-Whitney U test, p = 0.004; Fig. 4F), the increase in decay tau was not significant (CB: 20.05 6 2.84 ms; EGFP: 14.55 6 0.95 ms; Mann-Whitney U test, p = 0.126; Fig. 4G).

VGLUT1-cre mice injected with HACB2-AAV are less susceptible to acute seizures
To determine whether increased GABAergic inhibitory synaptic neurotransmission in CA1 pyramidal cells, seen in VGLUT1-cre mice injected with the HACB2-AAV, imparted neuroprotective effects against acute seizures, we subjected AAV-injected mice to intraperitoneal injections of PTZ. Based on initial studies we settled on a dose of PTZ (40 mg/kg), which induced seizures reliably in naive animals with minimal mortality. Three weeks after bilateral CA1 injection with HACB2-AAV (n = 15 mice) or a control EGFP-AAV (n = 12 mice), we assessed seizures along a modified Racine scale after a single dose of PTZ. The mice injected with the HACB2-AAV displayed lower cumulative seizure scores over the 30 min observed, compared with the EGFP-AAV control mice (Kolmogorov-Smirnov test, p , 0.001; Fig. 5A). HACB2-AAV-injected mice were less likely to have higher intensity seizures after PTZ compared with the control mice (Fisher's exact, p = 0.020, p = 0.019, and p = 0.002; seizure scores 4, 5, and 6, respectively; Fig. 5B). When comparing the onset latency to the first seizure of each intensity score (Fig.  5C), we found an increased latency period for the HACB2-AAV mice for seizure scores 1 and 2 (CB: 216.1 6 16.6s, EGFP: 76.4 6 2.6s; and CB: 260.0 6 22.4s, EGFP: 83.3 6 2.8s; unpaired t test, p = 0.031 and p = 0.042, respectively). However, the increased latency times were not statistically significant at the higher seizure scores because of the low numbers of HACB2-AAV mice having high intensity seizures (note the diminishing number of data points in Fig. 5C).

Discussion
We evaluated the effects of overexpressing CB in specific neuron types in the adult dorsal hippocampus. While CBSH31 splice variants are the far more abundantly expressed isoforms of CB (Harvey et al., 2004)  The percentage of animals that had a seizure at each intensity score. Asterisks indicate significance using Fisher's exact test; *p , 0.05, **p , 0.01. C, The average latency to the first seizure of each intensity was recorded. Error bars indicate SEM, and each dot represents one animal. Note that at higher intensity seizure scores, there are fewer mice that seize, particularly in the CB category. Most mice seize with lower intensities. Asterisks indicate significance using unpaired t test; *p , 0.05. al., 2014). The constitutively active CBSH3-isoforms, however, are capable of inducing the accumulation of gephyrin and GABA A Rs at postsynaptic sites (Chiou et al., 2011;Tyagarajan et al., 2011;Fekete et al., 2017). We found that overexpression of CB2SH3-in hippocampal CA1 pyramidal neurons resulted in significantly increased CB, gephyrin and GABA A R clustering in perisomatic regions (within SP). Overexpression of CB2SH3-in CA1 interneurons also resulted in increased CB expression, and larger perisomatic gephyrin and GABA A R clusters. The increase in postsynaptic GABAergic receptors was accompanied by a significantly increased mIPSC amplitude and frequency. Overexpression of CB2SH3-in excitatory neurons in CA1 of the dorsal hippocampus also provided resilience against PTZ-induced seizures.
In CA1 pyramidal cells, overexpression of CB2SH3-resulted in an increase in gephyrin and GABA A R cluster size. The majority of these large gephyrin clusters were apposed to VGAT terminals, indicating localization to GABAergic synapses. This is consistent with what has been seen in cultured hippocampal neurons transfected to overexpress CB2SH3- (Chiou et al., 2011;Tyagarajan et al., 2011;Körber et al., 2012) and in CB2SH3-transfected pyramidal neurons of the cerebral cortex of in utero electroporated (IUE) rats (Fekete et al., 2017). In HACB2-AAV transduced hippocampal pyramidal cells, the increase in amplitude of spontaneous mIPSCs further indicates an accumulation of GABA A Rs at the postsynapse.
Immunofluorescence revealed an increase in synaptic g 2-containing GABA A Rs in SP, but not in SO or SR, despite CA1 pyramidal cells receiving 92% of their GABAergic synaptic input at their dendrites (Megías et al., 2001). Each cell compartment is typically innervated by different interneuron subtypes, with parvalbumin (PV)1 and cholecystokinin (CCK)1 basket cells providing the majority of GABAergic innervation to perisomatic regions (Miles et al., 1996;Salesse et al., 2011). In addition, PV1 chandelier cells almost exclusively innervate the axon initial segments of pyramidal cells, which lie at the border of SP and SO (Buhl et al., 1994). The increased GABA A R g 2 clustering in SP suggests that the perisomatic synapses, in particular, are strengthened by HACB2-AAV transduction, although the longer rise times of spontaneous mIPSCs raise the possibility of increased GABA A R clustering at more distal synapses, or the addition of GABA A Rs of alternative compositions to the perisomatic synapses. At these synapses, GABA A Rs predominantly have a1b 3g 2 and a2b 3g 2 subunit composition (Nusser et al., 1996;Nyíri et al., 2001;Kasugai et al., 2010). The increase in postsynaptic gephyrin cluster size might attract extrasynaptic a5b 3g 2 GABA A Rs into the postsynapse. The a5b 3g 2 GABA A Rs are mostly extrasynaptic but can also be synaptically localized (Brünig et al., 2002;Christie et al., 2002;Schweizer et al., 2003;Alldred et al., 2005;Serwanski et al., 2006). The incorporation of a5b 3g 2 GABA A Rs into the synapse is consistent with the longer rise times and very slow deactivation of a5b 3g 2 GABA A Rs compared with a1b 3g 2 and a2b 3g 2 (Mortensen et al., 2012). In hippocampal pyramidal neurons, the a5, a2, and a1 are the three most abundant a GABA A R subunits, in that order (Wisden et al., 1992;Fritschy and Mohler, 1995;Nusser et al., 1996;Brünig et al., 2002;Serwanski et al., 2006;Kasugai et al., 2010). Immunofluorescence revealed that the effects of HACB2-AAV on enhancing postsynaptic GABA A R clustering are more subdued than the effects on gephyrin clustering. We attribute it to the fact that gephyrin clustering in these cells is largely concentrated on GABAergic synapses, while g 2containing GABA A Rs are both synaptic and extrasynaptic.
We did not observe any changes in density, MFI, or size of VGAT puncta, which is consistent with previous studies indicating that CB is not synaptogenic: knocking out CB in mice, while disruptive of gephyrin clusters in certain brain regions, does not affect presynaptic GABAergic innervation (Papadopoulos et al., 2007), nor does knockdown of CBSH3-or CBSH31 isoforms (George et al., 2021) or overexpression of CB (Chiou et al., 2011;Fekete et al., 2017). Nevertheless, we observed a left-shift in interevent interval and increase in frequency of mIPSCs in CB2SH3-overexpressing cells. Since this is not accompanied by increased presynaptic GABAergic innervation, as shown by VGAT immunofluorescence, the result is consistent with an increased probability of spontaneous vesicular GABA release from presynaptic terminals onto HACB2-AAV-transduced cells. Therefore, increased postsynaptic clustering of gephyrin/GABA A Rs seems to have a transsynaptic effect on spontaneous GABA release. PV1 and CCK1 basket cells in the hippocampus have distinct mechanisms for modulating presynaptic release. While CCK1 basket cell terminals are covered in cannabinoid and GABA B Rs (Marsicano and Lutz, 1999;Sloviter et al., 1999;Liu et al., 2019), M 2 muscarinic and m-opioid receptors modulate activity at PV1 basket cell terminals (Hájos et al., 1998;Drake and Milner, 2006;Freund and Katona, 2007). The increase in spontaneous mIPSC frequency is contrary to the decreased frequency observed in pyramidal cells of rat cerebral cortex transfected by IUE to overexpress CB2SH3- (Fekete et al., 2017). This difference could be attributed to: (1) being neurons in different brain areas innervated by different interneurons; or (2) the IUE study involved CB2SH3-overexpression in a sparse population of pyramidal neurons in cerebral cortex since embryonic age, which has the potential for altering synaptic function in the adult.
The possibility of enhancing synaptic GABAergic neurotransmission in hippocampal circuits is of particular interest in regards to circuits in which there is an excitation/ inhibition imbalance. In the current study, we used the chemoconvulsant PTZ to induce acute seizures. The mechanism of action for PTZ is not well understood, but it has been demonstrated to cause calcium and sodium influx in neurons, thereby resulting in depolarization (Papp et al., 1987). PTZ been used to model acute seizures with a single dose, and to study epileptogenesis in chemical kindling models with multiple subthreshold doses. In VGLUT1-cre mice injected with HACB2-AAV, latency to PTZ-induced seizure events was increased and mice were less likely to develop severe seizures, thereby demonstrating that enhancing the strength of GABAergic synaptic inhibition through HACB2-AAV transduction provided protection against convulsions. Our results bring attention to CB as a possible target for therapeutic intervention. They suggest that enhancing CB expression or enhancing the activity of existing endogenous CB in a selected population of hyperactive hippocampal neurons may decrease their activity and prove useful in ameliorating seizures originating in hippocampal structures such as temporal lobe epilepsy.
In VGAT-cre mice, we quantified interneurons with soma residing within the SP. While the majority of these cells are PV1 basket cells, there are many additional interneuron types whose soma can be found within or immediately adjacent to the SP, including CCK1 basket cells, PV1 bistratified or chandelier cells, ivy cells, and vasointestinal active peptide (VIP)1 type 1 or three interneuron selective interneurons (Klausberger, 2009;Pelkey et al., 2017). Many additional interneuron types were transduced in the SO and SR that may be bistratified cells, oriens-lacunosum moleculare cells, or CCK1/ VIP1 basket cells (Sik et al., 1995;Baude et al., 2007). The vast majority of boutons from pyramidal layer interneurons terminate on pyramidal cells within the SP or onto the soma of other basket cells or bistratified cells within the SP (Sik et al., 1995;Földy et al., 2010). CB may be differentially expressed between various interneuron types. It has been shown that PV1 interneurons in the hippocampus and cerebral cortex contain higher levels of endogenous CB (Arhgef9) transcripts and protein than other interneurons (Paul et al., 2017;George et al., 2021). We found that transduction with the HACB2-AAV resulted in increased expression of CB in transduced interneurons compared with control EGFP-AAV transduced interneurons.
Transduction of interneurons with HACB2-AAV resulted in roughly two-fold increases in both size and density of perisomatic gephyrin and modest increases in perisomatic GABA A R clusters. This effect was not accompanied by any changes to size, density, or MFI of presynaptic VGAT input onto the soma, indicating that CBSH3-overexpression does not recruit additional GABAergic innervation to these cells. To maintain consistency in our analysis, we restricted ourselves to quantification in cells within the center of SP. However, nearly every HACB2-AAV-transduced CA1 interneuron, including those in SO and on the perimeter of SR, displayed a noticeably strong perisomatic ring of gephyrin clusters. This was rarely seen in non-transduced interneurons of the contralateral CA1 or in EGFP-AAV-transduced CA1 interneurons. However, strong diffuse or punctate extrasynaptic immunoreactivity with the GABA A R g 2 antibody was frequently seen in CA1 interneurons, in VGAT-cre as well as VGLUT1-cre mice, and in both HACB2-AAV and EGFP-AAV injected and non-injected hemispheres, indicating that the g 2-containing GABA A Rs in interneurons are not only synaptic but also extrasynaptic.
The results of HACB2-AAV injection in VGAT-cre mice indicates that our approach can also be used for targeting GABAergic interneurons for the conditional overexpression of CB2SH3-. However, because of the heterogeneity of the transduced interneurons and the diversity of synaptic inhibitory GABAergic inputs from other interneurons, the VGAT-cre mice are not adequate to properly interrogate the functional outcomes, which would be hard if not impossible to interpret. Instead, CB overexpression can be tailored to target-specific classes of interneurons in the hippocampus or any region of the brain by using selected interneuron-type-specific cre-driver mouse lines (i. e., PV-cre, CCK-cre, Calbindin2-cre, SOM-Cre, and VIPcre among others). These future studies are beyond the scope of this article.
Enhancement of inhibitory synapses have also been accomplished by overexpression of NL2, a cell-adhesion molecule found selectively at inhibitory postsynapses which interacts with and activates CBSH31 (Poulopoulos et al., 2009;Soykan et al., 2014). It has also been demonstrated to interact with gephyrin and CB forming a tri-partite complex at GABAergic synapses with GABA A R-g 2 and the transmembrane GABA A R accessory protein LHFPL4 or GARLH (Varoqueaux et al., 2004;Davenport et al., 2017;Yamasaki et al., 2017). Two studies have investigated AAV-mediated overexpression of NL2 in the hippocampus. In one study, overexpression of NL2 results in increased gephyrin clusters and VGAT1 puncta by confocal microscopy immunofluorescence and an increase in the gephyrin, VGAT and GABA A R-g 2 subunit protein expression by western blot, suggesting a strengthening inhibitory synapses in the transduced neurons (Van Zandt et al., 2019). In the other study, increased NL2 expression leads to increased mRNA levels for GAD65, but no increase in neurexin-1, gephyrin, or GAD67 mRNAs (Kohl et al., 2013). Interestingly, the two AAV-mediated NL2-overexpression studies also revealed alterations to behavior, notably, reductions in social aggression and dominance behaviors but also diminished preference for social novelty. In another study, overexpression of NL2 in cortical pyramidal neurons after in utero electroporation shows an increase in VGAT which were associated to gephyrin clusters (Fekete et al., 2015). An increase in VGAT, but no change in gephyrin was reported in transgenic animals overexpressing NL2 (Hines et al., 2008). In the same study, it was determined that pyramidal neurons in the prefrontal cortex of these transgenic mice had increased mIPSC frequency but showed no effect on the amplitude of mIPSCs (Hines et al., 2008;Fekete et al., 2015). It is worth noting that in the two AAV-NL2 studies, NL2 was simultaneously overexpressed in both glutamatergic and GABAergic neurons. The NL2 overexpression studies strongly suggest that NL2 overexpression in neurons enhances GABAergic synaptic transmission and that if targeted to specific neurons might also be effective for seizure suppression.
The HACB2-AAV and described methodology can be generally applied for selective in vivo enhancement of GABAergic synaptic transmission in any neuronal type and brain region, by using the expanding collection of available Cre-driver mouse lines. Moreover, our DIO-AAV could also be used for the overexpression of CBSH3-in selected types of neurons in wild-type non-transgenic animal species besides mouse, by co-injecting it with another AAV carrying Cre-recombinase under the control of a neuron-type-specific modified promoter. This has been accomplished in rat and monkey (Mikhailova et al., 2016;Stauffer et al., 2016;Zalocusky et al., 2016).