Syngap1 Disruption Induced by Recombination between Inverted loxP Sites Is Associated with Hippocampal Interneuron Dysfunction

Abstract SYNGAP1 haploinsufficiency in humans causes intellectual disability (ID). SYNGAP1 is highly expressed in cortical excitatory neurons and, reducing its expression in mice accelerates the maturation of excitatory synapses during sensitive developmental periods, restricts the critical period window for plasticity, and impairs cognition. However, its specific role in interneurons remains largely undetermined. In this study, we investigated the effects of conditional Syngap1 disruption in medial ganglionic eminence (MGE)-derived interneurons on hippocampal interneuron firing properties and excitatory synaptic inputs, as well as on pyramidal cell synaptic inhibition and synaptic integration. We show that conditional Syngap1 disruption in MGE-derived interneurons results in cell-specific impairment of firing properties of hippocampal Nkx2.1 fast-spiking interneurons, with enhancement of their AMPA receptor (AMPAR)-mediated excitatory synaptic inputs but compromised short-term plasticity. In contrast, regular-spiking Nkx2.1 interneurons are largely unaffected. These changes are associated with impaired pyramidal cell synaptic inhibition and enhanced summation of excitatory responses. Unexpectedly, we found that the Syngap1flox allele used in this study contains inverted loxP sites and that its targeted recombination in MGE-derived interneurons induces some cell loss during embryonic development and the reversible inversion of the sequence flanked by the loxP sites in postmitotic cells. Together, these results suggest that Syngap1 plays a role in cell-specific regulation of hippocampal interneuron function and inhibition of pyramidal cells in mice. However, because of our finding that the Syngap1flox allele used in this study contains inverted loxP sites, it will be important to further investigate interneuron function using a different Syngap1 conditional allele.


Introduction
Neuronal computation relies on tight interactions between excitatory and inhibitory neurons. Cortical GABAergic cells constitute a heterogeneous population of inhibitory interneurons and represent 10-20% of the total neuronal population. Interneurons ensure the proper functioning of cortical network circuits, by gating information flow, sculpting neuronal patterns and preventing their hyperexcitability (Tremblay et al., 2016;Pelkey et al., 2017). Inhibitory cell dysfunctions are linked to several brain pathologies, such as epilepsy, autism, intellectual disability (ID), Rett syndrome, anxiety disorders and schizophrenia (Rossignol, 2011;Chattopadhyaya and Cristo, 2012;Marín, 2012).
Numerous studies with mouse models suggest that mutations in genes encoding proteins involved in synaptic structure and function are implicated in neurodevelopmental disorders like intellectual disability (ID) and autism spectrum disorders (ASDs; Südhof, 2008;Zoghbi and Bear, 2012). Dysfunction of synaptic proteins affect the organization and function of neuronal circuits in the developing and mature brain. For example, haploinsufficiency of the SYNGAP1 gene, encoding for such a protein, represents a common cause of ID with co-morbid ASD or/and epilepsy (Hamdan et al., 2009;Berryer et al., 2013Berryer et al., , 2016. SYNGAP1 codes for the synaptic GTPase-activating protein 1 (SYNGAP1) that negatively regulates Ras. In cortical excitatory neurons, SYNGAP1 is physically associated with the postsynaptic density 95 protein (PSD-95) and acts downstream of NMDA receptors (NMDARs; Kim et al., 1998;Walkup et al., 2016). SYNGAP1 limits the strengthening of excitatory synapses by restricting AMPA receptors (AMPARs) synthesis through inhibition of the Ras-ERK and mTOR signaling pathways (Wang et al., 2013;Araki et al., 2015) and by limiting the binding of auxiliary proteins (TARPs,LRRTM2, to the PDZ domains of PSD-95 (Walkup et al., 2016). These actions ensure proper refinement and maturation of synaptic connections. Indeed, in Syngap1 1/À mice, excitatory neurons show increased AMPAR density, larger spine heads and premature maturation of excitatory synapses (Clement et al., 2012(Clement et al., , 2013Aceti et al., 2015;Jeyabalan and Clement, 2016). These changes lead to altered excitatory/inhibitory (E/I) balance and abnormal critical periods of plasticity in somatosensory cortex (Clement et al., 2012(Clement et al., , 2013. Thus, Syngap1 haploinsufficiency in excitatory neurons is believed to underlie the synaptic and cognitive abnormalities observed in patients and mouse models of neurodevelopmental disorders. However, recent studies found that SYNGAP1 is also expressed in GABAergic interneurons, particularly in fastspiking parvalbumin-expressing (PV) interneurons (Berryer et al., 2016;Su et al., 2019), which are important for brain rhythms generation, working memory and attention control (Hu et al., 2014). Cortical interneurons can be classified into two groups based on their developmental origin from the medial ganglionic eminence (MGE) or caudal ganglionic eminence (CGE; Kepecs and Fishell, 2014). Syngap1 haploinsufficiency in MGE-derived interneurons has been reported to impair inhibitory synaptic transmission and cognition (Berryer et al., 2016), suggesting that SYNGAP1 in inhibitory cells ensures the proper formation of PV cell synaptic fields in a cell autonomous fashion (Berryer et al., 2016). In addition, the expression of the GluA2 subunit of AMPARs is upregulated in PV interneurons (PV-INs) in some brain regions like CA3 hippocampus and somatosensory barrel cortex of Syngap1 1/À mice (Sullivan et al., 2020). Although, these lines of evidence point to an important role of SYNGAP1 in interneuron development and function, its specific role in distinct types of inhibitory cells remains largely unknown.
Here, we used whole-cell patch-clamp recordings in slices to examine the role of SYNGAP1 in hippocampal CA1 MGE-derived Nkx2.1-expressing cells (Nkx2.1 1 ), more specifically fast spiking [FS; likely corresponding to parvalbumin (PV) expressing interneurons] and regular spiking (RS) interneurons [likely corresponding to somatostatin (SOM) expressing interneurons; Tricoire et al., 2011;Pelkey et al., 2017]. We report that Syngap1 disruption in these cells selectively impaired firing properties of Nkx2.1 1 FS cells, enhanced their AMPAR-mediated excitatory synaptic inputs, and compromised the short-term dynamics of their excitatory synapses. Interestingly, Nkx2.1 1 RS interneurons were largely unaffected. These changes in inhibitory interneurons were associated with a reduction of synaptic inhibition, as well as facilitation of excitatory postsynaptic response summation, in pyramidal cells. However, we found unexpectedly that the Syngap1 conditional allele used in this study contains inverted loxP sites and that its recombination in MGE-derived interneurons induces some cell loss during embryonic development and the reversible inversion of the sequence flanked by the loxP sites in postmitotic cells.
Electrophysiological data were recorded using Multiclamp 700A/B amplifiers (Molecular Devices) and digitized at 20 kHz using Digidata 1440A and pClamp 10 (Molecular Devices). Recordings were low pass filtered at 2 kHz. Series resistance (15-25 mX) was monitored regularly during experiments and data were included if the holding current and series resistance were stable (changes ,20% of initial value). Spontaneous EPSCs and sIPSCs were analyzed using Mini-Analysis 6.0.3 (Synaptosoft Inc.). For sEPSCs and sIPSCs, cells were recorded for a period of 5-10 min and 300 consecutive events were considered for analysis. For all experiments, data were acquired and analyzed blind to genotype.

Electrophysiological data analysis
Electrophysiological data were analyzed with Clampfit 10.5 (Molecular Devices), Mini-Analysis 6.0.3 and GraphPad Prism 8 (GraphPad Software Inc.). Membrane properties of EGFPpositive interneurons were measured in current-clamp (Tricoire et al., 2011;Artinian et al., 2019). Resting membrane potential (RMP) was measured with the holding current I = 0 pA immediately after cell membrane rupture. Input resistance (R in ) was measured using a linear regression of voltage deflections (610-15 mV) in response to 800-mslong current steps of 20 pA for fast-spiking EGFP-positive interneurons or 500-ms-long current steps of 10 pA for regular-spiking EGFP-positive interneurons from a holding potential of -60 mV. Rheobase was measured as the minimal current necessary to evoke an action potential. For FS-INs, the frequency-current (F-I) relationship for evoked firing was determined by injecting 800-ms somatic current steps of increasing amplitude (20-pA increments) to a maximum of 400 pA. For RS-INs, the F-I relationship was determined by injecting 500-ms somatic current steps of increasing amplitude (10-pA increments) to a maximum of 170 pA. The slope of the F-I relationship for each cell was calculated from a linear fit of the data for which the F-I relationship was approximatively linear.
AMPAR-mediated and NMDAR-mediated EPSCs were recorded in EGFP-positive interneurons using constant current pulse (50-ms duration) stimulation via an ACSFfilled theta-glass electrode, positioned ;100 mm from the recorded cell in CA1 stratum oriens for FS-INs, or at the border between CA1 stratum oriens and the alveus for RS-INs. Minimally evoked EPSCs were obtained using minimal electrical stimulation at 0.5 Hz (with a success rate of ;50%). Minimal EPSC amplitude (including successes and failures) and EPSC potency (including successes only) were calculated from 100-150 events. Paired pulse ratio of minimal EPSCs was calculated as a ratio of EPSC 2 to EPSC 1 (50-ms interstimulus interval). Synaptic currents input-output relationships were obtained by delivering stimulation current pulses of incremental intensity (50-600 mA) at 0.05 Hz, with five trials for each stimulation intensity. The slope (synaptic gain) and x-intercept (minimal stimulation intensity) of the linear regression of the input-output relationship were measured on averaged responses of individual cells. For short-term plasticity of EPSCs, five trains of 10 electrical stimuli at different frequencies (10, 20, and 50 Hz) were delivered at 20-s intervals. EPSC amplitude was normalized to the first EPSC amplitude. For EPSP-IPSP integration experiments, CA1 pyramidal cells were held at À60 mV and synaptic responses were obtained in current-clamp mode with electrical stimulation delivered by a concentric bipolar Pt/Ir electrode (FHC) placed in stratum oriens.
Immunostained slices were imaged using the Leica SP8 confocal microscope at 20Â (NA 0.75) at 1024 Â 1024, zoom = 1, z-step = 1.5 mm. Confocal stacks of ;20-mm depth were acquired from the CA1 region of the hippocampus from at least three coronal sections for each animal. At least 5 mice per genotype were used for the cell count analysis. All acquired images were analyzed in blind. Identical areas of 250 Â 250 mm 2 were analyzed semi automatically by either Neurolucida (MBF software) and or LAS X (Leica Application Suite X) software with two to three regions of interest (ROIs) per section (total = six to nine ROIs per animal). Total numbers of GFP 1 cells, PV 1 cells and both PV 1 GFP 1 as well as PV -GFP 1 cells were counted in the hippocampal CA1 area. Data are expressed as mean 6 SEM and analyzed using Prism 8 (GraphPad Software Inc). Normality and equal variance were tested and differences between two groups were assessed with the Mann-Whitney rank sum test for notnormally distributed data.

Whole genome sequencing
For whole genome sequencing (WGS), genomic DNA (gDNA) was extracted from the tail of a Syngap1 flox/flox mouse, using the PureLink Genomic DNA Mini kit (ThermoFisher). Paired-end (2 Â 150 bp) WGS was performed on a NovaSeq 6000 System (Illumina). Reads were first aligned onto the mouse reference genome GRCm38/mm10 using DRAGEN v.3.4.5 (Illumina). Variant calling was also done using DRAGEN v.3.4.5. Nonmatching sequences of over thirty nucleotides were identified from the list of called variants to locate the 34-bp loxP sequence insertions. The identity and position of these loxP sequences were confirmed with the Basic Local Assignment Search Tool (https://blast. ncbi.nlm.nih.gov/Blast.cgi) and by visual inspection using the Integrative Genomics Viewer tool version IGV_2.8.0 (http://www.broadinstitute.org/igv/).

Characterization of the Syngap1 flox allele by PCR
Various brain areas were isolated from Nkx2.1-Syngap1 f/1 and Syngap1 f/1 mice at 14 d of age [postnatal day (P)14], as well as from Emx1-Syngap1 f/1 and Emx1-Syngap1 f/f mice at 24 d of age (P24). Genomic DNA was purified from these samples using either the PureLink Genomic DNA Mini kit (ThermoFisher Scientific) or the DNeasy Blood and Tissue kit (QIAGEN) according to the manufacturers' instructions.
Cre-mediated inversion of Syngap1 floxed exons 4-8 was revealed by PCR using two primers (pair B), which were designed to anneal with the same orientation to the same (-) gDNA strand of the gene so that amplification only becomes possible if the primers face each other on opposite strands as a result of the inversion. The forward primer (59-GACTTGGATAGGCAGTCAGTCAG-39), named B-i3Rv1, anneals 59 to the loxP sequence in intron three on the (-) strand (directed toward exon 3) while the reverse primer (59-ACTCTTCTATGGCTTTAGTGGCG-39), named B-e9Rv2, anneals to the same strand 39 to the loxP site in exon 9.

Statistical analysis
Data are expressed as mean 6 SEM and analyzed using GraphPad Prism 8 (GraphPad Software Inc).
Normality and equal variance were tested using the Shapiro-Wilk test and Levene median test, respectively. Differences between two groups were assessed with the Student's t test for normally distributed data or with the Mann-Whitney rank sum test for not-normally distributed data. For multiple comparisons, mixed repeated-measures ANOVA were used for normally distributed data, and Friedman ANOVA was used for notnormally distributed data. Asterisks denote statistical significance as calculated by the specified statistical tests (*p , 0.05, **p , 0.01, ***p , 0.001; ns indicates not significant). Sample size required to reach significance were determined with a power analysis with power = 0.8 and a = 0.05 (G*Power software). Details of statistical analyses are given in Extended Data Figure 1-1.
Next, we assessed the effect of Syngap1 decrease in MGE-derived inhibitory cells on NMDAR-mediated EPSCs in Nkx2.1 1 FS-INs. To do this, electrically-evoked NMDAR-EPSCs were isolated pharmacologically by blocking GABA A R-mediated and AMPAR-mediated synaptic responses (100 mM PTX and 20 mM CNQX, respectively) and reducing extracellular Mg 21 . We found that NMDAR-mediated EPSCs in Nkx2.1 1 FS-INs were similar in Nkx2.1-Syngap1 f/1 and Nkx2.1-Syngap1 1/1 mice. The NMDAR-mediated EPSC amplitude I-O relationship and I-O slope were unchanged (Fig. 2l-n). In line with its role in excitatory pyramidal neurons, these results suggest that Syngap1 decrease in Nkx2.1 1 FS-INs specifically facilitates AMPAR-mediated but not NMDAR-mediated synaptic transmission.
Impairment of short-term dynamics of AMPARmediated synaptic excitation of Nkx2.1 fast-spiking interneurons Since Syngap1 decrease in MGE-derived interneurons affects synaptic excitation of Nkx2.1 1 FS-INs, we examined whether short-term dynamics of these synaptic responses were altered. We analyzed short-term plasticity of EPSCs in Nkx2.1 1 FS-INs during a train of 10 stimuli given at frequencies of 10, 20, and 50 Hz (Fig. 3a-c) by an extracellular stimulation pipette in stratum oriens. We found that in control Nkx2.1-Syngap1 1/1 mice, EPSCs showed short-term facilitation during the first few stimuli of the trains, which recovered to baseline level at the end of the train, for all stimulation frequencies. In contrast, in Nkx2.1-Syngap1 f/1 mice, EPSCs displayed short-term depression that reached a plateau during the stimulation trains, indicating that short-term dynamics of AMPARmediated synaptic excitation of Nkx2.1 1 FS-INs is deficient after conditional Syngap1 decrease in these cells.
Given the role of postsynaptic Ca 21 -dependent signaling in the regulation of AMPAR-mediated synaptic excitation in principal cells (Borgdorff and Choquet, 2002;Heine et al., 2008;Opazo et al., 2010;Constals et al., 2015) and FS-INs (Camiré et al., 2012), we tested whether preventing postsynaptic Ca 21 rises during high frequency stimulation could rescue the deficit in EPSC short-term dynamics in Nkx2.1-Syngap1 f/1 mice. We included the fast Ca 21 chelator BAPTA (10 mM) in the intracellular solution during whole cell recordings from Nkx2.1 1 FS-INs and measured EPSC short-term plasticity during a 50-Hz stimulation train (Fig. 3d). We found that in the presence of intracellular BAPTA, EPSCs from Nkx2.1-Syngap1 f/1 mice did not show depression, but instead showed facilitation during the first few stimuli and recovered to baseline level at the end of the train, similarly to EPSCs from Nkx2.1-Syngap1 1/1 mice (Fig. 3c,d). In the presence of intracellular BAPTA, EPSCs from Nkx2.1-Syngap1 1/1 mice displayed stronger facilitation. Thus, reducing postsynaptic Ca 21 rise rescued the deficit in short-term facilitation of AMPARmediated excitatory synapses in Nkx2.1 1 FS-INs from Nkx2.1-Syngap1 f/1 mice, suggesting that the deficit in short-term dynamics caused by conditional Syngap1 decrease in these cells involve postsynaptic Ca 21 -dependent signaling.
Next, we examined whether Syngap1 decrease in MGEderived interneurons produced cell type-specific effects on synaptic excitation. First, we measured spontaneous EPSCs in Nkx2.1 1 RS-INs. We found that both sEPSC frequency and amplitude were similar (Fig. 5a-c), indicating unaffected spontaneous excitatory synaptic activity in Nkx2.1-Syngap1 f/1 relative to Nkx2.1-Syngap1 1/1 mice. Second, we stimulated excitatory inputs onto Nkx2.1 1 RS-INs through an extracellular stimulation electrode positioned at the oriens/alveus border and recorded evoked EPSCs. We found that the input-output relationship of evoked EPSCs (Fig. 5d)  Next, we examined the effects of Syngap1 decrease in MGE-derived interneurons on the short-term dynamics of synaptic excitation of Nkx2.1 1 RS-INs. We compared shortterm plasticity of evoked EPSCs during brief trains at different frequencies (10 stimuli at 10, 20, and 50 Hz, Fig. 6a-c). In Nkx2.1 1 RS-INs of control Nkx2.1-Syngap1 1/1 mice, EPSCs amplitude increased during the first few stimulations of the train and then maintained a plateau of facilitation for all stimulation frequencies (Fig. 6a-c). In Nkx2.1-Syngap1 f/1 mice, EPSCs in Nkx2.1 1 RS-INs showed similar facilitation, except for a small decrease in facilitation with 50-Hz stimulation (Fig. 6c), compared with Nkx2.1-Syngap1 1/1 mice, suggesting mostly unchanged short-term plasticity in Nkx2.1 1 RS-INs, and a further indication of cell type-specific effects after Syngap1 decrease in MGE-derived interneurons.
Research Article: New Research and postsynaptic integration. First, we recorded spontaneous IPSCs (sIPSCs) in CA1 pyramidal neurons and found that sIPSC frequency and amplitude were reduced in Nkx2.1-Syngap1 f/1 mice relative to Nkx2.1-Syngap1 1/1 mice (Fig. 7a-c). These results are consistent with the previous report of decreased miniature IPSCs (Berryer et al., 2016), indicating a deficit in synaptic inhibition of pyramidal cells. Next, we assessed what is the net effect of the global changes in interneuron functions on postsynaptic integration in pyramidal cells and examined the summation of EPSPs in pyramidal neurons. To determine this, we first recorded postsynaptic potentials (PSPs) in pyramidal cells in current-clamp mode at a holding potential of À60 mV in response to electric stimulation delivered through a stimulation electrode in the stratum oriens (Fig.  7d, top). After adjusting the stimulation intensity to get PSPs of %2 mV amplitude (compound EPSP-IPSP; Fig.  7d, bottom), we delivered 10 stimuli at 50 Hz and measured EPSPs summation in pyramidal neurons (Fig. 7e). We found that the amplitude of EPSPs in pyramidal cells from both genotypes increase during the first few stimuli of the train, reaching a plateau level from the third stimulation of the train. Interestingly, in pyramidal cells from Nkx2.1-Syngap1 f/1 mice, the summation of EPSPs was facilitated relative to Nkx2.1-Syngap1 1/1 mice (Fig. 7e). Thus, Syngap1 decrease in MGE-derived interneurons results in facilitation of EPSP summation in CA1 pyramidal cells, consistent with a decrease in synaptic inhibition.
Decreased number of MGE-derived cells in the hippocampus of Nkx2.1-Syngap1 f/1 mice In the course of the electrophysiological experiments, we made the tentative observation that the number of EGFP 1 cells appeared decreased in the hippocampus of Nkx2.1-Syngap1 f/1 when compared with Nkx2.1-Syngap1 1/1 adult mice. In order to determine whether it was indeed the case, we used immunohistochemistry to quantify the number of EGFP 1 PV 1 and EGFP 1 PV À cells in the hippocampal CA1 and somatosensory cortex from both Nkx2.1-Syngap1 f/1 and Nkx2.1-Syngap1 1/1 mice. We found that the numbers of EGFP 1 PV 1 and EGFP 1 PV À cells were significantly decreased in Nkx2.1-Syngap1 f/1 mice when compared with their littermate controls (Fig. 8a,b; Extended Data Fig. 8-1a,b1,b2). Immunohistochemistry performed at P7 also showed a decrease of the number of these cells in the somatosensory cortex of Nkx2.1-Syngap1 f/1 mice when compared with Nkx2.1-Syngap1 1/1 mice, indicating that their loss occurred early during development (data not shown).
previously reported (Vazquez et al., 2004). However, the analysis revealed that these two loxP sites were in opposite orientations (Fig. 9). Previous studies have established that targeted recombination between inverted loxP sites in cis can result in the loss of the loxP-carrying chromosome in proliferating but not in postmitotic cells because of unequal crossover between sister chromatids after DNA replication and before entry into anaphase (Lewandoski and Martin, 1997;Grégoire and Kmita, 2008). Moreover, these studies have shown that the elimination of the loxP-carrying chromosome by targeted recombination may result in cell loss.
As the Nkx2.1-cre transgene is expressed in the ventricular layer of the MGE, one possibility is that recombination between these inverted loxP sites leads to the death of proliferating progenitors of MGE-derived hippocampal interneurons, explaining their decrease in Nkx2.1-Syngap1 f/1 adult mice (Xu et al., 2008). If this is the case, we would expect that targeted recombination would also induce the loss of other types of proliferating cells in Syngap1 f/1 and Syngap1 f/f embryos. In order to investigate this possibility, we induced targeted recombination of the Syngap1 flox allele using the Emx1-cre transgene, which is expressed in most excitatory neurons of the cortex and hippocampus and their proliferating progenitors (Gorski et al., 2002). Remarkably, we found that the dorsal telencephalon was severely under-developed in adult Emx1-Syngap1 f/1 and Emx1-Syngap1 f/f mice as well as in E13.5 Emx1-Syngap1 f/f embryos when compared with their Syngap1 1/1 and Syngap1 f/f counterparts (Fig. 10). This pattern of anomalies is consistent with the loss of proliferating cells on recombination of the Syngap1 flox allele Figure 8. Number of PV interneurons in the CA1 area of the hippocampus of PV-Syngap1 f/1 and Nkx2.1-Syngap1 f/1 mice. a, Confocal images of GFP 1 PV 1 cells (arrows) in the hippocampal CA1 from both Nkx2.1-Syngap1 1/1 (n = 3) and Nkx2.1-Syngap1 f/1 (n = 6) mice show significant decrease in the numbers of both GFP 1 PV 1 cells (b1; Mann-Whitney test: p = 0.0006) and GFP 1 PV À cells (b2; Mann-Whitney test: p = 0.0054) c. Confocal images of GFP 1 PV 1 cells (arrows) in the hippocampal CA1 from both PV-Syngap1 1/1 (n = 3) and PV-Syngap1 f/1 (n = 3) showed no change in the numbers of GFP 1 PV 1 cells (d; Mann-Whitney test: p = 0.7358). See Extended Data Figure 8-1, number of PV interneurons in the somatosensory cortex of PV-Syngap1 f/1 and Nkx2.1-Syngap1 f/1 mice. during telencephalon development. In contrast, targeted recombination of another Syngap1 conditional allele using the Emx1-cre transgene was not reported to affect the development of the telencephalon (Ozkan et al., 2014).
The Nkx2.1-cre transgene is also expressed postmitotically in the lineage of MGE-derived interneurons populating the hippocampus (Xu et al., 2008). Since targeted recombination between inverted loxP induces chromosomal elimination only in proliferating cells, cre expression in postmitotic cell should not affect the cellular composition of the hippocampus in Syngap1 flox/1 mice. We sought to test this hypothesis by counting the number of MGEderived cells in Syngap1 f/1 mice carrying the Pv-cre transgene, which drives the expression of cre in postmitotic PV 1 GABAergic interneurons starting on P12 (Yoon et al., 2000;Moon et al., 2002). We found that the number of EGFP 1 PV 1 and EGFP 1 PV À cells in the hippocampus and somatosensory cortex of adult Pv-Syngap1 f/1 mice was comparable to that of littermate controls (Fig. 8c,d; Extended Data Fig. 8-1c,d). Overall, these results are consistent with the possibility that targeted recombination of the Syngap1 flox allele in proliferating but not in postmitotic cells induces cell loss, explaining the reduction of the number of MGE-derived neurons in the hippocampus of Nkx2.1-Syngap1 f/1 mice.
Postmitotic cre-induced recombination does not induce chromosomal elimination in cells carrying inverted loxP sites, but it does result in the inversion of the sequences flanked by these sites. This inversion event being reversible because of the maintenance of the two loxP sites after recombination, ;50% of cre-expressing postmitotic cells are expected to carry the inverted DNA fragment at any given time point. We sought to determine whether such an inversion event is present in the brain of Nkx2.1-Syngap1 f/1 adult mice by performing PCR on genomic DNA (Fig. 11). As expected, we were able to amplify a PCR fragment corresponding to the nonrecombined Syngap1 flox allele from various brain areas of Nkx2.1-Syngap1 f/1 and Syngap1 f/1 adult mice, including the hippocampus, the cortex, the hypothalamus and the cerebellum (Fig. 11a,c). In contrast, we were able to amplify a PCR fragment corresponding to the inverted allele only from brain areas of Nkx2.1-Syngap1 f/1 Figure 9. Opposite orientations of loxP sites at the Syngap1 flox allele. a, Genomic organization of the Syngap1 gene. Shown are the 19 exons (boxes) of RefSeq NC_000083.6 that define the longest mRNA transcript, which codes for the Aa2 protein isoform. LoxP sites are found flanking exons 4-8 (numbered above the diagram). b, Syngap1 gene segment that harbors the two loxP insertions (yellow triangles) within introns 3 and 8 according to RefSeq (GRCm38). The italicized characters (E4-E10) underneath the scheme correspond to the exon numbering used by Vazquez et al. (2004) in their initial description of the Syngap1 flox allele. The half-filled box E4 represents exon 1 of the transcript coding for the B isoforms with its 59UTR on the clear side. The sequence of each insertion containing loxP sites (1 strand shown only) is shown as revealed by whole genome sequencing. The intron 3 insertion is 66 bp long whereas the intron 8 insertion is 163 bp. We confirmed that both insertions contain the 34 bp loxP sequence (in red). Each loxP site consists of an 8-bp central spacer region (bold and boxed) that gives the directionality of the site flanked by two 13-bp inverted repeats (italic). Note that each spacer sequence is a reverse complementary with respect to the other. mice that express the Tg(Nkx2.1-Cre) transgene postnatally or that have been populated by MGE-derived neurons, including the hippocampus, the cortex and the hypothalamus but not the cerebellum (Xu et al., 2008;Fig. 11b,d). We were not able to amplify this fragment from any brain areas of Syngap1 f/1 mice.
We also sought to determine whether the Syngap1 inversion is present in Emx1-Syngap1 f/1 adult mice by performing the same PCR experiments. We were able to amplify the PCR fragments corresponding to the nonrecombined and inverted Syngap1 flox allele from the olfactory bulb, ventral telencephalon, and kidneys of Emx1-Syngap1 f/1 and Emx1-Syngap1 f/f mice (Fig. 11a, e,f). In contrast, we were able to detect the nonrecombined but not the inverted allele from the cerebellum of these mice (Fig. 11a,e,f). These results are consistent with the observation that the Emx1-cre transgene is expressed in the olfactory bulb, ventral telencephalon, and kidneys but not in the cerebellum (Gorski et al., 2002).
Previous studies have shown that Nkx2.1 is expressed in all MGE progenitors and many, if not all, MGE postmitotic cells during early development (Sussel et al., 1999;Marin et al., 2000;Nóbrega-Pereira et al., 2008). However, MGEderived interneurons migrating toward the cortex and hippocampus rapidly downregulate Nkx2.1 expression after leaving the MGE (Nóbrega-Pereira et al., 2008). The expression of the Nkx2.1-cre transgene coincides with that of Nkx2.1 within the developing MGE but its expression later during development has not been characterized (Xu et al., 2008). Assuming that, like the endogenous gene, the expression of the transgene is downregulated in the MGE-derived interneurons migrating out of the MGE, we would not expect that recombination of the Syngap1 flox allele would occur after the late embryonic stages in these cells.
Recombination of the Syngap1 flox allele in postmitotic cells would result in the inversion of exons 4-8, which encode the entire PH and C2 domains of SYNGAP1. This inversion event is predicted to result in the production of a nonfunctional protein as the C2 domain has been shown to be critical for SYNGAP1 activity (Pena et al., 2008;Meili et al., 2021). Moreover, a previous study has established that targeted recombination of the Syngap1 flox allele in the adult hippocampus reduces SYNGAP1 protein production (Muhia et al., 2012). Interestingly, the reduction of protein levels observed in these experiments is relatively modest, which is consistent with the effect of a cre-induced inversion that would be reversible (see Discussion). All together, these results strongly suggest that the population of hippocampal MGE-derived interneurons of Nkx2.1-Syngap1 f/1 mice is divided into two groups, some with two functional copies of Syngap1 and some with only one copy.

Discussion
Deleterious mutations in SYNGAP1 gene cause ID, epilepsy, and ASD in humans (Hamdan et al., 2009(Hamdan et al., , 2011Berryer et al., 2013). Syngap1 haploinsufficiency in mice recapitulates several abnormalities observed in human with SYNGAP1 pathogenic variants, such as deficits in learning and memory and epileptic seizures (Komiyama et al., 2002;Clement et al., 2012;Nakajima et al., 2019). Syngap1 1/À mice display accelerated maturation of excitatory synapses during sensitive periods of development, leading to an imbalance in excitation-inhibition and an impairment of cognitive processes (Clement et al., 2012(Clement et al., , 2013. The role of SYNGAP1 in excitatory neuron network development is well established (Ozkan et al., 2014), but recent reports indicate that it also plays a critical role in the development and maturation of inhibitory synaptic circuits (Berryer et al., 2016;Su et al., 2019). Here, we used a conditional mouse model to uncover the impact of a decrease of Syngap1 expression in hippocampal inhibitory interneuron function.
Inverted loxP sites in the Syngap1 flox allele Unexpectedly, we found that the Syngap1 flox allele used in this study contains loxP sites that are oriented in opposite directions. The presence of inverted loxP sites at the Syngap1 locus has two potential consequences: (1) This configuration may induce the elimination of the Syngap1-carrying chromosome and subsequent cell loss on recombination between loxP sites in proliferating cells. Indeed, we found that Nkx2.1-cre expression resulted in a significant decrease of the number of MGE-derived interneurons in the CA1 hippocampus of Syngap1 f/1 mice. However, targeted recombination does not affect the PCR primers belonging to the same pair are illustrated as arrows of the same color with their orientation and approximate annealing location aligned on the corresponding strand. c-f, Agarose gel analyses of PCR products amplified from the region surrounding the loxP site in intron 8 in mice bearing the Syngap1 flox allele with or without transgenes expressing the cre recombinase (Nkx2.1-or Emx1-). c, e, Amplification using primer pair A (blue) showing products corresponding to the nonrecombined Syngap1 flox allele (464 bp) or to the wild-type allele (302 bp). d, f, Amplification using primer pair B (red) showing the presence or absence of a product predicted to be 429 bp, corresponding to the recombined (inverted) Syngap1 flox allele. Arrows on the left indicate the expected molecular weight of amplicons. Symbols on DNA molecular weight ladder: dot (500 bp) and star (300 bp). Abbreviations: H (hippocampus), C (cerebral cortex), Hy (hypothalamus), Cb (cerebellum), OB (olfactory bulb) VT (ventral telencephalon), K (kidney), bp (base pair). development of all MGE-derived interneurons, as a substantial proportion of these cells are generated and populate the hippocampus of Nkx2.1-Syngap1 f/1 mice. We hypothesize that MGE progenitors undergo a limited number of cell divisions in the presence of cre, allowing a subset of them to escape cell death. We have recently generated a new mouse line using a different Syngap1 conditional allele (Syngap1 floxGR ), where loxP sites flank exons 6-7 of the Syngap1 gene (Ozkan et al., 2014). We found no significant decrease of cortical and hippocampal MGE-derived interneurons numbers in these Tg(Nkx2.1-cre);Syngap1 floxGR/1 mice when compared with wild-type littermates, further supporting our conclusion that targeted recombination of the Syngap1 flox allele can induce the death of proliferating cells (unpublished data).
(2) Cre-induced recombination between the inverted loxP sites in postmitotic cells is predicted to result in the reversible inversion of the Syngap1 flox allele. As expected, we were able to detect both the nonrecombined and the inverted Syngap1 flox allele in brain areas that express the Nkx2.1-cre or Emx1-cre transgene or that have been populated by cells derived from lineages expressing these transgenes. These results strongly suggest that there are two populations of MGE-derived neurons in the hippocampus of Nkx2.1-Syngap1 f/1 adult mice: approximately half of them would carry two functional copies of Syngap1 while the other half would carry only one functional copy.
Our discovery that the Syngap1 flox allele is composed of inverted loxP sites sheds some light on previously published studies involving this mouse line. For instance, Knuesel et al. (2005) reported that the expression of the Camkii-cre transgene, which induces targeted recombination in excitatory forebrain neurons beginning approximately one week after birth, did not reduce the level of SYNGAP1 in Syngap1 ff mice below that of the heterozygous KO (Knuesel et al., 2005). Similarly, bilateral injection of a cre-expressing adenoviral vector into the hippocampus of Syngap1 ff mice led only to a modest reduction (24-42%) of SYNGAP1 protein levels (Muhia et al., 2012). It is possible that a greater reduction of SYNGAP1 protein levels was not achieved in these two experiments because the inversion induced by the recombination of the Syngap1 flox allele is reversible in postmitotic cells, resulting in about half of the cells with two functional copies of the Syngap1 gene.
Cell type-specific functional alterations in inhibitory interneurons of Nkx2.1-Syngap1 f/1 adult mice A previous study reported that Nkx2.1-Syngap1 f/1 mice showed reduced GABAergic neurotransmission, cognitive impairments and social behavioral deficits (Berryer et al., 2016). By performing whole-cell recordings in slice, we found here that MGE-derived interneurons from the hippocampus of these mice display cell-specific impairments. More specifically, we found that Nkx2.1 fastspiking interneurons (likely corresponding to parvalbumin (PV) expressing interneurons) are characterized by impaired membrane properties and evoked firing (Fig. 1), facilitated AMPAR-mediated, but not NMDAR-mediated, synaptic excitation (Fig. 2) and compromised short-term plasticity of excitatory inputs via a postsynaptic Ca 21 -dependent mechanism (Fig. 3). In contrast, Nkx2.1 regularspiking interneurons [likely corresponding to somatostatin (SOM) expressing interneurons] showed intact membrane and firing properties (Fig. 4), synaptic excitation (Fig. 5), and short-term plasticity of excitatory inputs (Fig. 6) comparable to controls. At the network level, the hippocampus of Nkx2.1-Syngap1 f/1 mice is characterized by impairment of pyramidal cell synaptic inhibition and enhancement of postsynaptic summation of excitatory responses (Fig. 7).
Overall, our results raise the possibility that conditional decrease of Syngap1 expression specifically in MGE-derived neurons causes cell type-specific changes that compromise fast-spiking interneuron firing and synaptic function and is associated with impaired pyramidal cell inhibition and enhanced synaptic excitation. Thus, cell-specific alterations in inhibitory interneuron function may contribute to the network hyper-excitability and cognitive impairments in mice models with global Syngap1 haploinsufficiency (Clement et al., 2012(Clement et al., , 2013Ozkan et al., 2014;Sullivan et al., 2020). Inhibitory interneuron dysfunction may also contribute to seizure susceptibility and cognitive deficits in humans with SYNGAP1 mutations (Hamdan et al., 2009(Hamdan et al., , 2011Berryer et al., 2013).
Previous results involving the Syngap1 flox allele (Berryer et al., 2016) and the results reported here have to be interpreted in light of our observation that the Syngap1 flox allele contains inverted loxP sites. On the one hand, it is possible that the decrease of the number of MGE-derived interneurons contributes to the behavioral and cellular deficits of Nkx2.1-Syngap1 f/1 mice. For instance, a decrease of these interneurons could contribute to the reduction of the synaptic inhibition of pyramidal cells that we observed in the hippocampus of Nkx2.1-Syngap1 f/1 mice. On the other hand, the presence of a nonfunctional allele in only half of the MGE-derived neurons of Nkx2.1-Syngap1 f/1 mice could attenuate the effect size of the deficits. However, as outlined below, several of the changes that we documented in MGE-derived interneurons are reminiscent of the role of Syngap1 in principal excitatory neurons (Clement et al., 2012;Ozkan et al., 2014) or were observed in interneurons in the context of global Syngap1 haploinsufficiency (Sullivan et al., 2020), suggesting that our observations might be relevant for the understanding of the role of Syngap1 in inhibitory interneurons.
Our results suggest that hippocampal fast-spiking interneurons from Nkx2.1-Syngap1 f/1 mice display intact resting membrane potential but reduced input resistance and impaired firing. The deficit in evoked firing may arise from the changes in input resistance. Interestingly, similar changes in membrane and firing properties were reported in prefrontal cortex PV fast-spiking interneurons in global Syngap1 1/À mice (Ozkan et al., 2014). Our findings of impaired membrane and firing properties of hippocampal fast-spiking interneurons in Nkx2.1-Syngap1 f/1 mice, suggest that the deficits observed in global Syngap1 1/À mice may result from reduced Syngap1 expression in inhibitory cells.
At the synaptic level, fast-spiking interneurons from Nkx2.1-Syngap1 f/1 mice display a potentiation of AMPAR-mediated synaptic responses with impaired short-term plasticity because of postsynaptic calcium-dependent mechanisms. cognitive alterations, including social behavioral deficits, supporting the hypothesis that Syngap1 plays a role in the establishment of GABAergic circuit function (unpublished data; Berryer et al., 2016;Su et al., 2019). This mouse line would thus be suitable for further studying the role of Syngap1 in inhibitory interneurons using whole-cell recordings.