Developmental loss of MeCP2 from VIP interneurons impairs cortical function and behavior

Rett Syndrome is a devastating neurodevelopmental disorder resulting from mutations in the gene MECP2. Mutations of Mecp2 that are restricted to GABAergic cell types largely replicate the behavioral phenotypes associated with mouse models of Rett Syndrome, suggesting a pathophysiological role for inhibitory interneurons. Recent work has suggested that vasoactive intestinal peptide-expressing (VIP) interneurons may play a critical role in the proper development and function of cortical circuits, making them a potential key point of vulnerability in neurodevelopmental disorders. However, little is known about the role of VIP interneurons in Rett Syndrome. Here we find that loss of MeCP2 specifically from VIP interneurons replicates key neural and behavioral phenotypes observed following global Mecp2 loss of function.

mature function. However, nothing is known about the contribution of VIP interneurons to neurodevelopmental dysregulation in Rett Syndrome. Using a mouse model, we generated conditional mutations of Mecp2 in VIP interneurons and compared these (i) with a conditional pan-interneuron mutation using the Dlx5/6 promoter to drive embryonic deletion in three major interneuron classes (VIP, parvalbumin-expressing interneurons [PV], and somatostatin-expressing interneurons [SST]) and (ii) with two conditional mutations in discrete interneuron populations (PV, SST). To identify the distinct contributions of each interneuron class, we assayed mortality, cortical activity, locomotor and anxiety phenotypes, and social behavior. Loss of MeCP2 selectively from VIP interneurons replicated key physiological and behavioral phenotypes observed in the pan-interneuron Dlx5/6 mutants, including altered firing rates, disruption of high-frequency cortical local field potential (LFP) patterns, and loss of state-dependent modulation of cortical activity. VIP interneuron-specific mutants further phenocopied impairments in marble burying and social behavior observed in the Dlx5/6 mutants. Overall, our findings suggest an unanticipated role for VIP interneuron dysfunction in the Mecp2 loss-of-function model of Rett Syndrome.

MeCP2 expression in PV, SST, and VIP interneurons
To confirm that MeCP2 is expressed in three major populations of GABAergic interneurons, we costained sections of cortex from adult mice with antibodies for interneuron markers and MeCP2 (Figure 1, Figure 1-figure supplement 1). As reported previously (Ito-Ishida et al., 2015), nearly all PV and SST interneurons expressed MeCP2. In addition,~80% of VIP interneurons expressed MeCP2, suggesting a previously unappreciated potential role for this signaling pathway in VIP interneuron development and function.

Seizure incidence following Mecp2 mutation
Previous work identified a characteristic seizure phenotype resulting from Mecp2 deletion in the brain (Chao et al., 2010) and found that loss of MeCP2 from SST-expressing cells may confer a lateonset tendency towards seizure (Ito-Ishida et al., 2015). We therefore evaluated the incidence of seizure in each of the three interneuron-specific Mecp2 deletion lines from weaning through to late adulthood. We compared the impact of MeCP2 depletion from VIP, PV, or SST interneuron populations with simultaneous depletion from all three interneuron classes using the Dlx5/6 Cre line. We further compared the interneuron-specific mutation mice with Mecp2 f/y littermate controls. To identify the relative impact of Mecp2 loss of function in GABAergic cells compared to loss of function of Mecp2 in all cells, we also compared seizure incidence in the interneuron-specific deletion mice with that in mice carrying a complete knockout of the Mecp2 gene (Mecp2 -/y ).
MeCP2 depletion from specific interneuron populations had markedly different effects on seizure incidence ( Figure 1-figure supplement 2A-B). We found that 100% of Mecp2 -/y (n = 10) and Dlx5/ 6 mutant (n = 13) mice exhibited at least one seizure, compared to only 52.9% of SST mutants (n = 36), 35.0% of PV mutants (n = 22), and 37.5% of VIP mutants (n = 8). In comparison, the seizure rates in Mecp2 f/y controls and in wild-types were 17.1% (n = 38) and 0% (n = 20), respectively, suggesting a small contribution of the floxed allele to the seizure phenotype (Ito-Ishida et al., 2015). The mean age of initial seizure was significantly earlier in Mecp2 -/y mutants, but not the Dlx5/6, PV, SST, or VIP mutants, compared to Mecp2 f/y controls (Figure 1-figure supplement 2B). Cumulative distribution plots of survival for controls (CON; black; n = 38), wild-types (WT; magenta; n = 10), and Mecp2 -/y (KO; gray; n = 10), Dlx5/6 (DLX; red; n = 13), VIP (orange; n = 8), PV (green; n = 22), and SST (cyan; n = 36) mutants. *, p<0.05. The online version of this article includes the following source data and figure supplement(s) for figure 1:   Altered patterns of cortical and hippocampal activity following Mecp2 mutation To examine the cellular and local network consequences of MeCP2 loss, we performed electrophysiological recordings in the cortex and hippocampus of awake animals. MeCP2 loss from interneurons caused alterations in the cortical LFP, a measure of local network activity (Figure 2A-B). MeCP2 depletion caused a modest change in LFP power, measured during periods of quiescence, in the 3-6 Hz range in the Dlx5/6 mutants (p=0.03; Kruskal-Wallis test with Dunn's post-test), but not in other groups ( Figure 2C). However, we observed a robust broadband decrease in high-frequency LFP activity in the Dlx5/6 mutants that was replicated in the VIP mutants, but not PV or SST mutants ( Figure 2A). Quantification of high-frequency activity around the gamma (40-55 Hz) band revealed a significant decrease in gamma-range activity in both the Dlx5/6 (p=0.005) and VIP mutants (p=0.04; Kruskal-Wallis test with Dunn's post-test; Figure 2D). We further found a loss of spike-field coherence in the gamma band in Dlx5/6 and VIP mutants ( Figure 2-figure supplement 1). By contrast, hippocampal recordings revealed a loss of gamma-range LFP power in the Dlx5/6 mutants that was replicated by the SST mutants ( Figure 2-figure supplement 2), suggesting potentially distinct celltype-specific roles in different brain areas. Perturbation of interneurons during development can result in elevated firing rates due to loss of synaptic inhibition and the reorganization of neural circuits (Close et al., 2012;Rossignol et al., 2013;Batista-Brito et al., 2017). We therefore recorded cortical firing activity in awake mice with Mecp2 mutations ( Figure 2B). Single-unit recordings revealed that loss of MeCP2 in the Dlx5/6 mutants led to a three-fold increase in the firing rates of regular-spiking, putative excitatory pyramidal neurons as compared to that in control animals (p=0.004), and this finding was replicated in the VIP (p=0.03) and SST (p=0.002; Kruskal-Wallis test with Dunn's post-test) mutants ( Figure 2E).
In previous work, we found that cortical firing rates are robustly modulated by changes in behavioral state, and are typically increased at the onset of locomotion (L) as compared to quiescence (Q) . Loss of normal VIP interneuron activity reduces this state-dependent cortical modulation (Fu et al., 2014;Batista-Brito et al., 2017). To determine whether MeCP2 loss from VIP interneurons impairs this function, we examined state-dependent modulation in the Mecp2 mutants. Both pan-interneuron MeCP2 loss in the Dlx5/6 mutants (p=0.004) and MeCP2 loss specific to VIP interneurons (p=0.02; Kruskal-Wallis test with Dunn's post-test), but not other interneuron populations, led to decreased state-dependent modulation of single-unit cortical firing rates, measured as a modulation index [(FR L -FR Q )/(FR L +FR Q )] ( Figure 2F). We did not find any differences in neural activity between Mecp2 f/y controls, wild-types, and Vip Cre controls, suggesting that these results arise from cell type-specific Mecp2 loss of function ( Figure 2-figure supplement 3).

Behavioral phenotypes associated with interneuron-specific Mecp2 mutation
Previous work has linked Mecp2 mutations in GABAergic interneurons to key impairments in motor, repetitive, and social behaviors (Moretti et al., 2005;Chahrour and Zoghbi, 2007;Samaco et al., 2008;Chao et al., 2010;Kaufmann et al., 2012;He et al., 2014;Ito-Ishida et al., 2015). We therefore examined the behavioral consequences of VIP-specific Mecp2 mutations. In agreement with previous work (Chao et al., 2010;Ito-Ishida et al., 2015), we found no significant impairments in percentage of time spent in the open arms of the elevated plus maze task, a measure of anxiety (Carobrez and Bertoglio, 2005), for the Dlx5/6, VIP, PV, or SST mutants compared to controls ( Figure 3A). Likewise, we found no impairment in locomotor behavior in the open field assay in any of the mutants (Figure 3-figure supplement 1A). By contrast, we found a significant impairment in marble burying, a measure of motor and repetitive behavior (Thomas et al., 2009;Silverman et al., 2010), in the pan-interneuron Dlx5/6 mutants compared to the controls (p=0.007). The deficit observed in the Dlx5/6 mutants was fully replicated in the VIP mutants (p=0.001; Kruskal-Wallis test with Dunn's post-test), but not in the PV or SST mutants ( Figure 3B). We tested social interaction behaviors in each Mecp2 deletion line using the three-chamber sociability task (Nadler et al., 2004). Control animals exhibited a significant preference for a chamber containing a conspecific over an empty chamber ( Figure 3C, Figure 3-figure supplement 1B). By contrast, the pan-interneuron Dlx5/6 mutants exhibited a reverse preference for the empty chamber (p=0.002). The VIP mutants (p=0.002; Kruskal-Wallis test with Dunn's post-test), but not the PV or SST mutants, fully replicated the effects of the pan-interneuron deletion, showing a significant preference for the empty chamber over the conspecific.
In addition to altered overall social preferences, Mecp2 deletion affected the number of approaches mice made towards conspecifics. Control animals made more approaches to the conspecific than to the empty holding cage ( Figure 3C). By contrast, the Dlx5/6 mutants made more approaches to the empty holding cage than to the conspecific animal (p=0.01). The VIP mutants (p=0.01; Kruskal-Wallis test with Dunn's post-test), but not the PV or SST mutants, fully replicated these effects of pan-interneuron deletion, approaching the empty holding cage more than the conspecific. Together, these data suggest that VIP interneurons may contribute to the deficits in social behavior caused by global Mecp2 dysfunction. We did not find any differences in behavior between Mecp2 f/y controls, wild-types, and Vip Cre controls (Figure 3-figure supplement 2), suggesting that these behavioral impairments arise as a consequence of cell-type-specific MeCP2 deletion.

Discussion
Our results reveal an unanticipated role for VIP interneurons in the Mecp2 loss-of-function model of Rett Syndrome. On the basis of previous characterizations of the Mecp2 model (Chahrour and Zoghbi, 2007;Samaco et al., 2008;Chao et al., 2010;He et al., 2014;Ito-Ishida et al., 2015), we examined interneuron contributions to several major categories of neural dysregulation: mortality and seizure, cortical activity patterns, anxiety and repetitive behaviors, and social behavior. We found that Mecp2 deletion from VIP interneurons recapitulates major phenotypes observed following pan-interneuron Mecp2 deletion at the levels of both neural activity and behavior (Supplementary file 1).
Patients who have Rett Syndrome exhibit respiratory impairments and seizure (Chahrour and Zoghbi, 2007), and these phenotypes are replicated in mouse models following deletion of Mecp2 from all GABAergic cells (Chao et al., 2010;Ito-Ishida et al., 2015). In agreement with previous work (Ito-Ishida et al., 2015), we found that Mecp2 loss of function in SST interneurons conferred a late seizure phenotype. However, conditional mutations of Mecp2 in VIP interneurons did not increase seizure or mortality rates. Mecp2 knockout animals had significantly earlier seizure onset and mortality than the Dlx5/6 mutants, supporting previous findings that global loss of MeCP2 in excitatory neurons also contributes to seizure (Goffin et al., 2014;Meng et al., 2016) and that respiratory impairments in the Mecp2 knockouts may increase early mortality (Chen et al., 2001;Guy et al., 2001). In comparison, a previous study did not observe a seizure phenotype in Dlx5/6 mutants, possibly as a result of methodological differences (Chao et al., 2010). We observed a low rate of late-onset seizure in the MeCP2 f/y control animals, indicating that the decrease in MeCP2 levels associated with the conditional allele (Samaco et al., 2008) may be associated with epileptogenic consequences in addition to some mild behavioral phenotypes.
Recordings in the cortex of awake mice revealed a robust impact of Mecp2 deletion on the pattern of cortical activity. We found a decrease in high-frequency LFP activity in the Dlx5/6 mutants that was replicated in the VIP mutants. In particular, VIP mutants exhibited a decrease in cortical gamma-range activity, which is associated with cognition and sensory processing (Cardin, 2016;Cardin, 2018a) and which may be impaired in Rett Syndrome (Peters et al., 2015). Although hippocampal gamma-range activity was also affected in the Dlx5/6 mutants, these effects were associated with SST rather than with VIP interneuron mutations, suggesting potential heterogeneity of the circuit-level impact of Mecp2 deletion across brain areas. Loss of MeCP2 in the Dlx5/6 mutants was associated with a three-fold increase in cortical firing rates, and this increase was replicated in the VIP and SST mutants.
In the healthy cortex, transitions between behavioral states, such as quiescence and arousal or locomotion, are associated with robust modulation of cortical firing rates. (Niell and Stryker, 2010;McGinley et al., 2015;Vinck et al., 2015;Tang and Higley, 2020). However, loss of MeCP2 in the VIP and Dlx5/6 mutants caused a profound dysregulation of this state-dependent modulation. This loss of modulation is unlikely to result from a 'ceiling effect' caused by increased overall firing rates, as putative excitatory neurons in the SST and VIP mutants exhibited equally enhanced firing but only the VIP mutants showed a loss of state-dependent modulation. VIP interneurons are thought to play a key role in regulating the state-dependent modulation of cortical circuits, partly via strong inhibition of SST interneurons and consequent disinhibition of pyramidal neurons (Pi et al., 2013;Fu et al., 2014;Prö nneke et al., 2015;Karnani et al., 2016;Muñoz et al., 2017). We had previously found that developmental perturbation of VIP interneurons by conditional deletion of the schizophrenia-associated gene ErbB4 caused a similar loss of state-dependent cortical regulation (Batista-Brito et al., 2017). Together, these findings suggest that disruption of state-dependent cortical dynamics may be a common outcome of disease mechanisms affecting VIP interneurons.
Previous work has highlighted alterations in repetitive and motor behaviors, but not anxiety, following GABAergic deletion of Mecp2 (Chao et al., 2010;Ito-Ishida et al., 2015;Ure et al., 2016). We therefore examined the impact of Mecp2 deletion from VIP interneurons on locomotor and anxiety phenotypes. None of the Mecp2 loss-of-function mutations resulted in anxiety-related or locomotor phenotypes in the elevated plus maze or the open field, respectively, in agreement with previous work (Chao et al., 2010;Ito-Ishida et al., 2015). However, the pan-interneuron Dlx5/6 mutants exhibited deficits in marble burying, a task that is susceptible to altered anxiety and OCDlike behaviors as well as changes in fine motor function (Thomas et al., 2009;Silverman et al., 2010). Cell-type-specific mutations of Mecp2 in VIP interneurons, but not in the PV or SST populations, phenocopied this behavioral impairment.
Abnormal or reduced social behavior is a hallmark of many autism spectrum disorder models, and has previously been shown in mice lacking MeCP2 in all GABAergic populations (Chao et al., 2010;Ito-Ishida et al., 2015). We found that the pan-interneuron Dlx5/6 Mecp2 mutants exhibited a reversal of normal social preferences in the three-chamber sociability assay, preferring an empty chamber to one containing a conspecific. Notably, Mecp2 deletion from VIP, but not from PV or SST, interneurons fully replicated this phenotype. SST-specific deletion led to loss of any social preference, suggesting a potential contribution of both VIP-and SST-expressing cells to deficits in social behavior following global Mecp2 loss of function. Together, these results suggest a previously unknown and potentially important role for VIP interneuron dysregulation in social behavior deficits in the Mecp2-deletion model.
Overall, our behavioral findings in the Dlx5/6, PV and SST mutants are in general agreement with previous work (Supplementary file 1). The Pvalb Cre line used here largely expresses Cre recombinase in PV interneurons, along with some thalamocortical projection neurons (Hippenmeyer et al., 2005;Cardin et al., 2009). In comparison, the Pvalb-2A-Cre line used in some previous work (Goffin et al., 2014;Ito-Ishida et al., 2015) also expresses Cre in a subset of cortical pyramidal neurons and additional thalamic nuclei (Madisen et al., 2010). These differences may contribute to the more severe behavioral phenotypes and early mortality previously observed in the Pvalb-2A-Cre line (Ito-Ishida et al., 2015). Other work examining the consequences of Mecp2 deletion in the Pvalb Cre line likewise found only mild behavioral phenotypes (He et al., 2014). However, as the PV promoter only becomes active postnatally, our findings do not preclude a substantial contribution of embryonic Mecp2 deletion from PV interneurons to Rett Syndrome phenotypes.
We find a unique impact of Mecp2 deletion from VIP interneurons. Despite being few in number (Rudy et al., 2011), VIP interneurons are targets of multiple neuromodulatory systems and play critical roles in state-dependent regulation of local neural circuits (Pi et al., 2013;Fu et al., 2014;Garcia-Junco-Clemente et al., 2017;Muñoz et al., 2017), making them a potential key point of vulnerability in neurodevelopmental disease. Although our electrophysiology results are specific to cortex and hippocampus, the three interneuron classes examined here exhibit distinct cellular-and circuit-level properties and play key roles across many brain areas that contribute to behavior. In addition, dysregulation of one GABAergic population is probably amplified by extensive synaptic connectivity with other inhibitory interneuron classes (Pfeffer et al., 2013;Cardin, 2018a). Indeed, our previous work suggests that developmental disruption of VIP interneuron activity may have multiple circuit-level consequences, including loss of synaptic inhibition of other interneurons, altered experience-dependent plasticity, and dysregulated cortical circuit maturation, in addition to ongoing perturbation of normal adult function. Mecp2 loss of function across multiple GABAergic interneuron classes may thus exert diverse influences on neural and behavioral deficits in Rett Syndrome.

Materials and methods
Animals All experiments were approved by the Institutional Animal Care and Use Committee of Yale University. We used the Dlx5/6 Cre (JAX#008199; Monory et al., 2006), Pvalb Cre (JAX#008069; Hippenmeyer et al., 2005), Sst Cre (JAX#013044; Taniguchi et al., 2011), and Vip Cre (JAX#010908; Taniguchi et al., 2011) mouse lines to target all forebrain GABAergic interneurons, parvalbuminexpressing interneurons (PV), somatostatin-expressing interneurons (SST), and vasoactive intestinal peptide-expressing interneurons (VIP), respectively. We crossed each Cre line to the conditional Mecp2 line (Mecp2 f/f ; JAX# 007177; Guy et al., 2001). In each case, we assayed male mice that were hemizygous for the floxed Mecp2 allele and heterozygous for Cre. All crosses were made on a C57BL/6J background (JAX#000664). Control animals were Cre-negative male mice that were hemizygous for the floxed Mecp2 allele (Mecp2 f/y ). We further compared Mecp2 f/y mice with age-matched wild-type C57Bl/6 mice (JAX#000664) and Vip Cre mice (JAX#010908). In a subset of experiments, we compared the interneuron-specific crosses with male mice from the Mecp2 knockout line (Mecp2 -/y ; JAX#003890; Guy et al., 2001). All behavioral assays were performed at P120 except for those involving the Dlx5/6 Cre+/-Mecp2 f/y animals, which were assayed at P90 due to their early morbidity. All behavioral and electrophysiological assays were carried out in animals with no prior seizure incidence.

Immunohistochemistry
For immunofluorescent staining of brain tissue, mice were perfused with 4% paraformaldehyde and post fixed for an hour before transferring into successive sucrose solutions at 15% and 30%. 20 mm thick cryosections were prepared for immunohistochemistry (IHC). Tissue was incubated with 1.5% normal goat serum (NGS) (Life Technologies) and 0.1% Triton X-100 (Sigma) in PBS for 60 min at room temperature. Sections were incubated with primary antibodies (Rat Anti-Somatostatin 1:250 [Millipore MAB354]; Anti-parvalbumin 1:1000 [Sigma P3088]; Anti-VIP 1:250 [ImmunoStar 20077]; Anti-MeCP2 1:250 [Millipore 07-013]) in the blocking buffer overnight at 4˚C. After washing three times with buffer, sections were incubated with secondary antibodies for 1 hr at room temperature (secondary antibodies: Alexa Fluor 488, 594 or 647 [Life Technologies, 1:1000]). Finally, coverslips were mounted using ProLong Gold Mounting Medium with DAPI (Life Technologies) and imaged at 10x. Quantifications were performed in Adobe Photoshop. Pictures were divided into a grid measuring 1 Â 1 mm in total and cells were counted in each grid square. The number of cells positive for antibody staining against MeCP2 was counted to assay the proportion of co-expressing cells.

Headpost surgery and wheel training
For recordings performed in awake animals, mice were initially handled for 5-10 min/day for 5 days prior to a headpost surgery. On the day of the surgery, the mouse was anesthetized with isoflurane and the scalp was shaved and cleaned three times with betadine solution. An incision was made at the midline and the scalp resected to each side to leave an open area of skull. Two skull screws (McMaster-Carr) were placed at the anterior and posterior poles. Two nuts (McMaster-Carr) were glued in place over the bregma point with cyanoacrylate and secured with C&B-Metabond (Butler Schein). The Metabond was extended along the sides and back of the skull to cover each screw, leaving a bilateral window of skull uncovered over primary visual cortex. The exposed skull was covered with a layer of cyanoacrylate. The skin was then glued to the edge of the Metabond with cyanoacrylate. Analgesics were given immediately after the surgery and on the two following days to aid recovery. Mice were given a course of antibiotics (Sulfatrim, Butler Schein) to prevent infection and were allowed to recover for 3-5 days following implant surgery before beginning wheel training.
Once recovered from the surgery, mice were trained with a headpost on the wheel apparatus. The mouse wheel apparatus was 3D-printed (Shapeways Inc) in plastic with a 15 cm diameter and an integrated axle and was spring-mounted on a fixed base. A programmable magnetic angle sensor (Digikey) was attached for continuous monitoring of wheel motion. Headposts were customdesigned to mimic the natural head angle of the running mouse, and mice were mounted with the center of the body at the apex of the wheel. On each training day, a headpost was attached to the implanted nuts with two screws (McMaster-Carr). The headpost was then secured with thumb screws at two points on the wheel. Mice were headposted in place for increasing intervals on each successive day. If signs of anxiety or distress were noted, the mouse was removed from the headpost and the training interval was not lengthened on the next day. Mice were trained on the wheel for up to 7 days or until they exhibited robust bouts of running activity during each session. Mice that continued to exhibit signs of distress were not used for awake electrophysiology sessions.

Locomotion detection
Wheel position was extracted from the output of a linear angle detector. We used a change-point detection algorithm that detected statistical differences in the distribution of locomotion velocities across time (see Vinck et al., 2015;Batista-Brito et al., 2017). Quiescent periods that lasted longer than 20 s were selected for analysis. For analysis of modulation with changes in behavioral state, we selected trials for which the preceding quiescent period lasted longer than 20 s, average speed until the next locomotion offset point exceeded 1 cm/s, and running lasted longer than 2 s.

Extracellular recordings
LFP recordings were made with tetrodes (Thomas Recording GMBH, Germany) targeted to layers 2/ 3 and 5 of visual cortex and to the CA1 field of the dorsal hippocampus (AP: +1.5-2 mm; ML: 1.2-1.75, Paxinos and Franklin, 2001). Signals were digitized and recorded with a DigitalLynx 4SX system (Neuralynx, Bozeman MT). All data were sampled at 40 kHz and recordings were referenced to the cortical surface. LFP data were recorded with a bandpass 0.1-9000 Hz filter.

Spike sorting
Spikes were clustered using previously published methods Batista-Brito et al., 2017). We first used the KlustaKwik 3.0 software (Kadir et al., 2013) to identify a maximum of 30 clusters using the waveform energy and the energy of the waveform's first derivative as clustering features. We then used a modified version of the M-Clust environment to separate units manually. Units were accepted if a clear separation of the cell relative to all the other noise clusters was observed, which generally was the case when isolation distance (ID) (Schmitzer-Torbert et al., 2005) exceeded 20 . We further ensured that maximum contamination of the ISI (inter-spike-interval) histogram did not exceed 0.1% at 1.5 ms.

Electrophysiology analysis
The firing rate was computed by dividing the total number of spikes a cell fired in a given period by the total duration of that period. To examine whether firing rates were significantly changed around locomotion onset, we computed the firing rate in the [À0.5, 0.5] s window around locomotion onset (L; as in Vinck et al., 2015) and compared this to the firing rate in the [À5,-2] s quiescence (Q) period before locomotion onset by computing a modulation index ([FR L -FR Q ]/[FR L +FR Q ]). All LFP power analyses were made using data from quiescent periods after animals had been stationary for a minimum of 20 s and excluding data from within 10 s of the next locomotion bout. Relative power in the specified frequency bands was measured as a ratio between power in those bands and the total power. Spike-field coherence measures were performed as previously described (Miri et al., 2018), analysis code available at https://github.com/jesscardin/Miri-Vinck-et-al (Cardin, 2018b; copy archived at https://github.com/elifesciences-publications/ Miri-Vinck-et-al). Power spectra were normalized to total power for visualization purposes.

Seizure detection
All mice were handled for at least 10 min each day throughout the study. During the daily handling regime, the mice were assessed for seizures. If a seizure did occur, the mouse was immediately returned to its home cage. Seizure severity was scored using the Racine scale (1: Mouth and facial movement, 2: Head nodding, 3: Forelimb clonus, 4: Forelimb clonus and rearing, 5: Forelimb clonus, rearing, and falling). Seizures were defined as events reaching Racine scale levels 4 or 5, with animals exhibiting rearing and forelimb clonus or rearing, forelimb clonus, and falling.

Morbidity analysis
After weaning at P21, all mice were monitored every day throughout the study. All deaths were noted, and animals were tracked daily until P500.

Behavioral analysis
The elevated plus maze, marble-burying, and sociability assays were performed under low-level (20-25 lux) illumination. In each assay, mice were given 15 min to acclimate to the behavioral assay room. In all cases, the researcher was blind to the genotypes of the mice until after all behavioral data were scored.

Elevated plus maze
Custom Labview software was used to control a camera recording the mouse's locomotion in the maze. At the beginning of the session, the mouse was placed at the center of the maze and allowed to freely move on either arm for five minutes. At the end of the session, the mouse was returned to the home cage and the maze was cleaned for the next mouse. Video recordings of mouse behavior were hand-scored to determine the amount of time spent in the open and closed arms of the maze.

Open field
The open field assay was performed in a 30 cm square box divided into nine quadrants. Custom software (Labview) was used to control a camera recording the mouse's path in the box. At the start of the session, the mouse was placed in the center quadrant of the box and allowed move freely for 20 min. After the time elapsed, the mouse was returned to the home cage and the box was cleaned for the next mouse. ImageJ software was used to analyze the total distance traversed by the mouse.

Marble burying
12 marbles were evenly placed in a cage with 1 inch of clean bedding. Custom Labview software was used to control a camera recording the mouse's activity in the cage. The mouse was placed in the center of the cage with the marbles and allowed to explore the cage for 20 min. At the end of the session, the mouse was returned to the home cage and the marbles were cleaned with a 10% bleach solution.
The proportion of marbles buried was analyzed in ImageJ using the 'analyze particles' function to compare the initial and final exposed surface area of the marbles.

Sociability
The sociability apparatus was divided into three equal areas with Plexiglas dividers, each with an opening allowing access to neighboring chambers. Custom Labview software was used to control a camera that recorded the mouse's activity in the chamber. An unfamiliar age-and sex-matched conspecific mouse (reared in separate cages, C57BL/6 genotype) was placed into a small cylindrical holding cage in one side of the chamber and an identical empty holding cage was placed in the other side. The location of the conspecific was randomly varied across trials. At the beginning of the session, the test mouse was placed in the central chamber and habituated to the central chamber for ten minutes. The dividers were then removed to allow the mouse to move freely among all the chambers for ten additional minutes. At the end of the session, the mice were returned to their home cages and the apparatus was cleaned. Video recordings of the mouse's behavior were scored to determine the amount of time spent in each of the three partitions and the number of approaches that the test mouse made to the conspecific and the empty holding cage. An approach was defined as the test mouse coming within a 5-cm radius of a cage or making contact with a holding cage. Social preferences were calculated both as comparisons of raw values and as index values for time spent ([Time C -Time E ]/[Time C +Time E ]) and approaches ([App C -App E ]/[App C +App E ]), where C denotes conspecific and E denotes the empty container.

Statistical analysis
Paired and unpaired non-parametric tests generated in GraphPad Prism (version 8 for Mac; San Diego CA) were used throughout the study to accommodate non-normal data distributions. Animals were used as the 'n' in all analyses. Exact p values and estimation statistics are reported in the source data files for all tests. All group data are shown as box-and-whisker plots in which the bars denote the minimum and maximum of the distribution and the box denotes the first and third quartiles and the median. experiments were approved by the Institutional Animal Care and Use Committee of Yale University (#11317).

Additional files
Supplementary files . Supplementary file 1. Consequences of cell-type-specific Mecp2 deletion. Summary table of phenotypes observed in conditional deletion mice with loss of Mecp2 function in GABAergic interneurons (upper) and glutamatergic excitatory neurons (lower). Observed phenotypes are noted as Y/N, assays that were not performed in a given study are left blank. The two distinct Pvalb Cre mouse lines used by different studies are identified by their JAX line numbers.
. Transparent reporting form Data availability Source data files are included for each figure and supplementary figure. Analysis code is available at https://github.com/jesscardin/Miri-Vinck-et-al (copy archived at https://github.com/elifesciencespublications/ Miri-Vinck-et-al). All data included in this study will be freely available upon request, as the data files and associated intermediate analysis files are very large (400GB) and depositing the full data is not feasible.