Dlx5/6 Expression Levels in Mouse GABAergic Neurons Regulate Adult Parvalbumin Neuronal Density and Anxiety/Compulsive Behaviours

Neuronal circuits integrating Parvalbumin-positive GABAergic inhibitory interneurons (PV) are essential for normal brain function and are often altered in psychiatric conditions. During development, Dlx5 and Dlx6 (Dlx5/6) genes are involved in the differentiation of PV-interneurons. In the adult, Dlx5/6 continue to be expressed at low levels in most telencephalic GABAergic neurons, but their importance in determining the number and distribution of adult PV-interneurons is unknown. Previously, we have shown that targeted deletion of Dlx5/6 in mouse GABAergic neurons (Dlx5/6VgatCre mice) results in altered behavioural and metabolic profiles. Here we evaluate the consequences of targeted Dlx5/6 gene dosage alterations in adult GABAergic neurons. We compare the effects on normal brain of homozygous and heterozygous (Dlx5/6VgatCre and Dlx5/6VgatCre/+ mice) Dlx5/6 deletions to those of Dlx5 targeted overexpression (GABAergicDlx5/+ mice). We find a linear correlation between Dlx5/6 allelic dosage and the density of PV-positive neurons in the adult prelimbic cortex and in the hippocampus. In parallel, we observe that Dlx5/6 expression levels in GABAergic neurons are also linearly associated with the intensity of anxiety and compulsivity-like behaviours. Our findings reinforce the notion that regulation of Dlx5/6 expression is involved in individual cognitive variability and, possibly, in the genesis of certain neuropsychiatric conditions.


Introduction
Brain function depends on neuronal microcircuits composed of excitatory neurons responsible for long-and short-range signal transmission and regulatory GABAergic inhibitory interneurons. The tuning of these neuronal networks affects most brain functions, including cognition, perception [1] and social behaviour [2]. GABAergic interneurons constitute a heterogeneous class of cells which can be classified on the basis of their peculiar anatomical, biochemical and physiological features [3]. They were divided into three major classes characterised by the expression of either parvalbumin (PV), somatostatin (SST) or serotonin receptor 3A (5HTr3A), although the great diversity of GABAergic interneurons is only now appreciated due to single cell transcriptomic analysis and has resulted in the identification of more than 20 sub-classes [4,5]. These diverse adult GABAergic subtypes are generated through well-defined transcriptional trajectories in which the sequential expression of groups of transcription factors (TFs) leads to the progressive differentiation The level of Dlx5/6 expression seems, therefore, to be essential for maintaining the equilibrium between the populations of inhibitory PV interneurons and that of excitatory neurons. DLX5/6 could therefore play a role in determining the excitatory/inhibitory (E/I) balance, a parameter which has been associated to neuropsychiatric conditions such as ASD and schizophrenia [36,37].

Animals
Mice were housed in light, temperature (21 °C) and humidity (50-60% relative humidity) controlled conditions. Food and water were available ad libitum. Mice were individually identified by a microchip postnatally implanted 3 weeks. Litter sizes and genotypes were recorded. WT animals were from Charles River, France. Slc32a1 tm2(cre)Lowl knockin mice (here referred as Vgat cre/+ mice) were purchased from Jackson Laboratories through Charles River, France. In the developing telencephalon, Vgat was expressed in post-mitotic GABAergic neurons staring at E11.5 [38]. All mutant strains were backcrossed and bred on a mixed C57BL6/N X DBA/2N genetic background.
To obtain mice carrying the R26R CAG-flox-Dlx5/+ allele, an F3/FRT-flanked cassette containing the CAG promoter, a floxed stop sequence, flag-tagged mouse Dlx5 cDNA and a poly(A) additional signal were inserted into the targeting vector pROSA26-1 (P. Soriano, Mount Sinai School of Medicine, New York, NY, USA) (Addgene, plasmid 21714) [35]. R26R CAG-flox-Dlx5/+ mice were crossed with Vgat cre/+ to induce GABAergic-specific expression of Dlx5 (GABAergic Dlx5/+ mice) and then backcrossed for more than 10 generations on a mixed C57BL6/N X DBA/2N genetic background ( Figure 1A).  [35] and Vgat cre/+ mice to induce GABAergic-specific expression of Dlx5 and then backcrossed for more than 10 generations on a mixed C57BL6/N X DBA/2N genetic background. (B) qPCR analysis of Dlx5 expression level in the parietal cortex of control and GABAergic Dlx5/+ littermates. The level of Dlx5 expression almost doubled in animals compared with controls. (C-F') Dlx5 expression was revealed by in situ hybridisation performed in parallel and with identical conditions on serial sections of young adult brains from control (C-F) and GABAergic Dlx5/+ (C'-F') mice. Strong induction of Dlx5 expression was observed in all GABAergic areas, including the olfactory bulb (C,C'), the parietal cortex (D,D'), the hypothalamus (E,E') and the cerebellum (F,F'). Dlx5 expression was also detected in GABAergic neurons not normally expressing the gene, such as, for example, Purkinje cells in the cerebellum (F,F'). cb, cerebellum; ht, hypothalamus; ob, olfactory bulb; pc, parietal cortex; Pcl, Purkinje cells layer. Bar: 250 µm C-E'; 100 µm F-F'. Levels of significance (** p ≤ 0.01).
Mice of both sexes were used. In all experiments, littermates with no Vgat cre alleles were used as controls. Results obtained with controls from GABAergic Dlx5/+ and Dlx5/6 VgatCre mice for either cortical cell counting or behavioural tests were compared and no significant differences were observed.

Behavioural Tests
Behavioural procedures were conducted between 9 a.m. and 5 p.m. in a dim and quiet room, not housing any other animal. Observers were blind to the experimental design. Mice were taken to the test room 30 min before the test and left in the absence of the observer.

− Open Field Test with object exploration
We used an open field test (OFT) with a centrally located object to measure anxiety-like and exploratory behaviours of mice placed in a novel environment [39]. The equipment consisted of a closed square arena (72 × 72 cm). The computer defined the grid lines dividing the box floor into 16 equal-sized squares, with the central four squares regarded as the central region. First, mice were familiarised with the empty arena for 10 min and placed back in their home cage for 2 min. A cylindrical plastic tube (diameter 3 cm, height 7 cm) (object, OB) was placed in the centre of the arena. Each mouse was then placed at one corner of the arena facing the wall and tracked and recorded for 10 min. Films were analysed by Ethovision system (Noldus). Latency to the first entry in the centre, number of entries in the centre, latency to interact with the OB, number of interactions with the OB and duration of OB sniffing were analysed. To eliminate olfactory cues, the equipment was thoroughly cleaned between each test.

− Marble burying test (MBT)
We used the marble-burying test (MBT) to measure anxiety-and compulsive-like behaviours. A clear Plexiglas box (36.5 cm long × 20.7 cm wide × 14 cm high) was filled with 3 cm of standard bedding. A total of 20 glass marbles were placed on the surface of the shavings. Mice were individually placed in the centre of the box and left for 10 min. At the end of the session, a picture of the marbles was taken, and the marbles buried index was counted with the Fiji (ImageJ) image-processing program.

− Nest building test
Each mouse, aged less than one year, was housed in a single cage before testing. During the test, a paper towel (30 cm × 21,5 cm) was placed in the cage and left for one week. The nest quality was scored daily at 10 a.m. into four categories as shown in [15]: 1-no interaction with intact paper, 2-paper partially torn, 3-paper completely torn, and 4-nest completely built.

Immunohistochemistry
Animals were deeply anesthetised and perfused intracardially with 4% paraformaldehyde in phosphate buffer. Brains were removed and postfixed overnight at 4 • C in the same fixative and cryoprotected by immersion in 30% sucrose. Cryoprotected brains were embedded in OCT and 60-micron thick free-floating cryostat sections were prepared. Immunohistochemical detection of parvalbumin (PV) was performed on these floating cryosections. All sections were washed twice with PBS and then treated in PBS Triton 0.1% H 2 O 2 overnight at 4 • C. Sections were pre-treated in PBS Triton 0.1% H 2 O 2 overnight at 4 • C, blocked in PBS 1×, 2% gelatine and 0.25% Triton and incubated with 1:1000 mouse anti-PV primary antibody (P3088, Sigma, France) overnight at 4 • C. Sections were incubated for 2 h in peroxidase-coupled goat anti-mouse antibody (1:300 Vector Laboratories, France) and revealed with 3,3 -diaminobenzidine. Sections were collected on Super Frost Ultra Plus slides (Thermo Fisher Scientific, Illkirch-Graffenstaden, France), dehydrated and mounted in Eukitt ®® mounting medium (03989, Sigma, St. Quentin Fallavier, France).

Quantification of PV Neuronal Density
After immunochemistry, brain sections were photographed and specific cortical areas were delimited. PV-neurons were counted manually with the "Fiji" program counting tool and their density was calculated by dividing this number by the surface area for the frontal and somatosensory cortex. Neuronal counting in the hippocampus was normalised with the length of the CA or DG. This operation was repeated on the left and right hemi-cortices and the results were averaged.

Reverse Transcription Quantitative PCR (RT-qPCR)
Cortical fragments were microdissected under a stereomicroscope of 1mm thick sections in cold phosphate buffer and immediately frozen in dry ice. Total RNA was isolated from fragments dissected from control, GABAergic Dlx5/+ , Dlx5/6 VgatCre/+ and Dlx5/6 VgatCre mice using a RNeasy minikit (Qiagen) according to the manufacturer instructions. Oncolumn deoxyribonuclease digestion (Qiagen) was applied after the RNA isolation procedure to remove potential genomic DNA contamination. cDNA was synthesised from 1 µg of RNA (Invitrogen). Real-time PCR was performed using the SYBR Green method according to the manufacturer's instructions (SYBR Green I master, Light cycler 480, Roche Diagnostics). The comparative Ct method on MxPro qPCR software (Agilent Technologies) was used to determine the normalised changes of the target gene relative to a calibrator reference. mRNA quantification samples were normalised to peptidylprolyl isomerase A (PPIA), hypoxanthine phosphoribosyltransferase (HPRT) and phosphoglycerate kinase 1 (PGK1) levels. As a calibrator reference, we referred to Ct from RNAse-free treated water animal samples.
Dlx5 transcripts were analysed using the following primers (Table 1):

Deregulation of Dlx5/6 Expression in Mouse GABAergic Neurons
To generate GABAergic Dlx5/+ mice, in which the expression of Dlx5 is ectopically induced in all GABAergic neurons, we crossed ROSA CAG-flox-Dlx5/+ mice, which express Dlx5 in a Cre-dependent manner [35], with Vgat cre/+ mice in which an IRES-Cre recombinase cassette is inserted downstream of the stop codon of the endogenous Vgat (vesicular GABA transporter) gene ( Figure 1A). In Vgat-cre mice, Cre-recombinase expression is observed in all GABAergic neurons but not in other cell types [41]. qRT-PCR analysis showed that the level of Dlx5 expression in the adult parietal cortex of GABAergic Dlx5/+ mice was almost doubled when compared with control littermates ( Figure 1B). In situ hybridisation of serial brain sections demonstrated an increased expression of Dlx5 in telencephalic regions known to present an endogenous level of expression in the adult (see also [15], Figure 1), such as the olfactory bulb ( Figure 1C,C'), the parietal cortex ( Figure 1D,D') and the hypothalamus ( Figure 1E,E'). Remarkably, in GABAergic Dlx5/+ mice, Dlx5 expression was also observed in non-telencephalic GABAergic neurons where the gene is not normally expressed during development and in the adult such as, for example, Purkinje cells of the cerebellum ( Figure 1F,F'). GABAergic Dlx5/+ mice did not present any obvious anatomical, metabolic or motility problems.
Dlx5 and Dlx6 were both inactivated in GABAergic interneurons since these two closely associated genes present similar regulations and redundant functions [16,17]. To this end, Dlx5/6 flox/flox mice, in which the homeodomain-encoding regions of both Dlx5 and Dlx6 are flanked by non-compatible lox sequences [42] were crossed with Vgat cre/+ mice as previously described [15]. RT-PCR analysis has shown that Dlx5 exon II and Dlx6 transcripts are strongly reduced in the cortex of heterozygous Dlx5/6 VgatCre/+ mice and virtually absent in homozygous Dlx5/6 VgatCre mice [15].

Effects of Dlx5/6 Deregulation in GABAergic Neurons on Adult PV-Interneurons Density
Dlx5/6 are known to play a role in the development of PV-interneurons, however their effects on this population of cells in the adult are still partially understood.
Serial sections of control, GABAergic Dlx5/+ , Dlx5/6 VgatCre/+ and Dlx5/6 VgatCre adult mouse brains were immuno-stained with anti-PV antibodies and PV-positive cells were counted in defined brain regions. No significant difference in brain morphology of the cortical layer thickness was observed. Analysis of a primary cortical region and the parietal cortex, did not show any significant variation in PV-interneurons density even when each individual cortical layer was counted. As we had previously described an inhibition of threat response and reduction in anxiety-like and obsessive-compulsive activities in Dlx5/6 VgatCre adult mice, we focused on the prelimbic cortex and on the hippocampus, two brain regions known to interact in the process of fear inhibition [43].

Behavioural Consequences of Dlx5/6 Expression Deregulation in GABAergic Neurons
We analysed the effects of modulating Dlx5/6 expression in GABAergic interneurons in three experimental settings related to anxiety-like and obsessive-compulsive behaviours.

− Open Field Test with object exploration
Novel stimuli, such as unfamiliar environments or objects, are known to create conflict in rodents, concomitantly evoking exploratory and avoidance behaviours [44,45]. The latter is often interpreted as "anxiety-like" behaviours and reflect the animal's fear of novelty.
In a previous paper, we have analysed the response of control, Dlx5/6 VgatCre and Dlx5/6 VgatCre/+ mice placed in a 72 × 72 cm square flat arena for 10 min (Open Field Test, OFT), showing that the reduction in Dlx5/6 expression in GABAergic neurons is associated with a significant decrease in anxiety-like behaviours as both homozygous and heterozygous mutant mice entered the arena more promptly and spent more time in the centre than control animals (on average 50 s vs. 100 s latency to enter the centre) ( [15] Figure 4). In this study, after a 10 min period of familiarisation, the mice were reintroduced in the same arena, but in the presence of a novel object placed in the centre [46]. This experimental paradigm stimulates a higher approach and exploratory behaviours compared to the OFT. By comparing different responses of mice exposed to the OFT and to the "OFT with object exploration" it is possible, in principle, to discern exploratory from anxiety-like behaviours [44]. Indeed, in the "OFT with object exploration" it took about 50 sec for control mice to reach the object in the centre of the arena ( Figure 5A) whereas a similar control group took an average of 100 s to enter the centre in the absence of an object [15]. Remarkably this enhanced exploratory behaviour, induced by the object, was exacerbated in Dlx5/6 VgatCre mice, and on average, entered the central area in less than 10 s ( Figure 5A) compared to the about 50 s in the absence of the object [15]. On the contrary, GABAergic Dlx5/+ mice hesitated on average more than two minutes before entering the centre ( Figure 5A) and did not display any different behaviour in the presence or in the absence of the object, suggesting a high level of anxiety which prevailed over exploratory inputs. However, after entering the centre the first time, within 10 min of the test, the number of entries and the time spent in the centre were not significantly different between genotypes ( Figure 5B,C). Overexpression of Dlx5 was associated with a delay in first contact with the object, but with a prolonged time of examination whereas Dlx5/6 VgatCre mice went almost immediately to sniff the object but rapidly lost interest ( Figure 5D-F).

− Marble burying test (MBT)
The consequences of Dlx5/6 inactivation in GABAergic neurons on stereotyped repetitive behaviour were assessed through the Marble Burying Test (MBT) ( Figure 6A).
Similarly to our previous report, during the 10 min test, both Dlx5/6 VgatCre/+ and Dlx5/6 VgatCre animals buried a significantly lower number of marbles than control littermates; remarkably, 44% (8/18) of the Dlx5/6 VgatCre animals displaced less than one marble or no marbles at all (5/18). On the contrary, overexpression of Dlx5 in GABAergic Dlx5/+ animals resulted in a significant increase in the number of buried or displaced marbles. It should be noted that also in the MBT, the number of buried marbles was linearly correlated to the number of Dlx5 alleles still present in GABAergic neurons (Figure 4).

− Nest building test
Nest building is an important natural behaviour occurring without the intervention of the experimenter. Whereas Dlx5/6 VgatCre/+ and Dlx5/6 VgatCre animals tended to make incomplete and poorly structured nests, even after seven days, overexpression of Dlx5 in GABAergic Dlx5/+ animals only marginally affected this specific behaviour. Essentially, GABAergic Dlx5/+ animals had a small delay in starting nest construction, but then generated relatively normal nests ( Figure 6B). By the end of the test, none of the Dlx5/6 VgatCre animals had built a high-quality nest, whereas all control and GABAergic Dlx5/+ mice had completed nest construction. -

Marble burying test (MBT)
The consequences of Dlx5/6 inactivation in GABAergic neurons on stereotyped repetitive behaviour were assessed through the Marble Burying Test (MBT) ( Figure 6A).
Similarly to our previous report, during the 10 min test, both Dlx5/6 VgatCre/+ and Dlx5/6 VgatCre animals buried a significantly lower number of marbles than control littermates; remarkably, 44% (8/18) of the Dlx5/6 VgatCre animals displaced less than one marble or no marbles at all (5/18). On the contrary, overexpression of Dlx5 in GABAergic Dlx5/+ animals resulted in a significant increase in the number of buried or displaced marbles. It should be noted that also in the MBT, the number of buried marbles was linearly correlated to the number of Dlx5 alleles still present in GABAergic neurons (Figure 4). -

Nest building test
Nest building is an important natural behaviour occurring without the intervention of the experimenter. Whereas Dlx5/6 VgatCre/+ and Dlx5/6 VgatCre animals tended to make incomplete and poorly structured nests, even after seven days, overexpression of Dlx5 in GA-

Discussion
Altered function of prefrontal cortical neuronal networks is associated with several psychiatric conditions, including autism and schizophrenia. It has been proposed that changes in the relative activity of excitatory and inhibitory neurons (E/I ratio), which regulates network tuning, can be at the origin of the symptoms observed in these conditions [8,47,48]. PV-positive GABAergic inhibitory interneurons constitute a relatively small part of the global neuronal population, but play a central role in determining the E/I ratio [49]. Several types of cortical PV interneurons can be identified on the basis of their morphological or molecular characteristics [4,5]. PV expressing interneurons are classified into basket, axo-axonic (chandelier), and bistratified cells [50]. These cells are involved in regulating local circuit function and rhythmogenesis and modulate information processing. PV interneurons, are involved in a variety of functions, including local circuit operations, learning and memory, sensory processing, and critical period plasticity. Dysfunctions in PV-interneurons are clearly implicated in autism and schizophrenia [48,51]. Post-mortem studies on patient brains have revealed an important reduction in the density of inhibitory PV-positive neurons both in autism and schizophrenia [52][53][54]. Remarkably, several mouse models of both autism and schizophrenia present a reduction in cortical PV neuronal density reminiscent of what has previously been observed in psychiatric patients (see, for example: ( [55][56][57][58][59]). In these models, the reduction in PV neuronal density is due to reduce Parvalbumin expression and not due to neuronal cell death [55].
Dlx5 and Dlx6 homeobox genes are expressed by developing and mature GABAergic cortical interneurons. Transplantation of immature neurons lacking Dlx5 or Dlx5/6 into normal brain has shown a specific reduction in their differentiation in mature PVinterneurons [19]. In the same study, no difference in the density of PV-interneurons was detected in the somatosensory cortex of mice heterozygous for a systemic Dlx5/6 deletion (Dlx5/6 +/− ); this finding is in line with our present observation that PV-interneurons density is normal in the SSC of Dlx5/6 VgatCre/+ mice.
The importance of Dlx5/6 expression in the prefrontal cortex (PFC) has been shown in Dlx5/6 +/− mice, [20] in which the activity of fast-spiking PV interneurons, that generate gamma oscillations, becomes abnormal after adolescence in parallel with the onset of cognitive inflexibility. Although these findings suggest that a reduction in Dlx5/6 expression could represent a good functional model of schizophrenia [20], the density of PV-interneurons in the prefrontal cortex of Dlx5/6 +/− mice was not measured.
Together these results suggest that Dlx5/6 expression levels in the PFC could be associated with autistic and/or schizophrenic phenotypes through regulations affecting the PV interneuronal population. In this study, we have targeted the induction of Dlx5 or deletion of Dlx5/6 expression specifically to GABAergic neurons to explore the role of these genes on PFC PV neuronal density and behaviour.
Our results reveal a linear correlation between Dlx5/6 level of expression and PV neuronal density in the prelimbic cortex and in the hippocampus (Figure 4), but not in the somatosensory cortex. After homozygous deletion of Dlx5/6 in GABAergic neurons in Dlx5/6 VgatCre mice, the density of PV positive neurons in the prelimbic cortex is less than half of that observed in control mice (Figure 2), a proportion similar to what had been observed grafting Dlx5/6-null immature neuroblasts [19].
The contrasting results obtained in the frontal and in the somatosensory cortex could indicate that Dlx5/6 expression is one among other regulators of PV expression, the weight of these factors being different in a primary sensory region such as the somatosensory cortex and in an associative region such as the frontal cortex. Another explanation could be that at least two populations of PV interneurons exist, one sensitive and the other insensitive to Dlx5/6 expression. In this case, the difference in PV neuron density between regions would mirror differential proportions of these PV neurons subclasses.
In the PFC, Parvalbumin levels of expression are variable in adults, and are regulated by several factors, including neuronal activity, serotonin levels and ageing [60], expression variations can even be recorded within a day [61].
In our experiments, the comparison of PV neuron density between mice with Dlx5 overexpression and Dlx5/6 invalidation could be indicative of the plasticity of PV neuronal proportion that can be mobilised, or decreased, depending on physiological and environmental conditions. Together, our results suggest that genetic variations in the DNA regions governing Dlx5/6 brain expression could contribute to individual differences in cognition [62] and, in certain cases constitute genetic risk factors for psychiatric diseases. The involvement of the DLX5/6 locus in human psychiatric conditions has been recently shown, integrating genetic and genomic data in a transcriptome-wide association study that provides a comprehensive resource for mechanistic insight and therapeutic development [24]. Molecular pathways or bioactive compounds capable of modulating Dlx5/6 expression could therefore act on the mind and, reciprocally, existing psychotropic drugs could act on Dlx5/6 regulations.