Serotonin receptor 3A controls interneuron migration into the neocortex

Neuronal excitability has been shown to control the migration and cortical integration of reelin-expressing cortical interneurons (INs) arising from the caudal ganglionic eminence (CGE), supporting the possibility that neurotransmitters could regulate this process. Here we show that the ionotropic serotonin receptor 3A (5-HT3AR) is specifically expressed in CGE-derived migrating interneurons and upregulated while they invade the developing cortex. Functional investigations using calcium imaging, electrophysiological recordings and migration assays indicate that CGE-derived INs increase their response to 5-HT3AR activation during the late phase of cortical plate invasion. Using genetic loss-of-function approaches and in vivo grafts, we further demonstrate that the 5-HT3AR is cell autonomously required for the migration and proper positioning of reelin-expressing CGE-derived INs in the neocortex. Our findings reveal a requirement for a serotonin receptor in controlling the migration and laminar positioning of a specific subtype of cortical IN.

C ortical interneurons (INs) are key cellular components involved in the assembly and function of cortical circuits 1 . Alterations in their development and functional integration are thought to play a critical role in the emergence of psychiatric disorders 2 . In mice, cortical INs mainly originate (about 70%) from the medial ganglionic eminence (MGE), which generates parvalbumin (PV)-expressing basket cells and chandelier cells or somatostatin (SST)-expressing Martinotti cells [3][4][5] . In addition to the MGE, the caudal ganglionic eminence (CGE) [6][7][8][9][10][11][12] and to a smaller extent the preoptic area 13,14 contribute to the remaining 30% of cortical INs. CGEderived interneurons (cINs) are comprised of diverse IN subtypes divided into two main classes: bipolar or double-bouquet INs that express the vaosintestinal peptide (VIP) and neurogliaform or multipolar INs that express reelin and/or neuropeptide Y (NPY) [15][16][17] .
Recent in vivo data have demonstrated that neuronal excitability controls the laminar positioning of reelin-expressing CGE-derived INs 18 . These data support the possibility that cellextrinsic factors such as neurotransmitters could regulate early activity of migrating cINs and control their final laminar positioning. Interestingly, cINs express the serotonin receptor 3A (5-HT 3A R) during the process of neuronal migration 15,16 . The 5-HT 3A R is the only known ionotropic serotonergic receptor 19,20 and is expressed exclusively in inhibitory GABAergic INs in the adult cortex 17,21 . Serotonin is detected as early as E10.5 in the embryonic forebrain 22 and has been shown to regulate a variety of developmental processes involved in the construction of cortical circuits 23 , including neuronal migration 24,25 and thalamo-cortical pathfinding 26 . In addition, early-life serotonin dysregulation plays a key role in vulnerability to psychiatric disorders 27,28 . Here we determined whether 5-HT 3A R-dependent signalling controls the migration of cINs.
The migration of INs is a multistep process regulated by the combinatorial expression of transcription factors and by a variety of cell-extrinsic factors including the extracellular matrix and guidance cues [3][4][5] . To populate the neocortex, INs exit the subpallium and migrate tangentially through migratory streams mainly located in the marginal zone (MZ), subplate and the subventricular zone (SVZ) [2][3][4]29 . Following this initial phase of tangential migration, INs migrate radially and progressively invade the cortical plate (CP) [2][3][4]29 .
To investigate the molecular mechanisms regulating different steps in the migration of cINs, we performed a microarray gene expression analysis on INs preferentially derived from the CGE at three distinct developmental stages of the migratory process. Using this approach, we find that the Htr3a is upregulated in INs during the process of CP invasion. Calcium imaging and electrophysiological recordings indicate that the 5-HT 3A R is functional in cINs during CP invasion and that 5-HT 3A R activation increases their migratory speed during this late phase of migration. Genetic loss-of-function experiments combined to time-lapse imaging and in vivo grafts reveal that the 5-HT 3A R regulates the entry of cINs into the CP in a cell-autonomous manner. Finally, we find that genetic deletion of the 5-HT 3A R led to the persistent laminar mispositioning of reelin-expressing CGE-derived INs.

Results
Migrating cINs specifically express functional 5-HT 3A R. To study the migration of CGE-derived INs, we used GAD65-green fluorescent protein (GFP) mice that preferentially label CGEderived INs and rarely MGE-derived INs expressing PV and SST (o5% at P21) 30,31 (Supplementary Fig. 1). To identify timespecific genes regulating the phase of cortical invasion, microarray-based gene expression analysis of GAD65-GFP þ INs was performed at three different developmental time points corresponding to distinct steps of the migratory process (Fig. 1a). GAD65-GFP þ cells in the process of tangential migration were obtained by microdissection of the E14.5 caudal pallium containing the SVZ migratory stream. GAD65-GFP þ cells invading the cortex were obtained by microdissection of the developing cortex at E18.5 (excluding the SVZ), whereas INs terminating migration were obtained by microdissection of the cortex at P2.5 (excluding the SVZ). GAD65-GFP þ INs were then isolated using fluorescent-assisted cell sorting (FACS; Supplementary Fig. 2a,b), total RNA was extracted and microarrays were performed. Data were analysed in order to identify genes specifically upregulated during cortical invasion, that is, genes that significantly increased their expression from E14.5 to E18.5 (Po0.01 and 42-fold change) and significantly decreased their expression from E18.5 to P2.5 (Po0.01 and 42-fold change). Using this unbiased approach, we found that Htr3a is the third top candidate gene, displaying a 4.9-fold upregulation from E14.5 to E18.5 (Fig. 1b Supplementary Fig. 2c,d). In line with these results, we observed that at E14.5, Htr3a-GFP þ cells were mainly found populating the CGE and not the MGE (Fig. 1d, e, Supplementary Fig. 2f). In the CGE, the majority of Htr3a-GFP þ cells expressed COUP-TFII and/or SP8 (Fig. 1d), two transcription factors preferentially expressed in this region during the embryonic period 33,34 and only very rarely NKX2.1 ( Supplementary Fig. 2e). Htr3a-GFP þ INs were observed at E14.5 exiting the subpallium by forming a tangential stream of migratory INs located in the pallial SVZ and invading the CP by E17.5 (Fig. 1e). In situ hybridization targeting the Htr3a mRNA confirmed the labelling pattern observed in Htr3a-GFP þ brains at E17.5 (Fig. 1f) and revealed strong expression in the CGE but not MGE at E14.5 ( Supplementary  Fig. 2f).
To determine whether the 5-HT 3A R is functional in migrating cINs, we isolated the CGE from E14.5 Htr3a-GFP slices and platted cINs in vitro in the presence of dissociated and isochronic cortical cells in order to allow Htr3a-GFP þ INs to migrate on a homogeneous substrate 35 (Supplementary Fig. 3a). Calcium imaging revealed that application of the 5-HT 3A R agonists mchlorophenylbiguanide (mCPBG) or SR57227 could repeatedly induce calcium events in migrating Htr3a-GFP þ INs and this could be observed in the presence of the sodium channel blocker tetrodotoxin (TTX), the NMDA (N-methyl-D-aspartate) antagonist D(-)-2-amino-5-phosphonovaleric acid (D-AP5), the AMPA/kainate antagonist 2,3-Dioxo-6-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide (NBQX) and the GABA-A antagonist gabazine ( Supplementary Fig. 3b-e). Furthermore, a significant time-dependent increase in 5-HT 3A R-induced calcium transients was observed in migrating cINs from E14.5 day in vitro 1 (þ DIV1) to E14.5 ( þ DIV3) and this was linked to a significant increase in protein expression of the 5-HT 3A R ( Supplementary Fig. 3f,g). To demonstrate that 5-HT 3A R on cINs are directly activated in acute cortical slices, electrophysiological recordings were performed in Htr3a-GFP þ INs. Recordings revealed that 5-HT 3A R activation directly triggers inward currents in migrating cINs in the CP of cortical slices and this was observed in the presence of TTX, D(-)-2-amino-5phosphonovaleric acid, NBQX and gabazine (Supplementary  Table 2). Interestingly, 5-HT 3A Rmediated inward currents were detected in Htr3a-GFP þ INs migrating in the E18-E19 CP but not in cINs migrating tangentially in the SVZ at E18-E19 or IZ/SVZ at E14-E15 (Supplementary Fig. 4 and Supplementary Table 2), indicating that Htr3a-GFP þ INs migrating in the CP become functionally responsive to 5-HT 3A R activation. At P2.5, 5-HT 3A R activation reliably induced currents in Htr3a-GFP þ cells located in the CP, suggesting that although Htr3a mRNA is downregulated in cINs from E18.5 to P2.5 targeting of the 5-HT 3A R to the cell membrane may increase through post-translational modifications 20 . Taken together, these data confirm previous findings 15,16 , indicating that functional 5-HT 3A R is specifically expressed in CGE-derived but not MGE-derived INs. Furthermore, functional upregulation of 5-HT 3A R in cINs during the process of cortical invasion suggests that 5-HT 3A R could regulate this migratory phase. SVZ/IZ stream or invading the E17.5 cortex were isolated through microdissection and time-lapse imaging of INs was performed in cortical cultures at E14.5 ( þ DIV1) and E17.5 ( þ DIV1) in a serotonin-free medium 26 (Fig. 2a,b). This in vitro strategy has successfully been used to unmask the effects of GABA-A stimulation on cortical IN migration 35 . Furthermore, it has the advantage to allow the recordings of INs in a serotoninfree environment, thus removing the potential confounding effects of endogenous levels of serotonin contained in acute slices. Indeed, it has been shown that it is not possible in cortical slices to unmask the role of 5-HT 3A R on the migration of cINs 25 . Using this in vitro strategy, time-lapse recordings revealed that the 5-HT 3A R agonist SR57227 but not vehicle exposure significantly increased the mean migratory speed and significantly decreased the pausing time of GAD65-GFP þ INs at E17.5 ( þ DIV1; Fig. 2c, Supplementary Fig. 5a, Supplementary Movie 1). In contrast, 5-HT 3A R-stimulation did not affect the migratory speed or pausing time of cINs at E14.5 ( þ DIV1; Fig. 2d, Supplementary Fig. 5b, Supplementary Movie 2). Given that it has been reported that high concentrations of serotonin (200-400 mM) inhibit cIN migration possibly through a 5-HT 6 R-mediated mechanism 25 , we aimed to determine whether more physiological concentrations of serotonin (nanomolar range) could have a pro-migratory effect and mimic the effects of 5-HT 3A R-stimulation. Indeed, exposure to serotonin (100 nM) significantly increased the mean migratory speed and significantly decreased the pausing time of GAD65-GFP þ INs at E17.5 ( þ DIV1) but not at E14.5 ( þ DIV1; Supplementary Fig. 5c-f). Taken together, these functional investigations indicate that 5-HT 3A R activation increases the migratory speed of CGE-derived INs during the phase of CP invasion.
Growth cones located at the tip of the leading processes of migrating INs play a key role in sensing the local environment.
To determine whether 5-HT 3A R growth cone responsiveness was specific to cINs migrating in the CP, we performed simultaneous time-lapse imaging of the growth cones of MGE-derived INs (mINs) and cINs as they invade the CP (Fig. 3a). MGE-derived INs were labelled by focal electroporation of a Tomato-expressing plasmid in the MGE of Htr3a-GFP þ cortical slices at E14.5 and growth cone imaging of cINs and mINs was simultaneously performed in the CP at E17.5 ( þ DIV3) in serotonin-free medium ( Fig. 3a-c). Using this paradigm, we found that the size and growth cone extension speed of cINs were specifically increased after mCPBG exposure, whereas no effect of the drug was observed in mINs ( Fig. 3d,e, Supplementary Movies 3 and 4). To confirm the specificity of the 5-HT 3A R agonist, similar experiments were performed on the growth cones of Htr3a-ko; GAD65-GFP þ INs. Quantification revealed that 5-HT 3A R activation failed to increase the GC speed and size of Htr3a-ko; GAD65-GFP þ INs ( Supplementary Fig. 5g,h). Taken together, these data indicate that the 5-HT 3A R specifically regulates the growth cone dynamics of cINs but not mINs during CP invasion.  Fig. 5i). Isochronic graft experiments indicated that 5-HT 3A R regulates cortical invasion in a cell-autonomous manner. Indeed, at E19, the cortical distribution of E14.5 CGE-isolated Htr3a-ko; GAD65-GFP þ INs grafted into the CGE of wild-type E14.5 embryos was significantly altered in comparison with grafted GAD65-GFP þ INs (Fig. 4d). Taken together, these results indicate that the 5-HT 3A R controls the migration and positioning of cINs into the developing cortex.

5-HT
5-HT 3A R is required for the laminar positioning of cINs.
To determine whether 5-HT 3A R deletion leads to persistent alterations in the distribution of cINs, the laminar positioning of CGE-derived INs was analysed in the somatosensory cortex of Htr3a-ko mice at P21. Quantification revealed a significant decrease in the percentage of Htr3a-ko; GAD65-GFP þ INs at P21 in superficial layers of the somatosensory cortex compared with GAD65-GFP þ INs (Fig. 5a). To determine whether the positioning of a specific subtype of CGE-derived INs was altered in Htr3a-ko mice, GAD65-GFP þ sections were stained for reelin, VIP and NPY, which label different CGE-derived IN populations. Quantification revealed that 5-HT 3A R deletion leads to the persistent laminar mispositioning of CGE-derived reelin þ but not VIP þ or NPY þ /Htr3a-ko; GAD65-GFP þ INs in superficial cortical layers as compared with GAD65-GFP þ INs (Fig. 5b-d). Taken together, these data indicate that the 5-HT 3A R regulates the laminar positioning of a specific subclass of reelin-expressing CGE-derived INs.

Discussion
Here we report that the 5-HT 3A R, a specific marker of CGEderived cortical INs, is functionally upregulated in cINs during the process of cortical invasion and that 5-HT 3A R activation stimulates the migration of cINs during the phase of cortical invasion. Furthermore, 5-HT 3A R-mediated activation specifically regulated the growth cones dynamic of cINs but not of mINs. Time-lapse imaging, graft experiments and in vivo quantification in Htr3a-ko mice demonstrated that the 5-HT 3A R is cell autonomously required for positioning of cINs into the nascent cortex. Finally, alterations in the early process of CP invasion lead to a persistent mispositioning of the reelin-expressing cINs subtype in superficial cortical layers of Htr3a-ko mice.
The migration of cINs into the cortex is a multi-step process. After exiting the subpallium, INs disperse into the pallium through specific tangential migratory streams mainly located in the MZ, subplate and SVZ [3][4][5]29 . The selection of migratory streams by mINs requires responsiveness to the chemokine Cxcl12 through the G-protein-coupled receptors Cxcr4 and Cxcr7 (refs [36][37][38][39]. It has been hypothesized that the loss of responsiveness to Cxcl12 could allow mINs to become responsive to a yet unknown chemoattractant located in the CP. The chemoattractant properties of the CP have not been elucidated and it is unknown whether cINs are responsive to the same set of guidance cues as mINs. The fact that the 5-HT 3A R is required for proper CP invasion and that serotonin levels increase during the late phase of cortical invasion 22 suggests that serotonin signalling through the 5-HT 3A R could specifically regulate the responsiveness of cINs to guidance cues located in the developing cortex. Serotonin has been reported to regulate the migration of different non-neuronal cell types including eosinophils, pulmonary artery smooth muscle cells, aortic endothelial cells and cranial neural crest cells [40][41][42][43][44] . Depending on the concentration of serotonin, cell types, experimental paradigms and embryonic age of the cells, serotonin was shown to either stimulate or inhibit migration. For example, differential effects of serotonin on the migration of neural crest cells were observed depending on the concentration of serotonin and embryonic age of the cells 41 . Here we report that serotonin levels at the nanomolar range increase the migration speed of cINs in contrast to high levels of serotonin (micromolar range), which have previously been shown to inhibit migration possibly through at 5-HT 6 R mechanism 25 . A requirement for serotonin in regulating neuronal embryonic migration in vivo was first described in the nematode Caenorhabditis elegans 45 . In mice, time-lapse imaging on cortical slices revealed that an excess of serotonin decreased the migratory speed of INs and pyramidal neurons in a reversible manner 24,25 . Interestingly alterations in the positioning of cINs in the developing cortex in vivo were reported in a pharmacological model of serotonin depletion 46 as well as in a genetic model of serotonin excess 25 . During the phase of cortical invasion, serotoninergic raphe fibers located in the intermediate zone and in the MZ 47 are the main source of serotonin to the developing cortex 26 . Our current model proposes that following the switch from tangential to radial migration cINs increase their responsiveness to serotonin signalling through 5-HT 3A R upregulation in order to respond to guidances cues located in the developing cortex. In a related developmental process, that is, embryonic thalamo-cortical wiring, a role for serotonin in modulating growth cone responsiveness to the guidance cue netrin1 has been described 26 . In migrating cINs, we find that the effect of 5-HT 3A R activation on growth cones dynamics is delayed, suggesting that 5-HT 3A R activation may have indirect effects on growth cone dynamics by modulating responsiveness to guidance cues. Although the guidance cues regulating the migration of cINs into the CP remains to be discovered, the identification of genes specifically upregulated in cINs during CP invasion provides a set of candidate genes that may specifically regulate this process. Among candidate genes from our microarray, PlexinA4 displayed a time-specific twofold upregulation during CP invasion, suggesting that this receptor could play a role in regulating CP invasion of cINs. Interestingly, PlexinA4/Semaphorin6A signalling was shown to regulate the migration of oligodendrocyte precursor cells in the developing cortex 48 . Further studies are required to establish whether PlexinA4 expression is regulated by 5-HT 3A R signalling and regulates the migration of cINs.
The 5-HT 3A R is a cation-selective ligand-gated ion channel that mediates neuronal depolarization and excitation 20 . Previous work in the field has demonstrated that neuronal excitability is required for the proper positioning of CGE-derived INs in superficial cortical layers 18 , supporting the possibility that neurotransmitters such as serotonin could regulate the migration of CGE-derived INs through 5-HT 3A R-dependent signalling. Here we show that 5-HT 3A R activation enhances the migratory speed and growth cone dynamics of cINs during the phase of CP invasion. The 5-HT 3A R is closely related to the anion selective GABA A receptor. In conditions of low-expression levels of the potassium-chloride co-transporter KCC2, GABA A stimulation induces an increase in calcium transients and promotes the migration of cINs and their entry into the cortex 35,49 . In addition, other neurotransmitter systems have been shown to regulate the migration of cortical INs such as AMPA and NMDA receptors [50][51][52] . Overall, in vitro data obtained in cINs but also in other neuronal cell types such as cerebellar granule cells indicate that neuronal migration involves the modulation of calcium transients through the opening of NMDA and voltage-gated calcium channels 35,[53][54][55] . It is thus hypothesized that early activity in neuronal networks generates ambient levels of neurotransmitters that regulate IN migration by modulating their intracellular calcium transients. Our data add to this model the possibility that ambient levels of serotonin could specifically regulate the migration of cINs into the CP through the regulation of 5-HT 3A R-dependent calcium transients.
Developmental insults including alterations in the migration and maturation of cINs contribute to the emergence of psychiatricrelevant functional alterations later in life 2 . Early-life serotonin dysregulation in rodents following genetic deletion or pharmacological blockade of the serotonin transporter by antidepressants leads to long-term psychiatric-related phenotypes 27 . In this perspective, the 5HT 3A R represents a novel developmental target that regulates several cellular events involved in neural circuit formation. Here we reveal that the 5-HT 3A R is required to control the migration and positioning of cINs into the CP. In addition to its importance in regulating cIN subtypes, the 5HT 3A R is expressed in Cajal-Retzius cells and regulates the apical dendritic growth of pyramidal neurons in the cortex through a reelin-dependent mechanism 56 . Taken together, our study provides further insights on the importance of the 5-HT 3A R in controlling developmental processes involved in the formation of cortical circuits. It proposes that 5-HT 3A R-mediated signalling constitutes a molecular mechanism controlling the migration and laminar positioning of a reelin-expressing CGE-derived IN subtype and thus further supports the important role of early activity in the development of reelin-expressing CGE-derived INs 18 .

Methods
Mice. Animal experiments were conducted according to the Swiss and international guidelines and approved by the local Geneva animal care committee. Adult timed pregnant mice were obtained by overnight (ON) mating and the following morning was counted as embryonic day (E) E0.5. To study CGE-derived INs, we used transgenic mice expressing GFP under the control of the GAD65 regulatory sequences (GAD65-GFP) 25,30 and transgenic mice expressing the enhanced GFP under the control of the Htr3a regulatory sequences (Htr3a-GFP) provided by the GENSAT Consortium 16 . Both strains were maintained on a C57Bl/6 background. C57Bl/6 wild-type mice were used for in vivo graft experiments. The Nkx2.1-Cre mice have been previously described 32 and the R26R-tdTOM fl/fl reporter mice were obtained from Jackson Laboratory. Htr3a-GFP mice were crossed to R26R-tdTOM fl/fl mice to obtain Htr3a-GFP; R26R-tdTOM fl/fl mice. Htr3a-ko mice have been previously described 57 and crossed to GAD65-GFP mice to obtain Htr3a-ko; GAD65-GFP animals.
In situ hybridization. Sections were hybridized as described previously 25 with the respective DIG-labelled RNA probes. The Htr3a plasmid probe 16 was linearized with HindIII-HF for RNA probe synthesis by T7 polymerase (kind gift from Dr B. Emerit). The Lhx6 plasmid probe 58 (kind gift from Dr M. Denaxa) was linearized with Not1 for antisense RNA probe synthesis by T3 polymerase. The unbound probe was washed and slices were incubated with alkaline phosphataseconjugated anti-DIG antibody (1:2,000; Roche) ON at 4°C. NBT/BCIP (Roche) was then used as an alkaline phosphatase substrate to reveal the hybridized probe. Fast Red (Roche) was used as an alkaline phosphatase fluorescent substrate.
FACS and microarrays. Acute cortical slices were prepared as described above and under a fluorescence dissecting scope (Leica, M165 FC), the E14.5 pallium including the SVZ stream and the E18.5 and P2.5 developing cortices excluding the SVZ stream were isolated from GAD65-GFP þ caudal coronal slices. Each time point was obtained in triplicate from three different litters and consisted of pooled embryos or pups (n ¼ 4-5). Following dissociation of the tissue, GAD65-GFP þ cells were isolated using FACS with the FACS VantageSE and collected in RNA Later solution (Invitrogen). RNA from the cells was extracted using the RNeasy Mini kit (Qiagen) and the quality was checked on an Agilent 2100 Bioanalyser (Agilent Technologies). A small-scale protocol from Affymetrix (High Wycombe) was used to reproducibly amplify and label total RNA. Briefly, approximately 100 ng total RNA was converted into double-stranded cDNA using a cDNA synthesis kit (Superscript; Invitrogen) with a special oligo(dT)_24 primer containing a T7 RNA promoter site added to the 5 0 poly(T) tract. After the first cRNA amplification by in vitro transcription using the Ambion MEGAscript T7 kit (Ambion), 400 ng cRNA was once more reverse transcribed, and biotinylated cRNAs were generated from double-strand cDNAs using an in vitro transcription labelling kit from Affymetrix. For each probe, 20 mg of the second amplification biotinylated cRNA was fragmented and hybridized to Mouse Genome 430 2.0 Array (Affymetrix) following the standard protocols. GeneChips were incubated at 45°C for 16 h with biotin-labelled cRNAs probes, and washed and stained using a streptavidin-phycoerythrin conjugate with antibody amplification using Affymetrix GeneChip Fluidis Station 450. GeneChips were scanned on a GC3000 scanner (Affymetrix). The signal intensities were analysed using the Partek Genomics suites (version 6.6 beta) and the data were normalized using robust multi-array average (RMA) 59 . The messenger RNA expression levels of genes (ProbeSet) were compared using a two-way analysis of variance model and contrasted with uncorrected P values set at o0.01 threshold. Selection of target genes specifically increasing their expression during CP invasion at E18.5 was based on 42-fold intensity increase from E14.5 versus E18.5 and 42-fold intensity decrease from P2.5 versus E18.5.
ELISA quantification of protein expression. Microdissection of the CGE was performed at E14.5, trypsinized dissociated cells were plated on 96-well plates and cultured in NMB supplemented with mCPBG (10 mM; Tocris) and the mitotic inhibitor arabinofuranosyl cytidine (1 mM). Quantification of protein levels was performed using ELISA kit (Thermo scientific) and 5-HT 3A R antibody (1:1,000, LSBio) on E14.5 day in vitro 1 ( þ DIV1) or E14.5 ( þ DIV4) cells. Absorbance levels were measured on a SpectraMax Paradigm Multimode Microplate detection reader (Molecular Devices). Absorbance levels for the 5HT 3A R were normalized to absorbance levels calculated for the total amount of cells via Janus Green whole-cell staining.
Focal electroporation on slices. For focal electroporation on slices, acute coronal slices of E14.5 Htr3a-GFP þ embryos were transferred on membranes (nucleopore Track-Etch; 1 mm pore size, Whatman) floating on NBM in an incubator (37°C, 5% CO 2 ) for at least 2 h following Vibratome sectioning. A plasmid encoding for a red fluorescent protein (pUBI-tdTOM) was focally injected at the concentration of 1 mg ml À 1 through a beveled glass micropipette into the MGE using a nanoinjector (B203XVZ; WPI) and precision of the injection site was controlled by addition of 10% Fast Green (Sigma) to the plasmid. The slices were kept moisturized in HBSS and electroporated between platinum Petri dish and cover square platinum electrodes (CUY701-P5E and CUY701-P5L; Nepa Gene) with two unipolar square pulses (DP: 100 V; duration: 5 ms; interval: 500 ms) generated by a CUY21-SQ electroporator (Nepa Gene). Slices were then cultured 3 days in vitro ( þ DIV3) in an incubator (37°C and 5% CO 2 ).
Time-lapse imaging and analysis. Time-lapse imaging of migrating GAD65-GFP þ cells in cultures was performed using an automated microscope (ImageXpress Micro XL; Molecular Devices) equipped with objective Â 10 (Plan Fluor Â 10/0.30 Ph1), which allows the simultaneous recording of migrating cells in multiple wells at 37°C under 95% O 2 and 5% CO 2 . Following a control time-lapse imaging period of 360 min, SR57227 hydrochloride (100 nM; Tocris), serotonin (5-HT, 100 nM; Tocris) or vehicle (NBM) was added to the medium and recordings were performed for an additional 360 min. Using Metamorph software (version 7.4; Molecular Devices), cells were tracked at E14.5 ( þ DIV1) and at E17.5 ( þ DIV1). Time-lapse imaging of cortical invasion of cINs was performed on cortical slices at E17.5, placed on porous nitrocellulose (Millicell-CM, Millipore) inserts in Fluorodishes (WPI) filled with NBM and with an inverted confocal microscope (Nikon A1R) equipped for live imaging (Life Technologies) with a long working distance Â 20 objective (CFI Plan Fluor ELWD C Â 20/0.45, Nikon). The microscope incubation chamber temperature was kept at 37°C with a constant flux (25 l h À 1 ) of 5% CO 2 humidified at 96%. 50-mm-thick stacks (3 mm-stepped) were acquired every 10 min during 10-12 h, with resonant laser scanning to reduce toxicity and to avoid bleaching. The first 90 min of movies were removed from analysis to avoid bias in measurements because of settling of the slices in the microscope chamber. For quantification of cortical invasion, the intermediate zone (260-410 mm from the pia) and cortical (50-220 mm from the pia) compartments were manually drawn in Metamorph and GFP þ INs from GAD65-GFP þ and Htr3a-ko; GAD65-GFP þ slices located in the IZ compartment at start were tracked during 8 h. The percentage of GAD65-GFP þ INs entering the cortex and not going back into the IZ or further migrating in the MZ during the 8-h time period was calculated. Migration speed was calculated as the total distance travelled by GAD65-GFP þ INs divided by total imaging time excluding the pausing time. The percentage pausing time was calculated as the fraction of time the cell nucleus was stationary during the imaging period. For simultaneous imaging of growth cones of CGE INs and MGE INs, E14.5 ( þ DIV3) Htr3a-GFP þ slices that were focally electroporated in the MGE (see Focal electroporation on slice) were placed in the bath chamber (RC-26; Warner Instruments) and were immobilized with an anchor. Imaging was done in the CP with a continuous superfusion of oxygenized ACSF (95% O 2 , 5% CO 2 ) under an inverted confocal microscope (Nikon, A1R) equipped for live imaging with a Â 60 oil-immersion objective (Plan Apo VC H Â 60/1.4, Nikon) with a Â 3 numerical zoom. ACSF and microscope incubation chamber temperatures were kept at 37°C. 10-mm-thick stacks (1 mm-stepped) were acquired every 2 min with resonant laser scanning. Dual imaging was done in channel series mode to avoid fluorescence crosstalk. Imaging sessions were systematically started 20-30 min after immersion of the slice in ACSF to avoid axial drift, loss of focus and bias in measurement because of settling of the slice. A standard imaging session was composed of a baseline period of 14 min followed by 14 min drug application (mCPBG, 100 mM; Tocris) and 2 Â 14 min wash. Time-lapse stacks were aligned using Metamorph software (version 7.4). For quantification, contours of the growth cone were manually drawn using Metamorph and their speed of progression as well as their area was measured. The speed was calculated by tracking the base of the growth cone. The leading process was taken as an axis of growth so that all movements towards the growth cone were considered positive and those towards the cell body negative.
In vivo grafts. The CGE of E14.5 GAD65-GFP þ or Htr3a-ko; GAD65-GFP þ mice was isolated, cells were dissociated and resuspended to obtain a cell suspension of approximately 100,000 cells per ml. 10% Fast Green (Sigma) was added to the cell suspension as a tracer. Using a nanoinjector (WPI, B203XVZ), approximately 1 ml of the cell suspension was injected into the CGE of embryos from pregnant E14.5 WT using beveled glass pipettes with a diameter of 70-100 mm. Embryos from host mouse were then collected at E19 for tissue processing and analysis. Only embryos with grafted cells observed in the CGE at E19 were used for the quantification.
Quantification of IN identity and distribution. Images were acquired using an epifluorescence microscope (Nikon, Eclipse 90i) equipped with a Â 10 objective (Plan Apo Â 10/1, Nikon) or a confocal (Nikon, A1R) microscope equipped with dry Â 10 and Â 20 objectives (CFI Plan Apo Â 10/0.45 and CFI Plan Apo VC; Â 20/0.75, Nikon) and oil-immersion Â 40 and Â 60 objectives (CFI Plan Fluor Â 40/1.3 and CFI Plan Apo VC H Â 60/1.4, Nikon). Quantification of the distribution of IN subtypes in the cortex was done by apposing a 10-bin grid at the level of the primary somatosensory cortex and bins corresponding to cortical layers were pooled.
Statistical analysis. No statistics were used to determine optimal group sample size; however, sample sizes were similar to those used in previous publications from our group and others. No samples were excluded from statistical analysis. Statistical analyses (GraphPad Prism software, version 6.0) were performed using paired or unpaired Student's t-test, one-way or two-way analysis of variance with Tukey's or Bonferroni's multiple comparisons test.