Mapping combinatorial expression of non-clustered protocadherins in the developing brain identifies novel PCDH19-mediated cell adhesion properties

Non-clustered protocadherins (ncPcdhs) are adhesive molecules with spatio-temporally regulated overlapping expression in the developing nervous system. Although their unique role in neurogenesis has been widely studied, their combinatorial role in brain physiology and pathology is poorly understood. Using probabilistic cell typing by in situ sequencing, we demonstrate combinatorial inter- and intra-familial expression of ncPcdhs in the developing mouse cortex and hippocampus, at single-cell resolution. We discovered the combinatorial expression of Protocadherin-19 (Pcdh19), a protein involved in PCDH19-clustering epilepsy, with Pcdh1, Pcdh9 or Cadherin 13 (Cdh13) in excitatory neurons. Using aggregation assays, we demonstrate a code-specific adhesion function of PCDH19; mosaic PCDH19 absence in PCDH19+9 and PCDH19 + CDH13, but not in PCDH19+1 codes, alters cell–cell interaction. Interestingly, we found that PCDH19 as a dominant protein in two heterophilic adhesion codes could promote trans-interaction between them. In addition, we discovered increased CDH13-mediated cell adhesion in the presence of PCDH19, suggesting a potential role of PCDH19 as an adhesion mediator of CDH13. Finally, we demonstrated novel cis-interactions between PCDH19 and PCDH1, PCDH9 and CDH13. These observations suggest that there is a unique combinatorial code with a cell- and region-specific characteristic where a single molecule defines the heterophilic cell–cell adhesion properties of each code.


Introduction
Brain development and neuronal network formation require strict coordination of neuronal migration and interaction.These processes are largely controlled by adhesion molecules, including the protocadherin (Pcdh) family [1][2][3].Pcdhs are calcium-dependent cell adhesion molecules that are predominantly expressed during nervous system development and early post-natal stages [3].Although PCDH functions in axon outgrowth, dendritogenesis, synaptogenesis and neuronal migration have been widely studied [4][5][6][7][8] and the effect of mutations in individual Pcdh genes linked to neurological phenotypes [1,4,[9][10][11][12][13][14], it remains unclear how the absence of a single molecule, albeit in a cellular mosaic state, can alter cell-cell communication and brain function.Given the spatio-temporal overlapping expression of non-clustered Protocadherins (ncPCDH) family members in vivo [15], the combinatorial expression of these adhesion molecules, at a single-cell level, could play a critical role in the formation and maintenance of brain structural integrity [16].On the other hand, missing a key molecule of a specific combinatory code could disrupt vital neuronal processes leading to a pathological phenotype [10,17].
Genomic variants in the X-linked PCDH19 gene cause PCDH19-clustering epilepsy (PCDH19-CE), one of the most common forms of inherited epilepsy, which is characterized by variable seizures and incompletely penetrant intellectual disability [25][26][27].PCDH19-CE has a unique mode of inheritance whereby heterozygous females and somatic mosaic males are affected while hemizygous males are clinically unremarkable [28].The proposed mechanism underpinning this synular interference where the coexistence of two cell populations, expressing either WT or variant PCDH19, results in abnormal cellular interactions and neuronal communication in affected individuals [9,29].Greater than 80% of PCDH19 disease variants are located in the region encoding for the EC domains, which are essential for homophilic interaction and cell-cell adhesion [30,31].
The aim of this study was to identify which ncPcdh family members are co-expressed with Pcdh19 at single-cell resolution in the developing hippocampus and cortex and to functionally test their role in combinatorial cell adhesion.Using probabilistic cell typing by in situ RNA sequencing (psicRNAseq) [32], we found that Pcdh19 was predominately co-expressed with Pcdh1, Pcdh9 or Cdh13 in Gria2+ glutamatergic excitatory neurons in developing mouse brain.Using cell aggregation assays, we demonstrated a critical role for PCDH19 in PCDH19+9 and PCDH19 + CDH13 combinatorial codes, suggesting that the mosaic absence of PCDH19 in the central nervous system (CNS) of PCDH19-CE-affected individuals could alter interactions of cells expressing these codes.Collectively, this study defines the spatial distribution of the Pcdh19-ncPcdhs combinatorial expression in the developing cortex and hippocampus and provides new insight into PCDH19-mediated cell-cell adhesion.

Identifying
Pcdh19 combinatorial expression at single-cell level in the developing mouse brain ncPcdhs are widely expressed in CNS during development and in adulthood [18,33] (electronic supplementary material, figure S1a).Although it has been recently proposed that a single cell can express more than one Pcdh [10,17], cell-specific co-expression of these molecules in the cortex and hippocampus in vivo is yet to be determined.Using the in situ seq method, we investigated co-expression of the 11 ncPcdhs in the developing mouse brain (18.5 dpc) at single-cell resolution (figure 1a-c).Consistent with previous reports, we found that ncPcdhs are highly expressed in mouse brain [4,10,34] (electronic supplementary material, figure S1a).We observed that a high number of cells in the cortex and hippocampus express Pcdh1 or Pcdh9; a medium-high number of cells express Pcdh7, 8, 10, 17, 18 or 19, whereas Pcdh11, 12 or 20 were detected only in a limited number of cells (electronic supplementary material, figure S1a).
Next, we investigated the co-expression of Pcdh19 and the other 10 ncPcdhs.In the cortex and hippocampus, Pcdh19 was co-expressed with most other ncPcdh family members, although the number of cells expressing a specific combination was highly variable (figure 1a-c and electronic supplementary material, figure S2a,b).Interestingly, in these brain regions, the highest number of cells co-expressed Pcdh19 and Pcdh9, and Pcdh19 and Pcdh1 (figure 1 and electronic supplementary material, figure S2), whereas only a relatively moderate number of cells expressed Pcdh19 together with its previously identified partners, Pcdh17 and Pcdh18 [10].
To further characterize the Pcdh19 co-expression code in the cortex and hippocampus, we next evaluated the number of other ncPcdhs expressed in Pcdh19 positive cells.A very low number of cells expressed Pcdh19 and two additional ncPcdhs, and no cells were found to express Pcdh19 and three or more ncPcdhs (figure 1d).These observations suggest that in the developing cortex and hippocampus, Pcdh19 combinatorial expression is mostly represented by only one additional ncPcdh.
To determine the neural cell types that express the Pcdh19 combinatorial code, we used expression markers to identify excitatory neurons (B-cell lymphoma/leukemia11B (Bcl11b/Ctip2), Calmodulin 2 (Calm2) and Glutamate Ionotropic Receptor AMPA Type Subunit 2 (Gria2)), interneurons (Somatostatin (Sst) and Parvalbumin (Pvalb)) and astrocytes (Aquaporin 4 (Aqp4) and Glial fibrillary acidic protein (Gfap)) (figure 2 and electronic supplementary material, figure S1b).Pcdh19 is predominantly expressed in glutamatergic Gria2 neurons.Commonly, excitatory neuronal subtypes express more than one cellular marker.Here, we have found that Pcdh19-expressing cells are positive for at least two excitatory neuronal markers (figure 2).Because the highest number of neurons expressing Pcdh19 are Gria2 positive, we then investigated the combinatorial expression of Pcdh19 and other ncPcdhs in this glutamatergic neuronal subtype (figure 3a,b).In cortical excitatory neurons, Pcdh19 was predominately co-expressed with Pcdh9.In addition, Pcdh1, 7, 10, 17 or 18 were also co-expressed with Pcdh19 in a moderate-low number of cells, whereas Pcdh19+8, 11, 12 or 20 were only detected in a few isolated excitatory neurons (figure 3a,c and electronic supplementary material, figure S3a).In hippocampal glutamatergic neurons, Pcdh19 was mostly co-expressed with Pcdh9 and Pcdh1.A medium-low number of cells were found to co-express Pcdh19 with Pcdh7, 8, 10, 17 or 18 (figure 3b,c and electronic supplementary material, figure S3b).We also investigated combinatorial Pcdh19 expression in interneurons and astrocytes (figure 2 and electronic supplementary material, figure S1b).Co-expression of Pcdh19 with other ncPcdh(s) was identified in a few Sst-positive interneurons, but not in Pvalb interneurons or astrocytes (Gfap+ or Aqp4+) (electronic supplementary material, figure S1a,b).Collectively, these observations reveal a region-specific and cell type-specific expression of Pcdh19+ncPcdhs combinatorial code.

Combinatorial expression of Pcdh19 and Cdh13 in cortex and hippocampus
CDH13 is expressed at very high levels in the developing cortex and hippocampus during embryogenesis and has been shown to play a role in neurogenesis and synaptogenesis [35].In situ RNA seq analysis showed that in the cortex a high number of cells co-express Cdh13 and ncPcdhs (figure 4 and electronic supplementary material, figure S4).Interestingly, although Pcdh19 is expressed in a moderate number of Cdh13-positive cells in the cortex (figure 4a), in the hippocampus, only a few isolated cells express Pcdh19 + Cdh13 (figure 4b).Analysis of the cell type expressing Cdh13 and Pcdh19 in the cortex and hippocampus showed that Pcdh19 + Cdh13 are mostly co-expressed in Gria2-positive excitatory neurons (figure 4c).To further compare Pcdh19 and Cdh13 expression, we performed co-immunostaining of CDH13 and PCDH19 using our previously established Pcdh19-HA/ Flag mouse model [10].Extensive co-localization of PCDH19 (HA) and CDH13 was detected in the developing cortex (figure 4d), consistent with our in situ RNA seq data.

Unique adhesion function of PCDH19 in PCDH19 + CDH13 and PCDH19+9 combinatorial codes
We and others have shown that cell populations expressing multiple ncPCDH can only completely intermix if they express the same combination of PCDH molecules and that cells expressing a 'partially overlapping code' cannot [10,17].Prompted by our in vivo co-expression data, we investigated the combinatorial adhesion properties of PCDH19 with PCDH1 or PCDH9 using K562 cells (which do not express endogenous cadherins) [10,17].Aggregation assays were performed by overexpressing PCDH19 with PCDH1 or PCDH9 with green fluorescent protein (GFP) and mCherry (MCH) fluorescent reporters to label each cell population (electronic supplementary material, figure S5).
To investigate the PCDH combinatorial activity, we performed mismatch co-aggregation assays in which populations of K562 cells expressing PCDH19+1 or PCDH19+9 were mixed with cells expressing individual proteins of each combination (figure 5a,b).To quantitate the degree of cell mixing, we performed Pearson's correlation coefficient analysis [10].The maximum level of intermixing (100%) was established using cells expressing the same combination of adhesion molecules: PCDH19+1(GFP) and PCDH19+1(MCH) or PCDH19+9(GFP) and PCDH19+9(MCH).Firstly, we assessed the PCDH19 adhesion function in PCDH19+1 combinatorial code (figure 5a).An approximately 70% decrease in relative mixing was observed between cells expressing the complete code (PCDH19+1) and those expressing PCDH19 only.However, no significant difference in the level of relative mixing was detected between cells expressing PCDH19+1 and PCDH1 only (figure 5a).
Next, we investigated the role of PCDH19 in PCDH19+9 combinatorial code.A significant (approx.85%) reduction in relative mixing was observed between populations expressing the complete code (PCDH19+9) and those expressing PCDH9 only (figure 5b).In contrast, relative mixing between cells expressing PCDH19+9 and PCDH19 was not significantly affected (figure 5b).Next, we evaluated the combinatorial CDH13+PCDH19 activity by performing mismatch co-aggregations assays as described above.Two populations of fluorescently labelled K562 cells expressing identical (PCDH19+CDH13 and CDH13+PCDH19) or different (PCDH19+CDH13 and PCDH19) or (PCDH19+ CDH13 and CDH13) combinations were mixed (figure 5c).Significant sorting between PCDH19+CDH13 and CDH13 compared with control (PCDH19 + CDH13 and PCDH19 + CDH13) was observed.In contrast, cell populations expressing PCDH19+CDH13 and PCDH19 did not show a significant decrease in cell mixing compared with control (figure 5c).Taken together these observations indicate that mosaic absence of PCDH19 in PCDH19+9 and PCDH19+CDH13, but not in PCDH19+1 codes significantly alter cellcell interaction.
To investigate the adhesion properties of PCDH19 in combination with CDH13, we performed aggregation experiments using K562 cells (figure 5d).Firstly, we compared cell populations expressing a single PCDH19 and CDH13 molecule labelled with different fluorescent markers (MCH/GFP) as described above.Extensive sorting was observed when cells expressing either PCDH19 or CDH13 were mixed (electronic supplementary material, figure S5).In contrast, combining cell populations expressing PCDH19+CDH13 showed extensive mixing confirming homotypic cell-cell adhesion activity of both adhesion proteins (figure 5d).Interestingly, CDH13 alone showed low cell-cell adhesion; however, its co-expression with PCDH19 significantly increased cell aggregation (figure 5d).
Next, we investigated whether PCDH19 as a key adhesion protein in two different combinatorial codes could promote a heterophilic interaction between them.We mixed two populations of fluorescently labelled K562 cells expressing PCDH19+CDH13 with PCDH19+PCDH9 (figure 6a).No significant difference in cell mixing was observed compared with control (PCDH19+CDH13 and PCDH19+CDH13).We also performed a reciprocal experiment where we mixed cells expressing combinatorial codes with different 'dominant' adhesion proteins: PCDH19+CDH13 or PCDH19+PCDH9 (PCDH19 is the key adhesion protein) with PCDH19+PCDH1 (PCDH1 is the key adhesion protein) (figure 6a).Near complete segregation of PCDH19+9 or PCDH19+CDH13 and PCDH19+1 cells versus control cells was observed.Altogether these observations suggest that the heterophilic adhesion between two different cell populations could occur only if the same dominant ncPCDH (PCDH19) is present.
Cadherins have previously been shown to undergo cis-and trans-interactions at the interface of adjacent cells [36].To evaluate PCDH19 cis-interaction with PCDH9, PCDH1 and CDH13, we performed immunoprecipitation analysis of differentially tagged proteins expressed in HEK293T cells (figure 6b-d).Our results revealed that PCDH19 could interact with both   δ1-PCDHs and CDH13.Taken together these observations suggest that despite the similar cis-interaction between PCDH19 and PCDH1/PCDH9/CDH13, each individual combinatorial code presents trans sensitivity only to a specific ncPCDH.

Discussion
Overlapping expression and spatial distribution of ncPCDHs support their potential combinatorial contribution to brain architecture and physiology [16][17][18]37].Although we [10] and others [17,18] were able to demonstrate the co-expression of ncPcdhs in the brain, the complexity of this expression code at single-cell resolution across specific brain regions remains poorly (a) understood.Using in situ RNA sequencing in vivo, we have found that most cortical and hippocampal excitatory neurons express two ncPcdhs (δ1 + δ2 or δ1 + δ1).Consistent with our observations, a recent study performed by Bisogni et al. revealed that olfactory neurons also mostly express two ncPcdhs, although some of these cells could express up to seven [17], something that we have not observed in the cortex and hippocampus.In contrast to our study, Bisogni et al. used the NanoString nCounter platform to identify ncPcdh combinatorial expression in post-natal olfactory neurons.Thus, the greater complexity of Pcdh expression identified by Bisogni et al. may reflect a difference in technical sensitivity or developmental stage specificity of ncPcdh expression.Another potential explanation is that the number of ncPcdhs expressed per cell is region-specific, perhaps reflecting requirements for local circuit formation and/or maintenance [15].Pcdhs have been implicated in multiple steps in excitatory neuronal circuit formation [1].In the present study, we focus on Pcdh19, a causative gene of PCDH19-CE, hypothesizing that perturbation of Pcdh19-associated combinatorial adhesion codes could compromise CNS function in affected heterozygous females.We have identified prevalent expression of Pcdh19+9 and Pcdh19+1 in the hippocampus and cortex which correlates with the relatively high expression of these proteins in the brain.Considering that PCDH9 and 19 are found in excitatory neuron synapses and variants in their cognate genes have been linked to neurodevelopmental disorders [9,29,38], it is possible that they control similar neurocircuit processes.In contrast, besides the higher expression of Pcdh1 in the mouse brain, mutations in PCDH1 have been exclusively linked to respiratory disorders in humans [39] suggesting a potential auxiliary cell adhesion role in the CNS, perhaps through other PCDHs.
It has been previously reported that Pcdh19 is expressed in a relatively high number of excitatory and interneurons, as well as in astrocytes at post-natal stages [40].In the developing brain, although Pcdh19 is expressed only in glutamatergic neurons and in some Sst-positive interneurons, but not in Pvalb-positive interneurons or astrocytes.Importantly, Pcdh19+ncPcdhs combinatorial expression was observed only in excitatory neurons, but not in Sst interneurons.To date, there is no evidence to support the existence of a combinatorial ncPcdh adhesion code in interneurons, perhaps reflecting a role for combinatorial expression of other cadherins, or even the complete lack of adhesion code in this cell type.Hence, the determination of combinatorial adhesion code in this cell type requires further investigation.On the other hand, the lack of Pcdh19 expression in both astrocytes and Pvalb interneurons could be due to the lower prevalence of these cell subtypes during brain development or the complete absence of Pcdh19 expression in these cells throughout embryogenesis.We [10] and others [17] have shown that missing a single protein in a PCDH combinatorial code significantly reduces intermixing with cells expressing the complete code.Here, we demonstrate that this is not always the case; for example, cells expressing PCDH19+9 completely intermix with cells expressing PCDH19 only.Similar results were reported by Bisogni et al., where the authors demonstrated that populations expressing the complete code in some cases could still intermix with cells expressing only a dominant protein [17].We have also shown that a key protein in one combinatorial code does not necessarily play a dominant role in a different combinatorial environment.However, considering the affinity differences and/or relative expression levels of ncPCDHs [17] mismatch aggregation assay may not comprehensively address these behaviours.In the case of PCDH19+1, missing PCDH1, but not PCDH19 has a significant impact on cell-cell adhesion between cells expressing the complete code.This finding may reflect a level of context-dependent redundancy in the combinatorial role of PCDH family members during CNS development in neuronal circuit assembly and maintenance.Interestingly, we have found that intermixing between cells expressing PCDH19 plus different adhesion molecules could occur only if PCDH19 is the dominant protein in both codes, as we have observed in the case of PCDH19+CDH13 and PCDH19+9.This finding could explain why missing a single adhesion protein causes such a diversity in the physiological outcome in a cell-specific manner.Considering that most of the neurons in the brain express several PCDHs and CDHs, the lack of one protein could be potentially compensated by other adhesion molecules.However, depending on whether a dominant protein of a combinatorial code is missing, will define whether the neuronal communication and brain circuit may be impacted.
Cis-interaction between PCDHs has been widely studied in cPCDHs; however, little is known about such interactions in ncPCDHs.It has been proposed for clustered and ncPCDHs that generate specific cell surface interaction codes using a 'zippering' mechanism where cis dimers of PCDHs from one cell interact in trans with complementary dimers from a second cell in antiparallel orientation [20,21].Interestingly, besides the structural differences, it seems that ncPCDHs behave in a similar way.To date, only PCDH10 and PCDH17 have been shown to interact in cis with PCDH19 [10].Although a recent biophysical study of EC domains of ncPcdhs has shown the lack of strong cis-interactions between members of the family [20], here using co-immunoprecipitation assays we were able to demonstrate an intracellular (cis) interaction between full-length ncPCDHs possessing a different number of EC domains: PCDH1 and PCDH19, and PCDH9 and PCDH19.Importantly, Harrison and collaborators focused on EC domains only, and in their original work, the authors do not exclude the possible contribution of other (transmembrane or cytoplasmic) domains in the establishment of these interactions [4,20].
In the last decade, it has been proposed that there is a close functional relationship between δ2-Pcdh members and classical cadherins [7,41].Intriguingly, PCDH19 has been found to interact with N-cadherin (NCAD) regulating cell movement during morphogenesis in zebrafish [42].In addition, a cis-interaction between PCDH19 and NCAD was proposed to be involved in hippocampal presynaptic function and cognitive behaviour [43].Here, we have shown that Pcdh19 and Cdh13 are highly co-expressed in glutamatergic cortical neurons and that these proteins can interact in cis.In the nervous system, CDH13 together with PCDH19 is found in synaptic space controlling vital neuronal processes including axonal outgrowth and synaptic formation, consistent with their genetic contribution to a number of neurodevelopmental disorders [4,9,29,44,45].
Interactions between ncPcdhs and cadherins have been shown to modulate their adhesion abilities [42].For instance, the lack of Pcdh19 or Ncad negatively affects cell adhesion.Importantly, CDH13 protein lacks the cytoplasmic domain characteristic of other cadherins, suggesting that its adhesion abilities could be promoted by additional interactors.Here, we have found that cis (intracellular) interaction between CDH13 and PCDH19 can impact CDH13 adhesion, as loss of PCDH19 in the same cell affects CDH13 cell-cell aggregation abilities suggesting that ncPCDHs and cPCDHs interactions may play a role in the modification of cell-cell adhesion properties in vivo.
It has been shown that mosaic expression of PCDH19 disrupts neuronal communication in PCDH19-CE [9,43].The critical role of PCDH19 in PCDH19+9 and PCDH19+CDH13 combinatorial codes suggests that missing PCDH19 in the mosaic population could disturb the adhesion properties between the cells expressing the complete code and the PCDH19 mutant cells expressing CDH13 or PCDH9 only.Considering the synaptic localization of PCDH9, CDH13 and PCDH19 [4,5,9,[46][47][48], a mismatch in expression could alter the synaptic contact adhesion leading to network dysfunction and cognitive impairment, both of which are features of PCDH19-CE pathology.

Experimental animals
In situ sequencing was performed using CD1 mice (Jackson Laboratory) housed at Karolinska Institute, Sweden, and all the procedures and experiments were performed in accordance with Swedish animal welfare laws authorized by the Stockholm Animal Ethics Committee: DNR 3796-2020.
The animals used for co-immunostaining of PCDH19 and CDH13 were housed at the South Australian Health and Medical Research Institute, Australia, and the experiments were conducted under SAM20-015.Experiments were performed using a previously established Pcdh19-Tag mouse model [10].
4.1.1.In situ sequencing and immunostaining of mouse samples Brain samples (18.5 dpc) were cryosectioned coronally at 10 µm (mouse brain atlas position 250-276), mounted onto Super Frost plus microscope slides (Thermo Fisher Scientific) and stored at −80°C.The slides were then transferred on dry ice to CARTANA (Solna, Sweden; now part of 10× genomics) for tissue fixation and in situ sequencing.The reverse transcription, probe ligation, rolling cycle amplification with reagents and according to the procedures supplied in the Neurokit (1010-01, CARTANA, Sweden) were performed at CARTANA followed by fluorescence labelling, and sequencing by sequential images.One probe was designed for each ncPcdh gene: Pcdh1, 7, 8, 9, 10, 11x, 12, 17, 18, 19 and 20, and Cdh13, and each cellular marker: Gira2, Calm4, Bcl11b (Ctip2), Stt, Pvalb, Aqp4 and Gfap (electronic supplementary material, figure S6).The data were collected over six rounds of five-colour microscopy images.The original images are taken using 40× objective.The result table of the spatial coordinates of each molecule of target genes together with the reference DAPI image per sample were provided by CARTANA.A customized script was used for data analysis from the original CARTANA script to approximate the locations of cells and assign individual CARTANA reads to specific cells.

In situ sequencing imaging analysis for cell type mapping
Cellular markers for excitatory neurons (Gria2, Calm2 and Bcl11b), inhibitory neurons (Sst and Pvalb) and astrocytes (Gfap and Aqp4) were used to define the cellular population expressing a single or combination of ncPCDHs.Each cellular marker was tested for expression of a combination of all 12 ncPcdhs simultaneously.The total number of cells expressing a combination of each of the two Pcdhs was analysed in the whole cortical and hippocampal areas/slice (left and right hemisphere) in each of the two sections per mouse (three mice in total).For generic co-expression of Pcdh19 and single members of ncPcdhs in the brain, Pcdh19 was used as a selection marker (figure 1).For expression of Cdh13 and single members of ncPcdhs, Cdh13 expressing cells were used as a selection marker.The data are presented as a heat map as an average of the three samples (two sections/mouse).The number of cells in the cortex expressing ncPcdhs and/or Cdh13 were counted and represented as an average of six samples.Due to a very close proximity and very high number of Gria2+ neurons expressing Pcdh1+9, Pcdh7+9, Pcdh9+10 or Pcdh9+18, and Cdh13 co-expressing neurons and Pcdh1, Pcdh9, Pcdh7 or Pcdh10 per brain (data not shown), their values are indicated as greater than 150 in the graphs in electronic supplementary material, figures S3 and S4.

Immunofluorescence of brain slices
E18.5 mouse brains were fixed in 4% PFA at 4°C overnight as previously described [49].Brains were cryoprotected in 30% sucrose and frozen in the Optimum Cutting Temperature embedding medium.Then 16 µm sections were prepared using a Leica CM1900 cryostat.Sections were blocked with 0.1% Triton X-100 and 10% horse serum in 1× PBS for 1 h at room temperature.Sections were then incubated overnight with rabbit anti-HA antibody 1:400 (Sigma) and anti-CDH13 1:300 (In Vitro Technologies) at 4°C.Slices were then washed three times with PBS and incubated with secondary donkey anti-rabbit Alexa 594 and anti-chicken 488 (Jackson ImmunoResearch) antibodies for 2 h at room temperature.Slides were mounted in Prolong™ Gold Antifade with DAPI (Invitrogen #P36931), and the images were acquired on a Nikon Eclipse Ti microscope and Nikon Digital Sight DS-Qi1 camera at 20×.

K562 mixing assay and cell aggregation imaging
Totally, 2 × 10 6 K562 cells were nucleofected as previously described in [10].Briefly, the cells were transfected using 5 µg of each plasmid DNA (10 µg total for co-nucleofections) and incubated for 24 h.Cells were harvested and centrifuged at 170g for 7 min, then the pellets were resuspended in culture media to obtain a single-cell suspension.From each sample, 2.5 × 10 5 cells (5 × 10 5 cells/well) were pooled and added to each well of a 24-well tray and allowed to aggregate for 3-4 h on a nutator at 37°C/5% CO 2 .Three to four images of aggregated cells were taken of each replicate using a Nikon Eclipse Ti2 microscope with a 10× objective.Each experimental condition was performed with a minimum of three biological replicates with three technical repeats.

Mixing assay quantification
Images (10×) were acquired and exported to Fiji software where they were subjected to an equal threshold transformation, and aggregate size was assessed using the Coloc2 plugin function.The Pearson's correlation coefficient for particle sizes from a minimum of three biological repeats (three images from three technical replicates in each) was calculated using the Coloc2

Figure 1 .
Figure 1.Expression profile of Pcdh19 + ncPcdhs in mouse cortex and hippocampus.Heat map of the mean number of cells co-expressing Pcdh19 and ncPcdhs in the cortex (a) and hippocampus (b).Each row represents the number of cells co-expressing Pcdh19 and an individual member of ncPcdhs (Pcdh1, 7, 8, 9, 10, 11, 12, 17, 18 and 20).(c) Representative phase contrast images of Pcdh19-only and co-expressing Pcdh1 or 7, 8, 9, 10, 11, 12, 17, 18 or 20 cells in a mouse brain at 18.5 dpc.The cortical areas highlighted in magenta and the hippocampal areas highlighted in white considered for the analysis in (a) and (b).(d) Representative images of Pcdh19 positive cells in the brain that co-express Pcdh10+17, Pcdh10+18, Pcdh17+18 and Pcdh1+9.The green 'X' in (c) and (d) represents the position of Pcdh-expressing cells in the brain based on the DAPI signal.Scale bar 500 µm.