WDFY3 cell autonomously controls neuronal migration

Proper cerebral cortical development depends on the tightly orchestrated migration of newly born 19 neurons from the inner ventricular and subventricular zones to the outer cortical plate. Any disturbance in this process during prenatal stages may lead to neuronal migration disorders 21 (NMDs), which can vary in extent from focal to global. Furthermore, NMDs show a substantial comorbidity with other neurodevelopmental disorders, notably autism spectrum disorders (ASDs). Our previous work demonstrated focal neuronal migration defects in mice carrying loss-of- function alleles of the recognized autism risk gene WDFY3 . However, the cellular origins of these defects in Wdfy3 mutant mice remain elusive and uncovering it will provide critical insight into WDFY3 -dependent disease pathology Our genetic approach revealed several cell autonomous requirements of Wdfy3 in neuronal development that could underly the pathogenic mechanisms of WDFY3 -related ASD conditions. The results are also consistent with findings in other ASD animal models and patients and suggest an important role for Wdfy3 in regulating neuronal function and interconnectivity in postnatal life. on neuronal morphology and circuit integration by evaluating maldevelopment neuronal migration defects of excitatory neurons. MADM technology in this study, the conclusion that these migration errors are driven by cell autonomous mechanisms that act directly on mutant neurons and their progenitors rather than impaired cell-environment interactions. In addition, we discovered alterations in mutant neuronal morphology consistent with other reports documenting decreased dendritic arbor complexity but increased spine density in mouse models and human cases. Our findings further underline the validity of Wdfy3 mice as ASD models and point to the significance of changes at the chemical synapse in

comorbidity with other neurodevelopmental disorders, notably autism spectrum disorders (ASDs). 23 Our previous work demonstrated focal neuronal migration defects in mice carrying loss-of-24 function alleles of the recognized autism risk gene WDFY3. However, the cellular origins of these 25 defects in Wdfy3 mutant mice remain elusive and uncovering it will provide critical insight into 26 WDFY3-dependent disease pathology . 27

Methods 28
Here, in an effort to untangle the origins of NMDs in Wdfy3 lacZ mice, we employed mosaic analysis 29 with double markers (MADM). MADM technology enabled us to genetically distinctly track and 30 phenotypically analyze mutant and wild type cells concomitantly in vivo using immunofluorescent 31 techniques. 32

Results 33
We revealed a cell autonomous requirement of WDFY3 for accurate laminar positioning of cortical 34 projection neurons and elimination of mispositioned cells during early postnatal life. In addition, 35 we identified significant deviations in dendritic arborization, as well as synaptic density and 36 morphology between wild type, heterozygous, and homozygous Wdfy3 mutant neurons in  MADM reporter mice at postnatal stages. 38 Background 56 57 During prenatal neurogenesis, newly born neurons are deployed from proliferative compartments 58 surrounding the ventricles towards the surface of the brain where they will settle into their proper 59 laminae and nuclei to form functional circuits [1][2][3][4]. Any disturbance of this tightly orchestrated 60 process can result in neuronal migration disorders (NMDs), birth defects with often devastating 61 dendritic arborization, spine density and morphology. Our results demonstrate a true cell-108 autonomous role of Wdfy3 in regulating either aspect of neuronal laminar and circuit integration. Mice carrying the Wdfy3 lacZ (Wdfy3 tm1a(KOMP)Mbp ) allele were generated and genotyped as 122 previously described [24] and maintained on C57BL/6NJ background. To generate Wdfy3 lacZ -123 MADM-5 mice, we crossed Wdfy3 +/lacZ with homozygous MADM-5 GT/GT mice [31,61]. 124 Subsequently, compound heterozygous offspring was crossed with homozygous MADM-5 TG/TG ; 125 in PBS three times for 5 minutes and tissue blocked with 10 % donkey serum in PBS with 1% 150 Tritton X-100 (PBST) for 1 h at room temperature. After being washed with PBS, sections were 151 incubated with chosen primary antibodies for 18 h at 4 °C, subsequently washed five times for 10 152 min each with PBS, and secondary antibodies applied for 2 h at room temperature. After secondary 153 incubation, slides were washed five times for 10 min each in PBS, and for 5 min submerged in a 154 0.1% 4′,6-diamidino-2-phenylindole (DAPI) (Thermo Fisher Scientific) to create a nuclear 155 counterstain. After being washed twice 10 min each with PBS, sections were covered with 156 Fluoromount-G (Thermo Fisher Scientific), mounted with cover glass (Thermo Fisher Scientific), 157 and then allowed to dry in a dark ventilated area for a minimum of 20 h before being imaged. 158 Agarose embedded sections (200 μm) were immunostained free floating on an orbital 159 shaker (Labnet) at 30 rpm using the process described above while omitting antigen retrieval. 160 Transfer of sections into washes, blocking, and incubation liquids were done using a paintbrush.  Microscopy and image processing. To assess genotype distribution and laminar positioning 182 fluorescent confocal microscopy and image acquisition was performed on an inverted Nikon C1 183 microscope with associated software, typically at 10x magnification. Images used for cell 184 morphological analysis, including dendritic arbor and spine analysis were obtained on a Nikon A1 185 microscope with associated software. Subsequently, images were uploaded and cells counted in 186 NIS-Elements. Cellular Morphology was analyzed in FIJI using the Simple Neurite Tracer plugin. 187 188 Statistical testing. All data were from processed, analyzed, and graphical figures created in 189 Graphpad Prism 9. To remove suspected outliers, data were initially processed with the ROUT 190 outlier test (Q value = 1%). Applicable data were then tested for normality, variance, and standard 191 deviation before analysis. Lamination analysis was done using Fisher's exact test, population 192 analysis was done using a unpaired student's t-test, Sholl profile analysis was statistically 193 evaluated by two-way Analysis of Variance (ANOVA), followed by a Tukey's multiple 194 comparison test. Analysis of bouton density was done using a one-way ANOVA with multiple 195 comparisons while bouton subtype distribution analysis was done using a two-way ANOVA with 196 multiple comparisons. When applicable and reasonable to add, data are reported as a mean with 197 standard deviation. Bar graphs with replicates are reported as mean with standard error of the 198 mean. Results were considered to be statistically significant if p ≤ 0.05. Individual data points 199 largely correspond to replicates of one brain. The extent of significance between groups is 200 indicated with one, two, three, or four asterisks if p-values were equal to or less than 0.05, 0. astrocytes within all cortical layers with cells distributed across labels associated with the 216 anticipated recombination of fluorophore encoding cassettes. We detected red tdT + cells expected 217 to be wild type (WT or +/+), green homozygous mutant GFP + cells (lacZ/lacZ), and yellow 218 heterozygous tdT + /GFP + cells (+/lacZ) (Fig. 1A, B). We proceeded to count all cells in ten sections 219 each of 3 male and 3 female brains per stage and converted the added numbers of each genotype 220 to ratios of a whole. In total, we counted at P8 685 tdT + cells, 1,330 tdT + /GFP + cells, and 658 GFP + 221 cells and at P30 614 tdT + cells, 1,236 tdT + /GFP + cells, and 262 GFP + cells. Genotype populations 222 were compared across each other and between the two time points (P8 and P30). In P8 brains, the 223 distribution of GFP labeled homozygous mutant neurons was ~24%, likely due to cell death 224 significantly decreasing to ~12% in P30 brains (0.2368 ± 0.029 at P8 and 0.1216 ± 0.028 at P30; 225 p < 0.0001) (Fig. 1C). This relative decrease in mutant cells at P30 was accompanied by a 226 significant relative increase of both WT (0.2634 ± 0.026 at P8 and 0.2949 ± 0.010 at P30; p = 0.02) 227 and +/lacZ cells (0.4998 ± 0.024 at P8 and 0.5835 ± 0.026 at P30; p = 0.002) (Fig. 1C). No 228 significant differences between the sexes were observed. 229 230

Cortical lamination errors occur more frequently in Wdfy3 homozygous mutant cells 231
Realizing that early postnatal development is associated with the disproportionate loss of lacZ/lacZ 232 neurons, we proceeded to examine whether this loss is associated with the laminar positioning of 233 these neurons. We performed immunofluorescent labeling targeting endogenous tdT and eGFP 234 expression produced as a result of the MADM system, and colabeled for cortical layer markers 235 Tbr1, Ctip2, and Brn2, each at a time. Tbr1 is predominantly expressed by layer VI neurons, Ctip2 236 by layer V neurons, and Brn2 by layer II/III neurons respectively. The approach allowed us to 237 identify neurons of either fluorescent marker and genotype and, if positive for one of the cortical 238 layer markers, assess whether they were also correctly positioned within their respective layer. 239 Analyzing a total number of 730 cells of six brains (3 male, 3 female) at P8, mutant neurons 240 in Wdfy3-MADM mice were mispositioned more often than their WT counterpart. Tbr1 + (VI) 241 mutant cells were correctly positioned in 46% of instances compared to ~91% of WT cells (WT 242 90.66%; lacZ/lacZ, 46%; p < 0.0001) (Fig. 2 A, G). Mutant Ctip2 + cells (V) were also significantly 243 less often correctly positioned compared with WT cells (WT, 78.79%; lacZ/lacZ, 29.76%; p < 244 0.0001) (Fig. 2 C, H). The same observation was made with Brn2 + cells (II/III) with mutant 245 neurons being significantly less often correctly located compared with WT neurons (WT, 89.78%; 246 lacZ/lacZ, 53.91%; p < 0.0001) (Fig. 2 E, I). Examining layer-specific positioning of 735 neurons 247 in six Wdfy3-MADM brains at P30 (3 male, 3 female), we found similar, but less exaggerated 248 discrepancies between WT and lacZ/lacZ neurons in P8 brains. Tbr1 + mutant cells were aligned in 249 the correct layer in ~63% of instances, compared to 87% for their WT counterpart (WT, 86.81%; 250 lacZ/lacZ, 62.82%; p = 0.0003) (Fig. 2 B sexes. For each genotype the mean number of intersections peeked at ~100 μm from the soma and 281 significant differences between genotypes were mostly confined to distances of 15 μm to 200 μm 282 from the soma. The number of maximum intersections for deep layer WT neurons at ~100 μm was 283 in average 18 while lacZ/lacZ neurons showed a peak average of 13.7 intersections (p < 0.05 -284 0.0001). Heterozygous neurons were slightly but not significantly more complex than WT with an 285 average of 19.5 intersections at ~100 μm distance from the soma (+/lacZ vs. lacZ/lacZ, p < 0.05 -286 0.001) (Fig. 3D). In upper layers, WT and +/lacZ neurons showed near identical Sholl profiles 287 with peak intersections at ~100 μm being on average 22.3 while lacZ/lacZ neurons showed a 288 highest average of 15.7 intersections at ~100 μm from the soma (WT or +/lacZ vs. lacZ/lacZ, p < 289 0.05 -0.0001) (Fig. 3E). 290 291

Wdfy3 mutation increases synaptic spine density 292
To assess dendritic spine density and subtype distribution, 200 μm sections of four male and three 293 female brains were immunolabeled and cleared via FOCM as described above. From acquired z-294 stacks of 42 labeled cells of each genotype, 10 μm (in z-depth) regions of dendrites were then 295 extracted and dendritic spines counted while also tracking morphological features associated with 296 widely described subtypes, including filopodia, long thin, thin, stubby, mushroom, and branched 297 [36]. 298 A decrease in Wdfy3 gene dosage led to an overall increase in spine density. In WT neurons, 299 spine density averaged 0.29 spines per 1 μm. In heterozygous neurons, density significantly rose 300 to an average of 0.42 spines per μm. In lacZ/lacZ cells, we recorded an even higher density of 0.54 301 spines per μm while no sex-specific differences were observed in any genotype. (WT, 0.2907 ± 302 0.1111; +/lacZ, 0.4188 ± 0.1004; lacZ/lacZ 0.5409 ± 0.1609; WT vs. +/lacZ p = 0.0286; WT vs. 303 lacZ/lacZ p = 0.0011; +/lacZ vs. lacZ/lacZ; p = 0.0287). 304 Dendritic spines can be distinguished by morphological criteria that associate with their 305 state of maturation and function. A well-replicated rank order of maturation distinguishes spines 306 that are filopodia, long thin, thin, stubby, mushroom shaped, and branched [36]. While assessing 307 overall spine density, we traced these different subtypes and converted counts to a percentage of a 308 whole to analyze distribution across genotypes. No significant differences between genotypes were 309 detected from a two-way analysis of variance (ANOVA), followed by a Tukey's multiple 310 comparison test (Fig. 4 H). Irrespective of genotype, the most prevalent spine types were long thin, 311 thin, and mushroom, comprising ~90% of all counted spines. In summary, we found Wdfy3 312 mutation to affect spine density in a gene dosage responsive manner, but to have no effect on spine 313 subtype distribution. During mouse cortical neurogenesis, WDFY3 is predominantly expressed by dividing 337 RGCs, the pia, and diffusely in the intermediate zone [24]. By using Emx1 Cre as a driver, this study 338 sought to isolate cell autonomous vs. non cell autonomous effects of WDFY3 loss on radial 339 migration by conditionally limiting mutation to cortical projection neurons and their progenitors. 340 Our results strongly suggest the predominance of intrinsic WDFY3 activity guiding migration and 341 possibly survival of cortical neurons in the developing cortex. Indeed, while at P8 WT cells 342 overwhelmingly localize within the cortical layers specific to their marker identity, lacZ/lacZ cells 343 do so at significantly reduced frequency, typically at only half the rate of WT cells. At P30, we 344 found the ratio of lacZ/lacZ neurons that are correctly positioned within layers VI (Tbr1 + ) and 345 II/III (Brn2 + ) to be elevated compared to P8. This finding very likely points to corrective 346 mechanisms that are designed to eliminate misplaced and/or inadequately circuit-integrated 347 neurons from the developing brain via programmed cell death. Indeed, TUNEL analysis of mouse 348 parietal neocortical fields 1, 3 and 40 (largely corresponding to the area examined in the present 349 study) revealed cell death to peak at P4, but also to continue well into the third postnatal week [32]. is controlled by electrical activity appears also strongly supported by elegant work that associated 356 electrical activity patterns with neuronal death rates in two distinct cortical areas (M1 and S1) [57]. 357 Thus, the suggestion that misplaced neurons may be especially targeted for removal as their 358 incorporation into physiologically functional networks is less likely, plausibly explains our 359 observations of progressively fewer lacZ/lacZ neurons located outside their correct layers.   show significantly higher bouton density compared with wild type. Homozygous mutant neurons 648 also display higher bouton density compared with heterozygous neurons (p ≤ 0.05). Scale bar is 20 649