DLC3/Cv-c function in testis development in humans and Drosophila: implication for variants of sex development

Identifying genes affecting gonad development is essential to understand the mechanisms causing Variants/Differences in Sex Development. Recently, a DLC3 mutation was associated with male gonadal dysgenesis in 46,XY DSD patients. We show that Cv-c, the Drosophila ortholog of DLC3, is also required to maintain testis integrity during fly development. We found that Cv-c and human DLC3 can perform the same function in fly embryos, as flies with wild type but not mutated DLC3 rescue gonadal dysgenesis, suggesting a functional conservation. Expression of different Cv-c protein variants demonstrated that the StART domain mediates the Cv-c function in the male gonad, independently from the GAP domain activity. This work demonstrates a role for DLC3/Cv-c in male gonadogenesis and highlights a novel StART-mediated function required for gonadal mesoderm-germ cell interaction during testis development. Significance Statement Associating rare human genetic variants to specific conditions is complex. An amino acid change in the StART domain of the RhoGAP DLC3 protein has been found in female DSD patients with a 46,XY male karyotype. We present a second DSD patient with a mutation in the same region and show that the Cv-c/DLC3 homolog is also required in Drosophila testis. In cv-c mutant embryos the testis mesodermal cells display various defects: the testes are unable to retain the germ cells due to their abnormal ensheathment by mesodermal cells and the mesodermal sheet of cells surrounding the testis is discontinuous resulting in the liberation of the germ cells outside the gonad. Defects can be rescued by gonadal expression of Cv-c or DLC3 but not by the patients’ DLC3 protein variant. Testis development requires the StART lipid binding domain but not the GAP domain, revealing a novel function of this RhoGAP family.

of the DLC3-StART domain and suggest that alterations of this Ω1 loop could impair the 164 domain's membrane interaction. 165

Expression of the cv-c RhoGAP gene in the Drosophila gonads 166
Structural analyses have shown that DLC3 and Cv-c are highly conserved ( Fig. 2A and  167 B) and that, in Drosophila, DLC3 can functionally substitute for Cv-c (15). However,

Cv-c functional requirement during male gonadogenesis in Drosophila 185
To test the functional significance of Cv-c expression in the fly gonads we analysed 186 embryos homozygous for the lethal nonsense cv-c M62 and cv-c C524 alleles where stop 187 codons result in truncated proteins lacking the GAP and StART domains (Denholm et al., 188 2005). We did not detect any major morphological defects in female gonads, confirming 189 cv-c is not required for embryonic ovary development (Fig. 3B). In contrast, homozygous 190 or hemizygous cv-c mutant male embryos have abnormal testes containing germ cells 191 that are not surrounded by the gonadal mesoderm ( Fig. 3C and E arrowheads compare with A and D respectively). This defect is unlikely to be due to the abnormal specification 193 of the gonad mesoderm cells, as using mesodermal specific markers we can observe 194 the presence of all cell types (Fig. 3C, E and G). However, the gonad mesoderm cells 195 are frequently displaced, with the pigment cells failing to completely surround the mutant 196 testis (Fig. 3E). 197 To test if these defects are due to the abnormal migration of the germ cells or the 198 mesoderm gonadal precursors during early gonad organogenesis [up to gonad 199 coalescence at stage 15 (st15)] or to later defects on testis maintenance, we labelled the  Analysis of fixed mutant testes shows that the extruded germ cells extend blebs that are 208 more characteristic of the earlier migratory phase (Fig. 3I to J arrowheads) (Jaglarz & 209 Howard, 1995). These blebs are not observed in the ovaries of cv-c mutant females 210 ( Supplementary Fig. 2B) nor in the wild type testis after gonad compaction (Fig. 3H). 211 212

Rescue of cv-c Drosophila testis defects with human DLC3 213
We have previously shown that DLC3 can rescue the mutant phenotypes caused by cv-c 214 mutations in the Malpighian tubules, the kidney-like structures of the fly, indicating these 215 homologous Drosophila and human proteins conserve similar functions (Sotillos et al., 216 2018). Therefore, given that Cv-c is expressed and required in the male Drosophila 217 gonad, and that DLC3 can functionally substitute for Cv-c in some tissues, we 218 investigated if DLC3 is also capable of rescuing the mutant gonadal defects observed in 219 cv-c C524 homozygous embryos ( Fig. 4A and D). Using the UAS/Gal4 system to express 220 wild type DLC3 protein with the pan mesodermal twi-Gal4 driver line in otherwise cv-221 c C524 mutant embryos, we efficiently rescued the testis defects (Fig. 4A, B and E) 222 pointing out to the conservation of DLC3/Cv-c function in gonadogenesis. In contrast,

Testis development in Drosophila requires the StART domain 227
To elucidate the molecular mechanisms mediating DLC3/Cv-c function in gonad 228 development, we analysed the capacity of different Cv-c protein variants to rescue the 229 testis defects of cv-c c524 homozygous embryos (Fig. 4A). 230 As mentioned above, StART domain mutations in DLC3 are the suspected cause of 231 gonadal dysgenesis in human patients. In agreement with this, we found that expression 232 of a UAS-cv-c ∆StART construct generating a Cv-c protein lacking the StART domain, does 233 not significantly normalize the testis defects ( Fig. 4C and I). In comparison, the 234 expression of the wild type Cv-c protein rescued the abnormal phenotypes in more than 235 50% of the testes ( Fig. 4C and G). Surprisingly, expression of a Cv-c mutant protein 236 substituting a highly conserved Arginine into Glutamine residue that has been shown to 237 block the GAP domain activity in vitro and the protein function in vivo (Leonard et al, 238 1998;Sotillos et al., 2013;Sotillos et al., 2018) rescued the gonadal phenotypes to a 239 better extent than the wild type protein ( Fig. 4C and H). This may be due to the 240 overexpression of a functional GAP protein resulting toxic, not allowing to appreciate the 241 full rescue of the StART-mediated function, a phenomenon that has been described 242 previously (Hendrick & Olayioye, 2019;Holeiter et al, 2012). This does not happen in Cv-243 c GAPmut nor in DLC3 WT , which may have a less efficient GAP function in a Drosophila 244 environment than Cv-c. 245 Moreover, analysis of embryos homozygous for cv-c 7 , an allele which carries that exact 246 GAP mutation in the endogenous gene (Denholm et al., 2005), showed normal testis 247 ( Fig. 5A and B), suggesting that Cv-c function in the gonad is not mediated through its 248 Cv-c in the testis gonadal mesoderm ( Supplementary Fig. S3). 254 activation, our experiments suggest that in the gonad these proteins do not require the 256 GAP function. To confirm that the testis phenotype is not due to Rho1 over-activation we 257 analysed if cv-c M62 gonad mutant phenotypes can be rescued by a Rho1 mutation. As 258 previously described, in Rho1 homozygous mutants the gonad precursor germ cell 259 migration is less efficient, giving rise to smaller male and female gonads (Kunwar et al, 260 2003). However, in Rho1 mutants, the cells reaching the gonads become ensheathed 261 and coalesce to form stable testes as in the wild type ( Fig. 5G and H). In These results show that the StART domain function is required for human and 267 Drosophila testis formation strongly supporting that the dysgenic gonad defects 268 observed in patients are caused by the DLC3 StART mutation and demonstrating that 269 Cv-c has a GAP-independent function that requires the StART domain. testes. In addition, we observe that extruded germ cells have almost no E-Cad on their 281 membranes ( Fig. 6B and B' arrowheads). We observe analogous, abnormal β-catenin 282 ( Fig. 3) localisation inside the testes, indicating that the relationship between the germ 283 cells and the surrounding somatic cells is not well established or poorly sustained (Fig.  284 embryos, localises to the cell extensions produced by the interstitial mesodermal cells 287 that ensheath the germ cells ( Fig. 6C and C') (Jenkins et al., 2003). In cv-c mutant testes 288 we find that Neurotactin expression is almost absent around the germ cells inside the 289 gonad ( Fig. 6D and D'). Accordingly, interstitial somatic gonad mesodermal cells 290 (labelled by the Traffic jam (Tj) antibody), which in the wild type gonad can be detected 291 distributed between GCs ( Fig. 6E and E'), were frequently displaced to the periphery in 292 cv-c mutant testes ( Fig. 6F and F'). 293 Finally, we studied the gonad integrity in cv-c mutants. Wild type testes are surrounded 294 by a Perlecan rich extracellular matrix located between the pigment cells and the 295 interstitial mesodermal cells that can be detected using a GFP insertion in the Perlecan 296 gene ( Fig. 6G and G'). In cv-c C524 mutants, the pigment cell layer and the ECM matrix 297 are discontinuous with the extruded germ cells locating where the matrix gaps appeared 298 this work, using Drosophila as a model, we confirmed the implication of DLC3 as a novel 317 DSD gene required for testis determination through the action of its StART domain. 318 DLC3 had been found to be mutated in two 46, XY DSD siblings presenting gonadal dysgenesis but no experimental evidence confirming causality had been provided 320 (Ilaslan et al., 2018). In this study we report a third DSD patient with a mutation in DLC 321 supporting the involvement of this gene in gonad development. First described in human 322 myeloid cells, DLC3 loss of expression was found in primary tumours from different 323 tissues (Durkin et al, 2007). This multidomain protein forms part of a RhoGAP conserved 324 family containing an N-terminus SAM domain followed by a serine-rich region, a catalytic 325 GAP domain, and a StART domain (reviewed in (Braun & Olayioye, 2015)). There are 326 also alternative isoforms lacking the SAM domain such as DLC3- (Durkin et al., 2007). 327 Despite the recent advances identifying its structure and spatial subcellular location 328 Our modelling analysis predicts that the S993N DLC3 mutation affects the Ω1-loop 356 structure of the StART domain. Ω-loops play multiple roles in protein function, often 357 related to ligand binding, stability, and folding (Fetrow, 1995). This loop is conserved in with the appearance of "blebs", suggest that the settling-down switch has not been 415 During the compaction stage, the Drosophila somatic cell's E-cad locates into thin 418 membrane extensions that surround each germ cell, which also expresses E-cad 419 (Jenkins et al., 2003). In our study, we observed that E-cadherin and β-catenin were 420 abnormally distributed in cv-c mutants, suggesting alterations of the connectivity 421 between PGCs and SGPs. Interestingly, this phenomenon was observed even under 422 Rho1 reduction, reinforcing the idea of a role of Cv-c in a Rho1-independent mechanism 423 that promotes the stabilization of AJ after gonad coalescence. The distance between the Cα atoms of the residues were computed every 100 ps and 493 the distribution was obtained using Kernel Density Estimation (KDE). Fig. 1C was 494 rendered using Visual Molecular Dynamics (VMD) (Humphrey et al, 1996). and a cut-off distance of 1.1 nm. A cutoff scheme was used for the vdW terms, with a 506 cut-off distance of 1.1 nm and Verlet cut-off scheme for the potential-shift (Verlet, 1967).  (Srinivasan et al, 2021). 512 between the protein and the membrane, computed with the gmx mindist tool in 514 GROMACS. Membrane-interacting residues were computed every 500 ps using an in-515 house script with the following protocol: a residue was considered to interact with the 516 membrane if the distance between any bead of the residue and any lipid bead was lower 517 than or equal to 0.5 nm. For each residue, we counted the instances of residue-518 membrane interaction during the trajectory, summed this value over all the replicas, and 519 computed a corresponding normalized value. Figures. 2A and 2D were rendered using 520 Visual Molecular Dynamics (VMD) (Humphrey et al., 1996). To generate UAS-Myc-DLC3 S993N a XhoI DLC3 WT fragment was subcloned into 542 pBlueScript (DLC3-Xho-pBs). DLC3-Xho-pBs was used as a template to mutate Ser at position 993 to Asn of human DLC3 (CCDS48134.1) using Pfu Polymerase. 544 The following primers were used: 545 Forward: 5'-TGTACCACTATGTCACCGACA-A-CATGGCACC-3' 546 Reverse: 5'-TGGGGTGCCATG-T-TGTCGGTGACATAGTG-3' 547 After PCR reaction DNA was incubated with DpnI during an hour at 37º to digest the 548 methylated template DNA and transformed. Clones were sequenced by standard 549 methods. 550  (purple) and anti-Eya and anti-β-catenin to label the msSGPs and the membranes 870 ensheathing the germ cells respectively (grey in lower panels). Scale bar: 10 µm Fisher-871 test; ns, p > 0.05; **p<0.001; ****p < 0.0001. 872