A Point Mutation in p190A RhoGAP Affects Ciliogenesis and Leads to Glomerulocystic Kidney Defects

Rho family GTPases act as molecular switches regulating actin cytoskeleton dynamics. Attenuation of their signaling capacity is provided by GTPase-activating proteins (GAPs), including p190A, that promote the intrinsic GTPase activity of Rho proteins. In the current study we have performed a small-scale ENU mutagenesis screen and identified a novel loss of function allele of the p190A gene Arhgap35, which introduces a Leu1396 to Gln substitution in the GAP domain. This results in decreased GAP activity for the prototypical Rho-family members, RhoA and Rac1, likely due to disrupted ordering of the Rho binding surface. Consequently, Arhgap35-deficient animals exhibit hypoplastic and glomerulocystic kidneys. Investigation into the cystic phenotype shows that p190A is required for appropriate primary cilium formation in renal nephrons. P190A specifically localizes to the base of the cilia to permit axoneme elongation, which requires a functional GAP domain. Pharmacological manipulations further reveal that inhibition of either Rho kinase (ROCK) or F-actin polymerization is able to rescue the ciliogenesis defects observed upon loss of p190A activity. We propose a model in which p190A acts as a modulator of Rho GTPases in a localized area around the cilia to permit the dynamic actin rearrangement required for cilia elongation. Together, our results establish an unexpected link between Rho GTPase regulation, ciliogenesis and glomerulocystic kidney disease.


Introduction
The dynamic regulation of actin polymerization is crucial for cell physiology and closely associated with cellular processes such as migration, survival, polarization and cytokinesis [1,2,3]. Among the main regulators of actin dynamics are the Rho family of GTPases, including RhoA, Rac1 and Cdc42, which act as molecular switches that cycle between the active, GTP-bound and inactive, GDP-bound forms [1,4,5,6]. The critical regulation of this cycle is provided by three classes of molecules: guanine nucleotide exchange factors (GEFs) exchange GDP for GTP, thereby activating Rho-family proteins, whereas GTPase activating proteins (GAPs) accelerate the intrinsic GTP hydrolytic activity of the Rho-family to attenuate signaling [6,7,8]. In addition, guanine nucleotide dissociation inhibitors (GDIs) bind to Rho GTPases and keep them in their inactive state, thereby preventing downstream effector activation. In spite of their extensive characterization in vitro and in cell culture models, the role of RhoGAPs is still poorly understood during animal development and associated diseases.
The RhoGAP protein p190A (Arhgap35, Grlf1) is a potent regulator of Rho GTPases. It is composed of a N-terminal GTPase domain necessary for its activity, a series of conserved FF domains, and a C-terminal RhoGAP domain [9,10,11]. p190A RhoGAP activity has been observed in vitro against the prototypical Rho GTPases, namely Rac1, Cdc42 and RhoA, with the strongest activity on the latter [12,13,14]. In line with this observation, the cellular activity of p190A seems primarily directed against RhoA [13]. Gene inactivation studies in the mouse identified an important role for p190A in neural tube closure, eye morphogenesis and mammary gland branching morphogenesis, differentiation and cell adhesion [15,16,17,18]. Little is known, however, about the cellular role of p190A in tissue morphogenesis, notably in the urogenital system.
In higher vertebrates, the formation of the definitive kidneys and urinary tracts commences with the budding and branching of the ureter tree by signals from the adjacent metanephric mesenchyme [19,20,21]. Conversely, nephron formation is initiated by signals from the ureter tip to induce metanephric mesenchyme condensation and epithelialization into renal vesicles, which proceeds through comma and s-shaped bodies [22]. Invasion of endothelial cells in the cleft of the s-shaped body results in the formation of the vascular and mesangial aspects of the glomerulus, while the epithelial component of the s-shaped body gives rise to nephron tubules, Bowman's capsule and the podocyte lineage that surround the vasculature and act as a filtration barrier [22,23]. In the adult kidney Rho family GTPases are required to maintain the glomerular slit barrier formed by podocytes. Alterations in the activity of RhoA, Rac1 and Cdc42 in differentiated podocytes leads to proteinuria and focal segmental glomerulosclerosis associated with foot process effacement [24,25,26,27,28,29]. Despite the relatively well-characterized role of the Rho proteins in adult glomerular maintenance, little is known about Rho GTPases and their regulators during nephron development.
Polycystic kidney disease is one of the most common renal pathologies [30]. While the underlying mechanism of cyst formation remains somewhat unclear, pathogenesis is often associated with defects in ciliogenesis or cilia function [31,32]. Primary cilia are microtubulebased cell surface projections that serve as signalling platforms for a number of developmentally important pathways [33]. Initiation of ciliogenesis requires vesicular trafficking of the mother centriole to, and subsequent docking with, the plasma membrane [34,35]. The docked centriole, known as the basal body, nucleates elongation of the ciliary axoneme by serving as a scaffold for ciliary-cargo containing vesicles [34,35]. Elongation of the axoneme occurs via microtubule trafficking by intraflagellar proteins attached to kinesin and dynein motors [36,37]. Interestingly, recent evidence suggests that regulation of actin cytoskeletal dynamics is required for the appropriate docking of the basal body and subsequent trafficking of ciliary vesicles [38,39,40,41,42,43].
Glomerulocystic disease is a subtype of polycystic kidneys that exhibit primarily glomerular cysts, characterized by severe dilation of Bowman's capsule and progressive glomerular atrophy [44]. It manifests in both familial and sporadic forms, and often occurs as a component of other genetic syndromes [44]. There are very few mouse models of glomerulocystic kidney disease [45,46], and although the two reported models implicate loss of primary cilia in cyst formation, the regulatory pathways involved are unclear. Here we performed a small-scale Nethyl-n-nitrosourea (ENU) mutagenesis screen [47] to identify new regulators of urogenital system development and disease. We identify a mutant strain with hypoplastic and glomerulocystic kidneys, resulting from a loss-of-function mutation in p190A RhoGAP (Arhgap35). We further demonstrate a crucial role for p190A in modulating actin dynamics during ciliogenesis.

Results
To uncover novel regulators of mammalian urogenital system development, we undertook a recessive ENU mutagenesis approach in which a total of 28 G1 C57BL/6J males (progeny of ENU treated G0 males) were out-crossed to wild type C3H/HeNCrl females to generate G2 mice (S1 Fig). Phenotypic screening was performed on E17.5 embryos from G2 females mated to their respective G1 males. Of the G1 lines tested, line D34 consistently harboured severe kidney defects, including hypoplastic kidneys with glomerular cysts (hypodysplastic kidneys) without associated ureter obstruction and a low penetrance of duplex urogenital systems (Figs 1A, 2C, 2I and 2I'). Interestingly, approximately half of embryos with urogenital system anomalies also exhibited neural tube closure defects, including exencephaly and spina bifida ( Fig  1A). Homozygous D34 mutation results in perinatal lethality, with few mutant animals surviving to weaning age.
Single nucleotide polymorphism (SNP)-based linkage analysis was undertaken to identify the genetic locus associated with kidney hypodysplasia in affected embryos. The characterization of 76 samples identified a single peak of approximately 18 cM on chromosome 7 (rs3675839-rs31924991) (Fig 1B). Additional linkage analysis was performed by restriction length fragment polymorphism (RFLP) on genomic DNA from 346 embryos using markers rs31200925 and rs31924991. This analysis revealed a strong association between hypodysplastic kidneys with the proximal region of chromosome 7 (Fig 1C). Given that there are over 170 genes contained within this chromosomal region, we performed whole exome sequencing on 5 affected embryos to narrow down possible causative mutations. We found a single candidate mutation (A to T) in Arhgap35 (p190A RhoGAP gene), which was located within the 18cM interval and occurred in all affected embryos. To further investigate the association between the altered allele and the kidney phenotype, we Sanger-sequenced Arhgap35 in 59 additional embryos. As expected, the vast majority of affected kidneys were homozygous for the altered Arhgap35 allele (Fig 1D) suggesting that it is likely to be causal in the renal abnormalities of line D34.

Genetic validation of Arhgap35 deficiency in glomerulocystic phenotype
To address whether the renal anomalies observed in the D34 line (hereafter referred to as Arh-gap35 D34 ) result from a loss-of-function mutation in the Arhgap35 gene, we obtained the previously generated Arhgap35 tmjset1 mutant line [16] (referred to as Arhgap35mice) and generated homozygous mutant and compound heterozygous mutant animals for both alleles. Gross histological characterization of the urogenital system of Arhgap35 D34/D34 kidneys at embryonic day 17.5 (E17.5) revealed a defined renal cortex and medulla, similar to control animals (Fig  2A, 2B, 2C, 2G, 2H and 2I). In line with this observation, differentiation markers showed no significant differences between control and mutant kidneys (S2A, S2B and S2C Fig). Remarkably, however, serial sectioning revealed that all Arhgap35 D34/D34 kidneys contained cystic glomeruli accompanied by occasional dilation of the nearby proximal tubule (Figs 2I, 2I', S2D and S3D). Arhgap35 -/kidneys exhibited similarly penetrant renal malformations not previously described, including kidney hypoplasia and agenesis, as well as cystic glomeruli (Fig 2E, 2K, 2K' and 2M and Table 1). As previously reported, these animals also harboured neural tube  Table 1) and a range of renal abnormalities including hypoplasia, agenesis and glomerular cysts at a similar frequency to both Arhgap35 D34/D34 and Arh-gap35 -/embryos ( Fig 2F, 2L, 2L' and 2M, and Table 1). Together, these results demonstrate that the D34 line is a novel Arhgap35 mutant strain and highlight the importance of p190A activity during kidney development.

Arhgap35 D34 is a loss-of-function allele
The phenotypic similarity and non-complementation between the Arhgap35and Arhgap35 D34 alleles suggested that Arhgap35 D34 is a loss-of-function mutation. The ENU-induced A to T nucleotide exchange alters amino acid leucine 1396 to glutamine in the RhoGAP domain of p190A ( Fig 3A). To gain insight into the functional significance of L1396, we first examined the interspecies conservation of the residue. Interestingly, L1396 is identical across species from mice to frogs and is located in a highly conserved region (Fig 3A), suggesting an important functional requirement for this domain. We next investigated the potential structural impact of the mutation on p190A function by examining the crystallized human p190A GAP- domain structure (PDB: 3FK2) ( Fig 3B and 3B'). Notably, L1396 is located on the interior of a tightly packed α-helical interface directly adjacent to the ordered surface formed by the highly conserved RhoA binding residues K1322, and N1395 and the catalytic arginine (R1284) required for p190A function [8,9,48,49] (Fig 3B). Modeling the conversion of L1396 to the bulkier glutamine reveals substantial steric clash with surrounding amino acids, irrespective of the position of the glutamine side chain (Fig 3B'). Based on this structural modeling, it is likely that the L1396Q substitution distorts the binding site on p190A, reducing its ability to act as an effective GAP.
As our genetic studies suggested a loss-of-function mutation, we predicted that it would result in reduced GTP hydrolysis by RhoA and/or Rac1, relative to wild type p190A. To examine this possibility we expressed either the wild-type GAP (p190A GAP WT ) or D34 mutant GAP (p190A GAP L1396Q ) domains as GST-fusion proteins for in vitro GAP activity assays. As expected, each small GTPase possessed little intrinsic GTP-hydrolytic activity alone, while addition of p190A GAP WT significantly enhanced the rate of GTP hydrolysis ( Fig 3C and 3D), reflecting an acceleration of Rho GTPase activity in line with previously published results [9,10,13]. Interestingly, p190A GAP WT showed a distinct preference for RhoA over Rac1, reflected in the increase in specific activity (Fig 3C and 3D). Strikingly, the mutant p190A GAP L1396Q protein severely impaired phosphate release for both substrates (Fig 3C and 3D), reflecting a significant reduction in GAP activity. In line with this observation, GTP-bound RhoA is enhanced in mouse embryonic fibroblasts derived from Arhgap35 D34/D34 animals, as compared to controls (S3D Fig). Together these results confirm Arhgap35 D34 as a loss-of-function allele likely resulting from a disruption of the p190A RhoGAP binding interface.

P190A is required for ciliogenesis
Given the relative paucity of mouse models of glomerulocystic disease, we decided to further characterize the molecular basis of glomerular cyst formation in Arhgap35-deficient kidneys. We first examined the expression of the podocyte marker Wt1, which showed a normal Arhgap35 +/x Arhgap35 +/- Arhgap35 +/D34 x Arhgap35 +/- The similar penetrance of these malformations in Arhgap35 D34/D34 , in an independent Arhgap35 mutant line, and in compound heterozygous animals confirms D34 as a Arhgap35 mutant strain and suggests a strong hypomorphic allele.
progression of glomerular development and overall density in Arhgap35 D34/D34 animals (S4 Fig). This result, together with the normal capillary loops, mesangial cells and podocytes in both cystic and normal glomeruli observed by H&E (Fig 2), indicated that the cystic phenotype is unlikely to result from defects in early glomerular development. We next asked whether Arh-gap35 mutation compromised the structure of the glomerular tuft. For this, we performed immunofluorescence staining for podocyte specific proteins, including nephrin, synaptopodin and Wt1, as well as the structural markers laminin, ZO-1 and F-actin ( Fig 4A and S4 Fig). We did not observe significantly altered expression or localization of any of the markers examined in non-cystic mutant glomeruli, indicating that prior to cyst formation Arhgap35 D34/D34 glomeruli are largely indistinguishable from controls. In addition, glomerular tufts of Arh-gap35 D34/D34 animals showed evidence of F-actin rich foot processes despite significant dilation of Bowman's capsule (Figs 4A and S4). Together, these results suggest that the glomerulocystic phenotype of Arhgap35 D34/D34 kidneys does not primarily result from glomerular morphogenesis defects. We next assessed the expression of Arhgap35 by in situ hybridization in developing kidneys. At E17.5, Arhgap35 expression was present but low in the glomerulus, which contrasted with its high expression levels in the proximal tubule (S5 Fig). This result raised the possibility that the glomerulocystic phenotype may be initiated in proximal tubules rather than in the glomerular epithelium.
Cystic renal phenotypes are often associated with defective ciliogenesis, resulting in dysregulation of downstream signaling pathways important for maintaining kidney epithelial homeostasis [31,50,51]. To investigate the possibility of a ciliogenesis defect, we visualized cilia by immunofluorescence against acetylated-α-tubulin and Arl13b, which mark the elongated axoneme, in combination with markers of different nephron segments. Whereas primary cilia appeared normal in glomeruli (S4E and S4F Fig), both the number of ciliated cells and the average length of the cilia were drastically reduced in the proximal tubules of Arhgap35 D34/D34 kidneys (Fig 4B, 4C and 4D). Interestingly, cilia formation appeared impaired in both s-shaped bodies and E14.5 proximal tubules of Arhgap35 D34/D34 mutant animals (S6 Fig), suggesting that defective proximal tubule ciliogenesis occurs during nephrogenesis, prior to the onset of cystic disease.
We next examined whether these ciliogenesis defects were secondary to abnormal basal body docking to the plasma membrane. Interestingly, γ-tubulin staining revealed appropriate positioning of the basal body in both control and Arhgap35 D34/D34 proximal tubules ( Fig 4B). Hence, the defect in cilia number and length observed in Arhgap35-deficient proximal tubules result from impairment in axoneme extension rather than basal body migration or positioning. P190A acts at the primary cilium to regulate cilia elongation To understand the mechanistic relationship between p190A and cilium formation, we turned to the more tractable system of primary mouse embryonic fibroblasts (MEFs), which reproducibly form cilia upon serum withdrawal. Whereas approximately 60% of wild type MEFs formed cilia, less than 20% of Arhgap35 D34/D34 MEFs produced elongated cilia and the cilia that did form were on average half the length of wild type (Figs 5A, 5E and S7A). Additionally, expression of a GFP-p190A WT full length construct, but not GFP-p190A L1396Q , in Arhgap35 mutant MEFs restored their ability to form elongated cilia to wild-type parameters (Fig 5D and 5E), further confirming that the cilium defect results from a deficiency in p190A activity. In line with the phenotype observed in proximal tubule cells, immunofluorescent staining of control and mutant MEFs with γ-tubulin showed a normal number of basal body-containing cells, which were docked appropriately at the plasma membrane (Fig 5A, 5B and 5C).
We next examined whether p190A was present at the cilium. Introduction of the GFP-p190A WT construct in wild type MEFs revealed an enrichment of p190A co-localizing with γtubulin at the base of cilia in 12% of cells (Fig 5F), consistent with a direct, and transient role for p190A in cilia formation. Interestingly, this enrichment was not observed in MEFs expressing the D34-mutant form of p190A (GFP-p190A L1396Q ) (Fig 5F), suggesting that a functional RhoGAP domain is necessary for p190A recruitment to basal bodies. In line with this result, p190A was also absent from the basal body in MEFs expressing a catalytically dead p190A variant (GFP-p190A R1284A ) (Fig 5F). We further generated deletion constructs fused to GFP, which identified the C-terminal GAP domain as sufficient for basal body localization (Fig 5F). Together, these results demonstrate the causal role of the p190A L1396Q mutation in the cilia phenotype of Arhgap35-deficient cells.

P190A regulates actin cytoskeletal dynamics to permit cilia elongation
Recently, the impact of the actin cytoskeleton on ciliogenesis has been garnering attention, notably on the balance between free versus filamentous actin in promoting cilia elongation [38,39,40,41,42,43,52]. Given that the p190A-L1396Q mutation results in a loss of GAP activity toward RhoA and Rac1 (Figs 3C, 3D and S3D), we predicted that their downstream pathways would be activated, potentially perturbing actin cytoskeletal dynamics. To evaluate the relative contribution of Rho and Rac signaling to ciliogenesis in the presence or absence of p190A function, we used small molecule inhibitors of Rho-kinases 1/2 (ROCK1/2) and Rac1. Remarkably, at concentrations that do not alter the onset of ciliogenesis in control MEFs, inhibition of ROCK1/2, using either Y27632 [53] or GSK 429286A [54], completely restored cilia numbers and cilia length of Arhgap35 D34/D34 MEFs to wild-type levels (Figs 6, S7B and S7C). In contrast, blocking Rac1 activation using NSC23766 [55] did not have a significant effect on overall cilia numbers but allowed for better elongation in forming cilia (Fig 6).
The complete rescue of ciliogenesis by ROCK1/2 inhibition pointed to an important role for RhoA signalling in the Arhgap35-deficient phenotype. Surprisingly, however, phalloidinlabelled F-actin formed normal cytoskeleton networks in both gain and loss of function analyses of p190A function (S7H Fig), suggesting that aberrations in ciliogenesis do not result from global alterations in stress fibre formation. Accordingly, inhibition of myosin light chain II activity by Blebbistatin [56] failed to rescue both the number and length of cilia in Arh-gap35 D34/D34 MEFs, indicating that in this system RhoA-Rock1/2 ectopic activation does not act through the actomyosin pathway (S7D and S7E Fig). To validate that the ciliogenesis defect was nonetheless a result of excess F-actin formation, we quantified cilium number and length in the presence of Cytochalasin D or Latrunculin A, two small molecules that prevent F-actin polymerization [41,57]. Strikingly, Cytochalasin D or Latrunculin A treatment restored both cilium number and length of Arhgap35 D34/D34 MEFs to wild type levels (Figs 6, S7F and S7G), suggesting that p190A may normally dampen Rho and Rac-dependent F-actin polymerization to permit cilia elongation. To determine whether reduced local actin polymerization also underlies the ciliogenesis defect observed in vivo, we examined the localization of Rac1, RhoA and F-actin in proximal tubules of control and Arhgap35 D34/D34 kidneys by immunofluorescent staining. In both wild type and Arhgap35 D34/D34 proximal tubules Rac1 and RhoA are enriched at the basal bodies to similar degrees (Fig 7A-7D). In contrast, Arhgap35 D34/D34 animals exhibited a marked increase in the amount of F-actin polymerized around the basal body compared to control animals (Fig 7E and 7F). Together, these results point to a permissive role for p190A specifically at the basal body to promote ciliogenesis through a local dampening of Rho GTPase-dependent F-actin polymerization (Fig 7G).

Discussion
Using an ENU mutagenesis approach, we have identified a new allele of the p190A RhoGAP gene Arhgap35, which leads to defects in renal morphogenesis. In particular, we found that Arhgap35 mutation results in hypoplastic kidneys with a fully penetrant glomerulocystic phenotype associated with defective cilia elongation. Together these results establish a strong and unexpected link between the modulation of Rho family GTPases by p190A RhoGAP, ciliogenesis and renal developmental diseases.
Using a combination of classical single nucleotide polymorphism (SNP) mapping and exome sequencing approaches, we have identified a point mutation in Arhgap35. The causal link between this sequence variant and the developmental phenotype was demonstrated both by genetic non-complementation with an independent Arhgap35 mutant strain [16], and by rescue of the ciliary defect in Arhgap35-deficient MEFs by reintroduction of wild type but not mutant p190A. The similar penetrance of neural tube and renal defects between Arhgap35 D34 and the previous reports on Arhgap35 mutant alleles [16,18] identifies D34 as a null or a strong hypomorph allele. In line with this, functional testing of the p190A L1396Q variant in vitro led to a significant loss of activity on the prototypical Rho family GTPases. Based on the structural and mechanistic similarity between Rho family GTPases and their cognate GAPs, three critical residues (R1284, K1322, N1395) have been identified that together define a shallow Rho-binding pocket on p190A comprised of two α-helices and the catalytic arginine loop [8,48,49]. Interestingly, the hydrophobic L1396 residue is buried between these helices. Modeling the mutant glutamine reveals steric clashes with the surrounding amino acids and likely results in disruption of this α-helical interface, and subsequent distortion of the Rho-binding surface. Furthermore, GTPase activity requires stabilization of the catalytic loop by N1395 to optimally position GTP for hydrolytic attack in the transition state intermediate [8,49]. It seems likely that substitution of the longer and hydrophilic glutamine side chain at position 1396, adjacent to N1395, affects this critical interaction point, thus lowering the efficiency of catalytic loop stabilization and resulting in the markedly reduced activity observed in vitro.
Unexpectedly, we found that the reduction in p190A GAP activity perturbs cilia elongation in both kidney proximal tubule cells and mouse embryonic fibroblasts, associated with dysregulated Rho GTPase signaling. The pharmacological manipulations of Arhgap35-deficient cells suggest that both Rac1/2 and RhoA may be substrates for p190A during cilia formation; however, while Rac inhibition only partially rescues ciliogenesis in the absence of p190A activity, ROCK inhibition completely rescues both cilia number and length. Together, these results raise the intriguing possibility that p190A plays a role at the docked basal body to downregulate active RhoA (and possibly Rac1) to allow the localized actin cytoskeleton rearrangement required for cilia elongation (Fig 7G). In support of this, full-length p190A and the RhoGAP domain alone are enriched at the base of the cilia, while p190A L1396Q and p190A R1284A that lacks RhoGAP activity fails to be recruited. Additionally, we find that the F-actin inhibitory drugs Cytochalasin D and Latrunculin A are able to rescue the defects in cilia number and length seen in the absence of functional p190A. Furthermore, as predicted, we observe increased F-actin around the basal bodies of Arhgap35 D34/D34 proximal tubules. These results confirm that p190A deficiency results in ectopic actin polymerization, leading to compromised cilia elongation. This finding is in line with other studies of ciliogenesis which identified Factin modifying proteins, including regulators of actin branch nucleation and filament stability, that restrict primary cilia elongation [38,39,40,41,42,43,52,58]. Notably, however, the upstream Rho GTPases involved remained unclear. Our study provides the first genetic demonstration that the negative regulation of Rho GTPases at the basal body is required for primary cilium elongation in multiple cell types.
Interestingly, cilia have been observed on the ureteric bud epithelium during branching morphogenesis, raising the possibility that earlier cilia defects may also be causative in the renal hypoplasia observed. Cilia elongation is also impaired in early nephrogenic structures upon loss of p190A activity; however, the functional relevance of this observation awaits future studies.
Together our results establish a link between Rho GTPase regulation, ciliogenesis and glomerular cystic defects. It provides a molecular and cell biological basis for understanding glomerulocystic kidney disease, a form of renal cystic disease that is less frequent and less well understood than polycystic kidney disease that primarily affects tubular nephrons and collecting ducts [59]. The finding that the ciliogenesis defect of Arhgap35-deficient kidneys primarily affects proximal tubule cells, which correlates with the site of high Arhgap35 transcriptional expression, raises the interesting possibility that glomerulocystic kidney disease can, in some instances, be secondary to proximal tubule deficiencies. This work additionally identifies Rho GTPases and their associated GAPs, and GEFs, as candidates in the etiology of glomerulocystic kidney disease.

Ethics statement
Experiments for this study were approved by the McGill Animal Care Committee, and were conducted in compliance with the Canadian Council of Animal Care ethical guidelines for animal experiments.

ENU mutagenesis screen
N-ethyl-N-nitrosourea (ENU, 90 mg/Kg) was administered as 3 weekly doses by intraperitoneal injections to 8-week old male C57BL/6J mice (Jackson Laboratories) to induce germline mutations as described previously [60]. G0 founder males were outcrossed to wild type C3H/ HeNrl females (Charles River Laboratories) to establish 30 G1 lines for screening. Phenotypes in the D34 line were analyzed between generations four and twelve, with establishment of two independent Arhgap35 D34 lines at generation seven by breeding Arhgap35 +/D34 males wild type females. Kidneys were identified as hypoplastic if they were <75% of littermate size. Restriction fragment length polymorphism (RFLP) analysis was used to genotype animals by PCR-amplifying 600bp fragments around the single nucleotide polymorphisms (SNPs) rs31200925 and rs31924991. Mice carrying the C3H/HeNCrl (wild-type) allele at these SNPs have restriction sites for Bgl II and Hind III, respectively, while animals with the C57BL/6J (mutated) allele do not. Restriction digest of the PCR product results in fragments of 600bp (C57BL/6J) and/or 400bp and 200bp (C3H/HeNCrl) allowing for genotype determination. When appropriate, Arhgap35 was also PCR-amplified for Sanger sequencing. All primers used are provided in S1 Table. Mutation identification SNP Array screening was performed on the Low-density Mouse Linkage Panel (LD panel, Golden Gate technology, Illumina). R-QTL mapping was performed using the R software platform [61]. Whole Exome sequencing was performed with Illumina Nextera Exome capture technology on 5 affected samples, 1 littermate control, 1 wild type C3H/HeNrl and 1 C57BL/6J sample. SNP Array and exome sequencing experiments were conducted by the McGill University and Genome Quebec Innovation Centre (Montreal, Canada).

P190A expression constructs
Full length p190A-pEGFP-C1 mammalian expression vector was generated as described [10]. The GFP-p190A L1396Q and GFP-p190A R1284A constructs were generated from the full-length wild-type construct by site directed mutagenesis. For the GFP-p190A domain constructs, the appropriate domains were PCR amplified from the full-length wild-type construct. For in vitro GAP assays, the GAP domain (nt 4417-5143) was PCR amplified from the either the full length GFP-p190A WT or GFP-p190A L1396Q constructs using specific primers containing BglII and EcoRI restriction sites, which allowed for cloning in the BamHI and EcoRI sites of the pGEX-6P1 bacterial expression vector. All constructs were sequence verified. All primers used are provided in S1 Table. In vitro p190A GAP assay Recombinant GST-tagged p190A GAP protein was produced in BL21(DE3) bacteria by conventional methods, and purified using Glutathione-agarose beads (Life Technologies). Cleavage of the GST tag, and elution of free p190A GAP protein, was achieved using PreScission protease (GenScript). Protein was quantified so that 1.5ug of GAP was added per GAP assay reaction. The in vitro GAP assay (Cytoskeleton) was performed according to manufacturer's instructions.

In vivo active RhoA assay
The amount of activated RhoA (RhoA-GTP) was assessed using the Rho Activation Assay Biochem Kit (Cytoskeleton) following manufacturer's instructions. Briefly, control and mutant MEFs were grown to 70% confluence, and serum starved for 16hr to assess the basal level of RhoA activation. Cells were lysed on ice, clarified by centrifugation and snap frozen. 400ug of each lysate was used for pulldown assay using the Rho binding domain from rhotekin conjugated to sepharose beads. The amount of activated RhoA was assessed by western blot, following standard procedures. The primary antibodies used were mouse anti-RhoA (1:500, Cytoskeleton) and rabbit anti-GAPDH (1:5000, Abgent). Secondary antibodies used were HRP-conjugated anti-mouse (1:10000, GE Healthcare) and HRP-conjugated anti-rabbit (1:10000, Cell Signaling Technology).

Structural analysis
To visualize the potential impact of the L1396Q substitution on the structure of p190A Rho-GAP, the Pymol software program was used to image the human p190A GAP domain (PDB: 3FK2). Substitution of glutamine for leucine 1396 was done using the mutagenesis function of Pymol, and the rotamer position of the glutamine side chain and potential steric clashes with surrounding amino acids were assessed for all possible conformations.

Histology, immunofluorescence and in situ hybridization
Dissected urogenital systems (UGS) were formalin-fixed, paraffin embedded and processed according to standard methods for Hematoxylin and Eosin staining at the McGill University Histology Facility. Section immunofluorescence and in situ hybridizations were performed on dissected UGS, fixed in 4% paraformaldehyde and cryo-sectioned as described previously [62]. Probes for Gata3, Ncc, and Wt1 have been described [63,64,65]. The Arhgap35 probe was generated from E12.5 whole embryo cDNA using specific primers containing a T7 promoter sequence (S1 Table). Probe was transcribed following standard procedures using T7 RNA polymerase and DIG-labeled dNTPs (Roche). For immunohistochemical stainings, sections were permeabilized with 0.3% Triton X-100 in PBS for 10 minutes, blocked with DAKO protein block (serum-free), and incubated with the following primary antibodies or Lectin overnight at 4°C: rabbit anti-laminin (

Imaging and statistical analysis
Images were acquired on a Confocor LSM 510 META Axiovert 200M inverted microscope at the Advanced BioImaging Facility of McGill University (Montreal, Canada). Cilia length and basal body positioning were determined using the measure function of the ImageJ image analysis program [66] from confocal projections. Statistical analyses were performed as either oneway ANOVA or unpaired, two-tailed Student's t-tests (as indicated) using GraphPad Prism 6 software. Data are presented as mean ± SEM; p<0.05 was considered significant.
Supporting Information S1 Fig. Breeding scheme of a recessive ENU mutagenesis screen. N-ethyl-N-nitrosurea (ENU)-mutagenized male C57BL/6J mice were outcrossed to wild-type C3H/HeNCrl females to produce first generation (G1) offspring. Sons from this cross were outbred to wild-type C3H/HeNCrl females to produce the second generation (G2), from which daughters were backcrossed to their father to recover recessive mutations in the third generation (G3). G3 progeny were dissected at embryonic day 17.5 (E17.5) for gross characterization that identified the D34 line. two-tailed Student's t-test). (C-D) Immunofluorescence staining for podocyte markers (Wt1, Synaptopodin) and structural markers (phalloidin, laminin) show no obvious misorganization in Arhgap35 D34/D34 glomeruli, either pre-or post-cyst formation (arrowheads). (E) Section immunofluorescence for cilia (acetylated α-tubulin) and the podocyte lineage (podocalyxin) shows normal cilia in Bowman's capsule cells (visualized with DAPI). (F) Quantification of cilia number per glomerulus from (F) shows no significant difference between control and Arh-gap35 D34/D34 animals, irrespective of glomerular dilation (unpaired, two-tailed Student's t-test). Immunofluorescence for phalloidin and acetylated α-tubulin in control and Arhgap35 D34/D34 mouse embryonic fibroblasts shows a normal F-actin cytoskeleton in ciliated cells. Staining for phalloidin in mouse embryonic fibroblasts overexpressing full length p190A WT -GFP shows no noticeable effect on F-actin organization. (E-F) Treatment with the myosin II inhibitor Blebbistatin does not rescue the defects in cilia number (D) and cilia length (E) in Arhgap35 D34/D34 mouse embryonic fibroblasts. (G-H) Treatment with the actin polymerization inhibitor Latrunculin A rescues the defects in in cilia number (G) and cilia length (H) in Arhgap35 D34/D34 mouse embryonic fibroblasts. Ã p<0.05, ÃÃ p<0.01, ÃÃÃ p<0.005, ÃÃÃÃ p<0.001 (one-way ANOVA) (TIF) S1 Table. Polymerase chain reaction primers used for genotyping and cloning of Arhgap35. (XLSX)