Abstract

Loss–of-function mutations in PAK3 contribute to non-syndromic X-linked intellectual disability (NS-XLID) by affecting dendritic spine density and morphology. Linkage analysis in a three-generation family with affected males showing ID, agenesis of corpus callosum, cerebellar hypoplasia, microcephaly and ichthyosis, revealed a candidate disease locus in Xq21.33q24 encompassing over 280 genes. Subsequent to sequencing all coding exons of the X chromosome, we identified a single novel variant within the linkage region, affecting a conserved codon of PAK3. Biochemical studies showed that, similar to previous NS-XLID-associated lesions, the predicted amino acid substitution (Lys389Asn) abolished the kinase activity of PAK3. In addition, the introduced residue conferred a dominant-negative function to the protein that drives the syndromic phenotype. Using a combination of in vitro and in vivo studies in zebrafish embryos, we show that PAK3N389 escapes its physiologic degradation and is able to perturb MAPK signaling via an uncontrolled kinase-independent function, which in turn leads to alterations of cerebral and craniofacial structures in vivo. Our data expand the spectrum of phenotypes associated with PAK3 mutations, characterize a novel mechanism resulting in a dual molecular effect of the same mutation with a complex PAK3 functional deregulation and provide evidence for a direct functional impact of aberrant PAK3 function on MAPK signaling.

INTRODUCTION

Mutations affecting the X-chromosome account for ∼8–12% of intellectual disability (ID) seen in males (1), with >100 disease genes identified thus far (2). Historically, X-linked ID (XLID) has been divided into syndromic and non-syndromic (NS-XLID) forms on the basis of the presence of clinical features other than cognitive impairment, such as dysmorphic, metabolic and/or behavioral characteristics (3). While this classification remains useful, it is a coarse representation of the clinical spectrum of XLID-associated phenotypes (4,5), as highlighted by the finding of mutations in a number of these disease genes that can drive both isolated and syndromic ID (2). For most genes and mutations, the underlying molecular basis is unclear, as are the genetic of biochemical drivers of variable expressivity that typifies this group of disorders (6,7).

Among XLID-associated genes, PAK3 has been reported to be mutated in five families with isolated cognitive impairment (8–12). PAK3 encodes a member of the p21-activated family of serine/threonine kinases (PAKs), which are downstream effectors for Rac/Cdc42 Rho GTPases. These proteins are involved in multiple intracellular processes, such as proliferation, cell cycle progression and cytoskeleton remodeling (13), and are known to control multiple signal transduction pathways, including the MAPK cascade (14,15). PAK3 is expressed predominantly in different regions of the brain (16), where its kinase activity contributes to spine morphogenesis and synaptic plasticity (17) by regulating actin cytoskeleton dynamics (18). NS-XLID mutations has been reported to cause the loss of PAK3 catalytic functions (19), which alters dendritic spine morphology and density, without producing overt defects in neuronal structures in mice (20).While multiple lines of evidence support a major role of PAK3 catalytic activity in controlling signal flow, its kinase-independent functions are also critical; enzymatically inactive PAK3 is able to regulate different cellular activities, such as transcription and postsynaptic transmission, by acting as scaffolding protein through its N-terminal, non-catalytic domain (21,22).

Here, we report a three-generation family showing X-linked recurrence of a previously undescribed neurocutaneous phenotype, caused by a novel missense mutation in PAK3 that both abolish the catalytic activity of the kinase and confer an additional dominant-negative function, thus leading to not only ID but also skin defects and brain structural anomalies.

RESULTS

Clinical evaluation revealed ID associated with brain structural anomalies, skin alterations and dysmorphic features in the affected males

At 3 months, the proband (III-2) presented his first epileptic episode, characterized by fixed gaze and asymmetric generalized clonic movements with dissociated autonomic manifestations in the right (pallor) and left (cyanosis) hemisoma. At 12 months of age, neurological examination showed psychomotor development delay, generalized hypotonia, hyperreflexia that was more evident in the legs, and afinalistic movements of hands and head. Brain MRI revealed agenesis of the corpus callosum (Fig. 1A and B), enlargement of occipital horns of lateral ventricles and cerebellar hypoplasia (Fig. 1A) involving both vermis and hemispheres, with mild enlargement of the cisterna magna and size reduction of the brainstem.

Figure 1.

Clinical features of the proband. (A and B) Sagittal (T1 weighted) and coronal (T2 weighted) sections of brain MRI, with arrows indicating agenesis of corpus callosum (thick in A), cerebellar hypoplasia (thin in A) associated with a widen cisterna magna and typical Probst bundles (B). (C) Proband's facial phenotype with left epicanthus, bilateral ptosis, convergent squint, depressed nasal bridge, broad nasal tip and long ears. (D and E) Severe ichthyosis with visible scales on the thorax (also showing pectus excavatum) and one arm.

Figure 1.

Clinical features of the proband. (A and B) Sagittal (T1 weighted) and coronal (T2 weighted) sections of brain MRI, with arrows indicating agenesis of corpus callosum (thick in A), cerebellar hypoplasia (thin in A) associated with a widen cisterna magna and typical Probst bundles (B). (C) Proband's facial phenotype with left epicanthus, bilateral ptosis, convergent squint, depressed nasal bridge, broad nasal tip and long ears. (D and E) Severe ichthyosis with visible scales on the thorax (also showing pectus excavatum) and one arm.

At the age of 2 years and 9 months, dysmorphic features and microcephaly (occipital frontal circumference =47.8 cm, <3rd centile for height age) were appreciated. Patient's height was 97.5 cm (>75th centile), while weight was 13.9 kg (25th centile for height age). Physical examination showed left epicanthus, bilateral ptosis, convergent squint, depressed nasal bridge, broad nasal tip, high palate, gingival overgrowth, mild microretrognathia, long ears, pectus excavatum, sinus pilonidalis, camptodactyly and subtle cutaneous syndactyly of the hands (Fig. 1C and D). Dermatological examination revealed small, white, semi-adherent scales of the trunk and limbs but sparing of flexures (Fig. 1D and E). A skin biopsy provided a histological diagnosis of ichthyosis, with diminished or absent granular layer and mild orthokeratotic hyperkeratosis of the stratum corneum. Electron microscopy showed typical alterations of keratohyalin granules in granular layer cells. The psychomotor development of the patient did not show any significant improvement in cognitive and motor abilities; the child could sit alone but never walked unassisted and did not develop speech. He died at 8 years of age from cardiorespiratory arrest during a prolonged epileptic seizure.

The first cousin of the proband (III-4), deceased at the age of 6 years, was affected by a similar condition, including congenital ichthyosis, psychomotor delay and onset of epilepsy during the first year of life. Similar to the proband, he had dysmorphic features that included microcephaly, epicanthus, mild hypertelorism, low and broad nasal bridge, hypoplastic nostrils, narrow palate, hypoplastic teeth, retrognathia, long ears and short neck, pectus excavatum. He had hypotonia and could never walk unassisted. He could speak a few words. Brain MRI showed hypoplasia of the corpus callosum (with agenesis of the splenium) and cerebellar hypoplasia. Dermatological examination showed medium to large, polygonal, semi-adherent, gray or white scales on trunk and limbs (Supplementary Material, Fig. S1A). Histology showed an expanded stratum corneum and reduced granular cell layer, while electron microscopy revealed small keratohyalin granules in granular layer cells.

The pregnancy of the fetus III-5 was interrupted at the 21st week of gestation, after ultrasound examination could not detect the corpus callosum. Postmortem examination showed that the fetus was a male with complete agenesis of septum pellucidum and corpus callosum, indicated further by the presence of Probst bundles at histological examination (Supplementary Material, Fig. S1B), that also revealed a hypoplastic left piramidal tract and poorly convoluted bulbar olives. Skin histology showed hyperkeratosis and desquamation of the epidermal layer compatible with congenital ichthyosis (Supplementary Material, Fig. S1C).

Obligate carrier females (individuals I-2, II-2 and II-3) only showed mild ichthyosis limited to the extensor surfaces of limbs, compatible with a balanced X-chromosome inactivation pattern (Supplementary Material, Table S1).

A novel missense mutation in PAK3 segregated with the disease trait

In the proband, karyotyping and 244K array-based comparative genomic hybridization (Agilent Technologies, Santa Clara, CA, USA) did not reveal any chromosomal abnormalities. We therefore performed linkage analysis. After genotyping of 18 ABI PRISM X-chromosome markers, LOD scores of ≥0 at recombination fraction = 0 for three consecutive markers (DXS990, DXS1106 and DXS8055) identified a candidate region of ∼40 cM between DXS986 and DXS1001 on Xq21.1–q24. We added a further 12 markers from the Marshfield Clinic Genetic Map and obtained an average inter-marker distance of ∼5 cM along the whole chromosome and of 2.5 cM across the candidate region. Linkage analysis using this denser map was performed including genotypes from an autoptic specimen of individual III-4 and from a chorionic villus sample of fetus III-5. The maximum possible attainable LOD score with this pedigree was 2.05 and this value was reached by markers between DXS8077 and DXS1001, defining a single candidate region of 26.8 cM on Xq21.3–q24 (Fig. 2A; Supplementary Material, Table S2).

Figure 2.

Haplotype analysis and genetic findings in the proband. (A) The family tree shows X-linked segregation of the disease, with the two affected males (III-2 and III-4), the fetus (III-5) and the obligate carriers (I-2, II-2 and II-3). The names of all genotyped markers on chromosome X are indicated on the left with their relative genetic distance in centiMorgan. Red boxes define the disease-associated region in Xq21.33q24 between DXS8077 and DXS1001 markers. (B) The screen-shot of Integrative Genomics Viewer (IGV) shows part of the 195/195 NGS reads supporting the hemizygous c.1167G > T nucleotidic change. (C) The electropherogram of PAK3 sequence shows the mutated nucleotide c.1167G > T (arrow).

Figure 2.

Haplotype analysis and genetic findings in the proband. (A) The family tree shows X-linked segregation of the disease, with the two affected males (III-2 and III-4), the fetus (III-5) and the obligate carriers (I-2, II-2 and II-3). The names of all genotyped markers on chromosome X are indicated on the left with their relative genetic distance in centiMorgan. Red boxes define the disease-associated region in Xq21.33q24 between DXS8077 and DXS1001 markers. (B) The screen-shot of Integrative Genomics Viewer (IGV) shows part of the 195/195 NGS reads supporting the hemizygous c.1167G > T nucleotidic change. (C) The electropherogram of PAK3 sequence shows the mutated nucleotide c.1167G > T (arrow).

The critical region did not encompass any gene(s) known to be associated with a neurocutaneous condition. To identify the driver mutation across the 287 genes (NCBI annotation release 104) within the region, we performed massive parallel sequencing of the entire protein coding region of the X chromosome in the proband (III-2). Three variants were retained after the filtration process (Supplementary Material, Table S3), but only the missense change (c.1167G > T; Lys389Asn) affecting PAK3 (NM_002578; Fig. 2B) mapped within the linkage interval. This variant, which was confirmed by Sanger sequencing (Fig. 2C), affected a highly conserved residue (Supplementary Material, Fig. S2) located in the catalytic core of the PAK3 kinase domain. The allele also co-segregated with the disease in the family, and was absent in 450 chromosomes of population-matched control individuals.

PAK3 is expressed in keratinocytes

The expression of PAK3 in the brain has been established (23,24) and correlates with ID associated with both published mutations and the nucleotide substitution found in our family. However, because of the presence of ichthyosis in our patients, we tested PAK3 expression in keratinocytes from breast and abdomen skin samples of control individuals. Western blot analysis with an anti-PAK3 monoclonal antibody confirmed the absence of expression of the kinase in primary fibroblasts and showed its expression in keratinocytes (Fig. 3A).

Figure 3.

In vitro and in silico analyses on endogenous and transfected wild-type and mutant PAK3 proteins. (A) Western blot reveals endogenous PAK3 protein in keratinocytes from abdomen and breast skin biopsies of control individuals. Ectopic expression of PAK3 is shown in V5-tagged PAK3 transiently transfected COS-1 cells as positive control. Fibroblasts and COS-1 cells do not physiologically express the gene, and were used as negative controls. NT, not transfected. WB, western blot. (B) Kinase assay results for PAK3K389N and NS-XLID mutants. COS-1 cells were cotransfected with wild-type PAK3 or mutants, in presence of either dominant negative (N17) or constitutively active (L61) YFP-Cdc42 vectors. 32P-MBP is the radioactive phosphorylated form of the MBP and it was detected by autoradiography (upper panel). Expression levels of transfected proteins were evaluated by western blotting, using anti-V5 and anti-Cdc42 antibodies (lower panels). β-Actin was used as loading control. KA, kinase assay; WB, western blot; ev, empty vector; NT, not transfected. (C) Ribbon representation of the three-dimensional model of PAK3. The structure of the PAK3 catalytic domain is composed of two subdomains, a smaller N-terminal lobe composed almost entirely by β-strands (dark blue in the lower part) and a larger C-terminal lobe formed almost uniquely by α-helices (light blue). The activation and the ATP-binding-loops are represented in yellow and red, respectively. The Thr421 residue, which in many kinases is phosphorylated in the active enzyme, is also represented. The catalytic loop is located between the two lobes, with Asp387 and Lys389 side chains represented in ball and stick. The mutated Asn389 residue is also shown, in overlap with the side chain of the wild-type Lys389. Oxygen, carbon and nitrogen atoms are represented in red, gray and blue, respectively. (D) Cycloheximide assay. After 16 h of incubation with cycloheximide, only the band corresponding to Lys389Asn PAK3 has an intensity similar to that obtained from untreated cells, while the other mutants show a significantly reduced band intensity.

Figure 3.

In vitro and in silico analyses on endogenous and transfected wild-type and mutant PAK3 proteins. (A) Western blot reveals endogenous PAK3 protein in keratinocytes from abdomen and breast skin biopsies of control individuals. Ectopic expression of PAK3 is shown in V5-tagged PAK3 transiently transfected COS-1 cells as positive control. Fibroblasts and COS-1 cells do not physiologically express the gene, and were used as negative controls. NT, not transfected. WB, western blot. (B) Kinase assay results for PAK3K389N and NS-XLID mutants. COS-1 cells were cotransfected with wild-type PAK3 or mutants, in presence of either dominant negative (N17) or constitutively active (L61) YFP-Cdc42 vectors. 32P-MBP is the radioactive phosphorylated form of the MBP and it was detected by autoradiography (upper panel). Expression levels of transfected proteins were evaluated by western blotting, using anti-V5 and anti-Cdc42 antibodies (lower panels). β-Actin was used as loading control. KA, kinase assay; WB, western blot; ev, empty vector; NT, not transfected. (C) Ribbon representation of the three-dimensional model of PAK3. The structure of the PAK3 catalytic domain is composed of two subdomains, a smaller N-terminal lobe composed almost entirely by β-strands (dark blue in the lower part) and a larger C-terminal lobe formed almost uniquely by α-helices (light blue). The activation and the ATP-binding-loops are represented in yellow and red, respectively. The Thr421 residue, which in many kinases is phosphorylated in the active enzyme, is also represented. The catalytic loop is located between the two lobes, with Asp387 and Lys389 side chains represented in ball and stick. The mutated Asn389 residue is also shown, in overlap with the side chain of the wild-type Lys389. Oxygen, carbon and nitrogen atoms are represented in red, gray and blue, respectively. (D) Cycloheximide assay. After 16 h of incubation with cycloheximide, only the band corresponding to Lys389Asn PAK3 has an intensity similar to that obtained from untreated cells, while the other mutants show a significantly reduced band intensity.

PAK3K389N is catalytically impaired and promotes MAPK pathway activation in vitro

Immunofluorescence confocal microscopy analysis showed that the disease-causing PAK3 mutant exhibited a subcellular localization comparable with that of the wild-type protein and did not affect the organization of actin cytoskeleton (Supplementary Material, Fig. S3). Next, we determined the kinase activity of PAK3K389N and that of NS-XLID mutants (PAK3R67C, PAK3A365E, PAK3R419X and PAK3W446S). We therefore co-transfected COS-1 cells with wild-type PAK3 or each mutant, in presence of either a dominant negative (Cdc42T17N) or constitutively active (Cdc42Q61L) Cdc42 isoform, and performed in vitro kinase assays (25), using myelin basic protein (MBP) as substrate. As shown in Figure 3B, the catalytic activity was found to be impaired in PAK3K389N and in three NS-XLID mutants (PAK3A365E, PAK3R419X and PAK3W446S), which were not able to phosphorylate MBP when stimulated by the constitutively active form of Cdc42, the major activator of PAK3. Differently from the other mutants, PAK3R67C, which was demonstrated to have reduced binding affinity for Cdc42 (19), showed kinase activity, possibly because overexpression of Cdc42 bypassed the reduced binding affinity, as pointed out by Kreis et al. (2007). We also observed that the kinase function of activated wild-type PAK3 is attenuated, but this is due to a physiologic degradation induced by PAK3 autophosphorylation through a feedback mechanism that mediates signaling switch off (26). As expected, PAK3 mutants, deprived of the kinase activity and the ability to induce this regulatory process, were not degraded (Fig. 3B).

To explore the possible impact of PAK3 mutations on MAPK signaling, we performed ERK phosphorylation assays in COS-1 cells co-transfected with wild-type or mutant PAK3, HA-tagged ERK and the constitutively active Cdc42Q61L isoform. We observed a significantly enhanced phosphorylation of ERK in cells expressing each of the catalytically inactive mutants, in contrast to the wild-type protein, which is degraded rapidly (Supplementary Material, Fig. S4). Taken together, these findings suggest that NS-XLID and Lys389Asn mutations abolish the enzymatic activity of PAK3 without altering the ability of the kinase to stimulate the MAPK pathway, that is probably mediated by a kinase-independent function.

PAK3K389N maintains a stable structure and persists within the cell

To explore possible differential perturbing effects of the Lys389Asn substitution and NS-XLID-associated mutations on PAK3 structure, we performed in silico structural modeling analyses. PAK3 exhibits high-sequence identity with PAK1 (78% overall and 96% within the kinase domain, amino acids 283–534), whose three-dimensional structure has been elucidated by crystallography (PDB id: 1f3m) (27). We therefore threaded a molecular model of PAK3 based on the PAK1 structure.

The PAK3 catalytic domain is composed of two subdomains (Fig. 3C): a smaller N-terminal lobe composed almost entirely by β-strands, and a larger C-terminal lobe formed almost uniquely by α-helices. The catalytic loop, spanning amino acid positions 387–392, is located between the two lobes and is characterized by the conserved consensus motif Asp-X-Lys-X-X-Asn. The ATP molecule is bound by a conserved Gly-rich loop, spanning positions 290–297 and is folded in front of the catalytic loop. By analogy to other kinases, the Asp residue is predicted to play the role of proton acceptor of the hydroxyl group of the substrate Ser/Thr residue undergoing phosphorylation, whereas Lys389 (mutated here to Asn) is likely important in the neutralization of the negative charge of the γ phosphate of the ATP molecule. The substitution of the basic Lys389 with the neutral Asn is therefore likely to abolish or diminish the efficiency with which PAK3 transfers the ATP γ phosphate to the Ser/Thr side chain of the target proteins. On the other hand, being the position of the mutated residue at the protein surface, the Lys389Asn change is not expected to influence the structural stability of the kinase C-terminal lobe. Among the NS-XLID mutations, Ala365Glu and Trp446Ser affect the larger C-teminal lobe of the kinase catalytic domain. Both Ala365 and Trp446side chains are buried in the protein core (Fig. 3C); the substitution of these hydrophobic residues with the hydrophilic Ser and Glu side chains is thus likely to destabilize the protein structure. An even more dramatic destabilization of the enzyme structure is expected to occur in the truncated PAK3R419X mutant, which lacks part of the activation loop and a large portion of the C-terminal lobe, with patches of the hydrophobic core residues exposed to the aqueous environment.

To confirm these in silico structural predictions, we assayed the stability of wild-type PAK3 and each mutant by inhibiting protein neosynthesis by cycloheximide treatment. We found that all the NS-XLID-associated mutants were degraded by 16 h, consistent with the notion they were incorrectly folded, unstable proteins. By contrast, the PAK3K389N mutant was stable, with no appreciable evidence of degradation throughout our time course study (Fig. 3D).

PAK3K389N induces craniofacial and brain alterations and promotes MAPK pathway activation in vivo

To assess whether the mutations tested earlier perturb the function of PAK3 in vivo and to test the postulated pathomechanism, we next turned to zebrafish embryos. Although a direct PAK3 ortholog does not exist in zebrafish, depletion of the related kinase PAK1 (74% homology) has been shown to cause convergent extension (CE) movements defects resulting in shortened body length in zebrafish embryos (28). In addition, given the craniofacial defects in our patients, we assayed for jaw formation defects, as determined by measuring the distance between the Meckel and ceratohyal cartilage (29). In comparison to wild-type PAK3, neither PAK3K389N nor the four NS-XLID mutants were able to rescue the altered phenotype produced by pak1 suppression (Fig. 4A and B; Supplementary Material, Fig. S5), confirming the expected (from prior work and our in vitro experiments) loss-of-function effect for all five mutants. Strikingly, however, overexpression of PAK3K389N, but not the four NS-XLID-associated alleles, caused significantly more severe CE and jaw defects (Figs 4C and 5A) compared with wild type, suggesting that Lys389Asn mutation might also trigger a dominant-negative effect for PAK3.

Figure 4.

Functional changes of K389N and NSID mutations during CE development. (A) Left panel: CE defects in pak1 morphants. Arrowheads show the body angle and bars show the width of somites. Expression of wild-type hPAK3 but not K389N and NS-XLID mutants can rescue the CE defects caused by pak1 suppression. ***P < 0.0001 in comparison with wild type. (B) Somitic development in zebrafish embryos. As indicated in the graph, morpholino and hPAK3 mRNA were injected to the zebrafish embryos at 1–4 cell stage. The images were captured with Nikon and the width versus length ratio was obtained by ImageJ. Depletion of pak1 results in the shift of width to length ratio, which can be rescued by wild type but not mutant RNA. Bar graph is represented as mean ± SEM. Class I and Class II embryos were merged for statistical analysis (*P < 0.05; **P < 0.01, in comparison with wild type). (C) Overexpression of K389N PAK3 significantly alters the CE severity.

Figure 4.

Functional changes of K389N and NSID mutations during CE development. (A) Left panel: CE defects in pak1 morphants. Arrowheads show the body angle and bars show the width of somites. Expression of wild-type hPAK3 but not K389N and NS-XLID mutants can rescue the CE defects caused by pak1 suppression. ***P < 0.0001 in comparison with wild type. (B) Somitic development in zebrafish embryos. As indicated in the graph, morpholino and hPAK3 mRNA were injected to the zebrafish embryos at 1–4 cell stage. The images were captured with Nikon and the width versus length ratio was obtained by ImageJ. Depletion of pak1 results in the shift of width to length ratio, which can be rescued by wild type but not mutant RNA. Bar graph is represented as mean ± SEM. Class I and Class II embryos were merged for statistical analysis (*P < 0.05; **P < 0.01, in comparison with wild type). (C) Overexpression of K389N PAK3 significantly alters the CE severity.

Guided by the clinical presentation of our patients, we broadened our phenotypic analyses further. Because of the cerebellar hypoplasia documented in the proband and his cousin, and the agenesis of corpus callosum reported in the two affected males and the fetus, we also examined the central nervous system (CNS), focusing on the cerebellum and the optic tectum, a site of early neurogenesis, as described (30). Similar to our CE and craniofacial assays, overexpression of PAK3K389N caused significantly more severe CNS defects, including a reduction in the size of optic tectum, cerebellar morphology defects, and loss of axonal tracks connecting the two hemispheres, as visualized by staining with an antibody against acetylated tubulin (Fig. 5B–D; Supplementary Material, Fig. S6). Finally, to test whether aberrant MAPK signaling might be driving these phenotypes, we injected embryos with the dominant human PAK3 allele and the MEK1 inhibitor PD684161 and asked whether that manipulation could rescue the observed neuroanatomical phenotypes induced by expressing dominant human PAK3 mutant. We observed complete rescue of the optic tecta and hemisphere connectivity phenotypes and significant rescue of the cerebellar disorganization defects (Fig. 5E).

Figure 5.

Expression of K389N PAK3 causes more severe phenotype in jaw layout and CNS structure. (A) Left panel represents the ventral view of Alcian blue stained jaw cartilage. Red line indicate the distance between MK and CH arches. Bar graph showed in right panel is represented as mean ± SEM. In comparison with wild type, ** and * indicate P < 0.01 and P < 0.05, respectively. (B) CNS pattern in zebrafish embryos expressing wild-type or mutant PAK3. Single red asterisk denotes the optic tectum; double red asterisks denote the connection between hemispheres; and red dashed box denotes cerebellum. (C and D) Expression of K389N disrupts cerebellum development and the connection between hemispheres. ***P < 0.0001 compared with wild type. (E) Co-injection of MEK inhibitor ameliorates K389N-associated CNS defects, including size of optic tecta, cerebellum development and the connection between hemispheres. Bar graph is represented as mean ± SEM, and P-value of each experiment is as indicated in the figure.

Figure 5.

Expression of K389N PAK3 causes more severe phenotype in jaw layout and CNS structure. (A) Left panel represents the ventral view of Alcian blue stained jaw cartilage. Red line indicate the distance between MK and CH arches. Bar graph showed in right panel is represented as mean ± SEM. In comparison with wild type, ** and * indicate P < 0.01 and P < 0.05, respectively. (B) CNS pattern in zebrafish embryos expressing wild-type or mutant PAK3. Single red asterisk denotes the optic tectum; double red asterisks denote the connection between hemispheres; and red dashed box denotes cerebellum. (C and D) Expression of K389N disrupts cerebellum development and the connection between hemispheres. ***P < 0.0001 compared with wild type. (E) Co-injection of MEK inhibitor ameliorates K389N-associated CNS defects, including size of optic tecta, cerebellum development and the connection between hemispheres. Bar graph is represented as mean ± SEM, and P-value of each experiment is as indicated in the figure.

Taken together, these findings demonstrate in vivo that Lys389Asn has a loss-of-function effect as the NS-XLID mutations, but it also confers a dominant-negative function to PAK3, which deregulates MAPK signaling, causing CNS and facial alterations.

DISCUSSION

We identified a novel missense (Lys389Asn) mutation in PAK3 responsible for a previously unrecognized neurocutaneous syndrome, in which ID is associated with agenesis of corpus callosum, cerebellar hypoplasia, seizure, microcephaly and ichthyosis. We obtained evidence that NS-XLID PAK3 mutants deprived of kinase activity are unstable and subsequently eliminated by the cell, consistent with the notion that PAK3 loss of function causes exclusively ID. This is in line with the PAK3 knockout mouse model that shows only cognitive deficiency, neither ichthyosis nor other neurological features (20). In contrast, PAK3K389N mutant is structurally stable and escapes physiologic turnover, being enzymatically inactive. The resulting persistence within the cell, coupled to the retention of a kinase-independent function, leads to enhanced stimulation of the MAPK signaling pathway and to in vivo alterations of cerebral and craniofacial structures. These findings are intriguing, since the phenotype associated with the Lys389Asn change shows partial clinical overlap (i.e. mental retardation, seizures, hypotonia, agenesis of corpus callosum, brainstem atrophy, ichthyosis and facial dysmorphism) with cardiofaciocutaneous syndrome (CFCS), a clinical entity belonging to the Rasopathies, a family of developmental disorders caused by altered MAPK signaling. Screening of all PAK3 coding exons in 60 patients with clinical features fitting or suggestive of CFCS, however, did not reveal any functionally relevant change, suggesting that dominant-negative-acting PAK3 mutations are likely to underlie a homogeneous, distinctive and rare phenotype.

Our data suggest a complex genetic contribution for PAK3 with markedly different clinical outcomes: while isolated loss of PAK3 kinase activity causes NS-XLID, a dual molecular effect is required to produce the syndromic neurocutaneous phenotype. We speculate that similar mechanisms might contribute to the phenotypic diversity of other X-linked syndromes and we predict that mutations in other genes known to be involved in non-syndromic phenotypes will also manifest syndromic pathology.

Another conclusion supported by our findings is that PAK3 biologic role is not limited to the fine regulation of dendritic spine dynamics and synaptic plasticity, but it might extend to the development of brain structures, such as corpus callosum and cerebellum, and to skin homeostasis.

Finally, our data highlight how a detailed understanding of the effect of deleterious alleles will be required to develop effective therapeutics treatments; in the present example, the dual activity of Lys389Asn is paradoxical to a predicted loss-of-function effect (by virtue of the mapping of the allele in the kinase domain) and suggests that pharmacological attenuation of MAPK signaling (here achieved with an MEK1 inhibitor) might be of clinical value.

MATERIALS AND METHODS

Clinical assessment

Medical and family history, with particular regard to neurological and cutaneous involvement, was obtained. Physical examination of the proband III-2 and of his cousin III-4 included assessment of height, weight and dysmorphic features. The neurological examination was especially focused on the evaluation of the electric activity of the brain by EEG and on the detection of cerebral malformations by MRI. The dermatological evaluation was supported by skin biopsy and was extended to carrier females. After abortion, the fetus III-5 was subjected to autopsy to confirm the brain anomaly previously detected by ultrasound scan and to reveal possible skin alterations.

Informed consent and permission for picture publication was obtained for individuals included in the study and molecular analyses were performed in this family as part of the CHERISH project, approved by the Ethical Review Board of Policlinico S.Orsola-Malpighi (Bologna).

Classical and molecular cytogenetics

Karyotyping was performed on G-banded chromosomes obtained from a peripheral blood sample of proband III-2 using standard procedures. The proband was also investigated by array-CGH at a resolution of ∼20 kb with the Agilent 244A kit (Agilent Technologies) to search for small chromosomal imbalances. This analysis was performed following the manufacturer's protocol.

Linkage analysis

Initial linkage analysis involved 13 samples from individuals I-2, II-1, II-2, II-3, II-4, II-5, II-6, II-7, III-1, II-2, III-3, III-6, III-7 which were genotyped for the 18 X-chromosome-specific microsatellite markers in the Panel 28 of the ABI PRISM Linkage-Mapping Set v.2 (Life Technologies, Carisbad, CA, USA). Microsatellite markers were amplified by PCR according to the manufacturer's protocol and PCR products were electrophoresed and detected on the 48-capillary ABI 3730 DNA analyzer (Life Technologies). Genotypes were determined with the support of the GeneMapper v.3.5 program (Life Technologies). Genotype inconsistencies were checked with PedStats (31). LOD scores were calculated with Merlin (32) modeling the disease as a completely penetrant X-linked recessive trait with a disease allele frequency of 1/1000 and equal marker allele frequencies.

Subsequently, 12 markers from the Marshfield Clinic Genetic Map (http://research.marshfieldclinic.org/genetics/GeneticResearch/compMaps.asp) were added to the map and DNA samples from III-4 to III-5 included in the analysis.

Next generation sequencing-based detection of variants affecting the X-chromosome-coding sequences

Using the Sure Select Human X-chromosome Panel kit by Agilent (Agilent Technologies) and a Illumina HiSeq2000 system (Illumina Inc., San Diego, CA, USA), we captured and sequenced the coding sequences of the proband's non-pseudoautosomal X-chromosome regions. We obtained >29M 100 bp paired-end reads. We aligned the reads against the hg19 reference genome with BWA (33). After removing duplicate reads with Picard tool's MarkDuplicates (http://picard.sourceforge.net/), a 55.6% of the total reads mapped to the X chromosome giving an average 292X coverage on the Agilent target. In order to exclude the possibility that a candidate variant was missed due to poor coverage, we checked for targeted regions having a mean coverage <10X. We detected 492 over the 7662 Agilent target regions (6.40%) showing such a low coverage. Of these, the 79% overlapped with Segmental Duplications according to the UCSC GenomicSuperDups database (http://genome.ucsc.edu/), and the 100% overlapped with regions that were poorly covered in all the other in-house X-chromosome exome samples. We therefore concluded that the probability of missing a candidate variant due to insufficient depth at the variant site was negligible. After performing local realignment around indels and base quality score recalibration using GATK (34), we called single nucleotide variants (SNVs) and small insertions/deletions (indels) using GATK Unified Genotyper, and filtered out the variants by quality using GATK VariantFiltrationWalker with parameters: -clusterWindowSize 10-filterExpression ‘MQ0>=4 && ((MQ0/(1.0 * DP))>0.1)’-filterExpression ‘QUAL<30||QD<5.0||HRun>5||SB>−0.10’ for SNVs and -filterExpression ‘MQ0 >=4 && ((MQ0/(1.0 * DP)) > 0.1)’ -filterExpression ‘SB >=−1.0’-filterExpression ‘QUAL<10’ for indels. We obtained 947 variants which passed quality criteria, and annotated them against the NCBI RefSeq (http://www.ncbi.nlm.nih.gov/RefSeq/) and UCSC KnownGene (http://genome.ucsc.edu/) databases with ANNOVAR (35). We retained for downstream analysis all non-synonymous SNVs, indels across the coding sequences and variants located within 5 bp from the intron–exon junctions of coding exons. Non-‘novel’ variants [i.e. not in publicly available database as dbSNP135 (http://www.ncbi.nlm.nih.gov/projects/SNP/), 1000genomes (http://www.1000genomes.org/), NHLBI Exome Sequencing project (http://evs.gs.washington.edu/EVS/) or in-house X-chromosome exomes] were filtered out. Variants within segmental duplications according to the UCSC GenomicSuperDups database were also excluded as likely false-positive mutation signals.

The variant found through this analysis process has been submitted to the Leiden Open Variation Database (http://databases.lovd.nl/shared/variants/0000022950).

Sequencing analysis of PAK3

Specific PCR with primers flanking exon 15 of PAK3 (NM_002578) and Sanger sequencing were performed to confirm the mutation found through NGS analysis on genomic DNA (∼20 ng) of the proband, to check the segregation within the family and to analyze over 450 × chromosomes derived from Italian control individuals.

All PAK3 exons were amplified and analyzed in 60 patients with CFC-like clinical features without a molecular diagnosis.

Standard PCR conditions were as follows: 95°C for 5 min, 40 cycles of 95°C for 30 s, 56°C for 30 s, 72°C for 30 s, followed by an extension step at 72°C for 7 min. PCR products were sequenced with an ABI Prism Big Dye Terminator v1.1 Cycle Sequencing Kit (Life Technologies) as instructed by the manufacturer and run on the 48-capillary ABI 3730 DNA analyzer. Sequences were analyzed with Sequencher software 4.9 (Gene Codes Corporation, Ann Arbor, MI, USA). The conservation of the Lys389 residue was assessed through Multiz Alignment tool from UCSC Genome Browser.

Cloning and mutagenesis

The wild-type cDNA of human PAK3 isoform (GenBank: BC152761.1), purchased from Life Technologies, was cloned into a pcDNA3.1/V5-His-TOPO (Sigma-Aldrich, St. Louis, MO, USA) expression vector. The mutant constructs carrying the amino acid substitution p.Lys389Asn and those associated with non-syndromic mental retardation (p.Arg67Cys, p.Ala365Glu and p.Trp446Ser) were generated by site-directed mutagenesis using the QuikChange XL kit (Agilent Technologies) in accordance with the manufacturer's protocol. A deleted cDNA clone containing the first 418 codons in frame with the V5 epitope was created to study the p.Arg419Stop mutation, detected in a family with isolated mental retardation. All generated constructs were checked by direct sequencing.

Cell cultures and transfection

Both primary fibroblast and keratinocyte cells were obtained from human skin biopsies and grown in high-glucose Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FCS, 2 mml-glutamine and 10 U/ml penicillin/streptomycin (Sigma-Aldrich).

COS-1 cells were grown in DMEM medium supplemented with 10% FCS, 2 mml-glutamine and 10 U/ml penicillin/streptomycin (Sigma-Aldrich). Each of the V5-tagged PAK3 mutants, wild-type protein or empty vector (pcDNA3.1/V5-His, 3 µg) was cotransfected with either a constitutively active (Leu61, L61) or a dominant-negative (Asn17, N17) pAc-YFP-Cdc42 mutant and HA-ERK for ERK phosphorylation experiments, using Fugene 6 (Roche, Basel, Switzerland). Twenty-four hours after transfection, cells were washed twice in cold PBS, and lysed (50 mm TRIS–Cl, pH 7.4, 150 mm NaCl, 0.25% NaDeoxycholate, 1 mm EDTA, 1 mm EGTA, 1% Triton X-100, 0.5% IGEPAL, 1 mm Na3VO4, 2 mm NaF, protease inhibitor cocktail). Cell lysates were cleared, and processed for immunoprecipitation and/or immunoblotting.

Immunofluorescence analysis and confocal laser scanning microscopy

Analysis was performed as previously described (36). In brief, COS1 cells were seeded on glass slides, and transfected 24 h after plating with K389N/V5 and WT/myc-tagged PAK3 plasmids. Twenty-four hours following transfection, cells were washed, fixed (4% paraformaldehyde, 20 min, 4°C) and washed again. After 5 min permeabilization with 0.5% Triton X-100, cells were incubated overnight at 4°C with 5 µg/ml mouse anti-V5 and rabbit anti-myc primary antibodies or an equal amount of isotype controls (DAKO). Then, cells were washed, incubated for 1 h at room temperature with 5 µg/ml anti-mouse Alexa Fluor®633 and anti-rabbit Alexa Fluor®568 secondary antibodies (Life Technologies) and washed twice. After 30 min incubation with Alexa Fluor®488-Phalloidin (Life Technologies), coverslips were washed and then mounted on glass slides with 5 µg/ml Hoechst33342 (Life Technologies) nuclear dye in anti-fade reagent (Life Technologies) and analyzed. No labeling was detected in isotype controls. Imaging was performed on an Olympus FV1000 apparatus, utilizing excitation spectral laser lines at 405, 488 and 633 nm. Signals from different fluorescent probes were taken in sequential scanning mode.

PAK3 immunoprecipitation, kinase assay and immunoblotting

Proteins (200 or 350 µg) were immunoprecipitated at 4°C by overnight incubation with 0.5 µg anti-V5 antibody (# R960, Life Technologies) or 1 µg anti-HA antibody (# H3663, Sigma-Aldrich), respectively. Twenty or 35 µl of slurry of protein A/G ultralink resin (#53 132, Thermo Fisher Scientific, Waltham, MA, USA) was then added, and samples were mixed at 4°C for 2 h. Anti-V5 or anti-HA immunocomplexes were washed twice with IP buffer (25 mm Tris–HCl, pH 7.4, 150 mm NaCl, 1 mm EDTA, pH 8, 1 mm EGTA, pH 7.5, 0.1% Triton X-100, 5 mm 2-glycerophosphate, 1 mm Na3VO4, 10 mm NaF and 1:200 protease inhibitor cocktail). In vitro kinase assay was performed as previously described (25). Briefly, anti-V5 immunocomplexes were washed twice with kinase assay buffer (20 mm Hepes, pH 7.5, 10 mm MgCl2, 10 mm 2-glycerophosphate, 1 mm DTT), and reactions were performed by resuspending beads in 40 µl of buffer supplemented with 10 µg of MBP (#M1891, Sigma-Aldrich), 10 µCi of [γ −32P]ATP (3000 Ci/mmol) and 40 µm ATP. Following incubation for 20 min at 30°C, reactions were stopped by the addition of 30 µl of 4× SDS sample buffer, and boiled for 5 min. After centrifugation, supernatants were resolved by 12% SDS–PAGE. Gels were stained with Coomassie Blue, dried and exposed to autoradiography. PAK3 and ERK (cell lysates and immunoprecipitated proteins), and phospho ERK, YFP-Cdc42, α-tubulin and β-actin (cell lysates) expression levels were evaluated by western blotting, using an anti-PAK3 (# 04395, Millipore, Billerica, MA, USA), anti-V5 and anti-HA (see above), anti phospho p44/42 MAPK (ERK1/2) thr202/tyr204 (#9106S, Cell Signaling Technology, Danvers, MA, USA), anti-Cdc42 (#2462, Cell Signaling Technology), α-tubulin (#T4026, Sigma-Aldrich) and anti-β actin (#A5316, Sigma-Aldrich) antibody, respectively. Anti-PAK3 monoclonal antibody (#04395, Millipore, Billerica, MA, USA) was used to reveal the endogenous protein from control keratynocytes lysates. Optical density measurements of western blot bands were performed with the ImageJ software.

For the ERK phosphorylation assay, the significance of the differences in mean values was determined by using the Student's t-test. P < 0.05 was considered significant. Mean + SEM values from three independent experiments of ERK phosphorylation (relative to total ERK) were expressed as fold increase over control cells (COS1 cells cotransfected with pcDNA3.1/V5-His empty vector, Cdc42-L61 and HA-ERK).

In silico structural analysis

Visual analysis was carried out with the molecular graphics suite O (37). The molecular model of PAK3 was built by homology based on PAK1 with the use of the SwissModel web server (38). The obtained model was further energy minimized by using simulated annealing as implemented in CNS (39).

Protein stability assay

Cycloheximide is an inhibitor of cytoplasmic protein synthesis and is used to reveal protein degradation induced by an uncorrected folding that in vitro could be masked by plasmid overexpression.

pcDNA3.1/V5 PAK3 constructs (WT, Lys389Asn, Ala365Glu, Arg419Stop and Trp446Ser) were transfected in HEK293 cells at 60% confluence, seeded in two distinct six-dishes plates and grown in DMEM supplemented with 10% fetal bovine serum, 2 mmol/l l-glutamine, 100 units/ml penicillin and 100 µg/ml streptomycin, using X-tremeGene HP reagent (Roche). Twenty-four hours after transfection, cells of one plate were collected, washed in PBS and stored at −80°C, while cells of the other plate were treated with 50 µg/ml cycloheximide (Sigma-Aldrich) for 16 h, then collected and washed in PBS. Treated and untreated cells were lysed in RIPA buffer (50 mm Tris–HCl, pH 7.5, 150 mm NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate, 1 mm EDTA) with protease (complete mini protein inhibitor coktail, Roche Diagnostics) and phosphatase (phosphostop, Roche) inhibitors, and then centrifugated at 4°C at 13 000 rpm for 30 min. The surnatants were quantified by Lowry method (Bio-Rad, Hercules, CA, USA) and 30 µg of purified proteins were separated by 10% SDS–PAGE (Bio-Rad). PAK3 proteins and β-actin were revealed by primary anti-V5 (Life Technologies) and anti-actin (Sigma-Aldrich) antibodies in a chemiluminescence assay (Life Technologies). Protein stability was assessed comparing band intensity for each PAK3 protein with and without cycloheximide treatment, normalizing for the endogenous β-actin. The results were confirmed in three independent experiments.

Zebrafish model

We obtained the published pak1-MO, and in parallel, we inserted human PAK3 into the pcDNA3.1/V5-His TOPO vector and synthesized mRNAs with the T7mMessage mMachine kit (Life Technologies). We then injected 50–100 embryos at the 1–4 cells stage with morpholino (2.5 ng) and either wild-type or mutant mRNA and scored for CE phenotypes at the mid-somitic stage. Whole mount in situ hybridization with cocktail riboprobes (myoD, krox2 and pax2) was then carried out in 10–12 somites stage embryos, as described previously (40). The yolk sacs of embryos were then removed and flat-mounted with glycerol. Images were captured using Nikon AZ100 and NIS Elements software. Pairwise comparisons to WT for rescue efficiency were determined using a χ2-test using previously established CE morphometrics (shortened anterior–posterior axis, widened notochord and defective eye development) (41–43).

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This work was supported by the European Commission's FP7 program, CHERISH project (grant agreement no. 223692); the Italian Ministry of Health Strategic Program, Genetic Basis of Birth Defects (grant number RFPS-4-631972); the Italian Ministry of Health, Young Investigators Award (grant number GR-2009-1574072); European Research Area Network for research programs on rare diseases (E-Rare) 2009 (NS_EuroNet) and Telethon-Italy (grant number GGP13107). N.K. is a Distinguished Brumley Professor.

ACKNOWLEDGEMENTS

We are grateful to Prof Michele De Luca (Centre for Regenerative Medicine, Modena, Italy) and Dr Emmanuel Lemichez (INSERM U452, Nice, France) who kindly provided skin biopsies for PAK3 expression studies and pAcYFP-Cdc42 plasmids, respectively. We thank Dr Roberta Zuntini for performing X-inactivation analysis and the family for its kind cooperation in the study.

Conflict of Interest statement. None declared.

REFERENCES

1
Ropers
H.H.
Genetics of intellectual disability
Curr. Opin. Genet. Dev.
 , 
2008
, vol. 
18
 (pg. 
241
-
250
)
2
Lubs
H.A.
Stevenson
R.E.
Schwartz
C.E.
Fragile X and X-linked intellectual disability: four decades of discovery
Am. J. Hum. Genet.
 , 
2012
, vol. 
90
 (pg. 
579
-
590
)
3
Gécz
J.
Shoubridge
C.
Corbett
M.
The genetic landscape of intellectual disability arising from chromosome X
Trends Genet.
 , 
2009
, vol. 
25
 (pg. 
308
-
316
)
4
Frints
S.G.
Froyen
G.
Marynen
P.
Fryns
J.P.
X-linked mental retardation: vanishing boundaries between non-specific (MRX) and syndromic (MRXS) forms
Clin. Genet.
 , 
2002
, vol. 
62
 (pg. 
423
-
432
)
5
Chelly
J.
Khelfaoui
M.
Francis
F.
Chérif
B.
Bienvenu
T.
Genetics and pathophysiology of mental retardation
Eur. J. Hum. Genet.
 , 
2006
, vol. 
14
 (pg. 
701
-
713
)
6
Franek
K.J.
Butler
J.
Johnson
J.
Simensen
R.
Friez
M.J.
Bartel
F.
Moss
T.
DuPont
B.
Berry
K.
Bauman
M.
, et al.  . 
Deletion of the immunoglobulin domain of IL1RAPL1 results in nonsyndromic X-linked intellectual disability associated with behavioral problems and mild dysmorphism
Am. J. Med. Genet. A
 , 
2011
, vol. 
155
 (pg. 
1109
-
1114
)
7
Tran Mau-Them
F.
Willems
M.
Albrech
T.B.
Sanchez
E.
Puechberty
J.
Endele
S.
Schneider
A.
Ruiz Pallares
N.
Missirian
C.
Rivier
F.
, et al.  . 
Expanding the phenotype of IQSEC2 mutations: truncating mutations in severe intellectual disability
Eur. J. Hum. Genet.
 , 
2014
, vol. 
22
 (pg. 
289
-
292
)
8
Allen
K.M.
Gleeson
J.G.
Bagrodia
S.
Partington
M.W.
MacMillan
J.C.
Cerione
R.A.
Mulley
J.C.
Walsh
C.A.
PAK3 mutation in nonsyndromic X-linked mental retardation
Nat. Genet.
 , 
1998
, vol. 
20
 (pg. 
25
-
30
)
9
Bienvenu
T.
des Portes
V.
McDonell
N.
Carrié
A.
Zemni
R.
Couvert
P.
Ropers
H.H.
Moraine
C.
van Bokhoven
H.
Fryns
J.P.
, et al.  . 
Missense mutation in PAK3, R67C, causes X-linked nonspecific mental retardation
Am. J. Med. Genet.
 , 
2000
, vol. 
93
 (pg. 
294
-
298
)
10
Gedeon
A.K.
Nelson
J.
Gécz
J.
Mulley
J.C.
X-linked mild non-syndromic mental retardation with neuropsychiatric problems and the missense mutation A365E in PAK3
Am. J. Med. Genet. A
 , 
2003
, vol. 
120
 (pg. 
509
-
517
)
11
Peippo
M.
Koivisto
A.M.
Särkämö
T.
Sipponen
M.
von Koskull
H.
Ylisaukko-oja
T.
Rehnström
K.
Froyen
G.
Ignatius
J.
Järvelä
I.
PAK3 related mental disability: further characterization of the phenotype
Am. J. Med. Genet. A
 , 
2007
, vol. 
143
 (pg. 
2406
-
2416
)
12
Rejeb
I.
Saillour
Y.
Castelnau
L.
Julien
C.
Bienvenu
T.
Taga
P.
Chaabouni
H.
Chelly
J.
Ben Jemaa
L.
Bahi-Buisson
N.
A novel splice mutation in PAK3 gene underlying mental retardation with neuropsychiatric features
Eur. J. Hum. Genet.
 , 
2008
, vol. 
16
 (pg. 
1358
-
1363
)
13
Bokoch
G.M.
Biology of the p21-activated kinases
Annu. Rev. Biochem.
 , 
2003
, vol. 
72
 (pg. 
743
-
781
)
14
Beeser
A.
Jaffer
Z.M.
Hofmann
C.
Chernoff
J.
Role of group A p21-activated kinases in activation of extracellular-regulated kinase by growth factors
J. Biol. Chem.
 , 
2005
, vol. 
280
 (pg. 
36609
-
36615
)
15
Slack-Davis
J.K.
Eblen
S.T.
Zecevic
M.
Boerner
S.A.
Tarcsafalvi
A.
Diaz
H.B.
Marshall
M.S.
Weber
M.J.
Parsons
J.T.
Catling
A.D.
PAK1 phosphorylation of MEK1 regulates fibronectin-stimulated MAPK activation
J. Cell. Biol.
 , 
2003
, vol. 
162
 (pg. 
281
-
291
)
16
Rousseau
V.
Goupille
O.
Morin
N.
Barnier
J.V.
A new constitutively active brain PAK3 isoform displays modified specificities toward Rac and Cdc42 GTPases
J. Biol. Chem.
 , 
2003
, vol. 
278
 (pg. 
3912
-
3920
)
17
Boda
B.
Alberi
S.
Nikonenko
I.
Node-Langlois
R.
Jourdain
P.
Moosmayer
M.
Parisi-Jourdain
L.
Muller
D.
The mental retardation protein PAK3 contributes to synapse formation and plasticity in hippocampus
J. Neurosci.
 , 
2004
, vol. 
24
 (pg. 
10816
-
10825
)
18
Lim
L.
Manser
E.
Leung
T.
Hall
C.
Regulation of phosphorylation pathways by p21 GTPases. The p21 Ras-related Rho subfamily and its role in phosphorylation signalling pathways
Eur. J. Biochem.
 , 
1996
, vol. 
242
 (pg. 
171
-
185
)
19
Kreis
P.
Thévenot
E.
Rousseau
V.
Boda
B.
Muller
D.
Barnier
J.V.
The p21-activated kinase 3 implicated in mental retardation regulates spine morphogenesis through a Cdc42-dependent pathway
J. Biol. Chem.
 , 
2007
, vol. 
282
 (pg. 
21497
-
21506
)
20
Meng
J.
Meng
Y.
Hanna
A.
Janus
C.
Jia
Z.
Abnormal long-lasting synaptic plasticity and cognition in mice lacking the mental retardation gene Pak3
J. Neurosci.
 , 
2005
, vol. 
25
 (pg. 
6641
-
6650
)
21
Thévenot
E.
Moreau
A.W.
Rousseau
V.
Combeau
G.
Domenichini
F.
Jacquet
C.
Goupille
O.
Amar
M.
Kreis
P.
Fossier
P.
, et al.  . 
p21-Activated kinase 3 (PAK3) protein regulates synaptic transmission through its interaction with the Nck2/Grb4 protein adaptor
J. Biol. Chem.
 , 
2011
, vol. 
286
 (pg. 
40044
-
40059
)
22
Chan
C.P.
Siu
Y.T.
Kok
K.H.
Ching
Y.P.
Tang
H.M.
Jin
D.Y.
Group I p21-activated kinases facilitate Tax-mediated transcriptional activation of the human T-cell leukemia virus type 1 long terminal repeats
Retrovirology
 , 
2013
, vol. 
10
 pg. 
47
 
23
Manser
E.
Leung
T.
Salihuddin
H.
Zhao
Z.S.
Lim
L.
A brain serine/threonine protein kinase activated by Cdc42 and Rac1
Nature
 , 
1994
, vol. 
367
 (pg. 
40
-
46
)
24
Teo
M.
Manser
E.
Lim
L.
Identification and molecular cloning of a p21cdc42/rac1-activated serine/threonine kinase that is rapidly activated by thrombin in platelets
J. Biol. Chem.
 , 
1995
, vol. 
270
 (pg. 
26690
-
26697
)
25
Tartaglia
M.
Niemeyer
C.M.
Fragale
A.
Song
X.
Buechner
J.
Jung
A.
Hählen
K.
Hasle
H.
Licht
J.D.
Gelb
B.D.
Somatic mutations in PTPN11 in juvenile myelomonocytic leukemia, myelodysplastic syndromes and acute myeloid leukemia
Nat. Genet.
 , 
2003
, vol. 
34
 (pg. 
148
-
150
)
26
Weisz Hubsman
M.
Volinsky
N.
Manser
E.
Yablonski
D.
Aronheim
A.
Autophosphorylation-dependent degradation of Pak1, triggered by the Rho-family GTPase, Chp
Biochem. J.
 , 
2007
, vol. 
404
 (pg. 
487
-
497
)
27
Lei
M.
Lu
W.
Meng
W.
Parrini
M.C.
Eck
M.J.
Mayer
B.J.
Harrison
S.C.
Structure of PAK1 in an autoinhibited conformation reveals a multistage activation switch
Cell
 , 
2000
, vol. 
102
 (pg. 
387
-
397
)
28
Lightcap
C.M.
Kari
G.
Arias-Romero
L.E.
Chernoff
J.
Rodeck
U.
Williams
J.C.
Interaction with LC8 is required for Pak1 nuclear import and is indispensable for zebrafish development
PLoS ONE
 , 
2009
, vol. 
4
 pg. 
e6025
 
29
Chassaing
N.
Sorrentino
S.
Davis
E.E.
Martin-Coignard
D.
Iacovelli
A.
Paznekas
W.
Webb
B.D.
Faye-Petersen
O.
Encha-Razavi
F.
Lequeux
L.
, et al.  . 
OTX2 mutations contribute to the otocephaly-dysgnathia complex
J. Med. Genet.
 , 
2012
, vol. 
49
 (pg. 
373
-
379
)
30
Margolin
D.H.
Kousi
M.
Chan
Y.M.
Lim
E.T.
Schmahmann
J.D.
Hadjivassiliou
M.
Hall
J.E.
Adam
I.
Swyer
A.
Plummer
L.
, et al.  . 
Ataxia, dementia and hypogonadotropism caused disordered ubiquitination
New Eng. J. Med.
 , 
2013
, vol. 
36
 (pg. 
1992
-
2003
)
31
Wigginton
J.E.
Abecasis
G.R.
PEDSTATS: descriptive statistics, graphics and quality assessment for gene mapping data
Bioinformatics
 , 
2005
, vol. 
21
 (pg. 
3445
-
3447
)
32
Abecasis
G.R.
Cherny
S.S.
Cookson
W.O.
Cardon
L.R.
Merlin--rapid analysis of dense genetic maps using sparse gene flow trees
Nat. Genet.
 , 
2002
, vol. 
30
 (pg. 
97
-
101
)
33
Li
H.
Durbin
R.
Fast and accurate long-read alignment with Burrows-Wheeler transform
Bioinformatics
 , 
2010
, vol. 
26
 (pg. 
589
-
595
)
34
DePristo
M.A.
Banks
E.
Poplin
R.
Garimella
K.V.
Maguire
J.R.
Hartl
C.
Philippakis
A.A.
del Angel
G.
Rivas
M.A.
Hanna
M.
, et al.  . 
A framework for variation discovery and genotyping using next-generation DNA sequencing data
Nat. Genet.
 , 
2011
, vol. 
43
 (pg. 
491
-
498
)
35
Wang
K.
Li
M.
Hakonarson
H.
ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data
Nucleic Acids Res.
 , 
2010
, vol. 
38
 pg. 
e164
 
36
Coppola
S.
Narciso
L.
Feccia
T.
Bonci
D.
Calabrò
L.
Morsilli
O.
Gabbianelli
M.
De Maria
R.
Testa
U.
Peschle
C.
Enforced expression of KDR receptor promotes proliferation, survival and megakaryocytic differentiation of TF1 progenitor cell line
Cell Death Differ.
 , 
2006
, vol. 
13
 (pg. 
61
-
74
)
37
Jones
T.A.
Kjeldgaard
M.
Essential O
 , 
1993
Uppsala, Sweden
38
Schwede
T.
Kopp
J.
Guex
N.
Peitsch
M.C.
Swiss_Model: an automated protein homology-modeling
Nucl. Acid Res.
 , 
2003
, vol. 
31
 (pg. 
3381
-
3383
)
39
Brünger
A.T.
Adams
P.D.
Clore
G.M.
DeLano
W.L.
Gros
P.
Grosse-Kunstleve
R.W.
Jiang
J.S.
Kuszewski
J.
Nilges
M.
Pannu
N.S.
, et al.  . 
Crystallography & amp; NMR system: A new software suite for macromolecular structure determination
Acta Crystallogr. D Biol. Crystallogr.
 , 
1998
, vol. 
54
 (pg. 
905
-
921
)
40
Thisse
C.
Thisse
B.
High-resolution in situ hybridization to whole-mount zebrafish embryos
Nat. Protoc.
 , 
2008
, vol. 
3
 (pg. 
59
-
69
)
41
Leitch
C.C.
Zaghloul
N.A.
Davis
E.E.
Stoetzel
C.
Diaz-Font
A.
Rix
S.
Alfadhel
M.
Lewis
R.A.
Eyaid
W.
Banin
E.
, et al.  . 
Hypomorphic mutations in syndromic encephalocele genes are associated with Bardet-Biedl syndrome
Nat. Genet.
 , 
2008
, vol. 
40
 (pg. 
443
-
448
)
42
Zaghloul
N.A.
Liu
Y.
Gerdes
J.M.
Gascue
C.
Oh
E.C.
Leitch
C.C.
Bromberg
Y.
Binkley
J.
Leibel
R.L.
Sidow
A.
, et al.  . 
Functional analyses of variants reveal a significant role for dominant negative and common alleles in oligogenic Bardet-Biedl syndrome
Proc. Natl. Acad. Sci. USA
 , 
2010
, vol. 
107
 (pg. 
10602
-
10607
)
43
Zaghloul
N.A.
Katsanis
N.
Zebrafish assays of ciliopathies
Methods Cell. Biol.
 , 
2011
, vol. 
105
 (pg. 
257
-
272
)

Author notes

These authors contributed equally to this work.
These authors jointly supervised this study.

Supplementary data