Recurrent De Novo Mutations Disturbing the GTP/GDP Binding Pocket of RAB11B Cause Intellectual Disability and a Distinctive Brain Phenotype

Abstract

The Rab GTPase family comprises ∼70 GTP-binding proteins, functioning in vesicle formation, transport and fusion. They are activated by a conformational change induced by GTP-binding, allowing interactions with downstream effectors. Here, we report five individuals with two recurrent de novo missense mutations in RAB11B; c.64G>A; p.Val22Met in three individuals and c.202G>A; p.Ala68Thr in two individuals. An overlapping neurodevelopmental phenotype, including severe intellectual disability with absent speech, epilepsy, and hypotonia was observed in all affected individuals. Additionally, visual problems, musculoskeletal abnormalities, and microcephaly were present in the majority of cases. Re-evaluation of brain MRI images of four individuals showed a shared distinct brain phenotype, consisting of abnormal white matter (severely decreased volume and abnormal signal), thin corpus callosum, cerebellar vermis hypoplasia, optic nerve hypoplasia and mild ventriculomegaly. To compare the effects of both variants with known inactive GDP- and active GTP-bound RAB11B mutants, we modeled the variants on the three-dimensional protein structure and performed subcellular localization studies. We predicted that both variants alter the GTP/GDP binding pocket and show that they both have localization patterns similar to inactive RAB11B. Evaluation of their influence on the affinity of RAB11B to a series of binary interactors, both effectors and guanine nucleotide exchange factors (GEFs), showed induction of RAB11B binding to the GEF SH3BP5, again similar to inactive RAB11B. In conclusion, we report two recurrent dominant mutations in RAB11B leading to a neurodevelopmental syndrome, likely caused by altered GDP/GTP binding that inactivate the protein and induce GEF binding and protein mislocalization.

Main Text

The application of whole-exome sequencing (WES) as a genetic test for intellectual disability (ID) and developmental delay has increased the number of genetic causes to more than 1,500.^{1, 2} This has significantly enhanced our knowledge of molecular processes that regulate learning and memory as well as brain development, although the roles of many genes involved in these processes have not yet been defined. Using trio-based WES in diagnostic and research settings,^{1, 2} we identified two different heterozygous de novo missense mutations in RAB11B (GenBank: NM_004218.3; MIM: 604198), c.64G>A; p.Val22Met and c.202G>A; p.Ala68Thr, in five unrelated individuals with severe ID (Figures 1A and 1B; Table 1). Written consent was obtained from the legal guardians for all individuals and the study was given IRB approval. Following identification of the RAB11B mutations of the first two individuals at Radboud University Medical Center, Nijmegen (individual 1: p.Val22Met; individual 4: p.Ala68Thr), a query via GeneMatcher³ resulted in the identification of individual 2 at Haukeland University Hospital, Bergen, and individual 3 at Seoul National University College of Medicine, Seoul. Both individuals harbored the variant p.Val22Met, similar to individual 1. A fifth individual (individual 5; DECIPHER ID: 263643), reported before as part of a study sequencing a large ID cohort,⁴ was identified through international collaboration. Interestingly, this individual had the same variant, p.Ala68Thr, as individual 4 (Figure 1B). Neither identified missense mutation was present in the ExAC database or in-house control databases and both variants affect highly conserved amino acids in 90% of Rab family members⁵ and among different species (Figure 1B). Additionally, the ExAC database has shown that RAB11B is significantly depleted of rare missense variants (Z score 3.48) in healthy controls.⁶

Figure 1.

RAB11B Mutations in Five Individuals with Severe ID

(A) cDNA composition of RAB11B and location of exons (dark gray) and identified mutations (green stars).

(B) Amino acid sequence of exon 2 of RAB11B. Both identified variants (green) localize at one of the two GTP binding sites (red stripes) and p.Ala68Thr also involves the switch 2 region (black stripe). Both identified mutations affect highly conserved amino acid residues among different species (light gray box). The known inactive variant p.Ser25Asn and known active variant p.Gln70Leu are marked orange.

(C–E) Structural characterization of mutations. In all panels, the overall Protein DataBank structure 2f9m of RAB11B¹⁴ is colored gray and shown in ribbon representation, GNP (phosphoaminophosphonic acid-guanylate ester, a non-hydrolyzable analog of GTP) is multicolored in a spacefill representation with the guanidine ring in blue shades and the phosphates in red and yellow. The magnesium molecule is represented as a purple sphere. The side chains of the affected amino acids in the affected individuals are green and for the GDP/GTP-locked mutants are orange. (C) Overview of RAB11B bound to GNP and magnesium and the affected amino acids are highlighted. (D) Close up of the p.Val22Met variant shows the side chains of valine at position 22 in green and the methionine in red. (E) Close up of the p.Ala68Thr variant shows the side chains of alanine at position 68 in green and threonine in red, and the mutated glutamine at position 70 which is mutated in the GTP-locked mutant is in orange.

Table 1.

Clinical Features of Individuals with De Novo Mutations in RAB11B

	Individual 1 (Nijmegen)	Individual 2 (Bergen)	Individual 3 (Seoul)	Individual 4 (Nijmegen)	Individual 5 (DDD)
Gender	Female	Female	Male	Male	Female
Age of examination	13 years	4.5 years	8 years, 5 months	11 years	8 years, 8 months
Mutation (GenBank:NM_004218.3)
Chromosome position (Hg19)	g.8464770G>A	g.8464770G>A	g.8464770G>A	g.8464908G>A	g.8464908G>A
cDNA change	c.64G>A	c.64G>A	c.64G>A	c.202G>A	c.202G>A
Amino acid change	p.Val22Met	p.Val22Met	p.Val22Met	p.Ala68Thr	p.Ala68Thr
Growth
Height	152 cm (−1 SD)	−2 SD	112 cm (−1.9 SD)	141 cm (- 1.5 SD)	121 cm (−1.5 SD)
Weight	47.8 kg (+1.5 SD)	NR	34.1 kg (+1.7 SD)	34.5 kg (+0.7 SD)	25.7 kg (−0.5 SD)
Head circumference	49 cm (−3 SD)	−4 SD	52 cm (+0.6 SD)	50 cm (−2.2 SD)	48 cm (−3 SD)
Development
Intellectual disability	Yes – severe/profound	Yes	Yes	Yes – severe/profound	Yes- severe/profound
First words	Absent speech	Absent speech	Absent speech	Absent speech	Absent speech
First steps with support	3 years	3 years	6 years	Unknown	8 years
Neurological
Epilepsy	Possibly – to investigate	No	Yes	Yes	Single generalized seizure
Hypotonia	Yes – childhood	Yes	Yes	Yes	No
Spasticity	NR	No	No	Yes	Yes
Dystonia	NR	No	No	Yes	Yes
Abnormal gait	Yes – ataxic; broad-based	NR	NR	Yes – ataxic; broad-based	Yes – broad-based
Nystagmus	Yes	NR	NR	No	Yes - horizontal
Opthalmological abnormalities
Refraction abnormality	Hypermetropia – Mild	NR	No	Hypermetropia	No
Strabismus	Yes	NR	No	No	Yes
Other	Delayed visual maturation	Reduced vision	None	None	Optic atrophy
Musculoskeletal abnormalities
Developmental hip dysplasia	Yes – mild, non-progressive	Yes – right-sided, non-progressive	Yes	No	Yes – requiring surgery
Tapering fingers	Yes	Yes	Yes	Yes	Yes
Other	Pes cavus; shortened achilles tendons; prominent steloideus ulnae	2 cm anisomelia; adducted thumbs; bilateral club foot	None	None	Long fingers
Other
	Drooling; Simean crease; neonatal feeding difficulties	Bilateral palsy nervus laryngeus recurrens, Diabetes Mellitus Type 1, hydrocephalus	Acanthotic skin, Epidermal nevus in face, neck, trunk; Short neck; Obstructive sleep apnea; cryptorchidism	Simean crease	Bruxism.

Abbreviations: NR, Not Reported; SD, Standard Deviation.

RAB11B is a member of the large Rab GTPase protein subfamily of RAS GTPases, consisting of almost 70 small ∼21 kDa monomeric GTP-binding proteins. They serve as molecular switches that function in vesicle formation, transport, tethering, and fusion.^{7, 8} The tightly regulated spatiotemporal activity of Rabs is controlled by guanine nucleotide exchange factors (GEFs) that catalyze the GDP/GTP-exchange and GTPase activating proteins, which catalyze the hydrolysis of GTP into GDP.^{9, 10, 11} Rabs mainly interact with downstream GTPase effector proteins in their GTP-bound active conformation.^{10, 12} Rab GTPases and their effector proteins have been described to play a role in neuronal development and the shaping of cognitive functions.¹³ Several genes within this Rab GTPase family have been associated with neurodevelopmental disorders and micro- or macrocephaly, for example RAB18 [MIM: 602207] and RAB3GAP1 [MIM: 602536] are associated with Warburg micro syndrome [MIM: 600118], and RAB39B [MIM: 300774] is associated with Waisman syndrome [MIM: 311510] and Mental Retardation, X-linked 72 [MIM: 300271].

All five individuals that carried de novo mutations in RAB11B showed severe ID with motor delay and absent speech. A variety of neurological problems were present in all affected individuals, including hypotonia (4/5), epilepsy (3/5), spasticity (2/4), dystonia (2/4), broad-based (3/3), and ataxic gait (2/3) and nystagmus (2/3) (Table 1). Re-evaluation of brain MRI images of four individuals showed similar brain imaging abnormalities (Figure 2 and Table S1). All had severely decreased white matter volume (cerebral cortex more severely affected than cerebellum), thin corpus callosum, and mild ex vacuo lateral ventriculomegaly affecting the frontal horns and body of the ventricles more than the occipital and temporal horns. Other features that were scorable in a subset of images included cerebellar vermis hypoplasia (3/3), thin brainstem (3/3), early global white matter signal abnormalities consistent with delayed myelination (2/2), later patchy white matter signal abnormalities consistent with injury (4/4), optic nerve hypoplasia (2/2), and what appears to be atypical partial rhombencephalosynapsis (1/3). Although two other affected individuals did not have rhombencephalosynapsis (partial or complete absence of the cerebellar vermis with fusion of the cerebellar hemispheres), the width of the cerebellar vermis was subjectively narrower than typical. Besides the neurological phenotype, ophthalmological and musculoskeletal abnormalities were present in the majority of individuals. Microcephaly was observed in three individuals (Table 1).

Figure 2.

Brain Imaging Features and Facial Photographs of Individuals with RAB11B-Related Intellectual Disability

(A, F, G, J) Markedly decreased cortical white matter volume in individuals 1, 3, 4, and 5.

(B, H, K) Markedly thin corpus callosum (arrows), mildly thin brainstem, and mildly small, atrophic-appearing cerebellar vermis (bracket) in individual 1, 4, and 5.

(C) Increased T2/FLAIR signal in the cortical white matter (asterisks) in individual 1.

(D and E) Atypical partial rhombencephalosynapsis in individual 1 (arrowheads). (A, D, F, G, and H) are T2-weighted axial images. (B and H) are T1-weighted sagittal images. (C) is a T2/FLAIR-weighted axial image. (E) is a reformatted inversion-recovery coronal image. (K) is a T2-weighted sagittal image.

(I and L) Facial photographs show upward slanted palpebral fissures, periorbital fullness, full nasal tip, and hypotonic face in individual 4 and upward slanted palpebral fissures, deep set eyes, and short philtrum in individual 5.

To study the effects of the identified mutations on the well described three-dimensional protein structure of RAB11B, we modeled the substitutions on Protein Data Bank entry PDB: 2F9M¹⁴ using YASARA software,¹⁵ and compared to the described active, GTP-locked RAB11B mutant p.Gln70Leu and inactive, GDP-locked RAB11B mutant p.Ser25Asn (Figure 1C).¹⁶ The p.Ser25Asn variant was shown to disrupt the binding of a magnesium molecule which is essential for GTP binding, consequently locking the GTPase in an inactive GDP bound, non-membrane-associated state.^{17, 18, 19} In contrast, the p.Gln70Leu variant affects a conserved residue in the flexible switch II region, which is essential for catalysis, and therefore this variant is predicted to fix the GTPase in a GTP-bound state, which is constitutively active (Figures 1B and 1C).^{19, 20} Both mutations identified in this study are situated in close proximity to the binding pocket for GTP/GDP and specifically on the side where the phosphate groups of GTP/GDP are positioned (Figure 1C). The valine residue at position 22 is located at the base in the middle of the GTP/GDP binding pocket, thereby being responsible for its shape (Figures 1B and 1D). Substitution into methionine brings in a larger amino acid which causes a predicted reorganization of the region around this residue, altering the shape of the binding pocket. This likely disrupts binding of GDP and GTP, resulting in a nucleotide-free, inactive state of RAB11B. The second mutation affects an alanine at position 68 and is located within a flexible loop of the switch II region (Figures 1B and 1E), which together with the switch I region specifically interacts with effector proteins when GTP is bound.¹² This flexible loop provides more space for a bigger amino acid residue, however, the mutant residue is positioned closely to the magnesium molecule, as well as the phosphate groups of GTP/GDP, and thereby alters the binding pocket and is predicted to disrupt nucleotide binding. Furthermore, the location within the switch region suggests that the mutation causes altered binding with effector proteins. Taken together, both identified mutations are predicted to alter the GTP/GDP binding pocket of RAB11B, hence it is likely that the functionality of the protein is affected.

RAB11B is one of three genes in the Rab11 GTPase subfamily, together with RAB11A [MIM: 605570] and RAB25 [MIM: 612942], and this subfamily specifically associates with recycling endosomes.^{7, 21} RAB11A is the best-characterized family member and is ubiquitously expressed,²² in contrast to specific expression of RAB25 and RAB11B in epithelial tissue²³ or in heart, testis, and brain,²⁴ respectively. The proteins have been implicated in the regulation of vesicular trafficking between the recycling endosome compartment and early endosomes to the trans Golgi network and plasma membrane.^{17, 25} Furthermore, it has been shown that RAB11 is essential for ciliogenesis,^{26, 27, 28} and that the proteins are localized to peri-centrosomal recycling endosomes concentrated at the base of the cilium.²⁹

To evaluate whether the identified mutations affect the sub-cellular localization of RAB11B, potentially at the cilium, we transfected human TERT-immortalized retinal pigment epithelium 1 (hTERT RPE1) cells with cDNA expression constructs encoding WT (UniProt: Q15907) or mutant RAB11B, fused to an N-terminal 3xFLAG tag. A marker for the peri-centrosomal region (Pericentriolar Material 1; PCM1) was used to assess the effect on the peri-centrosomal localization (Figures 3A–3F), and the marker acetylated tubulin to asses cilium morphology (Figure S1). Wild-type RAB11B localization was scattered throughout the cell with a punctuated pattern as shown in Figure 3A. For some cells, puncta were enriched, as expected, at the peri-centrosomal region. We observed that the constitutively active mutant RAB11B-p.Gln70Leu had a strong organization in puncta throughout the cytoplasm and a consistent localization at the peri-centrosomal region in all cells (Figure 3B). In contrast, GDP-bound inactive RAB11B-p.Ser25Asn showed a generally dispersed cytosolic localization, losing its association with the peri-centrosomal region, but with enrichment near the nucleus suggesting Golgi localization (Figure 3C). This was confirmed using Golgi marker GM130 (Figure 3G–3L). Our results upon expression of recombinant WT, p.Ser25Asn, and p.Gln70Leu RAB11B confirmed earlier studies on the effect of these variants on the localization of this GTPase.^{17, 18} Localization of both RAB11B-p.Val22Met (Figure 3D) and RAB11B-p.Ala68Thr (Figure 3E) showed homogeneous localization in the cytoplasm, no association with the peri-centrosomal region and co-localization with Golgi marker GM130 (Figures 3J and 3K), similar to the GDP-bound inactive mutant p.Ser25Asn. Quantification of the co-localization patterns with PCM1 and GM130 confirmed our observations (Figure 3F and 3L). The morphology of the cilium appeared normal in cells transfected with each of the four mutant constructs (Figure S1). We reproduced the localization patterns at the base of the cilium or at the Golgi for all RAB11B variants in a different cell line: mouse inner medullary collecting duct 3 (IMCD3) cells (Figure S2). In short, introduction of either one of the identified RAB11B variants results in a similar localization pattern as the GDP-bound inactive form, without an obvious difference between the two. Interestingly, previous studies showed that GDP-locked GTPases lose their membrane association.^{17, 19} For the p.Ser25Asn mutant of RAB11B, we indeed observe a localization pattern that is similar to that of the published RAB11B-ΔC mutant where five C-terminal amino acids were substituted to abolish prenylation, and consequently disturbed vesicular membrane association.¹⁸ The fact that both identified variants result in a comparable RAB11B localization pattern as the p.Ser25Asn mutant, suggests that in the affected individuals, the association of RAB11B with (vesicular) membranes is affected, contributing to the pathogenic effects of the identified variants.

Figure 3.

Localization of Wild-Type and Mutant RAB11B in hTERT-RPE1 Cells

Shown in green is the expression of recombinant 3xFLAG-RAB11B wild-type (A and G) and 3xFLAG-RAB11B variants p.Gln70Leu (B and H), p.Ser25Asn (C and I), p.Val22Met (D and J), p.Ala68Thr (E and K) in hTERT RPE1 cells, detected using anti-FLAG antibodies (rabbit polyclonal, #F7425, Sigma Aldrich; dilution 1:500 in PBS). N-terminal tagging was performed to not disturb the prenylation of the C terminus of RAB11B, which is required for membrane association.^{10, 18} Recombinant protein expression, immunocytochemistry, and image capture was performed as described previously.⁴⁴ All secondary antibodies were Alexa Fluor conjugates (Thermo Fisher Scientific). The following transfection efficiencies were obtained per 3xFLAG-RAB11B construct: wild-type, 14%; p.Gln70Leu, 19%; p.Ser25Asn, 9%; p.Val22Met, 10%; p.Ala68Thr, 14%. Staining of Pericentriolar Material 1 (PCM1) using anti-PCM1 antibodies (goat polyclonal, #SC-50164, Santa Cruz Biotechnology; dilution 1:250 in PBS) was used to mark the peri-centrosomal region (red) and magnifications are shown in the insets (A–E). Staining of GM130 using anti-GM130 antibodies (mouse monoclonal, #610822, BD Biosciences; dilution 1:250 in PBS) was used to mark the Golgi apparatus (in red; G-K). Co-localization with PCM1 (F) or GM130 (L) was quantified by calculating the Pearson’s correlation coefficient (PCC) using the JACoP plugin in ImageJ (N = 12 cells for each construct, mean in red and error bars represent the SD).^{45, 46, 47} The significance of difference with WT was calculated with a Student’s t test (ns: not significant (p > 0.05); ^∗: p < 0.05; ^∗∗: p < 0.005; ^∗∗∗: p < 0.0005). Wild-type and GTP-bound active RAB11B (p.Gln70Leu) are showing punctuated localization with an enrichment at the peri-centrosomal region. Both identified variants p.Val22Met and p.Ala68Thr show similar localization as GDP-bound inactive RAB11B (p.Ser25Asn) with dispersed localization throughout the cytoplasm and enrichment at the Golgi apparatus. In all pictures, nuclei were stained with DAPI (blue). Scale bars represent 10 μm.

To further assess the effect of the identified variants on the functionality of RAB11B, we screened for binary interaction partners of RAB11B by using a GAL4-based yeast two-hybrid screen of cDNA libraries from neuronal tissues (brain and retina) as previously described.³⁰ In physiological conditions, RAB11B is either GDP or GTP bound. Therefore we used both p.Ser25Asn and p.Gln70Leu mutant constructs as baits to screen for potential interactors, which identified two (p.Ser25Asn) and eight (p.Gln70Leu) different potential interactors (Figure 4 and Figure S3). All interactions were confirmed using independent co-transformation assays with validation of reporter gene activity (Figure S3). The identified interactors were all previously reported as Ral/Rac or Rab11 specific interacting partners (Table S2), validating our assay. Interestingly, three direct interactors identified in our yeast screen are encoded by genes where loss of function mutations are associated with neurodevelopmental disorders: CNKSR2 [MIM: 300724] with X-linked Intellectual Disability,^{31, 32, 33} MYO5A [MIM: 160777] with Griscelli syndrome type 1 [MIM: 214450], and TRAPPC9 [MIM: 611966] with Recessive Mental Retardation 13 [MIM: 613192] (Table S2).

Figure 4.

Identified Interactors of RAB11B Using Yeast Two-Hybrid cDNA Library Screening and Their Affinity for RAB11B Variants

Co-transformations were performed in PJ69-4α yeast strains with the identified clones from the library screens fused to pAD together with WT or mutant RAB11B constructs fused to pBD. The clones are sorted according to the library in which they have been identified, as indicated in the far left column. The quantification of the selection of the growth of two-hybrid clones grown on medium lacking leucine, tryptophan and histidine with 3mM 3AT is quantified in the “growth” column. α-Galactosidase reporter gene activation (α-gal column) or β-galactosidase reporter gene activation (β-gal column) are quantified as well. Quantifications are on a scale from 0-3; with 0 for no reporter gene activation and 3 for highest reporter gene activation. Details of the identified clones in the cDNA library screens can be found in Table S2, including quantification of reporter gene activation on –LWHA medium. Original images are shown in Figure S2.

To assess whether the identified variants affect the affinity of RAB11B to any of the interactors, we co-expressed p.Val22Met and p.Ala68Thr mutant constructs with each interactor and compared the binding affinities semiquantitatively by evaluation of the reporter gene-activation levels (Figure 4 and Figure S3, and Table S2). Wild-type, GDP-, and GTP-bound RAB11B were taken along as controls. Because GTP is more abundantly present in the cytosol than GDP, the WT RAB11B construct is expected to be predominantly GTP-bound. We indeed observed that the interaction pattern of the WT was comparable to the constitutively active mutant (Figure 4 and Figure S3). When either one of the identified variants was introduced, the binding pattern was mostly similar to that of WT RAB11B, but one important difference could be distinguished: both mutant proteins were able to bind to the one bona fide GEF identified in our screen, SH3BP5, while WT RAB11 could not (Figure 4 and Figure S3).

The affinity of the p.Val22Met mutant protein to SH3BP5 was strong and comparable to the GDP-locked mutant protein p.Ser25Asn, while the affinity of the p.Ala68Thr mutant protein was somewhat lower. Based on the 3D modeling and our localization assay, we already suggested that the p.Val22Met variant results in a nucleotide-free state by the inability of the protein to bind GDP or GTP. The interaction data support our hypothesis that variant p.Val22Met causes a nucleotide-free state, because GEFs are required for the stabilization of a GTPase if not bound to GTP or GDP.^{11, 19} A somatic substitution affecting the adjacent conserved glycine residue (RAB11B position 23; Figure S4) has been described in RAS-GTPase RHOA (p.Gly17Val),^{34, 35, 36} resulting in a nucleotide-free state of RHOA. As a result, mutant RHOA acts in a dominant-negative manner because it sequesters GEFs, which prohibits that these GEFs are available to activate wild-type RHOA.³⁷ With (1) the position of the p.Val22Met variant adjacent to this reported RHOA substitution, (2) the 3D modeling suggesting a nucleotide-free state of RAB11B (Figure 1C), and (3) the strong affinity of the mutant to bind the RAB11B GEF SH3BP5 (Figure 4 and Figure S3), it is likely that the identified p.Val22Met variant acts in a dominant-negative manner as described for RHOA. Interestingly, we observed that this mutant RAB11B is still able to interact with effectors as well as GEFs (Figure 4 and Figure S3). Also, for the p.Ala68Thr variant, the binding to SH3BP5 is much less pronounced and only detectable under less stringent assay conditions (-LWH + 5mM 3AT, Figure 4 and Figure S3). It has been described that a specific dominant-negative mutant of Ras (Ras15A), member of the RAS-GTPase family as well, has stronger affinities for GEFs with more defective nucleotide binding, compared to another dominant-negative mutant of Ras (Ras17N) where a neighboring residue is affected.³⁷ This highlights the possibility of variable effects between variants in the GTP/GDP binding pocket on the affinity for GEFs and nucleotides. We argue that, despite these slight differences in affinity, both variants have similar consequences on RAB11B function, since individuals who harbor the p.Ala68Thr variant display strong phenotypic overlap with individuals carrying the p.Val22Met variant. Furthermore, both variants caused a similarly disturbed localization (Figure 3). Therefore, we hypothesize that p.Ala68Thr could act in a dominant-negative manner as well, with the same phenotypic consequences as the p.Val22Met variant.

The second GEF for RAB11B identified in our screen is TRAPPC9, member of the TRAPP II complex. This group of proteins is involved in intracellular membrane trafficking processes^{10, 38} and acts as a GEF for RAB11B orthologs in yeast.^{39, 40, 41} The fact that the p.Val22Met and p.Ala68Thr mutant proteins are not able to associate with TRAPPC9 seems to contradict the results with the GEF SH3BP5. However, SH3BP5 was shown to be a bona fide GEF for RAB11,⁴² while TRAPPC9 requires the other subunits of the TRAPP II complex to act as a GEF. As the identified RAB11B mutations do not induce the TRAPPC9 binding, our data also suggest that TRAPPC9, as a single protein, does not act as a RAB11 GEF in contrast to SH3BP5.

It is likely that RAB11B disruption also has molecular consequences other than disturbed protein interactions that might alter several cellular mechanisms such as calcium influx, synaptic function, and neuronal migration,¹³ that could contribute to the underlying pathology. One explanation could be found in the mislocalization of RAB11B mutant proteins, which were able to bind effector proteins based on our yeast two-hybrid data. As a result of the membrane unbound situation suggested by the localization data, mislocalization of mutant RAB11B might therefore result in mislocalization of bound effectors, sequestering them from their usual sites of function.³⁷ Our data also show that RAB11B localization is altered at the basal body of the cilium (Figure S1), and RAB11B is a direct interactor of the ciliary RAB Rabin8 (Figure 4 and Table S2),⁴³ although we observed no morphological changes of the cilium in hTERT-RPE1 cells. More subtle changes in cilium morphology could have been missed in our assay or cilia might have an altered signaling function without morphological abnormalities. Although speculative at this point without an animal model, cilia could be mainly affected in brain tissue, given the predominant expression of RAB11B in the brain.^{13, 24}

In conclusion, we identified two recurrent de novo missense mutations in RAB11B in five unrelated individuals with severe ID and specific brain abnormalities. We show that (1) both mutations affect the GDP/GTP binding pocket of this small GTPase, (2) the association of the mutant proteins with (vesicular) membranes is affected in localization studies, and (3) both mutations have limited effect on RAB11B effector protein binding, but can enhance the affinity to the GEF SH3BP5. We propose that these effects together cause distinct defects in several neuronal developmental processes, a combination that results in a neurodevelopmental syndrome in human.

Published: October 26, 2017

Web Resources

• ExAC Browser, http://exac.broadinstitute.org/gene/ENSG00000185236/

• GeneMatcher, https://genematcher.org/

• OMIM, http://www.omim.org/