Article Text

Download PDFPDF

Original research
Biallelic VPS35L pathogenic variants cause 3C/Ritscher-Schinzel-like syndrome through dysfunction of retriever complex
  1. Kohji Kato1,2,3,
  2. Yasuyoshi Oka4,5,
  3. Hideki Muramatsu2,
  4. Filipp F Vasilev6,7,
  5. Takanobu Otomo6,
  6. Hisashi Oishi8,
  7. Yoshihiko Kawano3,
  8. Hiroyuki Kidokoro2,3,
  9. Yuka Nakazawa4,5,
  10. Tomoo Ogi4,5,
  11. Yoshiyuki Takahashi2,
  12. Shinji Saitoh1
  1. 1 Department of Pediatrics and Neonatology, Nagoya City University Graduate School of Medical Sciences and Medical School, Nagoya, Aichi, Japan
  2. 2 Department of Pediatrics, Nagoya University Graduate School of Medicine Faculty of Medicine, Nagoya, Aichi, Japan
  3. 3 Department of Pediatrics, Toyota Memorial Hospital, Toyota, Aichi, Japan
  4. 4 Department of Human Genetics and Molecular Genetics, Nagoya University, Nagoya, Aichi, Japan
  5. 5 Department of Genetics, Research Institute of Environmental Medicine, Nagoya University, Nagoya, Aichi, Japan
  6. 6 Department of Pathophysiology and Metabolism, Kawasaki Medical School, Kurashiki, Okayama, Japan
  7. 7 International Research Fellow of Japan Society for the Promotion of Science (Postdoctoral Fellowships for Research in Japan (Standard)), Tokyo, Japan
  8. 8 Department of Comparative and Experimental Medicine, Nagoya City University Graduate School of Medical Sciences and Medical School, Nagoya, Aichi, Japan
  1. Correspondence to Dr Shinji Saitoh, Department of Pediatrics and Neonatology, Nagoya City University Graduate School of Medical Sciences and Medical School, Nagoya 464-8601, Japan; ss11{at}


Background 3C/Ritscher-Schinzel syndrome is characterised by congenital cranio-cerebello-cardiac dysplasia, where CCDC22 and WASHC5 are accepted as the causative genes. In combination with the retromer or retriever complex, these genes play a role in endosomal membrane protein recycling. We aimed to identify the gene abnormality responsible for the pathogenicity in siblings with a 3C/Ritscher-Schinzel-like syndrome, displaying cranio-cerebello-cardiac dysplasia, coloboma, microphthalmia, chondrodysplasia punctata and complicated skeletal malformation.

Methods Exome sequencing was performed to identify pathogenic variants. Cellular biological analyses and generation of knockout mice were carried out to elucidate the gene function and pathophysiological significance of the identified variants.

Results We identified compound heterozygous pathogenic variants (c.1097dup; p.Cys366Trpfs*28 and c.2755G>A; p.Ala919Thr) in the VPS35L gene, which encodes a core protein of the retriever complex. The identified missense variant lacked the ability to form the retriever complex, and the frameshift variant induced non-sense-mediated mRNA decay, thereby confirming biallelic loss of function of VPS35L. In addition, VPS35L knockout cells showed decreased autophagic function in nutrient-rich and starvation conditions, as well as following treatment with Torin 1. We also generated Vps35l−/− mice and demonstrated that they were embryonic lethal at an early stage, between E7.5 and E10.5.

Conclusions Our results suggest that biallelic loss-of-function variants in VPS35L underlies 3C/Ritscher-Schinzel-like syndrome. Furthermore, VPS35L is necessary for autophagic function and essential for early embryonic development. The data presented here provide a new insight into the critical role of the retriever complex in fetal development.

  • C16orf62
  • autophagy
  • retriever complex
  • chondrodysplasia punctata
  • knockout mouse

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


VPS35L (also known as C16orf62) is highly evolutionarily conserved protein associated with the endosomal cargo recycling system of integral membrane proteins, which maintains homeostasis of the plasma membrane.1 In this endosomal cargo recycling system, SNX17 with the retriever complex or SNX27 with the retromer complex promotes selective cargo entry into the recycling pathway instead of degradation in lysosomes, interacting with the WASP and Scar homologue (WASH) and CCC (CCDC93, CCDC22 and COMMD) complexes.2–7 VPS35L forms the retriever heterotrimer protein complex with VPS29 and DSCR3. The SNX17 retriever axis is associated with numerous cell surface protein recycling, and SNX17 knockdown is reported to affect expression levels of these integral membrane proteins.1

In this endosomal cargo recycling gene network, CCDC22 and WASHC5 are the responsible genes for 3C (cranio-cerebello-cardiac)/Ritscher-Schinzel syndrome (3C/RSS (MIM:220 210)).8–11 3C/RSS is characterised by distinctive craniofacial features, cerebellar anomalies and congenital heart defects. Craniofacial features, including arched eyebrow, prominent forehead, hypertelorism, down-slanted palpebral fissures, upturned nose and thin upper lip are important for clinical diagnosis. In addition, cardiac defects and cerebellar anomalies are complicated in approximately 80% of cases.12 The molecular mechanisms of endosomal cargo recycling systems have been intensively researched, and retromer dysfunction has been linked to a growing number of neurological disorders, including Alzheimer disease and Parkinson disease.13–15 However, the association between retriever complex dysfunction and clinical disorders is not yet fully understood.

Here, we report two siblings, elder girl and younger boy, who presented clinical manifestations similar to, but distinct from, 3C/RSS, including cranio-cerebello-cardiac anomalies, coloboma, microphthalmia, chondrodysplasia punctata, complicated skeletal malformation, periventricular nodular heterotopia and proteinuria. We performed exome sequencing on the affected siblings and identified compound heterozygous pathogenic variants in VPS35L. These results link VPS35L dysfunction to a 3C/RSS-like congenital malformation syndrome and provide new insight into the critical role of VPS35L and the retriever complex in fetal development.


Genetic analysis

Exome sequencing was performed on the affected siblings and their parents as previously described.16 Genomic DNA was extracted from peripheral blood or umbilical cord, partitioned using the SureSelect XT Human All Exon V5 capture library (Agilent Technologies, Santa Clara, California, USA), and DNA sequenced using 100 bp paired end reads with the Illumina Hiseq 2500 sequencer. Following alignment to the reference genome (Hg19) and variant calling, we removed non-functional mutations except for non-synonymous SNVs, insertions and deletions (indels), and splice site variants. We also removed variants with allele frequency >1% at ExAC (Exome Aggregation Consortium, Cambridge, Massachusetts, USA (URL: except for those also identified as pathogenic variants in the National Center for Biotechnology Information Clin Var and HGMD databases. According to the pedigree, we focused on the recessively inherited trait. The identified variants were confirmed by Sanger sequencing of PCR-amplified products. RNA-seq was performed using RNA extracted from skin fibroblast derived from individual-2 and other controls as previously described.17–19 To identify the differentially expressed genes in patient-derived cells compared with controls, we used Wald-Log test.20

This study was approved by the Ethical Committee for the Study of Human Gene Analysis at Nagoya University Graduate School of Medicine. Written informed consent was obtained from the parents.


A pcDNA3-DYK-tagged clone of the coding sequence of VPS35L was obtained from GenScript (OHu03265, Piscataway, New Jersey, USA). Site-directed mutagenesis (KOD-Plus Mutagenesis Kit, Toyobo, Osaka, Japan) was performed to generate the VPS35L-A919T amino acid substitution construct. All constructs were verified by DNA sequencing. Plasmids were transfected into cells using Lipofectamine 3000 Transfection Reagent (Thermo Fisher Scientific, Waltham, Massachusetts, USA) as per manufacturer’s guidelines. To achieve equal expression levels of wild-type and A919T mutant VPS35L, we administered 1.5 times more of the VPS35L-A919T-expressing plasmid compared with the wild-type VPS35L.

Protein analysis

Indicated amounts of cell extracts were separated by SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes (Millipore, Billerica, Massachusetts, USA). After blocking with 5% skimmed milk in PBST, membranes were incubated with primary antibodies, followed by incubation with horseradish peroxidase–conjugated secondary antibody (GE Healthcare, Little Chalfont, UK). Primary antibody against VPS29 (sc-398874) was purchased from Santa Cruz Biotechnology (Texas, USA), C16orf62 (ab97889) from Abcam (Cambridge, UK), GAPDH (no 5174) and DYKDDDDK tag (no 2368) from Cell Signaling Technology (Danvers, Massachusetts, USA), and DSCR3 (ABN87) from Merck KGaA (Darmstadt, Germany). Densitometric quantification was performed using ImageJ software.

For cycloheximide (CHX) chasing assay, cells transfected with the wild-type or the VPS35L-A919T construct were incubated for 48 hours followed by treatment with 100 µg/mL CHX (Merck kGaA). Protein was extracted at 0, 20 and 40 min after CHX treatment using RIPA buffer (FUJIFILM Wako Pure Chemical, Osaka, Japan) containing protease inhibitor (Roche, Basel, Switzerland).

For DYK-tag immunoprecipitation, HEK293T cells transfected with the wild-type VPS35L or the VPS35L-A919T were lysed using IP Lysis Buffer (Thermo Fisher Scientific). Lysates were cleared by centrifugation and then incubated with anti-DDDDK-tag pAb-Agarose (PM020; MBL, Nagoya, Japan) for 1 hour at 4 ºC.

CRISPR/Cas9-mediated gene edition

A HEK293T cell line lacking VPS35L was generated using CRISPR/Cas9 technology.21 For generating VPS35L-A919T point mutation (knock-in) in HEK293T cell line, we used CRISPR/Cas9-mediated homology-directed repair pathway. Target sequence was TCTCCTGGCTCATGGCGGAGAGG in exon 28, and we used single-stranded DNA donor with G>A substitution mimicking patient-derived A919T mutation as template. Clones were isolated and gene disruption or point mutation were validated by PCR and PCR-based sequencing.

Generation of mutant animals

Vps35l mutant mice were created with the CRISPR/Cas9 system. The target sequence (5′-CCATCCGGGAACTCATTCCAAGA-3′) on exon 10 was selected for the inactivation of Vps35l gene. The sequence was inserted into the entry site of pX330 plasmid, gifted from Feng Zhang (Addgene plasmid no 42230).22 Vps35l targeting vector (pX330-Vps35l) for microinjection was isolated with FastGene Gel/PCR Extraction Kit (Nippon genetics, Tokyo, Japan) and diluted to 5 ng/µL with deionised distilled water. Then the vector was microinjected into the male pronuclei of fertilised oocytes which were harvested from superovulated mated C57BL/6N females. Surviving one-cell embryos were transferred into the oviduct of pseudopregnant ICR females. Two independent mouse lines (1 and 2), each with a frameshift indel in exon 10 of mouse Vps35l, were bred as heterozygotes. Animals were housed in accordance with protocols approved by the Institutional Animal Care and Use Committee at Nagoya City University. PCR genotyping of mice was performed with the use of genomic DNA from yolk sac or tail biopsies.

Assays for autophagy

Cells were washed twice with warm phosphate-buffered solution (PBS) and incubated in regular medium or Earle’s Balanced Salt Solution with or without 125 nM Bafilomycin A1 (BafA1) and 250 nM Torin 1. After 2 hours, cells were washed with PBS, collected and lysed in SDS sample buffer. Samples were subsequently subjected to immunoblotting to detect LC3 and β-actin. Antibody against β-actin (M177-3) and LC3 (PM036) were purchased from MBL. To ensure that these results were not from off-target effects, we designed another gRNA sequence to generate additional VPS35L-knockout HeLa cell line (VPS35L-KO no 2). HeLa cell lines lacking VPS35L were generated using CRISPR/Cas9 technology.21

Statistical analysis

Results are presented as the mean±SE measurement. Two-sided Student’s t-test was performed to compare the means between two groups. When the means of three groups were to be compared, one-way analysis of variance (ANOVA) with post hoc Tukey’s honestly significant difference calculator test was used. Statistics were calculated using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan).23 P value of <0.05 was considered significant.


Clinical findings

The elder girl, affected individual-1, and younger boy, individual-2, are the first and second children of unrelated, healthy Japanese parents, who had no remarkable medical histories. The siblings presented mostly overlapping multiple congenital anomalies of craniofacial, ophthalmological, central nervous, cardiac, genital and skeletal systems (table 1). The girl had craniofacial features, including dilated anterior fontanel, arched eyebrow, downslanting palpebral fissures, upturned nose, thin upper lip and micrognathia (figure 1A). She presented with vermian hypoplasia, bilateral buphthalmos, left choroidal coloboma, atrioventricular septal defect and brachytelephalangic chondrodysplasia punctata with first metacarpals and metatarsals shortening (figure 1B–D). She died at 28 days of life. The boy was 3 years at the latest examination, and he had severe postnatal growth retardation. His length was 75 cm (−6.2 SD), body weight was 8.2 kg (−3.8 SD) and head circumference was 47 cm (third percentile). He had craniofacial features, including dilated anterior fontanel, prominent forehead, arched eyebrow, downslanting palpebral fissures, upturned nose, thin upper lip, and micrognathia (figure 1E,F). His skeletal complication affected the whole body with vertebral body hypo-ossification, breast bone aplasia, shortened ulnae with radial bowing, short limbs and chondrodysplasia punctata with brachytelephalangy and first metacarpals and metatarsals shortening (figure 1G). Ophthalmological examinations revealed left microphthalmus and right choroidal coloboma, and orbital MRI showed atrophy of the optic nerve (figure 1H). Brain MRI showed vermian hypoplasia (figure 1I) and bilateral periventricular nodular heterotopia (PNH; figure 1J). He developed focal onset impaired awareness seizures at 18 months of age, which was controlled by administration of phenobarbital. He had proteinuria of 3+ by a urine test strip, but renal biopsy was not performed. His global development was severely delayed: he started to roll over at 2 years, but he was not able to sit at 3 years of age. In relation to cognitive development, he could understand several words and responded to bye-bye with waving of a hand. The craniofacial, cardiac and cerebellar anomalies presented by patients were typical symptoms in 3C/RSS. On the other hand, skeletal complications and postnatal growth retardation presented by patient 2 seemed to be more severe than those previously documented in patients with 3C/RSS.12 24–26 In addition, symptoms of microphthalmus, PNH and proteinuria are not typical in 3C/RSS (table 1).

Figure 1

Clinical features of the affected siblings. (A) Facial photograph of the affected girl taken at neonatal period, showing distinctive craniofacial features including arched eyebrow, downslanting palpebral fissures, upturned nose and micrognathia. (B–D) X-ray image showing brachytelephalangic chondrodysplasia punctata with first metacarpals and metatarsals shortening. (E) Photograph of the affected boy taken at 3 years of age showing distinctive craniofacial features similar to those of affected girl. (F) Breast bone aplasia and thoracic instability caused retractive breathing. (G) X-ray image showing shortened ulnae with radial bowing. (H) Axial slice of T1-weighted orbital MRI image showing left microphthalmus and optic nerve atrophy. (I) Sagittal slice of T1-weighted image show vermian hypoplasia. (J) Axial slice of fluid attenuation inversion recovery image shows periventricular nodular heterotopia. (A–D) Images of the findings in the affected girl, and (E–J) images of the findings in the affected boy. The parents of the siblings gave a written consent for the publication of the photographs.

Table 1

Comparison of clinical features in our two patients, and previously reported 3C/RSS patients with CCDC22 or WASHC5 pathogenic variants

Exome sequencing identifies compound heterozygous variants in VPS35L

We performed exome sequencing to identify the causative gene variants underlying each patient’s phenotype. According to the pedigree, we focused on recessively inherited traits (figure 2A), and in only one gene, VPS35L, we identified a compound heterozygous variant in both of the affected siblings (NM_020314.5: c.1097dup, p.Cys366Trpfs*28; c.2755G>A, p.Ala919Thr; figure 2B). The Ala919 residue is evolutionarily highly conserved (figure 2C) and located at the VPS29 binding domain.1 The missense variant of p.Ala919Thr has not been reported in any public database. The other alternations in this residue, p.Ala919Pro and p.Ala919Ser, are listed in gnomAD. However, they are very rare with allele frequency of 7.43E-5 and 3.98E-6, respectively, and no individuals with homozygous alternate allele are listed. RT-PCR and direct Sanger sequencing of cDNA from EB virus-transformed lymphoblastoid cell lines (LCLs) established from the affected boy showed the homozygous alternate residue base (A) at c.2755. This indicates a markedly decreased expression of the c.1097dup frameshift variant allele (figure 2B). In addition, RNA-seq data obtained from patient-derived skin fibroblast showed that mRNA expression level of VPS35L was reduced by half compared with other controls (online supplementary table S1). In respect to DSCR3 and VPS29, the other members of the retriever heterotrimer complex, no possibly pathogenic mutations were identified in exome sequencing. Furthermore, the mRNA expression level of DSCR3 and VPS29 was not changed in patient-derived skin fibroblast compared with controls in RNA-seq data.

Supplemental material

Figure 2

Genetic analysis of the affected siblings. (A) Pedigree analyses with the affected siblings represented by filled symbols. VPS35L genotypes of the family members are given: p.C366Wfs*28; p.A919T; wild-type, WT. (B) Genomic DNA (left) sequence chromatograms illustrate the compound heterozygous variants in VPS35L (c.1097dupG; c.2755G>A) in the affected boy. In addition, cDNA (right) sequence indicate mRNA decay in the frameshift variant allele. (C) Phylogenic conservation was observed for the mutated amino acid residue (p.A919T), marked by an asterisk.

VPS35L-A919T protein is unstable due to a loss in retriever complex formation

To assess the VPS35L function, western blot analysis was performed using proteins obtained from LCLs established from the affected boy, his parents and healthy control leucocytes. VPS35L protein expression was significantly decreased in the affected boy compared with his parents and healthy controls (figure 3A). In respect to the other members of the retriever heterotrimer complex, DSCR3 expression was also significantly decreased in the affected individual, but the VPS29 expression level was unchanged (figure 3A). To understand why VPS35L expression was markedly decreased in the affected boy even though one allele has a missense variant, we validated the impact of the VPS35L-A919T amino acid substitution on the retriever complex. Immunoprecipitation and immunoblot analyses indicated that the endogenous VPS29 protein was not co-immunoprecipitated with the VPS35L-A919T mutant while it was co-immunoprecipitated with the wild-type VPS35L (figure 3B). Next, we tested the stability of VPS35L-A919T protein in the presence of CHX. We observed a rapid reduction of VPS35L-A919T protein compared with the wild-type (figure 3C,D, one-way ANOVA with post hoc Tukey’s honestly significant difference calculator test; df=5, F=4.43, p<0.05).

Figure 3

Decreased protein levels of VPS35L and DSCR3 in patient cells due to the inability to form retriever complex in VPS35L-A919T. (A) Representative immunoblots of VPS35L, DSCR3 and VPS29 expression in LCLs derived from the affected boy, his parents and healthy controls (GAPDH as a loading control; n=3). (B) VPS35L-A919T mutant protein did not interact with VPS29. HEK293T cells were transfected with vectors expressing the indicated DYK-tagged proteins (DYK, empty; VPS35L-WT, wild type; VSP35L-A919T, mutant). The DYK-tagged proteins were immunoprecipitated and detected by immunoblotting with anti-DYK antibodies (n=3). (C) HEK293T cells expressing the DYK-tagged wild type, VPS35L or VPS35L-A919T were treated with CHX). Protein extracts from the CHX-treated cells at indicated time points were analysed by immunoblotting. (D) Quantification of VPS35L protein levels as a proportion of GAPDH levels averaged over three experiments. Relative densities of the bands were determined relative to 1 at time point 0 for each experiment and were normalised using GAPDH loading controls. VPS35L-A919T was unstable compared with the wild-type VPS35L. Bar graphs, means and SE measurements are shown; *p<0.05. (E) Schematic illustrating how the retriever complex interacts with the CCC and WASH complexes in the endosomal cargo recycling system. The CCC and WASH complexes include CCDC22 and WASHC5, known causative genes in 3C/RSS. VPS35L, encoding a core protein of the retriever complex, is a novel causative gene of 3C/RSS-like congenital malformation syndrome. VPS35L-A919T mutant protein is unstable due to a loss in retriever complex formation. 3C/RSS, 3C (cranio-cerebello-cardiac)/Ritscher-Schinzelsyndrome; CHX, cycloheximide; LCL, lymphoblastoidcell line.

To further ensure the pathogenicity of VPS35L-A919T, we generated VPS35L-KO and VPS35L[A919T/fs] cells (online supplementary file 2). We confirmed a homozygous frameshift variant in VPS35L-KO cells, and compound heterozygous variants in VPS35L[A919T/fs] cells including the patient-derived missense variant (c.2755G>A; p.Ala919Thr) and the frameshift variant (online supplementary file 2). The genotype of VPS35L[A919T/fs] cells mimics that of present patients. As the previous study indicated that retriever complex is required for the recycling of numerous membrane proteins and VPS35L suppression decreased expression level of membrane proteins, we compared the expression level of ITGA5 and ITGB1 between wild-type and gene-edited cells. As a result, western blot analysis using membrane fractions showed decreased expression levels of ITGA5 and ITGB1 in both VPS35L-KO and VPS35L[A919T/fs]. Furthermore, similar to patient-derived cells, both VPS35L-KO and VPS35L[A919T/fs] showed decreased expression level of DSCR3, although VPS29 expression level was unchanged (online supplementary file 2, one-way ANOVA with post hoc Tukey’s honestly significant difference calculator test; df=2, F=56.8, p<0.01 for ITGA5, df=2, F=313, p<0.01 for ITGB1, df=2, F=419, p<0.01 for DSCR3 and df=2, F=0.527, p=0.615 for VPS29). Overexpression of wild-type VPS35L in VPS35L-KO cells was able to restore expression levels of ITGB1 and ITGA5, and interestingly, this recovery was also confirmed by VPS35L-A919T expression (online supplementary file 2). These results suggest that VPS35L-A919T is not null, but a hypomorphic mutant due to its instability.

Supplemental material

Figure 3E schematically illustrates how the retriever complex interacts with CCC complex and WASH complex in the endosomal cargo recycling system. CCDC22 and WASHC5, members of CCC and WASH complex, respectively, are genes known to cause 3C/RSS. As described above, compound heterozygous loss-of-function variants in VPS35L were identified in affected individuals. The identified frameshift variant induced non-sense-mediated mRNA decay. Furthermore, the mutant VPS35L-A919T protein does not bind to VPS29, thereby the retriever heterotrimer complex is not formed. Instead, the mutant protein becomes unstable along with DSCR3. Interestingly, the affected siblings showed phenotypic overlap with patients with 3C/RSS. Hence, we concluded that the biallelic variants identified in the VPS35L gene cause 3C/Ritscher-Schinzel syndrome-like features in the affected siblings.

VPS35L is necessary for autophagy

Proper endosomal function is necessary for autophagy, and a VPS35 pathogenic variant was reported to impair autophagy.27 28 Therefore, we next tested whether VPS35L is associated with autophagy or not. We investigated autophagic flux in wild-type and VPS35L knockout HeLa cells. To avoid off-target effect, we designed two kinds of gRNA to make VPS35L-knockout HeLa cell line (online supplementary figure S2A,B). In this context, we analysed autophagic flux assay in two cell lines independently. Autophagic flux was estimated by the amount of LC3-II. LC3-II specifically associates with autophagosomes, the number of LC3-positive vesicles and the amount of LC3-II which represents a specific and sensitive method to make inferences about autophagic activity.29 We incubated cells with the lysosomal inhibitor, BafA1, which causes the accumulation of undegraded LC3-II and Torin 1, a potent and selective mTOR inhibitor.30 31 As in figure 4A,B and online supplementary figure S2C,D, a significant decrease of autophagy flux was detected in VPS35L-knockout cells under nutrient-rich condition (two-sided Student’s t-test; p<0.05). Also, we found a substantial decrease of autophagy flux in all VPS35L-depleted cells after starvation (two-sided Student’s t-test; p<0.05; figure 4C,D, and online supplementary figure S2E,F). Western blotting showed that autophagy flux reduced VPS35L knockout compared with wild-type control cells after treatment with Torin 1 (two-sided Student’s t-test; p<0.05; figure 4E,F, and online supplementary figure S2G,H). Notably, each cell line showed a similar autophagic flux response. These data demonstrated that depletion of VPS35L caused impairment in the autophagy.

Supplemental material

Figure 4

Autophagic functions in VPS35L knockout cells. (A), (C), (E) Representative immunoblots with anti-LC3 antibodies. The amount of LC3-II was quantitated as an indicator of autophagic vacuoles. The difference in LC3-II intensities between the presence and absence of BafA1 represent autophagic degradation activities (flux) during the assay period (2 hours). Cells were incubated in regular medium, EBSS (starvation condition), or 250 nM Torin 1 with or without 125 nM BafA1. (B), (D), (F) The LC3-II protein expression was quantified from three independent sets of immunoblots and normalised with β-actin. Significant decrease of autophagy flux was detected in the VPS35L-knockout cells under nutrient-rich condition (A, B), after starvation (C, D) and after treatment with Torin 1 (E, F). Mean and SE measurements are shown; * p<0.05. BafA1, BafilomycinA1; EBSS, Earle’s Balanced Salt Solution; WT, wild type.

Vps35l−/− mice are embryonic lethal at an early embryonic stage

To elucidate the function of Vps35l during embryonal development, we generated mice deficient in this protein by CRISPR/Cas9 system. The targeting construct for the disruption of mouse Vps35l was designed to make a double-strand break in exon 10, corresponding to the exon where the frameshift variant was detected in the affected siblings (figure 5A). We generated two heterozygous mutant male mice with frameshift variants, one is with a two-base deletion and the other with a two-base deletion and a seven-base insertion (figure 5B). Heterozygous mutant mice appeared normal and fertile (figure 5C,D and online supplementary figure S3). RT-PCR demonstrated that the expression level of mouse Vps35l mRNA in heterozygous mutant mice was reduced to almost half that of wild-type mice (two-sided Student’s t-test, p<0.01; figure 5E). This result is consistent with the frameshift variant inducing mRNA decay. Heterozygotes were then intercrossed to yield homozygous mutant mice. However, to date, no homozygous mutants have been detected among 85 newborn animals from heterozygote crosses, indicating that Vps35l / genotype is embryonic lethal (figure 5F). To determine the stage of embryonic lethality, embryos were dissected for genotyping. At E7.5, we identified Vps35l / embryos, however, no Vps35l / embryos were detected at E10.5 (figure 5F). These results suggested that Vps35l is essential for early embryonic development.

Supplemental material

Figure 5

Generation of Vps35l gene knockout mice. (A) gRNA target sequence for the mouse Vps35l gene knockout by CRISPR/Cas9 gene editing. (B) Genomic DNA sequence chromatograms indicate the frame shift indel #1 with two base deletion and #2 with a two base deletions and seven base insertions, respectively. Red underline indicates PAM sequence. (C) Representative photos of the wild-type (WT, left) and the Vps35l +/ (heterozygous, right) mice at P0. Vps35l +/ mice did not show any characteristic features compared with the WT. (D) Body weight at 4 weeks and 8 weeks was not significantly different between the WT and the Vps35l +/ mice (male at 4 weeks: WT, n=11, Vps35l +/, n=19; female at 4 weeks: WT, n=11, Vps35l +/, n=16; male at 8 weeks: WT, n=11, Vps35l +/, n=16; female at 8 weeks: WT, n=8, Vps35l +/, n=10). (E) Expression of Vps35l mRNA in the WT (n=3) or Vps35l +/ (n=3) mice. Quantitative RT-PCR demonstrated that Vps35l mRNA expression in heterozygous mutant mice was almost half that of wild-type (two-sided Student’s t-test; p<0.01). Total RNA was extracted from cerebellum. (F) Genotype analyses of offspring from Vps35l +/ intercross. Numbers indicate pups born or embryos at different stages of gestation.


This is the first report of germline biallelic loss-of-function variants in VPS35L that cause a 3C/RSS-like congenital malformation syndrome. The siblings in the current study showed a broad phenotypic overlap with 3C/RSS described in the literature, including craniofacial features, cerebellar anomalies and congenital heart defects. In addition, VPS35L, CCDC22 and WASHC5, two other known disease-causing genes for 3C/RSS, are members of the retriever-CCC-WASH axis, which plays a pivotal role in the endosomal membrane protein recycling system (figure 3E). Defects in the retriever-CCC-WASH axis, therefore, could cause congenital malformation syndromes, including 3C/RSS.

The affected siblings described here showed typical features of 3C/RSS, including cranio-cerebello-cardiac complications; however, there are some additional characteristic features, which have not yet been reported in conventional 3C/RSS cases. One such feature is bone and cartilage involvement. Both of the siblings presented tibia-metacarpal or brachytelephalangic type of chondrodysplasia punctata, which has not previously been reported in cases of 3C/RSS syndrome.8–10 25 In addition, the affected boy showed severe hypo-ossification in vertebrae and breast bones, proteinuria and PNH. Furthermore, postnatal growth retardation was more severe than that observed in previously reported cases of 3C/RSS. These discrepancies may illustrate the different impact of the VPS35L defect to the previously reported CCDC22 or WASHC5 deficiency in fetal development; for instance, VPS35L possibly plays more important role in skeletal development. The alternative possibility is the residual function. VPS35L-A919T mutant is shown to be a hypomorphic mutant, and it is possible that less residual function of a mutant protein give rise to more severe clinical presentation. Further analyses are required to clarify the molecular mechanisms underlying the clinical diversity.

Our investigation suggested that VPS35L is necessary for autophagic function. Interestingly, Parkinson’s disease-associated pathogenic variants in VPS35 have been reported to impair autophagy.27 This pathogenic variant was suggested to affect the trafficking of autophagy protein, ATG9A, resulting in autophagic dysfunction. Retriever complex is associated with numerous membrane protein trafficking components, like retromer complex, therefore, it is possible that some autophagy-related protein trafficking impairment reduced autophagic function in VPS35L knockout cells. Further investigation is required to determine what molecule is responsible for autophagic inhibition and at what stage autophagic function is impaired in the VPS35L-deficient condition. These data will provide a greater understanding of the molecular mechanism underlying the disease causes in patients with retriever dysfunction.

Vps35l homozygous knockout mice resulted in embryonic lethality, which suggests Vps35l is essential for the developing fetus. To our knowledge, this is the first reported generation of a Vps35l knockout mouse. VPS35L is a core protein of the retriever complex, which plays a critical role in maintaining the expression levels of numerous integrated membrane proteins. Previous reports suggested that loss of function of the retriever complex reduces expression levels of integrated membrane proteins, which may affect receptor functions. Impairment of autophagy is thought to be one of the phenotypes of retriever complex dysfunction, and a number of signal transduction pathways might be also affected. In this context, we anticipate that the decreased expression levels of membrane proteins associated with a variety of cellular functions cause fetal developmental arrest in the Vps35l−/− mice. Similarly, homozygous deletion of Vps35 in mice was reported to be lethal at an early embryonic stage.32 33 VPS35 is a core protein of the retromer complex, which is structurally and functionally analogous to the retriever complex and it is involved in the recycling of endosomal membrane proteins. The retromer and retriever complexes have been reported to traffic cargo specific for each of them, as well as for both of them.1 34 Collectively, it is suggested that both the retriever and retromer complexes are essential for fetal development.

In conclusion, we have identified VPS35L as a novel causative gene for a congenital malformation syndrome similar to, but distinct from, typical 3C/RSS. VPS35L is suggested to be necessary for autophagy and essential for fetal development. Further molecular studies and clinical information from more cases with mutations in VPS35L and its associated genes are required to uncover the as-yet-unknown biological roles of the retriever complex, such as in fetal development.


The authors thank the patient and his parents for participating in this study. We also thank Yusuke Okuno for technical assistance with exome sequencing.



  • Contributors KK, FFV and TaO performed functional analyses. KK and HO generated the mutant mice. YO, HM, YN, ToO and SS analysed and interpreted the data. KK, HK and YK contributed the clinical data. KK and SS drafted the manuscript. TaO, ToO and YT revised the manuscript.

  • Funding This study was partially supported by JSPS KAKENHI Grant Number JP16K15530 (to SS), JP17H05088 (to ToO) and by the Program for an Integrated Database of Clinical and Genomic Information from the Japanese Agency for Medical Research and Development (to SS).

  • Competing interests None declared.

  • Patient consent for publication Parental/guardian consent obtained.

  • Ethics approval This study was approved by the institutional review board of Nagoya City University Graduate School of Medical Sciences and Nagoya University Graduate School of Medicine.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information.