Bi-allelic loss-of-function variants in JKAMP cause a neurodevelopmental syndrome associated with dysregulation of GPR37 trafficking

Summary

The endoplasmic reticulum (ER) serves as a key hub for protein homeostasis, maintaining a strict quality-control system that ensures only properly folded proteins reach their destinations, while misfolded proteins are degraded via ER-associated degradation (ERAD) or selective ER-phagy. JKAMP, which encodes an ER-resident transmembrane protein involved in ERAD, has not previously been associated with human disease. Here, we report bi-allelic loss-of-function variants in JKAMP in 14 affected individuals from 10 unrelated families presenting with a neurodevelopmental syndrome characterized by intellectual disability, developmental delay, seizures, hypotonia, microcephaly, and dysmorphic features. An in vivo zebrafish model lacking jkamp recapitulated key aspects of the human disorder, including developmental abnormalities and impaired myelin production, further corroborating its pathogenic role. Mechanistic studies identified GPR37, a brain-enriched orphan G protein-coupled receptor (GPCR) and known JKAMP interactor, as a critical downstream effector. GPR37 plays essential roles in dopaminergic signaling, inflammatory pain regulation, neuroprotection, and myelination. Loss of JKAMP resulted in defective folding and degradation of GPR37, leading to its accumulation within the ER and impaired trafficking to the plasma membrane, likely due to impaired ER quality control. These findings establish JKAMP as a previously unrecognized contributor to human neurodevelopment and uncover a pathogenic mechanism linking ER protein quality control to GPCR regulation and neurological disease.

Bi-allelic loss-of-function variants in JKAMP cause a neurodevelopmental syndrome. A zebrafish jkamp knockout recapitulates key disease features. Functional studies show that JKAMP deficiency impairs folding and trafficking of the brain-enriched orphan G protein-coupled receptor GPR37, linking endoplasmic reticulum protein quality-control dysfunction to neurodevelopmental disease.

Introduction

The endoplasmic reticulum (ER) is the most extensive organelle in the cell, playing a critical role in protein synthesis, folding, and transport as well as in lipid synthesis and calcium storage.¹ The ER functions primarily as a site for the synthesis of secreted and integral membrane proteins along with certain cytosolic proteins. Properly synthesized proteins undergo folding and modifications, such as N-linked glycosylation, disulfide-bond formation, and oligomerization, mediated by molecular chaperones and folding enzymes. Once correctly folded, these proteins are released by chaperones, packaged, and transported through the Golgi to their final destinations.^2,3 However, a fraction of proteins fails to achieve their native conformation, becoming misfolded or aggregated. These proteins are either retained in the ER or directed to the ER-associated degradation (ERAD) pathway, ensuring they do not enter the secretory pathway.^3,4,5,6 ERAD is tightly regulated by an ER quality-control system that detects proteins that cannot properly fold or assemble into multiprotein complexes and marks them for ubiquitination-dependent degradation.^2,3,4,5,7

Failure in this process leads to the accumulation of misfolded proteins, triggering the unfolded protein response (UPR).^4,8 Initially, the UPR expands the ER and increases folding capacity but, if proper folding cannot be restored, persistent accumulation of misfolded proteins can result in cell death or apoptosis. Thus, maintaining ER homeostasis is essential for cellular survival.^4,8 Over the last decades, numerous additional proteins have been identified that participate in and contribute to the regulation of ERAD, intervening at various stages of the process.^5,9 Among these, JKAMP (JNK1 [Jun N-terminal kinase 1]-associated membrane protein) has been identified as a participant in ERAD. JKAMP is a ubiquitously expressed protein of 311 amino acids. It associates with JNK1 activation under stress conditions.¹⁰ JKAMP is an ER-resident transmembrane protein¹¹ containing seven predicted transmembrane domains, a zinc-finger domain, and an N-glycosylation site. Based on topology predictions, the C-terminal domain localizes to the cytosol while the N terminus resides within the ER lumen, supporting its role in integrating luminal and cytosolic signaling cues.^10,12

JKAMP participates in ERAD processes by facilitating the clearance of misfolded proteins through its association with multiple components of the proteasome, thereby promoting proteasome recruitment to the ER.^11,12 Through this mechanism, JKAMP forms complexes with ERAD components and helps maintain protein homeostasis.^11,12 JKAMP’s contribution to ERAD may be substrate specific, as initial studies showed that it is effective in promoting the proteasomal degradation of membrane proteins but not of misfolded proteins localized in the ER lumen or in the cytosol.^12,13

Like other multi-spanning proteins, G protein-coupled receptors (GPCRs) generally need to localize to the plasma membrane to be activated by their ligands and initiate signaling pathways. To ensure proper function, they are subject to ER quality control, since only well-folded and matured receptors can traverse the entire secretory pathway to be expressed at the plasma membrane.^14,15,16,17 However, the balance of ER-retained to surface-expressed GPCR varies depending on the type of receptor. GPCRs represent highly valuable pharmacological targets for several reasons, such as their presence on the cell surface, their inherent ability to be activated by small molecules, and their expression patterns being often restricted to specific tissues, which enables the development of tissue-selective therapies.¹⁸ Roy and colleagues showed that JKAMP controls the degradation of specific GPCRs such as the prostaglandin D₂ receptor (DP2),¹⁹ a mediator of inflammation expressed on the membrane of key inflammatory cells,²⁰ and β₂-adrenergic receptor (β₂AR),¹⁹ which, acting on the endothelium and vascular smooth muscle cells, mediates vasodilation, contraction, and angiogenesis.²¹

Using a yeast two-hybrid approach, Sokolina et al. identified interacting partners for 48 clinically relevant full-length human GPCRs, including the G protein-coupled receptor 37 (GPR37).²² Interestingly, in this study, GPR37 was found to interact with JKAMP.²² GPR37 (MIM: 602583) encodes an orphan GPCR expressed almost exclusively in the nervous system, specifically in both neurons and oligodendrocytes (OLs), where its expression is highest.^23,24,25 GPR37 has attracted significant interest over the past 20 years, particularly after it was identified as a substrate of the E3 ubiquitin ligase Parkin.^26,27,28 It seems to be one of the key substrates implicated in the pathogenesis of autosomal-recessive juvenile Parkinson disease (MIM: 600116) caused by Parkin variants.^26,27,28 It also plays crucial roles in the regulation of neuronal and glial physiology, such as dopaminergic signaling,^29,30,31 inflammatory pain,³² and myelination by OLs.^33,34 In particular, genetic deletion of Gpr37 has been found to alter OL differentiation.^33,34 Yang et al. reported that Gpr37 is enriched in OLs and that its expression increases during their differentiation into myelin-forming cells³³ and that genetic deletion of Gpr37 in mice results in precocious OL differentiation and hypermyelination.³³ Smith and colleagues showed that GPR37 interacts directly with the myelin-associated glycoprotein (MAG),³⁴ which is known to promote the stability of myelinated axons,³⁵ and that a deletion of Gpr37 in mice reduces MAG protein levels, rendering OLs more vulnerable to demyelination without affecting the number of both precursors and mature OLs.³⁴ GPR37 undergoes co-translational folding within the ER.³⁶ If proper folding is unsuccessful or excessively delayed, GPR37 is targeted for proteasomal degradation via the ERAD pathway.³⁶

Here, we report 14 individuals with bi-allelic loss-of-function variants in JKAMP (MIM: 611176). They share an overlapping phenotypic spectrum comprising intellectual disability, developmental delay, hypotonia, seizures, microcephaly, and dysmorphic features. We generated a jkamp zebrafish model that recapitulates the phenotype observed in affected individuals, with defects in the forebrain, further supporting the pathogenicity of JKAMP loss. Mechanistic studies identified GPR37 as a critical target of JKAMP-mediated ER quality control. We demonstrate that JKAMP monitors GPR37 degradation and that loss of JKAMP function results in ER retention of GPR37 and reduced cell surface targeting, indicating a disruption in the quality-control processes.

Collectively, our data provide compelling evidence that JKAMP loss-of-function variants result in an autosomal-recessive neurodevelopmental disorder associated with aberrant GPCR trafficking.

Subjects, material, and methods

Ascertainment and study approval

Informed consent was obtained from all participating families. Individuals from families 2 through 10 were identified through clinical evaluation, while individual F1(II-1) was recruited via the Telethon Undiagnosed Diseases Program (TUDP). Collaborative efforts between clinical and research teams were facilitated through the web-based platform GeneMatcher.³⁷ All procedures were conducted in accordance with the ethical standards of the responsible institutional and national committees on human experimentation for all individual studied. Consent for participation and publication was provided by the parents and/or legally authorized representatives of all affected individuals. Written consent for the publication of photographs was obtained for individuals F1(II-1), F2(III-1), F2(III-2), F3(II-4), F4(II-1), F9(II-1), and F10(II-1).

Variant identification and classification

Molecular genetic analyses were conducted across multiple research and clinical diagnostic centers. Detailed methodologies are provided with the supplemental information. JKAMP variants were identified primarily through exome sequencing, except for individual F9(II-1), for whom genome sequencing was utilized. Trio or quad-based sequencing approaches were employed for families 1, 5, 7, 8, and 10, whereas individuals F2(III-2), F3(II-4), F4(II-1), F6(II-1), and F9(II-1) underwent proband-only sequencing. Variant filtering was performed using institution-specific pipelines, which accounted for variant calling quality, predicted functional impact, and familial segregation. Clinical data were collected using standardized templates, completed by participating clinicians. Variant nomenclature was based on the JKAMP transcript GenBank: NM_016475.5. Classification of variants followed American College of Medical Genetics and Genomics (ACMG) guidelines³⁸ and ClinGen recommendations³⁹ using data obtained from Varsome (https://varsome.com/).⁴⁰ In addition, variant pathogenicity was assessed using several publicly available tools, including CADD v1.7,⁴¹ PolyPhen-2,⁴² AlphaMissense,⁴³ REVEL,⁴⁴ and SIFT.⁴⁵ Allele frequencies in the general population were retrieved from the gnomAD database.⁴⁶

Protein structure prediction

The three-dimensional structure of the JKAMP protein was obtained from the AlphaFold Protein Structure Database⁴⁷ (https://alphafold.ebi.ac.uk/entry/Q9P055; AlphaFold: AF-Q9P055-F1). Structural visualization and analysis were performed using PyMOL software⁴⁸ (version 2.5; Schrödinger, New York, NY, USA). The JKAMP structure and polar interactions between amino acids were examined using the mutagenesis wizard tool within PyMOL. The protein structure was optimized, and amino acid residues within a 4-Å distance were highlighted using stick representations. Additionally, the effects of the missense variants on the protein were assessed using in silico tools DynaMut2⁴⁹ and mCSM-membrane.⁵⁰

Zebrafish husbandry and embryo maintenance

Adult zebrafish (Danio rerio) and embryos were maintained at 28.5°C on a 14 h light/10 h dark cycle with sunrise and sunset, with the following water parameters: pH 7.5, conductivity 900 μS. Embryos were raised in egg medium containing methylene blue. Fish were housed in the animal facility at Institut de la Vision (Paris, France), which was built according to the local animal welfare standards. All animal procedures were performed in accordance with French and European Union animal welfare guidelines and were approved by the Committee on Ethics of Animal Experimentation of Sorbonne Université (APAFIS#21323–2019062416186982 and APAFIS #49266–2024050214115020).

Zebrafish jkamp mutants

Generation of jkamp F0 knockout larvae

jkamp F0 knockout larvae refer to wild-type Tüpfel long fin (TL) embryos that were injected at the single-cell stage with gRNA/Cas9 ribonucleoproteins (RNPs) targeting three loci on distinct exons of jkamp. The CRISPR-Cas9 RNP was made of three components purchased from IDT: the crRNA (Alt-R CRISPR-Cas9 crRNA) and tracrRNA (Alt-R CRISPR-Cas9 tracrRNA, 1072533), together forming the gRNA, and the Cas9 (Alt-R S.p. Cas9 Nuclease V3, 1081059). The crRNA is the only component of the RNP specific to the target locus. The three jkamp crRNAs were designed as previously described.^51,52 The CRISPR-Cas9 RNPs were prepared and injected as described in Kroll et al.⁵¹ Control larvae were injected with three non-targeting, “scrambled” RNPs. Knockout larvae were sequenced using Illumina MiSeq to confirm mutagenesis (see supplemental methods). Sequences of the crRNAs are available in Table S1. At 2 days post fertilization (dpf), larvae were placed in grooves made in a 2% agarose plate and photographed with a Leica MZ10-F microscope mounted with a Leica DFC310 FX camera. Morphological analysis used Fiji’s area measurement tool as previously described.⁵³

Whole-mount in situ hybridization

A plasmid encoding mbpa^54,55 was linearized overnight then used to synthesize anti-sense digoxigenin (DIG)-labeled RNA hybridization probes using a DIG RNA Labeling Kit (Roche, 11175025910).

Whole-mount DIG) in situ hybridization was performed according to standard protocols.⁵⁶ Briefly, after injection of CRISPR-Cas9 RNPs, embryos were raised until 4 dpf in egg water with phenylthiourea (0.003%, Sigma P7629) to prevent pigment formation then fixed overnight in 4% paraformaldehyde (Electron Microscopy Sciences, 15710). After permeabilization, the samples were incubated in a solution of DIG-labeled probe overnight at 68°C. Excess probes were removed by several washes. An alkaline phosphatase-conjugated anti-DIG-AP Fab fragments antibody (1:2,000, Roche, 11093274910) was used against DIG and a chromogenic NBT-BCIP substrate (1:50, Roche, 11681451001) was used to detect hybridization. Whole larvae were imaged on a Leica MZ10-F microscope mounted with a Leica DFC310 FX camera.

Quantitative RT-PCR

For each condition, total RNA was extracted from pools of 2-dpf larvae with TRIzol reagent (Invitrogen, 155960) followed by a TURBO DNA-free treatment (Ambion, AM1907). 1 μg of total RNA was reverse transcribed using random hexamers and the SuperScript III First-Strand Synthesis system (Invitrogen, 18080051) according to the manufacturer’s protocol. For quantitative RT-PCR, SYBR Green PCR Master Mix (Applied Biosystems, 43687) was used according to the manufacturers protocol, and the PCR reaction was performed with the QuantStudio 6 PCR System instrument (Applied Biosystems). eef1a1l1 was used as a reference gene as described previously.⁵⁷ All assays were performed in triplicate using 11.25 ng of cDNA per reaction. The mean values of triplicate experiments were calculated according to the deltaCT quantification method. Sequences of the primers are available in Table S1.

Cell culture and treatments

HeLa and SH-SY5Y cell lines were from the American Type Culture Collection (ATCC). Fibroblasts were derived from a skin biopsy from individual F1(II-1). HeLa cells and fibroblasts were cultured in Dulbecco’s modified Eagle’s medium (Gibco, 41965-062) supplemented with 10% FBS, 1% L-glutamine, 1% antibiotics (penicillin/streptomycin), and 1% sodium pyruvate. SH-SY5Y cells were cultured in Dulbecco’s modified Eagle’s medium/Nutrient Mixture F-12 (1:1) (Gibco, 11320-033) supplemented with 10% FBS, 1% L-glutamine, 1% antibiotics (penicillin/streptomycin), 1% sodium pyruvate, and 2.438 g/L sodium bicarbonate buffer. Cells were grown at 37°C and 5% CO₂. Cells were plated the day before to a final confluence of 70% and treated with 100 μg/mL of cycloheximide (CHX) (Sigma, C7698-5G).

Cloning procedures, DNA mutagenesis, and cell transfections

The human JKAMP (GenBank: NM_016475.5) FLAG-tagged open reading frame (ORF) clone construct from Sino Biological (HG17044-CF) was used to generate the following variants in JKAMP: c.110G>A (p.Gly37Glu), c.380A>G (p.Asp127Gly), c.443A>G (p.Glu148Gly). Variants were introduced using the PrimeSTAR Max DNA Polymerase (Takara, R045Q) and the In-Fusion HD Cloning Kit (Takara, 639648) following the manufacturer’s instructions. Mutations were verified by whole plasmid sequencing. Cells (∼200,000/well) were plated into six-well plates for 16 h, transfected with Lipofectamine LTX Reagent with PLUS Reagent (Invitrogen, 15338-100) following the manufacturer’s instructions, and collected 24 h after transfection. The sequences of primers used for mutagenesis in this study are listed in Table S1.

RNAi

100 nM ON-TARGETplus small interfering RNA (siRNA) against JKAMP (SMARTpool, L-016876-02-0010, Dharmacon) or control siRNA (non-targeting pool, D-001810-10-20, Dharmacon) were mixed with Lipofectamine RNAiMAX Reagent (Thermo Fisher Scientific, 13778075) according to the manufacturer’s instructions. Reverse transfection was performed on HeLa cells and maintained for 96 h.

Generation of JKAMP full knockout SH-SY5Y cells using CRISPR-Cas9

Full knockout of JKAMP in SH-SY5Y cells was achieved using the CRISPR-Cas9 system. The guide RNA (gRNA) targeting JKAMP was designed using the CRISPOR online tool (http://crispor.tefor.net/crispor.py). SH-SY5Y cells were electroporated with the gRNA and HiFi Cas9 nuclease (IDT, 1081060) using the Amaxa nucleofection system (Lonza, VCA-1003). Following electroporation, cells were single-cell sorted by fluorescence-activated cell sorting (FACS) into 96-well plates to obtain clonal populations. Genomic DNA was extracted from individual clones, and the targeted region was amplified by PCR. PCR products were analyzed by Sanger sequencing. Clones carrying homozygous insertion/deletion (indel) mutations introducing a premature stop codon were selected and expanded for further studies. Sequences of the gRNA targeting JKAMP and of the PCR primers for the clone screening are available in Table S1.

Western blot

Cells were lysed in RIPA buffer and incubated for 30 min on ice, followed by centrifugation at 15,000 rpm for 15 min at 4°C. Western blotting was performed as previously reported.⁵⁸ Primary antibodies used were anti-β-actin (Sigma, A5441), anti-FLAG (Sigma-Aldrich, F1804), anti-GAPDH (Santa Cruz, 32233), anti-GPR37 (Proteintech, 14820-1-AP), anti-JKAMP (Novus, NBP2-36446SS), anti-Na⁺/K⁺ ATPase (Abcam, ab7671). Horseradish peroxidase (HRP)-conjugated secondary antibody were used (Amersham ECL. NA931, NA934). ImageJ (Gels and Plot Lanes plugin) was used for immunoblotting band quantification.

Quantitative RT-PCR

Total RNA extraction, reverse transcription reactions, and quantitative RT-PCR analysis were performed as previously reported.⁵⁹ Human GAPDH and B2M were used for normalization. Results are shown as means ± SD of at least three independent experiments. Primer sequences are reported in Table S1.

Immunofluorescence

Permeabilized cells were grown on glass coverslips, fixed for 10 min in 4% paraformaldehyde in PBS, washed three times with PBS, and permeabilized for 45 min using 0.05% saponin, 0.5% BSA, 50 mM NH₄Cl in 1× PBS (blocking buffer), as previously described.⁶⁰ Cells were then incubated with the following primary antibodies: anti-Calnexin (Santa Cruz, 23954), anti-FLAG (Sigma-Aldrich, F1804), anti-GPR37 (Proteintech, 14820-1-AP), anti-LAMP1 (R&D Systems, AF4800), anti-Na⁺/K⁺ ATPase (Abcam, ab7671), anti-vesicle-associated membrane protein-associated protein A (VAPA) (in-house M.A.D.), and anti-vesicle-associated membrane protein-associated protein B (VAPB) (Proteintech, 66191-1-Ig). Secondary antibodies conjugated to Alexa Fluor 488 or Alexa Fluor 594 (Invitrogen, A32766; Invitrogen, A322754) were used. DNA was stained with Hoechst (Sigma-Aldrich, 33342). After three PBS washes, coverslips were mounted using Mowiol (Millipore, 475904).

Non-permeabilized cells were grown on glass coverslips and incubated on ice for 1 h with primary antibodies: anti-GPR37 (Proteintech, 14820-1-AP), anti-LAMP1 (R&D Systems, AF4800), and anti-Na⁺/K⁺ ATPase (Abcam, ab7671). Cells were then fixed with 2% paraformaldehyde in PBS for 15 min at room temperature and washed three times with 50 mM NH₄Cl in PBS. Blocking was performed for 30 min at room temperature using a solution containing 5% FBS, 20 mM PIPES (pH 6.3), 137 mM NaCl, 27 mM KCl, and 50 mM NH₄Cl. Slides were incubated with appropriate secondary antibodies conjugated to Alexa Fluor 488, 594, 647, (Invitrogen, A32766; Invitrogen, A322754; Invitrogen, A21448) or with wheat germ agglutinin (WGA) Alexa Fluor 488 conjugate (Invitrogen, W11261). After three washes with 50 mM NH₄Cl in PBS, coverslips were mounted with Mowiol (Millipore, 475904). All the experiments were performed at least three times.

Confocal fluorescence microscopy, image processing, and fluorescence quantification

Cells were imaged using a Plan-Apochromat 63× or 100×/1.4 oil objective on a Zeiss LSM800 confocal system and controlled by the Zen blue software. For quantitative immunofluorescence, 10–15 fields (10–20 cells per field) were randomly selected for each experiment, and 90 cells were analyzed per condition. Fluorescence micrographs are representative of images collected from at least three independent experiments. The images used for phenotype quantification were acquired with the same parameters (i.e., digital gain, laser power, magnification) and analyzed with the Fiji (ImageJ; National Institutes of Health) software. To quantify the level of colocalization of different proteins, confocal serial sections were acquired from 90 cells per experimental condition, using a 0.5-μm z-interval (for a total stack thickness varying between 2.5 and 4.5 μm depending on the cell). Briefly, image stacks were processed to obtain the maximum projection and reconstructed in Zen software. Then, an intensity threshold of 100 (in a range of 0–65536) for each analyzed channel was applied. The percentage of colocalization represents the percentage of total pixels in the reference channel that were positive for the other channel, and that percentage was calculated using the Zen software colocalization tool. Brightness and contrast were adjusted with Adobe Photoshop, and figure panels were assembled with Adobe Illustrator (Adobe).

Plasma membrane isolation

Plasma membranes were isolated using a plasma membrane protein extraction kit (Abcam, ab65400) according to the manufacturer’s instructions. Briefly, cells were lysed on ice in homogenization buffer supplemented with protease inhibitors. Homogenates were centrifuged at 700 × g for 10 min at 4°C to remove nuclei and debris. The resulting supernatant was centrifuged at 10,000 × g for 30 min at 4°C to pellet membrane fractions (plasma membrane and organellar membranes), leaving the cytosolic fraction in the supernatant. For further enrichment of the plasma membrane, pellets were resuspended in the upper-phase buffer and mixed with the lower-phase buffer, then centrifuged at 1,000 × g for 5 min at 4°C. The upper phase (plasma membrane enriched) was diluted 1:5 with ice-cold water and incubated at 4°C overnight. The plasma membrane fraction was finally collected by centrifugation at 21,000 × g for 10 min at 4°C. Pellets were solubilized in 0.1% SDS in PBS for western blotting.

Statistical analysis

All statistical analyses were performed in GraphPad Prism 7.⁶¹ When comparing two samples, unpaired t tests or Wilcoxon-Mann-Whitney tests were used, while one-way ANOVA tests were used when comparing more groups. The multiple unpaired t test was performed for pairwise group comparison. Details of statistical analysis and exact values of numbers quantified in each experiment can be found in the figure legends.

Results

Bi-allelic variants in JKAMP are associated with intellectual disability, developmental delay, epilepsy, hypotonia, microcephaly, and dysmorphisms

Fourteen individuals affected by a neurodevelopmental disorder were identified as harboring bi-allelic variants in JKAMP. Nine individuals were male and five were female. They ranged in age from 18 months to 25 years. The cohort included individuals from two major ethnic backgrounds, Arab and European, based on self-reported family origin and grouped into broad ancestral categories. All affected individuals, except for families F1 and F8, came from consanguineous parents (F2, F3, F4, F5, F6, F7, F9, and F10). Family 2 (F2) has two affected members (F2[III-1], F2[III-2]), F6 has two (F6[II-1], F6[II-2]), and F7 has three (F7[II-1], F7[II-2], F7[II-3]) affected individuals. Variants are shown for the GenBank: NM_016475.5 transcript and represent homozygous or compound heterozygous states with the respective protein changes in each individual: F1(II-1) (c.380A>G [p.Asp127Gly]; c.562C>T [p.Arg188Ter], compound heterozygous); F2(III-1) and F2(III-2) (c.871dup p.Tyr291[LeufsTer27], homozygous); F3(II-4) (c.243dup [p.Lys82GlufsTer16], homozygous); F4(II-1) (c.243dup [p.Lys82GlufsTer16], homozygous); F5(VI-3) (c.243dup [p.Lys82GlufsTer16], homozygous); F6(II-1) and F6(II-2) (c.110G>A [p.Gly37Glu], homozygous); F7(II-1), F7(II-2), and F7(II-3) (c.243dup [p.Lys82GlufsTer16], homozygous); F8(II-1) (c.443A>G [p.Glu148Gly]; c.878del p.[Phe293SfsTer34], compound heterozygous); F9(II-1) (c.640 + 1G>T [p.?], homozygous); and F10(II-1) (c.458 + 1G>C [p.?], homozygous). All affected individuals shared moderate-to-profound neurodevelopmental delay and intellectual disability. Speech was particularly impaired, with six subjects (F2[III-1], F2[III-2]), F6[II-1], F7[II-2], F7[II-3], and F9[II-1]) being non-verbal at their most recent assessment and the remaining only capable of pronouncing a few words. Neurodevelopmental regression was observed in five of the 14 affected individuals (F1[II-1], F7[II-2], F7[II-3]), F8[II-1], and F9[II-1]). Early-onset epilepsy was observed in all individuals (median age at first seizure: 6.5 months). In six subjects, epilepsy initially presented with febrile seizures (F1[II-1], F2[III-1], F2[III-2], F3[II-4], F4[II-1], and F5[VI-3]), later progressing to non-febrile, complex seizures. Electroencephalogram (EEG) findings were normal only in F8(II-1). In the affected individual F3(II-4), antiepileptic therapy resulted in normalization of the abnormal EEG. EEG abnormalities in the remaining individuals were predominantly nonspecific. Sudden death was reported in three individuals (F5[VI-3], F6[II-2], and F7[II-1]); in at least two of these cases (F6[II-2] and F7[II-1]), death was associated with seizures or status epilepticus. A fourth individual (F6[II-1]) experienced three episodes of cardiopulmonary arrest but was successfully resuscitated. Hypotonia was documented in 13 affected individuals. Ocular involvement occurred in one-third of individuals, manifesting as refractive errors (astigmatism in individual F1[II-1]), strabismus (individuals F1[II-1], F3[II-4], and F4[II-1]), nystagmus (individual F9[II-1]), and complex malformations (individual F7[II-2] showed bilateral microphthalmia, cataracts, retinal hypoplasia, and optic nerve enlargement). Microcephaly was reported in five individuals (F1[II-1], F2[III-1], F6[II-1], F7[II-2], and F7[II-3]), although available data did not always allow distinction between congenital and acquired forms. Other auxological parameters (height and weight) were within the normal ranges. Cardiac malformations and arrhythmias were reported in three subjects (F1[II-1], F7[II-1], and F9[II-1]). All individuals were found to have normal hearing except for F1(II-1) who had mild sensorineural hearing loss. Additional skeletal abnormalities, including mild scoliosis, structural vertebral anomalies (F5[VI-3]), and tapering fingers, pes cavus, talipes calcaneovarus, syndactyly, sandal gap (F8[II-1]), and fifth-digit clinodactyly (F10[II-1]) were identified. Genitourinary malformations were observed in three individuals: cryptorchidism and hypospadias in F6(II-1) and F6(II-2), and micropenis in F7(II-3). From a dysmorphological perspective, affected individuals had several shared craniofacial features including round face with full cheeks, synophrys, epicanthus, bulbous nasal tip, long and smooth philtrum, full lips/open mouth with everted lower lip, and widely spaced teeth. Representative facial photographs of a subset of affected individuals are shown in Figure 1A.

Figure 1.

Clinical and genetic features of JKAMP neurodevelopmental syndrome

(A) Frontal and lateral photographs of affected individuals from different families and ages: F1(II-1) at 16 months (a and b) and 8 years (c and d), F2(III-1) at 9 years (e) and 11 years (f), F2(III-2) at 5 years (g) and 7 years (h), F3(II-4) at 4 years (i), F4(II-1) at 2 years (j and k), F9(II-1) at 6 months (l) and 1 year (m),and F10(II-1) at 4 years (n). Frequent facial features include a round face, full cheeks, depressed nasal bridge, bulbous nasal tip, bilateral epicanthus, synophrys, peaked eyebrows, and low-set ears.

(B) Brain MRI of F2(III-1) at 14 months (a and b) showing generalized brain atrophy (widening of CSF spaces) and reduced volume of the white matter on T2-weighted image. MRI of F2(III-2) at 26 months (c), T2 sequence, demonstrating more prominent gray matter atrophy (further widening of CSF spaces), reduced white matter volume with deepened sulci, and increased white matter signal. Brain MRI of F4(II-1) at 15 months (d–f): axial T1-weighted images (T1W) (d and e) and axial T2-weighted image (T2W) (f) show mild to moderate delayed myelination for age, with occipital and parietal subcortical white matter appearing poorly hyperintense on T1W and showing incomplete myelination on T2W in the subcortical white matter of the calcarine areas and anterior limbs of the internal capsule. Bilateral choroid plexus cysts are also noted (f).

(C) Pedigree of 14 affected individuals from 10 unrelated families.

(D) Schematic representation of the human ER-resident protein JKAMP, illustrating the seven transmembrane domains and the JKAMP variants identified in our cohort (referenced to transcript GenBank: NM_016475.5), shown as homozygous (HOM) or compound heterozygous (CH): F1(II-1) (p.Asp127Gly, p.Arg188Ter – CH), F2(III-1) and F2(III-2) (p.Tyr291LeufsTer27 – HOM), F3(II-4) (c.243dup [p.Lys82GlufsTer16] – HOM), F4(II-1) (c.243dup [p.Lys82GlufsTer16] – HOM), F5(VI-3) (c.243dup [p.Lys82GlufsTer16] – HOM), F6(II-1) and F6(II-2) (c.110G>A [p.Gly37Glu] – HOM), F7(II-1), F7(II-2), and F7(II-3) (c.243dup [p.Lys82GlufsTer16] – HOM), and F8(II-1) (c.443A>G; c.878del [p.Glu148Gly]; p.Phe293SfsTer34 – CH). Family (F) numbers are indicated.

(E) Amino acid sequence alignment for various JKAMP vertebrate orthologs in the region of interest. H. sapiens (RefSeq: NP_057559.2), P. troglodytes (RefSeq: XP_063649549.1), M. mulatta (RefSeq: NP_001247922.1), C. lupus (RefSeq: XP_038529696.1), M. musculus (RefSeq: NP_077167.1), R. norvegicus (RefSeq: NP_001100208.1), G. gallus (RefSeq: NP_001264483.1), D. rerio (RefSeq: NP_956776.2), X. tropicalis (RefSeq: XP_002933780.2), O. latipes (RefSeq: XP_020569534.1). Gray residues vary from human reference. Yellow residues are those affected by the variants described (Gly37, Asp127, and Glu148).

Magnetic resonance imaging (MRI) scans revealed structural abnormalities of the brain in most affected individuals: cortical and/or cerebral atrophy was observed in 11/14 (78.6%) individuals and delayed myelination in 6/14 (43%). MRI scans were unremarkable in only two subjects (F1[II-1] and F9[II-1]). One subject underwent brain computed tomography (CT) scan, which was normal (F8[II-1]). Figure 1B shows the brain MRI of individuals F2(III-1)/F2(III-2) and F4(II-1). The MRI of individual F2(III-1) at 14 months (a and b) displayed a generalized brain atrophy with widening of cerebrospinal fluid (CSF) spaces and white matter hypomyelination, along with deepened sulci. The MRI of individual F2(III-2) at 26 months (c) revealed more pronounced gray matter atrophy, further widening of CSF spaces, reduced white matter volume, and increased white matter signal. The MRI of individual F4(II-1) at 15 months (d–f) showed mild to moderate delayed myelination, with poorly hyperintense occipital and parietal subcortical white matter and bilateral choroid plexus cysts.

Phenotypic and genetic information of all affected individuals is shown in Figures 1A and 1B, and can be found in Table 1. Figure 1C shows the pedigree of each family. Figure 1D provides a schematic representation of the ER-resident protein JKAMP, highlighting its seven transmembrane domains and the variants identified in our cohort. Details on JKAMP variants are reported in Figure 1D and Table 2. For additional details, please refer to Supplemental Note: Case Reports, Molecular Studies, and Table S2 contained within the supplemental information. The alignment of protein sequences in Figure 1E shows the evolutionary conservation of the residues that are mutated in the JKAMP missense variants p.Gly37Glu, p.Asp127Gly, and p.Glu148Gly.

Table 1.

Loss-of-function variants in JKAMP cause a neurodevelopmental disorder, featuring intellectual disability, developmental delay, epilepsy, hypotonia, microcephaly, and dysmorphisms

Individual	F1(II-1)	F2(III-1)	F2(III-2)	F3(II-4)	F4(II-1)	F5(VI-3)	F6(II-1)	F6(II-2)	F7(II-1)	F7(II-2)	F7(II-3)	F8(II-1)	F9(II-1)	F10(II-1)	Total
Sex	female	female	male	male	male	male	male	male	female	female	male	female	male	male	–
Ethnicity	European	Arab	Arab	European	Arab	European	Arab	Arab	European(Mennonite)	European (Mennonite)	European (Mennonite)	European	Arab	European	–
Current age	10 years	15 years	11 years	8 years	3 years	deceased	8 years	deceased	deceased	25 years	23 years	23 years	18 months	4 years	–
Age at last examination	6 years	9 years	5 years	6 years	20 months	16 years	7 years	10 months	3 years	11 years	9 years	22 years	17 months	4 years	–
gDNA (Hg38/GRCh38)	chr14:59495146:A>G; chr14:59498830:C>T	chr14:59504001:C>CT	chr14: 59504001:C>CT	chr14:59487816:C>CG	chr14:59487816:C>CG	chr14:59487816:C>CG	chr14:59487687:G>A	chr14:59487687:G>A	chr14:59487816:C>CG	chr14:59487816:C>CG	chr14:59487816:C>CG	chr14:59495209:A>G; chr14:59504012:GT>G	chr14:59498909:G>T	chr14:59495225:G>C	–
cDNA (GenBank: NM_016475.5)	c.380A>G; c.562C>T	c.871dup	c.871dup	c.243dup	c.243dup	c.243dup	c.110G>A	c.110G>A	c.243dup	c.243dup	c.243dup	c.443 A>G; c.878del	c.640 + 1G>T	c.458 + 1G>C	–
Protein (RefSeq protein: NP_057559)	p.Asp127Gly; p.Arg188Ter	p.Tyr291LeufsTer27	p.Tyr291LeufsTer27	p.Lys82GlufsTer16	p.Lys82GlufsTer16	p.Lys82GlufsTer16	p.Gly37Glu	p.Gly37Glu	p.Lys82GlufsTer16	p.Lys82GlufsTer16	p.Lys82GlufsTer16	p.Glu148Gly; p.Phe293SfsTer34	p.(?)	p.(?)	–
Variant type	missense; nonsense	frameshift indel	frameshift indel	frameshift indel	frameshift indel	frameshift indel	missense	missense	frameshift indel	frameshift indel	frameshift indel	missense; frameshift indel	splicing	splicing	–
Inheritance	comp het	hom	hom	hom	hom	hom	hom	hom	hom	hom	hom	comp het	hom	hom	–
MD	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	14/14
Speech delay	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	14/14
Intellectual disability	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	14/14
Seizures	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	14/14
Hypotonia	yes	yes	yes	no	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	13/14
CA	yes	no	no	no	no	ND	no	no	yes	no	no	no	yes	no	3/14
Brain MRI	normal (hemosiderin deposit)	mild cortical atrophy, delayed myelination	mild cortical atrophy, delayed myelination	mild cerebral atrophy	global brain atrophy, delayed myelination, bilateral choroid plexus cysts	mild dilatation of frontal liquor spaces, delayed myelination	global cerebral atrophy	global cerebral atrophy	diffuse cerebral atrophy	global cerebral atrophy and diffuse demyelinating disease	mild atrophy	ND	normal	mild cortical atrophy, delayed myelination	11/14
Head circumference (SD)	birth: −0.6; 2 years: −4; 6 years: −4	9 years: −3.1	5 years: −1.6	6 years: −1.25	birth: 0; 20 months: +0.2	16 years: −0.41	1 year: +1.6; 3 years 6 months: −2.2	ND	ND	birth: −1.16; 11 years 8 months: −7.2	birth: +0.03; 9 years 9 months: −3.1	22 years: −1.3	birth: +2; 17 months: +0.9	4 years: −1.6	–
Visual impairment	yes	no	no	yes	yes	no	no	no	no	yes	no	no	yes	no	5/14
Skeletal abnormalities	no	no	no	no	no	yes	no	no	no	no	no	yes	no	yes	3/14
Facial dysmorphisms	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	14/14

ND, not defined; comp het, compound heterozygous; hom, homozygous; MD, motor delay; CA, cardiac anomalies. A dash (−) in the last column indicates data not applicable.

Table 2.

Details of the missense JKAMP variants

Individual	Variant	Variant type	gnomAD allele frequency	SIFT		Polyphen-2		AlphaMissense		CADD		REVEL
				Score	PE	Score	PE	Score	PE	Score	PE	Score	PE
F1(II-1)	c.380A>G (p.Asp127Gly)	missense	NR	0	PP3-moderate	1	PP3-moderate	0.99	P	28	PP3-supporting	0.8	PP3-moderate
F6(II-1) F6(II-2)	c.110G>A (p.Gly37Glu)	missense	NR	0	PP3-moderate	1	PP3-moderate	0.95	P	29	PP3-moderate	0.8	PP3-moderate
F8(II-1)	c.443A>G (p.Glu148Gly)	missense	NR	0	PP3-moderate	0.99	PP3-moderate	0.99	P	26	PP3-supporting	0.8	PP3-moderate

PE, prediction effect; P, pathogenic; NR, not reported. AlphaMissense scores range from 0 (benign) to 1 (pathogenic). PP3-moderate and PP3-supporting are assigned as previously suggested.³⁹

Details of the missense variants identified in the current cohort of affected individuals. Variants are reported according to transcript GenBank: NM_016475.5.

Pathogenic missense variants in JKAMP cause structural alterations and self-aggregation

JKAMP is a membrane-associated protein predicted by AlphaFold⁴⁷ (AlphaFold: AF-Q9P055-F1 to contain seven transmembrane domains composed of α helices, flexible loops, and some β sheets (Figure 2A). Structural modeling of the missense variants p.Gly37Glu, p.Asp127Gly, and p.Glu148Gly revealed distinct conformational alterations (Figure 2A). Specifically, the p.Gly37Glu variant resulted in new polar interactions between Glu37, Glu110, and Tyr151. In contrast, the p.Asp127Gly and p.Glu148Gly variants led to a loss of critical interactions: Gly127 retained interaction only with Met124, while Gly148 maintained contacts only with Ala149 and Tyr151. Pathogenicity prediction and structural modeling analysis using DynaMut2⁴⁹ and mCSM-membrane⁵⁰ tools indicated that all the three variants are probably pathogenic and disruptive to the native JKAMP structure, with predicted changes in Gibbs free energy (ΔΔG) supporting a destabilizing effect. For p.Gly37Glu, the ΔΔG was −1.41 kcal/mol according to DynaMut2 and –0.19 kcal/mol using mCSM-membrane. The p.Asp127Gly variant showed a ΔΔG of −0.41 kcal/mol (DynaMut2) and −0.66 kcal/mol (mCSM-membrane). In addition, p.Glu148Gly was associated with a ΔΔG of −1.53 kcal/mol and −0.51 kcal/mol in the respective prediction tools. Overall, these findings suggest that variants in JKAMP compromise the structural integrity of the JKAMP protein by disrupting key polar interactions.

Figure 2.

Missense variants in JKAMP induce conformational changes and self-aggregation

(A) Predicted impact of JKAMP variants on protein structure. Top: 3D AlphaFold-predicted JKAMP structure (AlphaFold: AF-Q9P055-F1) highlighting wild-type amino acids Gly37, Asp127, and Glu148. α helices in red, β sheets in yellow, and loops in green. Bottom: 4-Å zoomed-in focuses of both wild-type (WT) and mutated residues (p.Gly37Glu, p.Asp127Gly, p.Glu148Gly), with polar interactions indicated by red dashed lines.

(B) Representative confocal images of HeLa cells transfected with FLAG-tagged JKAMP constructs (WT, (p.Gly37Glu, p.Asp127Gly, or p.Glu148Gly), stained with anti-FLAG and anti-VAPA (ER marker) antibodies. The panels on the right show magnifications and single-color channels of the boxed areas. Green, anti-FLAG; red, anti-VAPA. Scale bar, 20 μm.

Immunofluorescence in HeLa cells showed that wild-type JKAMP localizes to the ER, as previously reported,^10,12,19 with clear colocalization with the ER marker VAPA (Figure 2B). In contrast, mutant JKAMP constructs carrying p.Gly37Glu, p.Asp127Gly, or p.Glu148Gly substitutions formed prominent JKAMP self-association aggregates (Figure 2B). In F1(II-1)-derived fibroblasts (compound heterozygous p.Asp127Gly; p.Arg188Ter), JKAMP protein levels were markedly reduced compared to controls (Figure S1A). Western blot analysis detected only the band corresponding to the missense variant p.Asp127Gly, while no bands were observed for the truncated JKAMP protein p.Arg188Ter (∼188 amino acids; ∼15 kDa) despite the antibody targeting an epitope between amino acids 20–150, suggesting nonsense-mediated mRNA decay (Figure S1A). This was further supported by the significant reduction in JKAMP mRNA levels (Figure S1B).

Zebrafish jkamp mutants resemble the human disease

To investigate in vivo the causal link between JKAMP variants and the defects observed in the affected individuals, we generated jkamp F0 knockout (crispant) zebrafish larvae. Using CRISPR-Cas9, we mutated the gene at three distinct exons (Figure S2), according to well-established protocols.⁵¹ The Cas9/guide RNA RNPs mutated virtually every copy of the genome at each exon (99% ± 2% of reads had a small indel) causing a majority of frameshift mutations (75% ± 34% of all reads had a frameshift mutation), cumulatively resulting in an estimated 97% probability of bi-allelic knockout (francoiskroll.shinyapps.io/frameshiftmodel/). Approximately half (54%, n = 80/148) of the mutant larvae appeared broadly similar to scrambled-injected control larvae, while the other half (46%, n = 68/148) had gross morphological defects. The presence of these two populations, which we termed “mild” and “severe” phenotypes, likely reflects genuine variability in the phenotype, akin to that observed in the affected individuals, rather than variability in knockout rates. Indeed, it was possible for a larva to be mildly affected but carry ∼100% frameshift mutations in jkamp (Figure S2).

To quantify those morphological phenotypes, we measured the head and eye sizes of the jkamp knockout larvae (Figure 3A). Compared to control larvae, larvae from both mild and severe phenotype populations had smaller eyes (mild: −31%; severe: −55% vs. scrambled) and heads (mild: −31%; severe: −47% vs. scrambled), recapitulating the microcephaly observed in a subset of individuals presenting JKAMP variants (Figure 3A). Since delayed myelination was a common finding from MRI scans of affected individuals, we tested whether this is the case in crispant mutants. We analyzed myelination in the jkamp knockout larvae by measuring mbpa (myelin basic protein a) mRNA,⁶² which encodes a structural component of myelin sheath.⁶³ On whole-mount in situ hybridization, mbpa expression was high along cranial nerves, the lateral line, and in a central population of OLs within the hindbrain (Figure 3B), as previously described.⁶² Expression was fainter in jkamp knockout larvae with the mild phenotype and minimal in those with the severe phenotypes (Figure 3B), likely mimicking the delayed myelination observed in affected individuals. To confirm the reduction in mbpa expression, we performed quantitative RT-PCR on mRNA extracted from whole jkamp knockout larvae. mRNA levels of two genes highly expressed in OL precursors—nkx2.2a and olig2—were unchanged in jkamp knockouts, but mbpa expression was reduced 3-fold and 7-fold in mutants of mild and severe phenotype, respectively (Figure 3C).

Figure 3.

Loss of zebrafish jkamp causes microcephaly and loss of myelin

(A) Top: lateral views of a scrambled-injected and two jkamp F0 knockout (KO) larvae (2 dpf) of mild and severe phenotypes, with the eye (orange) and head areas (black) outlined. Scale bar, 200 μm. Bottom: morphological measurements from uninjected, scrambled-injected, and jkamp F0 KO with mild and severe phenotypes. ^∗∗∗∗p ≤ 0.0001, ^∗p ≤ 0.05; ns, not significant by t tests.

(B) In situ hybridization analyses of mbpa expression in 4-dpf scrambled-injected and jkamp F0 KO larvae. mbpa mRNA is significantly reduced throughout the nervous system of jkamp F0 KO larvae in comparison to scrambled-injected controls. Insets highlight mbpa expression in OLs within the hindbrain.⁶⁴ Scale bar, 100 μm. ^∗∗∗∗p ≤ 0.0001; ns, not significant by t tests.

(C) Expression of nkx2.2a, olig2, and mbpa genes measured by quantitative RT-PCR. Expression was normalized within each group to eef1a1l1 expression used here as housekeeping gene. n = 4 pools of 25–30 larvae for each group. Measurements were repeated on an independent batch of larvae; means ± SEM; ^∗∗∗∗p ≤ 0.0001; ns, not significant by t tests.

Overall, bi-allelic loss-of-function mutations in zebrafish jkamp also causes morphological defects, including microcephaly, and reduced myelination.

JKAMP deficiency impairs GPR37 trafficking to the plasma membrane

These genetic and in vivo findings prompted us to investigate the mechanistic basis by which the absence of JKAMP perturbs cellular homeostasis. Among the interactors of JKAMP identified using the BioGRID database⁶⁴ (https://thebiogrid.org; BioGRID: 119590), GPR37 emerged as one of the most compelling candidates. GPR37 is an orphan receptor highly expressed in neurons and OLs,^23,24,25 and it has been extensively studied over the past two decades for its role in autosomal-recessive juvenile Parkinson disease.^26,27,28 Sokolina et al., using a yeast two-hybrid approach, demonstrated that GPR37 binds to JKAMP.²² Similarly to other GPCRs, GPR37 is synthesized and initially processed in the ER, where it undergoes proper folding and quality control before being transported to the plasma membrane.³⁶

We examined how loss of JKAMP affects GPR37 maturation and trafficking. To determine whether GPR37 reaches the plasma membrane, we performed live-cell surface staining under non-permeabilizing conditions, to preserve membrane integrity and detect only surface-exposed proteins, both in siRNA-mediated JKAMP-depleted HeLa cells (siJKAMP) and in F1(II-1) fibroblasts, with their respective controls. We validated membrane integrity using LAMP1 (lysosomal marker) staining as a negative control, as the antibody used to recognize it has been raised against a cytosolic portion of LAMP1 and the Na⁺/K⁺ ATPase or WGA as plasma membrane markers (Figure 4). Quantification of surface GPR37 revealed significantly reduced plasma membrane localization in both siJKAMP HeLa cells (83% Figure 4A) and F1(II-1) fibroblasts (about 52%, Figure 4B) compared to controls, indicating defective trafficking.

Figure 4.

JKAMP loss reduces GPR37 localization at the plasma membrane

(A) Left: representative confocal images of endogenous GPR37 localized at the plasma membrane (PM) in siJKAMP and control (Ctrl−) HeLa cells. Wheat germ agglutinin (WGA) was used to label the PM for immunofluorescence staining. Insets show the absence of cytosolic components, as indicated by the lack of LAMP1 lysosomal staining. Red, GPR37; green, WGA; blue, LAMP1. Scale bar, 10 μm. Right: quantification of amounts of GPR37 localized at PM normalized on WGA immunofluorescence signal is expressed as percentage of Ctrl− (mean ± SD), in siJKAMP and Ctrl− HeLa cells. Three independent experiments, n > 30 cells per experiment; n = 3 independent experiments. Wilcoxon-Mann-Whitney test, ^∗∗∗∗p ≤ 0.0001.

(B) Representative confocal images of endogenous GPR37 localized on PM in F1(II-1)-derived fibroblasts and control (Ctrl). Plasma membrane is stained with the Na⁺/K⁺ ATPase. Insets show the absence of cytosolic components, as indicated by the lack of LAMP1 lysosomal staining. Red, GPR37; green, Na⁺/K⁺ ATPase; blue, LAMP1. Scale bar, 10 μm. Right: quantification of amounts of GPR37 localized at PM. Three independent experiments, n > 30 cells per experiment; n = 3 independent experiments, means ± SD. Wilcoxon-Mann-Whitney test, ^∗∗∗∗p ≤ 0.0001.

(C) Left: western blot of GPR37 and Na⁺/K⁺ ATPase in the total lysate (Tot) from JKAMP KO and WT SH-SY5Y cells. Actin was used as a loading control. Middle: western blot of GPR37 and Na⁺/K⁺ ATPase in the PM fraction, isolated by density gradient. Right: quantification of amounts of GPR37 associated with PM fraction in JKAMP KO, normalized to Na⁺/K⁺ ATPase and expressed as fold change relative to WT. Scale bar, 10 μm.

Kruskal-Wallis statistics, n = 3 independent experiments, means ± SD. ^∗p ≤ 0.05. PM, plasma membrane; Tot, total lysate.

To biochemically validate these observations, we performed plasma membrane protein isolation in SH-SY5Y neuroblastoma cells with a complete JKAMP knockout (JKAMP KO) and control, followed by western blot analysis. We found significantly reduced GPR37 protein levels in the plasma membrane fraction of JKAMP KO cells (59%; Figure 4C), supporting the notion that JKAMP is required for efficient GPR37 trafficking. These biochemical results were also confirmed in both siJKAMP HeLa cells and F1(II-1) fibroblasts (Figure S3).

To determine whether this trafficking defect was restricted to GPR37 or reflected a broader impairment of ER function, we evaluated the secretion of procollagen type I (PCI), a large protein that serves as a readout for ER function. Collagen biosynthesis requires extensive folding, post-translational modifications, and proper trafficking through the ER-Golgi pathway.⁶⁵ Specifically, we assessed the secretion of PCI in control and F1(II-1) fibroblasts, as defects in ER folding or export would likely impair its processing and release. Cells were treated with ascorbic acid to promote procollagen folding via proline hydroxylation and CHX to block new protein synthesis. Western blot analysis of cell lysates and conditioned media collected at 0 and 3 h post treatment showed that PCI was efficiently secreted in both F1(II-1) and control cells, without any delay (Figure S4). These findings indicate that general ER trafficking remains functional in F1(II-1) fibroblasts, as demonstrated by the normal secretion of PCI (Figure S4).

Having established that delivery of GPR37 to the plasma membrane is impaired in JKAMP-deficient cells, we next investigated where the receptor accumulates within the cell. To map GPR37 subcellular localization, we performed immunofluorescence using ER markers. In wild-type cells, GPR37 showed modest colocalization with the ER markers VAPB and Calnexin, consistent with its transient ER residence, whereas overlap with the ER increased in JKAMP-deficient cells relative to controls (Figures 5A and 5B). Indeed, in siJKAMP HeLa cells, we observed a significant increase of 22% in ER-localized GPR37 compared to control cells (Figure 5A). This result was further confirmed in F1(II-1)-derived fibroblasts, which showed a 20% increase relative to controls (Figure 5B). These findings suggest that loss of JKAMP leads to ER retention of GPR37.

Figure 5.

Loss of JKAMP leads to ER retention of GPR37

(A) Left: representative confocal images of endogenous GPR37 and the ER marker VAPB colocalization in siJKAMP and in the negative control (Ctrl−) HeLa cells. Scale bar, 10 μm. Arrows indicate GPR37/VAPB-positive structures. Green, GPR37; red, VAPB. Right: quantification of GPR37/VAPB colocalization, expressed as a percentage of Ctrl−in siJKAMP and Ctrl− (mean ± SD).

(B) Left: representative confocal images of endogenous GPR37 and the ER marker Calnexin (CNX) colocalization in F1(II-1)-derived fibroblasts and control (Ctrl). Scale bar, 10 μm. Arrows indicate GPR37/CNX positive structures. Green, GPR37; red, CNX. Right: quantification of GPR37/CNX colocalization, expressed as percentage of Ctrl (mean ± SD) in F1(II-1) fibroblasts and Ctrl. The panels on the right show magnifications and single-color channels of the boxed areas. Arrowheads indicate colocalization.

Data are shown as mean ± SD, n > 30 cells per experiment; n = 3 independent experiments. Statistical analysis was performed using Wilcoxon-Mann-Whitney test, ^∗∗∗∗p ≤ 0.0001.

JKAMP controls GPR37 turn over

Given JKAMP’s role in ERAD as a proteasome recruiting factor,^10,12 the increased ER retention of GPR37 would be consistent with defective folding and reduced proteasomal clearance. We therefore hypothesized that the loss of JKAMP would alter GPR37 degradation. To test this hypothesis, we analyzed GPR37 protein levels in various JKAMP mutant cell lines. We observed a significant increase in GPR37 protein levels in siJKAMP cells compared to control cells (Figure 6A). To understand whether these observations are caused by alterations in the transcription of GPR37, we quantitatively analyzed GPR37 mRNA levels in both control and siJKAMP cells. The levels were found to be comparable (Figure 6A), suggesting a post-transcriptional regulatory mechanism. Notably, both F1(II-1) fibroblasts (Figure 6B) and JKAMP KO SH-SY5Y cells (Figure 6C) showed increased GPR37 protein levels and decreased GPR37 mRNA relative to control cells (Figures 6B and 6C), supporting the interpretation that JKAMP influences GPR37 protein stability rather than upregulating its gene transcription and indicating that the effect of JKAMP deficiency is reproducible across cell types. To investigate this, we performed a CHX chase assay, which revealed a significantly slower degradation rate of GPR37 in JKAMP KO SH-SY5Y cells compared to wild-type cells (Figure 6D). These results indicate that GPR37 is stabilized in the absence of JKAMP, pointing to a role for JKAMP in promoting GPR37 degradation. This conclusion was further validated in siJKAMP HeLa cells (Figure S5). Overexpression of JKAMP led to a reduction in GPR37 protein levels (Figure 6E), suggesting that JKAMP facilitates the degradation of GPR37, whereas its absence impairs this process.

Figure 6.

JKAMP controls GPR37 stability

(A–C) (A) Left: western blot of GPR37 in siJKAMP HeLa cells and the negative control (Ctrl−). Middle: GAPDH-normalized protein levels are expressed as fold change compared with control. Right: RT-qPCR analysis of mRNA expression levels of GPR37 in siJKAMP cells and Ctrl−, expressed as fold change after normalization on the reference gene (GAPDH mRNA). (B) Left: western blot of GPR37 in F1(II-1) derived fibroblasts and control (Ctrl). Middle: GAPDH-normalized protein levels are expressed as fold change compared with control. Right: RT-qPCR analysis of mRNA expression levels of GPR37 in F1(II-1) fibroblasts and Ctrl, expressed as fold change after normalization on the reference gene (GADPH mRNA). (C) Left: western blot of GPR37 in JKAMP KO and WT SH-SY5Y cells. Middle: ACTIN-normalized protein levels are expressed as fold change compared with WT. Right: RT-qPCR analysis of mRNA expression levels of GPR37 in JKAMP KO expressed as fold change compared to WT after normalization on the reference gene (GADPH mRNA). Values are expressed as the means of independent biological replicates (A–C); means ± SD; n = 3 independent experiments.

(D) Top: wild-type and JKAMP KO SH-SY5Y cells were incubated with 100 μg/mL cycloheximide (CHX). Immunoblots were probed with GPR37 and ACTIN antibodies. Bands are at same exposure and noncontiguous on the same gel. Bottom: quantification of ACTIN-normalized GPR37 levels vs. untreated conditions (−). Data are expressed as means ± SD; n = 3 independent experiments.

(E) Western blot of GPR37 in HeLa cells transfected with empty vector (EV) or JKAMP-FLAG (JKAMP).

Histograms show quantification of band intensities normalized vs. actin expressed as mean ± SD; n = 3 independent experiments. Wilcoxon-Mann-Whitney, unpaired t test, and multiple unpaired t test statistics were applied; ^∗p ≤ 0.05, ^∗∗∗p ≤ 0.001; ns, not significant.

Together, these data established JKAMP as a selective ERAD factor for GPR37, promoting its degradation and enabling efficient surface delivery.

Discussion

In this study, we established that bi-allelic loss-of-function variants in JKAMP cause an autosomal-recessive neurodevelopmental syndrome characterized by developmental delay, intellectual disability, seizures, hypotonia, microcephaly, and distinctive craniofacial features. JKAMP was originally identified as an ER-resident transmembrane protein that functions as a scaffold for JNK1 activation under stress conditions.¹⁰ Beyond its role in JNK1 signaling, JKAMP also contributes to ERAD by facilitating proteasome recruitment to the ER, thereby promoting the degradation of misfolded membrane proteins.^11,12 Our cohort comprised 14 affected individuals, including 12 from eight unrelated consanguineous families and two probands carrying compound heterozygous variants from two unrelated non-consanguineous families. Previously, in a larger sequencing study in Amish and Mennonite populations, an affected individual of family 7 had been reported to share a homozygous frameshift variant (c.243dup [p.Lys82GlufsTer16]) in JKAMP, which was indicated as a candidate gene for neurodevelopmental disorders.⁶⁶ Interestingly, the same homozygous JKAMP variant was identified in four out of 10 families in our cohort. These families are consanguineous and come from two main ethnic backgrounds, Arab and European. This finding across unrelated populations suggests the possibility of a founder effect, in which the variant may have originated from a common ancestor and been maintained through population bottlenecks or endogamy in different derived populations. Further haplotype analysis would be required to support this hypothesis.

In our cohort, we identified not only protein-truncating variants and canonical splicing variants but also missense variants predicted to be pathogenic. Our analysis suggests that these missense changes induce conformational and structural alterations (Figure 2A), which may compromise JKAMP function by impairing its ability to interact effectively with substrates or binding partners, while also promoting the formation of self-aggregates, as observed by immunofluorescence (Figure 2B).

JKAMP’s involvement in human disease has not been previously demonstrated. We generated a jkamp zebrafish knockout model to further demonstrate that loss of JKAMP causes a neurodevelopmental disorder. The zebrafish model recapitulated key aspects of the human disease, including microcephaly and smaller eyes. Microcephaly was observed in several affected individuals from our cohort, as well as ocular involvement, manifesting mainly as refractive or ocular motility abnormalities, while structural malformations such as microphthalmia were observed in only one case. However, the reduced eye size observed in zebrafish likely reflects early forebrain defects resulting from the tight developmental coupling between eye and forebrain formation in this species. In contrast, in mammals, eye and forebrain development are more temporally and anatomically independent, which may explain why JKAMP loss consistently leads to microcephaly but only rarely to structural eye anomalies in humans. Consistent with impaired neurodevelopment, the jkamp zebrafish model also showed reduced expression of the myelin marker mbpa, supporting a conserved role of JKAMP in neural and oligodendrocyte biology, as the MRI scans of most affected individuals also showed cerebral atrophy and delayed myelination. Zebrafish jkamp mutants did not show changes in either nkx2.2a or olig2 mRNA, both highly expressed in OL precursors. This suggests that loss of Jkamp may directly disrupt myelin sheath formation, rather than indirectly through impaired OL function. These results support an important and thus far unknown role of JKAMP in this process.

JKAMP’s involvement in ERAD appears to be selective rather than global, as it promotes proteasomal degradation primarily of membrane-associated substrates such as TCRα, CD3δ, and the mutant CFTRΔ508 protein but does not influence the degradation of misfolded luminal or cytosolic proteins.¹² The effect on clearance of membranal but not luminal proteins suggests that JKAMP activity may be limited to substrates of ubiquitin ligases/chaperones with which it associates, or to proteins located in close proximity to JKAMP in the ER.¹² Among these membrane-associated substrates, JKAMP has been shown to regulate the degradation of specific GPCRs, including DP2 and β₂AR.¹⁹ G protein-coupled receptors are a notable class of multi-spanning proteins that require stringent ER quality control to ensure that only properly folded receptors reach the plasma membrane.^14,15,16,17 By in vitro functional studies, we uncovered a critical role for JKAMP in regulating homeostasis and trafficking of GPR37, an orphan GPCR predominantly expressed in neurons and OLs.^23,24,25 Our findings confirm that JKAMP plays a selective role in ER-associated quality control, specifically regulating the degradation and trafficking of certain membrane proteins such as GPR37 rather than broadly affecting ER function or proteostasis. Differences in JKAMP variant type and genetic background effects, particularly those involving ER proteostasis or GPCR signaling, may contribute to the phenotypic variability observed among affected individuals.

Several peptides have been reported to bind to GPR37, including head activator peptide,^67,68 prosaposin, and prosaptide,^69,70 but activation of GPR37 signaling by these peptides is modest at best and not observed at significant levels in all studies,⁶⁹ so GPR37 is thus still considered an orphan receptor. Nevertheless, GPR37 has been extensively studied over the last 20 years and has been implicated in a range of neurological processes and disorders, such as dopaminergic signaling,^29,30,31 inflammatory pain,³² Parkinson disease,^26,27,28,71 brain tumors,⁷² as well as in synaptic plasticity and neuroprotection,^69,73,74 and the myelination axis.³³

Previous studies have identified GPR37 as an inhibitor of late-stage OL differentiation and myelination during development, specific to the central nervous system (CNS) as it is not present in myelinating Schwann cells.³³ In addition, adult Gpr37 −/− mice showed a striking increase in the extent of demyelination following cuprizone treatment, a demyelination inducer, making OLs more vulnerable to demyelination,³⁴ leaving overall numbers of OL precursors and mature OLs unaffected.³⁴ Loss of GPR37 may render OLs more vulnerable to demyelination,³⁴ suggesting that GPR37 may play a role in the stability of myelin or resistance of myelin itself to demyelination.

Several GPCRs regulate myelin formation, some positively, such as GPR56 and GPR126,^75,76 while others might inhibit glial differentiation (e.g., GPR17).⁷⁷ The sum of JKAMP potential interactions would shift the balance toward a hypomyelination phenotype following its loss of function, at least in humans and zebrafish. It is possible that JKAMP activity in myelin formation is dependent on its functional role on more GPCRs that promote myelinating glial development.

The functional relationship between JKAMP and GPR37 may have important neurobiological implications, as disruption of GPR37 trafficking could contribute directly to the neurodevelopmental phenotype observed in affected individuals. Interestingly, there are already molecules known to enhance GPR37 trafficking, raising the possibility that future therapeutic approaches could target this pathway to alleviate certain symptoms. For example, Dunham et al. demonstrated that GPR37 physically interacts with syntenin-1, a scaffold protein that strongly enhances its surface trafficking.⁷⁸ Syntenin-1 plays a key role in regulating receptor trafficking, including that of GPCRs, and influences membrane architecture, neuronal structure, synapse formation, and axonal elongation.⁷⁹ Given that syntenin-1 is highly expressed in OLs, where GPR37 is most abundant, this interaction suggests a critical function in myelination and cellular homeostasis.⁷⁸ Moreover, increased GPR37 cell surface expression might also be achieved through treatment with the neuroprotective agent 4-phenylbutyrate (4-PBA), a chemical chaperone that restores the trafficking of misfolded GPCRs to the plasma membrane, thereby preventing their accumulation in the ER.⁸⁰

Collectively, our study identifies JKAMP loss of function as a cause of human neurodevelopmental disease linked to disrupted ER quality control of specific GPCRs such as GPR37. These findings underscore the importance of precisely regulating the homeostasis of multi-spanning membrane proteins for proper CNS development. While targeted therapies are not yet available, elucidating the molecular pathway affected by JKAMP deficiency lays the groundwork for future efforts to restore ER proteostasis or compensate for impaired GPCR trafficking.

Further work will be essential to comprehensively define the range of JKAMP substrates in the CNS and to evaluate whether modulation of ERAD or proteostasis pathways could serve as a viable therapeutic strategy. Additionally, expanding the cohort of affected individuals will help refine the clinical spectrum and natural history of JKAMP-related disease.

In conclusion, our results establish loss of JKAMP as a genetic etiology for a neurodevelopmental disorder and highlight its pivotal role in maintaining ERAD and trafficking of critical neuronal receptors. This study emphasizes how disruption of selective ER quality-control mechanisms may contribute to complex human neurodevelopmental syndromes.

Data and code availability

The variants have been submitted to ClinVar: c.380A>G (SCV006275840), c.562C>T (SCV006301795), c.871dup (SCV006275841), c.243dup (SCV006279772), c.110G>A (SCV006279837), c.443A>G (SCV006279838), c.878del (SCV006279839), c.640+1G>T (SCV006279840), and c.458+1G>C (SCV006311624).

Consortia

The members of the Telethon Undiagnosed Diseases Program are Vincenzo Nigro, Manuela Morleo, Annalaura Torella, Carmine Spampanato, Michele Pinelli, Sandro Banfi, Angelo Selicorni, Milena Mariani, Cecilia Daolio, Valeria Capra, Andrea Accogli, Marcello Scala, Ferruccio Romano, Francesca Nardecchia, Serena Galosi, Mario Mastrangelo, Filippo Manti, Giacomina Ricciardi, Rossella Bove, Donatella Milani, Giuseppina Vitiello, Corrado Romano, Donatella Greco, Chiara Pantaleoni, Claudia Ciaccio, Stefano D’Arrigo, Arianna De Laurentiis, Nicola Brunetti Pierri, Giancarlo Parenti, Antonietta Coppola, Marcella Zollino, Domizia Pasquetti, Federica L'Erario, Simona Amenta, Albina Tummolo, Claudia Santoro, Anna Grandone, Livia Garavelli, Carla Marini, Stefania Bigoni, Alfonsina Tirozzi, Carmelo Piscopo, Antonio Trabacca, Marta De Rinaldis, Angela Peron, Silvia Maitz, Alice Donati, Diego Di Bernardo, Raffaele Castello, Mariateresa Zanobio, Giulio Piluso, Viviana Cetrangolo, Chiara De Leonibus, Margherita Scarpato, Pasquale di Letto, Maria Elena Onore, Valerio Bonolis, Francesca Del Vecchio Blanco, Carola Giacobbe, and Sarah Iffat Rahman.

Acknowledgments

The authors are grateful to the families involved. This work was supported by the Telethon Foundation (TUDP no. GSP15001 and no. TGMCBDM122TT); Solve-RD project (European Union’s Horizon2020 Research and Innovation Program no.779257); the European Union - Next Generation EU - NRRP M6C2 - Investment 2.1 Enhancement and strengthening of biomedical research in the NHS (MAD-2022-12376934); ANR-22-CE17-0028-04 “PEMGeT” and EIC Pathfinder “EdiGenT” grant no. 101070903; Fondation pour la Recherche Médicale (no. SPF202309017505); The Research, Development, and Innovation Authority (RDIA), Kingdom of Saudi Arabia (no. 12996-iau-2023-TAU-R-3-1-HW-); the collaboration between the Clinic for Special Children and the Regeneron Genetics Center (Tarrytown, NY USA). This research was also made possible through access to data generated by the 2025 French Genomic Medicine Initiative. We thank Alexandra Lubin and Katie-Jo Thorpe (UCL, London) for the Illumina MiSeq runs, Cathal Wilson for critically reading the manuscript, and the Advanced Microscopy and Imaging Core Facility and the Bioinformatics Core at TIGEM.

Declaration of interests

The authors declare no competing interests.

Published: February 4, 2026

Web resources

• BioGRID, https://thebiogrid.org

• CADD, https://cadd.gs.washington.edu/snv

• ClinVar, https://www.ncbi.nlm.nih.gov/clinvar/

• DynaMut2, https://biosig.lab.uq.edu.au/dynamut2/

• Franklin, https://franklin.genoox.com/clinical-db/home

• GeneMatcher, https://genematcher.org/

• gnomAD, https://gnomad.broadinstitute.org/

• Marrvel, http://marrvel.org/

• NCBI HomoloGene, https://www.ncbi.nlm.nih.gov/homologene

• mCSM-membrane, https://biosig.lab.uq.edu.au/mcsm_membrane/

• MutationTaster, https://www.genecascade.org/MutationTaster2025/

• OMIM, https://www.omim.org/

• PolyPhen-2, http://genetics.bwh.harvard.edu/pph2/

• PyMOL, https://pymol.org/2/

• UniProt, https://www.uniprot.org/