Recessive TMEM167A variants cause neonatal diabetes, microcephaly, and epilepsy syndrome

Abstract

Understanding the genetic causes of diseases that affect pancreatic β cells and neurons can give insights into pathways essential for both cell types. Microcephaly, epilepsy, and diabetes syndrome (MEDS) is a congenital disorder with two known etiological genes, IER3IP1 and YIPF5. Both genes encode proteins involved in endoplasmic reticulum (ER) to Golgi trafficking. We used genome sequencing to identify 6 individuals with MEDS caused by biallelic variants in the potentially novel disease gene TMEM167A. All had neonatal diabetes (diagnosed at <6 months) and severe microcephaly, and 5 also had epilepsy. TMEM167A is highly expressed in developing and adult human pancreas and brain. To gain insights into the mechanisms leading to diabetes, we silenced TMEM167A in EndoC-βH1 cells and knocked-in one patient's variant, p.Val59Glu, in induced pluripotent stem cells (iPSCs). Both TMEM167A depletion in EndoC-βH1 cells and the p.Val59Glu variant in iPSC-derived β cells sensitized β cells to ER stress. The p.Val59Glu variant impaired proinsulin trafficking to the Golgi and induced iPSC-β cell dysfunction. The discovery of TMEM167A variants as a genetic cause of MEDS highlights a critical role of TMEM167A in the ER to Golgi pathway in β cells and neurons.

Keywords: Beta cells; Cell biology; Endocrinology; Genetic diseases; Genetics; Neurodevelopment.

Figure 1. Identification of recessive TMEM167A variants in individuals with MEDS.

(A) Partial pedigrees and summary of clinical features of the 6 patients with recessive TMEM167A variants. Head circumference standard deviation (SD) below the mean is given in parentheses. DD, developmental delay. (B) Schematic representation of the TMEM167A transmembrane protein with variant positions indicated by black arrows. Domains are represented as predicted in UniProtKB (https://www.uniprot.org/).

Figure 2. Expression of TMEM167A in human brain and pancreas.

(A) TMEM167A, IER3IP1, and YIPF5 expression from the BrainSpan Atlas (https://www.brainspan.org/) across development in dorsofrontal cortex (DFC); lateral (LGE), medial (MGE), and caudal (CGE) ganglionic eminences; and striatum (STR). (B) TMEM167A mRNA expression in human pallium and subpallium at gestational weeks (GW) 11–13. Expression at GW11 was set to 1.0; other stages are relative. Violin plots show median ± quartiles of 4 pallium and 3 subpallium samples. (C) RNAscope and immunolabeling of GW11 cortex showing TMEM167A mRNA (green), CTIP2 (gray), and SOX2 (red). The left panel shows merged staining (left) and TMEM167A alone (right). Magnified views of CTIP2-positive (top) and SOX2-positive (bottom) regions are shown in the right panels, with adjacent images displaying TMEM167A only. Scale bars: 50 μm, 10 μm (close-up). (D) Quantification of TMEM167A mRNA dots per cell in SOX2- and CTIP2-positive cells within a square with 100 μmlong sides. Violin plots show median ± quartiles from two GW11 samples. (E) Immunolabeling of rosettes in cerebral organoids showing TMEM167A mRNA (green), TBR1 (gray), and SOX2 (red). Scale bars: 50 μm, 20 μm (close-up). (F) Quantification of TMEM167A mRNA dots per cell in SOX2- and TBR1-positive rosette cells from 3 organoids. Violin plots show median ± quartiles from 3 biological replicates, with each rosette per organoid represented by a uniquely colored dot. **P < 0.01. (G) Heatmap of TMEM167A, YIPF5, and IER3IP1 expression in human embryonic pancreas (GW7–12) based on microarray data from FACS-sorted epithelial and mesenchymal populations. GP2+ indicates pancreatic progenitors; Mes, mesenchymal cells; GP2−, ductal/endocrine progenitors; Elo, endocrine cells; and Ghi, acinar cells. INS, NEUROG3, CEL, and POSTN serve as markers of β, pancreatic progenitor, acinar, and mesenchymal cells, respectively. (H) RNAscope and immunolabeling of GW9 and GW13 + 5 days human pancreas showing expression of TMEM167A mRNA (green), insulin (gray), and SOX9 (red) and nuclear counterstaining (DAPI, blue). Scale bars: 50 μm, 10 μm (close-up).

Figure 3. The TMEM167A V59E variant does not impact β cell differentiation but lowers insulin content.

(A) Quantification of dispersed S7 iPSC-β cell aggregates stained for insulin (green), glucagon (red), and somatostatin (pink); nuclei were stained with DAPI (n = 7–11). Polyhormonal cells are shown in yellow. (B and C) Quantification of hormonal volume (B) and sphericity (C) by light sheet microscopy of S7 iPSC-β cell aggregates stained for insulin, glucagon, and somatostatin (n = 4). (D and E) Proinsulin (D) and insulin (E) content normalized to total protein content of S7 iPSC-β cell aggregates differentiated from control and mutant iPSCs (n = 6–8). (F) Quantification of dispersed long-term differentiated iPSC-β cell aggregates stained for insulin (green), glucagon (red), and somatostatin (pink); nuclei were stained with DAPI (n = 6–7). (G and H) Proinsulin (G) and insulin (H) content normalized to total protein content of long-term culture iPSC-β cell aggregates differentiated from control and mutant iPSCs (n = 8). (I) Colocalization at each time point of the RUSH assay by i1 intersection coefficient, i.e., percentage of intersecting volume/GM130 volume (n = 3–4) in dispersed S7 iPSC-β cells. Triangles represent mother cell line 1.023, squares cell line 1.023 mutTMEM167A V59E.37, and circles cell line 1.023 mutTMEM167A V59E.48. The median is shown by horizontal lines in the box plots; 25th and 75th percentiles are at the bottom and top of the boxes; whiskers represent minimum and maximum values. Error bars represent SEM. Statistical significance was assessed in C–E, G, and H by 2-tailed unpaired t test, and in B and I by 2-way ANOVA followed by Bonferroni post hoc tests for pairwise comparison. *P < 0.05, **P < 0.01.

Figure 4. TMEM167A V59E β cells are dysfunctional.

(A) Dynamic insulin secretion upon perifusion with 2.8 mM glucose (G2.8), followed by 16.7 mM glucose (G16.7) alone or with 12 nM exendin-4 and G2.8 plus 30 mM KCl (sampling every 4 minutes). Data are normalized to protein content (n = 8). (B–E) Quantification of AUC from A for each phase. (F and G) Kinetics of mean (± SEM) frequencies (F) and amplitudes (G) of β cell slow potentials (SPs) recorded with multi-electrode arrays, upon 2.8 mM glucose, 16.7 mM glucose, amino acid mix (AA), and forskolin (Fk; 1 μM). (H and J) Blood glucose (H) and plasma human C-peptide (J) levels during an intraperitoneal glucose tolerance test in mice 4 months after transplantation with S7 MACS-purified iPSC-β cell aggregates (n = 8–9). (I and K) AUC of H and J. Triangles represent mother cell line 1.023, squares 1.023 mutTMEM167A V59E.37, and circles 1.023 mutTMEM167A V59E.48. Each data point represents an independent experiment. Error bars represent SEM. Statistical significance was assessed in C–E, I, and K by Mann-Whitney test, and in H and J by 2-way ANOVA with Bonferroni’s correction. *P < 0.05, ***P < 0.001.

Figure 5. TMEM167A V59E β cells have inhibited ER to Golgi trafficking and insulin secretion gene signatures.

(A) Clusters identified from single-cell data. 9,721 long-term culture iPSC-aggregate cells were sequenced for cell line 1.023, 9,291 cells for 1.023 mutTMEM167A V59E.37, and 4,910 cells for 1.023 mutTMEM167A V59E.48. (B) Fractions of wild-type and mutant cells populating different clusters.

Figure 6. TMEM167A V59E variant–carrying β cells are sensitive to ER stress–induced apoptosis.

S7 iPSC-β cell aggregates were exposed or not for 24 hours to brefeldin A (0.025 mg/dL; A–E) or for 48 hours to thapsigargin (1 μM; F–J). Viability was assessed by Hoechst 33342/propidium iodide staining (n = 6–7). Gene expression was assessed by qPCR, normalized to reference genes ACTB and VAPA (n = 6–8). Triangles represent mother cell line 1.023, squares 1.023 mutTMEM167A V59E.37, and circles 1.023 mutTMEM167A V59E.48. Error bars represent SEM. Extremities of box-and-whisker plots are maximal and minimal values; horizontal line shows median. Each data point represents an independent experiment. Statistical significance was assessed in A, B, D, F, H, and J by 2-way ANOVA with Bonferroni’s correction. *P < 0.05, **P < 0.01.