YIPF5 mutations cause neonatal diabetes and microcephaly through endoplasmic reticulum stress

Abstract

Neonatal diabetes is caused by single gene mutations reducing pancreatic β cell number or impairing β cell function. Understanding the genetic basis of rare diabetes subtypes highlights fundamental biological processes in β cells. We identified 6 patients from 5 families with homozygous mutations in the YIPF5 gene, which is involved in trafficking between the endoplasmic reticulum (ER) and the Golgi. All patients had neonatal/early-onset diabetes, severe microcephaly, and epilepsy. YIPF5 is expressed during human brain development, in adult brain and pancreatic islets. We used 3 human β cell models (YIPF5 silencing in EndoC-βH1 cells, YIPF5 knockout and mutation knockin in embryonic stem cells, and patient-derived induced pluripotent stem cells) to investigate the mechanism through which YIPF5 loss of function affects β cells. Loss of YIPF5 function in stem cell-derived islet cells resulted in proinsulin retention in the ER, marked ER stress, and β cell failure. Partial YIPF5 silencing in EndoC-βH1 cells and a patient mutation in stem cells increased the β cell sensitivity to ER stress-induced apoptosis. We report recessive YIPF5 mutations as the genetic cause of a congenital syndrome of microcephaly, epilepsy, and neonatal/early-onset diabetes, highlighting a critical role of YIPF5 in β cells and neurons. We believe this is the first report of mutations disrupting the ER-to-Golgi trafficking, resulting in diabetes.

Keywords: Cell Biology; Cell stress; Diabetes; Genetics; Human stem cells.

Figure 1. Identification of homozygous YIPF5 mutations in 6 patients with neonatal diabetes, severe microcephaly, and epilepsy.

(A) Partial pedigrees and summary of clinical features of the 6 patients with homozygous YIPF5 mutations. Age at diagnosis of diabetes and head circumference standard deviation below the mean are given in parentheses. (B) Schematic representation of the YIPF5 ER transmembrane protein using the CCTOP in silico predictor (http://cctop.enzim.ttk.mta.hu/). Note that there is uncertainty regarding YIPF5 transmembrane predictions and the position of the p.Trp218 residue is predicted to be cytoplasmic by UniProtKB (https://www.uniprot.org/).

Figure 2. YIPF5 is expressed in human pancreatic tissue and brain.

(A) YIPF5 mRNA expression was measured by qPCR in human tissues (n = 2–3), EndoC-βH1 cells (n = 15), and human islets (n = 4) and normalized to the geometric mean of the reference genes ACTB, GAPDH, and OAZ1. (B) In situ hybridization of YIPF5 in human fetal cortex at gestational week 12. Expression is found in the ventricular zone (VZ), intermediate zone (IZ), and cortical plate (CP) as well choroid plexus (ch) (antisense probe, right). No signal was detected when the sense probe was used (negative control, left). Scale bar: 100 μm.

Figure 3. YIPF5 deficiency does not affect insulin secretion but sensitizes β cells to ER stress–induced apoptosis.

(A–C) EndoC-βH1 cells were transfected with siRNA against YIPF5 (si1) or control siRNA (siCT) for 48 hours and incubated with 0 or 20 mM glucose or 20 mM glucose plus 10 μM forskolin (FSK). (A) YIPF5 mRNA expression by qPCR. (B) Insulin content normalized for total protein content. (C) Insulin secretion expressed as percentage of total insulin content. (D and E) EndoC-βH1 cells were transfected with 2 siRNAs against YIPF5 (si1 and si2) or control siRNA (siCT) for 48 hours and exposed or not (CTL) to thapsigargin (Tha) for 40 hours or brefeldin A (BFA) for 16 hours (n = 4). Apoptosis was evaluated by staining with DNA-binding dyes (n = 4) (D) or luminescence produced by annexin V binding (RealTime-Glo Annexin V assay) at the indicated time points (n = 3) (E). Thapsigargin is presented by solid lines and nontreated cells by dashed lines. (F) Dispersed human islet cells were transfected with si1 or siCT for 48 hours and exposed or not to brefeldin A for 24 hours. Apoptosis was evaluated by staining with DNA-binding dyes (n = 4). (G and H) EndoC-βH1 cells were transfected with si1 or siCT for 48 hours and exposed to thapsigargin for the indicated times (n = 5–6). CHOP (G) and DP5 (H) mRNA expression was measured by qPCR, normalized to β-actin (ACTB). (I and J) EndoC-βH1 cells were transfected with siCT or si1 and/or siRNA against CHOP (siCHOP) (I) or DP5 (siDP5) (J) and treated or not with thapsigargin for 40 hours (n = 5 and n = 8, respectively). Apoptosis was examined by DNA-binding dye. Individual symbols represent independent experiments, and box plots show the median by a horizontal line, 25th and 75th percentiles at the bottom and top of the boxes, and minimum and maximum values by whiskers. In time course experiments, data are shown as mean ± SEM. Paired 2-way ANOVA or mixed-model analysis (in case of missing values) followed by Bonferroni post hoc test. *P < 0.05, **P < 0.01, ***P < 0.001 vs. siCT in respective condition; ^##P < 0.01, ^###P < 0.001 for treated vs. untreated cells; ^†††P < 0.001 as indicated.

Figure 4. Proinsulin accumulation, increased ER stress signaling, and reduced insulin content in YIPF5-knockout stem cell–derived β cells.

(A) Immunocytochemistry for proinsulin (PROINS) and insulin (INS) at stage 7 of in vitro differentiation for WT and YIPF5-KO cells. Scale bars: 25 μm. (B) Immunocytochemistry for BiP and insulin (INS) at stage 7 of in vitro differentiation. Scale bars: 25 μm. (C) Percentage of cytoplasmic area covered by proinsulin or insulin per insulin-positive cell (n = 3). (D) Percentage of INS⁺BiP^hi cells per total number of INS⁺ cells (n = 4–8). (E) Percentage of apoptotic cells (INS⁺TUNEL⁺) per total number of INS⁺ cells after treatment with vehicle (DMSO) and the ER stressors thapsigargin, tunicamycin, and brefeldin A (n = 3–5). (F) Static glucose-stimulated insulin secretion at stage 7 normalized to micrograms DNA of β cells (n = 3–7). (G) Insulin content of stage 7 differentiated cells normalized to micrograms DNA of β cells (n = 3–8). (H) Percentage of INS⁺ cells at week 2 of stage 7 (n = 3–4). Statistical significance was assessed in C, D, G, and H by 1-way ANOVA test with Bonferroni correction, and in E and F by 2-way ANOVA test with Bonferroni correction. **P < 0.01, ***P < 0.001, ****P < 0.0001. Error bars represent SD from the mean. (I) Transmission electron microscopy of WT, YIPF5-KO, and YIPF5^Ile98Ser stage 7 cells showing the cytoplasmic area of β and α cells. Yellow arrowheads point at insulin granules, red arrowheads at glucagon granules, and green arrowheads at ER. Scale bars: 1 μm.

Figure 5. Reduced C-peptide secretion and β cell numbers in implanted YIPF5 knockout and signs of YIPF5^Ile98Ser β cell failure.

(A) Human C-peptide levels measured in mouse serum through 3 months after implantation (n = 3–8). (B) Mouse blood glucose levels at 1 and 3 months after implantation (n = 3–10). (C) Percentage of INS⁺GCG^– cells per the total number of INS⁺ plus GCG⁺ cells (n = 3–5). (D) Percentage of cytoplasmic area covered by proinsulin or insulin in insulin-positive cells (n = 3–4). (E) Percentage of INS⁺BiP^hi cells per total number of INS⁺ cells (n = 3–6). (F–H) Immunohistochemistry of grafts for glucagon (GCG) and insulin (INS) (F), proinsulin (PROINS) and insulin (INS) (G), and BiP and insulin (INS) (H) 3 months after implantation. Scale bars: 100 μm (F); 25 μm (G); 100 μm (H, left); 25 μm (H, right). Statistical significance was assessed by 2-way ANOVA test with Bonferroni correction in A, by multiple t test with Bonferroni correction in B, and by 1-way ANOVA test with Bonferroni correction in C–E. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.00001. Error bars represent SD from the mean.

Figure 6. iPSCs from patients IIIa and IIIb differentiated into β cells are sensitive to ER stress–induced apoptosis.

(A) Representative immunostaining of dispersed stage 7 aggregates stained for insulin (INS, green) and glucagon (GCG, red). Nuclei were visualized with DAPI (blue). (B) Quantification of immunostained cells (expressed as percent of total cells) in dispersed stage 7 control (n = 25) and patient cells (n = 11). Blue squares represent patient cells (2 patients, 2 iPSC lines for each); black circles and squares represent healthy control (1 iPSC line) and corrected patient cells (2 iPSC lines from 1 patient), respectively. (C and D) Apoptosis was assessed by staining with DNA-binding dyes in vehicle- (DMSO-)treated, thapsigargin-treated, and tunicamycin-treated control and corrected (n = 10) and patient (n = 6–7) stage 7 aggregates (C) or by luminescence produced by annexin V binding in time course experiments (means ± SEM; n = 10 control and corrected lines and n = 5 patient lines) (D). (E) mRNA expression of CHOP, BiP, sXBP1, DP5, and PUMA assessed by qPCR in stage 7 aggregates from control and corrected (n = 4–8, black) and patient cells (n = 5–7, blue) exposed for 48 hours to vehicle (DMSO), thapsigargin, or tunicamycin. mRNA expression was normalized to the geometric mean of reference genes β-actin and GAPDH. The median is shown by a horizontal line in the box plots; 25th and 75th percentiles are at the bottom and top of the boxes; whiskers represent minimum and maximum values, and data points independent experiments. Comparisons were done by multiple t test followed by Bonferroni’s correction for multiple comparisons (B), ANOVA followed by Bonferroni’s correction for multiple comparisons (C and D), and paired-ratio t test (E). *P < 0.05, **P < 0.01, ***P < 0.001 treatment vs. DMSO; ^#P < 0.05, ^##P < 0.01 vs. control and corrected cells as indicated.