Endothelial IRE1α promotes thrombospondin-1 mRNA decay and supports metabolic stress adaptation of pancreatic islets

Abstract

Vascular endothelial cells (ECs) play pivotal roles in maintaining metabolic tissue homeostasis, and EC dysfunction is associated with obesity and metabolic disorders. The mammalian ER stress sensor IRE1α kinase/RNase responds to metabolic cues, but it remains unclear whether endothelial IRE1α is implicated in controlling systemic metabolism. Here we show that genetic depletion of IRE1α in ECs leads to maladaptation of pancreatic islets under obesity-associated metabolic stress. We find that in high-fat diet-fed male mice, loss of IRE1α in ECs has no significant impact upon adiposity, but unexpectedly results in glucose intolerance with impaired insulin secretion, accompanied by blunted intra-islet angiogenesis and compensatory islet growth. Mechanistically, IRE1α RNase decays the mRNA encoding the endogenous anti-angiogenic factor thrombospondin-1 (THBS1/TSP1) in islet ECs. These findings thus uncover a critical role of the endothelial IRE1α suppression of THBS1 in governing the vascular support that enables the functional adaptation of islets to metabolic stress.

Fig. 1. IRE1α ablation in ECs results in glucose intolerance with deficient insulin production in HFD-fed male mice.

a Schematic illustration of the creation of Tamoxifen (Tam)-inducible endothelial cell (EC)-specific IRE1α knock-out (Ern1^EC-KO) mice in adult Ern1^fl/fl;VE-cadherin-CreER^T2 mice produced by intercrossing floxed IRE1α mice (Ern1^fl/fl) with the VE-cadherin-CreER^T2 line. b–i Phenotypic analyses of male Ern1^EC-KO mice with diet-induced obesity. b Experimental procedure. Male Ern1^fl/fl;VE-cadherin-CreER^T2 mice and age-matched Ern1^fl/fl littermates at 4 weeks of age were injected daily with 2 mg Tam for 5 consecutive days. Mice were maintained on a normal chow (NC, 10% fat) diet, or fed a high-fat diet (HFD, 60% fat) starting at 8 weeks of age. Metabolic phenotyping was conducted at the indicated time. c Immunoblot analysis of IRE1α and CD31 protein in CD31-positive (CD31 + ) ECs and CD31-negative (CD31-) control cells isolated using anti-CD31 antibody and Dynabeads from the liver, lung and heart tissues from 8-week-old animals. β-Actin or α-Tubulin was used as the loading control. Shown also is the quantification of IRE1α protein levels in CD31+ ECs and CD31- control cells (n = 4 per genotype). d Body weight monitoring (n = 20 per group). e Body fat content and lean mass relative to body weight (BW) after NC- or HFD-feeding for 12 weeks (n = 12 per group). f Glucose tolerance test (GTT) following 12 weeks of NC or HFD feeding (NC, n = 23 per group; HFD, n = 25 per group). Shown also is the incremental area under curve (iAUC). g Insulin tolerance test (ITT) following 14 weeks of NC or HFD feeding (NC, n = 21 per group; HFD, n = 25 per group). h Serum insulin (NC, n = 12 per group; HFD, n = 19 per group) and glucose levels (NC, n = 20 per group; HFD, n = 25 per group) after a 4-hour fast following 16 weeks of NC or HFD feeding. i Glucose-stimulated insulin secretion. Serum insulin levels were measured following intraperitoneal injection of glucose after a 16-hour fast, and blood glucose levels were also simultaneously monitored for HFD-fed mice (NC, n = 10 per group; HFD, n = 15 per group). Data are presented as mean ± SEM by unpaired two-tailed Student’s t-test (c, f, i) or two-way ANOVA (e, h).

Fig. 2. IRE1α deficiency in ECs leads to impaired compensatory growth of pancreatic islets in HFD-fed male mice.

a Immunoblot analysis of IRE1α and CD31 protein in CD31-positive (CD31 + ) ECs and CD31-negative (CD31-) control cells isolated using anti-CD31 antibody and Dynabeads from pancreas tissues from 8-week-old NC-fed Ern1^EC-KO mice and their Ern1^fl/fl control littermates. β-Actin was used as the loading control. Shown also is the quantification of IRE1α protein levels in CD31+ ECs and CD31- control cells (n = 3 per genotype). b–i Male Ern1^EC-KO mice and their Ern1^fl/fl littermates were fed a NC diet or HFD for 16 weeks. b Representative immunohistochemistry (IHC) staining of insulin (Ins) from pancreas sections. Scale bar, 200 µm. Shown also are magnified graphs of the indicated islets. Scale bar, 10 µm. c Quantification of islet number and average islet area from the IHC staining (3 sections per mouse, n = 6 mice per group for islet number, and n = 5 per group for islet area). d Quantification of β cell mass by normalization to pancreatic mass (n = 6 per group). e Distribution of islets of various sizes in mice of the indicated group (n = 6 per group). f Representative images of whole-mount immunofluorescent Ki67/insulin staining of isolated primary islets from HFD-fed mice. Insulin (Ins, red), Ki67 (green) and DAPI (blue). Scale bar, 50 μm. Shown also are magnifications of the demarcated regions from the merged images. Scale bar, 20 μm. g Quantification of the percentage of total Ki67 + /Ins+ cells per islet (44 islets from 3 Ern1^fl/fl mice, 31 islets from 3 Ern1^EC-KO mice). Shown in the lower panel are percentages of Ki67⁺ /Ins⁺ cells in islets of differing sizes. h Representative images of whole-mount immunofluorescent TUNEL/insulin staining of isolated islets from HFD-fed mice. Insulin (Ins, red), TUNEL (green) and DAPI (blue). Scale bar, 50 μm. Magnifications of the demarcated regions are also shown from the merged images. Scale bar, 20 μm. i Quantification of the number of Ins⁺ /TUNEL⁺ cells per islet (26 islets from 3 Ern1^fl/fl mice, 34 islets from 3 Ern1^EC-KO mice). Data are presented as mean ± SEM by unpaired two-tailed Student’s t-test (a, g, i) or two-way ANOVA (c, d, e).

Fig. 3. IRE1α deficiency in ECs impairs insulin secretion capacity of islets from HFD-fed male mice.

Male Ern1^EC-KO and Ern1^fl/fl mice were fed a NC diet or HFD for 16 weeks. a Representative immunofluorescent staining of insulin/glucagon from pancreas sections. Insulin (Ins, red), Glucagon (Gcg, green), DAPI (blue). Scale bar, 50 μm. b, c Quantification of (b) Ins⁺ β-cell and (c) Gcg⁺ α-cell area per islet (n = 6 per NC group and n = 9 per HFD group). d, e Quantification of the area percentage of (d) Ins⁺ β-cells and (e) Gcg⁺ α-cells per islet (n = 6 per NC group and n = 9 per HFD group). f–k Primary islets were isolated after a 4-hour fast. f Glucose stimulated insulin secretion (GSIS) analysis. Islets pooled from 3 mice of each group were divided into the indicated replicates (30-50 islets per replicate) before incubation at 2.8 mM or 16.8 mM glucose for 1 hour. Insulin levels were then measured and are shown after normalization to the islet protein content. Data are representative of 2 independent experiments. g Islet insulin content was measured following GSIS, shown after normalization to islet protein content (n = 6 per group). h Perifusion analysis of insulin secretion dynamics for pooled islets from HFD-fed mice (n = 3 per group). Islets were stimulated sequentially by 16.8 mM glucose and KCl, and insulin levels are shown as fold changes relative to the unstimulated starting basal value. p.I: Phase I; p.II: Phase II. Shown are representative results from 2 independent experiments. i Quantification of glucose-stimulated biphasic insulin secretion and KCl-induced insulin release in (h) after normalization to the islet protein content (n = 6 mice per group from the 2 independent experiments). j Glucose-stimulated Ca²⁺ influx in whole islets from HFD-fed mice (n = 80 islets from 6 mice per genotype). Shown are representative fluorescent calcium signals in islets maintained at 2.8 mM glucose or following stimulation with 16.8 mM glucose for 150 seconds. Scale bar, 100 μm. k Dynamic monitoring of Ca²⁺ intensity in response to high glucose stimulation in (j). Data are presented as mean ± SEM by two-way ANOVA (b, f) or unpaired two-tailed Student’s t-test (i, k).

Fig. 4. Depletion of IRE1α limits the increase of intra-islet EC density in HFD-fed male mice.

Male Ern1^EC-KO mice and their Ern1^fl/fl littermates were fed a NC diet or HFD for 16 weeks. a Representative IHC staining of CD31 from pancreas sections of mice. Scale bar, 200 µm. b Quantification of intra-islet CD31⁺ signals in (a) (n = 6 per group). c Quantitative RT-PCR analysis of the abundance of Pecam1 mRNA from isolated islets (n = 5 per group). d Representative images of whole-mount immunofluorescent staining of CD31/Insulin in isolated primary islets from mice. Insulin (Ins, green), CD31 (red). Scale bar, 50 µm. e Quantification of intra-islet CD31⁺ signals relative to Ins⁺ signals ( ~ 15 islets per mouse, n = 3 mice per group). f Representative images of whole-mount immunofluorescent staining of CD31/Ki67 in isolated islets from HFD-fed mice. CD31 (red), Ki67 (green). Scale bar, 50 μm. Shown also are magnifications of the demarcated regions from the merged images. Scale bar, 20 μm. g Quantification of CD31⁺/Ki67⁺ double-positive signals per islet ( ~ 15 islets per mouse, n = 3 per group). h Representative images of whole-mount immunofluorescent staining of CD31/TUNEL in isolated islets from HFD-fed mice (40 islets out of 3 mice per group). CD31 (red), TUNEL (green). Scale bar, 50 μm. Magnifications of the demarcated regions are also shown from the merged images. Scale bar, 20 μm. Data are presented as mean ± SEM by unpaired two-tailed Student’s t-test (g) or two-way ANOVA (b, c, e).

Fig. 5. IRE1α ablation in ECs impacts islet transcriptomes with enhanced islet Thrombospondin 1 expression in HFD-fed mice.

a–d Global gene expression analysis by RNA-seq using total RNA extracts from isolated primary islets of Ern1^fl/fl and Ern1^EC-KO mice after 16 weeks of HFD feeding (n = 3 per group). a Gene ontology (GO) biological process (BP) analysis of 573 genes (average FPKM > 1; |log₂ fold-change (FC) | > 0.5) whose expression was significantly different as a result of IRE1α deficiency in ECs. b Volcano plot depicting the log₂(FC) values (Ern1^EC-KO versus Ern1^fl/fl) versus -log₁₀(P-values) for genes (FPKM > 1) from the RNA-seq analysis of islets. Highlighted are up-regulated (red) or down-regulated (blue) genes ( | log₂(FC) | > 0.5; P Value < 0.01) resulting from endothelial IRE1α deficiency. c List of top-ranked genes whose expression was significantly upregulated or downregulated in islets with EC IRE1α deficiency (FPKM > 1; |log₂(FC) | > 0.5; P Value < 0.01). d Heat maps of 7 genes (average FPKM > 1; |log₂(FC) | > 0.5; P Value < 0.01) involved in regulation of endothelial proliferation, which were differentially expressed in islets from HFD-fed Ern1^EC-KO mice versus their Ern1^fl/fl counterparts (n = 3 mice per group). e, f Primary islets isolated from Ern1^fl/fl and Ern1^EC-KO mice after 16 weeks of HFD feeding. e Quantitative RT-PCR analysis of the mRNA abundance of islet Thbs1 (n = 7 per group) and Egfl7 (Ern1^fl/fl, n = 6; Ern1^EC-KO, n = 7). f Immunoblot analysis of islet TSP1 protein levels (Ern1^fl/fl, n = 5; Ern1^EC-KO, n = 6). α-Tubulin was used as a loading control. Data are shown as mean ± SEM by unpaired two-tailed Student’s t-test (e, f).

Fig. 6. IRE1α suppresses Thrombospondin 1 expression to promote endothelial cell proliferation.

a–f MS1 endothelial cells (a, c, e) or primary HUVECs (b, d, f) were infected with lentiviruses encoding a shRNA directed against IRE1α (shErn1 or shERN1) or a scramble control (shCtrl) for 48 hours. a, b Quantitative RT-PCR analysis of Xbp1 mRNA splicing and Thbs1/THBS1 mRNA abundance in MS1 cells (a) or HUVECs (b) pre-cultured for 4 hours at 5 mM glucose or following stimulation for 24 hours with high glucose at 16 mM (n = 3 independent experiments). c, d Immunoblot analysis of IRE1α and TSP1 protein levels in MS1 cell lysates (c) or HUVEC lysates (d) when cultured at 5 mM versus 16 mM glucose for 24 hours. α-Tubulin was used as a loading control (n = 3 independent experiments). e, f Immunoblot analysis of TSP1 protein in culture medium of MS1 cells (e) or HUVECs (f). Cell lysate α-Tubulin was used as a loading control (n = 3 independent experiments). g Cell proliferation and viability analysis by cell-count assay or using the CCK-8 Kit for lentivirus-infected MS1 cells cultured at 16 mM glucose (n = 2 independent experiments). h, i MS1 cells were infected with shCtrl, shErn1, or both shErn1 and shThbs1 lentiviruses for 48 hours. h Immunoblot analysis of IRE1α and TSP1 proteins in MS1 cell lysates. i MS1 cell proliferation analysis by cell-count assay or using the CCK-8 Kit when cultured at 16 mM glucose (n = 2 independent experiments). j Proliferation analysis by the CCK-8 Kit of MS1 cells that were cultured at 16 mM glucose and treated with PBS (Vehicle, Veh.) or 1 μg/mL recombinant human TSP1 protein in the absence or presence of 1 μg/mL anti-CD47 neutralizing antibody for 48 hours (n = 6 independent treatment experiments). Data are shown as mean ± SEM by two-way ANOVA (a–d, g, i, * indicates shCtrl versus shErn1; # indicates shErn1 versus shErn1/shThbs1 in i) or unpaired two-tailed Student’s t-test (e, f, j).

Fig. 7. IRE1α suppresses Thrombospondin 1 expression through its RIDD activity.

a MS1 cells were pre-cultured at 5 mM glucose for 4 hours before stimulation with 16 mM glucose for 12 hours in the absence or presence of 10 μM 4μ8C. Quantitative RT-PCR analysis of Xbp1 mRNA splicing and Thbs1 mRNA abundance (n = 3 independent experiments). b HEK293T cells with IRE1α depletion (HEK293T-KO) were transiently transfected for 36 hours with vector control (-) or THBS1-Myc plasmid, or co-transfected with THBS1-Myc plus IRE1α or XBP1s-Flag plasmids. Cells transfected as indicated were also treated with 10 μM 4μ8C for 12 hours. Immunoblot analysis of IRE1α, XBP1s and TSP1-Myc protein. Shown also is quantification of TSP1-Myc protein level after normalization to β-Actin control (n = 3 independent experiments). c, d HEK293T-KO cells were transiently transfected for 36 hours with vector control, IRE1α-WT or its RNase-deficient K907A mutant plasmids. c Quantitative RT-PCR analysis of human XBP1s, BLOC1S1 and THBS1 mRNA levels (n = 3 independent experiments). d Immunoblot analysis of IRE1α, TSP1-Myc and XBP1 protein with α-Tubulin as a loading control. Shown also is quantification of TSP1-Myc protein levels (n = 3 independent experiments). e Consensus sequence (underlined) of the putative RIDD region in human and mouse TSP1 mRNAs within the stem-loop structure predicted by RNAstructure Version 6.2. f Agarose gel analysis of IRE1α-mediated cleavage of human THBS1 mRNA. In vitro transcription-derived mRNA fragments of THBS1 (nt 1-1770 and nt 1771-3510) were incubated for 1 hour with recombinant human IRE1α protein (1 μg) in the presence of DMSO (-) or 10 μM 4μ8C, followed by 3% agarose gel analysis. Human XBP1 mRNA was used as a positive control. Red arrows indicate the probable major RNA cleavage products. g Synthetic RNA substrates for wild-type THBS1^WT (nt 2771-2808) or THBS1^Mut with the indicated G-to-C mutation were incubated for 1 hour with human IRE1α protein (1 μg) pre-mixed for 1 hour with DMSO (-) or 10 μM 4μ8C, followed by 20% TBE-Urea PAGE gel analysis. The red arrow indicates the RNA cleavage product. Data are representative of 2 independent experiments in (f, g). Results are shown as mean ± SEM by unpaired two-tailed Student’s t-test (a–d).

Fig. 8. Depletion of Thbs1 in ECs corrects glucose intolerance and islet dysfunction of HFD-fed male mice with endothelial IRE1α deficiency.

Male Ern1^EC-KO;Thbs1^EC-KO mice and their Ern1^fl/fl;Thbs1^fl/fl littermates were fed an HFD for 14-18 weeks. a Body weight monitoring (Ern1^fl/fl;Thbs1^fl/fl, n = 8; Ern1^EC-KO;Thbs1^EC-KO, n = 10). b Averaged daily food intake (Ern1^fl/fl;Thbs1^fl/fl, n = 8; Ern1^EC-KO;Thbs1^EC-KO, n = 10). c, d GTT along with iAUC (c) and ITT (d) analyses (Ern1^fl/fl;Thbs1^fl/fl, n = 7; Ern1^EC-KO;Thbs1^EC-KO, n = 10). e Serum insulin levels along with blood glucose levels were measured upon intraperitoneal glucose injection (Ern1^fl/fl;Thbs1^fl/fl, n = 6; Ern1^EC-KO;Thbs1^EC-KO, n = 8). f Primary islets were isolated from mice following 16 weeks of HFD feeding. Representative images of whole-mount immunofluorescent staining of CD31/insulin in islets. Scale bar, 50 µm. Shown also is quantification of intra-islet CD31⁺ relative to Ins⁺ signals (10 islets per mouse, n = 3 mice per group). g Quantitative RT-PCR analysis of the mRNA abundance of Pecam1 and Thbs1 in islets (n = 4 per group). h Representative IHC staining of insulin in pancreas sections from HFD-fed mice. Scale bar, 200 µm. Shown also are magnifications of the indicated islets. Scale bar, 20 µm. i Quantification of the distribution of islets in the indicated range of sizes, shown as the islet number per mouse for each group ( ~ 600 islets; 6 slides per mouse; n = 3 mice per group). j GSIS analysis of isolated primary islets (pooled from 3 mice, 5 replicates for each group). Insulin levels were normalized to islet protein content. k Insulin content was measured by ELISA from islets after GSIS, shown after normalization to islet protein content (pooled from 3 mice, 5 replicates for each group). Data are shown as mean ± SEM by unpaired two-tailed Student’s t-test (g) or two-way ANOVA (j).