COUP-TFII controls mouse pancreatic β-cell mass through GLP-1-β-catenin signaling pathways

Abstract

Background: The control of the functional pancreatic β-cell mass serves the key homeostatic function of releasing the right amount of insulin to keep blood sugar in the normal range. It is not fully understood though how β-cell mass is determined.

Methodology/principal findings: Conditional chicken ovalbumin upstream promoter transcription factor II (COUP-TFII)-deficient mice were generated and crossed with mice expressing Cre under the control of pancreatic duodenal homeobox 1 (pdx1) gene promoter. Ablation of COUP-TFII in pancreas resulted in glucose intolerance. Beta-cell number was reduced at 1 day and 3 weeks postnatal. Together with a reduced number of insulin-containing cells in the ductal epithelium and normal β-cell proliferation and apoptosis, this suggests decreased β-cell differentiation in the neonatal period. By testing islets isolated from these mice and cultured β-cells with loss and gain of COUP-TFII function, we found that COUP-TFII induces the expression of the β-catenin gene and its target genes such as cyclin D1 and axin 2. Moreover, induction of these genes by glucagon-like peptide 1 (GLP-1) via β-catenin was impaired in absence of COUP-TFII. The expression of two other target genes of GLP-1 signaling, GLP-1R and PDX-1 was significantly lower in mutant islets compared to control islets, possibly contributing to reduced β-cell mass. Finally, we demonstrated that COUP-TFII expression was activated by the Wnt signaling-associated transcription factor TCF7L2 (T-cell factor 7-like 2) in human islets and rat β-cells providing a feedback loop.

Conclusions/significance: Our findings show that COUP-TFII is a novel component of the GLP-1 signaling cascade that increases β-cell number during the neonatal period. COUP-TFII is required for GLP-1 activation of the β-catenin-dependent pathway and its expression is under the control of TCF7L2.

Figure 1. Pdx1^CRE/- COUP-TFII^Fl/Fl mice have defective glucose homeostasis.

(A) Immunostaining of pancreatic sections from COUP-TFII^Fl/Fl control and pdx1^CRE/- COUP-TFII^Fl/Fl adult mice using mouse monoclonal antibodies against COUP-TFII. Scale bars =  100 µm (B) Fasting plasma insulin of 4 month and 6- week-old mice littermates. (C) Intraperitoneal glucose tolerance tests were performed in 4-month-old male pdx1^CRE/-, COUP-TFII^Fl/Fl and pdx1^CRE/- COUP-TFII^Fl/Fl mice that had been fasted for 14 h. Blood glucose levels, left panel. Rate of glucose disappearance (K), inset panel. Plasma insulin levels, right panel. Results are mean ± SEM (error bars) for 8 to 14 animals per genotype. Values for pdx1^CRE/- COUP-TFII^Fl/Fl are significantly different from values for COUP-TFII^Fl/Fl and pdx1^CRE/- control mice at P<0.05 (*) or P<0.03 (**).

Figure 2. Pdx1^CRE/- COUP-TFII^Fl/Fl mice have fewer β-cells at birth.

(A) Pancreatic sections from adult mice were immunostained for pancreatic β-cells and α-cells using respectively anti-insulin and anti-glucagon antibodies. Scale bars =  100 µm (B) Estimation of the relative β-cell area and β-cell mass in the pancreatic sections of pdx1^CRE/-, COUP-TFII^Fl/Fl and pdx1^CRE/-COUP-TFII^Fl/Fl in 4-month-old mice. Beta-cell mass was estimated by morphometric analysis. Relative β-cell size (mean of ≥ 1500 cells counted). (C) Estimation of β-cell number and β-cell mass in pancreatic sections of 3-week-old mice. (D) Ratio of Insulin-positive cross-sectional area relative to pancreas (DAPI-positive) cross-sectional area in one-day-old neonate animals. # The reduction of pdx1^CRE/-, COUP-TFII^Fl/Fl over the cumulated controls (pdx1^CRE/- and COUP-TFII^Fl/Fl) is already quite important (23%) and is significant with student t-test and non-parametric tests (2-tailed Mann-Whitney U test, which does not assume normal distribution: p = 0.0476). It is not significant if referred individually to the two control populations. Data are means ± SEM (error bars) of values from 4 animals of each genotype. *Values for pdx1^CRE/-, COUP-TFII^Fl/Fl mice is significantly different from the corresponding values for control mice (pdx1^CRE/- and COUP-TFII^Fl/Fl) at P<0.05.

Figure 3. Five-day-old pdx1^CRE/-, COUP-TFII^Fl/Fl mice have reduction of β-cell neogenesis and COUP-TFII knockdown increases 832/13 INS-1 β-cells apoptosis.

In pancreatic sections of 5-day-old mice (A) β-cell proliferation was assessed by double staining insulin/Ki67 (B) β-cell neogenesis activation was evaluated through quantification of duct-associated insulin+ β-cells per unit of total tissue area. 832/13 INS-1 β-cells were electroporated with scrambled or specific COUP-TFII siRNA and cultured in INS-1 medium for 24 h or 48 h. (C) Relative COUP-TFII mRNA levels determined by RT-qPCR at 48 h. (D) Effect of COUP-TFII siRNA on cell viability analyzed by MTT assay (n = 5 electroporations). (E) Effect of COUP-TFII siRNA on β-cell proliferation. 832/13 INS-1 β-cells were cultured in INS-1 medium for 48 h and stained with BrdU. Cells were analyzed by flow cytometry (n = 4 electroprations). (F) Effect of COUP-TFII siRNA on β-cells apoptosis and comparison of the apoptotic rate with cell treated with a rat cytokine mix containing 25 ng/ml TNF-α, 10 ng/ml IL-1β and 10 ng/ml INF-γ during 24 h. 832/13 INS-1 β-cells were electroporated with scrambled or COUP-TFII siRNA and 48 h later they were stained with DiOC6(3) and analyzed by flow cytometry (n = 4 electroporations). * Significant difference between values linked by brackets at P<0.03.

Figure 4. COUP-TFII regulates β-catenin signaling target genes in mouse islets and a β-cell line.

(A) 832/13 INS-1 β-cells were electroporated with scrambled (control) or COUP-TFII specific siRNA and cultured for 48 h. Levels of cyclin D1, c-myc, glutamine synthase, axin 2, and β-catenin mRNA were determined by RT-qPCR. Effects on Wnt/β-catenin signaling were assessed by the luciferase activity of a TOP/FOP-flash reporter. Results are means ± SEM of data from between 3 and 12 (mRNA levels) independent electroporation experiments. Significant differences between COUP-TFII knockdown and control values are shown by asterisks at the P value indicated. *P<0.05; **P<0.03. (B) A representative immunoblot of different experiments of 832/13 INS-1 β-cells that were electroporated with scrambled or COUP-TFII specific siRNA. 50 µ g of total protein extract was subjected to immunoblot analysis with antibodies against cyclin D1, CDK4 or β-actin. (C) RT-qPCR analysis of β-catenin and cyclin D1 mRNA levels of 832/13 INS-1 cells infected with Ad-GFP or Ad-hCOUP-TFII at 2 or 5 PFU per cell as indicated normalized to housekeeping gene cyclophilin. Results are means ± SEM of data from 3 independent infection experiments. *Significant difference between between COUP-TFII overexpression and control values at P<0.05 (D) RT-qPCR analysis of β-catenin, cyclin D1, axin 2, pdx1 and proinsulin I-II mRNA from islets from 5-week-old pdx1^CRE/-, COUP-TFII^Fl/Fl and pdx1^CRE/- COUP-TFII^Fl/Fl mice. Results are mean ± SEM (error bars) for 5 animals per genotype. *P<0.05 for pdx1^CRE/- COUP-TFII^Fl/Fl versus COUP-TFII^Fl/Fl and pdx1^CRE/- control mice.

Figure 5. GLP-1 signaling enhances cyclin D1 gene expression via COUP-TFII and β-catenin expression.

(A) GLP-1 and Exd4 stimulate IGF1-R expression in the 832/13 INS-1 β-cells. Wild-type 832/13 INS-1 β-cells were cultured in INS-1 medium and stimulated for 6 h with (+) or without (-) 100 nM GLP-1 or 10 nM Exd4 and IGF1-R mRNA levels were analyzed by RT-qPCR and normalized to housekeeping gene cyclophilin. Numbers over brackets indicate fold increase between Exd4 and GLP-1 treated samples and the control as indicated. Results are means ± SEM of data from 5 independent electroporation experiments. Significant differences between values linked by brackets are shown by asterisks at P value indicated. **P<0.03. ***P<0.001. (B) Absence of COUP-TFII abrogates Exd4 stimulated cyclin D1 expression in 832/13 INS-1 β-cells. 832/13 INS-1 β-cells were electroporated with scrambled or specific COUP-TFII siRNA, infected or not infected with recombinant β-catenin adenovirus and stimulated with (+) or without (-) 100 nM GLP-1 or 10 nM Exd4 for 6 h. Cyclin D1 mRNA levels was analysed by RT-qPCR. Numbers indicate fold-increase of cyclin D1 expression. Results are means ± SEM of data from 5 independent electroporation experiments. ***Significant differences between values linked by brackets at P<0.001. (C) In vivo crosstalk of GLP-1 and Wnt signaling pathways. RT-qPCR analysis of IGF-1R, cyclin D1 and axin 2 mRNA levels were performed on mouse islets isolated from GipR ^-/-; Glp-1-R^-/- transgenic mice normalized to housekeeping gene cyclophilin. Data are means ± SEM for 5 animals of each genotype. *Significant difference between transgenic and control values at P<0.05.

Figure 6. COUP-TFII controls GLP-1R expression in islet.

RT-qPCR analysis of IGF-1R and GLP-1R mRNA from islets from 5-week-old pdx1^CRE/-, COUP-TFII^Fl/Fl and pdx1^CRE/- COUP-TFII^Fl/Fl mice normalized to housekeeping gene cyclophilin. Results are mean ± SEM (error bars) for 5 animals per genotype. *P<0.05 for pdx1^CRE/- COUP-TFII^Fl/Fl versus COUP-TFII^Fl/Fl and pdx1^CRE/- control mice.

Figure 7. COUP-TFII expression is activated by TCF7L2 in β-cells.

(A) Fixed healthy human islets were triple-stained for COUP-TFII in red, for insulin in green and DAPI in blue. (B) 832/13 INS-1 β-cells were electroporated with scrambled (control) or specific TCF7L2 siRNA and cultured for 48 h. Relative TCF7L2, COUP-TFII and cyclin D1 mRNA levels were determined by RT-qPCR normalized to housekeeping gene cyclophilin. Results are means ± SEM of data from between 4 independent electroporation experiments. Significant differences between TCF7L2 knockdown and control values are shown by asterisks at the P value indicated. *P<0.05.

Figure 8. Proposed scheme of how COUP-TFII increases β-cell numbers through GLP-1 signaling cascade during the neonatal period in mice.

Positive regulation (arrows) of indirect COUP-TFII target's genes in pancreatic β-cells.