Fatty acid transport protein-2 inhibition enhances glucose tolerance through α-cell-mediated GLP-1 secretion

Abstract

Type 2 diabetes affects more than 38 million people in the US, and a major complication is kidney disease. During the analysis of lipotoxicity in diabetic kidney disease, global fatty acid transport protein-2 (FATP2) gene deletion was noted to markedly reduce plasma glucose in db/db mice due to sustained insulin secretion. To identify the mechanism, we observed that islet FATP2 expression was restricted to α-cells, and α-cell FATP2 was functional. Basal glucagon and alanine-stimulated gluconeogenesis were reduced in FATP2KO db/db compared to db/db mice. Direct evidence of FATP2KO-induced α-cell-mediated glucagon-like peptide-1 (GLP-1) secretion included increased GLP-1-positive α-cell mass in FATP2KO db/db mice, small molecule FATP2 inhibitor enhancement of GLP-1 secretion in αTC1-6 cells and human islets, and exendin[9-39]-inhibitable insulin secretion in FATP2 inhibitor-treated human islets. FATP2-dependent enteroendocrine GLP-1 secretion was excluded by demonstration of similar glucose tolerance and plasma GLP-1 concentrations in db/db FATP2KO mice following oral versus intraperitoneal glucose loading, non-overlapping FATP2 and preproglucagon mRNA expression, and lack of FATP2/GLP-1 co-immunolocalization in intestine. We conclude that FATP2 deletion or inhibition exerts glucose-lowering effects through α-cell-mediated GLP-1 secretion and paracrine ß-cell insulin release.

Keywords: Beta cells; Endocrinology; Glucose metabolism; Insulin; Metabolism.

Figure 1. Islet hypertrophy and increased β cell mass in FATP2-KO db/db mice.

Representative IHC images of pancreatic islets from db/db (A) and FATP2-KO db/db (B) mice. As described in Methods, α and β cells were labeled with glucagon and insulin antibodies, respectively. Scale bars: 100 μm. (C) β Cell mass was calculated as described in Methods for db/db and FATP2-KO db/db mice. Data represent the mean ± SEM. *P < 0.01, by Student’s t test.

Figure 2. FATP2 expression localizes to pancreatic α cells in mouse islets.

WT mouse pancreatic islets were immunohistochemically labeled for FATP2 expression in α, β, and δ cells as described in Methods. Merged images, representing cell-specific FATP2 expression, are shown in yellow. Micrometer scale bars are shown at the bottom right of each merged image.

Figure 3. FATP2 expression localizes to pancreatic α cells in human islets.

(A) Paraffin sections of human pancreas were immunohistochemically labeled for FATP2 expression in α and β cells as described in Methods. Merged images, representing cell-specific FATP2 expression, are shown in yellow. Micrometer scale bars are shown at the bottom of each image. (B) Gene expression correlation between GCG and FATP2 gene (SLC27A2) from 2 public normal human islet transcriptome datasets (GSE38642 and GSE50397) using online software (http://r2.amc.nl). Data were analyzed by linear regression and Pearson correlation.

Figure 4. FATP2 expression and function in α cells.

(A and B) FATP2 and loading control GAPDH mRNA expression was determined in human and mouse pancreatic tissue and α cell lines by RT-PCR (as described in Methods). Data are representative of 3 experiments per condition. (C) Mouse αTC1-6 cells were preincubated with lipofermata (1 hour, 37°C, 0–50 μM) in triplicate. BODIPY-labeled fatty acid uptake velocity was then determined, as described in Methods. Results show the mean ± SEM of 4 experiments.

Figure 5. Effects of FATP2 deletion on glucagon.

(A) Random (nonfasting) plasma glucagon concentrations in 4- to 6-month-old WT and db/db mice with or without FATP2 gene deletion. Each symbol in the scatter bars represents the mean from 1 sample assayed in duplicate (n = glucagon concentrations from 9–17 mice per genotype). *P < 0.05 compared with WT by ANOVA with Tukey’s post hoc test for multiple comparisons. (B) Serial glucose measurements in 4- to 6-month-old WT and db/db mice with or without FATP2 gene deletion (n = 3 mice per group) following alanine administration (2 g/kg i.p.). Data represent the mean ± SEM.

Figure 6. Blood glucose and plasma GLP-1 concentrations following OGTTs and IPGTTs.

(A) OGTTs and IPGTTs were conducted in db/db and FATP2-KO db/db mice, as described in Methods. Blood glucose levels were determined at the indicated times in 5 mice per group. (B) As an index of glucose disposal, the AUC corresponding to FATP2-KO db/db experiments in A was integrated using GraphPad Prism 7 software. (C) Plasma was obtained at baseline and at the 1-hour time point during the OGTTs or IPGTTs in FATP2-KO db/db mice. Data represent the mean ± SEM.

Figure 7. FATP2 and GLP-1 localization in the intestines.

Slc27a2 (A) and Gcg (B) mRNA expression levels were determined in mouse gut segments by qPCR, as described in Methods. Data were normalized to expression in stomach, which was defined as 1.0. Immunohistochemical labeling of FATP2 and GLP-1 in human distal ileum (C and D) (note that FATP2 is red and GLP-1 is green) and duodenum (E and F) (note that FATP2 is green and GLP-1 is red). Representative images from 5 mice are shown. Original magnification, ×200 (C and E) and ×600 (D and F). Data represent the mean ± SEM.

Figure 8. FATP2-KO/inhibition effect on glucose-stimulated GLP-1 and insulin secretion.

(A) Pancreatic GLP-1⁺ α cell mass was determined as described in Methods in db/db and FATP2-KO db/db mice. *P < 0.01 compared with the db/db group by t test. (B) Human islets were preincubated with or without lipofermata (LF) and then tested for glucose-stimulated GLP-1 secretion as described in Methods. *P < 0.01 compared with all other groups by ANOVA. (C) αTC1-6 cells were preincubated with or without lipofermata or palmitate (Palm) as indicated. Glucose-stimulated GLP-1 secretion was then measured as described in Methods. Each symbol in the scatter bars in B and C represents 1 sample that was assayed in duplicate (n = 3–6 samples per condition). *P < 0.05 compared lipofermata plus palmitate by ANOVA. (D) αTC1-6 cells coincubated in 5 mM or 25 mM glucose with or without 400 μM palmitate with or without 50 μM lipofermata for 16 hours were analyzed for Pcsk1 and Pcsk2 mRNA expression by qPCR. The results are expressed as the ratio relative to the 5 mM glucose-only condition. *P < 0.05 compared with other groups by ANOVA. (E) Glucose-stimulated insulin secretion was measured in human islets, which were preincubated with or without lipofermata and then exposed to exendin[9-39] (Ex) or palmitate, as described in Methods. Each symbol in the scatter bars represents 1 sample that was assayed in duplicate (n = 3 samples per condition). *P < 0.01 compared with 16.8 mM glucose plus the lipofermata group by ANOVA. Data represent the mean ± SEM.