GATA6 Controls Insulin Biosynthesis and Secretion in Adult β-Cells
Abstract
GATA4 and GATA6 play essential, but redundant, roles in pancreas formation in mice, and GATA6 mutations cause pancreatic agenesis in humans. GATA6 mutations have also recently been linked to adult-onset diabetes, with subclinical or no exocrine insufficiency, suggesting an important role for GATA6 in human β-cell physiology. To investigate the role of GATA6 in the adult endocrine pancreas, we generated mice in which Gata6 is specifically inactivated in the pancreas. These mice develop glucose intolerance. Islets deficient in GATA6 activity display decreased insulin content and impaired insulin secretion. Gata6-deficient β-cells exhibit ultrastructural abnormalities, including increased immature insulin granules, swollen mitochondria, and disorganized endoplasmic reticulum. We also demonstrate that Pdx1 expression in adult β-cells depends on GATA sites in transgenic reporter mice and that loss of GATA6 greatly affects β-cell–specific gene expression. These findings demonstrate the essential role of GATA6 in β-cell function.
Introduction
The formation of β-cells during embryonic development is regulated by a notable number of transcription factors that activate lineage-specific genes. Interestingly, many of these transcription factors are also important in pancreas function during adult life. Members of the GATA zinc finger transcription family are crucial for pancreas organogenesis in both human and mouse (1–3). Exome sequencing studies have revealed that heterozygous GATA6 mutations are the most common cause of pancreatic agenesis (3,4). Mutations in GATA4 have also been linked to neonatal diabetes and pancreatic agenesis in humans (5,6). In the mouse, GATA4 and GATA6 play redundant roles in embryonic pancreas formation. Single inactivation of Gata4 or Gata6 in pancreatic progenitor cells does not have a major effect in pancreas organogenesis. However, the simultaneous inactivation of both Gata4 and Gata6 results in pancreatic agenesis due to defects in pancreatic progenitor cell proliferation and differentiation (1,2).
The expression of these two GATA transcription factors is not restricted to the embryonic stages of pancreas formation. Thus, in the adult pancreas, Gata4 is expressed in the acinar compartment, whereas Gata6 is expressed in the endocrine and exocrine compartments. Interestingly, recent genetic studies have revealed that GATA6 mutations are also linked to adult-onset diabetes with subclinical or no exocrine insufficiency. Moreover, GATA6 deficiency in human pluripotent stem cells altered in vitro directed differentiation toward β-like cells as well as their subsequent functionality (7–9). Taken together, these studies indicate that in addition to pancreas development, GATA6 plays a critical role in β-cell function that remains uncharted. Here, we thoroughly analyze mice with single inactivation of Gata6 in the pancreas. We show that although Gata6-deficient mice are apparently normal at birth, as they age they develop glucose intolerance. Furthermore, loss of Gata6 causes impairment of β-cell insulin biosynthesis and secretion and markedly affects the islet transcriptome.
Research Design and Methods
Mice
Gata6flox/flox, Pdx1-Cre, Pdx1-wt-lacZ, and Pdx1-mut-lacZ mice have been previously described (1,10,11). All experiments using animals were approved by the University of Seville Institutional Animal Care and Use Committee, Seville, Spain.
Histology, Immunohistochemistry, and Immunofluorescence
Histological and immunohistochemical analyses were performed, as previously described (12), on at least four different sections from five different mice. Primary antibodies were used at the indicated dilution: anti-GATA6 (1:1,000, AF 1700; R&D Systems), anti-Pdx1 (1:200, ab47308; Abcam), anti-Nkx6.1 (1:50, F55A10; DSHB), anti-Nkx2.2 (1:25, 74.5A5; DSHB), anti-glucagon (1:100, G2654; Sigma-Aldrich), anti-insulin (1:500, I2018; Sigma-Aldrich), anti-GLUT2 (1:300, 07-1402; Millipore), anti–β-galactosidase (1:500, 559762; MP Biochemicals), anti–C-peptide (1:100, 05-1109; Millipore), anti-MafA (1:50; Bethyl Laboratories), anti-somatostatin (1:1,000; Santa Cruz Biotechnology), anti-pancreatic polypeptide (1:100; Millipore), and anti-inducible nitric oxide synthase (iNOS; 1:100; Thermo Fisher Scientific).
Transmission Electron Microscopy and Insulin Granules Quantification
For transmission electronic microscopy (TEM) analysis, isolated pancreatic islets were processed using a standard Spurr protocol (13). TEM images were taken with an EMCCD camera (TRS 2k × 2k). Quantifications of insulin granules were performed on 20 TEM images obtained from isolated islets of control and Gata6 knockout (KO) mice (n = 6), as previously described (14,15), using ImageJ software.
Islets Isolation
Islets were isolated from pancreata of male Gata6 KO and control mice by the collagenase digestion method (16). Islets were handpicked after several purification steps and used for RNA isolation or kept in culture overnight for insulin secretion tests.
Glucose and Insulin Tolerance Test
After a 16-h fast, mice were weighed, and an Accu-Chek Aviva glucometer was used to measure glucose levels. Mice were injected intraperitoneally with glucose (2 g/kg of body weight). Blood glucose levels were measured every 30 min. Blood for in vivo insulin measurements was collected from the tail vein before and at 5 and 30 min after the glucose injection. Insulin concentration was measured using the Insulin ELISA kit (Mercodia) following the manufacturer’s instructions. The insulin tolerance test was performed as previously described (17).
Insulin Secretory Response
For in vitro insulin secretion, isolated islets were incubated in 2.8 mmol/L glucose or 16.7 mmol/L glucose for 1 h at 37°C. Ten matched-size islets were used per triplicate per condition. After incubation, media was collected to measure insulin using the Insulin ELISA kit (Mercodia). Islets were collected, spun down, and lysed using a 30-gauge syringe with Tris-EDTA buffer for measuring total insulin content.
C-Peptide and Proinsulin Content
Blood was collected from the tail vein of mice after a 16–h fast, and C-peptide and proinsulin levels were measured using the C-peptide ELISA kit (Crystal Chem) and Proinsulin ELISA kit (Mercodia), respectively, following the manufacturers’ recommendations.
β-Cell Area Quantification
β-Cell area quantification was performed as described (18). Immunohistochemistry images were taken with a Nikon microscope and processed using ImageJ software. β-Cell area was calculated as the percentage of insulin area–to–total pancreatic area (marked by DAPI staining).
Pancreas and Islets Insulin Content
Dissected pancreata were minced using a Polytron homogenizer in Tris-EDTA buffer, and isolated islets were lysed in Tris-EDTA buffer using a 30-gauge syringe. Debris was spun down, and the supernatant was collected to measure insulin, as described above. Total pancreatic protein content and islets DNA was measured using Bradford reagent (Sigma-Aldrich) and the Quant-iT PicoGreen dsDNA Assay kit (Invitrogen), respectively.
Microarray Analyses and Quantitative PCR
Isolated islets from three control or three Gata6 KO mice were pooled to obtain RNA. Total RNA was isolated using RNeasy Plus Micro Kit (Qiagen). Gene profile was performed in three independent pools of control and four pools of Gata6 KO islets using the Affymetrix GeneChip Mouse Gene 2.0 ST Array. Differentially expressed genes were defined as those for which the nominal P was 0.05. Gene expression data are available through the Gene Expression Omnibus database (accession number GSE106316) (19). Quantitative (q)PCR was performed from at least seven independent pools of control and Gata6 KO islets. qPCR was performed using the following TaqMan probes (Applied Biosystems): Ins1: Mm01950294_s1; Ins2: Mm00731595_gH; Slc2a2: Mm00446224_m1; Pdx1: Mm00435565_m1; Abcc8: Mm00803450_m1; Nkx6.1: Mm00454961_m1; Nkx2.2: Mm03053916_s1; Pcsk1: Mm00479023_m1; Gcg: Mm00801714_m1; Sst: Mm00436671_m1; Gata6: Mm00802632_m1; Gata4: Mm00484689_m1; Irs2: Mm03038438_m1; Cacna1c: Mm01188822_m1; Nos2: Mm0040502_m1; and β-actin: Mm02619580_g1. The following primers were used to quantify gene expression by SYBR Green: MafA: forward 5′-GAGGAGGTCATCCGACTGAAA-3′, reverse 5′-GCACTTCTCGCTCTCCAGAAT-3′; Snap25: forward 5′-GAGAACCTGGAGCAGGTGAG-3′, reverse 5′-AGCATCTTTGTTGCACGTTG-3′.
Statistical Analysis
Data in scatter plots and graph bars are presented as mean ± SD and mean ± SEM, respectively. The Student t test was performed with a level of significance of P < 0.05. Two-way ANOVA with repeated measures with Bonferroni correction were used to analyze intraperitoneal glucose tolerance test data.
Results
Pancreas-Specific Inactivation of Gata6 Leads to Glucose Intolerance With Aging
To elucidate the role of GATA6 in β-cell function, we generated pancreas-specific Gata6 KO mice by crossing mice with a conditional (flox) allele of Gata6 (Gata6flox/flox) with mice that express the Cre recombinase under the control of the Pdx1 promoter (Pdx1-Cre mice) (1). Efficient Gata6 inactivation in the pancreas was demonstrated by immunohistochemistry and qPCR (Fig. 1A–C). Clear elimination of GATA6 was observed in exocrine and islet cells of Gata6flox/flox; Pdx1-Cre mice (Gata6 KO mice, hereafter) (Fig. 1B). Gross morphology and histological analysis did not reveal any apparent pancreatic abnormalities in newborn (Fig. 1D and E) or 2-month-old (Fig. 1F and G) Gata6 KO mice. As mice aged, we observed mild loss of acinar cells, acinar- to ductal-metaplasia lesions, and pancreatic lipomatosis in Gata6 KO mice (Fig. 2H–O), in agreement with previous studies describing GATA6 as an essential factor for the maintenance of acinar identity (20). However, acinar cell morphology was not largely disrupted, as assessed by cell polarity markers (Supplementary Fig. 1). Similarly, overall exocrine pancreatic function was not compromised, as determined by plasma amylase activity (Supplementary Fig. 1). Histological (Fig. 2A and H) and immunohistochemical analysis for pancreatic hormones revealed that although the overall architecture of Gata6 KO islets was not markedly affected, insulin levels appeared lower in Gata6-deficient β-cells (Fig. 2C–H). To determine the functional requirement for GATA6 in the adult endocrine pancreas, we examined glucose homeostasis in Gata6 KO mice. Male and female Gata6 KO mice at 2 and 4 months old displayed normal glucose tolerance (Fig. 3A–D and Supplementary Fig. 2). However, by age 6 months, male and female Gata6 KO mice displayed a significant impairment in glucose tolerance compared with control mice (Fig. 3E and F and Supplementary Fig. 2), a phenotype that was maintained at 12 months of age (Fig. 3G and H). Despite the abnormal glucose tolerance, no differences in basal glycemia under fasting conditions were observed between Gata6 KO and control mice (Fig. 3I). Because Gata6flox/flox, Gata6flox/+; Pdx1-Cre, Gata6flox/+, and Pdx1-Cre mice exhibited similar phenotypes in glucose tolerance (Supplementary Fig. 3), we only included Gata6flox/flox littermates as controls in all subsequent experiments. The glucose intolerance phenotype of Gata6 KO mice was not associated to insulin resistance, as shown by insulin tolerance tests (Fig. 3J). Thus, our data indicate that Gata6 inactivation in the pancreas compromises glucose homeostasis.
Gata6-deficient pancreata have disrupted acinar compartment but normal islet architecture. A: Immunohistochemistry analysis on pancreatic sections using an anti-GATA6 antibody reveal robust nuclear GATA6 protein in both acinar and islet tissue in 2-month-old control (Ctrl) mice. Efficient inactivation of Gata6 floxed allele is demonstrated by the lack of GATA6 protein in the endocrine and exocrine pancreatic compartments of Gata6 KO (G6KO) mice (B) and by the decrease of GATA6 transcripts measured by qPCR using TaqMan probes (C). Insets in A and B are magnifications of selected areas. Values are mean ± SEM for n = 7 pools of islets for each group (each pool contains islets from three mice). Statistical significance between groups was determined by Student t test. **P < 0.01. D–O: Hematoxylin and eosin staining of G6KO and Ctrl pancreatic sections of mice at postnatal day 1 (P1) and at 2, 4, 6, and 12 months of age are shown. Acinar tissue of G6KO pancreata is focally replaced by adipose as early as 4 months of age (I, arrowheads) and increases with age (arrowheads in K and O). L and M: Acinar-to-ductal metaplasia lesions are also focally observed. The boxed area in L is shown at higher magnification in M. Scale bars: 100 μm.
Islets from Gata6 KO (G6KO) mice display normal architecture. G6KO pancreata display normal islet architecture and distribution of endocrine cells compared with control (Ctrl) pancreata at 6 months of age, as shown by histological analysis (A and B) and immunofluorescence analysis using insulin, glucagon, somatostatin (Sst), and pancreatic polypeptide (PP) (C–H). H&E, hematoxylin and eosin. C–H: Insulin immunoreactivity is markedly decreased in G6KO islets compared with Ctrl islets. Scale bars: 25 μm.
Gata6 KO (G6KO) mice develop glucose intolerance. A–H: Intraperitoneal glucose tolerance tests in G6KO (■) and control (Ctrl; ▲) mice. Blood glucose concentration at baseline and at 30, 60, 90, and 120 min after glucose administration in control and G6KO mice at 2 (A), 4 (C), 6 (E), and 12 months of age (G) are shown (n ≥ 18 male mice for each genotype). Glucose total area under the curve (AUC) in arbitrary units (AU) is similar between Ctrl and G6KO mice at 2 and 4 months of age (B and D) but significantly increased in G6KO mice compared with Ctrl mice at 6 and 12 months of age (F and H). I: Basal glucose levels under fasting conditions are similar in Ctrl and G6KO mice at 6 and 12 months of age. J: Intraperitoneal insulin tolerance tests do not reveal differences between G6KO mice and Ctrl mice at 6 months of age (n = 6 mice for each genotype). Data are expressed as mean ± SEM. *P < 0.05; **P < 0.01.
Gata6 KO Mice Display Reduced Pancreatic Insulin Content and Defects in Insulin Secretion
The glucose intolerance phenotype observed in 6-month-old Gata6 KO mice could be due to decreased β-cell numbers. Nonetheless, quantification of β-cell area did not reveal significant differences between 6-month-old Gata6 KO and control littermates (Fig. 4A). Similarly, no differences in β-cell apoptosis and proliferation were observed between Gata6 KO and control islets (Supplementary Fig. 4). Because immunofluorescence analysis revealed a potential reduction in insulin accumulation in β-cells (Fig. 2C–H), we measured total pancreatic insulin content. A significant decrease in insulin levels was observed in the Gata6 KO pancreas compared with the control pancreas (Fig. 4B). Moreover, the insulin content of isolated Gata6 KO islets was reduced by 2.5-fold compared with isolated control islets (Fig. 4C).
Decreased insulin content and insulin secretion in response to glucose in Gata6 KO (G6KO) mice. A: β-Cell area is represented as the ratio of the area of insulin-positive cells per total pancreatic area using ImageJ software. No differences in β-cell area between 6-month-old control (Ctrl) and G6KO mice are observed (n = 5 mice each genotype). B: Whole pancreatic insulin content in G6KO mice is decreased compared with Ctrl mice at 6 months of age (n = 5 mice each genotype). C: Isolated G6KO islets (n = 13 mice) show a significant reduction in insulin content compared with Ctrl islets (n = 14 mice). D: Plasma insulin levels at baseline and after 5 and 30 min of intraperitoneal glucose load is significantly lower in G6KO mice than in Ctrl mice at 6 months of age (n = 10 mice each genotype). E: In vitro insulin secretion assay on isolated islets of 6-month-old Ctrl and G6KO mice demonstrates that G6KO islets secrete less of their total insulin content than the Ctrl islets at baseline and under high-glucose concentrations (16.7 mmol/L) (n = 10 mice each genotype). Data in A and B are expressed as mean ± SD and in C–E as mean ± SEM. *P ≤ 0.05.
We next investigated whether Gata6 inactivation affected insulin secretion. To this aim, we measured circulating insulin levels in 6-month-old mice subsequent to an intraperitoneal glucose tolerance test. Gata6 KO mice exhibited impaired insulin secretion under basal conditions and upon glucose challenge (Fig. 4D). To directly analyze β-cell function, insulin secretion assays were performed on islets isolated from 6-month-old control and Gata6 KO mice. Similar to what we observed in vivo, insulin secretion from Gata6 KO islets was reduced compared with control islets under basal and high glucose concentrations (Fig. 4E). Nonetheless, the difference between basal and high glucose was maintained in Gata6-deficient compared with control islets, suggesting a dampened yet preserved glucose-induced insulin response (Fig. 4E). To confirm these findings, we measured intracellular calcium flux in isolated islets at different glucose concentrations using fluorescent calcium fluorescence sensor Fura-2. No marked differences between Gata6 KO and control islets were observed (Supplementary Fig. 5). Altogether, our results indicate that the glucose intolerance phenotype of Gata6 KO mice might be due to defects in both insulin production and secretion.
Key Genes in Insulin Biosynthesis and Secretion Are Downregulated in Gata6-Deficient Islets
To gain insight into the role of GATA6 in islet function, we performed microarray analysis comparing Gata6 KO and control islets. To avoid any potential interference of glucose intolerance in islet expression, gene expression was analyzed in islets collected from 2-month-old mice when they still display normal glucose tolerance. The most downregulated genes were genes involved in different aspects of β-cell function, such as glucose sensing (Slc2a2, encoding for GLUT2 transporter), insulin biosynthesis (Pcsk1), insulin secretion (Abcc8; Cacna1c; SNAP-25), and major transcriptional regulators of adult β-cell function such as Pdx1, Nkx2.2, Nkx6.1, and MafA (21–24). Insulin transcripts (both Ins1 and Ins2) of Gata6 KO islets also displayed a notable decrease compared with control islets, in agreement with the observed reduction in pancreatic and islet insulin content observed in 6-month-old mice (Fig. 2B and C). Gene expression analysis by qPCR confirmed the marked downregulation of these key mature β-cell markers in Gata6-deficient islets (Fig. 5A). However, the expression of other islet hormone genes, such as glucagon and somatostatin, was not decreased in Gata6 KO islets (Fig. 5A). Of note, levels of Gata4 transcripts were similar between Gata6-deficient and control islets (Fig. 5A).
Changes in β-cell gene expression in Gata6-deficient islets. A: qPCR from total RNA of pooled of isolated islets. Transcripts levels of key genes involved in insulin synthesis and secretion are dramatically decreased in Gata6 KO (G6KO) islets compared with control (Ctrl) islets. No significant changes are observed in glucagon, somatostatin, and Gata4 transcripts. Ctrl mice: ■; G6KO mice: □. Gene expression data are normalized to β-actin mRNA levels and presented as mean ± SEM (n = 7 pools of isolated islets for each group; each pool contains islets from three mice). *P ≤ 0.05; **P ≤ 0.01. B–K: Immunofluorescence analysis confirms the reduced accumulation of Pdx1, Glut2, Nkx2.2, Nkx6.1, MafA, and C-peptide proteins in G6KO islets compared with Ctrl islets. Scale bars: 25 μm.
To determine whether these changes in gene expression in Gata6 KO islets correlated with changes in protein accumulation, we performed immunohistochemical and immunofluorescence analyses in 2-, 4-, and 6-month-old mice. At both 2 and 4 months of age, a mild reduction in insulin and C-peptide was observed in Gata6-deficient islets compared with control islets. However, plasma insulin and C-peptide levels were normal in fasted 2-month-old Gata6 KO mice (Supplementary Figs. 6 and 7). Moreover, no apparent decrease in the accumulation of other β-cell markers, such as Pdx1, Glut2, Nkx2.2, and Nkx6.1, was observed in Gata6 KO islets compared with control islets (Supplementary Fig. 6). At 6 months of age, Pdx1, Glut2, Nkx2.2, Nkx6.1, and MafA levels were markedly diminished in Gata6 KO islets compared with control islets (Fig. 5B–I). In addition, islets lacking GATA6 displayed a notable reduction in C-peptide (Fig. 5J and K). Circulating levels of insulin and C-peptide were also decreased in Gata6 KO mice (Supplementary Fig. 7). Thus, Gata6 inactivation significantly affects the expression of critical β-cell–specific genes.
We next determined whether Gata6 depletion in adult β-cells affected β-cell gene expression. To this end, we inactivated Gata6 in isolated islets from 2-month-old Gata6flox/flox mice by infection with adenovirus expressing Cre. Although an efficient Gata6 inactivation was achieved, the expression of key β-cell genes was unaffected in Gata6-depleted islets, in contrast to what it was observed in GATA6 KO mice of the same age (Fig. 2 and Supplementary Fig. 6).
Gata6-Deficient β-Cells Exhibit Increased Immature Insulin Granules, Swollen Mitochondria, and Disorganized Endoplasmic Reticulum
To further explore the role of GATA6 in β-cell function we used the Ingenuity Pathway Analysis software to identify pathways affected by Gata6 inactivation. Consistent with our previous results, pathways related to insulin and β-cell function, such as “maturity-onset diabetes of the young signaling” and “insulin receptor signaling,” were found among the top canonical pathways (Fig. 6A). Other pathways identified were related to translational control and, interestingly, mitochondrial dysfunction (Fig. 6A). Indeed, several genes involved in mitochondrial function, including respiratory chain components (i.e., Cox6a2, Cyp4f39) and mitochondrial transporters (i.e., Slc25a12, Slc25a22), were downregulated in Gata6 KO islets (Supplementary Table 1). Because a link between mitochondrial dysfunction and defective β-cell function has been proposed (25,26), we used TEM to perform ultrastructural analysis of 6-month-old Gata6 KO and control β-cells. Mitochondria of Gata6 KO β-cells appeared swollen, with disorganized inner-membrane cristae (Fig. 6C and E), in contrast to the dense and organized cristae of mitochondria in control β-cells (Fig. 6B and D). TEM analysis revealed additional ultrastructural abnormalities in Gata6-deficient β-cells compared with control mice (Fig. 6F and G). Thus, extension of the endoplasmic reticulum (ER) was observed in Gata6 KO β-cells, indicative of ER stress (Fig. 6G). An association among mitochondrial dysfunction, ER stress, and induction of iNOS has been reported (27,28). Consistent with this notion, we observed a marked increase in iNOS expression in Gata6 KO islets as assessed by immunofluorescence and qPCR analysis (Fig. 7). TEM analysis also showed a marked reduction in the total number of insulin granules in Gata6-deficient β-cells compared with control cells (Fig. 6B–K). The secretory insulin granules from control β-cells featured an electron-dense core and a translucent halo appearance, indicative of mature secretory granules (Fig. 6B, D, F, and H). However, Gata6 KO β-cells displayed a decreased number of dense granules and a prominent number of gray low-dense granules (Fig. 6C, E, G, I, and J), indicating an increase in immature insulin granules. Despite the increase of immature secretory granules, no differences in the ratio of proinsulin to insulin circulating levels were found between Gata6 KO and control mice (Supplementary Fig. 7). Altogether, our results indicate that GATA6 is necessary to preserve the integrity of the ER and mitochondria and the formation of insulin secretory vesicles in β-cells.
Ultrastructure analyses of Gata6-deficient islets. A: Ingenuity Pathway Analysis from microarray data reveals altered signaling pathways in Gata6-deficient islets. Representative images of TEM of control (Ctrl) (B, D, F, and H) and Gata6 KO (G6KO) (C, E, G, and I) β-cells from 6-month-old mice are shown. Lower magnification of TEM images of Ctrl (B) and G6KO β-cells (C). The mitochondria (m) of β-cells from G6KO mice are severely swollen with disordered cristae (E) compared with normal mitochondrial morphology in Ctrl β-cells (D). The endoplasmic reticulum (er) cisternae of β-cells lacking GATA6 appeared distended and extremely disorganized (G) compared with Ctrl β-cells (F). Dense core insulin granules (arrowheads) are reduced in G6KO β-cells (I) compared with Ctrl β-cells (H). Immature secretory granules are prominent in Gata6-deficient β-cells (arrows) (I). J: Quantification of insulin secretory granules according to the core; dense, gray, rod-shaped, or empty (Ctrl β-cells: ■; G6KO β-cells: □). K: Total insulin secretory granules are decreased in G6KO β-cells compared with Ctrl β-cells. Scale bars: 2 μm in B and C, 500 nm in D–G, and 1 µm in H and I.
Accumulation of iNOS in Gata6-deficient β-cells. Immunofluorescence analysis of iNOS in islets of control (Ctrl) and Gata6 KO (G6KO) mice at 2 months (A–F) and 6 months (G–L) of age. Scale bars: 25 µm. M: qPCR of iNOS expression in islets of Ctrl and G6KO mice at 6 month of age (n = 10 mice). Data are expressed as mean ± SEM. *P ≤ 0.05.
GATA6 Directly Regulates Pdx1 Expression in Adult Islets
We previously showed that GATA factors directly regulate Pdx1 expression in the pancreas during embryonic stages by binding to two deeply conserved GATA sites in area III of Pdx1 regulatory sequences (1). We decided to determine whether the GATA sites in area III also contribute to Pdx1 expression in adult islets by analyzing Pdx1-wt-lacZ mice, transgenic mice that harbor the cis-regulatory region of Pdx1-containing areas I, II, and III fused to the lacZ reporter gene. This transgenic construct faithfully recapitulated the endogenous expression of Pdx1 in adult islets in three independent Pdx1-wt-lacZ transgenic lines. Thus, β-galactosidase localization directed by Pdx1-wt-lacZ completely overlapped with PDX1 accumulation in pancreatic islets (Fig. 8A–D and I–L). A similar pattern of β-galactosidase expression was observed in islets of transgenic mice containing mutations in Pdx1-area III GATA sites at 11 weeks of age in two independent transgenic lines analyzed (Fig. 8E–H). However, mutations in Pdx1-area III GATA sites resulted in a dramatic decrease of β-galactosidase expression in 6-month-old mice, with only a few β-galactosidase–positive cells found in the Pdx1 expression domain (Fig. 8M–P). These results indicate that GATA factors directly regulate Pdx1 expression in adult islets through GATA sites present in area III of the Pdx1-cis–regulatory region.
Conserved GATA sites in area III are required for Pdx1 expression in adult islets. Immunofluorescence analysis on pancreatic sections of 2- (A–H) and 6-month-old (I–P) Pdx1-wt-lacZ mice reveals a complete overlapping expression of β-galactosidase and endogenous Pdx1 expression. E–H: β-Galactosidase expression in islets of Pdx1-mut-lacZ mice closely mirrors the Pdx1 endogenous expression in 2-month-old mice. M–P: By contrast, only a few Pdx1-positive cells express β-galactosidase in the islets of Pdx1-mut-lacZ mice at 6 months of age. Scale bars: 25 μm.
Discussion
Previous studies have shown that GATA4 and GATA6 play essential, albeit redundant, roles during embryonic pancreas development (1,2). Here we show that conditional inactivation of Gata6 in the pancreas results in compromised glucose tolerance, mainly due to defects in insulin production and secretion.
Gata6-deficient mice showed a significant reduction in β-cell insulin content, including decreased insulin transcription. Impaired formation of insulin secretory vesicles could also contribute to the reduced insulin secretion of Gata6-deficient islets. TEM analysis showed that β-cells lacking GATA6 displayed decreased insulin granules but, more interestingly, increased immature insulin secretory vesicles. Increased immature insulin secretory vesicles can be associated to defective insulin processing. However, we failed to observe significant changes in the ratio of proinsulin to insulin blood levels in Gata6 KO mice. Consistent with the defective insulin secretion phenotype, a decrease in the expression of SNAP-25, a t-SNARE protein that participates in the exocytosis of secretory granules (29), was observed in Gata6-deficient islets. Gata6-deficient islets also displayed reduced expression levels of Slc2a2 and Abcc8 that encode GLUT2 and SUR1, respectively, two key proteins critical for glucose-stimulated insulin secretion (30,31). However, although basal insulin secretion was diminished, stimulation of secretion by glucose appeared to be unaffected by Gata6 inactivation. In agreement with this, the intracellular calcium response to glucose was not affected in Gata6-deficient β-cells, indicating that GLUT2 and SUR1 expression was not reduced to levels low enough to affect glucose-induced insulin secretion.
Gata6 inactivation results in decreased expression of β-cell–enriched transcription factors, including Pdx1, Nkx6.1, and MafA, which are essential for insulin transcription and secretion (23,32,33). Thus, we cannot rule out that the effects of GATA6 activity on β-cell function could be indirect. We previously demonstrated that GATA6 directly regulates Pdx1 expression during early pancreatic development through GATA sites present in area III of the Pdx1 regulatory sequence (1). We now show that, interestingly, these GATA sites are also required for Pdx1 expression in adult β-cells. Altogether, our data reveal GATA6 as a key component of the transcriptional network regulating β-cell function.
Despite multiple defects in insulin transcription, biosynthesis, and secretion, Gata6 KO mice do not become overtly hyperglycemic. We did not observe any compensatory increase in β-cell area or insulin sensitivity that could explain the lack of hyperglycemia, but we cannot rule out other compensatory mechanisms such as neural regulation or incretin hormone secretion. Interestingly, inactivation of other β-cell transcription factors, such as Rfx6 and NeuroD, also results in β-cell dysfunction and glucose intolerance without developing overt hyperglycemia (34,35).
Our microarray analysis revealed a marked aspect of the gene expression profile of Gata6 KO islets, the impairment of mitochondrial gene expression. In agreement with these results, TEM analysis revealed swelling of mitochondria, exhibiting disrupted cristae in Gata6-deficient β-cells. Similar mitochondrial alterations have been associated to β-cell failure, as reviewed by Supale et al. (25). Thus, mitochondrial dysfunction might also underlie the impairment of β-cell function in Gata6 KO mice. ER abnormalities were also observed in Gata6-deficient β-cells in concordance with the increasing body of evidence suggesting that mitochondrial and ER function are interconnected in β-cells (36). Interestingly, a marked increase in iNOS expression was found in Gata6 KO islets, a protein associated with mitochondrial dysfunction and ER stress (iNOS) has been reported (27,28). The mTOR signaling may also play a role in the interplay between mitochondrial and ER function according to recent reports (37,38) Interestingly, Ingenuity Pathway analysis predicted activation of the mTOR signaling pathway in Gata6 KO islets. Further studies will be required to clarify the direct role of GATA6 in the regulation of these pathways in β-cells.
Our results are in apparent contradiction with two reports describing inactivation of Gata6 in mouse β-cells. The inactivation of Gata6 using a Ptf1a-Cre line did not have any effect in glucose homeostasis in 30-week-old mice (20). Similarly to the Gata6flox/flox;Pdx1-Cre mice used in this study, the inactivation of Gata6 in Gata6flox/flox;Ptf1a-Cre mice also occurred early in pancreatic progenitors but at slightly different times (39). A possible explanation to the discrepancy in the glucose tolerance phenotype between these mouse models is that GATA6 might be required for the proper maturation and function of adult β-cells during a specific developmental time window in which Pdx1, but not Ptf1a, is expressed. Differences in Cre recombinase efficiency in endocrine cells of the two mouse models might also contribute to the observed discrepancies (40).
In other study, the inactivation of Gata6 specifically in adult β-cells (using the Ins2-CreERT mouse strain) did not affect glucose homeostasis (41). However, the authors report inefficient Gata6 deletion that could explain the lack of metabolic phenotypes on their mice. In contrast, we observed substantial Gata6 inactivation using the Pdx1-Cre line. An alternative possibility is that Pdx1-Cre–mediated Gata6 inactivation during embryonic stages causes the observed defects later in adult β-cells. In this regard, we performed in vitro Gata6 depletion in adult mouse islets without any apparent effect on β-cell gene expression. Although these experiments are far from conclusive regarding a possible developmental origin of Gata6 KO defects, they suggest the Gata6 inactivation does not have an immediate effect on β-cell function. Interestingly, despite the early inactivation of Gata6 by using the Pdx1-Cre line, Gata6 KO mice do not become clearly glucose intolerant until 6 months of age. The progressive defect in glucose homeostasis is not uncommon in mouse models of β-cell dysfunction and might be due to increased metabolic stress associated to age and concomitant increase in body weight. However, 6-month-old Gata6 KO mice develop exocrine defects that could potentially affect islet function. Although we cannot formally rule out this effect, we consider this an unlikely possibility based on the following observations. First, compromised expression of β-cell genes is found in 2-month-old mice, before overt exocrine defects are observed. Second, plasma amylase levels in 2- and 6-month-old Gata6 KO mice are not increased, indicating the functional effect of pancreatic injury was minor. Finally, 30-week-old Gata6flox/flox;Ptf1a-Cre mice display extensive pancreatic exocrine abnormalities but no defects in glucose homeostasis.
Recent genetic studies have revealed that GATA6 mutations can cause adolescent/adult-onset diabetes with subclinical or no exocrine insufficiency. Unfortunately, little is known about deficits in β-cell function in this specific group of patients. However, two recent studies using in vitro–directed differentiation of β-like cells from human pluripotent stem cells indicate a critical role for GATA6 in β-cell function (7,8). GATA6 mutant β-like cells display impaired insulin secretion and downregulation of key genes involved in β-cell function. Our results are in agreement with these reports demonstrating that GATA6 activity is necessary for β-cell function in mice. The elucidation of the mechanisms underlying diabetes onset on patients with GATA6 mutations will shed light on the differential requirements of GATA6 for pancreas development versus adult β-cell function.
In summary, our results show that loss of Gata6 markedly decreases the expression of insulin, key components of insulin synthesis and secretion machinery, and β-cell–enriched transcription factors, demonstrating an essential role for GATA6 in β-cell function.
Article Information
Acknowledgments. The authors thank Antonio Cárdenas and Alberto Morante, from Centro Andaluz de Biología Molecular y Medicina Regenerativa (CABIMER), for their technical assistance. The authors also thank Juan Luis Ribas, from the University of Seville, for the electronic microscopy technical services, and Eloisa Andújar and Mónica Pérez, from the Genomic Unit of CABIMER. The authors thank the Andalusian Bioinformatics Platform at the University of Malaga (www.scbi.uma.es) for providing access to Ingenuity Pathway Analysis software.
Funding. L.V. was supported by a contract from Spanish Ministry of Economy and Competitiveness (RYC-2013-14533). E.R.-S. was supported by a “Juan de la Cierva” postdoctoral fellowship from Spanish Ministry of Economy and Competitiveness (IJCI-2014-19251). This work was supported by grants from ISCIII cofunded by Fondos FEDER (PI14/01015, RD/0019/0028, and RD16/0011/0034 to B.S. and PI14/0804 to A.R.) and “Ramón y Cajal” program from the Spanish Ministry of Economy and Competitiveness (PI14/0804, RYC-2013-14533 to A.R.). P.M. was supported by a Juan de la Cierva fellowship from Spanish Ministry of Science and Innovation and by a Stand-Alone Grant from the Austrian Science Fund (FWF P27361-B23). Work by I.Q. was supported by the Spanish Ministry of Economy and Competitiveness (BFU2016-77125-R). Work by D.A.C. was supported by the Nicolás Monardes program of Andalusian Ministry of Health (C-0015-2014) and the Andalusian Ministry of Science and Innovation (CTS-7478).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. L.V., E.R.-S., R.A., M.C., E.B.-T., and J.M.M.-G. performed the experiments. B.R.G., P.M., I.Q., B.S., F.M., D.A.C., and A.R. conceived the data. I.Q., D.A.C., and A.R. analyzed the data. D.A.C. and A.R. wrote the manuscript. All authors edited the manuscript. A.R. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Footnotes
This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db17-0364/-/DC1.
- Received March 23, 2017.
- Accepted December 13, 2017.
- © 2017 by the American Diabetes Association.
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.
References
In this Issue
