Microphthalmia Transcription Factor Regulates Pancreatic β-Cell Function

Mice with a point mutation (C to T at nucleotide 916) in the Mitf gene have previously been characterized (19). The Mitf cloudy eye (Mitf^ce/ce) mutation results in a lack of the Mitf zipper domain due to the presence of a stop codon between the helix-loop-helix and zipper motif. This leads to the synthesis of a truncated protein incapable of DNA binding (19). Mitf^ce/ce mice have white coat color and small eyes. Heterozygous and wild-type animals both have black coat color. For distinguishing between the latter, DNA samples were sequenced (forward, ggtccattgtcttgttttatcacag; reverse, gtatccaccctctgccat). Wild-type, heterozygous, and homozygous animals are distinguished by C, C/T, and T at bp 916 of the Mitf gene (http://www.informatics.jax.org/searches/accession_report.cgi?id=mgi:1856532). Wild-type and heterozygous Mitf animals were indistinguishable in physiological and histological studies (data not shown), and therefore the data for them are combined as “wt/het” data. All animal work was approved by a local ethics committee for animal research.

Pancreata from E15.5 and E18.5 embryos, postnatal day P0, P7, and P21, and 12 week-old mice were fixed for 2 h at 4°C in 4% paraformaldehyde in PBS, washed with PBS, dehydrated, and embedded in paraffin. Sections (6 μm) were used for immunohistochemical analysis as previously described (9).

Primary antibodies used were as follows: rabbit anti-Mitf (1:1,000; gift from S. Saule, Institut Curie, Paris), rabbit anti-Mitf (1:1,000; Abcam), guinea pig anti-insulin (1:2,000; Millipore), mouse anti-glucagon (1:2,000; SIGMA), sheep anti-somatostatin (1:2,000; American Research Products), guinea pig anti-PP (1:2,000; Millipore), goat anti-ghrelin (1:500; Santa Cruz), rabbit anti-Glut2 (1:500; Chemicon International), rabbit anti-MafA (1:1,000; Bethyl Laboratories), rabbit anti-Nkx6.1 (1:1,000; β-Cell Biology Consortium), rabbit anti-Pax6 (1:1,000; Covance), mouse anti-Pax6 (1:100; Developmental Hybridoma Bank), goat anti-Pdx1 (1:1,000; β-Cell Biology Consortium), mouse anti-islet 1 (1:100; Developmental Hybridoma Bank), and mouse anti-Synaptophysin (1:100; Abcam). Secondary antibodies used were as follows: (1:500) diluted CY2, CY3, CY5 conjugated anti-mouse, -goat, -guinea pig, -rabbit, and -sheep (Jackson ImmunoResearch). DAPI (Invitrogen) was used for nuclear counterstaining (1:6,000).

Random-fed blood glucose samples were collected at a consistent time of the day (1200–1300 h). Fasted blood glucose levels were measured after a 6-h fasting period (800–1400 h) with a handheld glucometer (OneTouch; Lifescan). Mean difference between wild-type and Mitf^ce/ce was tested with Student t test.

Intraperitoneal glucose tolerance test (IPGTT) was performed on 12-week- and 6-month-old Mitf^ce/ce and wild-type mice after overnight fasting (12 h) using an injection of 2 g glucose/kg body wt. Blood glucose levels were measured using a handheld glucometer (OneTouch; Lifescan). Blood was collected from an incision in the distal part of the tail. Measurements were taken at 0, 5, 15, 30, 60, and 120 min after glucose administration for the IPGTT and at 0, 2, 5, and 15 min for glucose-stimulated insulin secretion studies. For the latter, blood was collected into heparin-coated tubes (Sarstedt) and the serum fractions were analyzed with a Mouse Insulin ELISA (Mercodia) according to the manufacturer's instructions.

Islets were isolated by collagenase digestion and handpicked under a stereomicroscope (20). Batches (n = 8) of three islets for each condition were kept in secretion assay buffer (SAB) (114 mmol/L NaCl, 4.7 mmol KCl, 1.2 mmol/L KH₂PO₄, 1.16 mmol/L MgSO₄, 20 mmol/L HEPES, 2.5 mmol/L CaCl₂, 25.5 mmol/L NaHCO₃, and 0.2% BSA; pH 7.2) containing 2.8 mmol/L glucose for 60 min in an incubator at 37°C. Three islets per well were transferred to a 96-well plate containing 200 μL/well SAB but with the addition of 2.8, 8.3, and 16.7 mmol/L glucose with or without 35 mmol/L KCl. After transfer of all islets, the plate was placed in an incubator at 37°C. At 60 min, a buffer sample was removed for measurement of insulin by ELISA.

Blood samples from random-fed and 12-h-fasted wild-type and Mitf mutant mice were collected into heparin-coated tubes. Samples were chilled immediately and centrifuged at 4°C. Plasma aliquots were stored at −20°C. Glucagon concentrations were determined using a glucagon radioimmunoassay kit (Millipore). Statistical analysis was conducted using Student t test.

Islet isolation has previously been described (20). RNA from E18.5 and P7 pancreas and islets of 12 week-old mice was extracted with ToTally RNA (Ambion) or RNeasy mini kit (Qiagen). RNA concentrations were measured with a NanoDrop ND-1000 spectrophotometer (NanoDrop), and RNA quality was assessed using a Bioanalyzer (Agilent). Only samples with a RNA Integrity Number >7.5 were further analyzed. Quantitative PCR was performed in two steps. First, reverse transcription was carried out with SuperScript III (Invitrogen) using 500 ng RNA. For the quantitative PCR reaction, cDNA was diluted 1:200 and all assays were performed with Fast SYBR Green Master Mix on a StepOnePlus Real-Time PCR instrument (Applied Biosystems). Primer sequences are available upon request. Formation of expected PCR products was confirmed by electrophoresis and melting curve analysis. Gene expression data were normalized to the expression of the internal control gene hypoxanthine-guanine phosphoribosyltransferase (HPRT). Experiments were repeated at least five times—each in triplicate. Raw data from real-time PCR measurements were exported from StepOne Software v2.1 and analyzed in Microsoft Excel. The data are shown as mean expression with SEM. Graphs represent the fold change in comparison with the wild-type control samples, set as 1. Mean difference was tested with Student t test.

Immunofluorescence images were collected with Zeiss Axioplan 2 imaging (Zeiss, Jena, Germany) in AxioVision Rel 4.9 software. Immunohistochemistry and quantification were performed from at least three adult wild-type and Mitf^ce/ce pancreata. β-Cell mass was assessed by quantifying the pancreatic and insulin-stained area in sections at a 720-μm interval throughout the whole pancreatic organ. The percentage of endocrine cells expressing ghrelin was assessed by quantifying Pax6 and ghrelin-producing cells at a 96 (E18.5) or 120 (P7) μm interval throughout the whole pancreas. Synaptophysin-expressing cells were measured in every 20th section of adult pancreas and normalized to the pancreatic weight. Mean differences were tested with Student t test; n ≥ 3.

β-Cell–derived β-TC6 cells and α-cell–derived α-TC6 (American Type Culture Collection) were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% calf serum. Cells were transfected with FLAG-tagged Mitf expression plasmid (True ORF cDNA clone MR227086; ORIGene) using Metafectene Pro (Biontex). Cells were harvested 48 h after transfection, fixed in 1% formaldehyde in Dulbecco’s modified Eagle’s medium without serum, and treated with lysis buffer (10 mmol/L Tris-HCl, pH 8.0; 10 mmol/L NaCl; 3 mmol/L MgCl₂; 1% NP-40; 1% SDS; and 0.5% sodium deoxycholate [DOC]) and sonicated four times for 5 min on high (30 s on and 30 s off) settings in a Bioruptor sonicator (Diagenode). Protein/DNA chromatin fragments were immunoprecipitated with mouse anti-FLAG antibody (Sigma) or mouse IgG (Santa Cruz) as previously described (21).

Nonquantitative PCR reactions were performed using purified immunoprecipitated DNA with 0.8 μmol/L primers specific for Pax6^EE (AGGCACGTCCTGGATGTTAG and CCCCAACCTCATTCTTTTCA), Pax6^PE (ACTCAGGCCTGTGGTTATGC and TCAAGAGCGAAGCTGGAAAC), Pax4 (AGGGACAATTAGCCCCAAAC and ACAGAAGCTTTCGACCCAGA), Runx2 (GTTGTTTGATTTGTTTTGAAGG and TTTTACCTAAAATGTGGTTTTTG), and PEPCK (GAGTGACACCTCACAGCTGTGG and GGCAGGCCTTTGGATCATAGCC) transcriptional control region primers. PCR products were visualized on a 1.5% agarose gel in 1× Tris-Borate-EDTA buffer. Each chromatin immunoprecipitation (ChIP) experiment was repeated at least three times using independent chromatin preparations.

The Pax6 P0 sequence (−3.8 to 0.2 kb) was PCR amplified from mouse genomic DNA and inserted upstream of a promoterless firefly luciferase plasmid (pGL2basic; Promega). Human embryonic kidney (HEK)293 cells were transfected with pPax6P0-LUC, cytomegalovirus-driven Mitf, and/or Pax6, and Renilla luciferase plasmids (phRL-CMV; Promega) using Metafectene Pro reagent (Biontex). Extracts were prepared 40–48 h later and analyzed for firefly and Renilla luciferase activity using the Dual-Luciferase Reporter Assay System (Promega). Mean difference was tested with one-way ANOVA; n = 3.

Immunohistochemical analysis was performed to determine which cell types produce Mitf during pancreas development and in the adult. Mitf expression was not detected at E15.5 (data not shown) but was initially observed at E18.5 (Supplementary Fig. 1). At P0, Mitf was found in nuclei of endocrine, exocrine, and ductal cells (Fig. 1A–D and Supplementary Fig. 1). Postnatally, Mitf expression becomes restricted to islet endocrine cells (Supplementary Fig. 1). Analysis of adult wild-type pancreata showed that Mitf is expressed in all hormone-producing cells (Fig. 1E–H). Mitf expression is restricted to nuclei of insulin-producing cells, whereas it is also detected in the cytoplasm of ∼20% glucagon cells (Supplementary Fig. 2).

Blood glucose measurements showed that Mitf^ce/ce mice had significantly lower blood glucose levels than their wild-type littermates whether they were random-fed or fasted (Fig. 2A). In addition, Mitf^ce/ce animals had improved blood glucose clearance after glucose injection at 12 weeks and 6 months. Blood glucose levels stayed below 20 mmol/L even at 15–30 min after the glucose injection (Fig. 2C and D) and returned faster to homeostatic conditions. Glucose-stimulated serum insulin levels were measured to determine whether this improved glucose clearance is a direct result of altered β-cell function. Interestingly, Mitf mutant animals have significantly higher serum insulin levels in fasting conditions than wild-type animals (0 min [Fig. 2E]), which suggests that Mitf is required for proper β-cell function. In contrast, plasma glucagon levels were not significantly changed in Mitf^ce/ce animals compared with wild-type littermates in both fasted and random fed conditions (Fig. 2B). Stimulation of Mitf^ce/ce and wild-type isolated islets with 16.7 mmol/L glucose and 35 mmol/L KCl resulted in a 60% increased insulin secretory response (Fig. 2F). Our results demonstrate that Mitf mutant mice have enhanced glucose tolerance and are protected from high blood glucose levels by elevated insulin secretion.

Immunohistochemical analysis of wild-type and Mitf^ce/ce pancreatic sections showed no obvious changes in the appearance of pancreatic islets (Fig. 3A). For determination of whether the elevated serum insulin levels observed in Mitf^ce/ce mice result from an increase in β-cell mass, quantitative immunohistochemical analysis was performed. The percentage of insulin-labeled wild-type and Mitf^ce/ce cell area and total islet cell area were comparable (Fig. 3B and Supplementary Fig. 4C and D) suggesting that Mitf is not required for β-cell specification or regulation of β-cell mass. Insulin and PP mRNA levels were significantly increased in Mitf^ce/ce islets (Fig. 3C), while ghrelin transcription was decreased to only 20% of wild-type (Fig. 3C). Immunohistochemical quantification of ghrelin-producing cells revealed no differences in the percentage of ghrelin cells within the endocrine Pax6⁺ cell population at E18.5 and P7 (Supplementary Fig. 3A–E), while ghrelin mRNA levels were already reduced by P7 (Supplementary Fig. 3F). In accordance with previous reports (22) ghrelin cells were almost absent in adult wild-type and Mitf^ce/ce islets, which precluded a quantitative analysis at this time point. Quantitative immunohistochemical analysis showed no significant increase in PP cell number (data not shown) in Mitf^ce/ce islets. These findings suggest that insulin, ghrelin, and PP production is altered in individual hormone-expressing cells. Mitf^ce/ce and wild-type β-cells were analyzed using electron microscopy to determine whether the increase in insulin mRNA levels results in enhanced production of insulin secretory granules. Mitf^ce/ce β-cells have the same number of insulin granules per cell, and granule morphology is comparable with the wild-type (Supplementary Table 1 and Supplementary Fig. 4A and B). Our findings illustrate that Mitf is critical for proper hormone expression but not essential for endocrine cell specification and maintenance.

Gene expression analysis and immunohistochemistry were performed to further characterize Mitf^ce/ce islets. First, we analyzed the expression level and distribution of transcription factors involved in β-cell development and/or function. The cellular distribution of known β-cell factors Pax6, Pdx1, MafA, and Nkx6.1 was unchanged (Fig. 4A–H). However, a twofold increase in Pax6 and Pax4 mRNA levels was observed in Mitf^ce/ce islet samples (Figs. 4K and 5). Interestingly, β-Cell–specific expression of Glut2, a known Pax6 target gene (3) was increased twofold in mutant islets (Fig. 4O), while the transcription of other glucose-sensing, metabolizing, and insulin secretory genes was unaffected (Fig. 5). These results suggest that Mitf is controlling β-cell function by directly regulating genes required for β-cell function and differentiation.

Previous studies have established that Mitf acts as a transcriptional activator and repressor in the developing eye (13–15). We hypothesize that Mitf directly controls Pax6 transcription, since Pax6 mRNA levels were upregulated in Mitf^ce/ce mice (Fig. 4K) and Mitf overexpression prevents basic fibroblast growth factor (bFGF)-induced Pax6 upregulation in cultured retinal pigmented epithelial cells (14,23). The mouse Pax6 gene contains several regulatory elements that confer tissue-specific expression in the pancreas, eye, and brain (24). Among those, the P0 proximal regulatory element (PE) is required for embryonic pancreas-specific Pax6 expression at physiological levels (25), while ectodermal expression is controlled by the EE element (EE). ChIP analysis was performed in mouse α-TC6 and β-TC6 cells transfected with FLAG-tagged Mitf to determine whether Mitf occupied the Pax6 EE and PE sequences (Pax6^EE and Pax6^PE). An enrichment of both Pax6^EE and Pax6^PE was found using primer pairs that span the Pax6^PE and Pax6^EE regions compared with control sequences only after α-FLAG immunoprecipitation (Fig. 6B and C). A similar analysis demonstrated that flag-tagged Mitf also immunoprecipitates Pax4 regulatory elements in β-TC6 cells (Fig. 6B). For further assessment of functional regulation of the Pax6 P0 promoter, Pax6 and Mitf were cotransfected with a Pax6P0-driven reporter construct in HEK293 cells (Fig. 6A and D). Pax6P0 firefly luciferase (pPax6P0-LUC) activity was enhanced upon cotransfection with Pax6, while cotransfection with Mitf resulted in transcriptional repression. Cotransfection of Mitf and Pax6 resulted in loss of Mitf-mediated repression, suggesting that interactions between these factors are critical for the transcriptional control of the Pax6P0 regulatory region. The ChIP and luciferase reporter data are complimentary and demonstrate that Mitf directly binds to and regulates Pax6 regulatory elements, suggesting that increased Pax6 mRNA levels in Mitf^ce/ce are a direct effect of the loss of Mitf.

Previous studies have established Mitf as a master regulator of pigment cell differentiation (rev. in 26). Our previous results have shown that Mitf expression is reduced in MafA and MafB mutant embryonic pancreata (12) suggesting that Mitf may be involved in β-cell differentiation and function. In this study, we demonstrate that Mitf is initially expressed in the pancreatic epithelium at E18.5 and that its expression becomes restricted to adult islet cells. Mitf inactivation results in altered β-cell function, due to an enhanced insulin secretory response, without changes in β-cell mass. In contrast, the transcription of genes critical for the regulation of blood glucose levels (insulin and Glut2) and β-cell formation and function (Pax4 and Pax6) was significantly upregulated in Mitf^ce/ce islets, suggesting that Mitf is critical for adult β-cell function.

During development, Mitf was initially detected in all pancreatic epithelial cells from E18.5 onward (Supplementary Fig. 1). Interestingly, Mitf expression increases and becomes progressively restricted to islet cells postnatally and is not detected in ductal and exocrine cells after the first weeks of birth. Our findings are supported by a gene-profiling study of Ngn3⁺ endocrine progenitor cells and their descendants, which showed that Mitf transcription increases postnatally, with the highest expression observed in adult islet cells (27). The late onset of Mitf expression and localization in endocrine cells coincides with profound morphological and physiological changes in the pancreas: endocrine cells cluster into mature islets, with β-cells in the core and other hormone-producing cells at the periphery. In addition, and perhaps most importantly, the glucose responsiveness of β-cells is established at this time (8). During specification of zebrafish melanocytes, Mitf is specifically required for late steps of differentiation (28), which also supports a role for Mitf in β-cell maturation or function.

To determine Mitf’s function in the developing and adult pancreas, we analyzed a Mitf mutant mouse model. Mitf^ce/ce pancreata were histologically indistinguishable from wild-type littermates, β-cell area was comparable, and no changes in adult islet cell composition were observed. However, as evident from our in vitro studies on isolated islets, the insulin response to high glucose and KCl concentrations was significantly greater in Mitf^ce/ce animals than in wild-type, whereas insulin response to a stimulation with lower glucose concentrations was not significantly affected (Fig. 2F). This shows that the total β-cell secretory capacity is increased in mutant mice. Interestingly, Mitf^ce/ce animals also had fasting hyperinsulinemia in the presence of slight hypoglycemia (Fig. 2A and E), again showing increased insulin secretion. Plasma glucagon levels were unchanged in the fasted and fed condition (Fig. 2B) suggesting that α-cell function is not affected in Mitf^ce/ce animals. After intraperitoneal glucose challenge, glucose tolerance was improved (Fig. 2C and D), which also reflects increased glucose-stimulated insulin secretion. However, when glucose-stimulated insulin secretion was determined the only significant difference was observed at 0 min (Fig. 2E). Mitf function in peripheral tissues may also influence glucose responsiveness and partially account for the altered glucose clearance in Mitf^ce/ce. However, thus far, no data supporting a role of Mitf in muscle, liver, and adipose glucose metabolism have been reported, and transcription in these tissues is relatively low (gene expression atlas, http://www.ebi.ac.uk/arrayexpress/). Collectively, these observations suggest that loss of Mitf specifically alters β-cell function by increasing glucose responsiveness and insulin secretion.

Gene expression analysis showed increased transcription of key β-cell genes such as Pax4, Pax6, insulin, and Glut2, which could contribute to the increased insulin secretory response. Thus elevated insulin mRNA levels may result in increased total insulin content and in a larger pool of insulin granules that are ready to be released, which could influence insulin secretion. However, Mitf^ce/ce β-cells have no change in the total number of insulin granules (Supplementary Fig. 4 and Supplementary Table 1), and the percentage of ready releasable granules stays the same suggesting that mutant β-cell function is most likely improved owing to changes in glucose sensing. Thus, elevated Glut2 expression may positively influence this process, since it is the only glucose transporter expressed in β-cells that mediates glucose-stimulated insulin secretion (29,30).

Upregulation of Pax6 transcription may be the key to the improved β-cell function observed in Mitf^ce/ce, since it has been shown that this transcription factor regulates the expression of several β-cell genes like insulin, Glut2, PC1/3, and glucokinase (31). Interestingly, we observed enhanced transcriptional activity of only some Pax6 target genes like insulin and Glut2 but not others, which suggests that a twofold increase in Pax6 expression is not sufficient to augment the transcription of all Pax6 target genes in this context.

Loss-of-function studies have established a critical role for Pax4 in β-cell development (32), while ectopic Pax4 expression in adult β-cells is associated with increased proliferation and protection against apoptosis (33). Mitf binds to Pax4 regulatory elements, and Mitf^ce/ce islets have twofold higher Pax4 mRNA levels than wild-type islets but no difference in β-cell mass and thus no difference in β-cell proliferation and specification. This may be due to the moderate upregulation of Pax4 transcription in Mitf^ce/ce islets compared with Pax4 overexpression models used in recent work, which has studied the effects of 15-fold increased Pax4 (33). However, the moderately increased Pax4 expression in Mitf^ce/ce islets may still positively influence β-cell function and contribute to improved glucose clearance.

Previous studies have demonstrated that Mitf acts both as transcriptional activator and repressor (14,15,34): it activates genes required for cell pigmentation (13,14), while ectopic Mitf expression prevents Pax6 transcription in the developing eye (14,23) and Pax6 and Mitf cooperatively repress common target genes in the retinal pigmented epithelium (15). Thus far, Mitf repressor function has only been described in the context of coexpression with Pax6 but not with other transcriptional repressors like groucho (15,35). Here, we demonstrate that Mitf binds to and represses the Pax6 P0 regulatory region, which is required for high Pax6 expression during pancreas development (25,36). This finding suggests that Mitf is partially responsible for inactivation of the embryonic pancreatic Pax6 P0 promoter.

In vitro transfection and gel shift experiments have demonstrated that direct protein-protein interaction between Pax6 and Mitf results in an inactivation of both proteins (37). The Mitf^ce protein lacks the leucine zipper domain, which results in loss of DNA binding and reduced interaction with cofactors (19). The leucine zipper domain is also required for interactions with Pax6 (37); thus, Mitf^ce protein not only has lost its transcriptional activity but also fails to interact and repress Pax6 function in mutant β-cells. These in vitro studies support our observation that the expression of the newly identified direct Mitf target gene Pax6 is elevated in Mitf^ce/ce animals and that enhanced Pax6 levels and transcriptional activity may result in increased Glut2, Pax4, and insulin transcription.

Our data identify Mitf as a novel transcriptional repressor in adult β-cells and suggest that regulation of β-cell function depends on interplay between the Mitf and Pax6 transcription factors.

This work was supported by the Swedish Research Council, the JDRF, Diabetesfonden, Lindhés, and Jeanssons Stiftelse.

No potential conflicts of interest relevant to this article were reported.

M.A.M. designed research, researched data, and wrote the manuscript. M.W., E.G., and J.K.C. researched data. J.K.J. reviewed the manuscript and contributed to discussion. H.B. researched data. M.F. researched data, reviewed the manuscript, and contributed to discussion. U.A.N. reviewed the manuscript and contributed to discussion. I.A. designed research and wrote the manuscript. M.A.M. and I.A. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank S. Saule (Institut Curie), C. Wright (Department of Cell and Developmental Biology, Vanderbilt University Medical Center), and the Beta Cell Biology Consortium (BCBC) for providing antibodies; B. Ahrén (Department of Clinical Sciences, Lund University), H. Semb (Stem Cell Center, Lund University, and The Danish Stem Cell Center, Copenhagen University), and R. Stein (Department of Molecular Physiology and Biophysics, Vanderbilt University School of Medicine) for comments on the manuscript; M. Magnusson for mouse strain maintenance; B.-M. Nilsson (Clinical Research Center, Lund University) for plasma glucagon measurements; and R. Wallén (Lund University) for transmission electron microscopy analysis.