Islet-1 (Isl-1) is essential for the survival and ensuing differentiation of pancreatic endocrine progenitors. Isl-1 remains expressed in all adult pancreatic endocrine lineages; however, its specific function in the postnatal pancreas is unclear. Here we determine whether Isl-1 plays a distinct role in the postnatal β-cell by performing physiological and morphometric analyses of a tamoxifen-inducible, β-cell–specific Isl-1 loss-of-function mouse: Isl-1L/L; Pdx1-CreERTm. Ablating Isl-1 in postnatal β-cells reduced glucose tolerance without significantly reducing β-cell mass or increasing β-cell apoptosis. Rather, islets from Isl-1L/L; Pdx1-CreERTm mice showed impaired insulin secretion. To identify direct targets of Isl-1, we integrated high-throughput gene expression and Isl-1 chromatin occupancy using islets from Isl-1L/L; Pdx1-CreERTm mice and βTC3 insulinoma cells, respectively. Ablating Isl-1 significantly affected the β-cell transcriptome, including known targets Insulin and MafA as well as novel targets Pdx1 and Slc2a2. Using chromatin immunoprecipitation sequencing and luciferase reporter assays, we found that Isl-1 directly occupies functional regulatory elements of Pdx1 and Slc2a2. Thus Isl-1 is essential for postnatal β-cell function, directly regulates Pdx1 and Slc2a2, and has a mature β-cell cistrome distinct from that of pancreatic endocrine progenitors.
Compromised pancreatic β-cell function is a critical factor underlying the onset of diabetes (1). β-Cell functional capacity is regulated by extrinsic signaling pathways and an intrinsic network of transcription factors (2,3). It is established that β-cell–specific transcription factors like MafA and Pdx1 are essential components of this intrinsic transcriptional network (4–7). It is less clear how pan-endocrine transcription factors like Islet-1 (Isl-1) affect postnatal β-cell function. These factors are expressed in all postnatal pancreatic endocrine cell types, suggesting roles in general endocrine function, cell type–specific physiology, or both. The majority of in vivo studies investigating pan-endocrine transcription factors have conditionally ablated their respective genes prior to maturation of the pancreatic endocrine compartment (8–11). As a result, it remains unclear whether these factors have unique functional roles in the endocrine cell types of the postnatal pancreas. To this point, a recent study demonstrated that the pan-endocrine factor NeuroD1 is necessary for maintaining functional maturity of mouse β-cells (12). Given these findings, Isl-1 and other pan-endocrine transcription factors may have functional roles in the postnatal β-cell distinct from their well-established developmental roles.
Isl-1 is a Lin11, Isl-1, and Mec-3 homeodomain (LIM-HD) factor that is essential for the genesis of the dorsal pancreatic bud at E9.5, the survival of Pax6+ endocrine progenitors at E13.5, and the ensuing maturation of α-, β-, δ-, and pancreatic polypeptide (PP) cells (11,13). Isl-1 was identified as an Insulin enhancer binding protein (14) and was subsequently shown to directly interact with NeuroD1 to promote Insulin expression (15). While Isl-1 expression is conserved in a variety of adult neuroendocrine cell types (16), many of the identified Isl-1 target genes are associated with pancreatic endocrine function, including IAPP, Sst, Gcg, and Kcnj11/Kir6.2 (11,17–21). In adult mouse β-cells, Isl-1 was identified as a key downstream target of leptin-induced Janus kinase signal transducer and activator of transcription 3 signaling (22). Recently, a transgenic mouse with islet-specific overexpression of Isl-1 displayed improved β-cell function (23). Interest in the mechanisms whereby Isl-1 regulates postnatal β-cell function is further raised by type 2 diabetes linkage and genome-wide association studies that identified genetic markers in the chromosomal region encompassing the ISL-1 locus (24–27).
Despite genetic links to type 2 diabetes in humans and evidence that Isl-1 regulates key genes associated with pancreatic function, the in vivo requirement for Isl-1 in postnatal β-cell function has not been thoroughly investigated. Here we derived an inducible, β-cell–specific, Isl-1 loss-of-function mouse. By combining microarray analysis of Isl-1–deficient islets with Isl-1 chromatin immunoprecipitation (ChIP) sequencing (ChIP-Seq) of βTC3 mouse insulinoma cells, we constructed the transcriptional network controlled by Isl-1 and identified novel gene targets directly regulated by Isl-1 in postnatal β-cells.
Research Design and Methods
The Isl-1L/L and Pdx1-CreERTm mouse lines have been previously described (28,29). Mice were maintained on a mixed C57BL/6, CD1, and Sv129 background. The morning after birth was considered P0.5. Analysis was restricted to female mice. Tamoxifen (Tm; Sigma-Aldrich, T5648) at 50 μg/g mouse bodyweight was administered to 8-week-old mice via three intraperitoneal injections at 24-h intervals. Tm was dissolved in 90% sunflower seed oil (vehicle [Veh]; volume for volume), 10% ethanol (volume for volume). Unless otherwise stated, analysis of Tm-treated animals was performed 2 days after the third injection. The Children’s Hospital of Philadelphia Institutional Use and Care Committees approved all animal studies.
Immunohistochemical and Immunofluorescence Analyses
Pancreata were dissected, fixed in 4% paraformaldehyde (pH 7.0) for 6 h at 25°C, and embedded in paraffin or optimal-cutting-temperature compound (Tissue-Tek, 4583). Sections were blocked using CAS-Block (Invitrogen, 008120), and primary antibodies were applied overnight at 4°C. Primary and secondary antisera information is provided in Supplementary Tables 1 and 2, respectively. For immunofluorescence, Vectashield mounting medium with DAPI (Vector, H-1200) was used to counterstain nuclei, and fluorescein isothiocyanate tyramide signal amplification (PerkinElmer, NEL741001KT) was used for Isl-1 detection. For immunohistochemistry, signal was detected using Vectastain Elite ABC Kit (standard; Vector, PK-6100) and DAB Peroxidase Substrate Kit (Vector, SK-4100). Staining was visualized using a Leica DM6000 B microscope, and images were captured using the Leica LAS AF software and Leica DFC300 FX digital camera.
To quantify staining, slides were digitally scanned using an Aperio ScanScope CS2 or MetaMorph microscopy automation software and analyzed using ImageScope software. Isl-1 ablation efficiency for a hormone+ population was calculated as the percentage of Isl-1+, hormone+ cells per total hormone+ cells using Indica Laboratories image analysis algorithms. β-Cell mass was calculated by averaging the percentage of insulin-stained tissue area over three sections that were taken at 100 μm levels. The fraction of positive area was then multiplied by the wet mass of the dissected pancreas measured at tissue harvest. Terminal deoxynucleotidyl TUNEL was performed as described (30) on three sections taken at 40 μm from pancreata that were harvested 14 days after the first Tm injection. These sections were then costained for insulin. TUNEL+, insulin+ cells were counted manually and normalized to the number of total β-cells. The number of total β-cells was determined by counting the Nkx6.1+ nuclei on an adjacent section.
RNA Isolation, cDNA Synthesis, Quantitative PCR, and Microarray
Total RNA was extracted from pancreatic islets isolated by the standard collagenase P (Roche, 11 213 873 001) protocol (31) or whole pancreata. Total RNA preparation and cDNA synthesis were performed as described (23). Quantitative PCR (qPCR) reactions were performed using SYBR Green JumpStart Taq ReadyMix (Sigma-Aldrich, S4438) and a Stratagene Mx3005P qPCR system. Fold enrichment of mRNA message was calculated by normalizing to a reference gene (see Supplementary Table 3 for qPCR primers). Control and mutant-isolated islet total RNA extractions were matched for pancreatic endocrine purity as described (32). Microarray analysis was performed by the University of Pennsylvania’s Diabetes Research Center Functional Genomics Core. RNA was labeled with the Agilent Low Input Kit and hybridized, using a dye-switch design, to the Agilent 4 × 44K Whole Mouse Genome Microarray. Arrays were hybridized overnight and scanned using the Agilent Microarray Scanner. Data were normalized using normalizeBetweenArrays from the Limma package followed by SAMR to identify differentially expressed genes.
Western Blot Analysis
Western blots were performed as described (33) using isolated islet whole-cell lysates. Pdx1 (Santa Cruz, Pdx1 sc-14664, 1:200) and α-tubulin (Sigma-Aldrich, T5168, 1:3000) antisera were used.
Glucose and Hormone Assays
Plasma glucose and insulin were measured as described (23). Random-fed plasma glucose was assessed between 10:00 and 11:00 a.m. Glucose tolerance and glucose-stimulated plasma insulin levels were assessed as described (11). Isolated islet glucose-stimulated insulin secretion (GSIS) was assessed via static incubations using 20–50 islets of similar size as described (23). Values for islet insulin content and secretion were normalized to the total number of islets per incubation. Values for relative islet insulin secretion reflect islet insulin secretion normalized to islet insulin content. Mouse insulin concentration was determined by ELISA (Mercodia, 10–1247). Pancreatic insulin content was measured as described (11).
ChIP and ChIP-Seq
βTC3 insulinoma cells were grown in monolayer (∼4 × 106 cells), or mouse islets were isolated from CD1 mice. ChIP assays were performed as described (33). Normal mouse IgG (Santa Cruz, SC2025) and anti–Isl-1 (Hybridoma Bank UI, 39.4D5-C) were used to immunoprecipitate sheared chromatin. Enrichment was determined using qPCR. Values are presented as fold enrichment over normal mouse IgG. To account for background, values were normalized to enrichment at the Pepck locus, which is not bound by Isl-1. For ChIP primers, see Supplementary Table 3. Whole-genome ChIP-Seq analysis using anti–Isl-1 (Hybridoma Bank UI, 39.4D5-C) to immunoprecipitate sheared βTC3 chromatin was performed as described in conjunction with University of Pennsylvania’s Diabetes Research Center Functional Genomics Core (33).
Electrophoretic Mobility Shift Assays
Electrophoretic mobility shift assays (EMSAs) were performed as described using a pCS2-Isl-1-Myc plasmid (gift from Dr. Pfaff) as a template for in vitro translation (11). The radiolabeled probe was designed as MafA-Region 3 (11). Competition experiments were performed using 100-fold molar excess of unlabeled dsDNA oligonucleotides spanning the homeodomain binding elements (HBEs) in Pdx1 enhancer areas I, II, and IV. For dsDNA oligonucleotide sequences see Supplementary Table 4. Supershift analysis was performed as described using a cocktail of Isl-1 antisera (Hybridoma Bank UI, 39.3F7, 39.4D5, 40.2D6, and 40.3A4) or anti-Myc (Santa Cruz, sc-40) (11).
Luciferase Vector Construction and Reporter Assays
Sequences of interest were cloned into the pGL4.27 luciferase vector (Promega). To create HBE mutations, site-directed mutants were generated as described (33). Luciferase reporter assays were performed in βTC3 or HeLa as described (33). Exogenous Isl-1 protein was overexpressed using the pCS2-Isl-1-Myc vector. Transient transfection of all vectors was accomplished using Lipofectamine 2000 (Invitrogen). All pCR4-TOPO and pGL4.27 vectors containing wild-type or HBE-mutated Pdx1 area I, II, and IV and Slc2a2 Re1 and Re2 are available upon request.
Isl-1L/L; Pdx1-CreERTm Mice Exhibit a Baseline Level of Postnatal Isl-1 Ablation Prior to Administering Tm
To determine the functional requirement for Isl-1 in the postnatal β-cell, we derived an inducible, β-cell–specific, Isl-1 loss-of-function mouse model (Isl-1L/L; Pdx1-CreERTm). Because of a recent report demonstrating minimal Tm-independent recombination of the Rosa26 locus in Pdx1-CreERTm transgenic mice (34), we assessed baseline Isl-1 protein and transcript levels before administering Tm to 8-week-old female Isl-1L/L; Pdx1-CreERTm mice, hereon notated as Isl-1L/L; Pdx1-CreERTm(No Tm). Isl-1 immunohistochemistry in 8-week-old Isl-1L/L; Pdx1-CreERTm(No Tm) animals revealed pancreatic islets with multiple Isl-1− nuclei (Fig. 1A and B). When quantified, Isl-1L/L; Pdx1-CreERTm(No Tm) mice had a 20% decrease in Isl-1+ nuclei compared with controls (Fig. 1C). Using qPCR, we determined that the relative Isl-1 mRNA transcript was reduced by ∼38% in islets isolated from Isl-1L/L; Pdx1-CreERTm(No Tm) animals; however, this difference was not statistically significant (Fig. 1D).
Despite a reduction in Isl-1+ β-cells, gross pancreatic morphology, islet distribution, and insulin staining were indistinguishable between 8-week-old Isl-1L/L; Pdx1-CreERTm(No Tm) animals and controls (Fig. 1E and F and data not shown). Importantly, β-cell mass in 8-week-old Isl-1L/L; Pdx1-CreERTm(No Tm) pancreata, as reflected by insulin+ staining, was equivalent to that of age-matched controls (Fig. 1G). Despite maintaining normal β-cell mass, 8-week-old Isl-1L/L; Pdx1-CreERTm(No Tm) animals displayed reduced first-phase insulin secretion in response to acute glucose challenge (Fig. 1H). In 4-week-old Isl-1L/L; Pdx1-CreERTm(No Tm) animals, we also observed a moderate impairment of in vivo GSIS that was associated with mild glucose intolerance (Supplementary Fig. 1A and B). Intriguingly, 8-week-old Tm-treated Isl-1L/+; Pdx1-CreERTm mice did not display glucose tolerance defects (Supplementary Fig. 1C), suggesting that Isl-1 is haplosufficient in the β-cell, and any pathophysiological defects reflect complete Isl-1 ablation in the β-cell. Together, these control experiments demonstrate that the Pdx1-CreERTm strain directs limited, Tm-independent recombination of Isl-1 as early as 4 weeks of age.
Isl-1 is required for the differentiation and maturation of pancreatic endocrine precursors (11). Conditionally ablating Isl-1 in the pancreatic epithelium at E13.5 results in mice that are born without a mature endocrine compartment, are hyperglycemic by P7, and die between 3 and 8 weeks of age (11). Significant Tm-independent recombination of the Isl-1 locus during embryogenesis could confound analysis of postnatal animals. However, insulin+, Isl-1− cells were rarely detected by coimmunofluorescence at P0.5 in pancreata from Isl-1L/L(No Tm) or Isl-1L/L; Pdx1-CreERTm(No Tm) animals (Supplementary Fig. 2A and B). We also determined that the level of pancreatic Isl-1 transcript at P0.5 was indistinguishable between Isl-1L/L(No Tm) and Isl-1L/L; Pdx1-CreERTm(No Tm) animals (Supplementary Fig. 2C). At P5, Isl-1L/L; Pdx1-CreERTm(No Tm) mice displayed normal random-fed plasma glucose, their endocrine compartment was present, and gross pancreatic morphology and intraislet hormone distribution appeared normal (Supplementary Fig. 2D–F). Furthermore, Isl-1L/L; Pdx1-CreERTm(No Tm) animals showed no signs of morbidity (data not shown). These observations contrast with previously characterized mouse models that ablate Isl-1 in the developing pancreas (11,13). Altogether, Isl-1L/L; Pdx1-CreERTm(No Tm) animals have no evidence of significant Tm-independent recombination during pancreas development.
Ablating Isl-1 in β-Cells Impairs Glucose Tolerance and Insulin Secretion Without Impacting β-Cell Mass
To maximally ablate Isl-1 in adult β-cells, we administered Tm to 8-week-old Isl-1L/L; Pdx1-CreERTm and control Isl-1L/L animals, respectively notated Isl-1L/L; Pdx1-CreERTmIP(Tm) and Isl-1L/L IP(Tm) (Fig. 2A). The 5-day pulse-chase recombined most remaining β-cell Isl-1 alleles. Isl-1 mRNA expression in islets isolated from Isl-1L/L; Pdx1-CreERTmIP(Tm) animals was consistently reduced to ∼24% of Isl-1L/L IP(Tm) levels (Fig. 2B). A similar result was observed with Isl-1 immunostaining; almost all insulin+ cells lacked Isl-1 expression (Fig. 2C–E). The majority of the remaining Isl-1+ nuclei were located at the periphery of the islet where non–β-cell endocrine cell types (i.e., α-, δ-, ε-, and PP cells) are typically located in mouse islets (Fig. 2C–D). Accordingly, Isl-1 was not ablated in α-cells; however, Isl-1 was ablated in a significant percentage of δ-cells (Fig. 2C–E and Supplementary Fig. 2G and H), paralleling Pdx1 expression in adult δ-cells (35).
When compared with Isl-1L/LIP(Tm) animals, Isl-1L/L; Pdx1-CreERTmIP(Tm) mice had increased random-fed plasma glucose levels but maintained equivalent random-fed insulin levels (Fig. 2F and G). Following fasting, Isl-1L/L; Pdx1-CreERTmIP(Tm) mice displayed a robust glucose intolerance phenotype and impaired GSIS response (Fig. 2H and I). In agreement with our findings in the postnatal Isl-1L/L; Pdx1-CreERTm(No Tm) animals (Fig. 1 and Supplementary Fig. 1), the Veh-treated Isl-1L/L; Pdx1-CreERTmIP(Veh) animals also displayed a moderate but significant glucose tolerance phenotype (Fig. 2H). Within only 5 days, maximally ablating Isl-1 noticeably exacerbated the Isl-1L/L; Pdx1-CreERTmIP(Veh) glucose intolerance and GSIS phenotypes. Unlike 8-week-old Isl-1L/L; Pdx1-CreERTm(No Tm) animals (Fig. 1H), both the first and the second phase of the GSIS response were significantly reduced in Isl-1L/L; Pdx1-CreERTm IP(Tm) mice (Fig. 2I).
Isl-1 has been implicated as a survival factor in developing cell populations, including pancreatic endocrine progenitors (11,36,37). A substantial and rapid reduction in β-cell mass could account for the physiological defects observed in the Isl-1L/L; Pdx1-CreERTm IP(Tm) animals; however, there was no difference in the number of insulin+, TUNEL+ cells between Isl-1L/L; Pdx1-CreERTmIP(Tm) and Isl-1L/LIP(Tm) animals (Fig. 3A–C), and β-cell mass was not significantly reduced (Fig. 3D). Taken together, these analyses demonstrate that the glucose homeostasis defects in Isl-1L/L; Pdx1-CreERTmIP(Tm) mice are not due to a significant increase in β-cell apoptosis or a substantial reduction in β-cell mass.
We observed a 25% reduction in total pancreatic insulin content in Isl-1L/L; Pdx1-CreERTmIP(Tm) mice (Fig. 3E). Since Isl-1 has been demonstrated to regulate Insulin transcription (14,15), we quantified the level of total Insulin mRNA in isolated islets and observed a 75% reduction in Isl-1L/L; Pdx1-CreERTmIP(Tm) animals (Fig. 3F). Reduced pancreatic insulin content and islet Insulin transcripts without increased apoptosis prompted us to investigate the functional capacity of islets. To directly evaluate islet function, we isolated islets from Isl-1L/L; Pdx1-CreERTmIP(Tm) and Isl-1L/LIP(Tm) animals and performed static incubation assays with basal and stimulatory glucose concentrations (2.5 and 16.0 mmol/L). In response to 16.0 mmol/L glucose, islets from Isl-1L/L; Pdx1-CreERTmIP(Tm) animals secreted less insulin than islets isolated from Isl-1L/LIP(Tm) animals (Fig. 3G). However, the insulin content per islet was also significantly depleted in Isl-1L/L; Pdx1-CreERTmIP(Tm) animals (Fig. 3H). When we normalized islet secretion to islet content, the relative insulin secretion rate of Isl-1L/LIP(Tm) and Isl-1L/L; Pdx1-CreERTmIP(Tm) islets was similar at both 2.5 and 16.0 mmol/L glucose (Fig. 3I). Overall, our data demonstrate that ablating Isl-1 in the adult β-cell impairs glucose homeostasis and compromises β-cell insulin secretion primarily as a result of reduced insulin synthesis.
The Isl-1 Cistrome of the Mature β-Cell Is Distinct From That of the Developing Pancreatic Epithelium
To assess the gene expression changes occurring in Isl-1–deficient β-cells, a microarray was performed using RNA isolated from Isl-1L/L; Pdx1-CreERTmIP(Tm) and Isl-1L/LIP(Tm) islets. This analysis yielded 714 genes whose expression was significantly altered (Fig. 4A, Supplementary Fig. 3A, and Supplementary Table 5). We used Ingenuity Systems software to perform gene ontology (GO) analysis of this data set. Not surprisingly, genes involved in “glucose tolerance” and “quantity of insulin in the blood” were significantly enriched among affected genes (Fig. 4B). Interestingly, GO categories associated with aspects of neuroendocrine function were also distinguished through this analysis, including genes regulating hormone concentration, intracellular molecular transport, and secretion of molecules (Fig. 4B). To determine if any of these differentially expressed genes were direct targets of Isl-1 in the mature β-cell, we performed Isl-1 ChIP-Seq using chromatin extracted from mouse βTC3 insulinoma cells (βTC3 cells) (Supplementary Table 6). Meta-analysis was performed using the microarray and ChIP-Seq data sets to determine putative targets of Isl-1 transcriptional regulation (Fig. 4A and Supplementary Fig. 3A). From this analysis, we identified MafA, a known regulatory target of Isl-1 (11), as well as Slc2a2, the gene encoding the Glut2 glucose transporter essential for rodent β-cell GSIS. We confirmed downregulation of MafA and Slc2a2 at both the transcript and the protein level in islets isolated from Isl-1L/L; Pdx1-CreERTmIP(Tm) animals (Fig. 4C–G).
Gene network analysis of our microarray data set utilizing Ingenuity Systems software yielded a de novo network containing factors essential for β-cell function (Fig. 4H). This gene network included both MafA and Slc2a2. Pdx1 appeared in this network and was also identified in the ChIP-Seq. Intriguingly, Pdx1 was not differentially regulated when Isl-1 was conditionally ablated from the mouse pancreatic epithelium at E13.5 (11). In Isl-1L/L; Pdx1-CreERTmIP(Tm) animals, however, Pdx1 was by this time significantly downregulated at both the transcript and protein levels (Fig. 5A and B) to a degree commensurate with the pathophysiological reductions in Pdx1 protein previously described in heterozygous Pdx1 loss-of-function mutants (38). The cis-regulatory regions (areas I, II, III, and IV) for Pdx1 have been well characterized (Fig. 5C) (39,40). Statistical analysis of the Isl-1 βTC3 ChIP-Seq identified three peaks that corresponded to Pdx1 areas I, II, and IV (Fig. 5C). To confirm the peak-calling analysis, we performed Isl-1 ChIP followed by qPCR using chromatin extracted from βTC3 cells (Fig. 5D) and from CD1 mouse islets (Fig. 5E).
To determine if Isl-1 binds to putative HBEs (i.e., TAAT/ATTA-containing regions) within Pdx1 areas I, II, and IV, we performed EMSAs using Myc-tagged Isl-1 incubated with a 32P-radiolabeled MafA-Region-3 (Reg3) probe (Fig. 5F). Competition assays were performed using unlabeled oligonucleotide probes representing the putative Isl-1 sites within Pdx1 areas I, II, and IV. At least one competitor from each tested Pdx1 enhancer successfully reduced Isl-1 binding to MafA-Reg3. To determine if the HBEs within the bound competitors were required for Pdx1 expression, we performed luciferase reporter assays in βTC3 cells. Vectors containing the wild-type sequences of Pdx1 areas I, II, or IV elicited significant signal above the empty vector (Fig. 5G). From the oligonucleotides that successfully competed with the MafA-Reg3 probe, we selected one oligomer from each Pdx1 enhancer element: area I-4, area II-3, and area IV-4. While we saw no change in signal when mutating HBE area I-4, we did see a significant decrease in signal when either HBE area II-3 or area IV-4 was mutated (Fig. 5G). Lastly, we performed a luciferase reporter assay using the Pdx1-area II vector in HeLa cells. Exogenously overexpressed Isl-1 amplified wild-type Pdx1-area II vector reporter activity (Fig. 5H). Taken together, these experiments suggest that Isl-1 directly regulates the adult β-cell expression of Pdx1 through at least areas II and IV.
Isl-1 Directly Regulates Slc2a2A Through the Downstream Re2 Enhancer Element
Although previous work using βTC3 cells demonstrated that Isl-1 was enriched at two putative Slc2a2 cis-regulatory elements (Re1 and Re2) (Fig. 6A), Slc2a2 expression was unaltered in Isl-1L/L; Pdx1-Cre mice, a model that ablated Isl-1 in the pancreatic epithelium at E13.5 (11,21). Thus Slc2a2 represents another key β-cell gene that is putatively regulated by Isl-1 only in the mature β-cell. To further determine whether Re1 and Re2 are involved in mediating Slc2a2 expression, we used a luciferase reporter in βTC3 cells. Luciferase activity in βTC3 cells was only observed using the reporter plasmid containing Re2. Furthermore, Isl-1 overexpression was sufficient to increase the Re2-containing vector reporter activity (Fig. 6B). The Slc2a2-Re2 sequence is highly conserved when compared with rat and human genomes (Fig. 6C). Within Re2, five putative HBEs were identified (Fig. 6C), and mutational analysis of Slc2a2-Re2 sites 1, 2, and 5 reduced the reporter activity, whereas similar treatment to site 4 enhanced activity (Fig. 6D). Overall, this analysis strongly supports the notion that Isl-1 directly regulates Slc2a2 through cis-regulatory elements in Slc2a2-Re2.
To determine the requirement for Isl-1 in the postnatal β-cell, we used a Tm-inducible, β-cell-specific, loss-of-function mouse model. Ablating Isl-1 in the postnatal β-cell impaired glucose tolerance and GSIS without affecting β-cell survival. Moreover, loss of Isl-1 compromised β-cell insulin secretion and altered the islet transcriptome. By combining microarray and ChIP-Seq analysis, we constructed a β-cell transcriptional network for Isl-1. Meta-analysis of this network identified new direct transcriptional targets of Isl-1, including Slc2a2, the glucose transporter that mediates a critical upstream step in mouse β-cell GSIS, and Pdx1, an essential transcriptional regulator of postnatal β-cell function. Remarkably, both Slc2a2 and Pdx1 are regulated by Isl-1 in the postnatal β-cell but not in pancreatic endocrine progenitors.
This study also exposed a limitation of the Pdx1-CreERTm line. When crossed to Rosa-lacZ mice, the Pdx1-CreERTm line displayed negligible Tm-independent recombination (34). In our Isl-1L/L; Pdx1-CreERTm mice, however, we encountered a significantly greater degree of Tm-independent Isl-1 recombination. We demonstrated that this occurred postnatally and that our analysis was not confounded by developmental deletion of Isl-1. Notably, 8-week-old Isl-1L/+; Pdx1-CreERTmIP(Tm) animals displayed no phenotype. Therefore, the phenotype in the Isl-1L/L; Pdx1-CreERTm(No Tm) mice reflected the accumulation of β-cells with two recombined Isl-1 alleles. Overall, increased surveillance for postnatal Tm-independent recombination with appropriate controls is warranted when using the Pdx1-CreERTm line.
Previous in vitro studies have identified that Isl-1 regulates genes associated with pancreatic endocrine function but have not provided definitive insights into the requirement for Isl-1 in the adult endocrine pancreas (17–20). Following postnatal deletion of Isl-1, we observed decreased β-cell function without increased apoptosis. In line with these observations, transgenic mice overexpressing Isl-1 in the endocrine pancreas increased β-cell function without enhanced β-cell proliferation (23). Considering the established relationship between Isl-1 and Insulin transcription (14,15,22), these in vivo findings further demonstrate that Isl-1 is essential for β-cell functional capacity. Nonetheless, our findings are at odds with the observation that overexpression and knockdown of Isl-1 in ex vivo rat islets enhanced β-cell proliferation and apoptosis, respectively (41). We speculate that the apoptosis/proliferation phenotypes observed in rat islets are secondary to Isl-1–regulating β-cell function or that the insults of islet isolation and culture may have unveiled a prosurvival function of Isl-1 in β-cells.
It is becoming increasingly evident that specified β-cells undergo a final period of maturation before attaining complete physiological capacity (42). Our findings suggest that the requirement for Isl-1 in postnatal versus developing β-cells is distinct. While ablating Isl-1 in endocrine progenitors had no effect on fetal levels of Slc2a2 and Pdx1 (11,21), we demonstrate here that ablating Isl-1 in postnatal β-cells reduces expression of both genes. Distinct roles for a transcription factor in developing versus postnatal β-cells have been observed for MafA and NeuroD1 (5,43). Similarly, Arx is necessary for the establishment of α-cell fate during pancreas development but is dispensable for maintaining α-cell fate even though its expression is maintained (44). Determining what regulates these shifts in transcriptional influence will be essential in defining immature versus mature pancreatic endocrine cells.
Observing that Isl-1 directly regulates MafA and Pdx1 in postnatal β-cells is also noteworthy when considering the plasticity of the adult pancreatic endocrine compartment (42). Multiple studies have demonstrated that misexpression of MafA and Pdx1 is sufficient to drive expression of β-cell–specific genes in non–β-cells (6,45,46). Isl-1 also directly regulates Arx, a transcription factor necessary for directing α-cell fate (33,44). Since α-, β-, δ-, and PP cells appear to arise from a common progenitor pool (42), ubiquitous expression of Isl-1 in the adult endocrine pancreas suggests that regulatory mechanisms exist to restrict Isl-1 transcriptional targets among the pancreatic endocrine lineages. It is well established that LIM-HD transcription factors drive cell-fate decisions in progenitor populations by activating distinct expression profiles (47). This is accomplished by LIM domain binding (Ldb) adaptor proteins nucleating combinatoric, multimeric LIM-HD complexes (48,49). The building blocks for this mechanism are still expressed in the adult endocrine pancreas. Ldb1, like Isl-1, is ubiquitously expressed in the adult endocrine pancreas, and other members of the LIM-HD family of transcription factors in addition to Isl-1 are enriched in islets as well, including Lhx1 and Mnx1/Hb9 (21,50). The epigenetic landscapes of the pancreatic endocrine cell types may also play role in restricting Isl-1 regulatory targets. For instance, Dnmt1-mediated DNA methylation is required to maintain β-cell identity, in part, by repressing Arx transcription (51). Similarly, mapping histone epigenetic modifications has identified variable enrichment of activating and repressive marks at cell type–specific genes between human α- and β-cells (52). Moving forward, it will be of great interest to determine which of these mechanisms contribute to directing the Isl-1 cistrome in the distinct endocrine lineages that populate the adult pancreas.
Acknowledgments. The authors thank the members of the Molecular Pathology & Imaging Core (MPIC) in the Penn Center for Molecular Studies in Digestive and Liver Diseases (P30-DK050306), the Pathology Core Laboratory at the Children’s Hospital of Philadelphia Research Institute, and the Radioimmunoassay/Biomarkers Core of the Penn Diabetes Research Center (P30-DK19525) for sample processing. The authors also thank the members of the Functional Genomics Core of the Penn Diabetes Research Center (P30-DK-19525) for performing sequencing and data analysis. The authors are also grateful for the Pdx1-CreERTm and Isl-1L/L mice provided by Drs. Douglas Melton (Harvard University) and Sylvia Evans (University of California, San Diego), respectively.
Funding. This work was supported by DK078606, DK019525, and JDRF 2-2007-730 to C.L.M.; R01 DK068157 to D.A.S.; and DK078606 to R.S. B.N.E. was supported by T32-GM07229 and T32-HD007516-15, and C.S.H. was supported by DK007061 and DK083160.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. B.N.E. researched the data and wrote the manuscript. A.D., J.L., and E.R.W. researched the data. C.S.H. researched the data and reviewed and edited the manuscript. J.S. and K.H.K. aided in generation and meta-analysis of high-throughput data and reviewed the manuscript. R.S., D.A.S., and C.L.M. supervised the research and wrote the manuscript. C.L.M. and D.A.S. 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.
Prior Presentation. Parts of this study were presented at the European Molecular Biology Organization/European Molecular Biology Laboratory Symposium 2014, Translating Diabetes, Heidelberg, Germany, 30 April–3 May 2014; at Imaging the Pancreatic Beta Cell: 5th NIDDK Workshop, Bethesda, MD, 15–16 April 2013; and at Keystone Symposium: Advances in Islet Biology, Monterey, CA, 25–30 March 2012.