Islet-1 is Required for the Maturation, Proliferation, and Survival of the Endocrine Pancreas

The derivation of the Isl-1^LoxP and Pdx1-Cre transgenic line has been reported previously (28,29). All mice were kept on a mixed outbred CD1 background. Pdx1-Cre;Isl-1^L/+ and Isl-1^L/L mice were mated to generate Pdx1-Cre;Isl-1^L/L mutant mice. Littermate Isl-1^L/L, Isl-1^L/+, and Pdx1-Cre;Isl-1^L/+ were indistinguishable from mixed CD1 Isl-1^+/+ controls in our assays. Animal experiments were approved by the Children's Hospital of Philadelphia's institutional animal care and use committee.

Overnight fasted animals were injected intraperitoneally with 2 g glucose (Sigma) per kilogram of body weight. Blood glucose values were monitored at 0, 15, 30, 60, 90, and 120 min after injection using an automatic glucometer (One Touch Ultra; LifeScan). To prepare plasma, blood was collected in heparinized tubes (BD Microtainer), spun, and stored at −80°C until assayed. Plasma insulin levels were measured using a Luminex kit (Linco). Total pancreatic insulin and glucagon content were assessed by radioimmune assay of acid-ethanol extracts at the University of Pennsylvania Diabetes Center.

Tissues were fixed in 4% paraformaldehyde (PFA) overnight at 4°C and then embedded in either paraffin or optimal cutting temperature freezing medium. Slides (8–10 μm sections) were subjected to microwave antigen retrieval in 10 mmol/l citric acid buffer (pH6.0) and blocked with protein blocking reagent (Immunotech). Slides were incubated with primary antibodies overnight at 4°C, and appropriate secondary antibodies were added for 2 h at room temperature. The following primary antibodies were used: Glucagon (1:3,000; Linco), insulin (1:1,000; Linco), somatostatin (1:3,000; Santa Cruz), pancreatic polypeptide (1:50; Zymed), ghrelin (1:200; Santa Cruz), Isl-1 (1:50; Developmental Studies Hybridoma Bank 39.4D5 and 40.2D6), Pax6 (1:500; Covance), MafA (1:1,000; Bethyl Laboratories), and Pdx1 (1:200; Santa Cruz). Ngn3 (1:500; Developmental Studies Hybridoma Bank), Cy2, Cy3, and Cy5 conjugated secondary antibodies (1:600; Jackson ImmunoResearch). The Developmental Studies Hybridoma Bank Isl-1 monoclonal antibodies were developed under the auspices of the National Institute of Child Health and Human Development and are maintained by The University of Iowa, Department of Biology (Iowa City, IA).

Total RNA from homogenized pancreata was extracted using the RNA Easy Kit (Qiagen). RNA was reverse transcribed using 1 μg Oligo(dT) primer, Superscript II Reverse Transcriptase, and accompanying reagents (Invitrogen). Real-time PCR reactions were set up using the Brilliant SYBR Green PCR Master Mix (Stratagene). All reactions were performed in triplicate with reference dye normalization, and median cycling threshold values were used for analysis. Primer sequences are available upon request.

Pancreata were removed, weighed, fixed in 4% PFA overnight at 4°C, and embedded in paraffin. Sections (8–10 μm) with maximum footprint were used for insulin staining. Images were taken under 4× magnification, and pancreatic tissue positive for insulin staining was measured by using IP Lab software. The β-cell mass was obtained by measuring the fraction of insulin-positive staining to total cross-sectional area and multiplying by the pancreatic weight. One section was used per pancreas with at least four control and mutant mice analyzed at each time point.

Immunostaining of Pax6 was used to detect late endocrine progenitors (hormone⁺ or hormone⁻ (30); individual hormone immunostains were used to mark mature endocrine cell subtypes. Pancreatic cross-sectional area was measured using IP Lab software. At E15.5 (n = 6), two adjacent sections with the largest pancreatic footprint were selected. Pax6⁺ cells were counted, and the results were normalized to cross-sectional area. To measure the differentiation of endocrine cell subtypes, the number of hormone-expressing cells was expressed relative to the number of Pax6⁺ cells. The same technique was used to measure endocrine cell subtypes at E18.5 (n = 3), except that the average was taken of ∼10 sections taken at seven-section intervals through the block. At P4 (n = 4), Pax6⁺ cells were counted in four sections and normalized to pancreas cross-sectional area.

The rate of apoptosis was evaluated by transferase-mediated dUTP nick-end labeling (TUNEL) staining using the ApopTag Peroxidase In Situ Apoptosis Detection Kit (Chemicon). Sections were processed for the TUNEL assay, and adjacent sections were stained with an anti-Pax6 antibody. The percentage of TUNEL⁺ endocrine cells was generated by dividing the number of TUNEL⁺ cells to the number of Pax6⁺ cells. Insulin staining was performed on the same section as TUNEL assay to locate islets. For bromodeoxyuridine (BrdU) assays, pregnant animals (E18.5) and pups were injected with 10 μl/g of BrdU Labeling Reagent (Zymed) and killed 6 h later. BrdU staining was performed with the BrdU Staining Kit (Invitrogen). Sections were labeled with BrdU antibody, and adjacent sections were stained with an anti-Pax6 antibody. The percentage of BrdU⁺ endocrine cells was generated by dividing the number of BrdU⁺ cells into the number Pax6⁺ cells. Insulin staining was performed on the same section as BrdU staining to locate islets. At least 2,000 Pax6⁺ cells were counted for each animal for both TUNEL and BrdU assays, and the number of Pax6⁺ cells was used for normalization.

pCS2 rat Isl-1-Myc plasmid (gift from Dr. Pfaff) was used as a template for in vitro translation of Isl-1-Myc using Quick Coupled Transcription/Translation System rabbit reticulocyte lysate reagents (Promega). DNA binding reactions (20 μl final volume) included 1 μl of in vitro translated protein or 10 μg of Ins-1 β-cell line nuclear extract (24,–26) and 400 fmol of ³²P-end labeled oligo probe. The oligonucleotide sequences are as follows: MafA Isl-1 wild type, −7,822-CGTAACGTTAATGGAAGATGCTTGCTGCAG-7793, and MafA Isl-1 mutant, 5′-GCCGTAACGTGCCGGGAAGATGCT-3′, with the mutation underlined. The reactions were allowed to proceed at 30°C for 20 min in a buffer containing 10 mmol/l HEPES (pH 7.8), 75 mmol/l KCL, 2.5 mmol/l MgCl2, 0.1 mmol/l EDTA, 1 mmol/l dithiothreitol, 3% (v/v) Ficoll, 1 mg/ml BSA, and 1 μg poly(dI-dC). Competition experiments were performed using 100-fold molar excess of unlabeled wild-type or mutant oligos. Antibody supershift analyses were performed using a cocktail of four Isl-1–specific antibodies (Developmental Studies Hybridoma Bank, 39.3F7, 39.4D5, 40.2D6, and 40.3A4) preincubated with the Isl-1 protein or nuclear extract at room temperature for 20 min before adding the DNA probes. Reactions were separated on 6% nondenaturing polyacrylamide gels at 150V for 2 h in 0.5× tris-borate-EDTA buffer. Gels were dried and visualized by autoradiography.

Cultured βTC3 cells were transfected using Lipofectamine reagent (Invitrogen) with 1 μg of MafA region3:pTK (wild type and Isl-1 mutant) or the pTK(An) chloramphenicol acetyltransferase (CAT) vector and 1 μg of Rous sarcoma virus enhancer–driven luciferase (pRSV-Luc). Transfection reporter constructs for MafA were as described (31). Site-directed mutagenesis (Quickchange Mutagenesis kit, Stratagene) was used to create a noncomplementary transversional mutation in the Isl-1 binding site 5′-CACGGCCGTAACGTGCCGGGAAGATGCTTGCTGC-3′ (mutation is underlined). Cell lysates were prepared 48 h after transfection, and luciferase (LUC) (Promega) and CAT (32) assays were performed. CAT values were normalized to the LUC internal control. Each experiment was performed in triplicate with at least two independently isolated plasmid preparations.

βTC3 cells were cultured for 72 h at 4 × 10⁶ cells per 10-cm dish, then isolated and cross-linked with 1% formaldehyde in DMEM for 5 min at room temperature. Protein DNA chromatin fragments were prepared by sonication and precleared with protein A-agarose/salmon sperm DNA (Millipore, Temecula, CA) for 2 h at 4°C. The precleared chromatin was then incubated overnight at 4°C with a cocktail of Isl-1 monoclonal antibodies (Developmental Studies Hybridoma Bank 39.4D5, 39.3F7, 40.3A4, 40.2D6) or as controls, normal mouse immunoglobulin (Santa Cruz Biotechnology), or no antibody. Antibody-bound chromatin complexes were precipitated with protein A-agarose beads at 4°C for 4 h. Washed complexes were eluted from the beads and cross-links reversed. Quantitative real-time PCR was performed on ∼1:20 of the Isl-1–immunoprecipitated DNA using SYBR Green PCR mater mix (Applied Biosciences) and an ABI Prism 7900 instrument. Input DNA dilutions were used to generate a standard curve and to normalize amplification of the gene of interest (ΔCt). The enrichment of target control sequences in the chromatin immunoprecipitation (ChIP) DNAs (n = 3–4) was then calculated relative to the inactive phosphoenolpyruvate carboxykinase (PEPCK) promoter set as onefold (ΔΔCt). Nonquantitative PCR reactions were performed using 1:20 of the purified immunoprecipitated DNA with Taq polymerase hotstart master mix (5 Prime, Gaithersburg, MD) and 15 pmol of primer. ChIP PCR amplicons were visualized using a 1.5% agarose gel stained with ethidium bromide in 1× tris-acetate-EDTA buffer. Each ChIP experiment was repeated at least three times using independent chromatin preparations. The sequences of the real- time and standard PCR primers are available upon request.

To inactivate Isl-1 specifically in the developing pancreatic epithelium, mice carrying the floxed allele of Isl-1 (29) were mated to Pdx-1–driven Cre transgenic mice (28). We anticipated that Isl-1 inactivation would occur in the pancreatic epithelium around E12.5 based on the detection of Pdx1-Cre activity in Rosa26R mice (33). However, efficient loss of Isl-1 protein in Pdx1⁺ pancreatic progenitors was not observed until E13.5 (Fig. 1,A and B; data not shown), with the delay likely reflecting the accessibility of the Isl-1 locus to Cre and/or Cre expression levels in our strain background. By E13.5, greater than 90% of epithelial cells lacked Isl-1, as the percentage of Pdx1⁺Isl-1⁺ double-positive cells was reduced from 1.3% in control pancreata to 0.1% in the Pdx1-Cre;Isl-1^L/L pancreata (Fig. 1 C). The few remaining Isl-1⁺ cells in the Pdx1-Cre;Isl-1^L/L pancreas were either unrecombined Pdx1⁺ epithelial cells, first-wave endocrine cells, or mesenchymal cells (24) (data not shown).

Pdx1-Cre;Isl-1^L/L mice were born at the expected Mendelian ratio and did not differ from their littermate controls in size at birth (data not shown). There was also no change in total pancreas weight up to 3 weeks of age (Fig. 2,A). However, both male and female mutant mice displayed elevated random-fed glucose levels as early as postnatal day (P)0 that worsened with age (Fig. 2 B and C). Eventually, mutant mice died between 3 to 8 weeks after birth.

We next evaluated endogenous islet β-cell function by intraperitoneal glucose tolerance tests (IPGTT) in 3-week-old male and female Pdx1-Cre;Isl-1^L/L and control animals. Mutant mice had slightly increased blood glucose levels after an overnight fast that became dramatically elevated without returning to baseline after administration of exogenous glucose (Fig. 2 D). Although both male and female mutant mice showed similar β-cell dysfunction, male mutant animals also exhibited growth retardation shortly after week 3, whereas females remained indistinguishable from controls (data not shown).

To investigate if the change in β-cell function reflected alterations in islet cell number and/or size, we compared hematoxylin-eosin–stained pancreatic sections from control and Pdx1-Cre;Isl-1^L/L animals. Fewer and smaller islets were found in P4 mutant mice (Fig. 2,E and F), with an even more dramatic difference by P14 (Fig. 2,G and H). The number of late endocrine progenitors marked by Pax6 expression (30) was quantified at E15.5, E18.5, and P4 to determine the period of endocrine cell loss. A 27.2% reduction was observed by E18.5 (Fig. 3,B), with a 39.7% decrease in Pax6⁺ cells by P4 (Fig. 3 A–C). These data show that Isl-1 is involved in maintaining the Pax6⁺ endocrine progenitor population, with their loss in Pdx1-Cre;Isl-1^L/L animals resulting from either a defect in cell differentiation, proliferation, and/or survival.

To determine if endocrine cell proliferation and/or survival was affected in the mutant pancreata, we next performed BrdU incorporation and TUNEL to evaluate these processes. A decrease in the rate of endocrine cell proliferation was first observed after birth in the pancreata of Pdx1-Cre;Isl-1^L/L mice, with a 33.2% reduction in BrdU incorporating cells at P6 (Fig. 3,E–J). Similarly, although there was not a significant change in the number of apoptotic TUNEL⁺ cells at E18.5, a twofold increase was found in P4 mutants (Fig. 3,K–P). We next examined by real-time PCR if mRNA levels of the prosurvival and/or proapoptosis mediators Bid, Bad, Bax, and Bcl2 was changed by P7 (34). Only Bad was modestly but significantly increased (Fig. 3 D). Collectively, these results suggest that a major function of Isl-1 in postnatal animals is to control the proliferation and survival of endocrine cells.

To investigate if Isl-1 is involved in endocrine hormone production, we measured the total glucagon and insulin content of control and mutant pancreata at birth. Interestingly, glucagon and insulin content were reduced in the Pdx1-Cre;Isl-1^L/L pancreas by 78 and 69%, respectively (Fig. 4,A). These results suggest that the reduction in total hormone content cannot be solely caused by the loss of endocrine cell mass described above (Fig. 3,A–C). Rather, Isl-1 is likely to have an additional role in regulating endocrine hormone production itself. To identify the cause of this defect in mutant animals, we measured the cross-sectional area of pancreatic tissue stained for insulin as an indicator for mature β-cell mass at E18.5, P4, and P14 and found significant reductions in this population of cells in mutant animals at all stages examined (Fig. 4 B).

To distinguish between changes in endocrine cell number (both mature and immature) and defects in the differentiation of islet cell subtypes, we immunostained pancreatic tissues with Pax6 (30) in conjunction with hormone cocktails (glucagon, insulin, somatostatin, and ghrelin) at E15.5, E18.5, and P4. Unlike control pancreata, in which the majority of the Pax6⁺ endocrine cells also coexpressed hormones, most of the Pax6⁺ endocrine cells in mutant pancreata lacked hormone expression at all stages examined (Fig. 4 C–H). These data suggest that Isl-1 regulates the maturation of pancreatic endocrine cells.

To determine when the defects in endocrine cell differentiation were first detectable, we performed immunofluorescence analysis at several embryonic stages (E13.5, E15.5, and E18.5). E13.5 was the earliest time point at which a reduction in glucagon- and insulin-expressing cells was evident (Fig. 5,A and B) coinciding with the onset of Isl-1 deletion in our model (Fig. 1). The number of glucagon⁺, insulin⁺, and somatostatin⁺ cells at E15.5 was reduced in mutant mice by 83, 63, and 99%, respectively (Fig. 5,C–F and M). A similar pattern was obtained upon analysis of hormone mRNA expression by real-time PCR (Fig. 5,N). Notably, the few hormone-expressing cells remaining in the mutant pancreas were not Isl-1⁺ (data not shown). The number of pancreatic polypeptide⁺ cells was decreased in E18.5 mutant pancreata (Fig. 5,O), whereas islet ghrelin⁺ (ε)- and exocrine amylase⁺-expressing cell numbers were unchanged (Fig. 5 K, L, and O and data not shown). These data strongly suggest that Isl-1 is required to regulate the formation of respective hormone-producing cell types.

Many islet-enriched transcription factors are essential regulators of pancreatic development and/or hormone gene expression (35). Next, we examined if Isl-1 controls expression of these key regulators. A dramatic reduction in MafA mRNA levels was observed in the E15.5 Pdx1-Cre;Isl-1^L/L pancreata, while the expression of other transcription factors including Ngn3 was unaffected (Fig. 6,A). Quantitative analysis of Ngn3⁺ cells from control and mutant pancreata further confirmed Pdx1-Cre;Isl-1^L/L animals have normal numbers of early endocrine progenitors (Fig. 6,B–D). In addition, decreased MafA expression in β-cells was evident by immunostaining (Fig. 6,E and F). Notably, the known regulators of MafA transcription (MafB, Pax6, Pdx1, Foxa2, and Nkx2.2) (19,31) were unaffected in the mutant animals (Fig. 6 A, G, and H and data not shown). In summary, these results suggest that Isl-1 acts upstream of MafA in β-cells.

MafA was originally identified in the β-cell as a transactivator of the C1 element in the insulin enhancer (36). Because mRNA and protein levels of previously known MafA regulators were not affected in Pdx1-Cre;Isl-1^L/L mice (Fig. 6,A and data not shown), we hypothesized that Isl-1 directly controls MafA transcription. The MafA 5′ promoter has six evolutionarily conserved regions, of which “region 3” is not only the longest and most well conserved (Fig. 7 A) but uniquely capable of driving β-cell line–specific expression in vitro (31).

ChIP analysis was performed in mouse βTC3 cells to determine if Isl-1 occupied region 3 of the MafA gene. A significant and quantitative enrichment of region 3 was found compared with PEPCK control sequences by standard or real-time PCR only after α–Isl-1 immunoprecipitation (Fig. 7,B and C). Conversely, insulin enhancer sequences were not enriched in Isl-1 ChIP DNA (Fig. 7 C).

Next, we performed electrophoretic mobility shift assays with either in vitro translated Myc-tagged Isl-1 or Ins-1 β-cell line nuclear extract to analyze binding to the Isl-1-like site in a MafA region 3 base pair −7,816 to −7,808 spanning probe (Fig. 7,A). Isl-1 indeed bound to this site because the level of the protein DNA complex was reduced by addition of excess wild-type and not mutant competitors (Fig. 7,D). Additionally, antibodies against Isl-1 blocked binding to the MafA probe (Fig. 7,D). Mutation of the Isl-1 binding site in region 3 also reduced reporter activity in βTC-3 cells (Fig. 7 E). Collectively, these in vitro and endogenous binding data strongly suggest that Isl-1 is a direct regulator of MafA transcription in β-cells.

Ahlgren et al. (24) had shown previously through tissue recombination studies using Isl-1 null embryos that Isl-1 expression in the mesenchyme is essential for proper differentiation of the exocrine pancreas, while the epithelial expression of Isl-1 is required for early endocrine pancreas development). We have identified novel roles for Isl-1 later in the pancreatic epithelium. Our findings establish Isl-1 as transcriptional regulator required in the formation, proliferation, and survival of islet α-, β-, δ-, and PP cells after the secondary transition and in postnatal lives.

A striking phenotype of Pdx1-Cre;Isl-1^L/L animals is the dramatic loss of glucagon, insulin, somatostatin, and pancreatic polypeptide expression in developing cells, with a severe reduction in islet cell mass after birth. Lack of Isl-1 in the developing endocrine pancreas has a global impact on islet cells, with the exception of ghrelin expression. We showed that the decrease in α, β, δ, and PP production was not simply the result of a reduction in the Ngn3⁺ endocrine progenitor population but was because of the incomplete endocrine cell maturation. Although it is not clear from this study why ghrelin expression was not affected in the Pdx1-Cre;Isl-1^L/L animals, it is possible that ε-cells represent an immature endocrine population that does not require Isl-1 action. This notion is supported by the fact that ghrelin-expressing cells are present only during fetal development and in young animals but not in the mature pancreas (1).

Although Isl-1 was identified more than a decade ago, its direct downstream targets in the islet remain largely uncharacterized. We identified MafA as a novel transcriptional target of Isl-1. Notably, MafA mRNA expression was compromised very early and highly selectively upon Isl-1 inactivation in vivo. Furthermore, Isl-1 binding was detected within the principal MafA transcriptional control region in vitro and by ChIP in β-cell lines. Finally, the luciferase construct containing mutations in the Isl-1 response element exhibited lower activity than the wild-type region 3 reporter. However, Isl-1 was unable to independently activate transfected MafA promoter-driven reporters in non–β-cells, as, for example, the relative activity level of Isl-1 to the vector control with the MafA region 3 reporter equaled 0.76 (±0.10)-fold. Similarly, Pdx-1, E47, and NeuroD1/BETA2, widely accepted insulin gene activators, are also unable to activate insulin-driven reporters in non–β-cell lines on their own (37,–39). Moreover, Isl-1 is incapable of stimulating a variety of different promoter-driven reporters under circumstances wherein other LIM-homeodomain proteins could potentiate (i.e., with E47) (37). The coactivators that are necessary for mediating Isl-1 activation in β-cells or for region 3 specifically have not yet been identified. The loss of only MafA in β-cells cannot explain the pancreatic phenotype observed in our Pdx1-Cre;Isl-1^L/L animals. For example, although insulin transcription is reduced in MafA^−/− mice, their total insulin content remains normal. Again, unlike Pdx1-Cre;Isl-1^L/L mice, the MafA mutant has no developmental phenotype but displays compromised β-cell activity soon after birth because of defects in insulin secretion (40).

Isl-1 has been shown to regulate preproglucagon gene expression in vitro (41) and presumably directly influences preproglucagon mRNA levels in Pdx1-Cre;Isl-1^L/L mice. A small but significant decrease in the mRNA level of the α-cell–enriched Arx transcription factor was also observed in mutant animals. Although much is known about how Arx regulates α-cell specification (35,42,43), little is known about its significance to α-cell maturation and function. It is also remains to be determined how Isl-1 controls somatostatin and pancreatic polypeptide gene expression in islet δ- and PP cell, respectively.

Unlike first wave Isl-1⁺ cells that are postmitotic (24), subsets of Isl-1⁺ cells after the secondary transition have the ability to proliferate (data not shown). Although it is unknown whether proliferation and/or survival of early endocrine cells was affected in the Isl-1 null pancreata (24), our data demonstrated a crucial role of Isl-1 in regulating endocrine cell growth and survival in young animals. Growth and maintenance of β-cell mass is crucial to prevent or delay loss of β-cells in diabetic patients. We propose that there are at least three transcriptional mechanisms utilized by Isl-1 in islet β-cell development and function: 1) direct regulation of targets bound by Isl-1 alone; 2) targets bound by both Isl-1 and MafA (or other factors); and perhaps to a lesser extent 3) indirect regulation of targets, bound for example by MafA alone. The former two are likely involved in regulating genes involved in cell proliferation and survival, processes unaffected in MafA null mice. In this regard, Bad gene expression was downregulated in Pdx1-Cre;Isl-1^L/L mice. However, it is unclear if this is a direct or indirect response since we were unable to detect binding of Isl-1 to the proximal Bad promoter in ChIP assays. Combined genome-wide transcription factor binding and expression profiling studies will need to be performed on factors like Isl-1 to determine how islet cell differentiation is regulated in mammalian cells.

Severe hyperglycemia observed in the mutant animals is likely because of both the loss of islet mass and decreased hormone expression by the remaining islet cells. The decrease in islet cell number results from decreased cell proliferation and increased cell death, changes that became significant soon after birth. Isl-1 has important functions in the differentiation of hormone-producing cells beyond what was shown previously, but the exact mechanism of how Isl-1 regulates its target genes is largely unknown. In addition, future studies in which Isl-1 is deleted conditionally in mature β-cells will determine if this factor controls β-cell function as well as β-cell survival and proliferation in adult animals.

This work was supported by National Institutes of Health Grants DK078606, DK019525, and JDRF 2-2007-703 to C.L.M. C.S.H. was supported by National Institutes of Health Grants DK007061 and DK083160 and R.S. by DK078606.

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

We thank Klaus Kaestner, Joshua Friedman, and Rana Gupta for careful reading of the manuscript. We thank Gary Swain and the members of the Morphology Core in the Center for Molecular Studies in Digestive and Liver Disease (P30-DK050306) as well as the Radioimmunoassay and Biomarkers Core of the Penn Diabetes Center (P30-DK19525) for sample processing. We are also grateful for the Isl-1 expression vector obtained from Sam Pfaff and the Pdx1-Cre mouse from Pedro Herrera. We are also thankful to Jeffrey Raum, Min Guo, and Brian McKenna for reagents and assistance with the ChIP and EMSA experiments.