Estrogen Receptor α Regulates β-Cell Formation During Pancreas Development and Following Injury

Decreased functional β-cell mass is the major cause for hyperglycemia in diabetes. Restoration of the endogenous β-cell mass as a therapeutic strategy, however, requires a better understanding of signaling pathways that control β-cell growth and differentiation. Embryonic β-cells are generated by a developmental program executed through the timed action of a number of key transcription factors among which Neurogenin3 (Ngn3) is key for endocrine specification. Ngn3⁺ cells delaminate from pancreatic epithelium, are mitotically quiescent, and give rise to endocrine cells. Ngn3 cells appear maximally competent for driving β-cell formation at embryonic day (E) 14.5. Formed β-cells expand through self-replication, already evident at E18.5, and continue into early postnatal life (1). Also in adult mice with severely injured pancreas by partial duct ligation (PDL), Ngn3⁺ cells are generated near duct epithelium and can differentiate into β-cells (2). β-Cells are vastly generated through replication in PDL (3,4), but some derive from acinar (5) and duct (6) cells, apparently through an Ngn3⁺ stage (2,5) as in embryonic pancreas.

How the numbers of Ngn3⁺ endocrine progenitors and replicating β-cells are controlled in the embryonic or mature pancreas is uncertain. Identifying factors that control these processes and manipulating them may be of therapeutic advantage. What is known is that 17β-estradiol (E₂) enhances β-cell survival and glycemic control in various animal models (7,8) by signaling through estrogen receptor (ER) α (8,9) and/or ERβ (10).

However, little is known about the importance of estrogen and ER signaling for β-cell proliferation and differentiation. So far, no in vivo effects on β-cell formation have been reported for the ER antagonist tamoxifen (TAM), although this compound is used to conditionally activate Cre recombinase activity (Cre^ERT) in genetic tracing studies on embryonic and adult endocrine progenitor and β-cell ontogeny. We investigated whether ER signaling is important for β-cell replication and neogenesis during embryogenesis and in injured adult pancreas.

All mouse experiments were performed according to the guidelines of our institutional ethics committee for animal experiments and national guidelines and regulations. Balb-c mice were obtained from Janvier Labs (Le Genest-Saint-Isle, France). The mouse strains Ngn3^CreERT;R26R^YFP and RIP^CreERT;R26R^YFP and Ngn3 knock-add-on enhanced yellow fluorescent protein (YFP) (Ngn3^YFP) were previously described (3,11). ERα^−/− mice are infertile and were obtained by crossing ERα^+/− mice as previously described (12).

All PDL pancreatic surgeries in the current study were performed in 8-week-old male mice as previously described (2). Embryos and embryonic pancreas and duodenum were dissected as previously described (13). Mouse embryos were obtained after timed mating. Pregnant mice were killed by cervical dislocation, uteri dissected in ice-cold sterile Dulbecco’s PBS, and embryos removed from the deciduas. For immunohistochemistry studies, embryos were fixed in 10% neutral-buffered formalin overnight at 4°C and embedded in paraffin. For RNA analysis, embryos were kept on ice-cold Dulbecco’s PBS, and the gut was dissected to remove the pancreas buds. Pancreas and duodenum were collected in RNAlater (R0901; Sigma-Aldrich, Diegem, Belgium).

TAM (T5648; Sigma) was dissolved in corn oil (C8267; Sigma) at 20 mg/mL, and 4-hydroxytamoxifen (4OHT) (579002; Calbiochem, Darmstadt, Germany) dissolved in pure ethanol to 20 mg/mL was diluted to 5 mg/mL in corn oil (C8267; Sigma). E₂ (E2785; Sigma) dissolved in pure ethanol to 0.4 mg/mL was diluted to 0.4 ng/mL or 2 ng/mL in 0.9% NaCl for intrapancreas injection, and aromasin (ARO) (Pfizer, Brussels, Belgium) was dissolved to 12.5 mg/mL in 0.3% hydroxypropyl cellulose (191884; Sigma) in PBS.

For PDL studies in adult mice, TAM, 4OHT, ARO, and the corresponding vehicles (corn oil, corn oil, and 0.3% hydroxypropyl cellulose/PBS, respectively) were given subcutaneously. E₂ (20 or 100 pg in 50 μL) and vehicle (50 μL saline) were injected in the ligated tail portion of PDL pancreas. Before PDL and sham surgery or intrapancreatic injection, mice were sedated by ketamine 3.5 mg/kg body weight i.p. (Ceva Santé Animale, Brussels, Belgium) plus xylazine 0.5 mg/kg body weight i.p. (Bayer, Diegem, Belgium).

Samples for immunohistochemistry were fixed in 10% neutral buffered formalin overnight at 4°C and embedded in paraffin. Sections (4–5 μm) were deparaffinized and washed in 0.2% Tween 20 (P5927; Sigma) in PBS, antigen retrieval was performed in a steamer, and slides were blocked in 10% donkey serum (Jackson ImmunoResearch Inc., Suffolk, U.K.). Primary antibodies for insulin (1:3,000, guinea pig, generated at the Diabetes Research Center, Vrije Universiteit Brussel, Brussels, Belgium), ERα (1:200, rabbit, E1644; Spring Bioscience, Fremont, CA), green fluorescent protein (1:100, goat; Abcam, Cambridge, U.K.), 5-iodo-2'-deoxyuridine (IdU) (1:100, mouse; BD Biosciences, Erembodegem, Belgium), Ki67 (1:1,000, rabbit; Leica Microsystems, Diegem, Belgium), Ngn3 (1:2,000, rabbit; Millipore, Temecula, CA), and E-cadherin (1:50, mouse; BD Biosciences) were incubated overnight at 4°C at the indicated dilutions. Secondary antibodies for detection of guinea pig, rabbit, goat, or rat antibodies were labeled by fluorescence (Cy3, Cy2, or Cy5). Nuclei were stained by 4 mg/mL Hoechst 33342 (Sigma).

Total RNA was isolated from tissue (74104, RNeasy; QIAGEN, Venlo, the Netherlands). Only RNA with RNA integrity number R ≥7 (2100 Bioanalyzer; Agilent) was further analyzed. cDNA synthesis and real-time quantitative PCR were done as previously described (14) with TaqMan Universal PCR Master Mix on an ABI Prism 7700 Sequence Detector, and data were analyzed using Sequence Detection System version 1.9.1 software (all Applied Biosystems, Life Technologies, Ghent, Belgium). Mouse-specific primers and probes were for Ngn3 (5′-GTCGGGAGAACTAGGATGGC-3′ [forward primer], 5′-GGAGCAGTCCCTA GGTATG-3′ [reverse primer], and 5′-CGGAGCCTCG GACCACGAA-3′ [probe]), Tert (Mm.PT.53a.13753537), Lcn2 (Mm.PT.53a.10167155), ptma (Mm.PT.53a.30222141), IL-6 (Mm.PT.49a.10005566), E2F1 (Mm.PT.53a.33146682), C-Jun (Mm.PT.49a.8204422.g), Pax4 (Mm01159043_g1), Pax6 (Mm.PT.51.14285402), Pdx1 (Mm.PT.51.11487144), Esr2 (Mm.PT.49a.17681375.g) (all from Integrated DNA Technologies [IDT], Leuven, Belgium), and Esr1 (Mm00433149_m1, TaqMan Gene Expression Assay; Life Technologies). Quantitative PCR reactions were performed with cDNA corresponding to 20 ng RNA as previously described (3). Data were normalized to cyclophilin A (CycloA) (Mm02342429, IDT). Conventional PCR primers for ERα (Exon3) were P1: 5′-AGAATGGCCGAGAGAGACTG-3′ and P2: 5′-TTCTCTTAAAGAAAGCCTTGCAG-3′ (IDT). Primers (IDT) for genotyping ERα^+/+ (wild-type) and ERα^−/− mice and embryos were previously described (12).

Protein extracts were prepared as previously described (15). Fifty micrograms of protein were separated by SDS-PAGE and transferred onto a nitrocellulose membrane. ERα was detected with a rabbit polyclonal antibody (1:200, MC20; Santa Cruz Biotechnology, Heidelberg, Germany). β-Actin (sc1616; Santa Cruz) was detected for verification of protein loading.

The measurement of β-cell volume (in cubic millimeters) in PDL or sham tail pancreas was performed as described previously (2,3). The measurement for each pancreas tail was based on 4-μm sections equally spaced (116 μm apart) and spanning the whole tail tissue, corresponding to the analysis of 3% of the total volume. Insulin content of pancreatic tissue was measured as previously described (16). E₂ concentration in whole pancreas tail tissue was determined after tissue homogenization using a Retsch Mixer Mill MM 400 (R20.745.0001; Retsch GmbH, Haan, Germany) using 5-mm grinding metal beads (R22.455.0003; Retsch). Lysates were cleared by centrifugation after which 100-μL samples were ether extracted. E₂ was then assayed using a radioimmunoassay (RIA) kit (DIAsource ImmunoAssays, Louvain-la-Neuve, Belgium).

Images were obtained by fluorescence microscopy (Olympus [Aartselaar, Belgium] BX61 with Hamamatsu C10600 ORCA-R2 camera). Ki67⁺ or IdU⁺ β-cells (insulin-positive) and Ngn3⁺ or YFP⁺ E-cadherin–positive cells were quantified in a nonautomated manner (inspecting and counting individual cells). For the quantification of IdU⁺ or Ki67⁺ β-cells, 10 nonconsecutive sections of each PDL pancreas (head or tail) were studied, and at least 1,000 β-cells per sample were counted. For the quantification of Ngn3⁺ cells, 10 nonconsecutive sections of each embryonic pancreas were studied, and at least 4,000 E-cadherin–positive cells per sample were counted. Images were analyzed using SmartCapture3 version 3.0.8 (Digital Scientific UK, Cambridge, U.K.), ImageJ (National Institutes of Health, Bethesda, MD), and Photoshop CS version 1.3.1 (Adobe, San Jose, CA) software.

Data are expressed as mean ± SEM of at least three independent experiments. Groups were compared using unpaired two-tailed Student t test, one-way ANOVA, or two-way ANOVA with Bonferroni or Šidák post hoc tests. Differences were considered statistically significant if P < 0.05.

We investigated whether TAM influenced the outcome of PDL-induced β-cell proliferation. First, transgenic Ngn3^CreERT;R26R^YFP male mice of mixed genetic background as well as inbred Balb-c male mice underwent PDL surgery followed by subcutaneous injection of either TAM (4 mg per injection, 20 mg in total over a 14-day period) (Fig. 1A), vehicle, or nothing and application of IdU through drinking water. Although more β-cells were IdU⁺ in the tail portion of PDL pancreas (from here on termed “PDL tail”) than in the nonligated head portion (from here on termed “PDL head”) (Fig. 1B), β-cell proliferation was blunted in PDL tail of TAM-treated mice compared with nontreated or vehicle-treated mice (Fig. 1B and C and Supplementary Fig. 1). Thus, TAM inhibited β-cell proliferation in PDL tail regardless of the genetic background or the presence of Cre recombinase. In another cohort of Balb-c mice with PDL surgery but without nucleotide analogs, β-cell proliferation in PDL tail was assessed by immunofluorescence for proliferation marker Ki67. The percentage of Ki67⁺ β-cells in PDL tail was significantly lower in TAM- versus vehicle-treated mice (Fig. 1D), corroborating that β-cell proliferation was inhibited by TAM.

We then examined the effect of 4OHT, a selective ER modulator and metabolite of TAM that binds to endogenous ERs and inactivates them (17). At only 1 mg per injection (5 mg total), 4OHT and TAM were equally efficient in inducing YFP expression (>90%) in β-cells of adult Rip^CreERT;R26R^YFP mice (Supplementary Fig. 2), and β-cell proliferation in PDL tail was decreased to the same extent by both compounds (Fig. 1E). Additionally, a single subcutaneous injection of either 1 or 2 mg TAM at day 5 after PDL (Fig. 1F) resulted in significantly lower β-cell proliferation in PDL pancreas at day 7 (Fig. 1G). Because TAM can act as an agonist or antagonist of estrogen signaling, depending on the cell type, its effect on β-cell proliferation was also examined in the pancreas of pregnant mice. A single intraperitoneal injection of 1.5 mg TAM at gestation day (G) 13.5 caused a marked decrease of β-cell proliferation at G15.5, whereas no significant effect of TAM on the basal β-cell proliferation in nonpregnant females (Fig. 1H) was seen. Collectively, the data suggest that adult β-cell proliferation in PDL pancreas or during pregnancy is suppressed by antagonism of estrogen signaling through ERs.

Ngn3⁺ cells have been identified in the ductal lining of ligated pancreatic tail and are capable of differentiating in vivo or ex vivo to functional β-cells (2). Activation of Ngn3 gene expression and the presence of Ngn3⁺ cells are required for in vivo β-cell expansion in PDL pancreas (2,3). Because Ngn3 gene expression is regulated by estrogen in developing neurons (18), we assessed whether Ngn3 transcriptional level in PDL pancreas was influenced by modulation of ER signaling.

As expected, the amount of Ngn3 transcripts increased in PDL tail compared with PDL head in mice that received vehicle, but it was significantly reduced following administration of TAM (Fig. 2A). In addition, the constitutive expression of Ngn3 associated with enteroendocrine progenitor cells (19) was also decreased in the duodenua of TAM-treated mice (Fig. 2B). These data suggest that TAM decreases Ngn3 gene expression and/or the number of Ngn3-expressing cells in PDL pancreas and intestine.

We next evaluated the impact of TAM administration on increased β-cell volume in PDL pancreas. The total β-cell volume was significantly increased in PDL versus sham tails of pancreas from mice that did not receive TAM, and this was not the case in TAM-treated mice (Fig. 2C). There was also a trend toward a lower average β-cell volume in PDL tails of TAM-treated mice compared with PDL tails of vehicle-treated mice (Fig. 2C). The insulin content of PDL tail was doubled in vehicle-treated mice but significantly lowered following TAM treatment (Fig. 2D). These data indicate that TAM blunts the increase in PDL-induced β-cell volume and insulin content, albeit to a limited extent.

The aforementioned results suggest that ERs are involved in β-cell replication and Ngn3 gene induction in PDL pancreas. When ERs were studied at the mRNA level in PDL tail versus head, Esr1 (ERα) expression was fivefold higher in PDL tail, whereas subtype Esr2 (ERβ) mRNA was 18-fold lower in PDL tail (Fig. 3A). Because Esr2 mRNA was nearly undetectable in PDL tail tissue (C_t = 39.0 ± 0.97), ERα appears to be the prime ER subtype in PDL pancreas. Studies in murine mammary cells showed that ERα can promote proliferation in response to E₂ primarily when ERβ is lowly expressed (20). The ERα (Esr1) transcript levels in β-cells isolated from sham tail or PDL tail (6) were similar (Fig. 3B). To investigate whether ERα activity contributes to the increase of β-proliferation induced by PDL, we analyzed ERα^−/− (null mutant of ERα) mice obtained by breeding of ERα^+/− parents (12). The full disruption of ERα in the knockout mice was evidenced by the absence of ERα mRNA and polypeptide in the uterus of female ERα^−/− mice and by its abnormal development (Supplementary Fig. 3) (21). PDL was performed in 8-week-old male ERα^−/− and ERα^+/+ littermates, and the pancreas was studied following immune detection of insulin and Ki67 at day 14 after surgery. ERα gene inactivation diminished β-cell proliferation specifically in PDL tail (Fig. 3C). In addition, ERα^−/− mice had decreased Ngn3 mRNA in PDL tail (Fig. 3D) and duodenum (Fig. 3E) compared with ERα^+/+ littermates. Thus, genetic inactivation of ERα and administration of TAM had the same effect on β-cell proliferation and Ngn3 transcript level in PDL pancreas and duodenum, suggesting that ERα activity contributed to a higher Ngn3 gene expression and/or number of Ngn3⁺ cells and increased β-cell proliferation in PDL pancreas.

To investigate how ERα activity could be influenced by PDL, we assayed the level of its natural ligand E₂ by RIA in the pancreas tail of male Balb-c mice 7 days after sham or PDL surgery. Although the level of E₂ in sham tail was generally below the detection limit of the assay (2 ng/L), its concentration was 17 ± 2 ng/L in PDL tail (±20 pg/PDL tail) (Table 1). Furthermore, the expression of several estrogen-regulated genes (22–28) was increased in PDL tail (Fig. 4A), suggesting increased ER activity.

The activity of ERα in β-cells of PDL pancreas was investigated on the basis of its subcellular localization. In uterus, ERα was predominantly located in the nucleus (Supplementary Fig. 3D), which is consistent with other studies (29). ERα protein remained predominantly cytosolic (9,29) in β-cells of sham tail (Fig. 4B) or PDL head (Fig. 4C) but was nuclear in β-cells of PDL tail (Fig. 4D). The nuclear ERα signal in β-cells of PDL tail was strongly diminished following TAM administration (Fig. 4E), and ERα protein was absent from the pancreas of ERα^−/− mice (Fig. 4F). Thus, PDL increased E₂ in the ligated portion of the pancreas and stimulated a nuclear localization of ERα in β-cells, which was counteracted by TAM.

We next examined whether an artificial increase of E₂ could stimulate in situ β-cell proliferation and Ngn3 gene expression in PDL pancreas. E₂ or vehicle was injected on the day of surgery (day 0) and on day 3, and PDL tail was studied on day 7. E₂ further increased β-cell proliferation and Ngn3 mRNA level in PDL tail (Fig. 4G–I). To investigate whether endogenous synthesis of E₂ was required for increased β-cell proliferation in PDL, we subcutaneously administered ARO, an inhibitor of aromatase, the rate-limiting enzyme for E₂ synthesis (Fig. 4J). ARO blunted activation of the β-cell cycle (Fig. 4K). Thus, β-cell proliferation in PDL depended on E₂ synthesis, and increasing E₂ within the environment of PDL was sufficient for stimulating proliferation. In addition, both TAM administration and ERα knockout interfered with nuclear localization of ERα protein in β-cells and with β-cell proliferation in the PDL pancreas. These data indicate that estrogen regulation of β-cell growth is mediated by ERα. Because macrophages were suggested to promote β-cell proliferation during PDL (30), we analyzed the presence of myeloid cells in untreated or TAM-treated PDL pancreas by flow cytometry (31). No shift in the myeloid cell composition and no difference in the number of macrophages (Supplementary Fig. 4A) were induced by TAM treatment of mice with PDL. Moreover, no difference was observed for other immune cell types (T cells, B cells, dendritic cells) (data not shown). As expected, the level of mRNA encoding various macrophage markers (Supplementary Fig. 4B) and proinflammatory cytokines (Supplementary Fig. 4C) was increased in PDL tail compared with sham tail, but treatment with TAM did not significantly change the expression levels in PDL (Supplementary Fig. 4B and C). Thus, estrogen has no effect on myeloid cell infiltration or inflammatory cytokines in PDL pancreas. That the expression of the ERα-responsive gene IL6 was not significantly affected by TAM suggests that additional regulatory factors control IL6 expression in PDL.

We next examined whether ERα plays a role in embryonic β-cell proliferation and development. Proliferation was evaluated at embryonic stage E18.5 when massive β-cell replication occurs (32). TAM efficiently reached the embryonic β-cells of Rip^CreERT;R26R^YFP mice because 89 ± 1.9% of E18.5 β-cells were YFP⁺ when 1.5 mg TAM was administered at day E16.5 to the pregnant mother (Supplementary Fig. 5). Embryonic β-cell proliferation evaluated at E18.5 was markedly lowered by the TAM treatment (Fig. 5A), suggesting that ERs might be involved in embryonic β-cell expansion.

The majority of endocrine progenitor cells are generated at E15.5 and transiently express Ngn3 (33). Because Ngn3 gene expression is regulated by estrogen in developing neurons (18) and was reduced by ERα inactivation in PDL pancreas (Figs. 2A and 3D), we examined the Ngn3 mRNA level in E15.5 pancreas of ERα^+/+ and ERα^−/− embryos and found it significantly decreased in ERα^+/+ embryos that received TAM and in untreated ERα^−/− embryos compared with ERα^+/+ control embryos (Fig. 5B). Transcription factors Pax6 and Pax4, which are crucial for differentiation of the α- and β-cell lineages downstream (34), were also reduced by TAM or knockout of ERα (Fig. 5B), suggesting that embryonic endocrine cell specification was affected. In contrast, Pdx1, which is already expressed before Ngn3 in pancreas formation, was not influenced (Fig. 5B). Moreover, Ngn3 gene expression in endocrine cell progenitors of E15.5 duodenum was significantly lower in TAM-treated ERα^+/+ embryos and in ERα^−/− embryos compared with ERα^+/+ control embryos (Fig. 5C).

Ngn3⁺ progenitor cells were identified in E15.5 pancreas either by the presence of YFP reporter in Ngn3^YFP embryos (11) or by detection with an NGN3-specific antibody. The percentages of E-cadherin–positive cells that were YFP⁺ or NGN3⁺ in E15.5 pancreas were significantly reduced when TAM was administered to the mother at E13.5 (Fig. 5D), indicating reduced numbers of embryonic endocrine progenitor cells. Furthermore, the percentage of E-cadherin–positive cells that expressed Ngn3 protein was lower in the pancreas of E15.5 ERα^−/− versus ERα^+/+ embryos (Fig. 5E). Together, the data suggest that the generation of Ngn3⁺ cells and the subsequent β-cell proliferation in developing pancreas are both regulated by ERα.

The current study shows that signaling through ERα is involved in the control of β-cell progenitor activation and β-cell proliferation in embryonic and injured adult pancreas. Both genetic loss and competitive inhibition of ERα signaling decreased the number of Ngn3⁺ endocrine pancreas progenitors in embryonic mice, suggesting a regulatory role for estrogen during pregnancy. Specification of endocrine progenitor cells is counteracted by Notch signaling (35). Because ERα signaling decreases Notch transcriptional activity in breast cancer cells (36), it is possible that E₂-mediated regulation of Ngn3 gene expression and endocrine cell differentiation in developing pancreas occurs through Notch signaling. In addition, in the vertebrate brain, aromatase, which catalyzes estrogen synthesis, is expressed by neural progenitor cells (37) and regulates neural progenitor cell proliferation (38). E₂ has been speculated to lower Notch signaling during brain development, resulting in Ngn3-dependent neural cell differentiation (18). Of note, certain variants of the NGN3 gene are associated with more severe hyperglycemia in men but not in women with diabetes (39). Molecular pathways other than ERα signaling are likely because Ngn3⁺ cells are present, albeit in lower numbers, in the pancreas of ERα^−/− embryonic mice. In line with these data, the pancreatic β-cell mass has been found to be similar in ERα null mutant and wild-type mice (9,40), possibly by progenitor cell activation through Ngn3-independent pathways (41). Because ER signaling was decreased in the embryos of mice treated with TAM at a dose generally used for activation of Cre^ERT recombinase in transgenic mice, while this also lowered the number of Ngn3⁺ cells as well as the proliferation of β-cells, we question the accuracy of data on the ontogeny of embryonic β-cells or Ngn3⁺ cells generally obtained with Cre^ERT-expressing mice. The current data suggest at least a temporary inhibition of estrogen-regulated endocrine cell formation by TAM. We report a 50–75% decrease of β-cell proliferation and Ngn3 expression by TAM or ERα^−/−, whereas TAM treatment does not cause such major changes of the β-cell volume and insulin content in PDL tail. To further address the impact of ERα on β-cell expansion in PDL, it is necessary to study mice with conditional knockout of ERα in β-cells. In addition, whether the effects described here account for ERα activation in the pancreas is uncertain because this study is based on global ERα knockout or drug administration. Therefore, a study in mice with ERα knockout in pancreatic cells, particularly in β-cells, will be informative.

Circulating levels of prolactin, placental lactogen, and estrogen are high during late pregnancy (42) when an increase of β-cell mass compensates for the increased needs of insulin in the mother (43). ER signaling affects the maternal β-cell function under these conditions (9,40,44), and on the basis of the current data, it also contributes to the increase of β-cells in the fetal pancreas by stimulating neogenesis and proliferation. Further studies are needed to clarify the precise mechanism of steroid and lactogenic hormones in the regulation of fetal β-cell formation.

In the pancreas of adult mice severely injured by PDL, Ngn3 gene expression is activated in endocrine progenitor–like cells and β-cell proliferation is stimulated, resulting in increased β-cell volume (2,3). We report that ER signaling is increased under these conditions, regulates Ngn3 expression levels and β-cell proliferation, and thus affects regenerative processes in PDL pancreas. Using TAM-based genetic tracing is unlikely to be the most adequate approach for evaluating the extent of adult β-cell formation from alternative cell sources in PDL pancreas because TAM clearly interferes with ER signaling and β-cell formation. A washout procedure might diminish the suppressive effects of TAM on ER signaling, but it is not excluded that transient exposures to TAM may have long-lasting effects on β-cell differentiation.

Both isoforms of ER are present in β-cells and contribute to control of function, proliferation, and survival (8–10,29,40,45), although mechanistic details are lacking [reviewed in Tiano and Mauvais-Jarvis (46)]. PDL increased the number of transcripts encoding ERα while decreasing ERβ mRNA to a nearly undetectable level. Deletion of only ERα prevented PDL-induced proliferation of β-cells, and although this does not exclude a role for ERβ, the latter could not compensate for the defect. Of note, the cell cycle is activated through ERα signaling when ERβ is lowly expressed as in mouse mammary cells (20) and human islet–derived precursor cells (47). Additionally, studies in breast cancer with ER-dependent growth have shown that the abundance of ERα-encoding transcripts increases, whereas that of ERβ mRNA decreases (48,49), suggesting that tissues with an increased growth tendency may require elevated ERα expression and an increased ERα/ERβ ratio. Moreover, in the case of β-cells, several conditions now appear to stimulate growth through ERα, including PDL injury, embryonic pancreas, and pregnancy.

Compared with sham treatment, PDL tail shows increased estrogen concentration, proliferation, and nuclear ERα localization in β-cells. The latter two processes were blunted in ERα^−/− mice as well as by aromatase inhibition and enhanced by intrapancreas injection of E₂ at a dose similar to its concentration in PDL pancreas, suggesting a role for synthesis of E₂ possibly by newly formed adipocytes in situ (50). In addition, we showed that local administration of E₂ potentiated Ngn3 gene expression in PDL pancreas. Because estrogen and activation of ERα do not increase β-cell proliferation in rodent models for diabetes (51) and in human islets transplanted in diabetic mice (52), we cannot exclude that the estrogen effects described in the current study would be specific to the PDL model. Nevertheless, the data also suggest involvement of estrogen in other models of β-cell proliferation, namely in both the embryonic and the mature pancreas during pregnancy.

Estrogen effects also depend on dose and route of administration (53). Cell-specific delivery of E₂ stimulates beneficial activities without unwanted side effects. Estrogen conjugated to GLP-1, which is delivered preferentially to the pancreas, reduces adiposity and improves hyperglycemia, whereas systemic delivery of estrogen potentially increases its adverse effects, such as in breast cancer (54). In summary, the new effects of estrogen signaling in developing and regenerating pancreas support ERα as a candidate target for control of glucose homeostasis and β-cell formation in diabetes therapy.

Acknowledgments. The authors thank Pierre Chambon and Andrée Krust (Université Louis Pasteur, Illkirch, France) for providing the ERα mutant mice; Ann Demarré, Veerle Laurysens, Jan De Jonge, Erik Quartier (Diabetes Research Center, Vrije Universiteit Brussel, Brussels, Belgium), Ellen Anckaert, and Gaby Schoonjans (Department of Endocrinology, Universitair Ziekenhuis Brussel, Brussels, Belgium) for technical assistance; and Yves Heremans (Diabetes Research Center, Vrije Universiteit Brussel, Brussels, Belgium) for critical revision of the manuscript.

Funding. Financial support was from the Chinese Scholarship Council (Y.Y.), Institute for the Promotion of Innovation by Science and Technology in Flanders (V.C., H.H.), Beta Cell Biology Consortium (H.H.), Innovative Medicines Initiative Joint Undertaking under grant agreement 155005 (Improving β-cell Function and Identification of Diagnostic Biomarkers for Treatment Monitoring in Diabetes [IMIDIA]) comprising a financial contribution from the European Union’s Seventh Framework Programme (2007–2013) and European Federation of Pharmaceutical Industries and Associations in-kind contribution (H.H.), Stichting Diabetes Onderzoek Nederland (H.H.), National Fund for Scientific Research for Flanders (Fonds Wetenschappelijk Onderzoek-Vlaanderen [FWO]) (H.H.), Interuniversity Attraction Poles Programme (H.H.), and Vrije Universiteit Brussel Research Council (H.H., M.V.d.C.).

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

Author Contributions. Y.Y., Y.C., H.H., and M.V.d.C. contributed to the study concept and design and data acquisition, analysis, and interpretation. B.L., S.D.G., G.L., V.C., E.V.O., W.S., N.D.L., G.M., and J.A.V.G. contributed to the data acquisition, analysis, and interpretation. H.H. 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.