1773 (2007) 1455–1461www.elsevier.com/locate/bbamcr

Biochimica et Biophysica Acta

The transcription factors Stat5a/b are not required for islet development butmodulate pancreatic β-cell physiology upon aging

Ji-Yeon Lee a, Oksana Gavrilova b, Behrous Davani a, Risu Na a,Gertraud W. Robinson a, Lothar Hennighausen a,⁎

a Laboratory of Genetics and Physiology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health,8 Center Drive, Room 101, Bethesda, MD 20892-0822, USA

b Mouse Metabolic Core Facility, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892, USA

Received 25 January 2007; received in revised form 3 May 2007; accepted 3 May 2007Available online 29 May 2007

Abstract

In insulinoma cell lines proliferation and insulin gene transcription are stimulated by growth hormone and prolactin, which convey their signalsthrough the transcription factors Stat5a and 5b (referred to as Stat5). However, the contribution of Stat5 to the physiology of β-cells in vivo couldnot be assessed directly since Stat5-null mice die perinataly. To explore the physiological role of Stat5 in the mouse, the corresponding gene locustargeted with loxP sites was inactivated in β-cells using two lines of Cre recombinase expressing transgenic mice. While the RIP-Cre transgene isactive in pancreatic β-cells and the hypothalamus, the Pdx1-Cre transgene is active in precursor cells of the endocrine and exocrine pancreas.Mice carrying two floxed Stat5alleles and a RIP-Cre transgene developed mild obesity, were hyperglycemic and exhibited impaired glucosetolerance. Since RIP-Cre transgenic mice by themselves display some glucose intolerance, the significance of these data is unclear. In contrast,mice, in which the Stat5 locus had been deleted with the Pdx1-Cre transgene, developed functional islets and were glucose tolerant. Mild glucoseintolerance occurred with age. We conclude that Stat5 is not essential for islet development but may modulate β-cell function.© 2007 Elsevier B.V. All rights reserved.

Keywords: Stat5; Rip-Cre; Pdx1-Cre; Islets of Langerhans; β-cells; Glucose homeostasis

1. Introduction

Growth hormone (GH), prolactin (PRL) and placentallactogen (PL) are potent growth factors for insulinoma cells invitro [1–3]. In addition, these hormones stimulate insulinproduction in vitro, partially through the transcriptionalactivation of the insulin gene promoter [4]. These effects aremediated through the GH receptor (GHR) and the PRL receptor(PRLR), which activate the tyrosine kinase Jak2 and thetranscription factors Stat5a and Stat5b (referred to as Stat5)[5,6]. The mechanisms by which GH and PRL promote β-cellproliferation and gene expression, and the role of Stat5 have

Abbreviations: Stat, Signal transducer and activator of transcription; JAK,Janus kinase; GH, growth hormone; PRL, prolactin; PL, placental lactogen;GHR, GH receptor; PRLR, PRL receptor; FFA, free fatty acid⁎ Corresponding author. Tel.: +1 301 496 2716; fax: +1 301 480 7312.E-mail address: hennighausen@nih.gov (L. Hennighausen).

0167-4889/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.bbamcr.2007.05.010

been explored in vitro [7–9]. In insulin-producing INS-1 cellsand in cultured rat islets, GH and PRL induced the phos-phorylation of Stat5a and Stat5b and their nuclear translocation[10,11], suggesting their involvement in β-cell physiology. Insupport of this, PRLR−/− and GHR−/− mice exhibited a re-duction in islet density and β-cell mass [12,13]. Pancreaticinsulin mRNA levels were also reduced in adult PRLR-nullmice. In addition, PRLR−/− and GHR−/− mice exhibitedimpaired glucose tolerance and increased insulin sensitivity,respectively. These observations established a physiologicalrole for GH and PRL in β-cell function and glucosehomeostasis.

Although Stat5 mediates GH- and PRL-induced proliferationof insulinoma cell lines and insulin gene transcription, andexpression of dominant-negative Stat5 in transgenic miceresulted in increased body weight and impaired glucosetolerance [14], the physiological consequences of a completeabsence of Stat5 in β-cells remained elusive. Since Stat5−/−

1456 J.-Y. Lee et al. / Biochimica et Biophysica Acta 1773 (2007) 1455–1461

mice die perinatally [15] it is impossible to explore the functionof Stat5 in the physiology of β-cells. To address the significanceof Stat5 we deleted the Stat5 locus in the entire pancreas and inβ-cells of mice using Cre-mediated recombination.

2. Results

2.1. Deletion of the Stat5 locus in pancreatic β-cells and thehypothalamus altered islet morphology and content

Since the complete loss of Stat5 from the mouse genomeresults in perinatal lethality [15], we elected to delete Stat5specifically in the pancreas of the mouse using Cre-mediatedrecombination. Two lines of Cre expressing mice were used todelete the Stat5 locus bracketed by loxP sites. While the RIP-

Fig. 1. Targeted disruption of the Stat5 genes and assessment of Stat5 deletion in βStat5fl/f l;RIP-Cre mice. (B) Fluorescence immunohistochemical analysis of Stat5 (redpanel) mice. Pancreata from 7-month-old mice were used for immunohistochemical(C) Impaired glucose homeostasis in 4- to 5-month-old Stat5fl/fl;RIP-Cre mice and RIP-of 6–8 males of each group. (D) Insulin release from isolated islets. Islets were isolat5 months. Insulin secretion was induced by basal (3 mM) and 16.7 mM of glucose.

Cre transgene [16] is active in pancreatic β-cells and in thehypothalamus [17], the Pdx1-Cre transgene [18] expresses Crein pancreatic precursor cells, which results in the deletion offloxed genes in exocrine and endocrine cells. Mice weregenerated that carried two floxed Stat5 alleles and the RIP-Cretransgene (Stat5fl/fl; RIP-Cre mice) (Fig. 1A). Stat5b was detectedin β-cells throughout the islets of control mice but not inStat5fl/fl; RIP-Cre mice (Fig. 1B). There was no difference ininsulin staining between control and Stat5fl/fl; RIP-Cre mice andlevels of insulin mRNA in islets of these mice were comparable(data not shown). Loss of Stat5 resulted in a disruptedarchitecture of islets as evidenced by the migration ofglucagon-expressing α-cells into the central region of the islets(Fig. 1B, right panel). Deletion of the Stat5 locus was alsoobserved in mice older than 1 year (data not shown),

-cells of Stat5fl/fl;RIP-Cre mice. (A) Schematic of the construct used to generate) and glucagon (green) in control (Stat5fl/fl; left panel) and Stat5fl/fl;Rip-Cre (rightanalyses. Residual Stat5b-positive cells in Stat5fl/fl;RIP-Cre mice are non-β-cells.Cre transgenic mice. Results are expressed as average blood glucose level±SEMed from two animals per genotype which were used for glucose tolerance test at(∗, #)Pb0.05; (∗∗, ##)Pb0.01.

1457J.-Y. Lee et al. / Biochimica et Biophysica Acta 1773 (2007) 1455–1461

demonstrating that there was no selective advantage of cellscarrying a non-recombined Stat5 locus.

2.2. Stat5fl /f l; RIP-Cre mice developed mild obesity

Up to 12 weeks of age there was no significant difference inbody weight between Stat5fl/fl; RIP-Cre and control mice.However, older Stat5fl/f l; RIP-Cre mice developed mild obesity(Table 1) and by 9 months of age, mutant mice weighed on theaverage 20% more than their control litter mates. Since the RIP-Cre transgene is also active in the hypothalamus [17,19], thedeveloping obesity could be due to the hypothalamic loss ofStat5 and an altered metabolism. Late-onset obesity was notobserved in RIP-Cre transgenic mice or in Stat5f l/f l; Pdx1-Cre

mice, further implicating hypothalamic Stat5 in the regulationof metabolism. To explore the overall metabolic consequence inStat5fl/f l; RIP-Cre mice, metabolic responses were measured. At7 months of age, serum glucose levels were slightly elevatedand serum free fatty acids (FFA) and leptin levels weresignificantly higher in mutant mice but serum insulin levelswere normal (Table 1). Body composition was measured byNMR. While the lean mass was similar between mutant andcontrol mice the adipose tissue mass of Stat5fl/f l; RIP-Cre femalemice was significantly elevated (Table 1).

2.3. Stat5fl/f l; RIP-Cre mice failed to maintain glucosehomeostasis

To examine the impact of the loss of Stat5 on the response toan acute glucose challenge, glucose disposal was assessed infasted mice by glucose tolerance tests. At all ages examined(6 weeks to 11 months), Stat5f l/f l; RIP-Cre mice displayed higherglucose levels than the controls at all time points after glucoseinjection (Fig. 1C). The severity of impaired glucose responsewas increasing in both female and male mice as they aged. Aswe described recently [20], RIP-Cre transgenic mice in aC57BL/6 background display some glucose intolerance as earlyas 6 weeks of age. Although glucose intolerance of 4- to 5-month-old Stat5fl/f l; RIP-Cre mice was more severe than that seen

Table 1Body composition and metabolic parameters of Stat5f l/f l; RIP-Cre mice

Stat5f l/f l Stat5f l/f l; RIP-Cre

Body weight (g) 31.1±0.9 36.8±2.0Lean mass (g) 22.51±0.6 22.58±0.6Total adipose mass (g) 5.34±0.9 12.14±1.9 a

Total adipose mass (% body weight) 17.18±2.7 32.11±3.5a

Serum glucose (mg/dl) 189.1±2.8 210.2±8.9 b

Serum insulin (ng/ml) 1.2±0.9 1.2±0.2Serum TG (mg/dl) 131.6±22.6 156.0±18.9Serum FFA (mM) 0.4±0.04 0.6±0.05b

Leptin (ng/ml) 10±1.9 21.2±4.2

Weight and body composition was measured in 7-month-old female mice. Bloodwas collected from the tail vein directly into non-EDTA-coated capillary tubesand centrifuged to separate the serum, which was used to assay triglycerides,FFAs, glucose, insulin and leptin. Values shown are the mean±SEM; seven tonine mice were used for analysis.a Pb0.01 vs. Control (Stat5f l/f l).b Pb0.05 vs. Control (Stat5f l/f l).

in RIP-Cre mice (Fig. 1C), both showed reduced insulinsecretion from isolated islets upon challenge with 16.7 mM ofglucose (Fig. 1D).

2.4. Stat5fl/f l; Pdx1-Cre mice exhibit normal glucose homeostasis

Since the presence of the RIP-Cre transgene by itself causessome glucose intolerance [20,21], we were unable to draw firmconclusions about the contribution of pancreatic Stat5 in glucosehomeostasis. To bypass this problem, Stat5fl/f l; Pdx1-Cre mice, inwhich Cre activity is confined to the pancreas, were generated(Fig. 2A). Stat5f l/f l; Pdx1-Cremice were observed over a 15-monthperiod and their weights were indistinguishable from controls(Table 2). Moreover, levels of glucose, insulin and glucagonswere normal over the 15-month period (Table 2). Isletarchitecture and the presence of Stat5b in pancreatic β-cellswere evaluated by immunohistochemical analysis. In controlmice Stat5b was detected throughout the islets (Fig. 2B), but itwas largely absent in Stat5fl/f l; Pdx1-Cre mice (Fig. 2C). The fewremaining Stat5b-positive cells in Stat5fl/f l; Pdx1-Cre mice eitherexpressed glucagon (Fig. 2C) or somatostatin (data not shown).This suggests that the Stat5 locus failed to be deleted in a smallnumber of progenitor cells. In mutant mice, α-cells had migratedinto the central region of the islets (Fig. 2C), but not in controlmice (Fig. 2B). Real-time PCR analysis confirmed the loss ofStat5b mRNA in islets from Stat5fl/f l; Pdx1-Cre mice compared toStat5fl/f l controls. There was no apparent difference in insulinstaining between control and Stat5fl/f l; Pdx1-Cremice (Fig. 2D andE). Real time PCR confirmed the presence of comparable levelsof insulin mRNA in islets from control and mutant mice, withbeta actin as internal control (data not shown).

To investigate the consequences of the loss of Stat5 on theresponse to an acute glucose challenge, glucose disposal wasassessed in fasted mice by glucose tolerance tests. Glucoselevels in fasting Stat5fl/f l; Pdx1-Cre mice prior to the administra-tion (ip) of glucose were normal in all age groups tested(Fig. 3A and B). Glucose tolerance was overtly normal in2–3 months old Stat5fl/f l; Pdx1-Cre mice (Fig. 3A) but mild glucoseintolerance developed in mice older than 6 months (Fig. 3B).

Mouse studies have suggested a role for GH and PRL on β-cell function during pregnancy. To test whether this effect ismediated by Stat5, glucose tolerance tests were performed priorto pregnancy and on day 17 of pregnancy. Stat5fl/f l; Pdx1-Cre

female mice were slightly more glucose intolerant duringpregnancy than their control litter mates (Fig. 3D), but thedifferences were not statistically significant. These results implythat Stat5 signaling contributes to the optimal function of β-cells under conditions of stress, such as pregnancy and aging.

3. Discussion

Given the established role of GH, PRL and PL in inducingthe proliferation of insulinoma cell lines and insulin genetranscription and secretion [7,9–11,22], as well as theinvolvement of Stat5 in these signaling pathways, moreprofound defects in Stat5 mutant mice had been anticipated.However, studies demonstrating a role for Stat5 in the

Fig. 2. Targeted disruption of the Stat5 genes and assessment of Stat5 deletion in β-cells of Stat5fl/fl; Pdx1-Cre mice. (A) Schematic of the Stat5 locus targeted with loxPsites and deletion of the Stat5 genes using the Pdx1-Cre transgene. (B, C) Assessment of Stat5 protein levels. Pancreata from 3-month-old mice were used forimmunohistochemical analyses. Sections were stained with antibodies against glucagon (green) and Stat5b (red). Immunofluorescence was viewed under an OlympusBX51 microscope (Olympus America Inc., Melville, NY) and images were captured with a Nikon DXM1200 digital camera (Nikon Inc., Melville, NY). (D, E)Fluorescence immunohistochemical analysis of insulin (green) and glucagon (red) in control (d: Stat5fl/fl) and Stat5fl/fl; Pdx1-Cre mice (E).

1458 J.-Y. Lee et al. / Biochimica et Biophysica Acta 1773 (2007) 1455–1461

proliferation of insulinoma cell lines might not fully reflect thein vivo situation since immortalized cell lines are subject toaltered growth control. This study now demonstrates that acomplete absence of Stat5 from pancreatic β-cells in mice hasno discernable impact on islet development and function inyoung animals. However, Stat5 appears to be a modulatingcomponent in β-cell function in older mice.

Table 2

Stat5 fl/fl Stat5 fl/fl;Pdx1-Cre P-value

Body weight (g) 28.8±1.3 28.7±2.5 0.49Glucose (mg/dl) 129.5±11.5 139.7±7.9 0.23Insulin (ng/ml) 0.6±0.3 0.6±0.2 0.22Glucagon (ng/ml) 246.7±27.5 272.3±89.4 0.61

Blood was collected from the tail vein directly into non-EDTA-coated capillarytubes and centrifuged to separate the serum, which was used for assay. Valuesshown are the mean±SEM; four to five 10- to 15-month-old fed mice were usedfor analysis.

The obesity, glucose intolerance and metabolic defectsobserved upon deletion of the Stat5 locus using the RIP-Cretransgene are likely not only due to the loss of Stat5 frompancreatic β-cells, but also the results of other events. Inparticular, RIP-Cre mice by themselves develop some glucoseintolerance [21], and loss of Stat5 from the hypothalamuspossibly influences metabolic pathways. Thus these mice arenot suited to explore the role of Stat5 in the physiology ofpancreatic β-cells.

Tissue culture studies have suggested that Stat5 is central tothe GH- and PRL-mediated proliferation of β-cells and theactivation of the genes encoding insulin and cyclin D2 [7–9]. Insupport of a role for Stat5, expression of constitutively activeStat5b in cell lines stimulated GH-independent proliferation[7,23]. Since functional islets were present in Stat5 mutant miceanalyzed in this study, it is clear that the presence of Stat5 is notessential for the generation and proliferation of β-cells.However, no stereological analyses were performed and it ispossible that the β-cell mass in Stat5-mutant mice is slightly

Fig. 3. Results from glucose tolerance tests in Stat5fl/f and Stat5fl/fl; Pdx1-Cre mice. Glucose tolerance tests were performed on fasted mice after an i.p. injection of 2 g/kgBW glucose at 10 weeks (A) and 6 months (B) of age. Blood glucose values were measured immediately before and 15, 30, 60, and 120 min after glucose injection.Results are expressed as average blood glucose level±SEM of 6–8 males of each group. (∗) and (∗∗) means Pb0.05 and Pb0.01, respectively, fromindependent t-tests (unpaired and two-tailed).P-value fromANOVA (analysis of variance between groups, two-way repeated) was 0.8661 at 10weeks (B) and 0.0007 at6months (B) of age. (C,D) Glucose tolerance test was performed on 2- to 3-month-old females before pregnancy (C) and on day 17 of gestation (D).Micewere fasted 9 hand administrated 2 g/kg BW glucose loads at time zero. All results are reported as mean±SEM of 5 females of each group. (∗) means Pb0.05 from t-tests, but ANOVAshows no significant difference both before pregnancy (P=0.7127) and on day 17 of gestation (P=0.2021).

1459J.-Y. Lee et al. / Biochimica et Biophysica Acta 1773 (2007) 1455–1461

different. Since PRLR−/− and GHR−/− mice displayed areduction in islet density and β-cell mass [12,13] it is possiblethat β-cell proliferation is controlled by signaling pathwaysother than Stat5. GH and PRL stimulate insulin production invitro, in part through the activation of Stat5 and thus atranscriptional induction of the insulin gene promoter [4–6,10,11]. Again, we found that levels of insulin mRNA wereunaltered in mutant mice, further demonstrating a negligiblerole of Stat5 in the context of the pancreas in vivo.

A recent study also addressed the role of Stat5 in vivothrough the expression in transgenic mice of a dominant-negative mutant of Stat5b (DNSTAT5) or a constitutively activemutant of Stat5b (CASTAT5) under control of the rat insulin 1promoter (RIP) [14]. Mild glucose tolerance, increased bodyweight, increased insulin and leptin levels and an increased β-cell area were observed in DNSTAT5 transgenic mice only on ahigh-fat diet but not on a standard diet. Thus, the loss of Stat5 inislets due to the deletion of the genes or due to expression of adominant negative form results in similarly mild defects andsupports the notion that Stat5 modulates β-cell function onlyunder defined physiological conditions.

Although a wide range of tissue culture cell experimentshave linked Stat5 to many aspects of cell physiology, theanalysis of mice from which Stat5 has been deleted in definedtissues and cells has not confirmed all of these proposed roles.Two explanations can be offered for this discrepancy. It is

possible that the functions of Stat5 observed in tissue culturecells are unique to those experimental systems but do not reflectthe situation encountered in the context of an animal.Alternatively, it is possible that other members of the Statfamily, namely Stat1 and Stat3, compensate for the loss of Stat5.Since the Stat3 gene is juxtaposed to the Stat5a/b locus it isimpossible at this time to test this possibility through thedeletion of all three genes.

4. Methods

4.1. Generation of Stat5 fl/fl; RIP-Creand Stat5 fl/fl; Pdx1-Cre mice

Stat5fl/fl mice [15] were bred with RIP-Cre [16] or Pdx1-Cre[18] transgenic mice. Mice were maintained on a 12-h light, 12-h dark cycle and fed water and standard diets. The NIDDKAnimal Care and Use Committee approved all procedures andstudies were conducted in accordance with National Institutesof Health guidelines. Genotyping was performed by PCRamplification of tail DNA from each mouse at 3 weeks ofage. The primers for genotyping of the Stat5 floxed allele wereprimer 1 (5′-GAAAGCATGAAAGGGTTGGAG-3′), primer2 (5′-AGCAGCAACCAGAGGACTAC-3′) and primer 3 (5′-AAGTTATCTCGAGTTAGTCAGG-3′). Primer pair 1 and 2amplifies a fragment of 450 bp from wild-type mice and a 200-bp fragment was detected by primers 2 and 3; the latter is located

1460 J.-Y. Lee et al. / Biochimica et Biophysica Acta 1773 (2007) 1455–1461

in the pLoxpneo vector, in mice heterozygous or homozygousfor the Stat5 floxed allele. Primers for the RIP-Cre transgenewere 5′-CTC TGG CCATCT GCT GAT CC-3′, which binds inthe insulin 2 gene promoter and 5′-CGC CGC ATA ACC AGTGAA AC-3′, which binds in the Cre sequence. This pairamplifies a 550 bp Rip–Cre specific fragment. For the detectionof the Pdx1-Cre transgene, 5′-TTGAAACAAGTGCAGGTGTTC G-3′ and 5′-CCATGA GTG AACGAA CCT GGT CG-3′were used.

4.2. Preparation of pancreatic sections andimmunohistochemical analyses

Pancreas samples were dissected from fed mice and fixedwith 4% paraformaldehyde/PBS solution overnight at 4 °C.Five-micrometer longitudinal sections of paraffin-embeddedsections were rehydrated with xylene followed by decreasingconcentrations of ethanol. Antigen retrieval was performed byheat treatment with an antigen unmasking solution (VectorLaboratories Inc.), and tissue sections were blocked for 30 minin PBS-Tween (PBST) containing 3% goat serum. Sectionswere incubated with anti-Glucagon (Sigma) and anti-Stat5b (sc-836, Santa Cruz, Calif.) antibodies. The primary antibodieswere allowed to bind overnight at 4 °C. Secondary antibodieswere from Molecular Probes, Inc. (Eugene, OR) and includedgoat anti-rabbit immunoglobulin G (IgG) conjugated to AlexaFluor 594 and goat anti-mouse IgG conjugated to Alexa Fluor488. Sections were incubated in the dark for 30 min, washedtwice in PBST, and mounted with Vectashield (VectorLaboratories Inc.). Immunofluorescence was viewed under anOlympus BX51 microscope (Olympus America Inc., Melville,NY) and images were captured with a Nikon DXM1200 digitalcamera (Nikon Inc., Melville, NY).

4.3. Analytical assays and metabolic rates

Blood glucose levels were determined from the tail vein usinga glucometer (Accu-Chek, Roche, Indianapolis, IN). Seruminsulin and leptin levels were measured by radioimmunoassay(Linco Research Inc., St. Charles, MO). Serum fatty acids andtriglyceride levels were analyzed in fed mice using a commercialFA kit (Roche, Indianapolis, IN) and the GPO-Trinder kit (SigmaChemical Co., St. Louis, MO), respectively. Insulin and glucosetolerance tests were performed in fasted animals after i.p.injections of either insulin (0.75 U/kg body weight) or glucose(2 g/kg body weight). Blood glucose values were measuredimmediately before and 15, 30 and 60 min after insulin injection,and before and 15, 30, 60, and 120 min after glucose injection.For insulin release, glucose (3 g/kg body weight) or L-arginine(1 mg/g BW in PBS) [24] was injected IP, and venous blood wascollected at 0, 2.5, 5, 15, and 30 min in chilled heparinized tubes,immediately centrifuged, and the plasma stored at −20 °C.Insulin levels were measured with a RIA kit. Metabolic rateswere measured in fed female mice 7 months of age by indirectcalorimetry using the Oxymax system (Columbus Instruments,Columbus, OH) as described by Héron-Milhavet et al. [25]. Datawere collected for 24 h at room temperature (22 °C) and for 24 h

at thermoneutral temperature (30 °C). Data are expressed as theaverage of 24 h and normalized to (total body weight)0.75.

4.4. Body composition

Body composition was measured in awake mice using anNMR analyzer (Bruker Minispec mq10, Bruker Optics Inc.,Billerica, MA).

4.5. Insulin secretion from isolated islets

Islets were isolated from 4-month-old control (Stat5fl/f) andStat5fl/f;RIP-Cre mice by collagenase digestion (Liberase, RocheDiagnostics, IN) followed by centrifugation over a Histopaquegradient (Sigma, St. Louis, MO). Islets were handpicked undera stereo microscope and incubated overnight in RPMI 1640medium supplemented with 10% FBS, 2 mM L-glutamine,100 IU/ml penicillin, and 100 μg/ml streptomycin at 37 °C and5% CO2. For assay of insulin secretion, islets were handpickedand incubated in Krebs–Ringer solution (119 mM NaCl, 5 mMKCl, 2.2 mM CaCl2, 1.19 mM KH2PO4, 240 mM NaHCO3,1.19 mMMgSO4, 10 mM HEPES [pH 7.4], 0.1% bovine serumalbumin, and 3.3 mM glucose) at 37 °C for 45 min. At the endof the incubation period, islets were transferred to a new 12-wellplate and stimulated with various glucose concentrations(3.3 mM and 16.7 mM) for 1 h at 37 °C. Supernatants werecollected and insulin release was measured with a rat insulinELISA kit (Crystal Chem. Inc., Chicago, IL). Samples were runin triplicate, and the number of experiments was n=4–5. Totalislet insulin content was measured in batches of five islets. Theislets were sonicated twice in 200 μl acidic ethanol (0.25 mol/lHCl in 87.5% ethanol) and incubated overnight at 4 °C. Thesamples were centrifuged and total insulin was measured in thesupernatant.

4.6. Statistical analysis

All results are reported as mean±SEM for equivalent groups.Results were compared with independent t-tests (unpaired andtwo-tailed) reported as P-values.

Acknowledgments

We thank Shoshana Yakar, Dinesh Gautam, and AndrasBratincsak for expert technical advice and John Hanover for hishelp with the microscopic observation. We also thank DougMelton for providing the Pdx1-Cre mice. This work wassupported by the Intramural program of the NIH/NIDDK.

References

[1] T. Brelje, D. Scharp, P. Lacy, L. Ogren, F. Talamantes, M. Robertson, H.Friesen, R. Sorenson, Effect of homologous placental lactogens,prolactins, and growth hormones on islet B-cell division and insulinsecretion in rat, mouse, and human islets: implication for placentallactogen regulation of islet function during pregnancy. Endocrinology 132(1993) 879–887.

[2] J. Nielsen, E. Galsgaard, A. Moldrup, B. Friedrichsen, N. Billestrup,

1461J.-Y. Lee et al. / Biochimica et Biophysica Acta 1773 (2007) 1455–1461

J. Hansen, Y. Lee, C. Carlsson, Regulation of beta-cell mass by hormonesand growth factors, Diabetes 50 (2001) S25–S29.

[3] J. Nielsen, C. Svensson, E. Galsgaard, A. Moldrup, N. Billestrup, Beta cellproliferation and growth factors, J. Mol. Med. 77 (1999) 62–66.

[4] E. Galsgaard, F. Gouilleux, B. Groner, P. Serup, J. Nielsen, N. Billestrup,Identification of a growth hormone-responsive STAT5-binding element inthe rat insulin 1 gene, Mol. Endocrinol. 10 (1996) 652–660.

[5] V. Goffin, P. Kelly, The prolactin/growth hormone receptor family:structure/function relationships. J. Mammary Gland Biol. Neoplasia 2(1997) 7–17.

[6] K. Takeda, S. Akira, STAT family of transcription factors in cytokine-mediated biological responses, Cytokine Growth Factor Rev. 11 (2000)199–207.

[7] B. Friedrichsen, E. Galsgaard, J. Nielsen, A. Moldrup, Growth hormone-and prolactin-induced proliferation of insulinoma cells, INS-1, depends onactivation of STAT5 (signal transducer and activator of transcription 5),Mol. Endocrinol. 15 (2001) 136–148.

[8] T. Brelje, J. Parsons, R. Sorenson, Regulation of islet beta-cell proliferationby prolactin in rat islets, Diabetes 43 (1994) 263–273.

[9] T. Brelje, L. Stout, N. Bhagroo, R. Sorenson, Distinctive roles for prolactinand growth hormone in the activation of signal transducer and activator oftranscription 5 in pancreatic islets of Langerhans, Endocrinology 145(2004) 4162–4175.

[10] T. Brelje, A. Svensson, L. Stout, N. Bhagroo, R. Sorenson, Animmunohistochemical approach to monitor the prolactin-induced activa-tion of the JAK2/STAT5 pathway in pancreatic islets of Langerhans, J.Histochem. Cytochem. 50 (2002) 365–383.

[11] L. Stout, A. Svensson, R. Sorenson, Prolactin regulation of islet-derivedINS-1 cells: characteristics and immunocytochemical analysis of STAT5translocation, Endocrinology 138 (1997) 1592–1603.

[12] M. Freemark, I. Avril, D. Fleenor, P. Driscoll, A. Petro, E. Opara, W.Kendall, J. Oden, S. Bridges, N. Binart, B. Breant, P. Kelly, Targeteddeletion of the PRL receptor: effects on islet development, insulinproduction, and glucose tolerance, Endocrinology 143 (2002)1378–1385.

[13] J. Liu, K. Coschigano, K. Robertson, M. Lipsett, Y. Guo, J. Kopchick, U.Kumar, Y. Liu, Disruption of growth hormone receptor gene causesdiminished pancreatic islet size and increased insulin sensitivity in mice,Am. J. Physiol.: Endocrinol. Metab. 287 (2004) E405–E413.

[14] M. Jackerott, A. Moldrup, P. Thams, E.D. Galsgaard, J. Knudsen, Y.C.Lee, J.H. Nielsen, STAT5 activity in pancreatic {beta}-cells influences theseverity of diabetes in animal models of type 1 and 2 diabetes, Diabetes 55(2006) 2705–2712.

[15] Y. Cui, G. Riedlinger, K. Miyoshi, W. Tang, C. Li, C.X. Deng, G.W.Robinson, L. Hennighausen, Inactivation of Stat5 in mouse mammaryepithelium during pregnancy reveals distinct functions in cell proliferation,survival, and differentiation, Mol. Cell. Biol. 24 (2004) 8037–8047.

[16] C. Postic, M. Shiota, K. Niswender, T. Jetton, Y. Chen, J. Moates, K.Shelton, J. Lindner, A. Cherrington, M. Magnuson, Dual roles forglucokinase in glucose homeostasis as determined by liver and pancreaticbeta cell-specific gene knock-outs using Cre recombinase, J. Biol. Chem.274 (1999) 305–315.

[17] M. Gannon, C. Shiota, C. Postic, C. Wright, M. Magnuson, Analysis of theCre-mediated recombination driven by rat insulin promoter in embryonicand adult mouse pancreas, Genesis 26 (2000) 139–142.

[18] E. Lammert, G. Gu, M. McLaughlin, D. Brown, R. Brekken, L.C.Murtaugh, H.-P. Gerber, N. Ferrara, D.A. Melton, Role of VEGF-A invascularization of pancreatic islets, Curr. Biol. 13 (2003) 1070–1074.

[19] M. Inoue, J. Hager, N. Ferrara, H. Gerber, D. Hanahan, VEGF-A has acritical, nonredundant role in angiogenic switching and pancreatic beta cellcarcinogenesis, Cancer Cell 1 (2002) 193–202.

[20] J.-Y. Lee, L. Hennighausen, The transcription factor Stat3 is dispensablefor pancreatic [beta]-cell development and function, Biochem. Biophys.Res. Commun. 334 (2005) 764–768.

[21] J.-Y. Lee, M. Ristow, X. Lin, M.F. White, M.A. Magnuson, L.Hennighausen, RIP-Cre revisited, evidence for impairments of pancreaticbeta-cell function 1074/jbc.M512373200, J. Biol. Chem. 281 (2006)2649–2653.

[22] E. Galsgaard, B. Friedrichsen, J. Nielsen, A. Moldrup, Expression ofdominant-negative STAT5 inhibits growth hormone- and prolactin-induced proliferation of insulin-producing cells, Diabetes 50 (2001)S40–S41.

[23] B. Friedrichsen, H. Richter, J. Hansen, C. Rhodes, J. Nielsen, N.Billestrup, A. Moldrup, Signal transducer and activator of transcription 5activation is sufficient to drive transcriptional induction of cyclin D2 geneand proliferation of rat pancreatic beta-cells. Mol. Endocrinol. 17 (2003)945–958.

[24] R. Kulkarni, J. Bruning, J. Winnay, C. Postic, M. Magnuson, C. Kahn,Tissue-specific knockout of the insulin receptor in pancreatic beta cellscreates an insulin secretory defect similar to that in type 2 diabetes, Cell 96(1999) 329–339.

[25] L. Héron-Milhavet, M. Haluzik, S. Yakar, O. Gavrilova, S. Pack, W. Jou,A. Ibrahimi, H. Kim, D. Hunt, D. Yau, Z. Asghar, J. Joseph, M. Wheeler,A. NA, D. LeRoith, Muscle-specific overexpression of CD36 reverses theinsulin resistance and diabetes of MKR mice, Endocrinology 145 (2004)4667–4676.