Ronadip R. Banerjee,1,2 Holly A. Cyphert,3 Emily M. Walker,3 Harini Chakravarthy,1

Heshan Peiris,1 Xueying Gu,1 Yinghua Liu,1 Elizabeth Conrad,3 Lisa Goodrich,4

Roland W. Stein,3 and Seung K. Kim1,5,6

Gestational Diabetes Mellitus FromInactivation of Prolactin Receptor andMafB in Islet b-CellsDiabetes 2016;65:2331–2341 | DOI: 10.2337/db15-1527

b-Cell proliferation and expansion during pregnancyare crucial for maintaining euglycemia in response to in-creased metabolic demands placed on the mother. Pro-lactin and placental lactogen signal through the prolactinreceptor (PRLR) and contribute to adaptive b-cell re-sponses in pregnancy; however, the in vivo requirementfor PRLR signaling specifically in maternal b-cell adapta-tions remains unknown. We generated a floxed allele ofPrlr, allowing conditional loss of PRLR in b-cells. In thisstudy, we show that loss of PRLR signaling in b-cells re-sults in gestational diabetes mellitus (GDM), reduced b-cellproliferation, and failure to expand b-cell mass during preg-nancy. Targeted PRLR loss in maternal b-cells in vivo im-paired expression of the transcription factor Foxm1, bothG1/S and G2/M cyclins, tryptophan hydroxylase 1 (Tph1),and islet serotonin production, for which synthesis requiresTph1. This conditional system also revealed that PRLR sig-naling is required for the transient gestational expression ofthe transcription factor MafB within a subset of b-cellsduring pregnancy. MafB deletion in maternal b-cells alsoproduced GDM, with inadequate b-cell expansion accom-panied by failure to induce PRLR-dependent target genesregulating b-cell proliferation. These results unveil molec-ular roles for PRLR signaling in orchestrating the physio-logic expansion of maternal b-cells during pregnancy.

Pregnancy is a unique acquired physiologic state of increasedmetabolic demand, requiring increased output of insulin byislet b-cells. This adaptive response is accomplished both

through enhanced insulin secretion and by b-cell prolifera-tion and expansion in mice (1). Failure of adaptive b-cellexpansion underlies dysregulated glucose homeostasis andprogression to diabetes mellitus (2). Thus, the mechanismsunderlying gestational b-cell proliferation are a focus of in-tensive ongoing investigation.

Pregnancy hormones are known regulators of b-cellgrowth and function (3). The lactogenic hormones prolactinand placental lactogen signal through the prolactin receptor(PRLR) and are crucial regulators of pregnancy adaptation inmany maternal tissues (4). PRLR is expressed in both rodentand human pancreatic b-cells (5), and in vitro treatment ofislets with prolactin has established it as a potent b-cell mi-togen in both species (1). Although gene expression studies ofislets during pregnancy identified strong induction of someprolactin signaling targets (6–8), the mechanisms underlyinglactogen-stimulated changes in b-cells during pregnancy areincompletely understood. During transient b-cell proliferationand expansion during gestation, prior studies have reportedincreased expression of nuclear factors like FoxM1, the cyclin-dependent kinases cyclin A2 and cyclin B1, andMafB (6,9,10).However, the requirement for PRLR signaling to induce ex-pression of these factors and the physiologic significance ofthe gestational MafB+ b-cell subpopulation are unknown.

Studies of b-cells during pregnancy in humans are con-founded by practical and ethical challenges. Thus, animalstudies remain critical for understanding b-cell biology dur-ing pregnancy (11). Mouse genetic studies of PRLR supporta role in b-cell development and function; the global PRLR

1Department of Developmental Biology, Stanford University School of Medicine,Stanford, CA2Division of Endocrinology, Gerontology and Metabolism, Department of Medi-cine, Stanford University School of Medicine, Stanford, CA3Department of Molecular Physiology and Biophysics, Vanderbilt University,Nashville, TN4Department of Neurobiology, Harvard Medical School, Boston, MA5Division of Oncology, Department of Medicine, Stanford University School ofMedicine, Stanford, CA6Howard Hughes Medical Institute, Stanford University School of Medicine, Stan-ford, CA

Corresponding author: Seung K. Kim, seungkim@stanford.edu.

Received 4 November 2015 and accepted 11 May 2016.

This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db15-1527/-/DC1.

© 2016 by the American Diabetes Association. Readers may use this article aslong as the work is properly cited, the use is educational and not for profit, andthe work is not altered.

Diabetes Volume 65, August 2016 2331

ISLETSTUDIES

knockout has glucose intolerance and reduced b-cell mass(12). Unfortunately, the global Prlr knockout mouse is ster-ile, precluding pregnancy studies (13). Prlr+/2 mice studiedduring pregnancy develop glucose intolerance and reductionin b-cell proliferation and mass expansion (14). Neverthe-less, as loss of PRLR results in multiple abnormalities inother metabolic tissues that could indirectly influence b-cellfunction (15), it is essential to assess the consequences oftargeted PRLR inactivation specifically in b-cells. In thisstudy, we generated a conditional Prlr allele allowing Crerecombinase–mediated genetic ablation of PRLR signalingand identified a requirement for PRLR in molecular, hor-monal, and proliferative adaptations by maternal b-cellsin pregnancy. Collectively, our results suggest PRLR sig-naling is a master regulator of adaptive b-cell responsesduring pregnancy.

RESEARCH DESIGN AND METHODS

Creation of the Floxed Prlr Allele, bPRLRKO, andbMafBKO MiceA targeting vector containing Prlr genomic DNA encompass-ing exons 4 through 9 were subcloned into plasmid PL253containing a thymidine kinase cassette. Using recombineer-ing, loxP and FRT-neo-FRT-loxP cassettes were placed flank-ing exon 5. The targeting vector was electroporated intoC57BL/6J embryonic stem cells and clones selected us-ing G418; validated clones were injected into 129 blastocystsgenerating chimeric males (Stanford Transgenic Core). Germ-line transmission of Prlrf allele was identified by brown furpups after crossing male chimeras with C57BL/6J females.Mouse tail genomic DNA was digested with NheI and South-ern blotting performed with a 59 external probe to confirm acorrectly targeted allele. The FRT-flanked neo cassette wasremoved by crosses with FLPeR mice (The Jackson Labora-tory, Bar Harbor, ME), generating floxed PRLR mice (Prlrf/+),which were backcrossed with C57BL/6J mice (The JacksonLaboratory) for more than eight generations. Prlrf/+ micewere crossed with mice with a transgene encoding Crerecombinase from rat insulin promoter elements (RIP-Cre)(16). Subsequently, RIP-Cre;Prlrf/+ males were crossed withPrlrf/+ females, generating bPRLRKO (PRLRf/f;RIP-Cre) andlittermate Prlrf/+, Prlrf/f, RIP-Cre;Prlrf/+, and RIP-Cre;Prlr+/+

controls. MafBf/f mice (17) were crossed with RIP-CreM

mice (18) and backcrossed to B6J mice at least six gener-ations (see Supplementary Table 1: genotyping primers).

Mouse Husbandry, Breeding, and ExperimentationMice were weaned 21–25 days after parturition. To avoidpotential effects of altered maternal metabolism on off-spring during in utero exposure, bPRLRKO females weresolely used as experimental mice. Beginning at 8 weeks ofage, bPRLRKO and control females were mated with wild-type FVB males. Vaginal plugs were scored at gestationalday (GD) 0.5 and males removed. All experiments wererepeated in at least two independent cohorts of mice. Allprocedures involving mice were approved and conductedin accordance with the Stanford Administrative Panel on

Laboratory Animal Care or Vanderbilt Animal Care andUse Program.

Tolerance Testing and Serum Metabolite AnalysisIntraperitoneal glucose and insulin tolerance testing wereperformed as previously described (19). An oral glucosetolerance test (GTT) was performed using a 22-gauge rigidgavage needle to deliver an intragastric glucose bolus of2 g/kg body weight. Blood was collected by tail vein bleed-ing. Blood glucose levels were determined by glucometer(Bayer Contour; Bayer). Ad libitum fed or overnight (16 h)–fasted blood glucose levels were measured at 8:30 A.M. Se-rum insulin levels were performed by ELISA (Crystal Chem)following the manufacturer’s directions.

Islet Isolation and CultureIslets were isolated using retrograde perfusion of thepancreatic duct with collagenase, purified using densitycentrifugation, and cultured as previously described (20).Recombinant mouse prolactin (R&D Systems) was dilutedin culture media to a final concentration of 500 ng/ml.Culture media was changed daily.

ImagingMicroscopy was performed on a Zeiss AxioM1 Fluores-cence microscope with AxioVision software (Carl Zeiss).Confocal images were obtained using a Leica Sp2 micro-scope (Beckman Cell Sciences Imaging Facility; LeicaMicrosystems).

Quantifying b-Cell Mass and ProliferationAfter weighing, the pancreas was fixed, embedded inoptimal cutting temperature compound, and then frozen.Tissues were sectioned at 10-mm intervals using a Leica3050S Cryostat (Leica Microsystems). We immunostainedtissue with anti-insulin antibody and DAPI and then an-alyzed sections separated by 200 mm. b-Cell mass wasmeasured using ImageJ (National Institutes of Health)to quantify insulin+ and total pancreas area in each section;b-cell area was the average of nine measured sections. Thecalculated b-cell mass = (b-cell area/total pancreas area) 3(total pancreas weight). To measure proliferation, weimmunostained tissues using antibodies against insulinand Ki-67 and the proportion of Ki-67+ b-cells calculatedas a percentage. At least 1,500 b-cells were counted foreach animal; only clusters containing at least 20 b-cellswere counted. The total number of islets per mouse wasthe sum of clusters counted in nine analyzed sections.Antibodies used are listed in Supplementary Table 2.

Quantitative PCRTotal RNA was isolated using TRIzol (Invitrogen) andreverse transcribed using the Ambion Retroscript Kit(Ambion) according to the manufacturer’s instructions.Quantitative RT-PCR (qPCRs) were performed withFastStart Universal Probe Master (Roche) and TaqManassays (see Supplementary Table 3) on an ABI 7500 RealTimePCR System (Applied Biosystems). Assays were performedin technical replicates and normalized to mouse actin as areference standard.

2332 b-Cell PRLR KO and Gestational Diabetes Mellitus Diabetes Volume 65, August 2016

Western BlottingIslets were lysed in 23 Bio-Rad loading buffer (Bio-Rad)containing 10% BME and then boiled for 5 min. Followingelectrophoresis on Bio-Rad TGX 4–20% gels (Bio-Rad),proteins were transferred onto nitrocellulose using thetank apparatus (Bio-Rad). Chemiluminescent detection wasperformed with SuperSignal West Pico (Thermo) and Care-stream Kodak BioMax XAR film (Sigma-Aldrich).

Statistical AnalysisData are presented as means 6 SEM. Statistical significancewas determined by unpaired two-tailed Student t test,repeated-measures ANOVA (RM-ANOVA), or x2 test as in-dicated using GraphPad Prism 6 (GraphPad Software, LaJolla,CA) or Microsoft Excel (Microsoft, Redmond, WA).

RESULTS

Conditional Prlr Inactivation in Islet b-CellsTo permit targeted PRLR signaling disruption, we gener-ated mice harboring a Cre recombinase-sensitive Prlrf al-lele (Fig. 1A and B) (see RESEARCH DESIGN AND METHODS).Intercrosses of Prlrf/+ mice generated Prlrf/f mice and sib-lings with expected genotypes in Mendelian ratio (Fig. 1Cand data not shown). To generate mice lacking PRLR inpancreatic b-cells (abbreviated bPRLRKO mice), we bredRIP-Cre mice to Prlrf mice (see RESEARCH DESIGN ANDMETHODS).qPCR revealed that Prlr mRNA levels were reduced .95%in islets from bPRLRKO females as compared with littermatePrlrf/f controls (Fig. 1D). Immunohistology and Western blot-ting demonstrated that b-cell PRLR protein was reduced oralmost undetectable in virgin or pregnant female bPRLRKOmice (Fig. 1E and F). Conception, nursing, and rearing of pupsby bPRLRKO mothers was similar to that of controls (seeGESTATIONAL DIABETES MELLITUS IN bPRLRKO MICE). Thus, the Prlrf/f

mice permitted conditional genetic inactivation of Prlr in vivo.

Gestational Diabetes Mellitus in bPRLRKO MiceWe next investigated the consequences of disrupting b-cellPRLR signaling on glucose homeostasis. Adult virgin femaleand male bPRLRKO mice did not show alterations in glu-cose tolerance, weight, or b-cell mass, in contrast to defectsobserved in global Prlr knockout mice (Fig. 2A–C) (data notshown). Ad libitum fed glucose and insulin levels, insulintolerance, and total pancreas mass also did not differ in2- to 3-month-old virgin bPRLRKO females compared withage-matched controls (Supplementary Fig. 1A–D), the agewhen all subsequent pregnancy studies were performed.

By contrast, we observed pronounced defects of glucoseregulation in pregnant bPRLRKO females. Maternal in-sulin demand peaks late in pregnancy, ;GD16.5 (10).On GD16.5, we observed marked glucose intolerance inbPRLRKO females (Fig. 2D) compared with control females(Prlrf/+ and Prlrf/f). We also observed glucose intolerance inRIP-Cre;Prlrf/+ mice (Fig. 2D). Blood glucose levels during adlibitum feeding were also elevated in GD16.5 bPRLRKOmice (Fig. 2E). Expression of Prlr mRNA in RIP-Cre;Prlrf/+

mice was intermediate between bPRLRKO and control fe-males (Supplementary Fig. 2A). However, weight gain in

pregnancy and litter sizes of bPRLRKO mothers were in-distinguishable from that of controls (Supplementary Fig.2B and C). Following parturition, bPRLRKO mothers lac-tated normally and nursed pups to weaning without impacton offspring survival rates (data not shown). Subsequentpostpartum analyses showed that glucose intolerance andad libitum glucose abnormalities in pregnant bPRLRKOmice resolved (Fig. 2F and G), a characteristic feature ofgestational diabetes mellitus (GDM). Because some mousestrains harboring a RIP-Cre transgene develop glucose ho-meostasis abnormalities (21), we also evaluated RIP-Cre lit-termate controls but did not observe any differences inglucose tolerance in virgin, pregnant, or postpartum micecompared with controls (Fig. 2D, F, and H). Thus, bPRLRKOmice developed GDM.

Because GDM increases the risk of dysglycemia in futurepregnancies, we investigated the effects of additional preg-nancy on glucose regulation in bPRLRKO multigravida. Dur-ing a second gestation in bPRLRKO females, glucoseintolerance (Fig. 2H) and hyperglycemia in ad libitum feedingrecurred (Fig. 2I). bPRLRKO females also developed evidenceof worsened glucose control, including glycemia.200 mg/dLupon completion of the GTT (Fig. 2H) and fasting hypergly-cemia (Fig. 2I), features not observed in the first pregnancy(Fig. 2D). Thus, we observed that glucose control worsenedwith additional pregnancies in bPRLRKO mice, similar tohumans with multigravidity or GDM (22).

Failure to Proliferate and Expand b-Cells in PregnantbPRLRKO MiceTo identify the basis for GDM in bPRLRKO mothers, weassessed insulin output and b-cell growth, features knownto undergo adaptive enhancement in pregnancy (1). Wefound reduced serum insulin levels in bPRLRKO mice dur-ing GTT in late gestation (GD16.5), consistent with relativeinsulin deficiency (Fig. 3A and B). We sought to understandthe basis for this relative insulin deficiency by examininginsulin secretion in islets isolated from mice at GD16.5.Islets from bPRLRKO or control mice showed similar basaland glucose-stimulated insulin secretion (GSIS), normalizedto insulin content (Fig. 3C), indicating the insulin de-ficiency observed in vivo was not a consequence of reducedGSIS. Insulin tolerance was also unchanged in bPRLRKOmothers at this gestational stage, excluding altered pe-ripheral insulin sensitivity (Supplementary Fig. 3A). Next,we used the cellular marker Ki-67 to quantify b-cell pro-liferation and observed a threefold reduction of b-cell label-ing in islets of GD16.5 bPRLRKO mice compared withcontrols (Fig. 3D and E). No change in total islet numberwas observed (Supplementary Fig. 3C).

Consistent with the observed reduction of Ki-67+ b-cellsin pregnant bPRLRKO mice, b-cell area and mass werereduced approximately twofold compared with pregnantPrlrf/f and RIP-Cre controls (Fig. 3F and G and Supplemen-tary Fig. 3C and D). Compared with nonpregnant bPRLRKOfemales (Fig. 2C) (0.89 6 0.06 mg), pregnant bPRLRKOfemales (Fig. 3G) (1.91 6 0.16 mg) exhibited a significant

diabetes.diabetesjournals.org Banerjee and Associates 2333

increase of b-cell mass during pregnancy (P = 0.004), sug-gesting PRLR-independent mechanisms of gestationalmass expansion are active yet insufficient to compensatefor the loss of PRLR signaling. Thus, loss of PRLR led toreduced b-cell proliferation and mass expansion in preg-nant bPRLRKO.

Disrupted b-Cell Cycle Gene Expression and SerotoninSignaling in bPRLRKO MiceTo investigate the molecular basis of reduced gestationalb-cell proliferation in bPRLRKO mice, we measured gene

expression. Gestational induction of known PRLR signal-ing targets like Prlr itself and cytokine-inducible Src ho-mology 2–containing protein (Cish), a negative regulatorof PRLR signaling (8), was markedly reduced as expected(Fig. 4A). Tryptophan hydroxylase 1 (Tph1) is highly in-duced by pregnancy in b-cells (6,23) and catalyzes therate-limiting step of islet serotonin synthesis. In turn,serotonin contributes to b-cell proliferation through auto-crine and paracrine effects (6). In bPRLRKO pregnantmice, islet Tph1 induction was eliminated (Fig. 4B), andimmunoreactive serotonin in islets was absent during

Figure 1—Conditional inactivation of PRLR signaling in pancreatic b-cells (bPRLRKO). A: Schematic diagram (not to scale) depicting thePrlr gene locus, targeting vector sequence, and Prlrf allele structure (see RESEARCH DESIGN AND METHODS). B: Southern blot of genomic DNAfrom mice demonstrating germline transmission of Prlrf. Introduction of a NheI site downstream of the loxP cassette resulted in a 9-kb(targeted allele) rather than a 16-kb fragment (wild-type [WT] allele). C: Genotyping PCR of a litter depicting representative progeny of allpotential Prlr genotypes following a cross of Prlrf/+ mice. D: mRNA levels of Prlr in bPRLRKO females (RIP-Cre;Prlrf/f) and control littermates(Prlrf/f). n = 6 mice/group. E: Immunofluorescence images of islets from bPRLRKO female and control mice stained for PRLR and insulin(scale bar, 100 mm). F: Western blot of PRLR expression from islets isolated from virgin bPRLRKO female and controls. Actin was used as aloading control. **P # 0.01 by t test.

2334 b-Cell PRLR KO and Gestational Diabetes Mellitus Diabetes Volume 65, August 2016

gestation (Fig. 4C). Thus, b-cell PRLR is required for isletserotonin production during pregnancy. Prior work sug-gests that lactogens alter expression of genes encodingb-cell cyclins, cyclin-dependent kinase inhibitors, andtranscription factors Foxm1 and Foxd3 (4,10,24–26). InbPRLRKO pregnant females, qPCR with isolated islets atGD16.5 revealed reductions in Ccna2, Ccnb1, Ccnb2, andCcnd1 (which, respectively, encode cyclins A2, B1, B2, andD1), and Foxm1, all previously implicated as regulators of

gestational b-cell proliferation (6,8,10) (Fig. 4D). By con-trast, we found that expression of Ccnd2 (Fig. 4D), Men1,Cdkn2c (p18), Cdkn1b (p27), and Foxd3 (SupplementaryFig. 4A), which prior studies indicate are regulated by lac-togen signaling in pregnancy (24–27), were unaffected inbPRLRKO islets at GD16.5. Moreover, reduced Foxd3 ex-pression during gestation (24) still occurred in bPRLRKOfemales, indicating PRLR signaling is not required for Foxd3downregulation in pregnancy (Supplementary Fig. 4A).

Figure 2—bPRLRKO females have GDM. Virgin bPRLRKO females and control littermates were followed through pregnancy, postpartum,and a second pregnancy. A: GTT. B: Body weights. C: b-Cell mass in virgin females. D–I: Glucose tolerance with corresponding area underthe curve (AUC) for each genotype and ad libitum fed and overnight fasted glucose for bPRLRKO females and controls. D and E: Firstpregnancy. F and G: Postpartum. H and I: Second pregnancy. n = 5–8 mice/group. *P < 0.05, **P # 0.01, ***P # 0.001 for RM-ANOVA(GTT) or t test (blood glucose).

diabetes.diabetesjournals.org Banerjee and Associates 2335

Collectively, our in vivo targeted genetic studies demon-strate that PRLR signaling is required for serotonin sig-naling and gene expression changes controlling adaptivematernal b-cell expansion.

To assess the formal possibility that GDM might indi-rectly alter gene expression in pregnant bPRLRKOmice, we measured the response of isolated bPRLRKOislets to lactogen treatment in vitro. PRLR activates sev-eral intracellular pathways, including Jak2/Stat5, a majormediator of its signal transduction (28). As expected,

purified mouse prolactin robustly induced phosphoryla-tion of Stat5 in control islets (Supplementary Fig. 4B).By contrast, phospho-Stat5 was reduced in islets fromRIP-Cre;Prlrf/+ mice and absent in bPRLRKO islets treatedwith prolactin (Supplementary Fig. 4B). We next exam-ined gene expression in cultured islets treated with pro-lactin for 24 h in concentrations comparable to peaklactogen levels in pregnancy (29). In controls, PRLR targetgenes, including Prlr, Cish, and Tph1, were induced tolevels similar to those at GD16.5; this induction was

Figure 3—bPRLRKO females have reduced b-cell proliferation and mass during pregnancy. Serum insulin levels during GTT (A) andcorresponding area under the curve (AUC) measurements (B) in GD16.5 pregnant bPRLRKO females and controls (Prlrf/f). n = 4 mice/group.**P# 0.01 by t test. C: GSIS on islets isolated from GD16.5 bPRLRKO females (n = 4) and controls (n = 6). D: Representative islet images ofGD16.5 pancreas sections from bPRLRKO females and controls with Ki-67 immunofluorescence, insulin, and DAPI (scale bar, 100 mm).E: The percentage of b-cells with Ki-67+ nuclei staining in nonpregnant (N) and pregnant (P) (GD16.5) bPRLRKO females and controls.F: Representative images of pancreas sections from GD16.5 bPRLRKO females and controls used for b-cell area. Insulin immunofluores-cence (green) and DAPI (blue) (scale bar, 2 mm). G: Calculated b-cell mass in GD16.5 bPRLRKO, RIP-Cre, Prlr+/+, and control (Prlr f/f)females. n = 3 or 4 mice/group. *P < 0.05 by t test.

2336 b-Cell PRLR KO and Gestational Diabetes Mellitus Diabetes Volume 65, August 2016

eliminated in bPRLRKO islets (Supplementary Fig. 4C).FoxM1 expression was reported to increase in whole isletsexposed in vitro to placental lactogen (10), but we did notdetect consistent FoxM1 induction after treatment of is-lets with lactogens up to 4 days in culture (data notshown). These in vitro data are consistent with gene ex-pression changes of bPRLRKO islets observed in vivo andsuggest that altered expression of PRLR target genes re-flects disrupted intrinsic islet signaling, not secondaryeffects of altered metabolism in GDM.

MafB Expression in b-Cells During PregnancyRequires PRLR SignalingIn pancreatic islets from adult mice, the transcriptionfactor MafB is exclusively expressed in a-cells (30), exceptduring pregnancy when we detected transient expressionin a subset of b-cells (9). This contrasts with MafB ex-pression within both a-cells and b-cells during develop-ment (30,31). The physiologic significance of MafBexpression in b-cells during pregnancy was unknown.We observed that up to 25% of maternal insulin+ (abbre-viated Ins+) cells produced MafB by GD16.5 (Supplemen-tary Fig. 5A). MafB+ b-cells diminished by postpartum day

7 and were undetectable by postpartum day 28 (Supple-mentary Fig. 5B). The maternal MafB+Ins+ cells at GC16.5also produced Pdx1, a key transcription factor and regu-lator of b-cells (Supplementary Fig. 5C). MafB expressionin adult a-cells promotes glucagon production, but we didnot observe bihormonal Ins+Glucagon+ cells in maternalislets (data not shown). Strikingly, we found that MafB+

b-cells were absent from maternal bPRLRKO islets atGD16.5 or other gestational stages (Fig. 4E and data notshown). Thus, b-cell MafB expression in pregnancy re-quires PRLR signaling.

MafB Loss Impairs Gestational b-Cell Expansion andPromotes GDMTo identify physiologic roles for MafB in maternal b-cells,we generated a b-cell–specific deletion of MafB (bMafBKO)(see RESEARCH DESIGN AND METHODS). Nonpregnant bMafBKOmice had normal glucose tolerance (Fig. 5A), but showedglucose intolerance during pregnancy compared with controlMafBf/f mice (Fig. 5B). This glucose intolerance was similarto that in bPRLRKO mice (Fig. 2). Additionally, both b-cellproliferation (Fig. 5C) and mass (Fig. 5D) were reduced inbMafBKO mice during pregnancy but not in virgin females.

Figure 4—Altered cell cycle regulators and serotonin signaling in bPRLRKO islets during pregnancy. mRNA levels of Prlr and Cish (A) andTph1 (B) in nonpregnant (N) or pregnant (P) bPRLRKO females (black bars) and controls (open bars). C: Immunofluorescence of islet serotonin,insulin, and DAPI (scale bar, 50 mm). D: Relative mRNA levels of Ccna2, Ccnb1, Ccnb2, Ccnd1, Ccnd2, and Foxm1 in bPRLRKO females andcontrols at GD16.5. E: Confocal immunofluorescence images of islets from GD16.5 bPRLRKO females and controls (original magnifica-tion 340). MafB (red), insulin (green), and DAPI (blue) with MafB+Ins+ cells highlighted (white arrows). For mRNA studies, pregnant sampleswere from GD16.5 and expression levels normalized to nonpregnant controls = 1. n = 3–8 mice/group. *P < 0.05, **P # 0.01 by t test.

diabetes.diabetesjournals.org Banerjee and Associates 2337

Like bPRLRKO mice, pregnant bMafBKO mice failedto induce Foxm1, Ccna2, or Ccnb1 in pregnancy (Fig. 5E).Although levels of Ccnb2 mRNA increased with pregnancyin bMafBKO females, this induction was blunted com-pared with pregnant controls (Fig. 5E). The RIP-CreM alleleused to generate bMafBKO mice also encodes humangrowth hormone that can activate PRLR signaling (32).Consistent with these prior findings, we observed signif-icant elevations of PRLR signaling targets Cish and Tph1in nonpregnant female RIP-CreM mice (Supplementary Fig.5D). By contrast, Prlr expression in nonpregnant bMafBKOwas similar to controls and was significantly induced during

pregnancy. However, the induction was blunted comparedwith pregnant controls, including RIP-CreM pregnant fe-males (Supplementary Fig. 5D). Despite this apparentpartial activation of PRLR signaling, female bMafBKOmice did not show elevated Foxm1 or Ccna2 levels atbaseline, and induction of both genes during pregnancywas blunted compared with pregnant RIP-CreM controls(Fig. 5E and Supplementary Fig. 5E). Moreover, despiteelevated Tph1 expression, bMafBKO islets showed clearreduction of serotonin compared with control mice atGD16.5 (Supplementary Fig. 5F). Thus, maternal isletserotonin production requires MafB expression in

Figure 5—bMafBKO mice have GDM, reduced b-cell proliferation, and gene expression changes that overlap with bPRLRKO mice. GTT inbMafBKO females (MafBf/f;RIP-CreM) and controls (MafBf/f) as virgins (A) and during pregnancy (B) at GD15.5. The percentage of Ki-67+ nucleiwithin b-cells (Ki-67+Ins+/total Ins+ cells) (C) and b-cell mass (mg) in virgin and GD15.5 bMafBKO and controls (D). E: Relative mRNA levels ofFoxM1, Ccna2, Ccnb1, and Ccnb2 in nonpregnant (N) or GD15.5 (P) bMafBKO females and controls. Expression levels were normalized to Ncontrols = 1. n = 3–8 mice/group. *P < 0.05, **P # 0.01 for RM-ANOVA (GTT) or t test.

2338 b-Cell PRLR KO and Gestational Diabetes Mellitus Diabetes Volume 65, August 2016

gestation. Together with our finding that MafB inductionfails in bPRLRKO mothers, these data indicate that MafBis regulated by PRLR signaling and required for b-celladaptations during pregnancy.

DISCUSSION

Genetic and cell biology studies have suggested thatPRLR signaling is a crucial regulator of b-cell expansion(12,14,29,33), but prior loss-of-function studies were per-formed in global PRLR knockouts rather than examiningPRLR loss specifically in b-cells. Our studies reveal a clearrequirement for PRLR in b-cell function in vivo. A priorstudy of conventional PRLR knockout mice identified glu-cose intolerance, reduced body weight, and reduced b-cellmass in adult mice of both sexes (12), but it was unclear ifthese reflected islet or nonislet effects. By contrast, weobserved normal glucose homeostasis and b-cell mass inadult prepartum bPRLRKO mice and no alteration in adultbody weight. Our results suggest that defective PRLRsignaling in nonislet tissues likely contributes to bodyweight and glucose homeostasis abnormalities previouslyobserved in nonpregnant mice with global PRLR knock-out. Some of these differences may also be related togenetic background differences between 129Sv and B6mouse strains, known to affect PRLR-dependent pheno-types (34). Additional studies have indicated a role forPRLR signaling in fetal and early postnatal islet develop-ment (35). RIP-Cre expression commences in the fetalpancreas and has been used by us and others to inactivateb-cell genes in the fetal and neonatal period (20). Thus,bPRLRKO mice might be predicted to display defectivedevelopmental expansion of b-cells; however, virgin adultbPRLRKO mice displayed no alteration in b-cell mass.Further studies are required to assess the role of PRLRin embryonic or perinatal b-cell development.

PRLR signaling is thought to involve transductionthrough multiple effectors, including signal transducerand activator of transcription 5 (Stat5). However, micecarrying a b-cell–specific deletion of Stat5 did not developGDM (36). Thus, although Stat5 is considered a principalmediator of PRLR signaling, our results suggest that othersignal transduction pathways such as mitogen-activatedprotein kinase/extracellular signal-regulated kinase orphosphatidylinositol 3-kinase/Akt (4) are critical for ges-tational b-cell adaptation. Supporting this model, recentstudies have demonstrated that prolactin induction ofTph1 is coordinated by activation of all three pathways(37). b-Cell–specific deletion of Tph1 has been reported,but gestational studies were not performed (38). Micewith pancreas-specific deletion of Foxm1 display profoundGDM, but also exhibit prepregnancy defects in b-cell massthat likely reflect Foxm1 roles in physiological b-cell pro-liferation (10,39). Collectively, these prior studies supportthe view that targets of PRLR signaling regulate gesta-tional b-cell expansion.

The significance of MafB+ b-cells during pregnancy waspreviously unknown. Our studies demonstrate intact PRLR

signaling is required for the induction of MafB in a subsetof b-cells during pregnancy. In this study, we report thatMafB expression is necessary for proliferation of b-cells inpregnancy. Moreover, MafB was required for pregnancy-induced islet serotonin production and Foxm1 upregulation.Prior studies identified roles for MafB in islet developmentand postnatal expansion, in which it regulates terminaldifferentiation and maturation (17,31); a role for MafBin proliferation has also been established in pathologicalsettings (40). Our studies reveal a role in physiologic pro-liferation of gestation, in which MafB+ b-cells maintaintheir hallmark features, including production of insulin andPdx1 and suppression of glucagon. This contrasts withre-expression of MafB in a-like b-cells observed afterPdx1 or Foxo1 loss, in which b-cell dedifferentiation con-tributes to diabetes pathogenesis (41). Thus, in normalpregnancy, transient expression of MafB by Ins+ cellsdoes not appear to reflect dedifferentiation. Future stud-ies will be necessary to identify the specific mechanismsby which MafB, Foxm1, and Tph1 coordinate and controlproliferation and mass expansion during gestation.

We and others have previously described additionalputative PRLR signaling targets, includingMen1, p18, p27,and Foxd3 (14,25,26,42). For example, we found prolactinwas sufficient to induce Men1 gene expression changes inislets and that Men1 induction was sufficient to impairmaternal b-cell expansion (25). However, in bPRLRKOpregnant mice, we did not detect altered expression ofMen1, p18, or p27, suggesting PRLR signaling is not nec-essary for regulating these factors. These findings mayreflect differences intrinsic to gain- and loss-of-functionmodels, in vitro and in vivo experimentation, mousestrain differences, or yet-unidentified modifying factorsthat contribute to the regulation of these genes duringpregnancy. Pregnancy is a complex, transient, polyhormo-nal state, and our findings highlight the importance ofin vivo loss-of-function models for genetic analysis ofputative target genes.

Brouwers et al. (32) report of human growth hormoneminigene effects on islets in mice harboring transgeneslike RIP-CreM highlights the importance of using RIP-CreM

control littermates in physiologic studies. The bPRLRKOmice lack the very receptor believed to mediate the aug-mentation of b-cell mass by human growth hormone (32).Thus, the bPRLRKO model should be resistant to possibleconfounding changes caused by the RIP-Cre transgene.

Prior studies suggest that human b-cells expand inpregnancy, but the molecular and cellular basis of this isnot clear (43,44). For example, Butler et al. (43) havesuggested that this expansion may reflect b-cell neogen-esis rather than proliferation, a possibility consistent withthe recent finding of maternal b-cell neogenesis in ro-dents (10,45). A recent study identified single nucleotidepolymorphisms in the 59 untranslated region of PRLRassociated with GDM risk in humans (46). Intriguingly,induction of TPH1 and islet serotonin also occurs in hu-man pregnancy (6), but it is not known whether this

diabetes.diabetesjournals.org Banerjee and Associates 2339

reflects increased b-cell PRLR signaling. However, PRLRexpression is much lower in human than rodent b-cells(47), and MafB expression, although restricted to a-cellsin adult rodents, is also present in b-cells from a nonpreg-nant human (48). Other correlative studies have implicateda more general role for PRLR signaling in human glucosecontrol and metabolism. Higher prolactin levels are associ-ated with a reduced risk of glucose intolerance and diabetesin both men and women (49). A human PRLR disease-associated mutation causing familial hyperprolactinemia(50) was recently described, although information on dia-betes or glucose homeostasis in this family was not report-ed. Thus, further studies are needed to assess the role ofPRLR signaling in human b-cell biology in both gestationaland nongestational settings.

In summary, we produced and characterized micelacking PRLR in b-cells during gestation and found thesemice developed hallmark features of GDM. These studiesestablish the in vivo requirement of PRLR in b-cells forserotonin production and for modulating expression ofgenes critical for b-cell proliferation during pregnancyand identify a novel role for MafB in gestational prolifer-ation. PRLR signaling is thought to modulate the devel-opment and function of many tissues (5), including bone,immunity, adipose metabolism, and breast and prostatecancer pathogenesis (15). Thus, the conditional Prlrf alleledeveloped in this study will be useful to examine PRLRsignaling in other settings relevant to human health.

Acknowledgments. The authors thank members of the Kim laboratoryand Drs. Hail Kim (Korea Advanced Institute of Science and Technology), JustinAnnes (Stanford University School of Medicine), and Mitchell Lazar (PerelmanSchool of Medicine, University of Pennsylvania) for helpful discussions andadvice. Kartik Viswanathan helped with initial stages of the project, and Dr. PeiWang (School of Medicine, UT Health Sciences Center, San Antonio, TX) providedhelp with design and construction of the Prlrf allele.Funding. R.R.B. is supported by a National Institute of Diabetes and Digestiveand Kidney Diseases Mentored Clinical Scientist Research Career DevelopmentAward (K-08 DK091359). H.A.C. is supported by National Institutes of Health(NIH) National Research Service Award 1F-32 DK-102283. H.C. is supported by aJDRF postdoctoral award. H.P. is supported by American Diabetes Associationgrant 1-16-PDF-086. Work in the laboratory of R.W.S. is supported by NationalInstitutes of Health/National Institute of Diabetes and Digestive and KidneyDiseases grants DK-090570, DK-089572, and DK-050203. The laboratory ofS.K.K. is supported by the Snyder Foundation and grants from the NIH, JDRF, andHelmsley Trust. S.K.K. is a Howard Hughes Medical Institute Investigator.Duality of Interest. No potential conflicts of interest relevant to this articlewere reported.Author Contributions. R.R.B., R.W.S., and S.K.K. conceived and designedthe experiments and wrote the manuscript. R.R.B., H.A.C., E.M.W., H.C., H.P., X.G.,Y.L., and E.C. performed experiments and analyzed data. L.G. provided MafB floxedmice. All authors reviewed and edited the manuscript. S.K.K. is the guarantor of thiswork and, as such, had full access to all the data in the study and takesresponsibility for the integrity of the data and the accuracy of the data analysis.

References1. Sorenson RL, Brelje TC. Adaptation of islets of Langerhans to pregnancy:beta-cell growth, enhanced insulin secretion and the role of lactogenic hormones.Horm Metab Res 1997;29:301–307

2. Halban PA, Polonsky KS, Bowden DW, et al. b-cell failure in type 2 diabetes:postulated mechanisms and prospects for prevention and treatment. J Clin En-docrinol Metab 2014;99:1983–19923. Ernst S, Demirci C, Valle S, et al. Mechanisms in the adaptation of maternalbeta-cells during pregnancy. Diabetes Manag (Lond) 2011;1:239–2484. Goffin V, Binart N, Touraine P, Kelly PA. Prolactin: the new biology of an oldhormone. Annu Rev Physiol 2002;64:47–675. Freemark M, Driscoll P, Maaskant R, Petryk A, Kelly PA. Ontogenesis ofprolactin receptors in the human fetus in early gestation. Implications for tissuedifferentiation and development. J Clin Invest 1997;99:1107–11176. Kim H, Toyofuku Y, Lynn FC, et al. Serotonin regulates pancreatic beta cellmass during pregnancy. Nat Med 2010;16:804–8087. Layden BT, Durai V, Newman MV, et al. Regulation of pancreatic isletgene expression in mouse islets by pregnancy. J Endocrinol 2010;207:265–2798. Rieck S, White P, Schug J, et al. The transcriptional response of the islet topregnancy in mice. Mol Endocrinol 2009;23:1702–17129. Pechhold S, Stouffer M, Walker G, et al. Transcriptional analysis of intra-cytoplasmically stained, FACS-purified cells by high-throughput, quantitativenuclease protection. Nat Biotechnol 2009;27:1038–104210. Zhang H, Zhang J, Pope CF, et al. Gestational diabetes mellitus resultingfrom impaired beta-cell compensation in the absence of FoxM1, a noveldownstream effector of placental lactogen. Diabetes 2010;59:143–15211. Pasek RC, Gannon M. Advancements and challenges in generating accurateanimal models of gestational diabetes mellitus. Am J Physiol Endocrinol Metab2013;305:E1327–E133812. Freemark M, Avril I, Fleenor D, et al. Targeted deletion of the PRL receptor:effects on islet development, insulin production, and glucose tolerance. Endo-crinology 2002;143:1378–138513. Ormandy CJ, Camus A, Barra J, et al. Null mutation of the prolactin receptorgene produces multiple reproductive defects in the mouse. Genes Dev 1997;11:167–17814. Huang C, Snider F, Cross JC. Prolactin receptor is required for normalglucose homeostasis and modulation of beta-cell mass during pregnancy. En-docrinology 2009;150:1618–162615. Ben-Jonathan N, Hugo ER, Brandebourg TD, LaPensee CR. Focus onprolactin as a metabolic hormone. Trends Endocrinol Metab 2006;17:110–11616. Herrera PL. Adult insulin- and glucagon-producing cells differentiate fromtwo independent cell lineages. Development 2000;127:2317–232217. Yu WM, Appler JM, Kim YH, Nishitani AM, Holt JR, Goodrich LV. A Gata3-Mafb transcriptional network directs post-synaptic differentiation in synapsesspecialized for hearing. eLife 2013;2:e0134118. Postic C, Shiota M, Niswender KD, et al. Dual roles for glucokinase inglucose homeostasis as determined by liver and pancreatic beta cell-specificgene knock-outs using Cre recombinase. J Biol Chem 1999;274:305–31519. Banerjee RR, Rangwala SM, Shapiro JS, et al. Regulation of fasted bloodglucose by resistin. Science 2004;303:1195–119820. Goodyer WR, Gu X, Liu Y, Bottino R, Crabtree GR, Kim SK. Neonatal b celldevelopment in mice and humans is regulated by calcineurin/NFAT. Dev Cell2012;23:21–3421. Lee JY, Ristow M, Lin X, White MF, Magnuson MA, Hennighausen L. RIP-Crerevisited, evidence for impairments of pancreatic beta-cell function. J Biol Chem2006;281:2649–265322. Solomon CG, Willett WC, Carey VJ, et al. A prospective study of pregraviddeterminants of gestational diabetes mellitus. JAMA 1997;278:1078–108323. Schraenen A, Lemaire K, de Faudeur G, et al. Placental lactogens induceserotonin biosynthesis in a subset of mouse beta cells during pregnancy. Dia-betologia 2010;53:2589–259924. Plank JL, Frist AY, LeGrone AW, Magnuson MA, Labosky PA. Loss of Foxd3results in decreased b-cell proliferation and glucose intolerance during preg-nancy. Endocrinology 2011;152:4589–4600

2340 b-Cell PRLR KO and Gestational Diabetes Mellitus Diabetes Volume 65, August 2016

25. Karnik SK, Chen H, McLean GW, et al. Menin controls growth of pancreaticbeta-cells in pregnant mice and promotes gestational diabetes mellitus. Science2007;318:806–80926. Arumugam R, Fleenor D, Lu D, Freemark M. Differential and complementaryeffects of glucose and prolactin on islet DNA synthesis and gene expression.Endocrinology 2011;152:856–86827. Huang C. Wild-type offspring of heterozygous prolactin receptor-null femalemice have maladaptive b-cell responses during pregnancy. J Physiol 2013;591:1325–133828. Rieck S, Kaestner KH. Expansion of beta-cell mass in response to preg-nancy. Trends Endocrinol Metab 2010;21:151–15829. Brelje TC, Allaire P, Hegre O, Sorenson RL. Effect of prolactin versus growthhormone on islet function and the importance of using homologous mammo-somatotropic hormones. Endocrinology 1989;125:2392–239930. Artner I, Le Lay J, Hang Y, et al. MafB: an activator of the glucagon geneexpressed in developing islet alpha- and beta-cells. Diabetes 2006;55:297–30431. Artner I, Blanchi B, Raum JC, et al. MafB is required for islet beta cellmaturation. Proc Natl Acad Sci U S A 2007;104:3853–385832. Brouwers B, de Faudeur G, Osipovich AB, et al. Impaired islet function incommonly used transgenic mouse lines due to human growth hormone minigeneexpression. Cell Metab 2014;20:979–99033. Vasavada RC, Garcia-Ocaña A, Zawalich WS, et al. Targeted expression ofplacental lactogen in the beta cells of transgenic mice results in beta cell pro-liferation, islet mass augmentation, and hypoglycemia. J Biol Chem 2000;275:15399–1540634. Binart N, Melaine N, Pineau C, et al. Male reproductive function is not af-fected in prolactin receptor-deficient mice. Endocrinology 2003;144:3779–378235. Auffret J, Freemark M, Carré N, et al. Defective prolactin signaling impairspancreatic b-cell development during the perinatal period. Am J Physiol Endo-crinol Metab 2013;305:E1309–E131836. Lee JY, Gavrilova O, Davani B, et al. The transcription factors Stat5a/b arenot required for islet development but modulate pancreatic beta-cell physiologyupon aging. Biochim Biophys Acta 2007;1773:1455–146137. Iida H, Ogihara T, Min MK, et al. Expression mechanism of tryptophanhydroxylase 1 in mouse islets during pregnancy. J Mol Endocrinol 2015;55:41–53

38. Kim K, Oh CM, Ohara-Imaizumi M, et al. Functional role of serotonin ininsulin secretion in a diet-induced insulin-resistant state. Endocrinology 2015;156:444–45239. Zhang H, Ackermann AM, Gusarova GA, et al. The FoxM1 transcriptionfactor is required to maintain pancreatic beta-cell mass. Mol Endocrinol 2006;20:1853–186640. Lu J, Hamze Z, Bonnavion R, et al. Reexpression of oncoprotein MafB inproliferative b-cells and Men1 insulinomas in mouse. Oncogene 2012;31:3647–365441. Talchai C, Xuan S, Lin HV, Sussel L, Accili D. Pancreatic b cell de-differentiation as a mechanism of diabetic b cell failure. Cell 2012;150:1223–123442. Hughes E, Huang C. Participation of Akt, menin, and p21 in pregnancy-induced beta-cell proliferation. Endocrinology 2011;152:847–85543. Butler AE, Cao-Minh L, Galasso R, et al. Adaptive changes in pancreatic betacell fractional area and beta cell turnover in human pregnancy. Diabetologia2010;53:2167–217644. Van Assche FA, Aerts L, De Prins F. A morphological study of the endocrinepancreas in human pregnancy. Br J Obstet Gynaecol 1978;85:818–82045. Hakonen E, Ustinov J, Palgi J, Miettinen PJ, Otonkoski T. EGFR signalingpromotes b-cell proliferation and survivin expression during pregnancy. PLoSOne 2014;9:e9365146. Le TN, Elsea SH, Romero R, Chaiworapongsa T, Francis GL. Prolactin re-ceptor gene polymorphisms are associated with gestational diabetes. Genet TestMol Biomarkers 2013;17:567–57147. Chen H, Kleinberger JW, Takane KK, et al. Augmented Stat5 signalingbypasses multiple impediments to lactogen-mediated proliferation in humanb-cells. Diabetes 2015;64:3784–379748. Dai C, Brissova M, Hang Y, et al. Islet-enriched gene expression andglucose-induced insulin secretion in human and mouse islets. Diabetologia2012;55:707–71849. Wang T, Lu J, Xu Y, et al. Circulating prolactin associates with diabetes andimpaired glucose regulation: a population-based study. Diabetes Care 2013;36:1974–198050. Newey PJ, Gorvin CM, Cleland SJ, et al. Mutant prolactin receptor andfamilial hyperprolactinemia. N Engl J Med 2013;369:2012–2020

diabetes.diabetesjournals.org Banerjee and Associates 2341