Growth hormone prevents the development ofautoimmune diabetesRicardo Villaresa,1, Dimitri Kakabadsea,1, Yasmina Juarranzb, Rosa P. Gomarizb, Carlos Martínez-Aa,and Mario Melladoa,2

aDepartment of Immunology and Oncology, Centro Nacional de Biotecnología, Consejo Superior de Investigaciones Científicas, E-28049 Madrid, Spain;and bDepartment of Cell Biology, Faculty of Biology, Complutense University of Madrid, E-28040 Madrid, Spain

Edited by Tak W. Mak, The Campbell Family Institute for Breast Cancer Research, Ontario Cancer Institute at Princess Margaret Hospital, University HealthNetwork, Toronto, ON, Canada, and approved October 18, 2013 (received for review August 9, 2013)

Evidence supports a relationship between the neuroendocrine andthe immune systems. Data from mice that overexpress or aredeficient in growth hormone (GH) indicate that GH stimulates Tand B-cell proliferation and Ig synthesis, and enhances maturationof myeloid progenitor cells. The effect of GH on autoimmunepathologies has nonetheless been little studied. Using a murinemodel of type 1 diabetes, a T-cell–mediated autoimmune diseasecharacterized by immune cell infiltration of pancreatic islets anddestruction of insulin-producing β-cells, we observed that sus-tained GH expression reduced prodromal disease symptoms andeliminated progression to overt diabetes. The effect involves sev-eral GH-mediated mechanisms; GH altered the cytokine environ-ment, triggered anti-inflammatory macrophage (M2) polarization,maintained activity of the suppressor T-cell population, and lim-ited Th17 cell plasticity. In addition, GH reduced apoptosis and/orincreased the proliferative rate of β-cells. These results supporta role for GH in immune response regulation and identify a uniquetarget for therapeutic intervention in type 1 diabetes.

beta cells | Tregs

Growth hormone (GH) is a pleiotropic hormone that affectsa broad spectrum of physiological functions, from carbohy-drate and lipid metabolism to the immune response (1). Severalstudies have linked GH with autoimmune diseases, although itseffects on the immune system are still debated. Whereas somereports using GH-deficient mice indicate that it does not affectimmune competence (2), others suggest that GH is necessary forcorrect immune system development (1, 3). The GH receptor(GHR) is expressed by several lymphocyte subpopulations (4).GH stimulates in vitro T and B-cell proliferation (5) and Igsynthesis (6); enhances human myeloid progenitor cell matura-tion (7); and modulates in vivo Th1/Th2 (8) and humoral im-mune responses (1). In addition, therapeutic activation of theGH/STAT5B axis is postulated as a target for restoring mucosaltolerance in Crohn disease (9, 10). A single point mutation inSTAT5B limits its DNA binding activity as well as maintenanceof FOXP3 expression by Treg cells in nonobese diabetic (NOD)mice (11). These mice develop type 1 diabetes, which is char-acterized by autoimmune destruction of pancreatic β-cells due tothe effect of environmental factors on genetically predisposedindividuals (12, 13). Although this murine model does notcompletely mimic the human disease, most steps in the patho-genesis, including prodromal and clinical symptoms, are closelycomparable (14).Despite the interdependence of GH and insulin regulation

and the known effects of GH and insulin-like growth factor 1(IGF1) on pancreatic β-cell survival, proliferation and neo-genesis (15, 16), hormone influences have not been described intype 1 diabetes; no specific studies have addressed the con-sequences of long-term GH replacement therapy in this disease.Here we show the effects of long-term GH supplementation asa tool to modulate autoimmune attack on pancreatic β-cells.NOD mice transgenic for bovine GH (NOD-Tg bGH) do not

develop type 1 diabetes, and show normal glycemia throughouttheir lives. Our histological analyses indicated that these micedevelop periinsulitis, but show little or no islet infiltration orβ-cell destruction. The mechanism involved specific GH-medi-ated effects on β-cells, where it influenced proliferation andapoptosis, and others that modulate the immune system. GHaffected Th17/Th1 plasticity, M1/M2 macrophage differentia-tion, and Treg cell function. Our findings show an unanticipatedGH effect on tolerization mechanisms that reduce type 1 di-abetes development, and underline the importance of neuroen-docrine regulation of the immune system.

ResultsSustained GH Expression Precludes Development of Overt Diabetes inNOD Mice. In our animal facility, >85% of virgin female NODmice develop overt diabetes before 40 wk of age (Fig. 1A). Tostudy the effects of sustained high levels of circulating GH, weobtained a mouse strain transgenic for bGH under the control ofthe rat phosphoenolpyruvate carboxykinase (PEPCK) promoter(17), on the NOD background. In this strain, as in the parentalC57BL/6-Tg bGH strain, circulating GH levels are constant(∼5 μg/mL). Our mice were healthy and showed no external signsof other autoimmune diseases or tumor development throughouttheir lives. Histological examination showed mild sialitis. Wemonitored glycemia in female NOD-Tg bGH mice and controllittermates for 60 wk, and found that the transgenic mice were

Significance

Although the relationship between endocrine and immunesystems is well documented, few studies have been performedon autoimmune disorders other than those that are sex hor-mone-related. We studied a murine model of autoimmunediabetes, showing that growth hormone (GH) modifies theimmune response to render diabetic mice resistant to diseasedevelopment. The mechanism involves a GH-mediated effecton β-cell survival and/or proliferation and a direct effect onimmune cells. GH triggers a cytokine environment that pro-motes anti-inflammatory macrophage polarization, maintainsthe activity of the suppressor T cells, and limits Th17 cell plas-ticity. This study provides evidence of the importance of endo-crine control of immune functions and indicates that therapiesbased on GH analogs should be considered for treatment ofautoimmune diabetes.

Author contributions: R.V., C.M.-A., and M.M. designed research; R.V., D.K., Y.J., andR.P.G. performed research; R.V., D.K., Y.J., R.P.G., and M.M. analyzed data; and R.V. andM.M. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1R.V. and D.K. contributed equally to this work.2To whom correspondence should be addressed. E-mail: mmellado@cnb.csic.es.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1314985110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1314985110 PNAS | Published online November 11, 2013 | E4619–E4627

IMMUNOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

21,

202

1

mailto:mmellado@cnb.csic.eshttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1314985110/-/DCSupplementalhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1314985110/-/DCSupplementalwww.pnas.org/cgi/doi/10.1073/pnas.1314985110

almost completely resistant to diabetes development (Fig. 1A),as reflected by their higher survival rate compared with NODmice (Fig. 1A). The results indicated significant lowering of cir-culating glucose levels in the transgenic mice with age (Fig. 1B),resulting in relative hypoglycemia. This observation contrastswith the susceptibility of C57BL/6-Tg bGH mice to type 2 di-abetes in a high-fat diet study (18). In histological analyses,NOD-Tg bGH mice showed giant islets with anomalous mor-phology and a periinsular mononuclear cell infiltrate that char-acterizes the initial stages of diabetes (Fig. 1 C–F). The irregularislet morphology suggests β-cell hyperproliferation and islet co-alescence (Fig. 1D), in agreement with the β-cell hyperpro-liferation observed in the presence of placental lactogen (19).Although the results were not statistically significant, the numberof apoptotic β-cells was reduced and Ki-67+ cells were increasedin NOD Tg-bGH pancreas (Fig. S1). These data suggest a role

for GH in protection of β-cells from apoptosis and stimulation oftheir proliferative capacity.

NOD-Tg bGH Mice Have Normal Delayed Type HypersensitivityResponses. Some reports indicate a role for GH as an immuneresponse activator (5, 20); we thus characterized immune cellpopulations in blood, spleen, and peripheral lymph nodes ofNOD and NOD-Tg bGH mice. Flow cytometry analysis showedsimilar immune cell populations in both mouse lines (Table S1).No differences were observed in activation markers (CD25,CD69, CD44, and CD62L) in CD3+ cells isolated from NODand NOD Tg-bGH lymph nodes (Fig. S2).To obtain an overview of immune function in NOD-Tg bGH

mice, we evaluated T-cell responses in a delayed type hyper-sensitivity (DTH) assay of sensitization and challenge with allo-geneic splenocytes. NOD-Tg bGH mice and control littermateswere sensitized by i.v. injection of C57BL/6 splenocytes andchallenged 6 d later by inoculating splenocytes into the right hindfootpad. Inflammation was measured every 24 h until remissionby comparing thickness of the inoculated to the contralateral,vehicle-inoculated footpad. Both mouse groups reacted similarly,with no significant differences in inflammation grade or resolu-tion time, with only a slight remission delay in the transgenicmice (Fig. 2A).

NOD-Tg bGH Mice Have Altered Serum Cytokine Levels. In type 1diabetes, a Th1-to-Th2 shift in the immune response is postu-lated to be protective for pancreatic islets (21). Indeed, theconversion of Th17 to Th1 is necessary to induce diabetes effi-ciently (22). To determine whether the GH-protective effectis associated with changes in circulating cytokines, we used aLuminex assay and detected higher peripheral blood levels ofIFN-γ, IL-4, IL-13, IL-17A, and TNF-α in sera from NOD-TgbGH mice than from NOD littermates (Fig. 2B). The Th2/Th1and Th17/Th1 cytokine ratios suggested a GH-mediated biasagainst a Th1 response in the transgenic mice (Fig. 2C).

Preclinical Type 1 Diabetes Symptoms in NOD-Tg bGH Mice. Early intype 1 diabetes progression, an inflammatory environment isestablished around the islets of Langerhans, usually interpretedas a result of anomalous macrophage activity during postnatalremodeling of the endocrine pancreas (23). Mononuclear cellsare recruited into the pancreas and situate around the pancreaticislets shortly after weaning. The insular parenchyma is invadedprogressively by immune cells, which destroy insulin-producingβ-cells. On serial pancreas sections from NOD-Tg bGH, NODlittermates, and control C57BL/6 mice at different ages, we usedH&E staining to evaluate islet size as well as insular and peri-insular infiltration (Fig. 1 C–F). In NOD-Tg bGH mice, insularinfiltration was delayed and most β-cells were conserved overtime (Fig. 3A). The islet-cell mass thus diminished progressivelyin NOD mice, as predicted, whereas it increased continuouslyin NOD-Tg bGH mice (Fig. 3B). These data concur with thereported GHR-dependent islet hyperplasia (24) and the com-pensatory hyperinsulinemia mechanism associated with GH-dependent insulin resistance (25).Using immunohistochemistry and flow cytometry, we analyzed

infiltrate composition in pancreata from 3- to 5-mo-old mice.Snap-frozen organs were sectioned and stained with anti-CD4,-CD8, -F4/80, -CD11c, and -B220 antibodies. Although in-dividual variation was broad, the inflammation grade in NOD-TgbGH mouse pancreas was lower than that of NOD mice (Fig.3A). Infiltrate composition was nonetheless grossly similar, witha predominance of T cells, mainly CD4+ (Fig. S3). We observedno immune cell infiltration in NOD-Tg bGH islet parenchyma,even in older mice in which infiltration was massive, but alwaysrestricted to the islet periphery (Fig. 3C). This inflammationcoincides with no apparent degradation of the periinsular laminin

Age (weeks)

Gly

cem

ia (m

g/dl

)

A

C D

E F

B

0

100

200

300

400

500

10 20 30 40 50 60

NODNOD-TgbGH

500 m 500 m

100 m100 m

600

20

40

60

80

100

0.0

0.2

0.4

0.6

0.8

1.0

0 10 20 30 40 50

Sur

viva

l rat

e (%

)

Diab

etes

incid

ence

Age (weeks)

Fig. 1. NOD-Tg bGH mice are protected against type 1 diabetes de-velopment. (A) Cumulative diabetes incidence in female NOD (continuousred line, n = 90) and NOD-Tg bGH mice (continuous green line, n = 89). Micewere maintained and glucose levels determined as described in Materialsand Methods. The survival rate of transgenic mice after 60 wk in the absenceof diabetes was 76% (dotted green line). The NOD mouse survival curve(dotted red line) reflects mouse sacrifice at disease onset. (B) Tendency tolower blood glucose levels with age in NOD-Tg bGH mice (green line; n = 10mice per group, Pearson’s correlation coefficient, r = −0.61, P < 0.01). Eachglucose measurement is plotted (twice monthly per mouse). Red, NOD mice;green, NOD-Tg bGH mice. (C–F) Immunohistochemistry of pancreas from 12-wk-old NOD (C and E) and NOD-Tg bGH (D and F) mice stained for lamininand counterstained with hematoxylin. (C and D) Low-magnification images(6×) showing mononuclear infiltrates (arrows) surrounding islets. (Scale bar:500 μm.) (E and F) High-magnification (40×) images showing damage in-duced in the NOD mouse in the laminin sheet (arrows) surrounding islets andcompared with its preservation in NOD-Tg bGH mice. (Scale bar: 100 μm.)

E4620 | www.pnas.org/cgi/doi/10.1073/pnas.1314985110 Villares et al.

Dow

nloa

ded

by g

uest

on

June

21,

202

1

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1314985110/-/DCSupplemental/pnas.201314985SI.pdf?targetid=nameddest=SF1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1314985110/-/DCSupplemental/pnas.201314985SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1314985110/-/DCSupplemental/pnas.201314985SI.pdf?targetid=nameddest=SF2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1314985110/-/DCSupplemental/pnas.201314985SI.pdf?targetid=nameddest=SF3www.pnas.org/cgi/doi/10.1073/pnas.1314985110

layer, which defines the basal lamina of the sheet of Schwanncells thought to be the first target of autoimmune attack (26)(Fig. 1 E and F). In some cases, we found intrainsular foci in maturemice (Fig. 3 C and D), although they must be considered peri-vascular, because they are surrounded by an intact laminin sheet.Because type 1 diabetes is also characterized by development

of a humoral response to islet antigens, we evaluated antiinsulinantibodies in prediabetic NOD-Tg bGH and control prediabeticNOD mice. Whereas NOD mice had high antiinsulin antibodytiters, indicative that insulin is the primary antigen for type 1diabetes in these mice (27), as it is in humans (28), NOD-Tg bGHmice had much lower titers (barely detectable even at 12 wk;Fig. 3E). Spleen B cells from NOD and NOD Tg-bGH mice wereactivated in vitro with anti-mouse IgM antibody (10 μg/mL, 180–360min, 37 °C) alone or with exogenous GH (5 μg/mL); activation wassimilar, as demonstrated by flow cytometry using anti-CD69 andanti-CD86 antibodies (Fig. S4). These findings rule out B-cell ac-tivation defects in NOD Tg-bGH mice.In addition to the maintenance of normal glycemia, the lack of

antiinsulin antibodies and the absence of islet-infiltrating, puta-tive antigen-specific CD4+ T cells in NOD-Tg bGH mice sug-gests that GH modulates the adaptive immune response in type1 diabetes.

Circulating GH Levels Are Responsible for the Protective Phenotype.For detailed characterization of the role of the adaptive immuneresponse in NOD-Tg bGH mice, we evaluated diabetogenic cellpopulations and suppressor cell activity. We transferred totalsplenocytes from NOD-Tg bGH or NOD mice into untreatedor sublethally (7 Gy) irradiated NOD or NOD-Tg bGH mice.Splenocytes from NOD-Tg bGH mice did not protect untreated

NOD mice from disease development (Fig. 4A), suggesting thatNOD-Tg bGH regulatory cells had insufficient suppressive ca-pacity in NOD mice. In addition, cells from the transgenic micedid not accelerate diabetes in sublethally irradiated NOD mice(Fig. 4B), indicating a lack of diabetogenic effector cells in theNOD-Tg bGH splenocyte population.Radiomimetic drugs trigger type 1 diabetes in NOD mice by

targeting the CD4+CD25+FoxP3+ T-cell population and impair-ing their recovery in pancreas infiltrates (29). To determine therole of Treg cells in our model, we transferred splenocytes (2 ×107) from a pool of three overtly diabetic 6-mo-old NOD miceinto 6-wk-old sublethally (7 Gy) irradiated NOD-Tg bGH mice.Diabetes did not develop in the recipients (Fig. 4C), whereas di-abetes was accelerated in irradiated control NOD littermates.These data suggest a resistant suppressive mechanism in irradiatedNOD-Tg bGH mice that blocked NOD effector cells.Because Treg cells are reported to be relatively radioresistant

(30), we transplanted diabetogenic splenocytes into lethally ir-radiated (12 Gy) NOD and NOD-Tg bGH recipients for a 3-wkradioprotection/accelerated diabetes assay. Recipient mice wererescued from lethality and, though NOD mice became hyper-glycemic within 7–10 d, transgenic mice remained normoglyce-mic throughout the experiment (Fig. 4D). Untransferred mice ofboth genotypes, used as a lethality control, died during the firstweek. The results imply a suppressive mechanism for diabetogeniccells in NOD-Tg bGH mice, which are resistant even to a highdose of full-body irradiation. The relative radioresistance of Tregcells and monocytes might account for this suppression.We tested whether transgenic splenocytes protect mice in an

accelerated diabetes model. Two groups of 12-wk-old NOD/SCID(severe combined immunodeficiency) mice were inoculated i.v.

95 100 105 110 115 120 125 130 135 140

24 48 72 96 Hours post challenge

Foot

pad

swel

ling

(%)

A C

0 1 2 IL17/IFN-

IL17/TNF-

IL4/IFN-

IL4/TNF-

IL13/IFN-

IL13/TNF-

Relative ratio

Cyt

okin

es

** **

**

**

B

0

2

4

6

8

NOD Tg

IL4

NOD Tg0

100

200

300

400IL13

0

75

150

225

300

NOD Tg

IL17

NOD Tg NOD Tg0

250

500

750

1000TNF-

0

15

30

45

60IFN-

pg/ml

*** *** *** * *

Fig. 2. T-cell response in NOD-Tg bGH mice. (A) NOD-Tg bGH mice show standard DTH responses. Footpad swelling at several times postchallenge in NOD-TgbGH (dark gray, n = 6) and NOD mice (light gray, n = 6). Swelling was calculated as the percentage of footpad thickness compared with the baseline at t = 0.Values shown are mean ± SD. Student t test showed no significant differences at any time. (B) Circulating cytokine levels in 4-mo-old NOD (n = 12) and NOD-Tg bGH (Tg) (n = 11) mice. Individual and mean values are shown. Student t test, *P < 0.05, ***P < 0.001. (C) Ratios between Th2/Th1 and Th17/Th1 profiles,assessed by circulating cytokines quantification, in NOD (light gray bars, n = 12) and NOD-Tg bGH (dark gray, n = 11) mice. Values have been normalized toNOD mice data. Student t test, **P < 0.01.

Villares et al. PNAS | Published online November 11, 2013 | E4621

IMMUNOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

21,

202

1

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1314985110/-/DCSupplemental/pnas.201314985SI.pdf?targetid=nameddest=SF4

with 2 × 107 splenocytes from a pool of two 5-mo-old diabeticNOD mice. One group also received 2 × 107 splenocytes froma pool of two 5-mo-old transgenic mice (Fig. 4E). All mice thatreceived NOD splenocytes alone developed hyperglycemia by4 wk posttransfer. The group inoculated with splenocytes fromNOD + NOD-Tg bGH mice showed a slight, nonsignificantdelay in hyperglycemia (log-rank test P = 0.094), suggesting lossof a hypothetical protective mechanism in the absence of circu-lating GH.

Suppressive Potential in NOD-Tg bGH Mice. Although Treg cells arebroadly implicated in type 1 diabetes development, their precise

function during the prediabetic stage is not well understood (31–34). CD4+CD25+ Treg cells control disease progression throughvarious potential mechanisms, inhibiting activation, proliferation,and/or migration of islet-specific T cells in lymph nodes and inpancreas (35). Because Treg cell suppressive potential is associ-ated with FoxP3 levels (36), we used flow cytometry to determineFoxP3 expression on CD4+CD25+ peripheral blood lymphocytesfrom C57BL/6, NOD, and NOD-Tg bGH mice at 2 mo of age,before hyperglycemia was detected (Fig. 5A), and at 5 mo, whenNOD mice were hyperglycemic (Fig. 5B). We found no differ-ences in FoxP3 expression in any 2-mo-old mice (Fig. 5 A and C),whereas CD4+CD25+ lymphocytes from 5-mo-old NOD miceshowed a clear reduction in FoxP3 levels compared with thosefrom B6 mice; this down-regulation was not observed for NOD-Tg bGH CD4+CD25+ Treg cells (Fig. 5 B and C). The resultssuggest a GH effect on the maintenance of Treg cell activity.In 2-mo-old mice, the percentage of CD4+CD25+FoxP3+

cells was higher in C57BL/6 than in NOD-Tg bGH mice or NOD

Fig. 3. Type 1 diabetes symptoms are largely suppressed in NOD-Tg bGHmice. (A) Severity of insulitis and destructive lesions in NOD, NOD-Tg bGH,and C57BL/6 (control) mice. After H&E staining, ∼50–100 random islets perpancreas were evaluated. Three mice were analyzed for each genotype andage group (13, 20, and 24 wk). The degree of mononuclear cell infiltrationwas graded independently by two observers as follows: 0, normal (white); 1,mild periinsulitis (light gray); 2, severe periinsulitis (dark gray); or 3, insulitis(black; see Materials and Methods for details). (B) Islet mass estimated frompercentage of endocrine area (relative to total pancreas surface) asa function of age for NOD (gray), NOD-Tg bGH (black), and C57BL/6 mice(white; n = 3 for each age group). (C and D) Late infiltration in islets fromNOD-Tg bGH mice. Frozen pancreas sections of 7-mo-old NOD-Tg bGH mice.Islets (I) and infiltrates (M) are labeled. (Scale bar, 100 μm.) (C) Mergedimage of H&E staining showing giant coalescent islets. Original magnifi-cation, 6×. (D) Section stained with antilaminin antibody to identify thebasal layer of pancreatic acini, blood vessels, and periinsular sheet ofSchwann cells. Hematoxylin counterstain shows perivascular and isletinfiltrates. Mononuclear infiltrates were surrounded by this laminin sheetand hence did not invade the islet parenchyma. Original magnification,10×. (E ) Evaluation of prodromal antiinsulin antibodies (NOD-Tg bGH, n = 5;NOD, n = 7; C57BL/6 control, n = 3) at 10 (light gray), 12 (dark gray), and 14wk (black). ELISA for human insulin did not detect autoantibodies in NOD-Tg bGH mice. Background optical density was subtracted. Student t test,**P < 0.01.

D

0 10 20 30 40

0 10 20 30 40

100

80

60

40

20

0

B100

80

60

40

20

0

A

0 10 20 30 40

100

80

60

40

20

0

E

0 10 20 30 40

100

80

60

40

20

0

C100

80

60

40

20

0

0 2 4 6 8 10 12 14Cum

ulat

ive

surv

ival

(%)

Age (weeks)

NS

NS

Fig. 4. Kaplan–Meyer analysis for adoptive transfer protocols. (A) Seven-week-old NOD mice (continuous line, n = 4) were inoculated with 2 × 107

NOD-Tg bGH mouse splenocytes; inoculated NOD-Tg bGH mice were used ascontrols (dashed line). No delay in hyperglycemia was observed comparedwith the normal behavior of NOD mice. (B) Seven-week-old NOD mice(continuous line, n = 6) were sublethally irradiated and inoculated with 2 ×107 NOD-Tg bGH mouse splenocytes. No delay was observed in manifesta-tion of overt diabetes compared with untreated NOD control mice (dashedline). (C) Six-week-old NOD-Tg bGH mice (dashed line, n = 5) were sub-lethally irradiated and inoculated with 2 × 107 diabetic NOD mouse sple-nocytes. Irradiated NOD littermates (continuous line, n = 4) were used ascontrols. Accelerated diabetes development was observed in NOD controls,whereas NOD-Tg bGH mice remained fully resistant. (D) The experiment in Cwas repeated using lethally irradiated NOD-Tg bGH mice (dashed line, n = 4).NOD mice (n = 2) were used as controls for accelerated diabetes development,and uninoculated mice (NOD, n = 2; NOD-Tg bGH, n = 2) as controls of lethality;all four untransferred mice died within 2 wk of irradiation. (E) NOD/SCID mousesensitivity to diabetes development after splenocyte transfer (NOD, continuousline; NOD + NOD-Tg bGH, dashed line; 12 wk old, n = 4 for both groups). Log-rank test P < 0.05 was considered significant. NS, not significant.

E4622 | www.pnas.org/cgi/doi/10.1073/pnas.1314985110 Villares et al.

Dow

nloa

ded

by g

uest

on

June

21,

202

1

www.pnas.org/cgi/doi/10.1073/pnas.1314985110

littermates, with no difference between the last two groups. Thepercentage of CD4+CD25+FoxP3+ cells was higher in 5-mo-oldNOD-Tg bGH mice than in NOD littermates, although in bothcases it was lower than that in C57BL/6 mice (Fig. 5D), whichsuggests higher suppressive T-cell activity in NOD-Tg bGH thanin NOD mice.

We used an in vitro suppression assay to test for a correlationbetween reduced FoxP3 expression on CD4+CD25+ lympho-cytes and a reduction in their suppressive capacity. Coculture ofCD4+CD25−NODT cells with varying proportions of CD4+CD25+

T cells from 5-mo-old NOD or NOD-Tg bGH mice led to slightbut significantly higher suppressive activity by NOD-Tg bGH-derived cells (Fig. 5E).These results indicate that GH participates in maintenance of

the suppressive potency of Treg cells as well as of the relativelyhigh FoxP3 levels in CD4+CD25+ cells. This effect might bepotentiated in vivo, because NOD-Tg bGH cells are continu-ously exposed to high levels of circulating GH. Treg cells in vivomight also be activated locally by antigen-presenting cells (37)specifically charged with islet antigens, which were absent in ourin vitro experiments.

GH Effect on Macrophage Polarization. Macrophages have a keyrole in pancreas remodeling (23) and are also present in pan-creas infiltrate in NOD and NOD-Tg bGH mice. There are twomain macrophage types: M1 or inflammatory macrophages,characterized by high NOS2 (inducible nitric oxide synthase)expression, and M2 or anti-inflammatory macrophages charac-terized by arginase-1 expression. We used quantitative real-timePCR to quantify M1/M2 macrophage markers in pancreaticlymph nodes from 3-mo-old NOD and NOD-Tg bGH mice. Wefound lower NOS2 and higher arginase-1 RNA levels in nodesfrom NOD-Tg bGH than from NOD mice (Fig. 5F). These datawere confirmed by immunohistochemical identification of argi-nase-1–producing cells in NOD and NOD-Tg bGH mouse pan-creas (Fig. 5 G and H). The results indicate a GH effect onmacrophage polarization toward the M2 phenotype in NOD-TgbGH mice.

GHModulates Th17 Cell Plasticity.Diabetes development is reportedto be associated with acquisition of a Th1-like phenotype by Th17cells (38), which then express T-bet and secrete IFN-γ. We testedthe GH effect on Th17/Th1 plasticity using RT-PCR to determineIL-17, IFN-γ, IL-2, IL-22, and GM-CSF mRNA levels in totalpancreas of age-matched NOD-Tg bGH and prediabetic NODmice, as well as of ROR-γT and T-bet transcription factors asspecific markers of Th17 and Th1 polarization, respectively (Fig.6A). To avoid interference due to different numbers of infiltratingcells in the pancreas, we compared the ratio for messages char-acteristic of Th17 and Th1 cells, which is more indicative of therelative abundance of nonpathogenic Th17 cells than are in-dividual Th17 or Th1 values (39). The results showed a higherRORC/Tbx21 (Fig. 6B) and IL-17/IFN-γ (Fig. 6C) ratios inNOD-Tg bGH mouse pancreas. In addition, we detected higherIFN-γ, IL-2, IL-22, and GM-CSF mRNA levels in NOD mousepancreas (Fig. 6A). These data indicate a significant reduction ofpathogenic Th17 cells infiltration and their plasticity to Th1 inNOD-Tg bGH mouse pancreas.

DiscussionType 1 diabetes is a multifactorial disease caused by the con-currence of genetic and environmental factors that include in-fectious agents, diet, and illness (40). Endocrinopathies charac-terized by chronic overproduction of hormones whose actionopposes that of insulin, such as epinephrine, glucagon, cortisol, orGH, generally cause diabetes by triggering insulin resistance (41);nonetheless, very little is known of their potential to influenceautoimmune diabetes. Here we observed that a transgenic mousestrain on the NOD background that expresses bGH under thecontrol of the rat PEPCK promoter (NOD-Tg bGH) did notdevelop type 1 diabetes. The pancreatic inflammatory phenotypecharacteristic of the NOD background was severely reduced inthe presence of GH. These mice also showed almost completesuppression of the adaptive immune response. We detected no

A B

FoxP3

C D

FE

0

1

2

HG

3

ARG1 iNOSRel

ativ

e ex

pres

sion

(2-

Ct )

2:1 4:1 8:1 C

*

*

*0

2

4

6

8

cpm

(x 1

04)

2 m 5 m

FoxP

3+ in

CD

4+ (%

)

0 100 200 300 400 500 600 700

2 m 5 m

FoxP

3 M

FI

NOD NOD-Tg bGH C57BL/6

I I

****

16

12

8

4

0

Fig. 5. Bias toward a regulatory phenotype in NOD-Tg bGH mice. Down-regulation of FoxP3 in NOD Treg cells. Blood samples from 2-mo-old (A) and5-mo-old mice (B) of C57BL/6, NOD, and NOD-Tg bGH genotypes were la-beled for CD4, CD25, and FoxP3; Treg cells were gated by light-scatterproperties (forward scatter and side scatter), CD25 and CD4. One represen-tative sample of nine is shown. C57BL/6, black line; NOD, red line; NOD-TgbGH, green line. (C) FoxP3 expression in CD4+CD25+ cells assessed by flowcytometry. FoxP3 mean fluorescence intensity in blood samples from 2- and5-mo-old mice. (D) Relative abundance of FoxP3+ cells in the blood CD4+

population, assessed by flow cytometry, at 2 and 5 mo. (C and D) NOD (lightgray), NOD-Tg bGH (dark gray), and C57BL/6 (white) mice mean values + SDfor four experiments (total n = 9 for each mouse group). Student t test, *P <0.5; **P < 0.1. (E) Suppressive activity of CD4+CD25+ splenocytes from NOD(light gray) and NOD-Tg bGH (dark gray) mice at various ratios of CD4+CD25−

effector cells (letter C indicates no suppressor cells added). [3H]thymidineincorporation after coculture is shown as a percentage of the value forcultured pure effector cells. (F) M1 and M2 macrophage marker expression.Quantitative RT-PCR was used to quantify relative levels of NOS2 and arginase-1 mRNA in pancreatic lymph nodes from 3-mo-old mice. Values (2−ΔΔCt) arerelative to the mean level of each message in samples from NOD mice. (G andH) Immunohistochemistry showing arginase-1 expression within the peri-insular infiltrate in young mice (8 wk). Anti–arginase-1 antibody was visualizedwith horseradish peroxidase and diaminobenzidine (arrows); hematoxylin wasused as counterstain (G, NOD; H, NOD-Tg bGH). Original magnification, 20×.

Villares et al. PNAS | Published online November 11, 2013 | E4623

IMMUNOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

21,

202

1

prodromal antiislet antibody production or diabetogenic cell ex-pansion; hence, there was no islet destruction in these mice.There is considerable evidence for distinct GH effects on

pancreatic β-cells (16), some direct and others through its mainmediator, IGF1 (42). Exogenous GH in rat islet cultures thusstimulates DNA synthesis and insulin production (43); IGF1 andGH signaling have mitogenic effects on INS-1 cells (44). Wedetected a large β-cell mass in NOD-Tg bGH mice that corre-lated with the higher proliferation rate and lower apoptotic β-cellnumbers detected in the pancreas of these mice, which couldexplain in part the lack of hyperglycemia in these mice. By ac-tivating the JAK/STAT pathway, GH can also stimulate β-cellsurvival. The JAK/STAT-activated suppressors of cytokine sig-naling block damage triggered by cytokines such as IFN-γ orTNF-α (45), and expression of a constitutively active form ofSTAT5b has a protective effect on β-cells in a model of strep-tozotocin-induced diabetes (46).NOD-Tg bGH mice showed periinsulitis, although we found

no sign of degradation of the periinsular laminin layer, and noantiinsulin antibodies in serum. The results confirm a GH effecton the immune system in addition to its effect on β-cells.Although differences in immune cell activity are associated

with high levels of circulating GH (47), we detected no majordifferences in any of the circulating cell populations in eithermouse type; B220+, CD3+, CD4+, CD8+, CD11b+, or Gr1+ cellnumbers were similar, as were the lymph node T-cell activationmarkers (CD25, CD69, CD44, and CD62L). This observationconcurs with the similar response in a DTH assay of NOD andNOD-Tg bGH mice to immunization with allogeneic splenocytes,ruling out general defects in the T-cell response and/or antigenpresentation. It is nonetheless postulated that APC, antigen pre-senting cell, defects are responsible for the lack of adequate reg-ulatory potential in NOD mice (48). It could be argued that thehalt observed at the periinsulitis checkpoint is due to a reduced

Th2 response and Ig production in NOD-Tg bGH mice. Over-expression of bGH in C57BL/6 mice alters the humoral responseto egg albumin by reducing Th2 cytokine production (1). Wedid not observe such Th2 defects on the NOD background, asassessed by measurement of circulating cytokine levels. In anycase, antiislet antibodies are not directly linked to β-cell de-struction, and the role of β-cells appears to be restricted to theirantigen-presenting activity (49). β-cells from NOD and NOD-TgbGH mice showed no defects in their in vitro activation by anti-IgM antibodies. We previously observed that after stimulationwith conventional antigens, the antigen response is reduced in TgbGH mice, with a IgG1 to IgG2 isotype shift (1). These findingsimplicate GH in altering T-cell function.Transferred NOD-Tg bGH splenocytes did not protect NOD

mice from diabetes, and NOD splenocytes did not promote di-abetes when transferred into NOD-Tg bGH mice, suggestingthat control of T-cell responses in NOD-Tg bGH mice is de-pendent on circulating GH levels.At 5 mo of age, with a well-developed inflammatory environ-

ment, CD4+CD25+ Treg cell numbers are maintained at stablelow levels in NOD mice, whereas they rise in NOD-Tg bGHmice. In NOD-Tg bGH mice, we did not observe the down-regulation of FoxP3 expression found in hyperglycemic NODmice. Because FoxP3 expression is directly linked to the regu-latory action of Treg cells (50), it is thus possible that GH reg-ulates FoxP3 activation via STAT5b or STAT3. A single pointmutation in STAT5b, which encodes a transcription factor in-volved in GH signaling (51), limits FoxP3 expression by Tregcells (11). Sustained activation of STAT3 (another transcriptionfactor involved in signaling through the GHR) is needed to main-tain FoxP3 expression by Treg cells (52). Through STAT5B orSTAT3 activation, GH might thus increase FoxP3 levels, regu-lating Treg cell activity.In the adoptive transfer experiments, sublethal and even lethal

irradiation did not alter the protective effect of GH expression inNOD-Tg bGH mice. GH promotes radioprotection in a varietyof cell types, and radiation sensitivity differs in some immunesystem niches. For example, GH and IGF1 enhance hemato-poietic stem cell radioresistance and proliferation (53). Theseresistance mechanisms nonetheless appear to be insufficient forthe rapid expansion of suppressive cells that would be needed tocontrol disease in an accelerated diabetes model. It is more likelythat GH increases the radioresistance of the Treg cell pop-ulation; lethally irradiated wild-type hosts transferred with scurfybone marrow cells did not develop autoimmune disease, due tosuppression of sf-derived T cells by radioresistant host FoxP3+

Treg cells (30).We tested the in vitro suppressive capacity of Treg cells from

NOD and NOD-Tg bGH mice and found slight but significantdifferences. These differences might be considered insufficient toexplain complete resistance to diabetes development; nonethe-less, T-cell differentiation and activity are dependent on antigen-presenting cell type and the microenvironment in which pre-sentation occurs, and Treg cells might be induced only locally byM2 macrophages (37). Macrophages have two distinct pheno-types (54): inflammatory (M1) macrophages participate in anti-gen recognition and secretion of inflammatory cytokines, andnoninflammatory (M2) macrophages are involved in tissue repairand remodeling (55). M1 polarization is induced by IFN-γ andcharacterized by high NOS2 expression and by secretion ofproinflammatory cytokines IL-1 and IL-12, whereas M2 are in-duced by IL-4 and IL-13 and are characterized by high arginase-1expression and by IL-10 secretion. Our data for pancreatic lymphnodes and pancreata from prediabetic NOD mice showed thepresence of M1 macrophages (arginase-1lowNOS2high), whereasage-matched NOD-Tg bGH mouse macrophages had an M2 phe-notype (arginase-1highNOS2low). This difference might be the result

RORC/Tbx21

0

50

100

150

200

250

**

mR

NA

RO

Rt

/mR

NA

T-b e

t

IL-17/IFN-

0

10

20

30

40*

mR

NA

IL-1

7/m

RN

AI F

N-

)B

A

C

* *** * *** *** * NS *

0% 20% 40% 60% 80% 100% 120%

RORC Tbx21 IL22 IL2 IFNG GM-CSF IL17 IL10

Relative expression

Fig. 6. Cytokine expression in the pancreas of NOD prediabetic and NOD-TgbGH diabetes-resistant mice. RNA from 13-wk-old mouse whole pancreaswas quantified by quantitative RT-PCR. Student t test (n = 6); *P < 0.05; **P <0.01; ***P < 0.001. (A) Relative expression (2−ΔΔCt) of characteristic Th1 andTh17 transcription factors and cytokines in NOD and NOD-Tg bGH micepancreas, relative to mean values in NOD. The ratios between mRNA expressionof RORC/Tbx21 (B) and IL-17/IFN-γ (C) in NOD (light gray) and NOD-Tg bGH (darkgray) mice are shown.

E4624 | www.pnas.org/cgi/doi/10.1073/pnas.1314985110 Villares et al.

Dow

nloa

ded

by g

uest

on

June

21,

202

1

www.pnas.org/cgi/doi/10.1073/pnas.1314985110

of the high circulating levels of IL-17 (56), and could also influenceincreased local Treg cell activity.Autoimmune diabetes is a well-characterized Th1 pathology.

We detected high IFN-γ and IL-2 mRNA levels in NOD mousepancreas, which were lower in NOD-Tg bGH pancreas. Bothmouse models also had high IL-17 mRNA levels, suggestingTh17 cell involvement in type 1 diabetes. Th17 cells are found insome autoimmune diseases such as experimental autoimmuneencephalomyelitis (57) and rheumatoid arthritis (58); NODand NOD-Tg bGH mouse pancreas also showed high ROR-γTmRNA levels.Recent evidence suggests that depending on the microenvi-

ronment, Th17 cells can alter their differentiation program toinduce protective or proinflammatory responses (59, 60). Wefound a significant reduction in mRNA levels of IFN-γ, IL-2, andGM-CSF, characteristic mediators for Th17 pathogenic cells(59), in NOD-Tg bGH pancreas compared with that of NODmice. Although, increasing evidences points to the role of Th17cells in NOD mice, it seems that the conversion of this cell subsetinto Th1 is more important (22, 38). How the inflammatorymicroenvironment modulates this Th17 differentiation is notcompletely understood, although our results suggest that GHmaintains a nonpathogenic profile of Th17 cells and reducestheir Th1 potential in the pancreas.We cannot rule out that GH might also contribute to trans-

genic islet resistance to immune attack via STAT3 activation.STAT3 activation in Treg cells, associated with type 1 diabetesresistance in NOD mice (61), controls the macrophage IL-10–mediated anti-inflammatory response (62) and is an importantsurvival factor in β-cells (63, 64). Our data show that GH-mediated interference in type 1 diabetes development involvesan increase in β-cell mass, protection of the periinsular lamininlayer, and a direct effect on immune cells—mainly macrophages,Th17, and possibly Treg cells. These results demonstrate theimportance of endocrine control of immune functions, and in-dicate that therapies based on GH analogs and/or their signalingcascades should be considered for treatment of autoimmunediabetes.

Materials and MethodsMice.Mice transgenic for bGH under the control of the phosphoenolpyruvatecarboxykinase promoter on a C57BL/6J × C3H/J hybrid background (65) werecrossed on the NOD background until NOD polymorphic alleles were stabi-lized (F0; NOD-Tg PEPCK-bGH/Ccnb), as assessed by the length of single se-quence repeats (66). The transgenic strain was maintained by continuousbackcrosses on NOD females. Experimental results were obtained from F5onward, always using strict littermates. Mice were fed a standard laboratoryrodent diet (Global Diet 2918, Harlan Iberica; 18.5% protein, 5.5% oils andfat) and tap water ad libitum. The mice were monitored twice a week withAccutrend kits (Roche Diagnostics) for development of hyperglycemia, anddeclared diabetic when glucose was >200 mg/dL in two consecutive mea-surements. Overtly diabetic mice were killed. Mice were handled accordingto national and European Union guidelines, and experiments were ap-proved by the Comité Ético de Experimentación Animal, Centro Nacional deBiotecnología.

Immunohistochemistry. Pancreata were embedded in optimal cutting tem-perature freezing medium (Sakura) and snap-frozen in cooled isopentane.Sections (7 μm) were cut, air-dried, and fixed in cold acetone. When storedat −80 °C, sections were postfixed in ethanol/acetone (1:3) before staining.Primary antibodies were guinea pig anti-porcine insulin antibody (Dako), anti–arginase-1 (BD Biosciences), rabbit anti-laminin (Sigma), and rabbit anti-Ki67(Novocastra antibodies); immunodetected with the tyramide signal amplifica-tion indirect staining kit (Perkin-Elmer); and visualized with diaminobenzidine.Hematoxylin was used for counterstaining.

β-Cell Mass Determination and Infiltration Level Count. Pancreata frommice ofdifferent ages (three per group) were extracted, weighed, formaldehyde-fixed, and paraffin-embedded. For quantification, one 5-μm section wasanalyzed every 100 μm. Sections were H&E stained and photographed at 10×

magnification. Composite images were generated with the photomergefunction of Adobe Photoshop CS5. On merged images, a grid was super-imposed and β-cell mass calculated from the ratio of intersections in endo-crine vs. total pancreas and pancreas weight (67). Separation between gridlines was 50 μm. The same slides were used to determine infiltration level inislets. Insulitis was scored by the following criteria: insulitis (grade 3), in-filtration in the islet parenchyma; severe periinsulitis (grade 2), three ormore rows of mononuclear cell infiltrate surrounding the islet; mild peri-insulitis (grade 1), less than three rows of periinsular infiltrating cells, and noinsulitis (grade 0), absence of cell infiltration. Apparent intraislet area wassometimes counted as periinsular when the insular parenchyma was notinvaded, as determined by the integrity of Schwann cell basal lamina.

Delayed Type Hypersensitivity. C57BL/6J splenocytes were isolated by me-chanical disaggregation and erythrocyte lysis with NH4Cl, washed once withPBS + 0.1% BSA and twice with PBS. The 6-wk-old mice were sensitized byi.v. injection of 2 × 105 C57BL/6J splenocytes, and challenged on day 6 in theright footpad with 1.5 × 107 cells in 50 μL PBS. Control left footpads received50 μL PBS. Footpad thickness was measured with a vernier caliper (MitutoyoJapan) at 24, 48, and 72 h after challenge. Results of footpad swelling werecalculated as the difference between challenged vs. prechallenged footpad,expressed as a percentage. Student t test P values >0.05 were considerednonsignificant.

Adoptive Transfer. Donor spleens were processed as above. The indicatednumber of splenocytes was injected i.v. into the tail of recipient mice. Whenindicated, host mice were previously irradiated lethally (12 Gy) or sublethally(7 Gy) with a single full-body dose from a 137Cs source. SPSS Statistics softwarewas used for statistical analyses.

Cytokine Measurement. Circulating cytokines were quantified in serum witha Bio-Plex kit and analyzer (BioRad). Serum samples were stored at −80 °Cuntil use. Antiinsulin and total antibodies in serum were measured bystandard ELISA at 495 nm. Plates were coated with 100 μL per well withhuman insulin (20 μg/mL; Novo-Nordisk) or goat anti-mouse Ig kappa chain(50 μg/mL) in PBS. Plates were incubated (overnight, 4 °C), washed with PBS,0.1% Tween 20, and blocked with 1% BSA, 0.05% Tween-20 in PBS [1 h,room temperature (RT)]. Serial serum dilutions were added to washed platesand incubated (2 h, RT); after washing, peroxidase-conjugated anti-Ig(1:2,000; Dako) was added. Plates were incubated (1 h, RT), washed, and 100μL orthophenylenediamine substrate solution (Sigma) was added. The re-action was terminated after 20 min and measured.

Cell Purification and Flow Cytometry. To prepare single-cell suspensions,spleens and lymph nodes were harvested and minced on a 40-μm nylon meshin RPMI medium 1640 (Lonza) supplemented with 10% FBS, 2 mM L-gluta-mine, and 50 μg/mL penicillin/streptomycin. For APC preparations, spleenswere predigested with collagenase A and DNase I. CD11c+, CD4+CD25+, andCD4+CD25− cell populations were enriched by CD11c Microbeads Mouse(Miltenyi Biotech), Dynabeads Flow Comp Mouse CD4+CD25+ Treg Cells Kit(Invitrogen), and an AutoMACS Cell Sorter (Miltenyi Biotech). When re-quired, murine B cells were purified using mouse pan-T Dynabeads (Invi-trogen) and T cells with mouse T-cell negative isolation kit (Dynal). Purity ofall cell preparations was routinely >95%. Blood samples were lysed withVersaLyse (Beckman Coulter).

Single-cell suspensions of lymphoid organs or blood leukocytes wereprepared and blocked with anti-CD16/32 (BD Pharmingen) to impede Fc-mediated nonspecific antibody binding. Samples were stained with antibodyconjugates by a standard procedure, using FITC anti-CD25, FITC anti-CD11b,and SPRD anti-Gr1 (Pharmingen); SPRD anti-CD4 (eBiosciences); FITC anti-CD3, FITC anti-CD8, PE anti-CD44, FITC anti-CD69, FITC anti-CD45, and APCanti-B220 (Beckman Coulter); PE anti-CD86 (BioLegend); and PE anti-CD62L(Southern). FoxP3 expression was determined after permeabilization andintracellular staining with a PE-labeled antibody (FoxP3 staining set; eBio-sciences). When necessary, naïve B cells (92–95% pure) were first activatedwith 10 μg/mL goat anti-mouse IgM Ab (Jackson ImmunoResearch; 3 or 6 h,37 °C), alone or with exogenous human GH (5 μg/mL, Genotonorm; Pfizer).Stained samples were analyzed on a flow cytometer (Cytomics FC 500; Beck-man Coulter). FACS data were analyzed with FlowJo and CytoSpec software.

In Vitro Suppression Assay. CD4+CD25− NOD splenocytes (5 × 104) werecocultured with variable ratios of CD4+CD25+ cells (2:1, 4:1, and 8:1) and withNOD spleen CD11c+ cells (5 × 104; previously γ-irradiated, 15 Gy) in the pres-ence of anti-CD3 (1 μg/mL). Cultures were prepared in triplicate in U-bottom96-well plates (Nunclon Surface) with RPMI-1640 medium supplemented with

Villares et al. PNAS | Published online November 11, 2013 | E4625

IMMUNOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

21,

202

1

10% FBS, 2 mM L-glutamine, 1 mM pyruvate, and 50 μM β-mercaptoethanol(72 h, 37 °C), and cells were pulsed (16 h) with 1 μCi [3H]thymidine (Perkin-Elmer) per well. The 3H incorporation was measured by liquid scintillationspectrometry using CytoScint mixture (MP Biomedical) and a 1450 MicroBetacounter (Perkin-Elmer).

Semiquantitative Real Time PCR. cDNA sequences were obtained from theGenBank database. PCR primers were designed from the cDNA sequencesusing Primer-BLAST (68) (Table S2). RNA (5 μg) was used for reverse tran-scription. cDNA was obtained by SuperScript II reverse transcriptase (Invi-trogen). Then cDNA was amplified by PCR analysis, using Power SYBR GreenPCR Master Mix (Applied Biosystems), 0.3 μM of primers, and three serialdilutions of RT products. Triplicate samples were quantified using the ABI

Prism HT7900 sequence detection system (Applied Biosystems). For relativequantification, we used the equation 2−ΔΔCt. We normalized each samplewith β-actin (ΔCt), and ΔΔCt represents the difference between the Ct fromeach gene expression of NOD at 13 wk and each datum.

ACKNOWLEDGMENTS. We are grateful for the gift of Genotonorm (PfizerEspaña). We thank R. Barroso and G. Cascio for technical support, L. Gómez foranimal handling, C. Bastos for secretarial assistance, and C. Mark for editorialassistance. Support for this work was provided by Comunidad de MadridContract S2011/BMD-2502 (to R.V.); a La Caixa Fellowship (to D.K.); SpanishMinistry of Science and Innovation Grant SAF 2011-27370; European UnionFP7-integrated project Masterswitch 223404; Fondo de Investigación Sanitaria,Instituto de Salud Carlos III RD12/009/009 and RD12/009/002; and the Comunidadde Madrid S2010/BMD-2350.

1. Gonzalo JA, et al. (1996) Enterotoxin septic shock protection and deficient T helper2 cytokine production in growth hormone transgenic mice. J Immunol 157(8):3298–3304.

2. Cross RJ, Bryson JS, Roszman TL (1992) Immunologic disparity in the hypopituitarydwarf mouse. J Immunol 148(5):1347–1352.

3. Murphy WJ, Durum SK, Anver MR, Longo DL (1992) Immunologic and hematologiceffects of neuroendocrine hormones. Studies on DW/J dwarf mice. J Immunol 148(12):3799–3805.

4. Dardenne M, Mello-Coelho V, Gagnerault MC, Postel-Vinay MC (1998) Growth hor-mone receptors and immunocompetent cells. Ann N Y Acad Sci 840:510–517.

5. Postel-Vinay MC, de Mello Coelho V, Gagnerault MC, Dardenne M (1997) Growthhormone stimulates the proliferation of activated mouse T lymphocytes. Endocri-nology 138(5):1816–1820.

6. Yoshida A, Ishioka C, Kimata H, Mikawa H (1992) Recombinant human growth hor-mone stimulates B cell immunoglobulin synthesis and proliferation in serum-freemedium. Acta Endocrinol (Copenh) 126(6):524–529.

7. Merchav S, Tatarsky I, Hochberg Z (1988) Enhancement of human granulopoiesis invitro by biosynthetic insulin-like growth factor I/somatomedin C and human growthhormone. J Clin Invest 81(3):791–797.

8. Takagi K, Suzuki F, Barrow RE, Wolf SE, Herndon DN (1998) Recombinant humangrowth hormone modulates Th1 and Th2 cytokine response in burned mice. Ann Surg228(1):106–111.

9. DiFedele LM, et al. (2005) Tumor necrosis factor alpha blockade restores growthhormone signaling in murine colitis. Gastroenterology 128(5):1278–1291.

10. Han X, et al. (2007) Tumour necrosis factor alpha blockade induces an anti-inflammatory growth hormone signalling pathway in experimental colitis. Gut 56(1):73–81.

11. Murawski MR, Litherland SA, Clare-Salzler MJ, Davoodi-Semiromi A (2006) Upregu-lation of Foxp3 expression in mouse and human Treg is IL-2/STAT5 dependent: Im-plications for the NOD STAT5B mutation in diabetes pathogenesis. Ann N Y Acad Sci1079:198–204.

12. King C, Sarvetnick N (2011) The incidence of type-1 diabetes in NOD mice is modu-lated by restricted flora not germ-free conditions. PLoS ONE 6(2):e17049.

13. Pollard KM (2012) Gender differences in autoimmunity associated with exposure toenvironmental factors. J Autoimmun 38(2-3):J177–J186.

14. Driver JP, Serreze DV, Chen Y-G (2011) Mouse models for the study of autoimmunetype 1 diabetes: A NOD to similarities and differences to human disease. Semin Im-munopathol 33(1):67–87.

15. Heit JJ, Karnik SK, Kim SK (2006) Intrinsic regulators of pancreatic beta-cell pro-liferation. Annu Rev Cell Dev Biol 22:311–338.

16. Nielsen JH, Svensson C, Galsgaard ED, Møldrup A, Billestrup N (1999) Beta cell pro-liferation and growth factors. J Mol Med (Berl) 77(1):62–66.

17. McGrane MM, et al. (1990) Developmental regulation and tissue-specific expressionof a chimaeric phosphoenolpyruvate carboxykinase/bovine growth hormone gene intransgenic animals. J Reprod Fertil Suppl 41:17–23.

18. Olsson B, et al. (2005) Bovine growth hormone transgenic mice are resistant to diet-induced obesity but develop hyperphagia, dyslipidemia, and diabetes on a high-fatdiet. Endocrinology 146(2):920–930.

19. Vasavada RC, et al. (2000) Targeted expression of placental lactogen in the beta cellsof transgenic mice results in beta cell proliferation, islet mass augmentation, andhypoglycemia. J Biol Chem 275(20):15399–15406.

20. Smaniotto S, et al. (2005) Growth hormone modulates thymocyte development invivo through a combined action of laminin and CXC chemokine ligand 12. Endocri-nology 146(7):3005–3017.

21. Lin MS, et al. (2011) A multivalent vaccine for type 1 diabetes skews T cell subsets toTh2 phenotype in NOD mice. Immunol Res 50(2-3):213–220.

22. Martin-Orozco N, et al. (2009) T helper 17 cells promote cytotoxic T cell activation intumor immunity. Immunity 31(5):787–798.

23. Charré S, et al. (2002) Abnormalities in dendritic cell and macrophage accumulation inthe pancreas of nonobese diabetic (NOD) mice during the early neonatal period.Histol Histopathol 17(2):393–401.

24. Wu Y, et al. (2011) Growth hormone receptor regulates β cell hyperplasia and glu-cose-stimulated insulin secretion in obese mice. J Clin Invest 121(6):2422–2426.

25. Valera A, et al. (1993) Glucose metabolism in transgenic mice containing a chimeric P-enolpyruvate carboxykinase/bovine growth hormone gene. FASEB J 7(9):791–800.

26. Winer S, et al. (2003) Autoimmune islet destruction in spontaneous type 1 diabetes isnot beta-cell exclusive. Nat Med 9(2):198–205.

27. Nakayama M, et al. (2005) Prime role for an insulin epitope in the development oftype 1 diabetes in NOD mice. Nature 435(7039):220–223.

28. Kent SC, et al. (2005) Expanded T cells from pancreatic lymph nodes of type 1 diabeticsubjects recognize an insulin epitope. Nature 435(7039):224–228.

29. Brode S, Raine T, Zaccone P, Cooke A (2006) Cyclophosphamide-induced type-1diabetes in the NOD mouse is associated with a reduction of CD4+CD25+Foxp3+regulatory T cells. J Immunol 177(10):6603–6612.

30. Komatsu N, Hori S (2007) Full restoration of peripheral Foxp3+ regulatory T cell poolby radioresistant host cells in scurfy bone marrow chimeras. Proc Natl Acad Sci USA104(21):8959–8964.

31. D’Alise AM, et al. (2008) The defect in T-cell regulation in NODmice is an effect on theT-cell effectors. Proc Natl Acad Sci USA 105(50):19857–19862.

32. Gregori S, Giarratana N, Smiroldo S, Adorini L (2003) Dynamics of pathogenic andsuppressor T cells in autoimmune diabetes development. J Immunol 171(8):4040–4047.

33. Kim D-H, et al. (2011) Inhibition of autoimmune diabetes by TLR2 tolerance.J Immunol 187(10):5211–5220.

34. Tarbell KV, Yamazaki S, Steinman RM (2006) The interactions of dendritic cells withantigen-specific, regulatory T cells that suppress autoimmunity. Semin Immunol 18(2):93–102.

35. You S, et al. (2008) Immunoregulatory pathways controlling progression of autoim-munity in NOD mice. Ann N Y Acad Sci 1150:300–310.

36. Manirarora JN, Kosiewicz MM, Parnell SA, Alard P (2008) APC activation restoresfunctional CD4(+)CD25(+) regulatory T cells in NOD mice that can prevent diabetesdevelopment. PLoS ONE 3(11):e3739.

37. Savage NDL, et al. (2008) Human anti-inflammatory macrophages induce Foxp3+ GITR+CD25+ regulatory T cells, which suppress via membrane-bound TGFbeta-1. J Immunol181(3):2220–2226.

38. Bending D, et al. (2009) Highly purified Th17 cells from BDC2.5NOD mice convert intoTh1-like cells in NOD/SCID recipient mice. J Clin Invest 119(3):565–572.

39. McGeachy MJ, et al. (2007) TGF-beta and IL-6 drive the production of IL-17 and IL-10by T cells and restrain T(H)-17 cell-mediated pathology. Nat Immunol 8(12):1390–1397.

40. Peng H, Hagopian W (2006) Environmental factors in the development of Type 1diabetes. Rev Endocr Metab Disord 7(3):149–162.

41. Yuen KC, Chong LE, Riddle MC (2013) Influence of glucocorticoids and growth hor-mone on insulin sensitivity in humans. Diabet Med 30(6):651–663.

42. Hill DJ, Petrik J, Arany E, McDonald TJ, Delovitch TL (1999) Insulin-like growth factorsprevent cytokine-mediated cell death in isolated islets of Langerhans from pre-diabetic non-obese diabetic mice. J Endocrinol 161(1):153–165.

43. Whittaker PG, Taylor KW (1980) Direct effects of rat growth hormone in rat islets oflangerhans in tissues culture. Diabetologia 18(4):323–328.

44. Rhodes CJ (2000) IGF-I and GH post-receptor signaling mechanisms for pancreaticbeta-cell replication. J Mol Endocrinol 24(3):303–311.

45. Flodström-Tullberg M, et al. (2003) Target cell expression of suppressor of cytokinesignaling-1 prevents diabetes in the NOD mouse. Diabetes 52(11):2696–2700.

46. Jackerott M, et al. (2006) STAT5 activity in pancreatic beta-cells influences the severityof diabetes in animal models of type 1 and 2 diabetes. Diabetes 55(10):2705–2712.

47. Clark R (1997) The somatogenic hormones and insulin-like growth factor-1: Stim-ulators of lymphopoiesis and immune function. Endocr Rev 18(2):157–179.

48. Alard P, et al. (2006) Deficiency in NOD antigen-presenting cell function may be re-sponsible for suboptimal CD4+CD25+ T-cell-mediated regulation and type 1 diabetesdevelopment in NOD mice. Diabetes 55(7):2098–2105.

49. Serreze DV, et al. (1998) B lymphocytes are critical antigen-presenting cells forthe initiation of T cell-mediated autoimmune diabetes in nonobese diabetic mice.J Immunol 161(8):3912–3918.

50. Wan YY, Flavell RA (2007) Regulatory T-cell functions are subverted and convertedowing to attenuated Foxp3 expression. Nature 445(7129):766–770.

51. Gebert CA, Park SH, Waxman DJ (1997) Regulation of signal transducer and activatorof transcription (STAT) 5b activation by the temporal pattern of growth hormonestimulation. Mol Endocrinol 11(4):400–414.

52. Pallandre J-R, et al. (2007) Role of STAT3 in CD4+CD25+FOXP3+ regulatory lympho-cyte generation: Implications in graft-versus-host disease and antitumor immunity.J Immunol 179(11):7593–7604.

53. Chen BJ, et al. (2010) Growth hormone mitigates against lethal irradiation and en-hances hematologic and immune recovery in mice and nonhuman primates. PLoSONE 5(6):e11056.

E4626 | www.pnas.org/cgi/doi/10.1073/pnas.1314985110 Villares et al.

Dow

nloa

ded

by g

uest

on

June

21,

202

1

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1314985110/-/DCSupplemental/pnas.201314985SI.pdf?targetid=nameddest=ST2www.pnas.org/cgi/doi/10.1073/pnas.1314985110

54. Martinez FO, Sica A, Mantovani A, Locati M (2008) Macrophage activation and po-larization. Front Biosci 13:453–461.

55. Gordon S (2003) Alternative activation of macrophages. Nat Rev Immunol 3(1):23–35.56. Zizzo G, Cohen PL (2013) IL-17 stimulates differentiation of human anti-inflammatory

macrophages and phagocytosis of apoptotic neutrophils in response to IL-10 andglucocorticoids. J Immunol 190(10):5237–5246.

57. Komiyama Y, et al. (2006) IL-17 plays an important role in the development of ex-perimental autoimmune encephalomyelitis. J Immunol 177(1):566–573.

58. Hirota K, et al. (2007) Preferential recruitment of CCR6-expressing Th17 cells toinflamed joints via CCL20 in rheumatoid arthritis and its animal model. J Exp Med204(12):2803–2812.

59. Ghoreschi K, Laurence A, Yang XP, Hirahara K, O’Shea JJ (2011) T helper 17 cellheterogeneity and pathogenicity in autoimmune disease. Trends Immunol 32(9):395–401.

60. Marwaha AK, Leung NJ, McMurchy AN, Levings MK (2012) TH17 Cells in Autoim-munity and Immunodeficiency: Protective or pathogenic? Front Immunol 3:129.

61. Wang R, et al. (2007) The role of STAT3 in antigen-IgG inducing regulatoryCD4(+)Foxp3(+)T cells. Cell Immunol 246(2):103–109.

62. Gaba A, et al. (2012) Cutting edge: IL-10-mediated tristetraprolin induction is part ofa feedback loop that controls macrophage STAT3 activation and cytokine production.J Immunol 189(5):2089–2093.

63. Hoang PT, et al. (2010) The neurosurvival factor Humanin inhibits beta-cell apoptosisvia signal transducer and activator of transcription 3 activation and delays andameliorates diabetes in nonobese diabetic mice. Metabolism 59(3):343–349.

64. Mori H, et al. (2007) Suppression of SOCS3 expression in the pancreatic beta-cell leadsto resistance to type 1 diabetes. Biochem Biophys Res Commun 359(4):952–958.

65. Cecim M, Kerr J, Bartke A (1995) Effects of bovine growth hormone (bGH) transgeneexpression or bGH treatment on reproductive functions in female mice. Biol Reprod52(5):1144–1148.

66. Dietrich WF, et al. (1996) A comprehensive genetic map of the mouse genome. Na-ture 380(6570):149–152.

67. Montanya E, Téllez N (2009) Pancreatic remodeling: Beta-cell apoptosis, proliferationand neogenesis, and the measurement of beta-cell mass and of individual beta-cellsize. Methods Mol Biol 560:137–158.

68. Ye J, et al. (2012) Primer-BLAST: A tool to design target-specific primers for poly-merase chain reaction. BMC Bioinformatics 13:134.

Villares et al. PNAS | Published online November 11, 2013 | E4627

IMMUNOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

21,

202

1