Long-acting progestin-only contraceptives impairendometrial vasculature by inhibiting uterinevascular smooth muscle cell survivalUmit A. Kayislia,1, Murat Basarb, Ozlem Guzeloglu-Kayislia, Nihan Semercia, Helen C. Atkinsonc, John Shapirob,Taryn Summerfieldb, S. Joseph Huangb, Katja Prelled, Frederick Schatza, and Charles J. Lockwooda,1

aDepartment of Obstetrics & Gynecology, Morsani College of Medicine, University of South Florida, Tampa, FL 33612; bDepartment of Obstetrics &Gynecology, College of Medicine, The Ohio State University, Columbus, OH 43210; cDepartment of Pediatric Endocrinology Research, School of Women’s &Infants’ Health, University of Western Australia, Crawley, WA 6009, Australia; and dDepartment of Safety Pharmacology, Bayer HealthCare, GDD-GED-TOX-Safety Pharmacology, 42096 Wuppertal, Germany

Edited by Jan-Åke Gustafsson, University of Houston, Houston, TX, and approved March 10, 2015 (received for review December 29, 2014)

Molecular mechanisms responsible for abnormal endometrialvasculature in women receiving long-acting progestin-only con-traceptives (LAPCs) are unknown. We hypothesize that LAPCsimpair vascular smooth muscle cell (VSMC) and pericyte prolifera-tion and migration producing thin-walled hyperdilated fragilemicrovessels prone to bleeding. Proliferating cell nuclear antigen(PCNA) and α-smooth muscle actin (αSMA) double-immunostainingassessed VSMC differentiation and proliferation in endometria fromwomen before and after DepoProvera (Depo) treatment and fromoophorectomized guinea pigs (OVX-GPs) treated with vehicle, estra-diol (E2), medroxyprogesterone acetate (MPA), or E2+MPA. Whole-genome profiling, proliferation, and migration assays were per-formed on cultured VSMCs treated with MPA or etonogestrel(ETO). Endometrial vessels of Depo-administered women displayedreduced αSMA immunoreactivity and fewer PCNA (+) nuclei amongαSMA (+) cells (P < 0.008). Microarray analysis of VSMCs identifiedseveral MPA- and ETO-altered transcripts regulated by STAT1 sig-naling (P < 2.22 × 10−6), including chemokine (C-C motif) ligand 2(CCL2). Both MPA and ETO reduce VSMC proliferation and migration(P < 0.001). Recombinant CCL2 reversed this progestin-mediatedinhibition, whereas a STAT1 inhibitor abolished the CCL2 effect.Similarly, the endometria of MPA treated OVX-GPs displayed de-creased αSMA staining and fewer PCNA (+) nuclei in VSMC (P <0.005). In conclusion, LAPCs promote abnormal endometrial vesselformation by inhibiting VSMC proliferation and migration.

progestin contraceptives | abnormal uterine bleeding | VSMC |impaired vascular maturation | proliferation

Long-acting progestin-only contraceptives (LAPCs) are safe,discrete, effective, and can be used by women in whom es-trogen use is contraindicated (1, 2). However, associated ab-normal uterine bleeding (AUB) reduces adherence (1–3). Theendometria of women receiving LAPCs contain elevated levels ofVEGF and angiopoietin-2 (Ang-2) (4). Like women, the guineapig (GP) exhibits spontaneous estrous cycling and hemochorialplacentation (5–7) and is a relevant animal model to study LAPC-induced AUB (1, 8–10). In humans and GPs, this hyperangiogenicmilieu creates endometria displaying superficial, irregularly dis-tributed enlarged, thin-walled capillaries and venules (2, 9, 10).Vascular smooth muscle cells (VSMCs) and pericytes main-

tain vascular tone, a critical determinant of vessel volume andblood flow (11). Endothelial cells of arteries and veins are sur-rounded by VSMCs, whereas those of capillaries are surroundedby pericytes (12). Completion of angiogenesis requires envel-opment of new vessels by pericytes or VSMCs, which expressα-smooth muscle actin (αSMA) and tropomyosin (11–13).Currently available LAPCs include Depo-Provera (Depo), an

injectable form of medroxyprogesterone acetate (MPA) andImplanon, a subdermally implanted rod that releases etonoges-trel (ETO) (3). Women administered LAPCs display significantly

reduced endometrial blood flow (14), consistent with constrictionof uterine arteries and arterioles. Although ETO administrationconstricts GP uterine radial arteries (8), it paradoxically induceshyperdilation in microvasculature in the functional endometrium(2, 9, 10). We postulate that LAPCs suppress VSMC and pericytedifferentiation, proliferation, and migration to create fragile, thin-walled hyperdilated vessels prone to bleed.

ResultsDecreased Endometrial VSMC Proliferation in Women ReceivingDepo. VSMC numbers and proliferation index were evaluatedin microvessels and spiral arteries of paired endometrial sec-tions from women before and after Depo (injectable MPA)after double-immunostaining for αSMA and proliferating cellnuclear antigen (PCNA). Staining intensity and numbers ofαSMA+ cells around microvessels (Fig. 1 A vs. B) and spiralarteries (Fig. 1 C vs. D) were lower in post- vs. pre-Depo–treated endometria (mean ± SEM: 117.8 ± 14.4 vs. 205.8 ±22.0, respectively; P < 0.008) (Fig. 1E). The VSMC proliferationindex was significantly lower in post- (Fig. 1 B and D) vs. pre-Depo

Significance

Over a million unintended pregnancies occur in the UnitedStates each year because of either discontinuation or misuse ofcontraceptives. The major reason for discontinuation of long-acting progestin-only contraceptives (LAPCs) is the occurrenceof abnormal uterine bleeding (AUB). Uncovering the mecha-nisms underlying LAPC-induced AUB is essential to prevent theirdiscontinuation. We found that LAPCs reduce proliferation ofhuman and guinea pig endometrial vascular smooth musclecells (VSMCs), resulting in production of thin-walled hyper-dilated fragile microvessels. In cultured VSMCs, chemokine (C-Cmotif) ligand 2 reverses LAPC-mediated inhibition of VSMCproliferation, suggesting that LAPCs impair endometrial vascu-lar integrity and that chemokine ligand 2 administration mayprevent LAPC-induced AUB.

Author contributions: U.A.K., M.B., O.G.-K., H.C.A., F.S., and C.J.L. designed research; U.A.K.,M.B., O.G.-K., N.S., H.C.A., J.S., T.S., S.J.H., and K.P. performed research; U.A.K., M.B., O.G.-K.,N.S., H.C.A., J.S., T.S., S.J.H., K.P., F.S., and C.J.L. analyzed data; and U.A.K., M.B., O.G.-K., N.S.,H.C.A., J.S., T.S., S.J.H., K.P., F.S., and C.J.L. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.

Data deposition: The data reported in this paper have been deposited in the Gene Ex-pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE55736).1To whom correspondence may be addressed. Email: uakayisli@health.usf.edu orcjlockwood@health.usf.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1424814112 PNAS | April 21, 2015 | vol. 112 | no. 16 | 5153–5158

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://crossmark.crossref.org/dialog/?doi=10.1073/pnas.1424814112&domain=pdfhttp://www.ncbi.nlm.nih.gov/geohttp://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE55736mailto:uakayisli@health.usf.edumailto:cjlockwood@health.usf.eduhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplementalhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplementalwww.pnas.org/cgi/doi/10.1073/pnas.1424814112

endometria (Fig. 1 A and C) (24.7% ± 3.5 vs. 57.8% ± 6.9;P < 0.003) (Fig. 1F).

LAPC-Regulated Genes Are Involved in VSMC Differentiation andProliferation. Microarray analysis revealed that MPA and ETOaltered transcription of 793 and 191 genes, respectively, in culturedVSMCs, whereas progesterone (P4) modified only 40 genes. (Seecomplete microarray results at www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE55736 in the Gene Expression Omnibus re-pository using access no. GSE55736). Of 793 genes differentiallyregulated byMPA, 392 are up-regulated and 401 are down-regulated.Of 191 genes differentially regulated by ETO, 82 are up-regulatedand 109 are down-regulated. Of 40 genes differentially regulatedby P4, 18 are up-regulated and 22 are down-regulated. TablesS1–S3 display the 40 most progestin-regulated genes. The Venndiagram displaying microarray data revealed that 612 genes areregulated solely by MPA, 13 solely by ETO, 16 solely by P4, and15 genes by all three progestins (Fig. 2). Of ETO-modulated genes,92% (175) are also regulated by MPA (Fig. 2) with the same di-rectional change [i.e., MPA and ETO induce transmembraneprotein 119 (TMEM119), a suppressor of myoblast differentiation,and suppress retinoic acid receptor responder (tazarotene induced)1 (RARRES1), a proliferation inducer] (Tables S1 and S2).

Among MPA- and ETO-regulated genes, H19 imprinted ma-ternally expressed transcript, noncoding RNA (H19), TMEM119,RARRES1, and chemokine (C-C motif) ligand 2 (CCL2) mediatedifferentiation, proliferation, and migration of VSMCs (15–18) andwere thus chosen for quantitative RT-PCR (qRT-PCR) confir-mation. In parallel with the microarray results (Tables S1 and S2),MPA or ETO up-regulated H19 mRNA and TMEM119 mRNA,and down-regulated RARRES1 mRNA and CCL2 mRNA levels(Fig. 2B). Further evaluation of microarray data by IngenuityPathway Analysis (IPA) software predicts that STAT1 inhibitionis associated with genes regulated in VSMCs by MPA or ETO(z-score and the overlap P value; −3.06, P < 8.5 × 10−14 and −2.43,P < 2.2 × 10−6, respectively) (Fig. 3A). Confirming this prediction,compared with control VSMCs, Western blotting found that MPAand ETO partially reduced phosphorylated but not total STAT1levels (Fig. 3B).

LAPC Inhibition of VSMC Proliferation Is Reversed by CCL2 via STAT1Signaling in Culture. In cultured VSMCs, 48-h treatment withMPA or ETO significantly reduced BrdU incorporation vs.control (mean ± SEM 0.31 ± 0.02 or 0.40 ± 0.03 vs. 0.81 ± 0.03respectively; P < 0.001), whereas P4 (0.71 ± 0.05) was ineffective(Fig. 4A). In contrast, none of the progestins altered apoptosis

Fig. 1. Reduced αSMA and PCNA immunoreactivity in VSMCs of endometria of women receiving Depo-Provera (Depo, Depot medroxyprogesterone acetate).Immunoreactivity for αSMA (red) and PCNA (brown) in VSMCs (arrowheads) surrounding pre- (A) vs. post-Depo (B) endometrial microvessels (single asteriskdenotes lumen); and pre- (C) vs. post-Depo (D) endometrial spiral arteries (double asterisks denote lumen). Note reduced αSMA immunoreactivity and fewerPCNA (+) cells among VSMCs in post-Depo administered endometria (B and D) vs. their corresponding pairs (A and C). Graphs display αSMA HSCOREs (E) andpercentage of PCNA positively stained nuclei (F) in VSMCs. Bars (n = 6) represent mean ± SEM; *P < 0.008 and #P < 0.003. (Scale bar, 40 μm.)

Fig. 2. Differentially regulated genes in VSMCs by progestins. Venn diagram analysis of genes regulated individually or in common by MPA or ETO or P4 (A).qRT-PCR analysis for regulation of H19, TMEM119, RARRES1, and CCL2 mRNA in cultured VSMC by progestins (B). Bars represent mean ± SEM (n = 3).C, control; P, progesterone (P4).

5154 | www.pnas.org/cgi/doi/10.1073/pnas.1424814112 Kayisli et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE55736http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE55736http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental/pnas.201424814SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental/pnas.201424814SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental/pnas.201424814SI.pdf?targetid=nameddest=ST3http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental/pnas.201424814SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental/pnas.201424814SI.pdf?targetid=nameddest=ST2http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental/pnas.201424814SI.pdf?targetid=nameddest=ST1http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental/pnas.201424814SI.pdf?targetid=nameddest=ST2www.pnas.org/cgi/doi/10.1073/pnas.1424814112

(Fig. 4B and Fig. S1). To determine if inhibition of VSMCproliferation by MPA or ETO is mediated by decreasedCCL2 expression, VSMCs were treated with vehicle or MPA orETO ± recombinant CCL2 (rCCL2) for 72 h. Compared withcontrol (0.86 ± 0.03), rCCL2 elicited a concentration dependentincrease in BrdU incorporation (1.05 ± 0.05 and 1.21 ± 0.05, re-spectively; P < 0.005) (Fig. 4C). Addition of 10 ng/mL rCCL2 re-versed the inhibitory effects of MPA or ETO, whereas 100 μmMTA, a specific STAT1 inhibitor (19), abolished these rCCL2 ef-fects (Fig. 4C).

LAPCs Inhibit VSMC Migration in Vitro. Compared with control(migration rate: 48.4%), 12-h incubation with MPA or ETOor P4 reduced migration of VSMCs (migration rate: 18.7% or29.6% or 39.4%, respectively; P < 0.001) (Fig. 5). In coincuba-tions, rCCL2 did not reverse inhibition of VSMC migration byETO or P4, but partially reversed the MPA effect (Fig. 5).

MPA Administration Decreases Endometrial VSMC Numbers inOophorectomized GPs. Endometrial sections from control, E2,E2+MPA, or MPA-treated oophorectomized (OVX) GPs weredouble-immunostained for αSMA and PCNA. Controls displayedmoderate to weak discontinuous αSMA immunoreactivity in vesselsacross the endometrium (Fig. 6 A and B). In comparison, E2-administered GPs displayed elevated numbers of αSMA immunore-active VSMCs with intense and continuous immunostaining acrossthe endometrium (Fig. 6 C and D). Addition of MPA blunted theseE2 effects and resulted in discontinuous αSMA immunostainingalong vessels in the basal layer (Fig. 6 C vs. E) and absence ofimmunoreactive cells around superficial layer vessels (Fig. 6F).Similarly, αSMA immunoreactivity was weaker (Fig. 6 A vs. G) andvirtually absent around superficial endometrial vessels in GPs ad-ministered MPA alone vs. control (Fig. 6 B vs. H). Semiquantitativeevaluation revealed that E2 elicited higher αSMA histologicalscoring (HSCOREs) vs. control (mean ± SEM; 212.7 ± 14.6 vs..135.2 ± 26.0, P < 0.005) (Fig. 7A), whereas E2+MPA significantlylowered αSMA HSCOREs vs. E2 alone (116.7 ± 13.84 vs.. 212.7 ±14.6, respectively; P < 0.005) (Fig. 7A). Similar effects were ob-served following administration of MPA alone vs. E2 alone (83.50 ±9.0 vs.. 212.7 ± 14.6, respectively; P < 0.001) (Fig. 7A). In double-stained sections, PCNA+ endometrial VSMC numbers are signifi-cantly higher in the E2 vs. control group (mean ± SEM; 36.8% ± 2.7vs.. 11.2% ± 2.3; P < 0.001) (Figs. 6 A vs. C or B vs. D and 7B).Administration of E2+MPA blocked E2-enhanced endometrial

VSMC proliferation (11.3% ± 2.5 vs. 36.8% ± 2.7, respectively; P <0.001) (Figs. 6 C vs. E or D vs. F and 7B). Administration of MPAalone vs. E2 produced similar effects (10.2% ± 2.8 vs. 36.8% ± 2.7,respectively; P < 0.001) (Figs. 6 C vs. G or D vs. H and 7B).

DiscussionThe endometria of women receiving LAPCs display irregularlydistributed hyperdilated thin-walled vessels with deficient base-ment membranes that are the site of AUB (14, 20). This study

Fig. 3. MPA and ETO inhibit STAT1 signaling in VSMCs. IPA identified a cluster of differentially regulated genes by MPA (Left) or ETO (Right) associated withinhibition of STAT1 signaling (A). Immunoblotting of total (To-) and phosphorylated (Ph-) STAT1 in VSMCs treated with control, ETO and MPA for 15 min (B).All gene symbols were abbreviated according to GenBank standard nomenclature.

Fig. 4. LAPCs inhibit VSMC proliferation. BrdU incorporation (A) or apo-ptotic index (B) in VSMCs treated with control (vehicle), 10−7 M P4, 10

−7 METO, or 10−7 M MPA for 48 h. BrdU incorporation in VSMCs treated withcontrol (vehicle), 1 or 10 ng/mL CCL2 or 10−7 M P4 or 10

−7 M ETO or 10−7 MMPA ± CCL2 (10 ng) ± 100 μM MTA (Stat1 inhibitor) for 72 h (C). Barsrepresent mean ± SEM, n = 3 with four replicates per experiment.

Kayisli et al. PNAS | April 21, 2015 | vol. 112 | no. 16 | 5155

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental/pnas.201424814SI.pdf?targetid=nameddest=SF1

reveals a novel mechanism that accounts for the development ofthis abnormal vasculature. Integration of our in vitro findings inhuman VSMCs with in situ observations in MPA treated OVX-GPs and in post-Depo (injectable form of MPA) -treated humanendometria demonstrated the inhibitory role of the progestincomponent of LAPCs on VSMC proliferation and migration.This accounts for the hyperdilated microvessels in the superficialendometrium, which are devoid of normal VSMC layer and ex-hibit increased susceptibility to bleeding.Previous studies investigated VSMC characteristics in the

endometrium of normal cycling and LAPC-treated women (21–23). As an early differentiation marker, αSMA is expressed byVSMCs of endometrial vessels, by pericytes attached to capil-laries, and by stromal myofibroblasts (21, 22). Increased αSMAexpression during the secretory phase (22) suggests a correlationbetween endometrial vascular maturation with αSMA expres-sion. In support of this thesis, the VSMC proliferation index (24)and the number of endometrial vessels enveloped by more thanone VSMC layer increases from the proliferative to secretoryphase (25). Moreover, αSMA expression is decreased aroundendometrial microvasculature of Norplant users exhibiting bleeding(23). Building upon these prior reports, PCNA and αSMA double-immunolabeling now reveals a significant decrease in endometrialVSMC proliferation index, accounting for the reduction in VSMCnumbers, reduced vascular wall thickness, and impaired structuralintegrity responsible for the hyperdilated thin-walled endometrialvessels seen in women receiving Depo. These pathological changesare consistent with lower VSMC proliferation in menorrhagic vs.normal endometrium (24). In mice, vascular leakage and bleedingfrom enlarged, thin-walled capillaries results from pericyte de-pletion caused by knockout of PDGF subunit B or PDGF receptorβ genes (26, 27), indicating the crucial role played by pericyte/VSMCs in maintaining vascular integrity.Given the lack of availability human endometrial VSMCs, we

used progesterone receptor-expressing VSMCs isolated from hu-man female aortas (28). Our results revealed both common andunique responses of VSMCs to P4, ETO, and MPA. Strikingly, wefound much greater numbers of genes modulated by MPA com-pared with ETO or P4. This strong impact of MPA on gene reg-ulation likely accounts for its greater local and systemic effects.However, we focused on common genes regulated by ETO andMPA because AUB is a common side-effect seen in most LAPCusers. Thus, we confirmed mRNA levels of H19 (15), TMEM119(17), RARRES1 (18), andCCL2 (16) because these genes are knownto mediate VSMC differentiation, proliferation, and migration.The present study demonstrates in situ inhibition of endo-

metrial VSMC proliferation in post-Depo users, and in VSMCmonolayers treated with MPA or ETO. This inhibition occurs in

parallel with down-regulation of CCL2 by both progestins.Conversely, induction of VSMC proliferation by CCL2 (16, 29)suggests that this cytokine could act as an adjuvant treatment toreverse LAPC inhibition of VSMC proliferation. This strategy issupported by the current finding that rCCL2 reverses MPA andETO mediated inhibition of VSMC proliferation in vitro.Results of IPA and immunoblotting demonstrated that MPA-

and ETO-suppressed STAT1 signaling mediates inhibition ofVSMC proliferation. In support of this mechanism, the STAT1inhibitor, MTA, abolishes proliferative effects of CCL2 onprogestin-treated VSMCs. These observations are consistent withreports that: (i) the CCL2 receptor CCR2 is present in human andrat VSMCs (30, 31), (ii) STAT1 signaling induces VSMC pro-liferation (32), and (iii) CCR2 activates STAT1 signaling (33).Compared with other rodents, GPs share multiple reproductive

tract anatomic and physiologic characteristics with humans (6, 7).Our previous studies (8, 9) demonstrated that, as in human endo-metria, LAPC administration to GPs results in such endometrialvascular changes as focal hemorrhage, increased oxidative stress,high apoptotic indices, and elevated lipid peroxidation. Similarly, asin endometria of women receiving Depo, the present study con-firmed reduced numbers of αSMA (+) cells and decreased VSMC

Fig. 5. LAPCs reduces VSMC migration. VSMCs treated with vehicle (control)or P4 or ETO or MPA in the presence or absence of 10 ng/mL rCCL2 seeded oninsert transwell membranes and allowed to migrate to lower wells for 12 h.Bars represent mean ± SEM, n = 3 with three replicates in each independentexperiments; *P < 0.05 and #P < 0.001.

Fig. 6. Endometria of OVX-GPs treated by MPA display reduced αSMA andPCNA immunoreactivity in VSMC. In double-immunostained sections, im-munoreactivity for αSMA (red) and PCNA (brown) in VSMCs surroundingvessels throughout the endometrium of OVX-GPs treated with vehicle(A and B), E2 (C and D), E2+MPA (E and F), or MPA (G and H) for 21 d. Startsindicate vessel lumens. Note that discontinuous αSMA immunoreactivity inOVX-GPs (B) becomes continuous and increases by the E2 administration (D),whereas immunoreactive αSMA disappears in superficial endometrial vesselsby E2+MPA (F) or MPA (H) administration. Greater numbers of PCNA (+)nuclei (brown) among VSMCs (red) are present in endometrium from GPsfollowing E2 administration compared with vehicle control (A and B),E2+MPA (E and F), or MPA (G and H) administration. (Scale bars, 40 μm.)

5156 | www.pnas.org/cgi/doi/10.1073/pnas.1424814112 Kayisli et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

021

www.pnas.org/cgi/doi/10.1073/pnas.1424814112

proliferation in MPA-treated GP endometria, and confirms that inGPs treated with MPA, the majority of superficial endometrialvessels are devoid of VSMCs.Although decreased proliferation and migration or increased

apoptosis may contribute to the decreased density of the VSMClayer observed in the superficial endometria of LAPC users, wenow show that apoptosis is unlikely to contribute to this abnor-mal vasculature because the apoptotic index was unaffected byMPA or ETO or P4. The partial reversal of MPA but not ETO orP4 effects on VSMC migration by rCCL2 is likely a consequenceof the mixed progestin-glucocorticoid properties of MPA vs.ETO, a purer progestin, or P4. Alternatively, the greater numberof genes regulated by MPA compared with ETO or P4 maymediate this partial reversal.In conclusion, our results reveal that LAPCs induce AUB by

unique molecular and cellular mechanisms that impair endo-metrial VSMC proliferation, migration, or differentiation, andthus produce insufficient VSMCs to maintain vascular integrity.This novel process accounts for emergence of the thin-walledhyperdilated fragile vessels and focal hemorrhage seen in theendometria of LACP users (Fig. 8). Our results also suggest thatcoadministration of CCL2 may prevent LAPC-induced AUB.

Materials and MethodsTissues. After receiving written informed consent at New York Universityunder Institutional Review Board approval, paraffin sections were derivedfrom banked paired endometrial tissues obtained before and after Depotherapy by Pipelle biopsy from each reproductive age woman with regularmenstrual cycles during the secretory phase (n = 6). Based on our establishedanimal model for LAPC-induced AUB in humans (9), nulliparous adult femalepigmented GPs (weighing 525–775 g) were used in experimental proceduresconforming to National Health and Medical Research Council Guidelines ofAustralia and approved by the Animal Ethics Committee of The University ofWestern Australia. GPs (n = 24) were subjected to bilateral OVX and givensubcutaneous 21-d time-release pellets (Innovative Research of America)containing MPA (2.5 mg/d; n = 6) or 17β-Estradiol (E2) (0.1 mg/d; n = 6), bothMPA and E2 (n = 6), or no pellets (n = 6). On day 20 of exposure tothe pellets, GPs were killed with pentobarbital and uterine tissues fixed in10% (vol/vol) formalin and processed for paraffin embedding. In vitroexperiments used aortic VSMCs from premenopausal women (n = 2) (LifeTechnologies).

Immunohistochemistry. Paraffin sections (5 μm) from paired pre- vs. post-Depo treated human or GP endometria were deparaffinized, boiled in cit-rate buffer (pH: 6.0) and then incubated in methanol containing 3% (vol/vol)H2O2. Following 10% (vol/vol) normal horse serum blocking (Vector Labs),slides were incubated overnight at 4 °C with mouse anti-PCNA monoclonalantibody (EMD Millipore). After several rinses in Tris-buffered salinecontaining 0.1% tween-20 (TBS-T; pH 7.6) slides were then incubated withbiotinylated anti-mouse IgG (Vector Labs) for 30 min, then with a strepta-vidin-peroxidase complex (Elite ABC kit; Vector Labs) for 30 min and withdiaminobenzidine (DAB; Vector Labs) for 3 min. These steps, unless other-wise specified, were repeated for double-immunolabeling using 10% (vol/vol)normal goat serum, rabbit anti-αSMA polyclonal antibody for 1 h (Abcam), agoat biotinylated anti-rabbit IgG (Vector Labs), and a streptavidin-alkalinephosphatase (ABC kit) and Vector Red (Vector Labs). For negative controls,normal mouse IgG2a and rabbit IgG were used at the same concentrations asthe primary antibodies. Slides were counterstained by Hematoxylin and eval-uated by histological scoring (HSCORE), as described previously (34). (See detailsfor HSCORE evaluation for αSMA immunoreactivity and PCNA quantitation inSI Materials and Methods).

Cell Cultures and Experimental Incubations. VSMCs were thawed and culturedin medium 231(M231; Life Technologies) with smooth muscle growth sup-plement [5% (vol/vol) FBS; 2 ng/mL human basic fibroblast growth factor,0.5 ng/mL human epidermal growth factor, 5 ng/mL heparin, 0.2 μg/mL insulinand 5 μg/mL BSA; Life Technologies] with 100 U/mL penicillin, 100 μg/mLstreptomycin, 0.25 μg/mL fungizone. Confluent VSMCs were washed twicewith 1× PBS to remove residual serum, passaged in 96-well or 6-well plates

Fig. 8. Model of aberrant vascular transformation and related AUB in women administered LAPCs. In normal endometrium, estradiol (E2) and progesterone(P4) regulate angiogenesis as well as VSMC proliferation and migration. Coordination of these steroid-mediated effects triggers vascular maturation resultingin a tight and continuous layer of VSMCs lying beneath endothelial cells of new vessels, which provides normal blood flow to induce endometrial growth. Incontrast, long-term effects on human endometrium by synthetic progestins elicit: (i) reduced blood flow which induces local hypoxia; (ii) induced decidu-alization and increased expression of such angiogenic factors as VEGF, IL8, and Ang-2 in stromal cells; and (iii) inhibited VSMC proliferation and migration byblocking CCL2-mediated STAT1 signaling. The first two mechanisms promote excess angiogenesis, whereas the third mechanism results in insufficient VSMCnumbers to cover and surround newly formed vessels. Consequently, impaired vascular maturation generates thin-walled hyperdilated fragile vessels whichare prone to leakage and AUB.

Fig. 7. Semiquantitative analysis of αSMA and PCNA immunoreactivity inendometrium of OVX GPs. Graphs represents HSCORE values for αSMA ex-pression (A) and percentage of PCNA (+) VSMCs (B). Bars represent mean ±SEM; n = 6 for each group; *P < 0.005 or #P < 0.001.

Kayisli et al. PNAS | April 21, 2015 | vol. 112 | no. 16 | 5157

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental/pnas.201424814SI.pdf?targetid=nameddest=STXT

and incubated in parallel in M231 with 1% (wt/vol) charcoal striped FBS with0.1% ethanol (vehicle, control) or 10−7 M MPA (Sigma-Aldrich) or 10−7 METO (Organon) or 10−7 M P4 (Sigma-Aldrich) for 15 min, 6, 48, and 72 h.VSMCs were then washed with ice-cold PBS and stored at −80 °C for furtherRNA and protein analyses.

Cell Proliferation and Apoptosis Assay. VSMCs were grown in 96-well plates or8-well chamber slides and incubated with 10−7 M of MPA (Sigma-Aldrich) orETO (Organon) or P4 (Sigma-Aldrich) or vehicle ± rCCL2 (R&D Systems) ±STAT1 inhibitor, 5′-methyl-thioadenosine (MTA; EMD Millipore) (19) for 48or 72 h. VSMCs were then analyzed for apoptosis by an ELISA based apo-ptosis kit (Roche) or cleaved-caspase 3 immunostaining. VSMC proliferationwas measured by BrdU incorporation Kit (Cell Signaling). (See further detailsin SI Materials and Methods).

Microarray Analysis. Total RNA was isolated using Qiagen miRNeasy Mini Kit(Qiagen) and RNeasy MinElute Cleanup Kit (Qiagen). The quality of isolatedRNAs was confirmed using an Agilent 2100 Bioanalyzer. Microarray analysiswas limited to those specimens in which RNA Integrity Number exceeded 8.Microarray processes were performed in the Keck Biotechnology ResourceLaboratory at Yale University (NewHaven, CT) using Illumina HumanHT-12 v4BeadChip (Illumina). Raw data without normalization were analyzed byGeneSpring GX12.5 software (Agilent Technologies-Silicon Genetics). Normali-zation was performed as described (35). Genes with a fold-change > 1.25 anda P value < 0.05 were considered differentially expressed. IPA software ex-plored molecular and biological networks related to differentially expressedgenes (Qiagen).

Western Blot Analysis. To confirm inhibition of STAT1 signaling in LAPC-treated VSMCs, detected by IPA, cell lysates from VSMCs were run in 10% (wt/vol)Tris·HCl gel (Bio-Rad). Transfer onto membrane, blocking, washing, chem-iluminescence development, and stripping/reprobing were performed as de-scribed previously (35). The membrane was incubated with antitotal STAT1 orantiphospho–STAT1 rabbit monoclonal antibodies (Cell Signaling) and then

with peroxidase-conjugated goat anti-rabbit IgG (Vector Labs). Signals weredeveloped by a chemiluminescence kit (GE Healthcare).

qRT-PCR. For qRT-PCR, 500 ng total RNA from each treatment was reversetranscribed by using RETROscript kit (Ambion). qRT-PCR was performed usingthe TaqMan Gene Expression Assay Kits for H19, TMEM119, RARRES1, andCCL2 (TaqMan ID # Hs00262142_g1, Hs01938722_m1, Hs00161204_m1 andHs00234140_m1, respectively; Life Technologies) as described previously(36). Expression of the target mRNAs was normalized to β-actin levels, andthe 2−ΔΔCT was used to calculate relative expression levels.

Transwell Insert Migration Assays. VSMCs (75,000 cells/200 μL) were sus-pended in M231 containing 1% charcoal striped FBS and either 0.1% etha-nol (vehicle control) or the corresponding steroids and seeded onto insertwells (transwell filter with 12-μm pores; EMD Millipore). The inserts werethen placed in 24-well plates containing 500 μL M231 with 1% charcoalstriped FBS. After 12 h, cells remaining on the upper surface of the filterwere removed with a cotton swab. Migrated cells were trypsinized, centri-fuged, and resuspended in 200 μL of M231 medium. The cell suspensionswere mixed with 200 μL 10% (vol/vol) ethanol containing 0.4% Trypan bluedye and the live cell number was counted by Countess Automated CellCounter (Invitrogen). See further details in SI Materials and Methods.

Statistical Analysis. HSCOREs were normally distributed with equal variancebetween comparison groups and compared by Student’s t-test. Western blotor qRT-PCR results were normally distributed as determined by Kolmogorov–Smirnov test and analyzed by one-way ANOVA followed by testing posthoc Holm–Sidak method using SigmaStat v3.0 software (Systat Software). AP value < 0.05 is accepted as statistically significant.

ACKNOWLEDGMENTS. This study was funded by National Institutes ofHealth/The Eunice Kennedy Shriver National Institute of Child Health andHuman Development Grant 2 R01 HD 033937. Partial financial support(F27229 6009) for the guinea pig studies was provided by Bayer Pharma.

1. Krikun G, et al. (2011) Long-term progestin-only contraception in humans versusanimal models. Ann N Y Acad Sci 1221:119–123.

2. Lockwood CJ (2011) Mechanisms of normal and abnormal endometrial bleeding.Menopause 18(4):408–411.

3. Affandi B (2002) Long-acting progestogens. Best Pract Res Clin Obstet Gynaecol 16(2):169–179.

4. Lockwood CJ, Schatz F, Krikun G (2004) Angiogenic factors and the endometriumfollowing long term progestin only contraception. Histol Histopathol 19(1):167–172.

5. Carter AM (2007) Animal models of human placentation—A review. Placenta28(Suppl A):S41–S47.

6. Hutz RJ, Bejvan SM, Durning M, Dierschke DJ (1990) Changes in follicular populations,in serum estrogen and progesterone, and in ovarian steroid secretion in vitro duringthe guinea pig estrous cycle. Biol Reprod 42(2):266–272.

7. Lee KY, DeMayo FJ (2004) Animal models of implantation. Reproduction 128(6):679–695.

8. Krikun G, et al. (2012) Effects of etonogestrel treatment in the reproductive organsand uterine arteries of nonoophorectomized guinea pigs. Reprod Sci 19(4):400–406.

9. Krikun G, et al. (2010) Long-term progestin contraceptives (LTPOC) induce aberrantangiogenesis, oxidative stress and apoptosis in the guinea pig uterus: A model forabnormal uterine bleeding in humans. J Angiogenes Res 2(8):1–7.

10. Lockwood CJ, Krikun G, Hickey M, Huang SJ, Schatz F (2009) Decidualized humanendometrial stromal cells mediate hemostasis, angiogenesis, and abnormal uterinebleeding. Reprod Sci 16(2):162–170.

11. Carmeliet P (2000) Mechanisms of angiogenesis and arteriogenesis. Nat Med 6(4):389–395.

12. Ribatti D, Nico B, Crivellato E (2011) The role of pericytes in angiogenesis. Int J DevBiol 55(3):261–268.

13. HellströmM, Kalén M, Lindahl P, Abramsson A, Betsholtz C (1999) Role of PDGF-B andPDGFR-beta in recruitment of vascular smooth muscle cells and pericytes duringembryonic blood vessel formation in the mouse. Development 126(14):3047–3055.

14. Hickey M, et al. (2006) Long-term progestin-only contraceptives result in reducedendometrial blood flow and oxidative stress. J Clin Endocrinol Metab 91(9):3633–3638.

15. Kim DK, Zhang L, Dzau VJ, Pratt RE (1994) H19, a developmentally regulated gene, isreexpressed in rat vascular smooth muscle cells after injury. J Clin Invest 93(1):355–360.

16. Selzman CH, et al. (2002) Monocyte chemotactic protein-1 directly induces human vas-cular smooth muscle proliferation. Am J Physiol Heart Circ Physiol 283(4):H1455–H1461.

17. Tanaka K, et al. (2012) Interaction of Tmem119 and the bone morphogenetic proteinpathway in the commitment of myoblastic into osteoblastic cells. Bone 51(1):158–167.

18. Wang X, et al. (2013) TIG1 promotes the development and progression of in-flammatory breast cancer through activation of Axl kinase. Cancer Res 73(21):6516–6525.

19. Mowen KA, et al. (2001) Arginine methylation of STAT1 modulates IFNalpha/beta-induced transcription. Cell 104(5):731–741.

20. Hickey M, Dwarte D, Fraser IS (2000) Superficial endometrial vascular fragility inNorplant users and in women with ovulatory dysfunctional uterine bleeding. HumReprod 15(7):1509–1514.

21. Abberton KM, Healy DL, Rogers PA (1999) Smooth muscle alpha actin and myosinheavy chain expression in the vascular smooth muscle cells surrounding human en-dometrial arterioles. Hum Reprod 14(12):3095–3100.

22. Kohnen G, Campbell S, Jeffers MD, Cameron IT (2000) Spatially regulated differen-tiation of endometrial vascular smooth muscle cells. Hum Reprod 15(2):284–292.

23. Rogers PA, Plunkett D, Affandi B (2000) Perivascular smooth muscle alpha-actin isreduced in the endometrium of women with progestin-only contraceptive break-through bleeding. Hum Reprod 15(Suppl 3):78–84.

24. Abberton KM, Taylor NH, Healy DL, Rogers PA (1999) Vascular smooth muscle cellproliferation in arterioles of the human endometrium. Hum Reprod 14(4):1072–1079.

25. Lash GE, et al. (2012) Localization of angiogenic growth factors and their receptors inthe human endometrium throughout the menstrual cycle and in recurrent mis-carriage. Hum Reprod 27(1):183–195.

26. Levéen P, et al. (1994) Mice deficient for PDGF B show renal, cardiovascular, andhematological abnormalities. Genes Dev 8(16):1875–1887.

27. Soriano P (1994) Abnormal kidney development and hematological disorders inPDGF beta-receptor mutant mice. Genes Dev 8(16):1888–1896.

28. Lee WS, Harder JA, Yoshizumi M, Lee ME, Haber E (1997) Progesterone inhibits ar-terial smooth muscle cell proliferation. Nat Med 3(9):1005–1008.

29. Watanabe T, Pakala R, Katagiri T, Benedict CR (2001) Monocyte chemotactic protein 1amplifies serotonin-induced vascular smooth muscle cell proliferation. J Vasc Res38(4):341–349.

30. Spinetti G, et al. (2004) Rat aortic MCP-1 and its receptor CCR2 increase with age andalter vascular smooth muscle cell function. Arterioscler Thromb Vasc Biol 24(8):1397–1402.

31. Hayes IM, et al. (1998) Human vascular smooth muscle cells express receptors for CCchemokines. Arterioscler Thromb Vasc Biol 18(3):397–403.

32. Seki Y, et al. (2000) Role of the JAK/STAT pathway in rat carotid artery remodelingafter vascular injury. Circ Res 87(1):12–18.

33. Hu X, Park-Min KH, Ho HH, Ivashkiv LB (2005) IFN-gamma-primed macrophages ex-hibit increased CCR2-dependent migration and altered IFN-gamma responses medi-ated by Stat1. J Immunol 175(6):3637–3647.

34. Budwit-Novotny DA, et al. (1986) Immunohistochemical analyses of estrogen receptor inendometrial adenocarcinoma using a monoclonal antibody. Cancer Res 46(10):5419–5425.

35. Guzeloglu-Kayisli O, et al. (2014) Long-acting progestin-only contraceptives enhancehuman endometrial stromal cell expressed neuronal pentraxin-1 and reactive oxygenspecies to promote endothelial cell apoptosis. J Clin Endocrinol Metab 99(10):E1957–E1966.

36. Lockwood CJ, et al. (2014) Interferon-γ protects first-trimester decidual cells againstaberrant matrix metalloproteinases 1, 3, and 9 expression in preeclampsia. AmJ Pathol 184(9):2549–2559.

5158 | www.pnas.org/cgi/doi/10.1073/pnas.1424814112 Kayisli et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

021

http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental/pnas.201424814SI.pdf?targetid=nameddest=STXThttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1424814112/-/DCSupplemental/pnas.201424814SI.pdf?targetid=nameddest=STXTwww.pnas.org/cgi/doi/10.1073/pnas.1424814112