Characterization of the Antiapoptotic Bcl-2 FamilyMember Myeloid Cell Leukemia-1 (Mcl-1) and theStimulation of Its Message by Gonadotropins in theRat Ovary*

CHANDRA P. LEO†, SHEAU YU HSU, SANG-YOUNG CHUN, HYUN-WOOK BAE,AND AARON J. W. HSUEH

Division of Reproductive Biology, Department of Gynecology and Obstetrics, Stanford UniversityMedical Center (C.P.L., S.Y.H., A.J.W.H.), Stanford, California 94305-5317; and the HormoneResearch Center, Chonnam National University (S.Y.C., H.W.B.), Kwangju 500–757, Republic of Korea

ABSTRACTThe majority of ovarian follicles undergo atresia mediated by ap-

optosis. Bcl-2-related proteins act as regulators of apoptosis via theformation of dimers with proteins inside and outside the Bcl-2 family.Previous studies have identified BAD as a proapoptotic Bcl-2 familymember expressed in the ovary. It is known that BAD phosphoryla-tion induced by survival factors leads to its preferential binding to14–3-3 and suppression of the death-inducing function of BAD. Toidentify ovarian binding partners for hypophosphorylated BAD, weperformed a yeast two-hybrid screening of a rat ovary complementaryDNA library using as bait a mutant BAD incapable of binding to14–3-3. Screening of yeast transformants yielded positive clones en-coding the rat ortholog of Mcl-1 (myeloid cell leukemia-1), an anti-apoptotic Bcl-2 protein. Amino acid sequence analysis revealed thatrat and human Mcl-1 showed a complete conservation of the Bcl-2homology domains BH1, BH2, and BH3. In the yeast two-hybridsystem, Mcl-1 binds to the hypophosphorylated mutant of BAD and

interacts preferentially with different proapoptotic (Bax, Bak, Bok,Bik, and BOD) compared with antiapoptotic Bcl-2 family members(Bcl-2, Bcl-xL, Bcl-w, Bfl-1, CED-9, and BHRF-1). Northern blot hy-bridization demonstrated expression of Mcl-1 transcripts of 2.3 and3.7 kb in the ovary and diverse other rat tissues. In immature rats,PMSG treatment led to a transient increase in the 2.3-kb Mcl-1transcript, peaking at 6 h after injection and returning to baselinelevels after 24 h. Moreover, the same transcript was induced in thePMSG-primed preovulatory rat ovary 6 h after the administration ofovulatory doses of either hCG or FSH. In situ hybridization studiesrevealed that the gonadotropin stimulation of ovarian Mcl-1 messageoccurs in both granulosa and thecal cells. In conclusion, rat Mcl-1 wasidentified as an ovarian BAD-interacting protein and the message forthe antiapoptotic Mcl-1 protein was induced after treatment withgonadotropins in granulosa and thecal cells of growing follicles. (En-docrinology 140: 5469–5477, 1999)

IN THE MAMMALIAN ovary, less than 1% of all folliclesreach ovulation, whereas more than 99% undergo atresia

during reproductive life. Recent studies have establishedthat apoptosis (programmed cell death) is the molecularmechanism underlying follicle atresia (1, 2). Moreover, go-nadotropins (as well as estrogens, various growth factors,and cytokines) were found to act as extracellular survivalfactors for early antral and preovulatory follicles by sup-pressing apoptosis and thereby rescuing follicles from atresia(3–11). However, the exact intracellular mechanisms bywhich gonadotropins suppress apoptosis remain largelyunknown.

One key group of intracellular factors regulating apoptosisis the Bcl-2 family of proteins (12). The members of this familycan be subdivided into antiapoptotic proteins (such as Bcl-2

and Bcl-xL) and proapoptotic proteins (such as Bax andBAD). It has been proposed that anti- and proapoptotic pro-teins regulate cell death by binding to each other and formingheterodimers (13, 14). According to this model, a delicatebalance between anti- and proapoptotic Bcl-2 family mem-bers exists in each cell, and the relative concentrations ofthese two groups of proteins determine whether the cellsurvives or undergoes apoptosis.

Of the more than 15 known Bcl-2 family members, eachtissue or cell type expresses a specific subset. This laboratoryhas previously shown that BAD is a proapoptotic Bcl-2 fam-ily member expressed in the rat ovary (15). The function ofBAD is controlled through its phosphorylation by survivalfactor-dependent kinases (16, 17). Only the hypophosphor-ylated form of BAD can heterodimerize with the antiapop-totic proteins Bcl-xL and Bcl-2, thereby leading to cell death.By contrast, the hyperphosphorylated BAD preferentiallybinds to 14–3-3 proteins, resulting in diminished cell killing(16, 18).

To identify the ovarian binding partners for hypophos-phorylated BAD, we screened an ovarian fusion comple-mentary DNA (cDNA) library using the yeast two-hybridsystem. As bait, we used a modified rat BAD molecule inwhich the critical phosphorylation site for 14–3-3 interactionwas mutated (serine 137 to alanine). This mutation abolishes

Received May 6, 1999.Address all correspondence and requests for reprints to: Dr. Aaron

J. W. Hsueh, Division of Reproductive Biology, Department of Gyne-cology and Obstetrics, Stanford University Medical Center, Stanford,California 94305-5317. E-mail: aaron.hsueh@stanford.edu.

* This work was supported by NIH Grant HD-31566 (to A.J.W.H.) andKorean Science and Engineering Foundation Grants 97–04-01–06-01–3and HRC-98k1–0405 (to S.Y.C.).

† Supported by a postdoctoral fellowship from the German AcademicExchange Service. Present address: Department of Obstetrics and Gy-necology, University of Leipzig, Leipzig, Germany 04207.

0013-7227/99/$03.00/0 Vol. 140, No. 12Endocrinology Printed in U.S.A.Copyright © 1999 by The Endocrine Society

5469

the interaction of BAD with 14–3-3 while allowing its inter-action with Bcl-xL and thus mimics the hypophosphorylatedstate of BAD (16, 18). The library screening yielded severalpositive clones encoding the full-length rat ortholog of hu-man Mcl-1 (myeloid cell leukemia-1).1 Mcl-1 was first dis-covered as an early induction gene during the differentiationof a human myeloblastic leukemia cell line (19). Subsequentstudies established Mcl-1 as an antiapoptotic Bcl-2 familyprotein with an expression pattern differing from that ofBcl-2 and capable of suppressing cell death induced by var-ious stimuli (20–23). The expression of Mcl-1 in hemopoieticcells can be induced by various survival factors, leading toenhanced cell viability (24–26). Here, we describe the cloningof the rat Mcl-1 cDNA, dimerization properties of Mcl-1 withdiverse pro- and antiapoptotic Bcl-2 family members, as wellas the cellular localization of Mcl-1 messenger RNAs(mRNAs) in the rat ovary and their induction by gonado-tropins, which are known follicle survival factors.

Materials and MethodsYeast two-hybrid screening of an ovarian cDNA library

The full-length open reading frame (ORF) of the rat BADS137A cDNA(18) was fused in-frame with the GAL4-binding domain into the pGBT9yeast shuttle vector (CLONTECH Laboratories, Inc., Palo Alto, CA). Thisvector was used to identify proteins interacting with the mutant BADmolecule by screening 1.5 3 106 transformants from a GAL4-activationdomain-tagged rat ovarian fusion cDNA library. Positive transformantswere isolated as described previously (18).

Protein interaction assays in the yeast two-hybrid system

Interactions between rat Mcl-1 and diverse other Bcl-2 members wereassessed in the yeast two-hybrid system using a pGADGH/Mcl-1 con-struct and pGBT9 vectors containing the cDNAs of various pro- andantiapoptotic Bcl-2 family proteins (27). Specific binding of differentprotein pairs was evaluated based on the activation of the GAL1-HIS3reporter gene in the presence of 30 mm 3-aminotriazole (18).

Animal treatment

For time-course analyses of mRNA expression, female Sprague Daw-ley rats (Simonsen Laboratories, Inc., Gilroy, CA) were injected sc with10 IU PMSG (Calbiochem, La Jolla, CA) at 26 days of age and receivedan ip injection of 10 IU hCG (Schein Pharmaceuticals, Florham Park, NJ)48 h later. Rats were killed at different time points, and the ovaries werecollected for total RNA extraction or were fixed for in situ hybridizationanalysis. For analysis of Mcl-1 message in enriched granulosa cells, totalRNA was extracted from granulosa cells isolated by needle puncture(28). All animal protocols were approved by the administrative panel onlaboratory animal care at Stanford University.

For quantification of Mcl-1 message in preovulatory follicles afterhCG or FSH administration, 25-day-old female rats received a sc injec-tion of 10 IU PMSG. In addition, sc injections of the GnRH antagonistOrg30850 (40 mg/kg BW; Organon, Oss, The Netherlands) were admin-istered on days 25 and 26 to suppress the endogenous secretion ofpituitary gonadotropins (29). On day 27, a single ip injection of 10 IUhCG or 30 IU recombinant FSH (Org32489E, Organon) was adminis-tered. These doses were chosen because they were found to reliablyinduce ovulation in this model system (30). Rats were killed before or6 h after gonadotropin administration, and ovaries were collected forRNA extraction.

RNA extraction and Northern blot analysis

Ovaries were dissected free of adherent tissue, snap-frozen in a dryice/ethanol bath, and stored at 270 C. Total RNA was extracted using

the TRIzol reagent (Life Technologies, Inc., Gaithersburg, MD) accord-ing to the manufacturer’s instructions. Twenty micrograms of total RNAper lane were run on agarose-formaldehyde gels before transfer tonitrocellulose membranes and UV cross-linking. Blots containing 2 mg/lane polyadenylated [poly(A)1] RNA extracted from various rat tissueswere obtained from CLONTECH Laboratories, Inc. cDNA probes for ratMcl-1, tissue plasminogen activator (tPA), b-actin, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were 32P radiolabeled by ran-dom priming using a commercial kit (Life Technologies, Inc.). Northernblots were prehybridized and hybridized using ExpressHyb solution(CLONTECH Laboratories, Inc.) following the manufacturer’s instruc-tions. After several washes in 0.1 3 SSC-0.5% SDS at 60 C, membraneswere exposed to film (at 270 C) or to phosphorimager screens.

Quantitation of Mcl-1 mRNA levels

After exposure of membranes to phosphorimager screens, the signalintensities for the Mcl-1 transcripts (as well as for the GAPDH transcriptfrom a subsequent hybridization) were quantified for each RNA sampleusing the Storm 860 PhosphorImager and ImageQuant image analysissoftware (Molecular Dynamics, Inc., Sunnyvale, CA). Alternatively,films were scanned on a GS-710 Imaging Densitometer and analyzedusing the Quantity One Software package (Bio-Rad Laboratories, Inc.,Hercules, CA). The values for the 2.3- and 3.7-kb Mcl-1 transcripts inRNA samples extracted from rat ovaries before and 6 h after injectionof hCG or FSH were then normalized to the respective GAPDH signal.The results are expressed as fold induction compared with the 0 hcontrol, which was arbitrarily set at 1.

In situ hybridization studies

Ovaries were fixed at 4 C for 6 h in 4% paraformaldehyde in PBS,followed by immersion in 0.5 m sucrose in PBS overnight. Cryostat sections(14 mm thick) were mounted on microscope slides coated with poly-l-lysine(Sigma Chemical Co., St. Louis, MO), fixed in 4% paraformaldehyde in PBS,and stored at 280 C until analyzed. To allow for direct comparison betweenovarian sections from different experimental groups, all slides were pro-cessed simultaneously under identical conditions. The hybridization pro-cedure was essentially the same as that previously described (31). In brief,sections were pretreated serially with 0.2 m HCl, 2 3 SSC, pronase (0.125mg/ml), 4% paraformaldehyde, and acetic anhydride in triethanolamine.Hybridization was carried out at 52–55 C overnight in a mixture containing35S-labeled rat Mcl-1 complementary RNA probe (108 cpm/ml), 50% for-mamide, 0.3 m NaCl, 10 mm Tris-HCl, 5 mm EDTA, 1 3 Denhardt’s so-lution, 10% dextran sulfate, 1 mg/ml carrier transfer RNA, and 10 mmdithiothreitol. Posthybridization washing was performed under stringentconditions that included ribonuclease A (25 mg/ml) treatment at 37 C for30 min and a final stringency of 0.1 3 SSC. Slides were dipped into NTB-2emulsion (Eastman Kodak Co., Rochester, NY) and exposed at 4 C for 3–4weeks before development. The slides were stained with hematoxylin andeosin and examined under a light microscope with bright- and darkfieldillumination.

Statistical analysis

Results are presented as the mean 6 sem. For the ovulation inductionexperiments using hCG or FSH, differences in normalized levels of Mcl-1transcript expression among groups were assessed by ANOVA followedby Fisher’s protected least significant differences post-hoc test. Differ-ences in Mcl-1 transcript expression before and after hCG treatment ingranulosa cells were analyzed using the unpaired t test. Statistical sig-nificance was inferred at P , 0.05.

ResultsIsolation of rat Mcl-1 in a yeast two-hybrid screening of anovarian cDNA library

To identify ovarian binding partners for the proapoptoticprotein BAD, a cDNA encoding the rat BADS137A mutant wasused as a bait to screen 1.5 3 106 yeast transformants froma rat ovary fusion cDNA library. The serine to alanine sub-stitution in BAD eliminates the critical phosphorylation site1 The GenBank accession number for rat Mcl-1 is AF115380.

5470 GONADOTROPINS STIMULATE Mcl-1 EXPRESSION IN OVARIAN FOLLICLES Endo • 1999Vol 140 • No 12

of BAD, thus mimicking the hypophosphorylated, apopto-sis-inducing form of BAD (16, 18). Based on activation of theGAL1-HIS3 reporter gene, several positive clones encodingBAD-interacting proteins were identified. In addition to mul-tiple clones representing P11 (18), three clones were found toencode a full-length protein with extensive homology to thehuman Mcl-1 protein (see Fig. 1).

Sequence analysis of rat Mcl-1

Comparison between the deduced amino acid sequencesshowed that rat and human Mcl-1 share 78% identity with acomplete conservation of the Bcl-2 homology domains BH1,BH2, and BH3 as well as the transmembrane region (Fig. 1). Inaddition, rat Mcl-1 retains the long N-terminal sequence con-taining a proline-glutamic acid-serine-threonine-rich (PEST)domain that has been implicated in targeting the protein forrapid intracellular turnover. Furthermore, use of the rat Mcl-1amino acid sequence as a query to search the GenBank ex-pressed sequence tag (EST) division identified a zebrafish EST(AI544581) that encodes a protein fragment with closest ho-mology to the mammalian Mcl-1 proteins. Figure 1 shows analignment of the deduced amino acid sequences for rat andhuman Mcl-1 as well as for the recently published partial se-quence of chicken Mcl-1 (32) and the zebrafish EST.

Interaction of Mcl-1 with diverse pro- and antiapoptoticBcl-2 family proteins in the yeast two-hybrid system

The dimerization properties of rat Mcl-1 with differentpro- and antiapoptotic Bcl-2 members were assessed in theyeast two-hybrid system (Fig. 2). In agreement with its iden-tification in the yeast two-hybrid screening, Mcl-1 interactedstrongly with the phosphorylation site mutant of BAD. How-ever, under the conditions employed, Mcl-1 did not bind towild-type BAD, which appears to be constitutively hyper-phosphorylated in yeast cells as suggested by its strong in-teraction with the u-isoform of 14–3-3. In addition, Mcl-1 alsointeracted with the proapoptotic Bcl-2 family members Bax,Bak, Bok/Mtd, Bik, and BOD/Bim. By contrast, Mcl-1 dimer-ized only weakly (Bcl-w, Bfl-1, CED-9, and BHRF-1) or notat all (Mcl-1, Bcl-2, and Bcl-xL) with the antiapoptotic Bcl-2family members tested in this assay.

Northern blot analysis of Mcl-1 expression in different rattissues and developmental expression in the ovary

To investigate the tissue expression pattern of Mcl-1, wehybridized Northern blots containing RNA extracted fromvarious rat tissues with a radiolabeled rat Mcl-1 cDNA probe.As shown in Fig. 3, two Mcl-1 transcripts with sizes of 2.3 and

FIG. 1. Comparison of the deducedamino acid sequences for rat, human,and chicken Mcl-1 and a zebrafish EST(AI544581). The PEST sequence, Bcl-2homology domains (BH1, BH2, andBH3), and transmembrane region (TM)are marked. As the chicken and ze-brafish sequences contain only partialORFs, the amino acid numbers (aster-isks) refer to the respective start of theknown coding sequence fragments.

GONADOTROPINS STIMULATE Mcl-1 EXPRESSION IN OVARIAN FOLLICLES 5471

3.7 kb, respectively, were expressed at varying levels in allrat tissues examined. The highest levels of expression werefound in spleen, lung, heart, and kidney [Fig. 3A, poly(A)1

RNA analyzed], whereas lower expression levels were ob-served in reproductive tissues, including ovary, oviduct,uterus, and testis (Fig. 3B, total RNA analyzed). To furtherinvestigate developmental expression of Mcl-1 in the femalegonad, total RNA extracted from ovaries at different pointsof postnatal development was hybridized with the sameprobe. As shown in Fig. 4A, the Mcl-1 message is expressedfrom day 3 after birth and increased after day 12 of age.

Gonadotropin induction of Mcl-1 message in the ovary

As it had previously been shown that the Mcl-1 messagecan be rapidly induced by survival factors in hemopoieticcells (24, 25, 33), we investigated whether gonadotropins,which are known to act as follicle survival factors, couldincrease Mcl-1 expression in the ovary. Immature rats re-ceived an injection of 10 IU PMSG, followed 48 h later by 10IU hCG to induce ovulation. Using Northern blot analysis,the expression of Mcl-1 message in rat ovaries collected atdifferent time points was assessed. As shown in Fig. 4, B andC, the two Mcl-1 transcripts were found to be expressedthroughout the treatment time course. After PMSG injection,

expression of the 2.3-kb transcript increased, peaking at 3–6h and then returning to baseline levels by 24 h. Two days afterPMSG priming, ovulation induction with hCG resulted in an8.5-fold induction (P , 0.05) of the 2.3-kb transcript (nor-malized to GAPDH expression; n 5 3) at the 6 h point,whereas expression of this transcript at 12 h was not signif-icantly different from the 0 h expression level (P . 0.10).There were no statistically significant differences among theexpression levels of the larger 3.7-kb transcript at 0, 6, and12 h after hCG administration (P . 0.05).

Both hCG and FSH cause a preovulatory induction of Mcl-1message in the rat ovary

As both hCG and FSH can act as follicle survival factorsin preovulatory follicles (7) as well as induce follicle rupturein PMSG-primed immature rats (30), we further compared

FIG. 3. Expression of the Mcl-1 mRNA in different rat tissues. North-ern blots containing RNA extracted from different nonreproductive[A; poly(A)1-selected RNA] and reproductive rat tissues (B; totalRNA) were hybridized with a radiolabeled Mcl-1 cDNA probe. Ar-rowheads indicate the two major Mcl-1 transcripts (2.3 and 3.7 kb).Sk. muscle, Skeletal muscle. The b-actin message is shown as acontrol for RNA loading.

FIG. 4. Regulation of Mcl-1 message in the rat ovary during devel-opment and after gonadotropin treatment. A, Developmental regu-lation of Mcl-1 message in the prepubertal rat ovary. B, Effect ofPMSG treatment on ovarian Mcl-1 message in the 26-day-old rat. C,Effect of hCG treatment (48 h after PMSG priming) on ovarian Mcl-1message. The GAPDH message is shown as a control for RNA loading.The results shown are representative of two (A and B) or three sep-arate experiments (C).

FIG. 2. Interaction of Mcl-1 with different Bcl-2 family proteins in theyeast two-hybrid system. The interactions between Mcl-1 and differ-ent pro- and antiapoptotic Bcl-2 family proteins are rated as strong(31), moderate (21), weak (1), or absent (2), based on the activationof the GAL1-HIS3 reporter gene. pGADGH and pGBT9 denote theempty activation domain and binding domain vectors, respectively,which were included as negative controls. The specific interactions ofwild-type BAD and BADS137A with 14–3-3 u and Mcl-1, respectively,illustrate the importance of serine phosphorylation at position 137 indetermining the binding partners for BAD (see text). All Bcl-2 familyproteins tested in this assay are of mammalian origin, except forCED-9, an antiapoptotic Bcl-2 family protein found in the nematodeCaenorhabditis elegans, and BHRF-1, a Bcl-2-homolog encoded by theEpstein-Barr virus.

5472 GONADOTROPINS STIMULATE Mcl-1 EXPRESSION IN OVARIAN FOLLICLES Endo • 1999Vol 140 • No 12

the effects of these two gonadotropins on the induction ofMcl-1 mRNA in the ovary. Immature rats were primed withPMSG, but additionally received two injections of the GnRHantagonist Org30850 to suppress endogenous pituitary go-nadotropin secretion. At 48 h after PMSG administration, therats were injected with an ovulatory dose of either hCG orFSH. As the earlier time-course studies had demonstrated amaximal induction of Mcl-1 message after 6 h, expression atthis time point was compared with levels before the injectionof hCG or FSH.

Northern blots of total RNA isolated from rat ovaries be-fore treatment or at 6 h after hCG or FSH administration werehybridized with a radiolabeled Mcl-1 cDNA probe. Subse-quently, the same blots were also hybridized with controlprobes for tPA and GAPDH (Fig. 5A). The message for tPAhad previously been shown to be induced by both hCG andFSH (30) and therefore served as a positive control, whereasthe housekeeping gene GAPDH was used as a control forRNA loading. The expression of each of the two Mcl-1 tran-scripts was quantified separately and normalized to theGAPDH signal. As shown in Fig. 5B, expression of the 2.3-kbMcl-1 transcript in the ovaries of GnRH antagonist-pre-treated rats was induced 3.3- and 2.4-fold (P , 0.01) by hCGand FSH, respectively, whereas expression of the 3.7-kb tran-script remained essentially unchanged (P . 0.05).

Localization of ovarian Mcl-1 message by in situhybridization and gonadotropin stimulation of Mcl-1expression in isolated granulosa cells

To further examine the expression pattern of Mcl-1 in thepreovulatory ovary, we performed in situ hybridization anal-ysis. As shown in Fig. 6A, ovaries from immature rats con-tained multiple early antral follicles, preantral follicles, andsome atretic follicles. Mcl-1 mRNA was mainly detected inthe thecal cells of early and preantral follicles (Fig. 6B). Al-though treatment with PMSG for 6 h did not cause obviouschanges in ovarian morphology (Fig. 6C), a major increase inMcl-1 message could be detected in both granulosa and the-

cal cells (Fig. 6, D and E). The specificity of the signal obtainedwith the antisense Mcl-1 probe was demonstrated by the lackof hybridization in sections treated with the sense Mcl-1probe (Fig. 6F). After PMSG priming for 48 h, the rat ovariescontained multiple large antral follicles, and the Mcl-1 mes-sage was again mainly expressed in the thecal cells of thesefollicles (Fig. 6, G and H). Six hours after the hCG injection,hybridization with the Mcl-1 antisense probe demonstratedan increase in Mcl-1 mRNA in both thecal and granulosa cellsof these follicles (Fig. 6, I and J).

As the Mcl-1 signal observed in granulosa cells wasweaker than that in thecal cells, we further substantiated theinduction of the Mcl-1 message in granulosa cells, the celltype undergoing apoptosis during follicle atresia in the rat(1). Therefore, enriched granulosa cells were isolated fromthe follicles of PMSG-primed rat ovaries by needle puncture.Northern blots containing total RNA extracted from granu-losa cells at 0 and 6 h after hCG injection were hybridizedwith the Mcl-1 cDNA probe. Although treatment with hCGdid not significantly change the expression of the 3.7-kbtranscript (P . 0.10), it caused an almost 3-fold induction(2.8 6 0.64 vs. 1.0 6 0.13; P , 0.05) of the 2.3-kb Mcl-1 messagein enriched granulosa cells, similar to the induction seen inthe preovulatory ovary as a whole.

Discussion

Follicle atresia is a hormonally controlled apoptotic pro-cess (1, 2). In particular, interference with gonadotropin se-cretion or receptor binding causes atresia of preovulatoryfollicles in vivo (34–36). Conversely, early atretic follicles canbe rescued by administration of exogenous gonadotropins(3). However, the exact intracellular mechanisms mediatingthe antiatretic and antiapoptotic effects of follicle survivalfactors such as gonadotropins have remained unclear. Here,we demonstrate that the rat ortholog of the antiapoptotichuman Mcl-1 molecule is capable of dimerization with dif-ferent proapoptotic Bcl-2 family proteins and that it is ex-pressed in thecal and granulosa cells of the ovary, where it

FIG. 5. Induction of Mcl-1 message in the preovulatory ovary by hCG or FSH. A, Representative Northern blot of total ovarian RNA extractedfrom PMSG-primed ovaries before or at 6 h after the injection of ovulatory doses of either hCG or FSH. The blot was successively hybridizedwith probes for rat Mcl-1, tPA, and GAPDH. B, Quantification of ovarian Mcl-1 mRNA induction by hCG or FSH. The phosphorimager signalfor each of the two Mcl-1 transcripts was quantified and normalized to the GAPDH signal. The results are expressed as the fold induction overthe 0 h control, which was arbitrarily set at 1. Data shown are the mean 6 SEM for five independent samples per group. Asterisks, Significantlydifferent from the 0 h control group, P , 0.01.

GONADOTROPINS STIMULATE Mcl-1 EXPRESSION IN OVARIAN FOLLICLES 5473

is induced by the survival-promoting gonadotropins PMSG,hCG, and FSH.

Among the intracellular regulators of apoptosis, the Bcl-2family of proteins occupies a central position by integratingdiverse survival and death signals and linking them to down-stream apoptotic events such as mitochondrial cytochrome crelease and caspase activation (12). The importance of Bcl-

2-mediated pathways to the regulation of follicle survivaland atresia is illustrated by the phenotype of mice overex-pressing Bcl-2 under the control of the inhibin-a promoter/enhancer. The targeted overexpression of this antiapoptoticgene in somatic ovarian cells leads to diminished follicularcell apoptosis and increased ovulation rates and litter sizesin transgenic animals (37). This phenotype is somewhat rem-

FIG. 6. In situ localization of Mcl-1mRNA in the immature rat ovary beforeand after PMSG and hCG treatment.Sections of ovaries collected from 26-day-old rats were hybridized with a 35S-labeled rat Mcl-1 antisense probe andprocessed for liquid emulsion autora-diography. Ovaries from rats before (Aand B) and 6 h after (C and D) PMSGinjection were analyzed. Brightfield (Aand C) and corresponding darkfield (Band D) photomicrographs are shown(350). E, Higher magnification view ofD (3125). F, Darkfield view of an ovar-ian section hybridized with an Mcl-1sense probe as a control (350). In ad-dition, ovarian sections from ratsprimed with PMSG for 48 h were ana-lyzed before (G and H) and 6 h after hCGinjection (I and J). Brightfield (G and I)and corresponding darkfield (H and J)photomicrographs are shown (350).PAF, Preantral follicle; EAF, early an-tral follicle; AtF, atretic follicle; LAF,large antral follicle; Oo, oocyte; Tc, the-cal cells; Gc, granulosa cells.

5474 GONADOTROPINS STIMULATE Mcl-1 EXPRESSION IN OVARIAN FOLLICLES Endo • 1999Vol 140 • No 12

iniscent of the effects seen in wild-type animals hyperstimu-lated with gonadotropins, raising the possibility that gonad-otropins might induce an antiapoptotic Bcl-2 family proteinin follicular cells.

Although the mRNAs for the antiapoptotic moleculesBcl-2 and Bcl-xL have been found in the ovary, their expres-sion appears to be unaffected by PMSG treatment in theimmature rat model (38). To identify additional antiapoptoticBcl-2 family members in the ovary, we performed a yeasttwo-hybrid screening of a rat ovary cDNA library using amutant BAD molecule as bait. The proapoptotic BAD is con-stitutively expressed in the rat ovary (15). Although pre-sumably hyperphosphorylated BAD preferentially binds dif-ferent isoforms of 14–3-3, a BAD mutant mimicking thehypophosphorylated form of BAD was found to interactwith P11, a survival gene induced by nerve growth factor inthe PC12 pheochromocytoma cell line (18). Here, we describethe identification of Mcl-1 as an additional interaction part-ner for this phosphorylation site mutant of BAD.

In living eukaryotic cells, Mcl-1 is capable of interactingwith the wild-type forms of several other proapoptotic Bcl-2family proteins (Bax, Bak, Bok/Mtd, Bik, and BOD/Bim), asdemonstrated by yeast two-hybrid assays. Both Bok/Mtd(27, 39) and BOD/Bim (40, 41) were previously shown to beexpressed in the rat ovary. In a recently performed yeasttwo-hybrid screening using BOD/Bim as bait, several of theclones with the strongest interaction were also found to en-code Mcl-1 (data not shown). The dimer formation betweenpro- and antiapoptotic Bcl-2 family proteins, mediated bytheir highly conserved Bcl-2 homology domains, is thoughtto be a critical determinant of their mutual functional an-tagonism (12–14). In this respect, the observation that Mcl-1can interact promiscuously with different proapoptotic Bcl-2-related proteins in yeast cells complements previous find-ings showing that Mcl-1 coimmunoprecipitates with andsuppresses apoptosis induced by diverse proapoptotic Bcl-2members (22, 27).

In addition to its heterodimerization with proapoptoticBcl-2 family members, Mcl-1 could inhibit apoptosis by atleast two other mechanisms. Similar to Bcl-xL and Bcl-2,Mcl-1 could form pores in the outer mitochondrial mem-brane to regulate cytochrome c release (42, 43) and/or inhibitthe activation of caspases by retaining the adaptor moleculeApaf-1 in an inactive conformation (12, 44). Although itsexact mechanism of action remains to be examined, Mcl-1 hasbeen shown to increase cell viability under different apop-tosis-inducing conditions, including growth factor depriva-tion and exposure to chemotherapeutic agents or UV irra-diation (21–23).

The complete conservation of the Bcl-2 homology domainsBH1, BH2, and BH3 (located in the C-terminal portion of theprotein) between rat and human is consistent with theirsuggested importance in providing the structural basis fordimerizations between Bcl-2 family proteins (12, 45). Thelong N-terminal portion of Mcl-1, which bears no homologyto other Bcl-2-related proteins, is also conserved between ratand human, although to a lesser degree. The functional sig-nificance of this domain is unclear, except for its PEST se-quence, which is thought to account for the short half-life ofthe protein (see below). The partial amino acid sequences,

deduced from a published chicken nucleotide sequence (32)and a recently released zebrafish EST, offer further insightsinto the conservation of Mcl-1-related sequences in nonmam-malian species. The chicken sequence and zebrafish ESTencode protein fragments with 61% and 53% identity to ratMcl-1, respectively. The zebrafish sequence described here,therefore, most likely represents the first Bcl-2 family gene tobe found in teleosts.

Several antiapoptotic Bcl-2 family genes (Bcl-2, Bcl-xL,A1/Bfl-1, and Mcl-1) are transcriptionally induced by spe-cific cytokines in hemopoietic cell types (12). Prosurvivalfactors inducing Mcl-1 in hemopoietic cells include granu-locyte-macrophage colony-stimulating factor, interleukin-1b, and vascular endothelial growth factor (24, 25, 46). Thestimulation of Mcl-1 expression by gonadotropins describedhere represents the first evidence of a Bcl-2 family gene beinginduced by follicle survival factors in the ovary. The rapidand transient increase in the ovarian Mcl-1 message aftergonadotropin treatment mirrors the pattern of induction ob-served in hemopoietic cells (24, 47). This pattern was shownto reflect a transient transcriptional activation of the Mcl-1gene, generating a message with a short half-life of less than2 h (47). In hemopoietic cells, these changes in Mcl-1 mRNAlevels closely correlate to similar alterations in the levels ofthe Mcl-1 protein, which is also rapidly turned over, prob-ably due to its N-terminal PEST sequence (24, 47, 48). It hasbeen suggested that the transient induction of Mcl-1 messageand protein by survival factors serves to protect cells duringspecific stages of differentiation when they are particularlyvulnerable to apoptosis (23).

Because of its induction by PMSG, FSH, or hCG, the go-nadotropin-dependent expression of Mcl-1 in ovarian cells isprobably mediated by a mechanism in the common down-stream signaling pathways of the LH/hCG and FSH recep-tors. The murine Mcl-1 promoter contains a DNA sequencesimilar to the consensus site recognized by the CRE-2-bind-ing protein, which is involved in transcriptional induction bycAMP (24, 49). However, a detailed molecular analysis of theMcl-1 gene promoter will be necessary to delineate potentialcis-responsive elements mediating the gonadotropin induc-tion of Mcl-1. Interestingly, the induction of the 2.3-kb Mcl-1transcript at 6 h after hCG injection was lower in Org30850-pretreated rats (3.3 6 0.30-fold) than in rats not pretreatedwith the GnRH antagonist (8.5 6 0.17-fold). This observationcould be due to the different age of animals in these twoseries of experiments (27 vs. 28 days) and/or differences intheir endogenous gonadotropin levels.

The two Mcl-1 transcripts observed in rat tissues (2.3 and3.7 kb) correspond in size to the two known human Mcl-1messages (2.5 and 3.8 kb) (19). Both Mcl-1 transcripts in therat are long enough to contain the full-length coding se-quence. Although one cannot exclude the possibility that therat mRNAs may encode different ORFs, the two humanMcl-1 transcripts probably result from the use of alternativepolyadenylation sites in the 39-untranslated region of thegene (19). Although the significance of selective induction ofthe 2.3-kb Mcl-1 transcript by gonadotropins remains un-known, it is interesting to note that a similar differentialinduction of two splicing variants by gonadotropins in gran-

GONADOTROPINS STIMULATE Mcl-1 EXPRESSION IN OVARIAN FOLLICLES 5475

ulosa cells has recently been observed for the kit ligand gene(50).

During follicle atresia in the rodent ovary, apoptotic celldeath is confined to granulosa cells (9, 11). The basal expres-sion and gonadotropin induction of Mcl-1 in granulosa cellsare therefore consistent with its potential role in regulatingfollicle atresia. Whether the levels of Mcl-1 expression inindividual follicles correlate with granulosa cell apoptosisremains to be determined. Interestingly, levels of Mcl-1mRNA expression in thecal cells, a cell type resistant toapoptosis in this model, were higher than those in granulosacells. A similar differential ovarian expression pattern hasbeen observed for the inhibitor of apoptosis proteins, Hiap-2and Xiap, which are also induced by gonadotropins (51). Itis likely that the high expression of antiapoptotic genes inthecal cells confers a protection from apoptosis.

From the data presented, we propose the following model.Mcl-1 is an antiapoptotic Bcl-2 member expressed in ovarianfollicles that can dimerize with BAD and other proapoptoticovarian Bcl-2 family members (e.g. Bax, BOD, and Bok). Go-nadotropins induce a transient increase in the expression ofMcl-1 and may thereby shift the balance of ovarian anti- andproapoptotic Bcl-2 family proteins to favor survival and res-cue follicles from atresia. The induction of Mcl-1 along withother antiapoptotic genes such as inhibitor of apoptosis pro-teins may therefore represent a molecular basis for the sup-pression of follicular apoptosis and atresia by gonadotropins.

Acknowledgments

We are grateful to Connie Schindler in the laboratory of AmatoGiaccia (Stanford, CA) for help with the phosphorimager analysis, andto Sameena Beguwala for assistance with the yeast two-hybrid assays.We thank the following individuals for the provision of cDNAs fordifferent proteins of the Bcl-2 family: M. Cleary (Stanford, CA; Bcl-2), S.Cory (Victoria, Australia; Bcl-w), G. Chinnadurai (St. Louis, MO;Bfl-1/A1 and Bik), A. Rickinson (Birmingham, UK; BHRF1), R. Horvitz(Cambridge, MA; CED-9), T. Chittenden (Cambridge, MA; Bak), and C.Thompson (Chicago, IL; Bcl-xL). We also thank Dr. Lenus Kloosterboarfrom Organon (Oss, The Netherlands) for providing the GnRH antag-onist Org 30850 and recombinant human FSH.

References

1. Hsueh AJW, Billig H, Tsafriri A 1994 Ovarian follicle atresia: a hormonallycontrolled apoptotic process. Endocr Rev 15:707–724

2. Kaipia A, Hsueh AJW 1997 Regulation of ovarian follicle atresia. Annu RevPhysiol 59:349–363

3. Braw RH, Tsafriri A 1980 Effect of PMSG on follicular atresia in the immaturerat ovary. J Reprod Fertil 59:267–272

4. Maxson WS, Haney AF, Schomberg DW 1985 Steroidogenesis in porcineatretic follicles: loss of aromatase activity in isolated granulosa and theca. BiolReprod 33:495–501

5. Tilly JL, Kowalski KI, Schomberg DW, Hsueh AJW 1992 Apoptosis in atreticovarian follicles is associated with selective decreases in messenger ribonucleicacid transcripts for gonadotropin receptors and cytochrome P450 aromatase.Endocrinology 131:1670–1676

6. Billig H, Furuta I, Hsueh AJW 1994 Gonadotropin-releasing hormone directlyinduces apoptotic cell death in the rat ovary: biochemical and in situ detectionof deoxyribonucleic acid fragmentation in granulosa cells. Endocrinology134:245–252

7. Chun SY, Billig H, Tilly JL, Furuta I, Tsafriri A, Hsueh AJW 1994 Gonad-otropin suppression of apoptosis in cultured preovulatory follicles: mediatoryrole of endogenous insulin-like growth factor I. Endocrinology 135:1845–1853

8. Luciano AM, Pappalardo A, Ray C, Peluso JJ 1994 Epidermal growth factorinhibits large granulosa cell apoptosis by stimulating progesterone synthesisand regulating the distribution of intracellular free calcium. Biol Reprod51:646–654

9. Chun SY, Eisenhauer KM, Minami S, Billig H, Perlas E, Hsueh AJW 1996

Hormonal regulation of apoptosis in early antral follicles: follicle-stimulatinghormone as a major survival factor. Endocrinology 137:1447–1456

10. Johnson AL, Bridgham JT, Witty JP, Tilly JL 1996 Susceptibility of avianovarian granulosa cells to apoptosis is dependent upon stage of follicle de-velopment and is related to endogenous levels of bcl-xlong gene expression.Endocrinology 137:2059–2066

11. Boone DL, Carnegie JA, Rippstein PU, Tsang BK 1997 Induction of apoptosisin equine chorionic gonadotropin (eCG)-primed rat ovaries by anti-eCG an-tibody. Biol Reprod 57:420–427

12. Adams JM, Cory S 1998 The Bcl-2 protein family: arbiters of cell survival.Science 281:1322–1326

13. Oltvai ZN, Milliman CL, Korsmeyer SJ 1993 Bcl-2 heterodimerizes in vivowith a conserved homolog, Bax, that accelerates programmed cell death. Cell74:609–619

14. Yang E, Zha J, Jockel J, Boise LH, Thompson CB, Korsmeyer SJ 1995 Bad, aheterodimeric partner for Bcl-XL and Bcl-2, displaces Bax and promotes celldeath. Cell 80:285–291

15. Kaipia A, Hsu SY, Hsueh AJW 1997 Expression and function of a proapoptoticBcl-2 family member Bcl-XL/Bcl-2-associated death promoter (BAD) in ratovary. Endocrinology 138:5497–5504

16. Zha J, Harada H, Yang E, Jockel J, Korsmeyer SJ 1996 Serine phosphorylationof death agonist BAD in response to survival factor results in binding to 14–3-3not BCL-X(L). Cell 87:619–628

17. Datta SR, Dudek H, Tao X, Masters S, Fu H, Gotoh Y, Greenberg ME 1997Akt phosphorylation of BAD couples survival signals to the cell-intrinsic deathmachinery. Cell 91:231–241

18. Hsu SY, Kaipia A, Zhu L, Hsueh AJW 1997 Interference of BAD (Bcl-xL/Bcl-2-associated death promoter)-induced apoptosis in mammalian cells by14–3-3 isoforms and P11. Mol Endocrinol 11:1858–1867

19. Kozopas KM, Yang T, Buchan HL, Zhou P, Craig RW 1993 MCL1, a geneexpressed in programmed myeloid cell differentiation, has sequence similarityto BCL2. Proc Natl Acad Sci USA 90:3516–3520

20. Yang T, Kozopas KM, Craig RW 1995 The intracellular distribution andpattern of expression of Mcl-1 overlap with, but are not identical to, those ofBcl-2. J Cell Biol 128:1173–1184

21. Reynolds JE, Li J, Craig RW, Eastman A 1996 BCL-2 and MCL-1 expressionin Chinese hamster ovary cells inhibits intracellular acidification and apoptosisinduced by staurosporine. Exp Cell Res 225:430–436

22. Zhou P, Qian L, Kozopas KM, Craig RW 1997 Mcl-1, a Bcl-2 family member,delays the death of hematopoietic cells under a variety of apoptosis-inducingconditions. Blood 89:630–643

23. Zhou P, Qian L, Bieszczad CK, Noelle R, Binder M, Levy NB, Craig RW 1998Mcl-1 in transgenic mice promotes survival in a spectrum of hematopoietic celltypes and immortalization in the myeloid lineage. Blood 92:3226–3239

24. Chao JR, Wang JM, Lee SF, Peng HW, Lin YH, Chou CH, Li JC, Huang HM,Chou CK, Kuo ML, Yen JJ, Yang-Yen HF 1998 mcl-1 is an immediate-earlygene activated by the granulocyte-macrophage colony-stimulating factor (GM-CSF) signaling pathway and is one component of the GM-CSF viability re-sponse. Mol Cell Biol 18:4883–4898

25. Moulding DA, Quayle JA, Hart CA, Edwards SW 1998 Mcl-1 expression inhuman neutrophils: regulation by cytokines and correlation with cell survival.Blood 92:2495–2502

26. Townsend KJ, Zhou P, Qian L, Bieszczad CK, Lowrey CH, Yen A, Craig RW1999 Regulation of MCL1 through a serum response factor/Elk-1-mediatedmechanism links expression of a viability-promoting member of the BCL2family to the induction of hematopoietic cell differentiation. J Biol Chem274:1801–1813

27. Hsu SY, Kaipia A, McGee E, Lomeli M, Hsueh AJW 1997 Bok is a pro-apoptotic Bcl-2 protein with restricted expression in reproductive tissues andheterodimerizes with selective anti-apoptotic Bcl-2 family members. Proc NatlAcad Sci USA 94:12401–12406

28. Hsueh AJW, Adashi EY, Jones PB, Welsh Jr TH 1984 Hormonal regulation ofthe differentiation of cultured ovarian granulosa cells. Endocr Rev 5:76–127

29. Deckers GH, de Graaf JH, Kloosterboer HJ, Loozen HJ 1992 Properties of apotent LHRH antagonist (Org 30850) in female and male rats. J Steroid BiochemMol Biol 42:705–712

30. Galway AB, Lapolt PS, Tsafriri A, Dargan CM, Boime I, Hsueh AJW 1990Recombinant follicle-stimulating hormone induces ovulation and tissue plas-minogen activator expression in hypophysectomized rats. Endocrinology127:3023–3028

31. Hsu SY, Kubo M, Chun SY, Haluska FG, Housman DE, Hsueh AJW 1995Wilms’ tumor protein WT1 as an ovarian transcription factor: decreases inexpression during follicle development and repression of inhibin-a gene pro-moter. Mol Endocrinol 9:1356–1366

32. Lee R, Gillet G, Burnside J, Thomas SJ, Neiman P 1999 Role of Nr13 inregulation of programmed cell death in the bursa of fabricius. Genes Dev13:718–728

33. Townsend KJ, Trusty JL, Traupman MA, Eastman A, Craig RW 1998 Ex-pression of the antiapoptotic MCL1 gene product is regulated by a mitogenactivated protein kinase-mediated pathway triggered through microtubuledisruption and protein kinase C. Oncogene 17:1223–1234

5476 GONADOTROPINS STIMULATE Mcl-1 EXPRESSION IN OVARIAN FOLLICLES Endo • 1999Vol 140 • No 12

34. Braw RH, Bar-Ami S, Tsafriri A 1981 Effect of hypophysectomy on atresia ofrat preovulatory follicles. Biol Reprod 25:989–996

35. Braw RH, Tsafriri A 1980 Follicles explanted from pentobarbitone-treated ratsprovide a model for atresia. J Reprod Fertil 59:259–265

36. Bill CH, Greenwald GS 1981 Acute gonadotropin deprivation. I. A model forthe study of follicular atresia. Biol Reprod 24:913–921

37. Hsu SY, Lai RJ, Finegold M, Hsueh AJW 1996 Targeted overexpression ofBcl-2 in ovaries of transgenic mice leads to decreased follicle apoptosis, en-hanced folliculogenesis, and increased germ cell tumorigenesis. Endocrinol-ogy 137:4837–4843

38. Tilly JL, Tilly KI, Kenton ML, Johnson AL 1995 Expression of members of thebcl-2 gene family in the immature rat ovary: equine chorionic gonadotropin-mediated inhibition of granulosa cell apoptosis is associated with decreasedbax and constitutive bcl-2 and bcl-xlong messenger ribonucleic acid levels.Endocrinology 136:232–241

39. Inohara N, Ekhterae D, Garcia I, Carrio R, Merino J, Merry A, Chen S, NunezG 1998 Mtd, a novel Bcl-2 family member, activates apoptosis in the absenceof heterodimerization with Bcl-2 and Bcl-XL. J Biol Chem 273:8705–8710

40. Hsu SY, Lin P, Hsueh AJW 1998 BOD (Bcl-2-related ovarian death gene) is anovarian BH3 domain-containing proapoptotic Bcl-2 protein capable of dimer-ization with diverse antiapoptotic Bcl-2 members. Mol Endocrinol12:1432–1440

41. O’Connor L, Strasser A, O’Reilly LA, Hausmann G, Adams JM, Cory S,Huang DC 1998 Bim: a novel member of the Bcl-2 family that promotesapoptosis. EMBO J 17:384–395

42. Schendel SL, Xie Z, Montal MO, Matsuyama S, Montal M, Reed JC 1997Channel formation by antiapoptotic protein Bcl-2. Proc Natl Acad Sci USA94:5113–5118

43. Kluck RM, Bossy-Wetzel E, Green DR, Newmeyer DD 1997 The release of

cytochrome c from mitochondria: a primary site for Bcl-2 regulation of apo-ptosis. Science 275:1132–1136

44. Hu Y, Benedict MA, Wu D, Inohara N, Nunez G 1998 Bcl-XL interacts withApaf-1 and inhibits Apaf-1-dependent caspase-9 activation. Proc Natl Acad SciUSA 95:4386–4391

45. Sattler M, Liang H, Nettesheim D, Meadows RP, Harlan JE, Eberstadt M,Yoon HS, Shuker SB, Chang BS, Minn AJ, Thompson CB, Fesik SW 1997Structure of Bcl-xL-Bak peptide complex: recognition between regulators ofapoptosis. Science 275:983–986

46. Katoh O, Takahashi T, Oguri T, Kuramoto K, Mihara K, Kobayashi M, HirataS, Watanabe H 1998 Vascular endothelial growth factor inhibits apoptoticdeath in hematopoietic cells after exposure to chemotherapeutic drugs byinducing MCL1 acting as an antiapoptotic factor. Cancer Res 58:5565–5569

47. Yang T, Buchan HL, Townsend KJ, Craig RW 1996 MCL-1, a member of theBCL-2 family, is induced rapidly in response to signals for cell differentiationor death, but not to signals for cell proliferation. J Cell Physiol 166:523–536

48. Rechsteiner M, Rogers SW 1996 PEST sequences and regulation by proteol-ysis. Trends Biochem Sci 21:267–271

49. Hyman SE, Comb M, Lin YS, Pearlberg J, Green MR, Goodman HM 1988 Acommon trans-acting factor is involved in transcriptional regulation of neu-rotransmitter genes by cyclic AMP. Mol Cell Biol 8:4225–4233

50. Ismail RS, Dube M, Vanderhyden BC 1997 Hormonally regulated expressionand alternative splicing of kit ligand may regulate kit-induced inhibition ofmeiosis in rat oocytes. Dev Biol 184:333–342

51. Li J, Kim JM, Liston P, Li M, Miyazaki T, Mackenzie AE, Korneluk RG,Tsang BK 1998 Expression of inhibitor of apoptosis proteins (IAPs) in ratgranulosa cells during ovarian follicular development and atresia. Endocri-nology 139:1321–1328

GONADOTROPINS STIMULATE Mcl-1 EXPRESSION IN OVARIAN FOLLICLES 5477