Original article 485

Bone morphogenetic protein 15 (BMP15) alleles predictover-response to recombinant follicle stimulation hormoneand iatrogenic ovarian hyperstimulation syndrome (OHSS)Francisco J. Morona, Francisco de Castrob, Jose L. Royoa, Luis Montorob,Emilia Mirac, Marıa E. Saeza, Luis M. Reala, Alejandro Gonzalezd,Santos Manesc and Agustın Ruiza

Objective Controlled ovarian stimulation (COS) using

recombinant follicle-stimulating hormone (rFSH) is the

main treatment in assisted reproduction. We performed a

pharmacogenetic analysis of bone morphogenetic protein

15 (BMP15) gene using single nucleotide polymorphisms

(SNPs) in COS. We also investigated the role of the BMP15

gene in ovarian hyperstimulation syndrome (OHSS).

Methods We analysed different intragenic SNPs located

within the BMP15 gene in 307 women treated with rFSH,

evaluating its involvement in COS outcome.

Results First, we analysed two polymorphisms, by applying

different tests for genetic association, and we found a

minimum P-value in patients producing Z12 follicles in

COS (high responders) in both polymorphisms of the

BMP15 gene. Using bi-directional DNA sequencing, we

identified two additional single nucleotide DNA variants.

Second, we conducted association studies with all

polymorphisms together, and noticed that none of them

seemed to fully explain the association of the BMP15 gene

with over-response to rFSH. However, N103S missense

mutation is predicted to disrupt the secondary structure of

human BMP15 protein and is weakly associated with

OHSS. This coding mutation of the BMP15 gene may

partially explain the results obtained during our research.

Using Thesias software, we reconstructed haplotypes

with the four intragenic variants and calculated their

frequencies in normal and over-responders to rFSH. The

haplotype TGGA was over-represented in high responders

when compared with the rest of patients. Moreover, this

association was higher in patients with OHSS, with a

significant global haplotypic effect of the BMP15 gene.

Conclusion Our results suggest a direct relationship

between increased follicle production during COS and

BMP15 alleles in response to rFSH in humans. The use of

BMP15 markers to prevent OHSS is also a possibility that

requires thorough evaluation. Pharmacogenetics and

Genomics 16:485–495 �c 2006 Lippincott Williams &

Wilkins.

Pharmacogenetics and Genomics 2006, 16:485–495

Keywords: Controlled ovarian stimulation (COS), BMP15, ovarianhyperstimulation syndrome (OHSS), pharmacogenetics, pyrosequencing

aDepartamento de Genomica Estructural, Neocodex, Sevilla, Spain, bUnidad deReproduccion Humana Asistida, Hospital Universitario Prıncipe de Asturias,Alcala de Henares, Madrid, Spain, cDepartamento de Inmunologıa y Oncologıa,Centro Nacional de Biotecnologıa, Universidad Autonoma de Madrid, Madrid,Spain and dUnidad de Reproduccion y Genetica Humana, Centro Avanzado deFertilidad (CAF), Jerez de la Frontera, Spain.

Correspondence and requests for reprints to Agustın Ruiz MD, PhD,Departamento de Genomica Estructural. Neocodex. Avda. Charles Darwin sn;Parque Tecnologico Isla de la Cartuja, 41092-Sevilla Spain.Tel: + 34-955-047618; fax: + 34-955-047325; e-mail: aruiz@neocodex.es

Sponsorship: This study was supported by the Ministerio de Ciencia y Tecnologıa(MCYT, Spain - grant numbers FIT-010000-2004–68, FIT-010000-2004–69,PTQ2003-0546, PTQ2004-0006 and PTQ2003-0783) and by the EuropeanCommission (QLK4-CT-2002–02403).

Received 21 October 2005 Accepted 30 January 2006

IntroductionTo obtain better pregnancy rates in assisted reproduction

techniques (ART), exogenous gonadotropins are widely

used during controlled ovarian stimulation (COS) proto-

cols. Nowadays, recombinant follicle-stimulating hor-

mone (rFSH) is the main treatment employed in ART

[1], and several studies have shown high variability in

clinical outcome in women undergoing FSH treatment.

This variability may depend on factors such as age of

patients, ovarian reserve, or endocrine status [2]. The

ovulation rate in mammals is determined by a complex

exchange of hormone signals between the pituitary gland

and the ovary, and by a paracrine crosstalk between the

oocyte and its adjacent somatic cells (granulosa and theca

cells). The use of gonadotropin-releasing hormone

(GnRH) agonists (GnRHa) is very important in ovarian

stimulation protocols to suppress oestrogen production in

the ovary via biochemical ablation of the hypothalamus–

hypophysis–ovary axis (pituitary suppression). Conse-

quently, COS pharmacogenetics provide an unique

opportunity to study the regulation of ovarian oestrogen

and follicle production. Calculation of the rFSH dosage in

a specific patient is performed empirically and/or on the

basis of previous results and therefore all oestrogen

production is controlled by exogenous rFSH pharmaco-

logically.

1744-6872 �c 2006 Lippincott Williams & Wilkins

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Usually, COS treatments give rise to different types of

ovarian response, there being two extreme situations:

patients with lower than normal response (poor respon-

ders) and patients with a high response to gonadotropins

(high responders). Moreover, these last patients have a

substantially higher risk of developing ovarian hyperstimu-

lation syndrome (OHSS) and associated complications [3].

OHSS is a well-known iatrogenic complication of ovula-

tion induction and it can be accompanied by severe

morbidity that could be fatal. The clinical course of

OHSS may involve, according to its severity and the

occurrence of pregnancy, cardiovascular effects, electro-

lytic imbalance, pulmonary manifestations, neurohormo-

nal and haemodynamic changes and can contribute to

arteriolar vasodilatation, hepatic anomalies, hypoglobuli-

naemia, febrile morbidity, thromboembolism, neurological

manifestations, and adnexal torsion [4]. Therefore, the

detection of patients with an increased risk of developing

OHSS is a serious clinical problem, and multiple

biomarkers in different and unrelated biochemical path-

ways, such as oestradiol [5], renin–angiotensin system

[6,7], cytokines [8–10], inhibins [11] and soluble

selectins [12], have been evaluated as potential OHSS

predictors. Ovarian hyperstimulation entails very high

serum oestradiol levels and altered progesterone/oestra-

diol ratios. So far, a limited number of fatal cases have

been reported [13–16].

We and others are interested in the study of genes

implicated in COS outcome using the pharmacogenetic

approach (reviewed in reference 3). There is enough

evidence of genetic markers involved in poor response

during COS treatment in women. Georgiu et al. studied

common PvuII and rare BstUI restriction fragment length

polymorphisms (RFLPs) within the ESR1 gene and

suggested that ESR1 alleles might affect pregnancy rate

and oocyte ratio during in-vitro fertilization (IVF) [17].

Sundarrajan et al. partially replicated this work and,

moreover, they found that ESR1 PvuII PP carriers had

lower follicle and oocyte numbers observed during COS

[18]. The Ser680Asn polymorphism in the FSHR gene

was also associated with FSH basal level and elevated

gonadotropin requirements during COS [19,20]. In

addition, we recently proposed that the same polymorph-

ism at the FSHR locus acts as a low-penetrance allele

related to COS outcome and could be dependent on the

basis of a complex multifactorial model [21]. We also

detected an oligogenic model, including specific FSHR,

ESR1 and ESR2 genotype patterns, related to the poor

response to FSH hormone during COS treatment [22].

More recently, Daelmans et al. have argued that the same

polymorphism (Ser680Asn) in the FSHR gene could be a

predictor of severity of symptoms among OHSS patients

[23]. Furthermore, three different natural mutations in

the FSHR gene were identified in patients presenting

with recurrent severe spontaneous OHSS [24–26].

During our search for new candidate genes involved in

COS outcome, we selected the bone morphogenetic

protein-15 gene (BMP15). A genetic study of a natural

mutation in Inverdale and Hanna sheep carrying the

fecundity X (FecXI and FecXH) in the BMP15 gene

provided evidence of the essential role of BMP15 in

folliculogenesis and fertility in mono-ovulating mammals

[27]. Females carrying homozygous mutations compro-

mising BMP15 activity displayed follicular development

arrests at the primary stage with an early block in

folliculogenesis, resulting in infertility. In contrast,

heterozygotes exhibited an increased ovulation rate and

multiple pregnancies. Therefore, BMP15 appears to be

associated with mechanisms of infertility and super

fertility in a dosage-sensitive manner [27].

BMP15 protein is a member of the transforming growth

factor-b (TGF-b) superfamily of proteins. This super-

family controls many aspects of development by binding

and activating two types of transmembrane serine/threo-

nine kinase receptors and SMAD (signalling mothers

against decapentaplegic homologue) proteins to regulate

cellular functions [28]. Given the relative importance of

these factors in the intraovarian regulation of follicular

development and its involvement in ovulation rate, we

considered BMP15 to be a good candidate gene for COS

pharmacogenetic analysis.

Here, we present a genetic association study of COS

outcome in 307 unrelated women with normal ovarian

function who underwent COS using rFSH. We analysed

four single nucleotide polymorphisms located at the

BMP15 gene in order to investigate the role of this gene

in relation to COS outcome. Our results support the

hypothesis that BMP15 alleles predict over-response to

rFSH and ovarian hyperstimulation syndrome in humans.

Materials and methodsPatients

The referral centre is the Hospital Prıncipe de Asturias

de Alcala de Henares (Madrid, Spain). 307 Caucasian

women undergoing IVF or intracytoplasmic sperm injec-

tion (ICSI) techniques were recruited retrospectively

during this work (all patients underwent IVF or ICSI, and

later on they were called and included in this study). The

mean ± SD age of patients was 32.6 ± 2.6 years. The

causes of infertility in selected women were male factor

(174 cases), tubal factor (90 cases) or both of them

(43 cases). Ovulation in all patients included was

conserved (patients with dysovulation were excluded).

Evaluation of the response to FSH in women was

performed using standard criteria and previously pub-

lished parameters [21].

The study design was retrospective. All patients were

treated with rFSH to achieve COS. GnRH analogue was

486 Pharmacogenetics and Genomics 2006, Vol 16 No 7

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

administered to all women included in our study

according to the minimal monitoring method described

by Wikland et al. [29]. Briefly, oestradiol levels of all

patients were down-regulated with a GnRH agonist

according to a long protocol permitting fairly precise

programming of the oocyte collection (triptorelin 0.1 mg/

day over 15 days to reach oestradiol values r 40 pmol/l).

Once minimal oestradiol values were obtained, the

women were treated with 100–200 IU/day of human

rFSH for 11 consecutive days. Follicular development and

growth were evaluated by transvaginal sonography after 6

days of stimulation and then every other day. Ovulation

was induced by a single injection (5000–10 000 IU) of

human chorionic gonadotropin in women with at least

three follicles with a diameter Z 18 mm. Oocytes were

obtained by conventional methods as previously reported

[21]. Conventional IVF or ICSI was carried out in

accordance with original protocols [30,31].

To perform phenotype–genotype correlations and endo-

cohort analyses, we decided to divide our population

according to follicle production in COS outcome. Thus,

we defined two groups: high responders, with patients

producing Z 12 follicles (90th percentile) (n = 35), and

no high responders, with patients producing < 12

follicles (n = 272) (Table 1). In order to test the

predictive value of genetic markers for OHSS, we also

analysed separately an additional group within high

responders in COS, consisting of those women showing

early symptoms or biochemical data of this rare adverse

effect as the strongest endpoint of rFSH over-response

(n = 11). Thus, the treatment was cancelled in all

patients who showed any early manifestation of OHSS.

This treatment cancellation was carried out when the

serum oestradiol levels are > 5000 pmol/l, or when these

levels are between 3000 and 5000 pmol/l, with low body

mass index and/or high ovarian reserve in the first

sonographic surveillance.

Written informed consent was obtained from all patients

participating in the present study. The protocol was

approved by the Institutional Review Board of the

Hospital Prıncipe de Asturias (Alcala de Henares, Madrid,

Spain).

DNA extraction

We obtained 5 ml of peripheral blood from all patients to

isolate germline DNA from leukocytes. DNA extraction

was performed according to standard procedures using

the Nucleospin Blood Kit (Macherey-Nagel, GmbH,

Duren, Germany). Aliquots of DNA at a concentration

of 5 ng/ml were prepared to perform PCR. The rest of the

stock was cryopreserved at – 201C.

Mutation analysis

The DNA sequence used to carry out this study

corresponds to the genomic sequence of BMP15. The

genomic sequence containing the BMP15 gene (Genomic

Contig NT_011638, locus ID 9210) was identified using

the BLAT tool at UCSC (University of California Santa

Cruz) Genome Bioinformatics server (http://www.ucsc.

edu). The gene is di-exonic (328 bp and 851 bp

respectively), spans 5823 bp and is located at Xp11.2

(Fig. 1a). Information concerning any single nucleotide

polymorphism (SNP) identified was compared with the

UCSC Genome Bioinformatics and also with Genbank

Database at the National Center for Biotechnology

Information (http://www.ncbi.nlm.nih.gov/).

Automated DNA sequencing methods were employed to

scan the entire coding region of the BMP15 gene in

selected specimens. Overlapping PCRs covering the

coding sequence (including 50 and 30 UTR regions) of

gene were designed. PCR products were purified and bi-

directionally sequenced using the corresponding primers

(Supplementary Table 1a). Sequencing reactions were

performed using the CEQ Dye Terminator Cycle

Table 1 Genetic and clinical profile of high responders versus no high responders undergoing controlled ovarian stimulation

Parameter All patients High responders (Z12follicles)

No high responders ( < 12follicles)

Statistical analysis

n (patients) 307 35 272– 9C > G* 172/120/15 14/15/6 158/105/9 P = 0.0001b

(CC/CG/GG) (0.72)a (0.12)a OR = 7.57, CI = 2.35–24.36– 673C > T*

(CC/CT/TT)IVS1 + 905A > G 140/141/26 11/17/7 129/124/19 P = 0.004b

(AA/AG/GG) (1.00)a (0.19)a OR = 4.32, CI = 1.49–12.51Age (years)w 32.6 ± 2.6 32.3 ± 2.7 32.7 ± 2.5 P = 0.61c

Cause of sterility (male/tubal/both) 172/92/43 21/8/6 151/84/37 P = 0.63d

Peak oestradiol (pmol/l)w 1200 ± 932 2513 ± 1488 1030 ± 672 P < 0.0001c

Days of stimulationw 10.8 ± 2 9.6 ± 1.6 10.5 ± 2.9 P < 0.0001c

FSH dose (IU)w 12.9 ± 6.7 10.7 ± 5.1 15.4 ± 7.5 P < 0.0001c

Follicle productionw 6.4 ± 3.6 14.3 ± 2.7 5.7 ± 2.7 P < 0.0001c

aDeviation from Hardy–Weinberg equilibrium in cases/controls of patients.bHomozygous tests adapted from Sasieni [33].cTest H Kruskall–Wallis.dChi-square with two degrees of freedom.* – 673C > T and – 9C > G loci are a syntenic variant with same allelic frequencies and in complete LD.wMean value ± standard deviation.

BMP15 gene in COS outcome Moron et al. 487

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Sequencing Quick Start Kit (Beckman Coulter, Inc.,),

according to the manufacturer’s instructions, and ana-

lysed using CEQ 8000 Genetic Analysis System (Beck-

man Coulter, Inc., Fullerton, California, USA).

Genotyping of polymorphisms within the BMP15 gene

using pyrosequencing technology

A total of four SNPs were analysed (Supplementary Table

1b). All genotypes were carried out by using the

pyrosequencing technology protocols [32]. Assays were

run on a thermal cycler machine (MJ Research Inc.,

Waltman, Massachusetts, USA) using a final volume of

20 ml. The selected primers for pyrosequencing analysis

and PCR conditions are shown in Supplementary Table

1(b). The pyrosequencing machine was programmed in

accordance with the manufacturer’s recommendation.

Figure 2(a) shows genotype results from pyrosequencing.

Constructs and cell culture transfections

The 50UTR (50untranslated region) of BMP15 was

amplified to a final volume of 10 ml containing 10 ng of

genomic DNA, 0.2 mmol/l each amplification primer,

1.5 mmol/l MgCl2, 75 mmol/l dNTP and 1 U of High

fidelity Taq Polymerase (Roche Diagnostics, Germany).

Primers, including XhoI and NcoI restriction sites, and

PCR conditions are described in Supplementary Table

1(c). Amplicons were purified, digested and cloned into

pGL3-Basic vector (Promega, Madison, Wisconsin, USA).

Ligation reactions were performed using a rapid ligation

kit (New England Biolabs, Beverly, Massachusetts, USA)

according to the manufacturer’s instructions. Then,

plasmids were chemically transformed to XL1-Blue

competent cells. Colonies were checked for the

presence of the insert and, positive constructs were

sequenced to discard the constructs bearing point

mutations incorporated during the PCR. As a result we

obtained pTG9 for studying the mutant haplotype and

pCC43 as a wild-type vector. These two plasmids were

also shuffled in order to isolate each single SNP within

the vector. Thus we obtained pTC2 and pCG15

(Supplementary Table 2).

Indeed, pGL3-control was digested with BamHI/XbaI and

a 0.3 kb fragment containing the SV40 enhancer was

isolated. This sequence was cloned downstream

of the Luc gene of pCC43, pTG9, pTC2, pCG15 and

Fig. 1

XhoI

TATA-box

−673C>T

ERE−9C>GNcoI

5′upstream sequence 3′downstream sequence

(a)

(b)

−673C>T

−9C>G

Exon 1

rs3810682 rs3897937IVS1+905>G

N103S

Exon 2

328bp

5.82Kb

851bp

tcg

(a) Diagram of BMP15 gene with polymorphisms studied. (b) Plasmid containing the 50UTR of BMP15 directing the expression of the Fireflyluciferase.

488 Pharmacogenetics and Genomics 2006, Vol 16 No 7

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

pGL3-basic vectors, thus generating pECC, pETG,

pETC, pEGG and pGL3-enhancer. Clones were se-

quenced and prepared for cell culture transfection.

Du145, MCF-7, K562 and GH3 cell lines were cultured in

Dulbecco’s modified Eagle’s medium (DMEM) or MEM

supplemented with 20% fetal calf serum at 371C in 95%

humidified air and 5% CO2.

Statistical analysis

To compare quantitative variables among groups, mean

values were analysed by using Analyse-It (version 1.65;

Analyse-It Software Ltd, Leeds, UK) software for the

Kruskall–Wallis test (for non-normally distributed data).

To perform statistical analysis of genotype distribution,

the test for deviation from Hardy–Weinberg equilibrium

or two-point association studies, we employed tests

adapted from Sasieni [33] without a Bonferroni-type

correction because the detected polymorphisms are in

complete linkage disequilibrium (LD). These calcula-

tions were performed at the online resource of the

Institute for Human Genetics, Munich, Germany (http://

ihg.gsf.de).

LD values between markers, haplotype construction and

haplotype analysis for dichotomic traits were performed

using Thesias software (http://www.genecanvas.org)

based on the SEM algorithm [34]. Age-adjusted haplo-

type analysis for quantitative traits (oestradiol levels and

number of follicles) was also carried out using Thesias

software. This method allows one to estimate haplotype

effects by comparison to a reference (the intercept)

taken here as the most frequent one.

Regarding the bioinformatics tools used for the sequence

analysis of the 50UTR in BMP15 gene, it should be stated

that Translational Element Search System (TESS)

software is available at http://www.cbil.upenn.edu/tess.

Dragon Oestrogen Response Elements (ERE) Finder

software was also online interrogated at http://sdmc.

lit.org.sg/ERE-V2/index.

ResultsGenotype analysis

We analysed 307 patients undergoing COS for the allelic

frequency and genotype distribution of two selected

polymorphisms located at the 50UTR, identified as

– 9C > G (rs3810682), and in intron 1, identified as

IVS1 + 905A > G locus (rs3897937) of the BMP15 gene.

Allelic frequencies in our sample were 0.76 for C allele

and 0.24 for G allele in the – 9C > G locus and 0.69 for

A allele and 0.31 for G allele in intron 1 IVS1 + 905A > G

locus. The observed heterozygosity was 0.39 and 0.46

respectively. We observed that the genotype frequencies

were in accordance with Hardy–Weinberg equilibrium

(P > 0.68).

Fig. 2

(a)

(b)

−9C>G

−673C>T

G/G IVS1+905 A>G

G/C

C/C

T/T

T/C

C/C

N103S

(a) Genotype analysis of – 9C > G and IVS1 + 905A > G polymorphisms using pyrosequencing technologies. (b) Sequence chromatographs of theBMP15 gene regions containing DNA variants.

BMP15 gene in COS outcome Moron et al. 489

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

A two-step analysis was conducted. In the first step, we

released a preliminary association studies dividing

patients into poor (r 3 follicles), high (Z 12 follicles)

and normal responders (between 4 and 11 follicles).

Differences in genotype distributions for two polymorph-

isms were explored using chi-squared tests adapted from

Sasieni [33]. Preliminary statistical results indicated that

high responder patients have a different allelic frequency

and genotype distribution in both markers of the BMP15gene when compared to normal responders. In contrast,

poor responders do not display any significant association

compared with normal responders (data not shown).

Therefore, we focused our studies only on patients with

high follicle production and treated women were divided

into two groups, patients producing Z12 follicles (high

responders) and patients producing < 12 follicles (no high

responders), as previously described under methods. The

chi-squared tests adapted from Sasieni (Table 1) showed a

minimum P-value, for risk allele G in both markers, applying

the homozygous test (OR = 7.57, CI = [2.35–24.36],

P = 0.0001 for – 9C > G and OR = 4.32, CI = [1.49–

12.51], P = 0.004 for IVS1 + 905A > G). These statistical

analyses point to the existence of a functional DNA variant

close to the BMP15 locus affecting recombinant FSH-

induced follicle production. For this reason we sequenced

the entire gene looking for functional variants.

Gene sequencing

We determined the complete coding sequence, 1 kb

upstream of ATG and 30UTR of the BMP15 gene in 35

high responders (including 11 OHSS women) using bi-

directional automated capillary DNA sequencing. We

identified two new single nucleotide DNA variants. The

first one appeared again in the 50UTR, and it was

identified as – 673C > T. The other mutation appeared

in the exon 1 c.308A > G locus and resulted in a missense

mutation in codon 103 of the protein (N103S) (Fig. 1).

A second step was conducted using these new variants,

and we again investigated its allelic frequency and

genotype distribution in 307 patients undergoing COS.

Allelic frequencies in our patients were 0.76 for C allele

and 0.24 for Tallele in the – 673C > T polymorphism and

the observed heterozygosity for the locus was 0.39

(similar results to the – 9C > G locus). In the N103S

locus the allelic frequencies were 0.94 for allele N103 and

0.06 for allele S103 and the observed heterozygosity for

this locus was 0.10. The genotype frequencies are in

accordance with Hardy–Weinberg equilibrium (P > 0.58).

The N103S variant does not show a different allelic

frequency and genotype distribution when we compare

high responders to the rest of patients. Interestingly, when

we separated the patients with early symptoms or

biochemical data of OHSS, because it is the hardest end-

point of FSH over-response, we observed that 3 out of 11

OHSS women (27.3%) carried the N103S variant in

heterozygous state. By contrast, 3 out of 24 high responders

(OHSS women excluded) (12.5%) and 27 out of 272 no

high responders (9.9%) carried the N103S variant. In this

regard, over-representation of N103S in OHSS women

compared with no OHSS women, is close to statistical

significance (OR = 3.53, CI = [0.88–14.15], P = 0.06 for

heterozygous test). However, these variants only partially

explain the results obtained during association studies. This

data points to the existence of other variants close to the

BMP15 gene involved in over-response to rFSH and OHSS.

For this reason, we decided to re-construct haplotypes of

the BMP15 gene using the four validated genetic variants

looking for haplotypes in LD with the putative functional

variant of the BMP15 gene.

Haplotype analysis

In order to analyse the degree to which these polymorph-

isms are in LD, we performed standardized pairwise LD

(D0) using Thesias software. The value D0 of ± 1

indicates complete LD, positive when the rare allele of

each polymorphism is preferentially associated and

negative when the rare and frequent alleles are associated

[35]. Thus, we verified that – 673C > T locus is a

syntenic variant of the – 9C > G locus with the same

allelic frequencies and in complete LD and the four

polymorphisms are in complete LD between them

(D0> 0.95 and P < 0.004).

Using Thesias software, we reconstructed haplotypes and

calculated their frequencies, although rare haplotypes

with minor-allele frequencies of less than 5% were

excluded. The program identified four distinct haplo-

types for combinations and permutations from each

genotype of four polymorphisms in our population,

TGGA, TGGG, CCGA and CCAA for – 673C > T,

– 9C > G, IVS1 + 905A > G and N103S polymorphism

respectively (Table 2).

Differences in haplotype distribution between high respon-

ders and no high responders was significant (Pearson Chi-

square with 3 df = 8.51; P = 0.04). Two haplotypes were

statistically significant: haplotype CCAA is a protective

allele (OR = 0.53, CI = [0.31–0.91], P = 0.02), whereas

haplotype TGGA, which is over-represented in high

responders when compared versus no high response

patients (30% versus 17.6%), is a risk allele (OR = 2.00,

CI = [1.10–3.60], P = 0.02) for women in COS over-

response. This phenomenon is due to ‘contrary genetic

sweep’, where the over-representation of one haplotype may

lead to the reduction of the other haplotype.

Intriguingly, the haplotype carrying N103S amino acid

change (TGGG in Table 2), though, it does not reach

statistical significance, has almost identical OR to TGGA

haplotype. In fact, analysis of the haplotypic background

490 Pharmacogenetics and Genomics 2006, Vol 16 No 7

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

showed no effect difference between presence of the A or

G allele at TGG–background (P = 0.95), suggesting that

TGG(G/A) haplotype (including TGGG and TGGA

alleles) carries a functional variant involved in rFSH

over-response. However, given the over-representation of

N103S in OHSS women compared with non-OHSS

women, observed in genotype analysis, and its low

population frequency, we cannot discard that the

N103S locus or other linked variant could modify rFSH

response in COS. Alternatively, and using our marker set,

we can not rule out the fact that both haplotypes (TGGG

and TGGA) might share the same founding mutation.

Given that N103S is not informative about the TGG–

background, we repeated this study with the other three

markers alone. We observed that the statistical signifi-

cance for the global haplotypic effect was slightly

increased with respect to the four markers haplotype

analysis (Pearson chi-square with 2 df = 8.56; P = 0.01).

The TGG haplotype, according to the previous analysis,

is associated with increased risk of rFSH over-response

(OR = 2.15, CI = [1.23–3.73], P = 0.005).

Furthermore, when we compared the patients with early

symptoms or biochemical data of OHSS, versus the rest

of the patients (no OHSS), we observed a strong positive

association with TGG haplotype (OR = 3.95, CI = [1.55–

10.10], P = 0.002). In fact, haplotype frequencies were

inverted in OHSS patients with a significant global

haplotypic effect on the BMP15 gene (Pearson chi-square

with 2 df = 11.21; P = 0.004) (Fig. 3). These last results

point to the presence of one or more functional mutations

in the haplotype background TGG(G/A) associated with a

high follicle production phenotype and OHSS in our

population. In addition, our results also suggest that the

putative functional mutations may have different mole-

cular mechanisms.

We found no associations with any haplotype in the QTL

analysis of peak oestradiol or follicle production. That

could be due to the fact that OHSS women are included

in the high responders group and we have no information

about the final number of follicles production in them,

due to early cancellation. However, a trend towards

association was obtained for peak oestradiol in haplotype

TGGA (n = 298, P = 0.08).

Functional studies

In addition to the missense mutation detected, DNA

variations located at 50UTR of the BMP15 gene might also

be functional. In order to perform a functional character-

ization of the sequence under study, we cloned 1 kb

upstream of the BMP15 gene of both CCAA/CCAA (wild-

type) and TGGA/TGGA (mutant) women. As stated in

Materials and Methods, four constructs were initially

tested and their luciferase activity was compared to the

promoter-less vector. pRL-null vector containing the

Renilla luciferase gene was included for internal normal-

ization, allowing inter- and intra-experimental comparison.

Different cell lines were transfected following previous

experimental procedures, and Luc activity was deter-

mined after a 24 and 48 h incubation. In order to test

different genetic backgrounds, four cell lines were

included in this study: Du145, MCF-7, K562 and GH3.

After three independent experiments, we concluded that

no expression differences could be detected among the

constructs after normalization. In addition, luciferase

activities were similar to those obtained with the pGL3-

basic (promoter-less control). This led us to propose that

specific activation signals may be required for triggering

expression from our vectors in transfected cells.

A detailed in-silico analysis of the BMP15 50UTR using

both TESS and Dragon ERE Finder software programs,

highlighted an oestrogen receptor element (ERE) located

80 bp upstream from the initial ATG (Fig. 1b). For this

reason we stimulated transfected MCF-7 cells with 5 nmol/

l of 17-alpha ethinyloestradiol [36]. Luciferase activity was

measured at 0, 0.5, 1, 4, 8 and 24 h post-induction. No

differences were observed between constructs bearing the

wild-type and mutant 50UTR haplotypes (data not shown).

Moreover, no up-regulation was detected during any

experiment. We also considered the possibility that the

ERE was exerting a negative effect on 50UTR activity. This

regulatory effect may be due to the oestrogenic activity of

the phenol-red from the culture media [37]. In order to

eliminate this possibility, transfected MCF-7 was cultured

in a red-free medium, with identical results.

Previous studies in mice linked BMP15 expression

via stimulation with FSH. Therefore, we looked for

Table 2 Comparison of BMP15 haplotype frequencies, using Thesias software among high responders and no high responders

Haplotype No high responders High responders P-value OR [CI]

Chromosomes % Chromosomes %

I TGGA 96 17.6 21 30 0.02 2.00 [1.10–3.60]II TGGGa 27 5 6 8.6 0.25 1.79 [0.58–4.65]III CCGA 38 7 4 5.7 1.00 0.81 [0.20–2.34]IV CCAA 382 70.2 39 55.7 0.02 0.53 [0.31–0.91]Total 543 99.8 70 100 0.04b

aN103S variant is tracked by this unique haplotype.bThe global haplotypic effect using standard Pearson chi-square with 3 df.

BMP15 gene in COS outcome Moron et al. 491

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

differential promoter activities of the candidate plasmids

under these conditions. After transfection, Du145 were

cultured in serum-free media with/without 0.1 U/ml FSH

(Sigma-Aldrich, St Louis, Missouri, USA) [38]. As for the

above-mentioned experiments, Luc expression was mea-

sured following a time-course kinetic. Again, no induction

was observed upon stimulation and no promoter activity

was detected in cultures transfected with our plasmids

series (data not shown).

Some authors have described the negative feedback

circuit linking the oocyte and the granulosa cells involving

Kit ligand [39]. Therefore, we proposed that by

stimulating with Kit ligand we might modulate BMP15promoter expression. For this purpose, we electroporated

the K562 cell line, which is known to express an active

Kit receptor [40]. Identical kinetic analyses were

performed and neither promoter activity nor differences

between constructs were observed.

Finally, each kinetic analysis was repeated using the

plasmids containing the SV40 enhancer. Because identical

results were obtained we concluded that: (1) the 1 kb

upstream from the initial ATG codon does not contain

the minimal promoter sequence of BMP15; or (2) that

specific transcription factors expressed in the oocyte are

required for triggering BMP15 expression; and/or (3) that

alternative (or additional) stimuli should be present in

order to promote transcription initiation of the sequence

under study. Finally, we should also consider the

possibility that the tight control of BMP15 expression

might be mediated through the action of specific

repressors present in non-oocyte cells.

DiscussionSeveral studies have shown high variability in clinical

outcome of women undergoing COS using rFSH hormone

[3]. In this regard, there are two opposite situations

involved in COS outcome, a poor or a high response

inducing OHSS. The existence of genetic factors

involved in poor response has been described in previous

studies [17–21] and we have presented direct evidence of

the polygenic nature of the COS trait in humans [22]. In

contrast, there are not many studies for genetic markers

associated with over-response to FSH or high follicle

production. However, Daelemans et al., in a recent

genetic study of the Ser680Asn polymorphism of the

FSHR gene, suggest that the genotype in position 680 of

the FSHR cannot predict which patients will develop

OHSS, but it could be a predictor of severity of symptoms

among OHSS patients [23].

In this way, and employing the candidate gene approach,

we selected the BMP15 locus to explore its involvement

in COS outcome in women undergoing assisted repro-

ductive techniques using rFSH hormone. BMP15 is a

good candidate gene because in animal models other

groups have demonstrated the important role in ovarian

development and folliculogenesis played by the action of

the BMP15 protein. Five separate point mutations in the

BMP15 gene have been identified affecting the ovulation

rate in ewes [27,41,42] and BMP15 knock-out female

mice are subfertile and show reduced ovulation rates after

gonadotropin treatment [43]. The analysis of the results

obtained in both models suggests that BMP15 may not

affect follicular development and ovulation rate in the

same way in all mammal species.

Fig. 3

OHSSCCG

CCG

CCA

TGG

CCA

TGG

CCG

CCA

TTG

Haplotype

1.00

0.009

0.002

Haplotype frequencies (%)

(n=11)

no OHSS(n=296)

0 20 40 60 80

P value

Global Chi-square with 2 df=11.21 and Global P value=0.004.TGG haplotype including TGGG and TGGA(N103S) haplotypes.

3.95 [1.55−10.10]

0.30 [0.12−0.77]

0.62 [0.024−4.08]

OR [CI]

Estimated haplotype frequencies in patients with OHSS versus rest of patients.

492 Pharmacogenetics and Genomics 2006, Vol 16 No 7

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

We analysed the clinical outcome of 307 COS cycles and

the role of four polymorphisms ( – 673C > T, – 9C > G,

IVS1 + 905A > G and N103S) located within the BMP15gene in rFSH efficacy. Our results support the hypothesis

that the BMP15 locus may not be implicated in poor

response (r 3 follicles) to COS. In contrast, we have

enough evidence regarding the existence of association

among three linked markers within the BMP15 gene with

high follicle production in women undergoing rFSH

stimulation (high responders, Z 12 follicles). This result

could be due to population stratification. To ensure that

our population did not present selection bias or popula-

tion stratification, we repeated the allele frequency and

genotype distribution in the BMP15 gene for a population

including 1210 unrelated postmenopausal women and we

observed a similar allele frequency and genotype

distribution (data not shown). This result is in accor-

dance with previous observations of Galloway et al. in

ewes with heterozygous FecXI mutant females that

exhibited increased ovulation and lambing rates [27].

These observations indicate that high responder women

might have a functional variant close to the BMP15 locus

involved in COS outcome. Moreover, an independent

study in a family affected by hypergonadotropic ovarian

failure identified a heterozygous non-conservative sub-

stitution (Y235C) in the pro-region of the BMP15 allele in

two sisters affected by this Mendelian phenotype [44].

This observation reinforces the role of our candidate in

ovarian physiology and underlines the fact that the

BMP15 signalling pathway is also important for folliculo-

genesis and/or ovarian development in humans.

When we sequenced the functional BMP15 gene in 35

over-responder women, the non-conservative mutation

described by Di Pasquale et al. (the only BMP15 mutation

described for humans to date) was not identified in the

Spanish population. However, a new missense variant,

N103S, was identified and, its association with iatrogenic

OHSS is close to reach statistical significance. Never-

theless, the functionality of the N103S variant needs

further characterization. Although statistical analyses

indicate that N103S might partially explain the genetic

association detected, results from the TGG(G/A) haplo-

type analysis clearly suggest that another functional

variant must be involved in COS over-response.

We suspect that some mutations of 50UTR of the BMP15gene could be related to the COS outcome phenotype.

Therefore we fused the 1 kb 50UTR plus the first four

BMP15 amino acids to the luciferase reporter gene in

order to determine the Luc enzymatic activity expressed

in both the wild-type and mutant BMP15 haplotypes.

However, no expression differences were detected among

the constructs after normalization. The reason for the

lack of functionality of these haplotypes in these

experiments requires detailed discussion. We feel that

specific repression sequences, including the YY1/delta

factor found in the 50UTR and promoter region of the

human BMP15 gene, impaired the proper activation of

the putative minimal promoter outside oocyte cells (data

not shown). In view of this, specific oocyte-derived

transcription factors might be necessary to activate the

human BMP15 promoter.

On the basis of this hypothesis, several mechanisms were

tested using previous results in animal models. BMP15 is

a potent stimulator of granulosa cell growth and promotes

the progression of folliculogenesis from the primary stage

to the FSH-dependent stage [45]. BMP15 has a double

function in granulosa cells, since there is a negative

feedback system between oocyte and granulosa cells,

where BMP15 stimulates Kit ligand expression and

mitotic activity in granulosa cells, whereas Kit ligand is

a potent inhibitor of BMP15 mRNA expression in oocytes

[39]. On the other hand, BMP15 inhibits FSH action by

suppressing FSHR expression [46] and FSH plays an

essential role in folliculogenesis and female fertility; for

example, FSH promotes proliferation and differentiation

of preantral follicles via paracrine factors such as insulin

growth factor-1 and activin [47,48]. In addition, FSH is

crucial for an appropriate modulation of Kit ligand and

BMP15 in promoting oocyte growth [49].

Our preliminary functional studies did not prove that a

functional mutation in BMP15 50UTR exists. It is possible

that our approach is not suitable for assessing the

function, due to the lack of expression outside of oocyte

cells. A possible reason for this is that an additional

sequence up- and/or down-stream would be necessary for

transcription initiation.

In conclusion, although we obtained enough genetic

evidence of BMP15 gene involvement in COS over-

response and OHSS, the molecular mechanism for the

genetic association observed remains unclear. Perhaps the

N103S variant itself, or a different variation linked to

TGG(G/A) haplotype in the BMP15 gene or in another

gene in tight linkage disequilibrium, could be the

principal cause of over-response in COS associated with

the BMP15 haplotypes that we detected.

For a complete understanding and corroboration of our

findings the analysis of the BMP15 gene in independent

populations would be necessary. Furthermore, we have

studied women with conserved ovulation (women with

dysovulation were excluded) since patients with poly-

cystic ovarian syndrome are considered at high risk for the

development of OHSS [4]. However, it could be of

interest to analyse the role of BMP15 genes in patients

with polycystic ovarian syndrome. Functional character-

ization of BMP15 N103S protein and risk haplotypes

detected would be necessary to delineate a mechanistic

hypothesis of BMP15 involvement in COS outcome in

humans.

BMP15 gene in COS outcome Moron et al. 493

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Our results suggest the existence of a new pathway for

activating oestrogen and follicle production in humans as

it occurs in other mammals. Thus, and according to data

provided by sheep and mouse models, a reduced level of

BMP15 might result in higher levels of FSHR in the

granulosa cells, high levels of oestrogens, activated Kit

ligand (a potent inhibitor of BMP15 mRNA expression in

oocyte) and increased follicle production with the

possible presentation of a side effect related to gonado-

tropin hypersensitivity and OHSS.

Predicting which patients will develop OHSS during COS

is one of the main challenges for the clinician involved in

assisted reproductive techniques [3]. Our results are the

first to show a direct link between increased follicle

production and BMP15 alleles during controlled ovarian

hyperstimulation and provide a new potential tool which

may be useful in human pharmacogenetics studies of

COS treatment and OHSS prediction.

AcknowledgementsWe are deeply grateful to the patients for their

participation in this study. We are grateful to Ana Salinas,

Eva Molero and Rocıo Pascual who provided excellent

technical, informatics and English grammar support,

respectively, during this work.

References1 Salha O, Balen AH. New concepts in superovulation strategies for assisted

conception treatments. Curr Opin Obstet Gynecol 2000; 76:1185–1190.2 Kligman I, Rosenwaks Z. Differentiating clinical profiles: predicting good

responders, poor responders, and hyperresponders. Fertil Steril 2001;76:1185–1190.

3 de Castro F, Moron FJ, Montoro L, Real LM, Ruiz A. Pharmacogenetics ofcontrolled ovarian hyperstimulation. Pharmacogenomics 2005; 6:629–637.

4 Delvigne A, Rozenberg S. Review of clinical course and treatment of ovarianhyperstimulation syndrome (OHSS). Hum Reprod 2003; 9:77–96.

5 Navot D, Relou A, Birkenfeld A, Rabinowitz R, Brzezinski A, Margalioth EJ.Risk factors and prognostic variables in the ovarian hyperstimulationsyndrome. Am J Obstet Gynecol 1988; 159:210–215.

6 Navot D, Margalioth EJ, Laufer N, Birkenfeld A, Relou A, Rosler A,Schenker JG. Direct correlation between plasma renin activity and severity ofthe ovarian hyperstimulation syndrome. Fertil Steril 1987; 48:57–61.

7 Morris RS, Paulson RJ. Increased angiotensin-converting enzyme activity in apatient with severe ovarian hyperstimulation syndrome. Fertil Steril 1999;71:562–563.

8 Levin ER, Rosen GF, Cassidenti DL, Yee B, Meldrum D, Wisot A, Pedram A.Role of vascular endothelial cell growth factor in ovarian hyperstimulationsyndrome. J Clin Invest 1998; 102:1978–1985.

9 Chen CD, Chen HF, Lu HF, Chen SU, Ho HN, Yang YS. Value of serum andfollicular fluid cytokine profile in the prediction of moderate to severe ovarianhyperstimulation syndrome. Hum Reprod 2000; 15:1037–1042.

10 Barak V, Elchalal U, Edelstein M, Kalickman I, Lewin A, Abramov Y.Interleukin-18 levels correlate with severe ovarian hyperstimulationsyndrome. Fertil Steril 2004; 82:415–420.

11 Enskog A, Nilsson L, Brannstrom M. Peripheral blood concentrations ofinhibin B are elevated during gonadotrophin stimulation in patients who laterdevelop ovarian OHSS and inhibin A concentrations are elevated afterOHSS onset. Hum Reprod 2000; 15:532–538.

12 Daniel Y, Geva E, Amit A, Eshed-Englender T, Bar-Am A, Lessing JB.Soluble endothelial and platelet selectins in serum and ascitic fluid ofwomen with ovarian hyperstimulation syndrome. Am J Reprod Immunol2001; 45:154–160.

13 Mozes M, Bogokowsky H, Antebi E, Lunenfeld B, Rabau E, Serr DM, et al.Thromboembolic phenomena after ovarian stimulation with humangonadotrophins. Lancet 1965; 2:1213–1215.

14 Cluroe AD, Synek BJ. A fatal case of ovarian hyperstimulation syndrome withcerebral infarction. Pathology 1995; 27:344–346.

15 Beerendonk CC, van Dop PA, Braat DD, Merkus JM. Ovarianhyperstimulation syndrome: facts and fallacies. Obstet Gynecol Surv 1998;53:439–449.

16 Semba S, Moriya T, Youssef EM, Sasano H. An autopsy case of ovarianhyperstimulation syndrome with massive pulmonary edema and pleuraleffusion. Pathol Int 2000; 50:549–552.

17 Georgiou I, Konstantelli M, Syrrou M, Messinis IE, Lolis DE. Oestrogenreceptor gene polymorphisms and ovarian stimulation for in-vitro fertilization.Hum Reprod 1997; 12:1430–1433.

18 Sundarrajan C, Liao WX, Roy AC, Ng SC. Association betweenestrogen receptor-beta gene polymorphisms and ovulatory dysfunctionsin patients with menstrual disorders. J Clin Endocrinol Metab 2001; 86:135–139.

19 Perez Mayorga M, Gromoll J, Behre HM, Gassner C, Nieschlag E, Simoni M.Ovarian response to follicle-stimulating hormone (FSH) stimulationdepends on the FSH receptor genotype. J Clin Endocrinol Metab 2000;85:3365–3369.

20 Sudo S, Kudo M, Wada S, Sato O, Hsueh AJ, Fujimoto S. Genetic andfunctional analyses of polymorphisms in the human FSH receptor gene. MolHum Reprod 2002; 8:893–899.

21 de Castro F, Ruiz R, Montoro L, Perez-Hernandez D, Sanchez-Casas Padilla E,Real LM, Ruiz A. Role of follicle-stimulating hormone receptor Ser680Asnpolymorphism in the efficacy of follicle-stimulating hormone. Fertil Steril 2003;80:571–576.

22 de Castro F, Moron FJ, Montoro L, Galan JJ, Hernandez DP, Padilla ES, et al.Human controlled ovarian hyperstimulation outcome is a polygenic trait.Pharmacogenetics 2004; 14:285–293.

23 Daelemans C, Smits G, de Maertelaer V, Costagliola S, Englert Y, Vassart G,Delbaere A. Prediction of severity of symptoms in iatrogenic ovarianhyperstimulation syndrome by follicle-stimulating hormone receptorSer680Asn polymorphism. Clin Endocrinol Metab 2004;89:6310–6315.

24 Smits G, Olatunbosun O, Delbaere A, Pierson R, Vassart G, Costagliola S.Ovarian hyperstimulation syndrome due to a mutation in the follicle-stimulating hormone receptor. N Engl J Med 2003; 349:760–766.

25 Vasseur C, Rodien P, Beau I, Desroches A, Gerard C, de Poncheville L, et al.A chorionic gonadotropin-sensitive mutation in the follicle-stimulatinghormone receptor as a cause of familial gestational spontaneous ovarianhyperstimulation syndrome. N Engl J Med 2003; 349:753–759.

26 Montanelli L, Delbaere A, Di Carlo C, Nappi C, Smits G, Vassart G,Costagliola S. A mutation in the follicle-stimulating hormone receptor asa cause of familial spontaneous ovarian hyperstimulation syndrome. J ClinEndocrinol Metab 2004; 89:1255–1258.

27 Galloway SM, McNatty KP, Cambridge LM, Laitinen MP, Juengel JL,Jokiranta TS, et al. Mutations in an oocyte-derived growth factor gene(BMP15) cause increased ovulation rate and infertility in a dosage-sensitivemanner. Nat Genet 2000; 25:279–283.

28 Chang H, Brown CW, Matzuk MM. Genetic analysis of the mammaliantransforming growth factor-beta superfamily. Endocr Rev 2002;23:787–823.

29 Wikland M, Borg J, Hamberger L, Svalander P. Simplification of IVF: minimalmonitoring and the use of subcutaneous highly purified FSH administrationfor ovulation induction. Hum Reprod 1994; 9:1430–1436.

30 Lopata A, Johnston IW, Hoult IJ, Speirs AI. Pregnancy following intrauterineimplantation of an embryo obtained by in vitro fertilization of a preovulatoryegg. Fertil Steril 1980; 33:117–120.

31 Edwards RG, Steptoe PC, Purdy JM. Establishing full-term humanpregnancies using cleaving embryos grown in vitro. Br J Obstet Gynaecol1980; 87:737–756.

32 Ronaghi M, Karamohamed S, Pettersson B, Uhlen M, Nyren P. Real-timeDNA sequencing using detection of pyrophosphate release. Anal Biochem1996; 242:84–89.

33 Sasieni PD. From genotypes to genes: doubling the sample size. Biometrics1997; 53:1253–1261.

34 Tregouet DA, Escolano S, Tiret L, Mallet A, Golmard JL. A new algorithm forhaplotype-based association analysis: the Stochastic-EM algorithm. AnnHum Genet 2004; 68:165–177.

35 Lewontin RC, Kojima K. The evolutionary dynamics of complexpolymorphisms. Evolution 1960; 14:450–472.

36 Falany JL, Falany CN. Regulation of oestrogen activity by sulfation in humanMCF-7 breast cancer cells. Oncol Res 1997; 9:589–596.

37 Ortmann O, Sturm R, Knuppen R, Emons G. Weak estrogenic activity ofphenol red in the pituitary gonadotroph: re-evaluation of estrogen andantiestrogen effects. J Steroid Biochem 1990; 35:17–22.

494 Pharmacogenetics and Genomics 2006, Vol 16 No 7

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

38 Ben-Josef E, Yang SY, Ji TH, Bidart JM, Garde SV, Chopra DP, et al.Hormone-refractory prostate cancer cells express functional follicle-stimulating hormone receptor (FSHR). J Urol 1999; 161:970–976.

39 Otsuka F, Shimasaki S. A negative feedback system between oocyte bonemorphogenetic protein 15 and granulosa cell kit ligand: its role inregulating granulosa cell mitosis. Proc Natl Acad Sci USA 2002;99:8060–8065.

40 Ramenghi U, Ruggieri L, Dianzani I, Rosso C, Brizzi MF, Camaschella C,et al. Human peripheral blood granulocytes and myeloid leukemic cell linesexpress both transcripts encoding for stem cell factor. Stem Cells 1994;12:521–526.

41 Bodin L, Lecerf F, Pisselet C, SanCristobal M, Bibe B, Mulsant P. How manymutations are associated with increased ovulation rate and litter size andprogeny of Lacaune meat sheep? Workshop on Major Genes and QTL inSheep and Goat, Toulouse, France, 8–11 December 2003.

42 Hanrahan JP, Gregan SM, Mulsant P, Mullen M, Davis GH, Powell R,Galloway SM. Mutations in the genes for oocyte-derived growth factorsGDF9 and BMP15 are associated with both increased ovulation rate andsterility in Cambridge and Belclare sheep (Ovis aries). Biol Reprod 2004;70:900–909.

43 Yan C, Wang P, DeMayo J, DeMayo FJ, Elvin JA, Carino C, et al. Synergisticroles of bone morphogenetic protein 15 and growth differentiation factor 9in ovarian function. Mol Endocrinol 2001; 15:854–866.

44 Di Pasquale E, Beck-Peccoz P, Persani L. Hypergonadotropic ovarian failureassociated with an inherited mutation of human bone morphogeneticprotein-15 (BMP15) gene. Am J Hum Genet 2004; 75:106–111.

45 Otsuka F, Yao Z, Lee T, Yamamoto S, Erickson GF, Shimasaki S. Bonemorphogenetic protein-15. Identification of target cells and biologicalfunctions. J Biol Chem 2000; 275:39523–39528.

46 Otsuka F, Yamamoto S, Erickson GF, Shimasaki S. Bone morphogeneticprotein-15 inhibits follicle-stimulating hormone (FSH) action by suppressingFSH receptor expression. J Biol Chem 2001; 276:11387–11392.

47 Adashi EY, Resnick CE, Hurwitz A, Ricciarelli E, Hernandez ER, Roberts CT,et al. Insulin-like growth factors: the ovarian connection. Hum Reprod 1991;6:1213–1219.

48 Miro F, Hillier SG. Modulation of granulosa cell deoxyribonucleic acidsynthesis and differentiation by activin. Endocrinology 1996; 137:464–468.

49 Thomas FH, Ethier JF, Shimasaki S, Vanderhyden BC. Follicle-stimulatinghormone regulates oocyte growth by modulation of expression of oocyteand granulosa cell factors. Endocrinology 2005; 146:941–949.

BMP15 gene in COS outcome Moron et al. 495

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.