An evolutionary conserved interaction between the Gcm transcription factor and the SF1 nuclear receptor in the female reproductive system

In Drosophila , the complete lack of the Gcm protein leads to embryonic lethality due to the transformation of glial cells into neurons 24 30 31 . Viable hypomorphic mutations, however, allow the analysis of gcm mutant animals at later stage 24 31 32 33 : the gcm rA87 allele is due to the insertion of a P-element containing the LacZ gene in the promoter of gcm and the gcmGal4 allele has been produced upon replacement of the LacZ by the Gal4 gene 24 31 32 33 . The gcmGal4 homozygous and the transheterozygous gcmGal4/gcm rA87 animals reach adulthood and display fertility defects. To assess whether the defects are sex specific, we crossed wild type (WT, Oregon-R) males with gcmGal4 homozygous or with transheterozygous females and found a significantly reduced number of offspring compared to that obtained in control crosses (<1% and 20% of the progeny, respectively, Fig. 1a ). In contrast, fertility assays on transheterozygous males showed no fertility defects (data not shown). Thus, Gcm is required in reproduction in females, in addition to its well-known role in glia and blood development 24 30 31 34 35 36 37 38 . To clarify the role of Gcm on fertility, we crossed the gcmGal4 driver, which faithfully mimics the expression of Gcm 32 33 , with a UAS-RFP reporter. RFP expression was detected in the adult spermatheca, while no labelling was observed in the ovaries nor in the oviduct ( Fig. 1b ). Gcm expression in the adult spermatheca was confirmed by qPCR assays ( Supplemental Figure S1 ). Finally, a GO-term analysis on a genome-wide screen aiming at identifying the direct targets of Gcm 39 specifically highlighted the genes involved in the reproductive system development as the most enriched class of genes after those involved in nervous system development, in line with the known role of Gcm at the glial determinant 24 30 31 ( Fig. 1c ). Comparison between this screen and the published transcriptome of the spermatheca 13 revealed that 387 direct targets of Gcm are expressed in this organ ( Fig. 1d , list in Supplemental Table S1 ). This data indicate that Gcm is necessary for female fertility and that it is expressed in the spermatheca. Two elegant studies 14 40 showed that the spermatheca of Drosophila contains a layer of lumen epithelial cells (LEC) expressing the Runt-domain transcription factor Lozenge (Lz), which is essential for the development of the whole spermatheca 14 ( Fig. 2a ). Surrounding the LEC is the layer of SC that express and require the transcription factor Hindsight (Hnt) 15 . Accessory cells are located basal (basal cells, BC) to the SC and apical (apical cells, AC) to the LEC. The AC are thought to secrete a cuticular canal that connects the secretory unit to the lumen of the spermatheca, which contains the spermatozoids. AC and BC undergo apoptosis during pupal spermatheca development, with some BC being still present in young adult females 14 . To assess the mode of action of Gcm, we analysed the morphology of the spermathecae in animals carrying altered levels of Gcm. The WT SC appear as a translucent layer of cells surrounding a dark cuticular structure that is produced by the LEC ( Fig. 2b ). In hypomorphic gcm conditions ( gcmGal4 homozygous animals), the SC layer is completely absent, leaving the dark cuticular structure relatively unaffected ( Fig. 2c ). Accordingly, immunolabelling assays show a complete lack of SC in homozygous gcmGal4 females ( Fig. 2f,h ), which leads to the absence of spermatozoids in the spermathecae ( Supplemental Figures S2a–d ). The lack of SC in gcm homozygous females is also observed in other hypomorphic gcm conditions such as transheterozygous gcmGal4/gcm rA87 animals and can be rescued by overexpressing Gcm ( Supplemental Figures S2e–g ). Of note, some spermathecae from transheterozygous gcmGal4/gcm rA87 animals show few remaining SC ( Supplemental Figure S2h’ ), explaining why this strain is not completely sterile. In addition, the number of SC significantly decreases when Gcm is knocked-down by RNAi ( gcm KD ) using the gcmGal4 as a driver ( Fig. 2g,h ). The egg laying rate is in agreement with this data. A positive correlation was previously made between the number of SC of the spermathecae and the number of eggs laid 15 and indeed the number of SC as well as the egg laying rate decrease in gcm hypomorphs ( Fig. 2h,i ). The reduction in SC number no longer persists in gcm KD spermathecae that also carry the UAS-gcm transgene. Indeed, these spermathecae carry supernumerary SC ( Fig. 2h ), suggesting that Gcm expression may be sufficient to induce the differentiation of the SC. Finally, to analyse the phenotype of a null gcm allele, MARCM clones were produced using the Df ( 2L ) 132 strain in which the gcm gene is completely deleted 33 41 ( Supplemental Figure S2i ). Similar clonal analyses were also performed using a gcm hypomorphic but lethal mutation induced by P-element mutagenesis, gcm 34  24 ( Fig. 2j–k ”’). Recombination was induced at the 3 rd instar larval stage prior to spermatheca differentiation. WT clones contain both cell types (SC and LEC), whereas Df(2L)132 and gcm 34 mutant clones contain LEC but completely lack SC. Thus, Gcm is necessary for the differentiation of SC. Given the strong phenotype observed in loss of function gcm alleles, we assessed the consequences of overexpressing Gcm in its own territory of expression in WT animals ( gcm  >  gcm GOF ). In these gain of function (GOF) animals, the dark cuticular structure and the LEC are present but the morphology of the spermatheca is altered ( Fig. 2d , Supplemental Figures S2j-j” ). In addition, these spermathecae display a very high number of SC ( Fig. 2h ). Altogether, this data clearly indicate that Gcm is expressed and required in the spermatheca to control SC differentiation. The mutant phenotype prompted us to assess the role and the mode of action of Gcm. Given the early and transient expression of Gcm in glial cells 24 30 31 , we analysed the mutant spermathecae and the profile of Gcm expression during development. Spermathecae develop during the pupal stage and the different cell types arise from multipotent precursors (MP) that express Lz and divide to produce the lumen epithelium precursors (LEP) also expressing Lz as well as the secondary precursors called Secretory Unit Precursors (SUP), which do not express Lz 14 . Each SUP divides and produces the AC and a tertiary precursor, which in turn divides and produces the SC as well as the BC that undergoes apoptosis at the adult stage ( Fig. 3h ) 14 . First, we knocked down Gcm expression using the lzGal4 driver ( lz  >  gcm KD ), which is active in the MP. Like in hypomorphic conditions and in gcm  >  gcm KD animals, RNAi-mediated down-regulation of Gcm in the MP leads to the decrease of the number of SC in the adult spermatheca ( Fig. 3a,b ) and the LEC are not impacted ( Fig. 3b–b ”). The similar phenotypes obtained with gcm > and lz >, a driver that is not active in the SUP 14 , suggest that the gcm promoter is already active in the MP that generates all cell types of the spermatheca (including SC and LEC). We then proceeded to overexpress Gcm under the control of the lzGal4 driver ( lz  >  gcm GOF ) and found that this leads to severe spermatheca defects including a deformed and almost absent cuticular structure. This phenotype is stronger than the overexpression of Gcm using the gcmGal4 driver ( gcm  >  gcm GOF ) in which the cuticular structure can still be observed (compare Fig. 3c’ and Supplemental Figures S2j–j” ). This indicates that premature Gcm expression prevents LEC development and suggests that Gcm is expressed below threshold levels in the MP. Following this, we tracked the lineage of the Gcm expressing cells by crossing the g-trace flies 42 with the gcmGal4 flies and found that both SC and LEC originate from cells expressing Gcm (white asterisks and dashed line, respectively, in Fig. 3d–d ”’). In addition, we tracked Gcm expression during spermatheca development using the gcm rA87 βGal reporter in heterozygous conditions. By 24hrs after puparium formation (APF), after the division of the MP, the SUP co-expresses Gcm and Hnt (full arrowheads in Fig. 3e–e”’ ) 15 and some MP can still be seen co-expressing Lz and βGal (empty arrowheads in Fig. 3e–e ”’). At 48 hrs and 72 hrs APF, three types of cells can be identified: the LEC expressing exclusively Lz, the cells expressing Hnt and low levels of Gcm, which comprise the SC ( Fig. 3f ”’), and the BC expressing Gcm and almost no Hnt ( Fig. 3f,g ). Few apoptotic AC can also be detected, expressing Hnt ( Fig. 3f ”’). This confirms that Gcm and Lz are transcribed in the MP and that Gcm remains expressed in the SUP and its offspring. Finally, in the adult spermatheca, cell-specific immunolabelling on animals carrying the gcmGal4 driver and the UAS-mCD8GFP reporter ( gcm  >  GFP , Supplemental Figures S3a, S3c–c” ) and anti-βgal labelling on the enhancer trap line gcm rA87 in heterozygous conditions ( Supplemental Figure S3b ) indicate that Gcm is expressed exclusively in the adult BC. These are the cells that undergo apoptosis 14 ( Supplemental Figures S3c–c” ), as shown by the decreased number of labelled cells in old gcm  >  GFP spermathecae compared to young ones ( Supplemental Figures S3d–e’ ). Of note, the number of BC decreases in the gcm  >  gcm GOF spermathecae that instead present a very high number of SC ( Fig. 2h , Supplemental Figure S3f ), suggesting that the BC may convert into SC in gcm  >  gcm GOF spermathecae. Collectively, our data show that Gcm starts to be expressed in the MP, specifies SUP differentiation and triggers the differentiation of the SC. Hr39 and Hnt are two transcription factors involved in the development of the spermatheca: knock out as well as KD of hnt and Hr39 lead to defective production of SC in the spermatheca 13 14 15 . In addition, they both contain canonical Gcm binding sites (GBS) 43 44 and were identified as direct targets of Gcm by the genome-wide screen using the DNA adenine methyltransferase identification (DamID) procedure 39 45 ( Fig. 4a,d ). To validate our data functionally, we analysed the regulation of Hnt and Hr39 by Gcm in S2 cells transfected with a Gcm expression vector. The levels of Hr39 transcripts are significantly induced by Gcm ( Fig. 4b ). Next, we built luciferase reporters carrying the two GBS present in the Hr39 locus where Gcm is binding according to the DamID screen and reporters carrying the mutated GBS. Upon co-transfection with the Gcm expression vector, both GBS present in the Hr39 locus induce luciferase activity and mutations of either GBS reduces the luciferase expression levels ( Fig. 4c ), indicating that Gcm induces Hr39 expression through these two GBS ( Fig. 4a–c ). The endogenous levels of Hnt are not significantly induced by Gcm in S2 cells ( Fig. 4e ), however, the hnt locus possesses one GBS in the promoter region ( Fig. 4d ) and a luciferase assay similar to that performed on Hr39 indicates a significant induction of hnt reporter expression by Gcm, which decreases upon GBS mutagenesis ( Fig. 4f ). Thus, Gcm is also able to induce the expression of Hnt through the GBS. The lack of induction of the endogenous Hnt in S2 cells is likely due to the absence of cofactors or to the unavailability of the enhancer region targeted by Gcm. In all cases, the mutation of the GBS does not abolish the induction of the luciferase activity completely. This may be due to an indirect effect of Gcm on these promoters or to the presence of non-canonical GBS. Overall, this data indicate that Gcm promotes Hr39 expression and likely contributes to the induction of Hnt expression as well. Finally, we complemented this data by assessing the biological relevance of the interaction between Gcm and Hr39. Since gcm KD in the MP ( lzGal4 driver) leads to a decrease in SC number at adult stage ( Fig. 3a,b ), we overexpressed Hr39 in lz  >  gcm KD spermathecae and found rescue of the mutant phenotype ( Fig. 4g,h ). The increased number of SC in the adult compared to that observed in animals that only express low levels of Gcm strongly suggests that Hr39 is indeed a major target of Gcm in the development of the female reproductive system. Of note, Hr39 is already detected in the genital discs of the late 3 rd instar larvae 13 14 suggesting that the role of Gcm is not to initiate Hr39 expression but to maintain or increase Hr39 expression during the first division of the MP after pupal formation. The closest mammalian orthologs of the Hr39 gene are Nr5a1 and Nr5a2, which code respectively for SF-1 13 and LRH-1 14 and are both involved in the formation and function of mammalian reproductive tissues 46 47 48 . We hence assessed whether the functional conservation includes the regulation of Nr5a1 and Nr5a2 by the orthologs of Gcm: mGCM1 and mGCM2. First, we measured the endogenous levels of hNR5A1 and hNR5A2 in HeLa cells (human) and those of mNR5A1 and mNR5A2 in mouse embryonic fibroblasts (MEF) upon transfection of mGCM1 and mGCM2 expression vectors. While the levels of expression of hNR5A2/mNr5a2 are not modulated by the mGCM proteins, the expression levels of the hNR5A1 / mNr5a1 transcripts significantly increase when either mGCM proteins are expressed ( Fig. 5a–e ). In HeLa cells, both mGCM1 and mGCM2 induce hNR5A1 expression at similar levels ( Fig. 5a ) and in MEF, mGCM2 induces mNR5A1 expression at higher levels than mGCM1 (9-fold increase compared to WT with mGCM1 versus 6E5-fold increase with mGCM2) ( Fig. 5b,c ). Then, quantitative PCR (qPCR) analyses indicate that mGcm1 , mGcm2 and mNr5a1 are expressed in the adult mouse uterus and that their levels of expression in this tissue are higher than those found in liver and testes ( Fig. 5f ). It is important to note, however, that their levels are one order of magnitude lower than the transcription factor Msx1 , which is known to be strongly active in the uterus 49 ( Fig. 5f ). In situ hybridisation assays confirm the expression of mGcm2 mostly in the stroma of the endometrium ( Fig. 5g ). No signal could be detected using the mGcm1 probe, likely due to the low levels of mGcm1 expression it that tissue. This suggests that the regulation of Hr39 expression by Gcm observed in Drosophila is conserved in evolution and that mGCM proteins might regulate the expression of mNR5A1 in the mouse uterus. Mammalian GCM proteins have been associated with DNA demethylation at the promoter of their target genes: hGCM1 affects Syncytin 2 demethylation in human placenta 50 and mGCM1 and mGCM2 affect Hes5 demethylation in the mouse embryo 29 . For this reason, it was proposed that mGCM proteins trigger DNA demethylation, even though the molecular mode of action was not understood. To further characterize the impact of the mGcm genes, we asked whether mNR5A1 regulation by mGCM1 and mGCM2 is associated with changes in the DNA methylation profile of the mNr5a1 gene using transfected cells. In human and mouse, the Nr5a1 genes contain a CpG island that covers the transcription start site (TSS) until the 3 rd exon ( Fig. 6a ). The methylation rate of each CpG within the regions covering the 2 nd exon and the TSS was estimated by bisulfite sequencing in MEF transfected with an empty vector (Control) or with expression vectors of mGCM1 or mGCM2. The three CpG located around the TSS are demethylated in the presence of mGCM1 or mGCM2 proteins compared to that observed upon transfecting the control plasmid ( Fig. 6b ). In addition, a significant increase in CpG methylation is observed in the exon 2 region upon mGCM1 or mGCM2 transfection ( Fig. 6c ). The highest levels of methylation are observed when the cells are transfected with mGCM2 ( Fig. 6c ), in agreement with the strong increase in mNr5a1 expression levels observed in MEF cells overexpressing mGCM2 ( Fig. 5e ). These data show that the mGCM proteins are not specifically involved in DNA demethylation and fit with the emerging view that gene expression is linked to DNA demethylation at the promoter and to DNA hypermethylation in the gene body 51 ( Fig. 6d ). Our data are also in line with the recent hypothesis that transcription factors can bind demethylated as well as methylated DNA 52 . Finally, the high levels of expression of hNR5A1/mNR5A1 observed in endometriotic tissues are also linked to high levels of CpG methylation around exon 2 and low levels around the TSS 53 54 55 56 . Overall, this data suggest that the mGcm genes induce the transcription of mNr5a1 and this is associated with important changes in the DNA methylation profile at the mNr5a1 locus.

Flies were raised on standard medium at 25 °C. The genotype and provenance of the strains are detailed in Supplemental experimental procedures . Fertility and egg laying assays are detailed in Supplemental experimental procedures . For fertility assays, the progeny produced in 12 days by 3 virgins of the indicated genotypes crossed with one male WT were counted and reported to number of progeny/female. For the egg laying assays, the number of eggs laid in 48 hrs by five females of the indicated genotypes crossed with ten males WT were counted and reported to number of eggs/females/days. The p-values were estimated after variance analysis using bilateral student test with equal variance. The spermathecae were labelled using standard immunolabelling protocol as described in ref. 39 . The list of antibodies and the labelling protocol are detailed in in Supplemental experimental procedures . For each spermatheca, the Hnt/DAPI positive cells (SC) were counted from the stack of six focal plans taken at 3 μm interval in the middle of the spermatheca (the plan giving the largest cross-section of the spermatheca). This was repeated in at least six independent spermathecae for each genotype. The average number of SC and the s.e.m. are represented in Figs 2 h and 4 h. The p-values were estimated as described for the fertility assays. The transfection of S2 cells, the quantitative PCR (qPCR) and the luciferase assay were performed as described in Cattenoz et al . 39 and detailed in Supplemental experimental procedures . Each experiment was carried out in triplicates. RNA in situ hybridisation with digoxigenin-labelled probes for mGcm2 transcripts was performed as described in Vernet et al . 86 with slight modifications detailed in Supplemental experimental procedures . The qPCR were carried out on C57BL/6 mouse uterus RNA extracted from 3 different animals with TRI reagent. HeLa cells transfection was performed as described in Cattenoz et al . 39 and MEF cells transfection was performed as detailed in Supplemental experimental procedures . 48 hrs after transfection, the cells were sorted according to GFP expression before RNA extraction. Reverse transcription and qPCR were carried out as described for the S2 cells with the primer pairs listed in Supplemental experimental procedures . MEF cells transfected and sorted as described above were used to analyse the methylation profile of mNr5a1 locus. The procedure is detailed in Supplemental experimental procedures . The loci of interest were then amplified by PCR, cloned and sequenced. At least 10 clones were sequenced per condition. The p-values were estimated after variance analysis using bilateral student test for paired samples.

How to cite this article : Cattenoz, P. B. et al . An evolutionary conserved interaction between the Gcm transcription factor and the SF1 nuclear receptor in the female reproductive system. Sci. Rep. 6 , 37792; doi: 10.1038/srep37792 (2016). Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

In this study, we discover a molecular cascade required in the Drosophila female reproductive system that may be conserved in mammals. The Drosophila transcription factor Gcm is expressed during the development of the SC of the spermatheca and mutations or knock-down of Gcm inhibit the development of these cells, leading to female sterility. Gcm acts by targeting the ortholog of the hNR5A1/mNR5A1 hormone receptor Hr39 . Finally, the orthologous genes mGcm1 and mGcm2 are expressed in the mouse uterus, induce the expression of hNR5A1/mNR5A1 in human and murine cell lines, respectively and modify the DNA methylation profile of mNr5a1 . This suggest that defects in the hGCM pathway may be associated with pathologies affecting women reproductive system. Gcm is required in the nervous, in the immune and in the reproductive systems. These Gcm dependent pathways display a common feature as, in all cases, a multipotent precursor gives rise to cells with different identities. In the nervous system, the neuroblast can produce glia or neurons, in the immune system the prohemocyte can produce plasmatocytes or crystal cells and in the spermatheca the MP can produce SC or LEC. Gcm is absolutely required to induce one fate over the other as gcm mutant animals lack glia and display supernumerary neurons 24 30 31 and the number of plasmatocyte decreases whereas that of the crystal cells increases 36 38 . In the spermatheca, the absence of SC in homozygous gcmGal4 animals is accompanied by an increase in LEC number ( Fig. 2h , Supplemental Figure S4 ), suggesting that Gcm induces the differentiation of the SC at the expense of the LEC. A second common feature between the three developmental events is the transient and early expression of Gcm. In the spermatheca, Gcm is expressed during the differentiation of the SC but no longer present in the adult SC. Similarly, Gcm is expressed early in the glial and in the hemocyte lineages but its transcripts are not detected in the mature cells 34 36 57 58 . Thus, the Gcm fate determinant provides a trigger that needs to be erased to allow terminal differentiation. In the nervous system, Gcm activates the transcription of its target gene repo , which remains expressed in glial cells until adulthood 59 . The Repo homeobox containing protein constitutes the pan-glial specific transcription factor that induces the expression of late glial genes, maintains the glial fate and actually contributes to Gcm degradation 57 59 (Trebuchet, unpublished results). In the spermatheca, Gcm induces the expression of the Hr39 transcription factor that is required for SC formation and that remains expressed in those cells until adulthood 13 14 15 , Hr39 may hence play a maintenance role similar to that played by Repo in the glial cells. Recent data suggest that early and transient expression of fate determinants may be a general rule that allows stable and terminal cell differentiation. Interestingly, the Drosophila proneural transcription factor Atonal (Ato) is expressed early during photoreceptor differentiation but needs to be switched off for normal eye development 60 . A third common feature between the three systems is the participation of the Notch pathway. In the spermatheca, the production of the SC from the initial MP encompasses three cells divisions. The first and third divisions involve the Notch pathway and trigger the differentiation of the LEP and the SC respectively 14 15 . In these two divisions, Notch is activate only in the cells that do not express Gcm suggesting that Notch and Gcm may interact negatively. Such negative interaction was previously reported during the differentiation of the adult sensory organ precursors (SOP). Constitutive activation of the Notch pathway in the SOP represses gcm expression and prevents the production of glial cells; accordingly, lack of Notch induces gcm expression and the production of glia at the expense of neurons 61 62 . Finally, during the development of the embryonic hemocytes, there is no report of interaction between Gcm and the Notch pathway, however Gcm is involved in plasmatocyte development and Notch in crystal cell development 34 36 63 . Importantly, several members of the Notch pathway are directly regulated by Gcm including the two ligands Serrate and Delta 39 , which suggests a strong interaction between Gcm and Notch that remains to be investigated. Our work also highlights the cell-specific nature of the Gcm differentiation pathways: while the Gcm transcription factor is required to induce several cell identities, its downstream factors are cell-specific. Repo expression is absent in the spermatheca and Hr39 expression is absent in glial cells. Moreover, the overexpression of Gcm in the spermatheca does not activate Repo expression in those cells nor does Gcm overexpression in the embryonic nervous system activate Hr39 expression in that territory (data not shown). Thus, although ‘master regulators’ are considered as simple molecular switches, this represents an oversimplified view of cell differentiation. The activity of such potent transcription factors rather relies on the history of a given cell, that is, its specific transcriptional and epigenetic asset. For example, the ectopic expression of the famous eyeless master gene induces the formation of ectopic eyes on wings, legs and antennae 64 , while in the embryonic nervous system its ectopic expression alters the axonal wiring of the ventral nerve cord 65 . Finally, the expression profile of Gcm gives an important insight on spermatheca differentiation. Our study shows that Gcm and Lz are co-expressed in the MP and that Gcm remains expressed exclusively in the SUP following the asymmetrical division of the MP whereas Lz is repressed in the SUP 14 . A comparable interaction between Gcm and Lz was observed during the differentiation of the embryonic hemocytes. Gcm is required for the differentiation of the plasmatocytes and Lz for the differentiation of the crystal cells 38 . Initially, Gcm is expressed in all prohemocytes but subsequently its expression fades away in the precursors of the crystal cells, which allows for the expression of Lz 34 37 38 . Thus, Gcm induces the plasmatocyte fate and inhibits the crystal cell fate through inhibition of Lz: as mentioned above, gcm mutant animals display supernumerary crystal cells and in addition ectopic Gcm expression in the crystal cell precursors using the lzGal4 driver prevents the expression of Lz and converts cells into plasmatocytes 36 38 . We propose that in the spermatheca, Gcm is expressed at low levels in the MP where it cohabits with Lz, then its expression progressively rises in the SUP until its levels become sufficient to repress Lz expression in this cell. SUP cells that express low levels of Gcm adopt the LEC fate. The absence of Gcm binding sites at the lz locus and the known role of Gcm as an activator of transcription prompt us to speculate that Gcm represses Lz expression indirectly. The transcriptional repressor Tramtrack (Ttk) 66 67 was already described as an inhibitor of Lz expression in the larval eye disc 68 , is a downstream target of Gcm 39 69 70 71 and is expressed in the spermatheca 72 . Future studies will determine whether Ttk could act as the intermediary protein between Gcm and Lz inhibition. Hr39 and NR5A1 transcription factors were proposed to share similar functions and to target similar genes for the development of specific secretory glands of the reproductive system (steroidogenic glands in mammals and spermathecae in Drosophila ) 13 14 73 . Our study suggests that the control of Hr39 and NR5A1 by the GCM protein family is also conserved. Gcm controls female fertility due to its effects on the SC in the spermathecae, and the lack of SC is explained by the lack of induction of Hr39. This regulation is conserved in mammals with the mGcm1 and mGcm2 genes inducing mNR5A1 expression in MEF cells and being expressed in the reproductive system. This represents the first evidence of functional conservation for GCM proteins in similar biological systems of Drosophila and mammals. Endometriosis 20 21 22 23 54 74 is an oestrogen-dependent disorder defined by the ectopic growth of endometrium-like tissue (reviewed in ref. 75 ), which represents the leading cause of women infertility 76 77 78 . A major feature of endometriotic tissues is the overexpression of hNR5A1 and the modification of the hNR5A1 DNA methylation profile in that tissue: the hNR5A1 TSS is demethylated and the CpG island covering exon 2 is hypermethylated 22 53 54 55 ; the present study shows that mNr5a1 expression and its DNA methylation profile are regulated by the two mGCM proteins ( Fig. 6d ). This suggests that the GCM protein family could be involved in the pathogenesis of endometriosis. Over the past ten years, several studies aimed at identifying the molecular basis of endometriosis by comparing the transcriptomes of healthy endometrium to endometriotic tissue 56 79 80 81 82 83 84 . hGCM1 and hGCM2 did not come out in any of these studies. The large majority of these reports used micro-array to profile gene expression and both hGCM1 and hGCM2 were below the detection range in all studies even in healthy tissues whereas we detected mGcm1 expression by qPCR and mGcm2 expression by qPCR and in situ hybridisation. Several factors may explain the difficulty to identify the hGCM genes in those analyses, among them the known instability of their RNA and their potential transient expression (reviewed in ref. 85 ). This indicates that the study of GCM1 and GCM2 in endometriosis should be carried out using highly sensitive methods and possibly during the development of the disease to catch the transient presence of their transcripts. Overall, our study suggests that the GCM regulatory network is robustly conserved and that Drosophila represents a model of choice to decipher this pathway in the reproductive system. Finally, this study indicates that the Gcm transcription factor has a much broader role than initially thought. We foresee that the deep analysis of its regulatory network will allow us to understand pleiotropic differentiation pathways and hence the role and mode of action of potent fate determinants.