Dlx5 and Dlx6 are two closely associated homeobox genes which code for transcription factors involved in the control of steroidogenesis and reproduction. Inactivation of Dlx5/6 in the mouse results in a Leydig cell defect in the male and in ovarian insufficiency in the female. DLX5/6 are also strongly expressed by the human endometrium but their function in the uterus is unknown. The involvement of DLX5/6 in human uterine pathology is suggested by their strong downregulation in endometriotic lesions and upregulation in endometrioïd adenocarcinomas. We first show that Dlx5/6 expression begins in Müllerian ducts epithelia and persists then in the uterine luminal and glandular epithelia throughout post-natal maturation and in the adult. We then use a new mouse model in which Dlx5 and Dlx6 can be simultaneously inactivated in the endometrium using a Pgrcre/+ allele. Post-natal inactivation of Dlx5/6 in the uterus results in sterility without any obvious ovarian involvement. The uteri of Pgrcre/+; Dlx5/6flox/flox mice present very few uterine glands and numerous abnormally large and branched invaginations of the uterine lumen. In Dlx5/6 mutant uteri, the expression of genes involved in gland formation (Foxa2) and in epithelial remodelling during implantation (Msx1) is significantly reduced. Furthermore, we show that DLX5 is highly expressed in human endometrial glandular epithelium and that its expression is affected in endometriosis. We conclude that Dlx5 and Dlx6 expression determines uterine architecture and adenogenesis and is needed for implantation. Given their importance for female reproduction, DLX5 and DLX6 must be regarded as interesting targets for future clinical research.
In adult eutherians, the endometrium consists of a monostratified columnar epithelium (luminal epithelium, LE) separating the lumen of the uterus from the stroma (S), a dense mesenchyme containing tubular endometrial glands lined by the glandular epithelium (GE) (1,2). The endometrium is surrounded by circular and longitudinal layers of smooth muscle, which support uterine contractions. The initial development of the uterus takes place prenatally from the central segment of the Müllerian ducts (3,4). Although most female reproductive tract (FRT) organs deriving from the Müllerian ducts are fully developed at birth, uterine morphogenesis and differentiation is only completed postnatally (5).
The neonatal mouse uterus consists of a simple LE supported by a rather undifferentiated mesenchyme and lacks endometrial glands (2). During the first three post-natal days (PN0–PN3), the mesenchyme gives rise to three separate layers: the radially oriented endometrial stroma, the inner circular myometrial layer and the prospective outer longitudinal myometrium. At PN6, the LE starts to form epithelial invaginations that prefigure the buds of future endometrial glands (5). By PN12, endometrial glands present a differentiated GE and extend from the LE into the surrounding endometrial stroma. In the mouse, the definitive adult structure of the uterus is attained at PN15 (5). Disruption of post-natal uterine adenogenesis and mesenchymal specification and differentiation can hamper fertility in the adult (6,7). Moreover, the presence of well-formed and functional endometrial stroma and myometrium is a prerequisite for endometrial receptivity and decidualization as well as for expulsion of the foetus at term (8). During the preimplantation period, endometrial glands secrete the factors required for uterine receptivity and embryo implantation (9–12). Post-natal uterine maturation is a complex morphogenetic process governed by stromal–epithelial interactions coordinated by dynamic signalling networks and by circulating hormones after puberty (13).
Dlx genes comprise a highly conserved family of homeobox genes homologous to the Distal-less (Dll) gene of Drosophila. During embryonic development, Dlx genes are expressed in craniofacial primordia, basal telencephalon and diencephalon, in the apical ectodermal ridge of the limb buds. Deletion of the coding and intergenic regions of Dlx5 and Dlx6 with a single targeting event in the mouse results in perinatal death, limb malformations reminiscent of the human ectrodactyly, Split Hand Foot Malformation type I (SHFM1) and in the transformation of lower jaws into upper-like jaws (14–16). Besides these patterning defects, these same mutants have also permitted to identify the roles for Dlx5 and Dlx6 in the control of cell differentiation. In particular, Dlx5/6 play a role in the control of bone development (17) and in the differentiation of Leydig cells (18). Dlx5 and Dlx6 are also expressed in granulosa cells of ovarian follicules and control follicular maturation and steroidogenesis. Heterozygous mutant females Dlx5/6+/− have a reduced fertility due to rapid follicular depletion (19). Several studies have shown that DLX5 is also expressed in the human endometrium. Indeed, high-throughput profiling experiments have shown that, beside the ovary, DLX5 and DLX6 are also upregulated during the secretory phase in the human endometrium (20). In endometrial pathologies, several studies have shown deregulation of DLX5 in comparison to healthy endometrium. For example in endometriosis, a condition characterized by the presence of uterine tissue outside the uterine cavity, DLX5 is one of the most downregulated genes in endometriotic lesion compared with eutopic endometrium (21) and in endometroid adenocarcinomas DLX5 expression is strongly upregulated (22,23).
Here, we have analysed the function of Dlx5 and Dlx6 in the mouse uterus. Simultaneous conditional inactivation of Dlx5 and Dlx6 in the uterus during post-natal maturation results in reduced adenogenesis and abnormal lumen architecture, with a large number of wide invaginations that extended along the uterus. The LE morphology is impaired and appears cuboidal. We provide evidence of the fact that DLX5 is also expressed by human uterine epithelia and that its expression is deregulated in certain forms of endometriosis suggesting that DLX5 and DLX6 could represent interesting targets to tackle this condition.
Dlx5 expression during uterine development
The distribution of Dlx5 during uterine development was analysed in Dlx5LacZ/+ mice. At 15.5 days post coitum (dpc), just after sexual differentiation, Dlx5 is expressed by the medial part of the Müllerian ducts which later in development, give rise to the uterus (Fig. 1A). At E16.5 and E18.5, Dlx5 expression persists in the uterus and appears also in the cervix (Fig. 1B and C). Histological analysis of Dlx5LacZ/+ mice and in situ hybridization show that, at E18.5, Dlx5 is specifically expressed by the uterine epithelium (Fig. 1D–F).
Dlx5 and Dlx6 are expressed in luminal and glandular epithelia in the post-natal uterus
At all stages of uterine maturation, Dlx5-lacZ is expressed by luminal and glandular epithelia: at P6 in the forming epithelial buds (Fig. 2A and A′), at PN8 in the first uterine GE (Fig. 2B and B′) and then in all luminal and glandular epithelia (Fig. 2C and C′). In the adult uterus, β-galactosidase activity persists in the GE and, non-uniformly, in the LE (Fig. 2D and D′). In situ hybridization confirms that both Dlx5 and Dlx6 transcripts are present in all luminal and glandular epithelia (Fig. 2E and E′, and F and F′).
Dlx5 and Dlx6 conditional inactivation in the uterus
We generated mice permitting the simultaneous conditional deletion of exon II of Dlx5 and Dlx6 (Dlx5/6flox/flox); to avoid cross-recombination, exon II of Dlx6 was flanked by lox5171 (24) sites which recombine with themselves but not with wild-type loxP sites (Fig. 3A). Using a generalized cre-recombinase mouse, we could recapitulate all aspects of the craniofacial phenotype of Dlx5/6−/− mice (16) validating this new mouse model. We then crossed Dlx5/6flox/flox mice with Progesterone receptor Cre knockin (PgrCre/+) mice (25) to obtain conditional ablation of Dlx5 and 6 in the uterus. The Pgr promotor is active from P3 in the uterine epithelium and from P6 in the uterine stroma (26). At this stage, Pgr is not expressed in the granulosa of immature follicles (25,27). At P25, in PgrCre/+; Dlx5/6flox/flox the recombinant allele was present in the uterus but not in the tail, or in the ovary (Fig. 3B). qPCR analysis confirmed the absence of Dlx5 and Dlx6 transcripts from the uterus of PgrCre/+; Dlx5/6flox/flox mice (Fig. 3C).
Pgrcre driven Dlx5/6 inactivation results in infertility
PgrCre/+; Dlx5/6flox/flox females are not fertile (Table 1). The ovaries of PgrCre/+; Dlx5/6flox/flox and WT mice did not present any obvious difference at P25 and in the adult. All classes of follicles, including corpora lutea in the adult, could be detected in WT and mutant animals suggesting normal ovarian function (Supplementary Material, Fig. S1). In adults, oestrous cycle is also normal in mutant and heterozygous mice. Despite these apparently normal features, after crossing with WT males, adult PgrCre/+; Dlx5/6flox/flox females never remained pregnant suggesting that their infertility depended from a defect in implantation. To check this hypothesis, we performed uterine embryotransfer. Seventy blastocysts were flushed from the uterine horns of six WT B6D2F1N donor females, randomly divided, and transferred into the uterus of six recipient pseudopregnant females, three PgrCre/+; Dlx5/6flox/flox and three control littermate females. All females were sacrificed 3 days after the transfer, to verify the presence of embryo implants in the uterine wall. None of the PgrCre/+; Dlx5/6flox/flox mice had implants, versus 20 implantation sites found in the uterine horns of control littermates.
|Total of females crossed||Number of plug||% of plug giving a litter||Average litter size (standard deviation)||Uterine embryo transfer (implants/embryo)|
|Pgr+/+; Dlx5/6flox/flox||16||61||65.2||8.6 (±2.17)||20/35|
|Pgrcre/+; Dlx5/6flox/+||7||13||53.8||6.3 (±2.98)||Not tested|
|Total of females crossed||Number of plug||% of plug giving a litter||Average litter size (standard deviation)||Uterine embryo transfer (implants/embryo)|
|Pgr+/+; Dlx5/6flox/flox||16||61||65.2||8.6 (±2.17)||20/35|
|Pgrcre/+; Dlx5/6flox/+||7||13||53.8||6.3 (±2.98)||Not tested|
Conditional inactivation of Dlx5/6 in the uterus results in impaired adenogenesis and abnormal luminal shape
Control Dlx5/6flox/flox mice present a small uterine lumen with few, short, invaginations which are often terminated by uterine glands (5). At PN25, PgrCre/+; Dlx5/6flox/flox present a significantly lower number or absence of uterine glands than in controls (1.681 ± 0.3064 versus 7.639 ± 1.328; P = 0.0024) (Fig. 4A–C), suggesting a severe defect of uterine adenogenesis. Furthermore, PgrCre/+; Dlx5/6flox/flox uteri present a larger lumen than undeleted littermates characterized by many large and branched invaginations extending deeply into the uterine stroma almost in contact with the myometrium (Fig. 4A). To better analyse the lumen phenotype, we performed 3D segmentation of a 512-µm-long uterine portion: in PgrCre/+; Dlx5/6flox/flox animals, invaginations are profound and continuous over large uterine segments and almost invariably present a bifurcated distal end. In contrast, in control animals invaginations are present only punctually along the uterus (Fig. 4D). At histological examination, the LE of PgrCre/+; Dlx5/6flox/flox uteri appears abnormal: control LE presents elongated nuclei localized in the basal pole while, in mutant animals, the cytoplasm is not elongated at the apical pole and the nucleus is round and localized in the centre of the cell (Fig. 4B–D).
Effect of Dlx5/6 inactivation on the expression of genes involved in uterine maturation
We then analysed the effect of Dlx5/6 conditional inactivation on the level of expression of key genes of uterine post-natal maturation (Fig. 5). As expected, in conditionally deleted uteri, Dlx5 and Dlx6 expression are not detectable. Several Wnt genes are known to orchestrate post-natal maturation (4). In particular, Wnt7a plays a role in glandular and myometrial formation (28) and regulates other Wnt genes and Hoxa10 and Hoxa11 (29), which play an important role in stromal proliferation. In PgrCre/+; Dlx5/6flox/flox uteri at PN25, Wnt7a, expressed in the LE, and Wnt11, expressed in both GE and LE, are upregulated. Msx1 and Msx2 are expressed by LE cells and are important for the process of implantation (30,31). In PgrCre/+; Dlx5/6flox/flox uteri, the expression of both Msx genes is deregulated: Msx1 is downregulated while Msx2 is upregulated. In line with the observation of a severe reduction in the number of uterine glands, Foxa2, normally expressed by uterine GE is downregulated in mutant uteri. Surprisingly leukemia inhibitory factor, which is mostly produced by uterine glands, is upregulated. Hoxa10 and Hoxa11, which are expressed in the stroma, are downregulated.
RNA-seq transcriptome analysis
To analyse the molecular regulations associated with the mutant phenotype, we have compared by RNA-seq the transcriptomic signatures of normal and mutant uteri at PN25.
In total, 424/16 748 genes are differentially expressed; 195 are over expressed and 229 are under expressed (Supplementary Material, Tables S1–S3). As histological inspection of the mutant uteri displayed a strong reduction in the number of uterine secretory glands, we also identified the subpopulation of secreted genes in differentially expressed genes (Supplementary Material, Table S4). We confirmed that between the most deregulated genes are those involved in uterine maturation (Wnt7a, Wnt11, Msx2, Foxa2), but also genes important for fertility and implantation (cystathionine beta synthase, ErbB4).
Surprisingly, the most upregulated gene in mutant uteri (log 4.7) codes for oviductal glycoprotein 1 (Ovgp1), a glycosylated protein found in the epithelial cells of the oviduct which is the only specific tubal marker so far (32) suggesting a partial ‘homeotic’ transformation of the uterus into an oviduct.
To assess if there was any enrichment of genes belonging to functionally related biological processes, we analysed the distribution of Gene Ontology (GO) terms in differentially expressed genes. Compared with the frequency of GO terms amongst all genes encoded by the genome, we observed significant (P ≤ 0.01) enrichment of terms for cell adhesion-migration (GO:0016477), angiogenesis (GO:0001525), epithelial cell differentiation (GO:0030855), regulation of cell proliferation (GO:0042127), immune system development (GO:0045859) amongst genes over expressed.
The most significantly enriched GO terms amongst the downregulated genes were cell cycle (GO:0007049), chromatin dynamics (GO:0000070), response to stress (GO:0033554), microtubule organization (GO:0032886), developmental process involved in reproduction (GO:0003006) (Fig. 6). STRING analysis of our findings confirms a decrease in cell proliferation (Supplementary Material, Fig. S2).
Altered expression of DLX5 in human endometriosis
It has been previously shown by transcriptome analysis that DLX5 is the most downregulated gene in ectopic endometriotic lesions (21). We show that, as in the mouse, in the human endometrium, DLX5 is present in the endometrial GE. By immunohistochemistry, a strong signal is detected (Fig. 7) in all normal human uterine GE with an apparent cytoplasmic and nuclear distribution similarly to what has been previously reported for Dlx5 in the developing brain (33). In contrast, in the epithelium lining endometriotic tissues, DLX5 immunostaining is weaker and discontinuous with segments of DLX5-negative and DLX5-positive epithelium. The epithelium of endometriotic lesions was often cuboidal and abnormal when compared with eutopic uterine GE. In regions presenting a strongly reduced DLX5 staining, epithelial cells were not coherent and tended to desquamate suggesting an important role of Dlx5/6 for epithelial integrity. It should also be noted that endometriotic lesions almost never present endometriotic glands with a phenotype reminiscent of PgrCre/+; Dlx5/6flox/flox uteri.
It has been previously show that Dlx5 and Dlx6 are expressed at several levels of the FRT. Although the function of Dlx5/6 in the ovary has been previously analysed (19), very few data were so far available on the distribution and function of these genes at other levels of the FRT. With this study we provide new insight to understand the physiological role of Dlx5/6 in the post-natal uterus. We show that conditional inactivation of Dlx5 and Dlx6 in during maturation of the mouse uterus results in reduced adenogenesis, abnormal lumen architecture and LE morphology. We also show that DLX5 is expressed by human uterine LE and GE and that its deregulation is associated with endometriosis. Collectively our data suggest that Dlx5 and Dlx6 could be important actors for healthy implantation and development of the embryo.
Dlx5 and Dlx6 are involved in uterine adenogenesis
The absence of Dlx5/6 provokes a severe reduction in UG formation. In line with the morphological observation, Foxa2 expression was strongly downregulated in PgrCre/+; Dlx5/6flox/flox uteri. Indeed, Foxa2 is a master gene for uterine adenogenesis (6) implicated also in the development of endoderm-derived organs such as the liver, pancreas, lung and prostate (34,35). Post-natal conditional inactivation of Foxa2 in the uterus leads to a total absence of UG formation and infertility (6). Our results suggest that Dlx5 and Dlx6 are upstream activators of Foxa2 during uterine glandular development.
Wnt family genes organize pre- and post-natal development of the uterus through complex epithelial–mesenchymal interactions and several members of the Wnt canonical pathway are involved in regulation of uterine adenogenesis (36).
Wnt7a is a key regulator of adenogenesis. Wnt7a inactivation (28) or conditional inactivation in the uterus (29) leads to defect in adenogenesis and provokes infertility associated to the downregulation of Foxa2 (29). We show that the inactivation of Dlx5 and Dlx6 provokes an upregulation of Wnt7a, suggesting that the homeostasis of uterine adenogenesis is assured through a complex regulatory network involving Foxa2 Wnt-dependent regulation. This hypothesis is supported by the fact that a similar set of genes is deregulated in PgrCre/+; Dlx5/6flox/flox as in Wnt7a conditional mutant (29), for example Msx1 is downregulated, Msx2 is upregulated, Hoxa10 and Hoxa11 are downregulated. It has been suggested that the downregulation of Wnt7a in single cells of the LE might determine the sites of future glandular development (28).
Deregulation of epithelial genes in the absence of Dlx5/6
Wnt11 and Wnt7a are upregulated in PgrCre/+; Dlx5/6flox/flox, suggesting that Dlx5 and Dlx6 are repressors of these genes. Wnt11 is expressed in LE and GE, and Wnt7a is expressed only in the LE. While, as stated before, inactivation of Wnt7a provokes the absence of glands, Wnt11 conditional inactivation does not produce any obvious uterine phenotype. The only deregulation observed in the absence of Wnt11 is the downregulation of Vangl2 (36), a gene known to act on the definition of cellular planar polarity. Indeed, Vangl2 mutants present an altered cell polarity in the LE (37). Therefore, the upregulation of Wnt11 observed in our Dlx5/6 conditional mutant might explain the abnormal cell polarity of the LE observed in PgrCre/+; Dlx5/6flox/flox uteri (Fig. 4).
Moreover, two other homeobox genes involved in the control of epithelial signalling are deregulated in PgrCre/+; Dlx5/6flox/flox uteri: Msx1 is downregulated and Msx2 is upregulated. Msx1 is expressed during peri-implantation in the LE. Msx1/2 conditional mutants have an implantation failure due to an absence of epithelial polarization similar to what is observed in the upregulation of Wnt5a. An elegant study by Cha et al. (38) has shown that, during implantation, Wnt5a plays a role in crypt formation and particularly in apical polarity of LE. Although the levels of expression of Wnt5a, ROR2, ROR1 and Hbegf did not change significantly in our experiments, it would be interesting to analyse more precisely the effects of implantation on their expression in PgrCre/+; Dlx5/6flox/flox uteri.
Analysis of our RNA-seq results indicates that: ROR2 is differentially expressed: P value adj = 0.54, log2 fold change −0.94, Hbegf is also differentially expressed: P value adj = 0.57, log2 fold change 0.57 while ROR1 is not differentially expressed.
As these genes have different roles and expression level during implantation, it would be interesting to look at 3,5-4 dpc.
The increasing number of luminal invaginations in Dlx5/6 mutant could be linked to the reduction of adenogenesis. These uterine invaginations are at the origin of the UG formation, the absence of Dlx5/6 could provoke an abnormal fusion of the LE at the bottom of these invaginations which would fail to form glands. Moreover, the change in epithelial cell morphology, suggests that Dlx5/6 affect the epithelial cytoskeleton permitting the formation of glands.
An intriguing hypothesis might be that in the absence of Dlx5/6 the uterus undergoes a partial ‘homeotic’ transformation into an oviduct. This possibility is suggested both by molecular indications including the surge of the oviduct-specific gene Ovgp1 (32) and the decreased expression of Hoxa10 and by the morphological similarity between the mutant uterus and the oviduct with the absence of glands and complex luminal invaginations.
Most of the results obtained through our qPCR candidate gene approach have been confirmed by transcriptomic analysis. The reduction of GO terms related to the cell cycle would suggest that cell proliferation is impaired in mutant uteri while epithelial cell differentiation and angiogenesis increase. The observed histological phenotype characterized by a larger surface of LE could well explain the upregulation of these genes.
Implications for human pathology
We have also shown that DLX5 is strongly expressed by human glandular uterine epithelium.
In a separate study, we are going to present a comparative analysis of the pattern of expression of DLX5 in normal and pathological endometria. Here we have focused our attention only on the distribution of DLX5 in endometriotic lesions, as it has been shown that, when comparing eutopic to ectopic tissues, DLX5 is the most downregulated gene (21). We observe that in endometriosis, changes in epithelial morphology are accompanied by a strong reduction of the expression of DLX5. Remarkably, in human endometriosis the epithelia lose their columnar organization and present an altered morphology which resembles closely that observed in conditionally inactivated mice. It seems, therefore, that DLX5 expression plays a role in determining epithelia polarity and integrity as also suggested by the association of Dlx5 with microtubule organization (GO:0032886) (Fig. 6).
Materials and Methods
Procedures involving animals were conducted in accordance with the directives of the European Community (council directive 86/609) and the French Agriculture Ministry (council directive 87–848, 19 October 1987, permissions 00782 to GL). Mice were housed in specific pathogen-free and light, temperature (21°C) and humidity (50–60% relative humidity) controlled conditions. Food and water were available ad libitum.
By targeted homologous recombination, we flanked exon 2 of Dlx5 and Dlx6 between two different types of LoxP sites to avoid cross-recombination. Neo and Hygro were used as selective markers, respectively, for Dlx5 and Dlx6 and were inserted in our targeting construct between FRT sites in order to be excised by flippase recombinase. The targeting protocol was performed in two stages: we first generated mice containing LoxP sites flanking Dlx5, then we used ES cells derived from these mice to introduce a Dlx6 allele in which exon 2 is flanked by two Lox5171 sites. Animals were backcrossed with wild-type B6D2F1/N from Charles River Laboratory. To obtain Dlx5LacZ/+ mice, we crossbreed Dlx5LacZ/+ males with B6D2F1N females. To obtain Pgrcre/+; Dlx5/6flox/flox mice, we crossbreed Pgrcre/+; Dlx5/6flox/+ males with Dlx5/6flox/flox females.
For genotyping, DNA was extracted from mice tails and uteri using a KAPA express extract kit (Kapa Biosystems). Mutant mice were genotyped by PCR using allele-specific primers using TAKARA Ex Taq (Takara) or KAPA2G Fast Genotyping PCR mix (Kapa Biosystems):
– Dlx5LacZ/+ mice: 1. Z1: 5′-GCGTTACCCAACTTAATCG-3′, 2. Z2: 5′-TGTGAGCGAGTA ACAACC-3′, a PCR product of 350 bp shows LacZ allele.
– Dlx5-floxed-delta: Er: 1. 5′-TTCCATCCCTAAAGCGAAGAACTTG-3′, Lf: 2. 5′-CCTCCCAGAAATACCCCTTCTCTTG-3′. PCR product of 753 bp shows wild-type allele, 943 bp product shows floxed allele and a 244 bp product shows the recombinant Δ allele.
– Dlx6-floxed-delta: Er2: 1. 5′-CTTTAGGCGTTGGGAAAAGCCAGG-3′, Lf2: 2. 5′-CTGGTCTCAGCTCATAAGTTTCCTTC-3′. PCR product of 856 bp shows wild-type allele, 1014 bp product shows floxed allele and a 193 bp product shows the recombinant Δ allele.
– PGR-Cre mice: 1. Cre-L1: 5′-GCCACCAGCCAGCTATCAACTC-3′, 2. Cre-R1: 5′-TTGCCCCTGTTTCACTATCCAG-3′. PCR product of 250 bp shows cre-recombinase gene.
For testing fertility, a total of 75 sexually mature female mice (2 month to 1 year old) were bred with WT males (B6D2F1N) for up to 6 months, and vaginal plugs were detected. The litter size per mice during this period was recorded for each genotype.
For embryo transfer, we used the standard procedure (39). Six WT B6D2F1N female mice 5- to 6-week old were superovulated and the blastocysts were collected at E3.5 dpc by flushing the uterine horns. The same day the collected blastocysts were reimplanted into the uterus of pseudopregnant recipient females of either Pgrcre/+; Dlx5/6flox/flox or control littermates at E2.5 dpc. Females were sacrificed at E5.5, the uterine horns removed and the implants checked.
Tissue samples and treatment
Uteri were dissected in phosphate buffer saline (PBS). Pictures of whole uterus were taken, and each uterine horn was measured for width (at two sites), length and area. Uteri were fixed in 4% paraformaldehyde for 1 day at 4°C or in Bouin's fixative solution (75% saturated picric acid, 20% formol 40%, 5% acetic acid) for 7 days at 4°C. Some of them were cryoprotected in 4% paraformaldehyde with 15% sucrose for 1 day, then in sucrose 30% for 1 day and stored at −80°C before further use. Fixed uteri were embedded in paraffin, and 8 µm-thick sections were prepared. Cryoprotected uteri were embedded in OCT (Leica) and 12 µm-thick sections were prepared. Uteri sections were stained with Mallory Trichromic protocol (Groat's haematoxylin, acid fuchsin, aniline blue/orange G) (40). Some other uteri sections were stained with haematoxylin and eosin.
Immunohistochemistry was performed on deparaffinized sections using standard protocols of the Dako Envision kit or the Dako ARK kit. Anti-DLX5 (NBP1-19547; Novus Biologicals, Littleton, CO) polyclonal rabbit primary antibody was used at 1/200 dilution, diluted in PBS with 5% fetal calf serum (FCS) concentrations.
In situ hybridization
Gene expression was analysed by in situ hybridization on floating sections of 50 µm using digoxigenin (DIG)-UTP labelled antisense RNA probes against Dlx5 and Dlx6. After hybridization, probes were detected using an alkaline-phosphatase-conjugated anti-DIG antibody and BM Purple substrate (Roche).
Total RNA was isolated from uterus of 6 PgrCre/+; Dlx5/6flox/flox and 6 Dlx5/6flox/flox at PN25 using an RNeasy minikit (Qiagen) according the manufacturer instructions. On-column deoxyribonuclease (Qiagen) digestion was incorporated into an RNA isolation procedure to remove potential genomic DNA contamination. RNA concentration and the ratio of the absorbance at 260 and 280 nm were measured using a NanoDrop 2000 spectrophotometer (Thermo Scientific). Reverse transcription was carried out using 600 or 200 ng total RNA and Superscript III (Invitrogen) or Primscript (Ozyme) reverse transcriptase to obtain cDNA. Quantitative real-time PCR (qPCR) was performed using 7300 Real-Time PCR (Applied Biosystems). The PCR programme consisted of 95°C for 10 min, 40 cycles 95°C of 15 s and 60°C for 10 min. Relative gene expression was measured with Taqman gene expression assays (Life Technologies, France). All primers were Taqman pre-designed primers (Life Technologies). To measure the relative amount of PCR products, the Ct of Sdha was used as a control gene and was subtracted from the Ct of genes of interest to derive ΔCt. The ΔCt of mutant animals was compared with ΔCt of control animals and the difference was assigned as ΔΔCt. The fold change between two samples was then calculated as 2−ΔΔCt.
RNA library preparation was carried out according to the Illumina TruSeq Stranded Total RNA Sample Prep protocol, following Ribo-Zero Gold Deplete procedure (Illumina, San Diego, CA). From a total amount of 1.5 µg of total RNA, both cytoplasmic and mitochondrial ribosomal RNA (rRNA) were removed using specific biotinylated oligos and Ribo-Zero rRNA removal beads. RNA, purified from uteri of 3 PgrCre/+; Dlx5/6flox/flox and 3 Dlx5/6flox/flox mice at PN25, was fragmented by the addition of divalent cations and the incubation at 94°C for 4 min. The first strand cDNA was synthesized using random primers and a reverse transcriptase. DNA Polymerase I synthesized the second cDNA strand and generated blunt-ended ds cDNA, while RNase H removed the RNA template. An A nucleotide was added to the 3′ ends, followed by the ligation of multiple indexing adapters that are required for the pooling and the hybridization with the flow cell. These DNA products were purified and only those that have adapter molecules on both ends were enriched with PCR by using a Cocktail primer that anneals specifically to the adapters. The obtained libraries were validated by Bioanalyzer (Agilent DNA 1000, Agilent Technologies), pooled and sequenced using the Illumina HiSeq 2000 sequencing system.
RNA-seq data analysis
The global quality of each library was checked using FastQC software (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). To reduce any PCR amplification bias, choice has been done to remove duplicated reads before the mapping based on a 100% sequence identity between reads. Then, five nucleotides were trimmed at both 5′ and 3′ ends of all remnant reads using FASTX-Toolkit v0.0.14 (http://hannonlab.cshl.edu/fastx_toolkit/index.html). Reads were mapped against the mouse mm10 genome (version GRCm38.p3) using TOPHAT2 software v2.0.10 (41) with default parameters. Only uniquely mapped reads were kept and were counted on genes exons using HTSeq-count software v0.6.1p1 (42) with ‘union’ mode using the mus musculus Ensembl annotation (release 79). Only codant and expressed genes were considered. ‘R’ package DESeq2 v1.8.0 (43) was used for statistical analyses to determine differential gene expression levels (cutoff: adjusted P-value ≤ 0.1). Samples homogeneity was checked realizing a principal component analysis with ‘R’ package FactoMineR (44), following which, one mutant mouse was not considered for further analysis. We finally performed a GO term enrichment analysis using DAVID tool (45,46). To analyse predicted protein interaction in deregulated genes, we used STRING tool (47). To sort secreted protein, we use COMPARTMENTS tool (48).
The Mann–Whitney unpaired test was conducted using Prism (Graphpad Software, La Jolla, CA) to calculate the differences between groups. All values are expressed as means ± SEM of combined data from replicate experiments. Values of P < 0.05 were considered statistically significant.
B.B. is recipient of a doctoral fellowship from the IDEAL European Consortium. This research was partially supported by the EU Consortia IDEAL (HEALTH-F2-2011-259679) and HUMAN (EU-FP7-HEALTH-602757). The international cooperation involved in the study was partially supported by the GDRI PhyGHA of the French CNRS.
We are grateful to Dr D Lydon (Huston, Texas) for the kind gift of PGR-cre mice. We thank the RDDM's Bioinformatics Core Facility of MNHN for technical support and Aurélie Hagneau and Ocilia Fernandes for technical assistance. We are obliged to Yorick Gitton and Eglantine Heude for interesting discussions and constructive criticism. We also thank the team in charge of mouse animal care and in particular M Stéphane Sosinsky and M Fabien Uridat. We thank Aicha Bennana and Lanto Courcelaud for administrative assistance.
Conflict of Interest statement. None declared.