Swiss outbred mice were purchased from Taconic Biosciences. B6SJLF1/J and Tg(Sp7-tTA,tetO-EGFP/cre)1Amc (Osx-Cre) transgenic mice were obtained from the Jackson Laboratory. Tg(Sp7/mCherry)2Pmay/J (Osx-Cherry) transgenic mice were produced using a CD1 outbred stock (Strecker et al., 2013). Osx-Cre and Osx-Cherry mice were subsequently outcrossed to Swiss Webster mice (Taconic Biosciences). Amhr2 ^ΔE1-6 (Amhr2-ΔE1-6; Mishina et al., 1996), Amhr2 ^tm2Bhr (Amhr2-lacZ; Arango et al., 2008), Gt(ROSA)26Sor ^tm1(lacZ)Sor (R26R-lacZ; Soriano 1999), and Mt1-AMH (Behringer et al., 1990) mice were maintained on a predominantly C57BL/6J genetic background. All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Texas MD Anderson Cancer Center. Studies were performed consistent with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

High density BAC clone arrays on filters from the RPCI-22 129/SvEv female mouse genomic library (BACPAC Genomics, Emeryville, CA) were screened with a³²P-labelled probe located ∼7 kb 5′ of Amhr2 exon 1, using a Megaprime DNA Labelling System kit (Amersham) (Mishina et al., 1996). Six positive clones were identified using the manufacturer’s grid key. Purified BAC clone DNA was digested with NotI, subjected to pulse-field agarose gel electrophoresis, and analyzed by Southern blot, using the same probe used in the initial screen. An additional probe within intron 6 of the Amhr2 locus was used (Mishina et al., 1996). This analysis suggested that all six BAC clones contained the Amhr2 locus. Sequencing of the ends of BAC clone RPCI-22 315D12 delineated the genomic region contained in the clone (NCBI37/mm9: Chr 15, 102,401,136–102,548,427). RPCI-22 315D12 was linearized with I-SceI. The linearized DNA (0.2 ng/ul) was injected into the pronuclei of B6SJLF2/J hybrid zygotes and then transferred into the oviducts of pseudopregnant Swiss surrogate mothers.

Osx-Cre, Osx-Cherry, Amhr2 ^ΔE1-6 , Amhr2-lacZ, R26R-lacZ were genotyped as previously described (Mishina et al., 1996; Soriano 1999; Rodda and Mcmahon, 2006; Arango et al., 2008; Strecker et al., 2013). Mt1-AMH transgenic mice were genotyped by PCR using the following primers: hMIS-Fw1: 5′ CCC TAG TGC TGT CTG CCC T 3′ and hMIS-Rv2: 5′ GGA GCT GCT GCC ATT GCT G, resulting in a 176 bp amplified DNA fragment. PCR genotyping for the Amhr2 BAC transgene was performed using the following pBACe3.6 vector primers: BAC-F: 5′ GTG ATA TCG CGG AAG GAA AA 3′ and BAC-R: 5′ AGG ATA TAC GGC AGG CAT TG 3’, resulting in a 499 bp amplified DNA fragment.

Embryos were stained for lacZ expression as described (Arango et al., 2008).

Ten µm sections were cut from paraffin embedded lacZ-stained tissues and counterstained with 0.1% Nuclear Fast Red. 10 µm frozen sections were cut from OCT embedded Osx-Cherry embryos and stained with DAPI (4′,6-diamidino-2-phenylindole).

Embryos were dissected in phosphate-buffered saline and visualized using a Leica MZ10F fluorescent dissecting microscope for Cherry fluorescence. Images were captured using a JENOPTIK GRYPHAX camera. Frozen sections of Osx-Cherry embryos were visualized for Cherry and DAPI fluorescence using an A1 Nikon confocal microscope.

The lack of cell type-specific regulatory sequences in the 500-bp 5′ region of Amhr2 (Kimura et al., 2017) motivated us to screen a larger region surrounding the locus. BAC RPCI-22 315D12 has a ∼147 kb (147,291 bp) region of mouse chromosome 15 from a 129S6/SvEvTac inbred mouse, containing the Amhr2 locus (Figure 1). There is ∼35 kb of sequence 5′ and ∼94 kb 3′ of the Amhr2 locus. The BAC clone contains 7 other genes (Figure 1). We hypothesized that Amhr2 Müllerian duct mesenchyme-specific regulatory sequences were located within this BAC clone. To test this idea, we generated transgenic mice with this unmanipulated Amhr2-containing BAC clone. Two independent transgenic mouse lines (Tg ^BAC-Amhr2 ) were established.

We next bred male mice from the two Amhr2 BAC transgenic lines to female mice heterozygous for an Amhr2-lacZ allele that is also a null allele (Arango et al., 2008). The Amhr2-lacZ allele encodes a convenient visual marker of the uterus and oviduct because lacZ is expressed in the myometrium and myosalpinx, respectively (Arango et al., 2008). From the progeny of that cross, Tg ^BAC-Amhr2 ; Amhr2 ^lacZ/+ male mice were identified and then bred to female mice homozygous for an Amhr2 deletion allele that is also a null allele (ΔE1-6; Mishina et al., 1996) to generate Tg ^BAC-Amhr2 transgenic males on an Amhr2 compound heterozygous null (ΔE1-6/lacZ) genetic background. Amhr2 ^ΔE1-6/lacZ males served as controls for persistent Müllerian duct derivatives, exploiting the lacZ allele to label Müllerian duct derivatives, i. e. uterus and oviducts (Mishina et al., 1996; Arango et al., 2008). At postnatal day 0 (P0) there was complete regression of the Müllerian system in control males (Figure 2A). Positive lacZ staining was detected in the persistent uterus/oviduct of an Amhr2 ^lacZ/ΔE1-6 male (Figure 2B). We found that Tg ^BAC-Amhr2 ; Amhr2 ^ΔE1-6/lacZ males were negative for lacZ staining because they lacked a uterus and oviduct (Figure 2C). The complete rescue of the Amhr2 mutant phenotype was observed for both Amhr2 BAC transgenic lines (Line 1, n = 2; Line 2, n = 2). This suggests that the ∼147 kb Amhr2-containing BAC clone contains all of the required regulatory regions for Müllerian duct mesenchyme expression to mediate AMH signaling for regression.

We next used a different BAC transgenic mouse approach to screen for Müllerian duct mesenchyme-specific regulatory regions for Osx. BAC RP23-399N14 has a ∼204 kb (204,096 bp) portion of mouse chromosome 15 from a female C57BL/6J mouse containing the Osx locus (Figure 1). There is ∼145 kb of sequence 5′ and ∼50 kb 3′ of the Osx locus. The BAC clone contains 10 other genes (Figure 1). Like our strategy to identify a Müllerian duct mesenchyme-specific regulatory regions for Amhr2, we tested the idea that an Osx Müllerian duct mesenchyme-specific regulatory sequences were located within BAC clone RP23-399N14. Transgenic mice carrying this Osx-containing BAC have been previously generated (Rodda and Mcmahon, 2006). The Osx-containing BAC was previously modified by recombineering to insert a Tet-off regulatable GFP:Cre fusion protein under the control of the Osx promoter (Osx-Cre). In the absence of doxycycline, Cre expressed from the Osx locus mediates recombination of floxed alleles in mice. The Osx-Cre BAC transgene was found to be active in the osteoblast lineage throughout embryonic and early postnatal development (Rodda and Mcmahon, 2006).

We used the R26R-lacZ Cre reporter mouse (Soriano 1999) to screen for Osx-Cre expression in the developing Müllerian ducts. Timed matings were established and E13.5 to 16.5 embryos were dissected and stained with X-gal. X-gal staining was detected initially at E14.5 in the developing Müllerian ducts of male embryos (n = 6) (Figure 3). Although Osx is expressed throughout the male Müllerian duct mesenchyme (Mullen et al., 2018), we observed strong lacZ expression in the anterior Müllerian duct but reduced lacZ expression in the posterior Müllerian duct (Figure 3B). X-gal staining was also observed at E15.5 in the male Müllerian duct, in the region adjacent to the testes (n = 4) (data not shown). A very small population of lacZ-expressing cells was observed in the Müllerian ducts of female embryos (Figures 3A,C). Histological sections demonstrated that X-gal staining in male embryos was restricted to the Müllerian duct mesenchyme (Figure 3D). This suggests that the ∼204 kb Osx-containing BAC contains regulatory information for male-specific Müllerian duct mesenchyme expression.

Transgenic mice have previously been generated using ∼39 kb of genomic sequence containing the Osx gene derived from BAC clone RP24-362M3 (C57BL/6J) (Strecker et al., 2013). This ∼39 kb region excluded other gene coding sequences and encompassed ∼20% of the BAC cloned used to generate Osx-Cre transgenic mice. A Cherry Fluorescent Protein reporter was inserted just upstream of the second translational start site of Osx. Cherry fluorescence was detected in developing and adult skeletal tissues (Strecker et al., 2013). To determine if the ∼39 kb region also contained Müllerian duct mesenchyme regulatory sequences, we examined E14.5 male and female embryos for Cherry fluorescence in the Müllerian ducts (Figure 4). At E14.5, we initially identified Osx-Cherry transgenic embryos because they express Cherry fluorescence in the developing skeleton. Male and female Osx-Cherry transgenic embryos were identified by brightfield assessment of the gonads. Although Cherry fluorescence was observed in the developing skeleton tissues of female Osx-Cherry transgenic embryos, no Cherry fluorescence was detected in the Müllerian ducts (n = 4) (Figures 4A,C). Likewise, Cherry fluorescence was detected in the developing skeleton tissues of male Osx-Cherry transgenic embryos (Figure 4B). In addition, Cherry fluorescence was detected in the Müllerian ducts (n = 6) (Figures 4B,D). Histological sections demonstrated that Cherry fluorescence was present in the male but not female Müllerian duct mesenchyme (Figures 4E,F). We also examined AMH-responsive organs, including testes, ovaries and uteri of adult Osx-Cherry transgenic and control mice but no Cherry fluorescence was detected compared to controls (Supplemental Figure S1). These results suggest that the ∼39 kb genomic region included in the Osx-Cherry transgene contains a male-specific, Müllerian duct mesenchyme regulatory sequences that direct Osx transcription.

Previously, we showed that an Osx-lacZ knock-in allele was expressed in male but not female Müllerian duct mesenchyme, reflecting the sex-specific pattern of Osx expression in the Müllerian ducts (Mullen et al., 2018). We also showed that ectopic expression of human AMH from an Mt1-AMH transgene could activate the Osx-lacZ knock-in allele in female embryos, indicating that Osx is an AMH-induced gene (Mullen et al., 2018). To determine if the Osx-Cherry transgene could also respond to ectopic human AMH in female embryos, we crossed Osx-Cherry females to Mt1-AMH males to generate E14.5 Mt1-AMH; Osx-Cherry double transgenic female embryos (Figure 5). We observed Cherry fluorescence in the Müllerian ducts of the Mt1-AMH; Osx-Cherry double transgenic female embryos but not in Osx-Cherry transgenic female embryos (n = 5) (Figures 5A,B). The Cherry fluorescence in the Müllerian ducts of the Mt1-AMH; Osx-Cherry double transgenic female embryos was localized to the mesenchyme (Figure 5D). These results demonstrate that the 39 kb Osx region contains a Müllerian duct mesenchyme regulatory sequences that are responsive to AMH signaling.

We found that transgenic mice generated with a mouse BAC clone containing the entire Amhr2 locus rescued the persistent Müllerian duct-derived tissues found in Amhr2-null males (Mishina et al., 1996; Arango et al., 2008). This suggests that the BAC clone contains regulatory sequences sufficient for Müllerian duct mesenchyme expression. It is also possible that Amhr2 expression from the BAC clone was expressed ubiquitously. However, this could lead to engagement of Type I receptors for AMH signaling in ectopic tissues that might lead to mutant phenotypes. The Amhr2 BAC transgenic mice in both lines were normal and fertile.

We also tested the 6.2 kb region 5′ of Amhr2 including the endogenous promoter for tissue-specific activity using a lacZ reporter. However, in 9 independent lines of transgenic mice there was no lacZ expression in E14.5 Müllerian ducts of both sexes, adult testes, ovary and uterus (unpublished observations). Thus, the immediate upstream region of Amhr2 is not sufficient for cell type-specific transcription (Kimura et al., 2017). While our current results are a step forwards, the BAC clone is ∼147 kb in length, creating a challenge to localize the Müllerian duct mesenchyme regulatory sequence to a discreet location.

Amhr2 transcription in the Müllerian duct mesenchyme is induced by WNT7a secreted by the Müllerian duct epithelium (Parr and McMahon, 1998). Wnt7a-null males do not express Amhr2 in the Müllerian duct mesenchyme, leading to a block in regression, resulting in the formation of a uterus (Parr and McMahon, 1998). Thus, WNT signaling lies upstream of Amhr2 transcription in the GRN (Moses and Behringer, 2019). In addition, beta-catenin is required in the ductal mesenchyme for Müllerian duct regression during male differentiation (Kobayashi et al., 2011). These observations suggest that there may be cis elements related to WNT signaling that are required for Amhr2 transcription.

Amhr2 is also expressed in Sertoli, granulosa cells and smooth muscle of the uterine myometrium (Arango et al., 2008). Amhr2 expression in these cell types have been shown to be regulated by multiple factors. A binding site for NR5A1 (also known as SF-1) located within ∼300 bp upstream of the transcription start site of the human AMHR2 gene was shown to be required for reporter expression in the NT2/D1 teratocarcinoma cell line (De Santa Barbara et al., 1998). Knockdown of Dmrt1 in male chick embryos caused a down-regulation of Amhr2 (Cutting et al., 2014). BMP4 and BMP15 were shown to enhance AMHR2 transcript levels in human and sheep ovarian granulosa cells in vitro (Pierre et al., 2016). This response to BMP signaling was mediated by a ∼2.2 kb region 5′ of the AMHR2 promoter (Pierre et al., 2016). A long non-coding (lnc) RNA, lnc-Amhr2 has been identified that enhanced Amhr2 promoter activity in a mouse ovarian granulosa cell line, OV3121 (Kimura et al., 2017). The factors that regulate Amhr2 transcription in Müllerian duct mesenchyme cells remain to be discovered.

The BAC used to generate the Osx-Cre line spans 204 kb of chromosome 15, from the 3′ end of Znf40 to the 5′ end of Sp1. Using the Osx-Cre BAC transgenic mouse line, we discovered male-specific Cre reporter gene expression in the Müllerian duct mesenchyme. Although the endogenous Osx gene is expressed throughout the entire male Müllerian duct mesenchyme (Mullen et al., 2018), we found that Osx-Cre directed reporter gene expression predominantly in the anterior Müllerian duct. It is possible that there are separate cis elements that direct anterior and posterior expression. However, these findings may be unique to this one transgenic mouse line.

The Osx-Cherry construct encompasses 39 kb within the 204 kb Osx-Cre BAC region, spanning Osx and intergenic sequences on both the 5′ and 3’ ends of the gene. Using the Osx-Cherry transgenic mouse line, we found male-specific Cherry fluorescence along the entire length of the Müllerian ducts, indicating that the 39 kb region contains a Müllerian duct mesenchyme-specific regulatory sequences. AMH signaling is necessary and sufficient for Osx expression in the Müllerian duct mesenchyme (Mullen et al., 2018). We found that ectopic AMH provided by an Mt1-AMH transgene could induce Osx-Cherry expression in the Müllerian duct mesenchyme of female embryos. This suggests that the 39 kb Osx region contains sequences that respond to AMH signaling. The sequences that direct Müllerian duct mesenchyme-specific transcription and those that mediate AMH signaling may be the same but await further study.

In the mouse genome, Osx and Amhr2 are linked, residing on chromosome 15 within 100 kb of each other. There is one gene, namely Sp1, located between Osx and Amhr2. Sp1 encodes a transcription factor that is expressed in most tissues (O’connor et al., 2016). Interestingly, the 39 kb Osx-Cherry region (NCBI37/mm9: Chr 15, 102,182,823–102,220,102) does not overlap with the Amhr2 BAC clone (NCBI37/mm9: Chr 15, 102,401,136–102,548,427). Therefore, the Müllerian duct mesenchyme regulatory region we identified for Osx-Cherry is distinct from the one identified within the Amhr2 BAC. The identification of two distinct Müllerian duct mesenchyme regulatory regions, one for Amhr2 and one for Osx, seems reasonable because Amhr2 is expressed in the mesenchyme of both male and female Müllerian ducts that is dependent on epithelial-derived WNT7A, whereas Osx expression is restricted to male Müllerian ducts that depends on Amhr2 and AMH secreted by the fetal testes (Baarends et al., 1994; Parr and McMahon, 1998; Arango et al., 2008; Mullen et al., 2018). Thus, Amhr2 is upstream of Osx in the GRN. Analysis of the chromatin landscape of E14.5 male and female Müllerian duct mesenchyme could point to Müllerian duct-specific regulatory elements. However, these types of studies have not been reported perhaps because there are very few cells per embryo and single cell approaches still require cell numbers greater than practically available. The identification of genomic regions that contain regulatory sequences for Müllerian duct mesenchyme expression contributes to the definition of this GRN of sexual development (Moses and Behringer, 2019).