Sox9 in mouse urogenital sinus epithelium mediates elongation of prostatic buds and expression of genes involved in epithelial cell migration

Abstract

Previous studies identified Sox9 as a critical mediator of prostate development but the precise stage when Sox9 acts had not been determined. A genetic approach was used to delete Sox9 from mouse urogenital sinus epithelium (UGE) prior to prostate specification. All prostatic bud types (anterior, dorsolateral and ventral) were stunted in Sox9 conditional knockouts (cKOs) even though the number of prostatic buds did not differ from that of controls. We concluded that Sox9 is required for prostatic bud elongation and compared control male, control female, Sox9 cKO male and Sox9 cKO female UGE transcriptomes to identify potential molecular mediators. We identified 702 sex-dependent and 95 Sox9-dependent genes. Thirty-one genes were expressed in both a sex- and Sox9-dependent pattern. A comparison of Sox9 cKO female vs control female UGE transcriptomes revealed 74 Sox9-dependent genes, some of which also function in cell migration. SOX9 regulates, directly or indirectly, a largely different profile of genes in male and female UGE. Eighty-three percent of Sox9-dependent genes in male UGE were not Sox9-dependent in female UGE. Only 16 genes were Sox9-dependent in the UGE of both sexes and seven had cell migration functions. These results support the notion that Sox9 promotes cell migration activities needed for prostate ductal elongation.

1. Introduction

The prostate originates from the urogenital sinus (UGS) in three sequential stages of development (specification, initiation and elongation). Prostate development initiates when fetal androgens bind androgen receptors in UGS mesenchyme (UGM). The activated androgen receptors stimulate release of paracrine signals, including fibroblast growth factors (FGFs), which act on UGS epithelium (UGE) to promote bud formation (Thomson and Cunha, 1999; Prins and Putz, 2008). Prostatic buds are specified across three UGE surfaces: anterior, dorsolateral and ventral. Prostatic buds are initiated as small protrusions extending from the UGE surface (Lin et al., 2003; Vezina et al., 2008). Prostatic buds elongate by invading the UGM, a process that continues postnatally, when buds also undergo extensive branching and canalization to form ductal networks of the anterior, dorsolateral and ventral prostate lobes (Sugimara et al., 1986).

Sry-box 9 (Sox9) is involved in a variety of developmental processes (Pritchett et al., 2011) and encodes a transcription factor essential for prostatic bud formation (Huang et al., 2012; Thomsen et al., 2008). Human and mouse prostatic SOX9 expression are predominantly epithelial (Huang et al., 2012; Wang et al., 2008) and are robust in buds during initiation and elongation (Huang et al., 2012; Thomsen et al., 2008; Wang et al., 2008). The best-characterized Sox9 regulated genes in cartilage and bone are extracellular matrix (ECM) genes, which participate in chondrogenesis (Akiyama, 2008; Oh et al., 2014; Ohba et al., 2015). Outside of cartilage and bone, however, the battery of Sox9 regulated genes varies between organs (Garside et al., 2015). Sox9 dependent genes have not until now been identified in vivo in the mouse UGE.

The etiology/progression of prostate cancer and benign prostate hyperplasia involve, to some degree, reactivation of signaling pathways, homologous in mouse and human, that direct prostate development (Schaeffer et al., 2008; Schrecengost and Knudsen, 2013; Cunha and Ricke, 2011; Cunha et al., 2018). Sox9 is essential for mouse prostate development and promotes prostate carcinogenesis in mouse models (Thomsen et al., 2008; 2010; Huang et al., 2012). Sox9 is expressed during human fetal prostate development and is associated with multiple measures of human prostate cancer (Wang et al., 2008; Schaeffer et al., 2008; Zhong et al., 2012; Qin et al., 2014). Therefore, understanding functional roles of Sox9 in prostatic bud formation may lead to new strategies or targets for treating prostate cancer and benign prostate hyperplasia.

Two previous studies deleted Sox9 in the mouse UGS and surprisingly, the outcomes were not the same. Thomsen et al. (2008) used an Nkx3-1^cre to delete Sox9 in the UGE and found that ventral prostatic bud formation was inhibited while anterior and dorsolateral bud formation was spared. Huang et al. (2012) used a ROSA26-ER^cre to delete Sox9 in UGE and UGM and found that ventral, anterior and dorslolateral bud formation was impaired. While both studies showed Sox9 plays a role in prostatic budding, the precise function of Sox9 remained uncertain.

The present study used a Shh^cre driver, deleting Sox9 from UGE only, to test five major hypotheses. Conditional Sox9 deletion from mouse UGE, prior to the start of bud formation: (1) disrupts one or more stages (specification, initiation and/or elongation) of bud formation, (2) disrupts formation of one or more prostatic bud types, (3) disrupts expression of Sox9-dependent genes in both male and female UGE, (4) disrupts one or more Sox9-dependent cellular functions in UGE necessary for bud formation, and (5) identifies genes and cellular functions important for budding by comparing male control and Sox9 cKO UGE transcriptomes to each other and to female control and Sox9 cKO UGE transcriptomes.

We found that Sox9 deletion from the male UGE does not affect bud specification or initiation, but instead impairs bud elongation culminating in “stunted buds”. This phenotype is manifested in all prostatic bud types (anterior, dorsolateral and ventral). We found that Sox9 dependent genes differ in the male and female UGS and identified candidate genes, including those involved in cell migration, which are likely to mediate Sox9-dependent prostatic bud elongation.

2. Results

2.1. SOX9 expression in lower urinary tract and Sox9 knockout in UGE prior to bud initiation

The lower urinary tract of male mouse fetuses was stained at multiple developmental stages to characterize the temporal expression pattern of SOX9. At E14.5, the earliest stage assessed, no SOX9 expression was observed (Supplementary Fig. 1). By E15.0, SOX9 was detected throughout the epithelium of the bladder, UGS, Wolffian ducts and urethra, and was also detected in mesenchymal cells largely located in close proximity to the UGE (Fig. 1A, left and right panels).

Fig. 1.

Expression of SOX9 in lower urogenital tract and prostatic buds of the male mouse fetus. (A, left panel) IHC showing SOX9 expression in a sagittal section of the lower urogenital tract at E15.0. Brown staining is seen throughout epithelium of the bladder (BL), Wolffian ducts (WD), UGS and urethra (UR), and in patches of UGS mesenchyme [40x]. (A, right panel) SOX9 expression in the UGS is visualized at higher magnification [100x]. (B, left panel) SOX9 expression is shown for two anterior buds in a transverse section of UGS at E18.5 [40x] and (B, right panel) in a single anterior bud (AB) at E18.5 [200x]. Nuclei were counterstained blue with hematoxylin. Other abbreviations: anterior (A), dorsal (D) and ventral (V) regions of the UGS. Images are representative of n=3 litter independent fetuses, where each fetus came from a different litter.

At E18.5, after prostatic buds have initiated and are elongating, SOX9 expression was observed most prominently in the UGE throughout the entire length of prostatic buds (Fig. 1B, left and right panels). This is shown at low magnification for two anterior buds (AB) in a transverse section of the UGS (left) and at high magnification for one of these ABs (right). Careful examination of SOX9 immunostaining in the single AB (right) shows SOX9 is expressed, not only in epithelial cells throughout the bud, but also in some mesenchymal cells in close proximity to the bud surface.

Previous studies have shown that Sox9 plays an important role in prostate budding/development, but the precise nature of its role is uncertain. To better define its role, we used Shh^cre to delete Sox9 from the UGE and tested the hypothesis that prostatic bud formation in vivo requires Sox9 in the UGE, prior to and during, bud specification, initiation and elongation.

ShhcreERT2/C activity was detected in the entire lower urinary tract epithelium of E9-13 mice (Seifert et al., 2009). This is before prostatic bud formation is initiated. Therefore, we used a Shh^cre/+ mouse to drive cre expression in UGE at a slightly later stage of development. Shh^cre/+ mice were mated to ROSA26 reporter mice and UGSs from E14.5 fetuses were harvested and stained for β–galactosidase activity using Bluo-gal as the chromogen. Blue staining showed that cre was expressed in bladder and urethral epithelium at E14.5 (Fig. 2A).

Fig. 2.

Conditional knockout of Sox9 in UGE and confirmation of loss of SOX9 expression in lower urinary tract epithelium. Conditional knockout of Sox9 (cKO Sox9) in the UGE was mediated by Shh driven cre recombinase expression. (A) Lower urinary tract of an E14.5 cre Reporter male fetus stained with Bluo-gal. Blue staining indirectly shows cre recombinase expression restricted to epithelium of the bladder, UGS, and urethra (20x). (B and C) IHC showing SOX9 expression (brown) in sagittal sections of the lower urogenital tract and testis (inset, bottom right) from a representative control (B) and Sox9 cKO (C) fetus at E16.5 (40x magnification; nuclei counterstained blue with hematoxylin). The UGS is identified by a red, rectangular box. In the control fetus (B), SOX9 expression is seen throughout epithelium of the lower urogenital tract. In the Sox9 cKO fetus (C) SOX 9 expression was completely absent from epithelium of the bladder (BL), UGS, and urethra (UR) prior to prostatic budding. Specificity of the conditional Sox9 knockout is demonstrated by normal SOX9 expression in Wolffian ducts (WD), mesenchymal patch (arrow), and Sertoli cells of the testis in Sox9 cKO fetuses. All images are representative of n = 3 litter-independent male fetuses.

To confirm SOX9 was deleted prior to initiation of prostatic budding, UGSs from E16.5 control (Shh^+/+; Sox9^fl/+), and Sox9 cKO (Shh^Cre/+; Sox9^fl/fl) fetuses (see Table 1) were sectioned and stained by IHC for SOX9. E16.5 is immediately prior to prostatic budding initiation. UGSs from control mice showed SOX9 expression throughout the UGE which tended to increase across the UGE from luminal to mesenchymal surface (Fig. 2B). Some peri-UGE mesenchymal cells of control mice also showed SOX9 expression. By comparison, no SOX9 was observed in lower urinary tract epithelium of Sox9 cKO mice (Fig. 2C). Importantly, SOX9 expression was seen in Wolffian duct epithelium, peri-UGE mesenchymal cells (arrow), and Sertoli cells in the testis (inset) of Sox9 cKO mice (Fig. 2C). These internal positive controls establish that Sox9 deletion was complete and specific to lower urinary tract epithelium.

Table 1.

Study design.

Mating Scheme	Offspring UGSa
Dam x Sire	Genotype 1	Genotype 2	Group	Experimental Useb (day)
ROSA26 Reporter x Shhcre/+	ROSA26-βgal +	Shhcre/+	Cre Reporter	β-Galactosidase Staining (E14.5)
Shh+/+; Sox9fl/fl x Shhcre/+; Sox9fl/+	Shh+/+	Sox9fl/+	Control	IHC (E16.5) WM IHC (E18.5-P0.5) Array (E16.75) qRT-PCR (E16.75)
	Shh+/+	Sox9fl/fl	Control	WM IHC (E18.5-P0.5)
	Shhcre/+	Sox9fl/+	Sox9 Het	WM IHC (E18.5-P0.5)
	Shhcre/+	Sox9fl/fl	Sox9 cKO	IHC (E16.5) WM IHC (E18.5-P0.5) Array (E16.75) qRT-PCR (E16.75)

^aUGSs collected from male fetuses at E14.5, 16.5, 16.75 or 18.5-P0.5 and from female fetuses at E16.75 were used in the present study.

^bAbbreviations: immunohistochemistry (IHC), whole mount immunohistochemistry (WM IHC), microarray (Array), and quantitative reverse transcription polymerase chain reaction (qRT-PCR). Embryonic day (E) or postnatal day (P) of UGS tissue harvest.

2.2. Sox9 knockout in male UGE inhibits elongation of all prostatic bud types

IHC staining, using CDH1 antibody to mark the epithelium, was performed on whole UGSs from control and Sox9 cKO male fetuses and neonates between E18.5 and P0.5 of development after which prostatic buds were counted. Total bud numbers did not differ between control (36.1 ± 3.9, n=9) and Sox9 cKO (32.3 ± 8.3, n=3) UGSs. The number of anterior buds (ABs, 3.2 ± 0.5 vs 3.7 ± 0.3), dorsolateral buds (DLBs, 27.9 ± 3.2 vs 26.7 ± 7.8) and ventral buds (VBs, 5.4 ± 0.8 vs 2.0 ± 1.2) also did not differ between control and Sox9 cKO UGSs. We conclude from these results that Sox9 in UGE is not required for prostatic bud initiation.

We next investigated whether Sox9 in UGE is required for bud elongation by assessing the length of prostatic buds. Sox9 cKO prostatic buds were noticeably shorter than controls. More specifically, all bud types (ABs, DLBs and VBs) in 100% of Sox9 cKO UGSs were shorter than in littermate controls.

Fig. 3 shows representative examples of control (A) and Sox9 cKO (B) male UGSs from P0.5 littermates. For each UGS, the right panel is an enlargement of the boxed area in the left panel. All bud types in the Sox9 cKO UGS failed to elongate and this stunted buds phenotype is most easily observed (B, right) for anterior buds (AB, pseudocolored blue) and ventral buds (VB, pseudocolored green). Note one AB in the control UGS (A, right) has already begun to bifurcate/branch.

Fig. 3. Conditional knockout of Sox9 in the male mouse UGE prevents prostatic bud elongation.

WM IHC for cadherin-1 (CDH1) stained epithelium in control male UGS (A) and Sox9 cKO male UGS (B). Sox9 is knocked out specifically in the UGE, before prostatic budding initiation begins. The representative control UGS (A) and Sox9 cKO UGS (B) were harvested at P0.5 from mouse neonates from the same litter. For each UGS, the right panel is an enlargement of the UGS (boxed area) in the left panel (20x). All bud types in the Sox9 cKO UGS failed to elongate. From this view (right panels), failed elongation of buds in the Sox9 cKO UGS (B) is most easily observed for anterior buds (AB, pseudocolored blue) and ventral buds (VB, pseudocolored green). Abbreviations: bladder and bladder neck (BL), seminal vesicle (SV), urethra (UR), dorsolateral buds (DLB) and urogenital sinus (UGS).

It is not known if the shorter prostatic buds in Sox9 cKO UGSs are permanently stunted or if elongation is severely delayed, because Sox9 cKO fetuses die shortly after birth, likely in part, to impaired lung development (Rockich et al., 2013). However, when comparing lengths of ABs, DLBs and VBs in UGSs from Sox9 cKO male fetuses at E18.5 to lengths of the same bud types at P0.5 there were no noticeable differences in lengths. This suggests that all prostatic bud types in Sox9 cKO male fetuses are permanently stunted.

UGSs of control and Sox9 hets did not differ in bud count (Supplementary Table 10) or visual appearance of bud length. This shows that one copy of Sox9 is sufficient to promote prostatic bud elongation.

2.3. GO analysis of microarray data from control and Sox9 cKO male UGE shows cell migration is impaired

To elucidate the molecular mechanism of Sox9 in the prostatic budding process, total RNA was isolated from UGE of E16.75 control male and Sox9 cKO male fetuses for microarray analysis. E16.75 was chosen because this is just prior to prostatic bud initiation, when control and Sox9 cKO male UGSs are still morphologically identical. Compared to control male UGE, 95 annotated transcripts were differentially regulated at least 1.3 fold (p ≤ 0.05) in male Sox9 cKO UGE: 63 downregulated and 32 upregulated (Supplementary Table 1). Additionally, 149 (61%) unannotated transcripts were significantly differentially regulated at least 1.3 fold: 71 downregulated and 78 upregulated (Supplementary Table 1).

GO analysis of these differentially regulated, Sox9-dependent genes in the male UGE identified 10 pathways or cell functions for which there was significant gene enrichment (Z score < −2.0 or ≥ ±2.0). Eight of the pathways/functions were related to cell migration (Table 2). The Z score for each of the 10 pathways/functions was negative, indicating downregulated or deficient activity.

Table 2.

Gene ontology analysis of male Sox9 cKO UGE transcriptome predicts decreased UGE cell migration.

GO Terms	Z Scorea	# Genes	Downregulated Genes	Upregulated Genes
Cell Movement	− 2.1	23	ALOX15, BARX2, CCND1, CDC42EP1, CXCL14, CXCR4, DPYSL3, IGFBP4, ITGB8, KCNN4, LRIG1, mir-23a, NOXA1, PLA2G7, SHH, SOX9, TGFBI, THBS1, TNS4	HDC, LEF1, mir-lOb, PTGDR
Migration of Cells	− 2.4	22	ALOX15, BARX2, CCND1, CDC42EP1, CXCL14, CXCR4, DPYSL3, IGFBP4, ITGB8, NOXA1, PLA2G7, SHH, SOX9, TGFBI, THBS1, KCNN4, mir23a, TNS4	HDC, PTGDR, LEF1, mir-10b
Chemotaxis of Cells	− 2.6	8	CCND1, CXCL14, CXCR4, KCNN4, PLA2G7, THBS1	LEF1, PTGDR
Cell movement of Phagocytes	− 2.0	7	CCND1, CXCL14, CXCR4, PLA2G7, THBS1	mir-10b, PTGDR
Chemotaxis of Phagocytes	− 2.4	6	CCND1, CXCL14, CXCR4, PLA2G7, THBS1	PTGDR
Chemotaxis of Myeloid cells	− 2.2	5	CCND1, CXCL14, CXCR4, PLA2G7, THBS1
Chemotaxis Antigen Presenting Cells	− 2.2	5	CCND1, CXCL14, CXCR4, THBS1	PTGDR
Adhesion of Tumor Cell Lines	− 2.1	5	ALOX15, CCND1, CXCR4, THBS1	LEF1
Glucose Metabolism Disorder	− 2.2	21	ADAMTS9, ALOX15, CAR13, CAR2, CCND1, COL14A1, CXCL14, CXCR4, DPYSL3, LRIG1, mir-23, Naip7, SLC27A1, TGFBI, HBS1	FLRT3, HIF1A, PNLIPRP1, PTGER3, RET, TACR3
Fibrogenesis	− 2.2	6	ALOX15, CCND1, DPYSL3, TGFBI, TNS4	EPAS1

^aZ score ≤ − 2.0 is significantly decreased.

2.4. Comparison of control male and female UGE transcriptomes

To identify sex-dependent genes that may regulate prostatic bud development, it is essential to compare transcriptomes at E16.75 of control male UGE (which forms prostatic buds) to control female UGE (which does not). This is the same time in development when the Sox9 cKO male transcriptome was compared to the control male transcriptome to identify Sox9-dependent genes in the male UGE.

In this comparison of UGE transcriptomes between sexes, upregulation indicates RNA was more abundant in the male UGE, while downregulation indicates it was less abundant compared to female. In control male UGE, compared to control female UGE, 702 annotated transcripts were significantly differentially regulated (p ≤ 0.05) at least 1.3 fold: 352 upregulated (RNA more abundant in male) and 350 downregulated (RNA less abundant in male) (Supplementary Table 2). Additionally, 519 (43%) unannotated transcripts were significantly differentially regulated at least 1.3 fold: 221 upregulated and 298 downregulated (p ≤ 0.05).

2.5. Discovering gene candidates in the UGE for regulating prostatic bud development

To identify UGE candidate genes for the regulation of prostatic bud development, gene expression in the UGE at E16.75 was compared between Sox9 cKO male (which forms stunted prostatic buds) and control male UGE (which forms elongated buds). This comparison revealed 95 genes that where significantly up or downregulated by Sox9 cKO in the male UGE (Supplementary Table 1). These genes are referred to as “Male Sox9-Dependent”. In addition, since there is a known sex difference in prostatic bud formation (UGE of the control male undergoes extensive bud formation while UGE of the control female does not) the 702 genes significantly up and downregulated in the male vs female UGE are referred to as “Sex-Dependent” (Supplementary Table 2).

Importantly, among the 95 “Male Sox9-Dependent” genes (Supplementary Table 1) and 702 “Sex-Dependent” genes (Supplementary Table 2), only 31 genes have expression that is both “Male Sox9-Dependent” and “Sex-Dependent” (union of Venn diagram in Fig. 4). Of these 31, 10 genes have expression in the UGE that is either increased in both a “Male Sox9-Dependent” and “Sex-Dependent” manner or decreased in both a “Male Sox9-Dependent” and “Sex-Dependent” fashion. Due to this expression pattern, these 10 genes are not considered “candidates” for regulating prostatic bud formation and they are listed in Supplementary Table 3.

Fig. 4. Expression of 31 genes in male mouse UGE is both Sox9-dependent and sex-dependent.

Union (light tan) of the Venn diagram shows 31 genes in the male UGE at E16.75 that have Sox9-dependent expression (tan, control male vs Sox9 cKO male) in common with sex-dependent expression (white, control male vs control female). Also 18 of these 31 genes are candidates for promoting bud formation and 3 are candidates for inhibiting bud formation (Table 3). Results for males are based on n = 6 litter independent, male UGEs for both control and Sox9 cKO groups, respectively, and results for females are based on n = 5 litter independent, female UGEs for both control and Sox9 cKO groups.

On the other hand, the remaining 21 genes are “candidates” for regulating prostatic bud formation and they are given in (Table 3). 18 of these genes may “positively regulate” bud formation. They are: [1] downregulated in Sox9 cKO male UGE (buds stunted) vs control male UGE (buds form) and [2] upregulated in control male UGE (buds form) vs control female UGE (buds do not form). These 18 genes include Sox9, a known “positive regulator” of bud formation and 17 other “candidate positive regulators” or promoters of prostatic bud formation. In addition, 3 of these genes may “negatively regulate” bud formation. They are: [1] upregulated in Sox9 cKO male UGE (buds stunted) vs control male UGE (buds form) and [2] downregulated in control male UGE (buds form) vs control female UGE (buds do not form). These 3 genes are “candidate negative regulators”, genes transcriptionally repressed by Sox9 that would otherwise inhibit bud formation.

Table 3.

Putative regulators of bud development.

ARRAY EXPERIMENTAL DESIGN		sox9 CKO MALE	vs	CONTROL MALE	CONTROL MALE	vs	CONTROL FEMALE
(Prostate Bud Development)		(Stunted)		(Normal)	(Normal)		(None)
MGIa Gene ID	Gene Symbol	Male SOX9 Dependent Fold Change			Sex Dependent Fold Change
Putative Promoters
99538	Acsm3	− 1.9			1.5
87997	Alox15	−1.5			1.5
109563	Cxcr4	− 3.1			2.8
88594	Cyp27a1	−1.7			1.9
5454178	Gm24401	−1.6			1.4
5455666	Gm25889	− 1.4			1.3
109615	Mia	− 2.2			1.8
2676897	Mir23a	− 1.4			1.4
2676912	Mir99a	−2.0			1.7
1298220	Naip5	− 1.3			1.7
1298222; 1858256	Naip6; Naip7	− 1.5			2.1
2449980	Noxal	− 1.4			1.5
1201784	Oit1	− 1.4			1.4
94860	Ppp1r1b	− 1.4			1.6
2684952	Sbspon	− 1.4			3.7
2145373	Slc25a48	− 1.6			1.5
98371	Sox9	− 2.0			1.4
1341828	St6galnac3	− 1.4			1.3
Putative Inhibitors
88179	Bmp3	1.5			− 1.3
5453643	Mir5619	1.4			−1.4
102476	mt-Tr	1.6			− 1.8

^aMouse Genome Informatics (www.informatics.jax.org).

2.6. PCR confirmation of Sox9 regulated genes in male UGE determined by microarray

Follow-up quantitative PCR, targeting 14 differentially regulated genes in the male UGE determined by microarray (control male vs Sox9 cKO male; Supplementary Table 1) was carried out using a different RNA pool that was collected from control and Sox9 cKO UGEs for PCR (Supplementary Table 9). Fig. 5 shows that the PCR results essentially confirmed microarray results. This was the case for genes with various putative functions including: cell migration, prostatic bud promotion, prostatic bud inhibition and other functions. Also for each of the 14 genes assessed by microarray and by PCR in Fig. 5, relative expression levels for “individual UGE samples” are shown as “red dots” in Supplementary Fig. 2 illustrating variance in the data.

Fig. 5.

Confirmation of Sox9-dependent genes in male mouse UGE by PCR and microarray. Comparison of microarray analysis to qRT-PCR in determining relative expression of Sox9-dependent transcripts in the UGE of control (Con) vs Sox9 cKO male mouse fetuses at E16.75. Genes selected for this comparison were initially identified by microarray analysis as being Sox9-dependent by comparing their relative transcript abundance between control male vs Sox9 cKO UGEs at E16.75. Some of these Sox9-dependent genes were then discovered by gene ontology analysis to function in cell migration or other functions while other Sox9-dependent genes in the male were identified as putative prostatic bud promoters and inhibitors. Sox9-dependent genes, within these functional groupings, were selected for confirmation of differential expression by qRT-PCR. UGEs were assessed by microarray (n = 6, both groups) and qRT-PCR (n = 4, both groups). UGEs analyzed by qRT-PCR were separate and distinct from those analyzed by microarray. The expression of these transcripts in control samples, determined by microarray from the bi-weight average signal, has been set to 1.0, while the expression in Sox9 cKO samples is shown as the fold-change of Sox9 cKO signal, to control signal. For each transcript analyzed with qRT-PCR, the average expression relative to Actb1 (2^−ΔCt) in control UGEs have been set to 1.0. Expression of these transcripts relative to Actb1 in Sox9 cKO samples is shown as the fold-change of the relative expression seen in controls (2^−ΔΔCt). For both methods error bars represent SEM. Asterisks indicate a significant difference between control and Sox9 cKO (p ≤ 0.05) when the same method of gene expression analysis is used; ns indicates no significant difference (p > 0.05). Abbreviations: conditional knockout (cKO), urogenital epithelium (UGE), standard error of the mean (SEM) and quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR).

2.7. GO analysis of microarray data from control and Sox9 cKO female UGE also shows cell migration is impaired

Having described the Sox9 deficient transcriptome of the male UGE (Supplementary Table 1) we did the same for the female UGE (Supplementary Table 4). Total RNA was isolated from UGEs of E16.75 control female and Sox9 cKO female fetuses for microarray analysis. Compared to the control female UGE, 74 annotated transcripts were differentially regulated at least 1.3 fold in the Sox9 cKO female UGE: 34 downregulated and 40 upregulated (p ≤ 0.05). Additionally, 124 (63%) unannotated transcripts were significantly differentially regulated at least 1.3 fold–57 downregulated and 67 upregulated. Gene ontology analysis of the differentially regulated genes in the female UGE identified three pathways or cell functions for which there was significant gene enrichment (Z score < −2.0 or ≥ +2.0). Two of these were related to cell migration (Supplementary Table 5).

2.8. Comparison of “Sox9-Dependent” genes in male and female UGE - notable similarities but marked differences

Comparing the 95 “Male Sox9-Dependent” annotated transcripts in the male UGE (Supplementary Table 1) to the 74 “Female Sox9-Dependent” annotated transcripts in the female UGE (Supplementary Table 4) revealed 16 annotated transcripts (and 8 unannotated transcripts) that are regulated, in common, by Sox9 in both sexes (union of Venn diagram in Fig. 6, Table 4 and Supplementary Table 6). Of these 16 genes, 14 were downregulated in the UGE of both sexes and 2 were upregulated. Of the 14 downregulated “Sox9-dependent” genes, 7 were shown by GO analysis (Cdc43ep1, Cxcl14, Cxcr4, Itgb8, Shh, Sox9, Tns4) to function in cell migration (Table 2; Supplementary Table 5).

Fig. 6. Only a few Sox9-dependent genes in mouse UGE are “common” to both sexes.

Union (dark tan) of the Venn diagram shows that 16 genes exhibit Sox9-dependent expression in both male and female UGEs at E16.75. The number of Sox9-dependent genes that are significantly upregulated and downregulated (± 1.3 fold, p ≤ 0.05) are given for male UGEs (tan) and female UGEs (red) based on a comparison of gene expression (control vs Sox9 cKO) within each sex. For each sex, the number of differentially regulated, annotated transcripts is shown at the top, and unannotated transcripts in parentheses at the bottom. Names of the 16 Sox9-dependent genes, common to both sexes (union, dark tan), are given in Table 4. Results for males are based on n = 6 litter independent, male UGEs for both control and Sox9 cKO groups, respectively. Findings for females are based on n = 5 litter independent, female UGEs for both control and Sox9 cKO groups.

Table 4.

SOX9 dependent genes differentially expressed, in common, in male and female UGEs at E16.75.

Groups Compared:		Sox9 cKO vs Control
MGIa Gene ID	Gene Symbol	Male Fold Change	Female Fold Change
98371	Sox9	−2.0	−2.1
Downregulated in	Sox9 cKOb
99538	Ascm3	−1.9	−1.8
1931322	Car13	−1.7	−1.4
1929763	Cdc42ep1	−1.3	−1.3
1341272	Col14a1	−2.1	−1.6
1888514	Cxcl144	−1.3	−1.6
109563	Cxcr4	−3.1	−1.6
106627	Gpx2-ps1	−1.5	−1.3
96223	Hr	−1.4	−1.3
1338035	Itgb8	−1.6	−1.3
98297	Shh	−1.4	−1.7
1915778	Smim6	−1.5	−1.4
106196	Stfa3	−1.4	−1.4
2144377	Tns4	−1.4	−1.3
Upregulated in Sox9 cKOc
1933157	Pdzrn3	1.3	1.3
97795	Ptger3	1.3	1.4

^aMouse Genome Informatics (www.informatics.jax.org).

The last major finding in comparing “Male Sox9-Dependent” to “Female Sox9-Dependent” genes in the UGE is that, the majority of these genes were differentially regulated in only the male or female. In the male UGE, 79 (83%) of the 95 “Sox9-dependent” genes (Fig. 6) were unique to males whereas in the female UGE, 58 (78%) of 74 such genes (Fig. 6) were unique to females. This sex-dependent difference in the Sox9 null UGE transcriptome at E16.75 is illustrated by the heatmap in Fig. 7. The highest fold change for the full set of differentially expressed genes, for both genders, was the unannotated transcript identified as TC0500002057.mm.1 in the microarray (5.35 in female and 4.1 in male) (Supplementary Table 7).

Fig. 7. Sex-dependent differences characterize the Sox9 deficient, UGE transcriptome of the mouse fetus.

Heatmap of 151 differentially expressed, Sox9-dependent genes are shown for the “Male” and “Female” mouse UGE at E16.75 (Supplementary Table 8). Fold change in expression of individual genes in the “Male” [Sox9 cKO male vs control male] and “Female” [Sox9 cKO female vs control female] UGE is depicted in the heatmap as yellow (upregulated) and blue (down-regulated). Of these 151 genes, 16 (11%) were significantly different from control in both sexes (regulated in common); 58 (38%) were significantly different from control only in the “Female” and 77 (51%) were different only in the “Male”. Results for males are based on n = 6 litter independent, male UGEs for both control and Sox9 cKO groups, respectively. Findings for females are based on n = 5 litter independent, female UGEs for both control and Sox9 cKO groups.

3. Discussion

3.1. Major conclusions about the role of Sox9 in prostate development

We identified Sox9 protein at an earlier stage of mouse prostate development than previously reported. No expression was observed at E14.5, but by E15.0 SOX9 was expressed in urethra, UGS and bladder epithelium and UGS mesenchyme. We used a Shh^cre driver to delete Sox9 from UGS epithelium prior to onset of budding. In contrast to a previous study describing a regionally restricted role of Sox9 in ventral prostate development (Thomsen et al., 2008), we show that Sox9 is required for elongation of all (anterior, dorsolateral and ventral) prostatic buds. Shhcre;Sox9 knockout mouse prostatic buds appropriately specify and initiate but fail to elongate, resulting in a “stunted buds” phenotype. We performed transcriptomic analysis to reveal possible mechanisms of Sox9 action. Our results support a model in which Sox9 functions uniquely in male UGS epithelium to mediate bud outgrowth by potentially enhancing epithelial cell migration.

3.2. A more widely distributed role for Sox9 in prostatic bud elongation than previously appreciated

Our conclusion that UGS epithelial Sox9 is required for outgrowth of all mouse prostatic buds differs from two previous reports. Thomsen et al. (2008) used a Nkx3-1cre;Sox9 knockout to conclude Sox9 acts regionally to control ventral prostate development without appreciably affecting anterior and dorsolateral prostate development. Nkx3-1 is first expressed in the UGS at E15.5 (Keil et al., 2012b) and ventral prostatic buds form after anterior and dorsolateral buds (Lin et al., 2003). A possible reason why ventral prostate development is selectively impaired in Nkx3-1cre;Sox9 knockouts is that the cre is active for a longer period of time in the ventral region, enabling more complete re-combination prior to ventral bud formation. In contrast, complete Sox9 deletion is attained at least as early as E12.0 in Shhcre;Sox9 knockout UGS epithelium (Seifert et al., 2009), resulting in widespread defects in prostatic bud elongation. Huang et al. (2012) concluded from the tamoxifen-inducible Rosa26ERCre-Sox9^flox/flox conditional knockout that Sox9 is essential for bud initiation and is a mediator of the UGE/prostate epithelial lineage. However, the cre driver used in their study is expressed in both UGS epithelium and mesenchyme, differing from the UGS epithelial-specific Sox9 knockout in our study. Their study was performed in vitro while ours was performed in vivo. Finally, it is possible that stunted buds developed in the Sox9 mutant UGS organ cultures of Huang et al. (2012), but due to their abnormally small size, were not detected.

3.3. Sox9-dependent transcriptome in UGS epithelium differs from that of other tissues

The Sox9-dependent transcriptome in male and female UGEs in the present study revealed few differentially expressed transcripts (< 0.63% of all transcripts) and the fold changes in gene expression that were significant, were modest (1.3 fold cut-off, p < 0.05). The developmental stage selected to assess gene expression may play a role. E16.75 is before the start of bud elongation (Lin et al., 2003; Vezina et al., 2008). If we had assessed Sox9-dependent gene expression one day later, when far more initiated buds were elongating, fold changes in expression may have been greater. Finally, modest changes in Sox9-dependent gene expression are not unique to the UGE. This was also observed in mouse atrioventricular canal and hair follicle stem cell transcriptomes following deletion of Sox9 (Garside et al., 2015; Kadaja et al., 2014). Thus, Sox9 functions in the UGS epithelium by modulating the expression of genes as opposed to strongly inducing or repressing their expression.

SOX transcription factors exert their activating or repressive functions on gene expression by binding protein partners (Kamachi and Kondoh, 2013). The combinatorial code of target specificity produced by SOXs and their partner proteins (Soxpartner code) (Kamachi et al., 2000; Kondoh and Kamachi, 2010; Kamachi and Kondoh, 2013) ensures stringent target gene selection. Accordingly, Sox9-dependent genes in male UGE differ from those identified in cartilage (Lefebvre et al., 1997; Bridgewater et al., 1998; Sekiya et al., 2000; Xie et al., 1999), chondrocytes (Oh et al., 2014), testes (de Santa Barbara et al., 1998), neural crest (Spokony et al., 2002), intestine (Blache et al., 2004), lung (Rockich et al., 2013), heart (Garside et al., 2015) and, except for Mia, conjunctiva (Chen et al., 2014).

3.4. Sox9 function in cell migration fits with Sox9’s role in extracellular matrix regulation

Many developing organs require cell migration (Hogan, 1999) including salivary glands (Hauser and Hoffman, 2015; Harunaga et al., 2011), ocular glands (Tsau et al., 2011; Garg and Zhang, 2017), mammary glands (Hinck and Silberstein, 2005), lungs (Weaver et al., 2000; Rockich et al., 2013), kidneys (Chi et al., 2009; Kuure et al., 2010) limbs (Wyngaarden et al., 2010; Hopyan et al., 2011; Hopyan, 2017) and neural crest derived tissues such as the nervous system (Minoux and Rijli, 2010). Sox9 is required for cell migration in lung buds (Rockich et al., 2013), in bladder, prostate and lung cancer cells (Ling et al., 2011; Wang et al., 2015; Cai et al., 2013), and in metastatic cell migration in breast, colon, prostate and skin (Chakravarty, et al., 2011; Bowen et al., 2009; Cai et al., 2013; Francis et al., 2018; Wang et al., 2008; Rao et al., 2010).

Our proposed mechanism of Sox9 as a mediator of epithelial cell migration in prostatic buds is consistent with the established role of Sox9 in ECM regulation. Migrating cells polarize and then form ECM adhesions before protruding in the direction of migration (Ganser et al., 1991; Hinck and Silberstein, 2005). ECM adhesions act as traction sites for forward movement (Ridley et al., 2003) and ECM components associate with the cell cytoskeleton and influence: cytoskeletal reorganization (Hay, 1982; Nishizaka et al., 2000) and cell migration (Ridley et al., 2003). Sox9 plays a pivotal role in both (Pritchett et al., 2011).

SOX9 directly binds loci of 18 ECM genes in chondrocytes (Oh et al., 2010) and regulates genes encoding ECM proteins and modifying enzymes (Oh et al., 2014). Sox9 organizes valvular ECM proteins in heart (Lincoln et al., 2007) and regulates ECM proteins and matrix metalloproteinases (MMPs) in gonads (Georg et al., 2012; Nakamura et al., 2012). Sox9 mediates fibrotic and sclerotic diseases by promoting excessive and inappropriate ECM deposition (Naitoh et al., 2005; Hanley et al., 2008; Bennett et al., 2007; Sumi et al., 2007; Airik et al., 2010; Schulick et al., 1998). Thus, regulation of the ECM is a critical Sox9 function.

We identified a subset of Sox9-dependent genes in male UGS epithelium (Table 2, Supplementary Table 5) that mediate cell migration. Cdc42 regulates cell polarity (Ridley et al., 2003) and Barx2 facilitates cell adhesion and ECM remodeling (Meech et al., 2005; Stevens and Meech, 2006); both crucial for ocular gland bud elongation (Tsau et al., 2011). CXCR4 modulates MMP expression and enhances cell migration (Singh et al., 2004) and Col14a1 has an adhesive role integrating collagen bundles (Schnittger et al., 1995). Cell migration and ECM remodeling are functions of integrin β8 (Mertens-Walker, et al., 2015) and Thbsl, an adhesive glycoprotein, mediates cell-matrix interactions (Hu et al., 2017). Tns4 binds β integrin, linking ECM to actin cytoskeleton, to promote cell migration (Haynie, 2014). TGFBI, secreted from UGE, interacts with ECM proteins and integrin receptors to decrease cell adhesion (Nummela et al., 2012).

3.5. Putative roles for Sox9 in UGS epithelial cell migration and distal progenitor cell maintenance

We raise the possibility that prostatic buds fail to elongate in UGE conditional Sox9 knockouts because of decreases in UGE cell migration, UGE cell proliferation, and/or distal progenitor cell maintenance. Without these potential Sox9-dependent functions in the UGS epithelium, we hypothesize that initiated prostatic buds do not elongate, culminating in a “stunted buds” phenotype. This proposed mechanism of action for Sox9 recognizes that an individual prostatic bud has two parts: proximal base (which attaches to UGS) and distal tip. This distinction is important as cell migration and proliferation during bud elongation are greater distally than proximally (Tsau et al., 2011; Sugimura et al., 1986) and so is SOX9 expression (Wang et al., 2008; Thomsen et al., 2008; Huang et al., 2012). The possibility that Sox9 is needed for migration of UGS epithelial cells is predicted by GO analysis of our microarray data collected at E16.75 (Table 2). The notion that Sox9 is required for proliferation of UGS epithelial cells is suggested by DNA synthesis being reduced in stunted buds of the ventral prostate at E18.5 when Sox9 is deleted from the UGE (Thomsen et al., 2008). Other studies also show Sox9 is required for epithelial cell proliferation and migration during bud elongation in the development of other branching organs and for metastasis in cancer (Rockich et al., 2013; Chakravarty, et al., 2011; Bowen et al., 2009; Cai et al., 2013; Francis et al., 2018; Wang et al., 2008; Rao et al., 2010).

4. Experimental Procedures

4.1. Mice, mouse husbandry and tissue collection

All mice were housed in polysulfone cages containing corn cob bedding and maintained on 12 h light/dark cycles at 21 ± 1°C and 20–50% relative humidity. Feed (Harlan Teklad Rodent Diet 8604, Harlan Laboratories Inc., Madison, WI) and water were available ad libitum. All procedures were approved by the University of Wisconsin Animal Care and Use Committee and conducted in accordance with the NIH Guide for Care and Use of Laboratory Animals.

The following mice were originally purchased from The Jackson Laboratory (Bar Harbor, ME) and maintained in small breeding colonies in an AAALAC-approved vivarium: C57BL/6J (stock no. 000664), ROSA26 reporter mice (B6.129S4-Gt(ROSA)26Sor^tm1Sor/J; Soriano, 1999), mice with a Shh^cre/+ targeted mutation (B6.Cg-Shh^{tm1(EGFP/cre)Cjt}/J; Harfe et al., 2004) and mice with a Sox9^fl/fl targeted mutation (B6.129S7-Sox9^tm2crm/J; Akiyama et al., 2002); all genetically engineered mice had a C57BL/6J background.

To obtain timed-pregnant dams, females were paired with males for 2–3 h; the time at which the mating pair was separated was considered embryonic day (E) 0. Pregnant dams were euthanized by CO₂ asphyxiation at different stages of fetal development. Fetuses and neonates (P0.5) were euthanized by decapitation. The entire genitourinary (GU) tract, containing the UGS, was removed from fetuses or neonates and fixed in 10% neutral buffered formalin for approximately 24 h at 4°C. For short-term storage, fixed GU tracts were kept in 70% ethanol at 4°C; for long-term storage, GU tracts went through graded dehydration steps and were kept in 100% ethanol or 100% methanol at −20°C.

4.2. Study design

Table 1 gives the general design, including mating schemes, experimental use and developmental stage of mice in each experimental group.

4.3. β-Galactosidase staining

E14.5 cre reporter mouse UGSs were placed in staining buffer (5mM potassium ferricyanide, 2 mM magnesium chloride, 0.01% sodium deoxycholate and 0.02% Nonidet P40 in PBS, pH 7.5). UGS tissues were incubated at 37°C for 5–10 min to equilibrate the UGS to the buffer. Bluo-gal (Gold Biotechnology, B-673–250), a β-galactosidase substrate, was used to stain UGS tissues. Bluo-gal staining solution, 100 mg/ml Bluo-gal dissolved in N,N-dimethylformamide (Sigma-Aldrich, 227056), was added to a final concentration of 1 mg/ml. UGS tissues were stained for 2 h at 37°C and then fixed with 10% neutral buffered formalin for approximately 24 h at 4°C.

4.4. Immunohistochemistry (IHC)

Control and Sox9 cKO UGSs were sectioned and immunostained to confirm Sox9 knockout. Fixed UGS tissues were embedded in paraffin, cut into 5 μm sagittal sections, deparaffinized and rehydrated, treated with 3% hydrogen peroxide for 10 min, boiled in 10 mM sodium citrate for 20 min, and allowed to cool to room temperature to unmask epitopes. Sections were blocked for 2 h with blocking solution: 5% goat serum (Sigma-Aldrich, G9023) and 1% bovine serum albumin (EMD Millipore, 2910) in phosphate buffered saline containing 0.05% Tween-20 (PBST; Sigma-Aldrich, P3563). The Rabbit Anti-SOX9 primary antibody (Abcam ab185230) was diluted 1:250 in blocking solution, applied to the section, and incubated overnight at 4°C. Sections were washed with PBST and subsequently incubated for 1 h with biotinylated goat anti-rabbit IgG secondary antibody (Vector Labs, BA-1000) diluted 1:250 in blocking solution. Sections were washed with PBST and incubated for 30 min with peroxidase-conjugated streptavidin (Vector Labs, PK-6100). After washing with PBST, staining was achieved by incubating sections with ImmPACT DAB solution (Vector Labs, SK-4105) for 2–5 min at room temperature. Sections were counterstained for 20–30 s with Hematoxylin QS (Vector Labs, H-3404) to label nuclei.

4.5. Whole mount (WM) immunohistochemistry (IHC)

Whole mount IHC staining of control, Sox9 het and Sox9 cKO UGSs, harvested at E18.5 or P0.5, was done to observe effects of Sox9 cKO on prostatic budding. Fixed UGSs were rehydrated into PBS and the bladder and urethra were trimmed off. The complete staining procedure was published previously (Keil et al., 2012a). The primary antibody, rabbit anti-mouse CDH1 antibody (Cell Signaling #3195, RRID:AB_2291471), was diluted 1:350 in blocking buffer. The secondary antibody, Horse Anti-rabbit IgG, ImmPRESS VR Reagent (Vector Labs, MP-6401, RRID:AB_2336529), is provided at a ready-to-use concentration. The chromogen used as the enzymatic substrate was ImmPACT DAB (Peroxidase Substrate Kit, Vector, SK-4105).

Stained whole mount UGSs were viewed under a dissecting microscope, and prostatic buds were counted by region – anterior, dorsolateral and ventral – then summed for total bud number. Unpaired Student’s t-tests were used to determine if observed differences in bud number between control and Sox9 cKO male UGSs were significant (P < 0.05).

4.6. Separation of UGE from UGS and isolation of RNA from UGE

UGS (with bladder and urethra attached) was harvested from male and female Sox9 cKO and littermate control fetuses on E16.75. Each UGS was immediately placed into a 1.5 ml microfuge tube containing 300μl of 1% trypsin (Difco, 215240) in PBS and incubated on ice for 30 min. Collagenase (Sigma, C9891) was added to a final concentration of 1 mg/ml, followed by an additional 30–45 min incubation on ice. A dissecting microscope was used to mechanically separate UGM from UGE, after which the bladder and urethra epithelium were removed, leaving only UGE. Total RNA was purified from each UGE using RNeasy system (Qiagen, Hilden, Germany) and analyzed using Agilent Bioanalyzer 2100 and Agilent RNA 6000 Pico Kit (Agilent Technologies, Santa Clara, CA, USA).

4.7. RNA isolation, microarray experiments, and data analysis

RNA was isolated from UGE of E16.75 control and Sox9 cKO male (N = 6 per genotype) and female (N = 5 per genotype) mouse fetuses. RNA quality assessment, labeling, and hybridization to microarray chips was performed at the University of Wisconsin-Madison Biotechnology Gene Expression Center in 2015. At that time, microarray analysis was widely used due to the expense of RNA sequencing. We understand microarray data can have increased false positives, lower detection range, and saturation of high signals compared to RNA-seq technologies. RNA quality was assessed for all 22 samples using an Agilent 2100 Bioanalyzer. The RNA Integrity Number (RIN) was ≥5.5 for each RNA sample, and the average RIN for all 22 RNAs was 7.3. 2ng of total RNA per sample was amplified and labeled using GeneChip WT Pico Reagent Kit (Affymetrix, Santa Clara, CA, USA), then hybridized to GeneChip Mouse Gene 1.0 ST Array (Affymetrix, Santa Clara, CA, USA) according to manufacturer’s protocol.

Microarray data were analyzed using Transcriptome Analysis Console (TAC; Affymetrix, Santa Clara, CA, USA). A transcript was considered differentially regulated (compared to a control) if it had an absolute fold change in abundance of ≥1.3 and p ≤ 0.05. Due to the low number of annotated differentially expressed genes, we used this cut-off as a broad screening tool for potential targets of interest. Also, due to tissue heterogeneity, low fold changes can have biological relevance which we wanted to capture. The stringency criteria of considering false positive rates of differentially expressed genes (p<0.05) as the minimum fold change to be significant was determined based on microarray validation and assessment described previously (Nobis et al., 2003; Huggins et al., 2008; Bigler et al., 2013, Laurent et al., 2013). An annotated transcript was defined as one having a gene symbol or description following TAC analysis; if these were absent the transcript was considered unannotated. Annotated transcripts from the “Male Sox9-Dependent” (control male vs Sox9 cKO male), or “Female Sox9-Dependent” (control female vs Sox9 cKO female) comparisons were entered into Ingenuity Pathway Analysis software (IPA; Qiagen, Hilden, Germany) for further analysis. This included gene ontology (GO) analysis for which an IPA-generated z score of ≤ −2.0 or ≥ 2.0 was considered significant (p ≤ 0.05). Lists of all differentially expressed genes including for each gene: bi-weight average signal (log2), standard deviation, fold change (linear), ANOVA p-value and FDR-adjusted p-value (q-value), are provided in Supplementary Tables 1, 2 and 4, respectively. All raw microarray data (significant and nonsignificant) from each one of the individual UGE samples used in the present study were uploaded, separately, to the NCBI GEO database (GSE113011; reviewer token mhizgciwxxcxbyl).

4.8. Heatmap construction

Microarray data were extracted from Supplementary Table 1 (Sox9 cKO male vs control male) and Supplementary Table 4 (Sox9 cKO female vs control female). The fold change, gene symbol, description, and grouping was extracted and merged together in Supplementary Table 7. This file was then filtered to remove any unknown gene symbols and saved as Supplementary Table 8. A heatmap was generated with the fold change and clustered using Euclidean distance (Fig. 7) from the data in Supplementary Table 8. All data processing and visualization was done using R and gplots.

4.9. Quantitative real-time polymerase chain reaction (qRT-PCR)

All UGE tissues used for qRT-PCR were separate and distinct from those used for microarray analysis and were used to confirm microarray results. For qRT-PCR, as done previously for microarray, UGEs isolated from male, Sox9 cKO and littermate control fetuses on E16.75 were homogenized and total RNA purified using RNeasy (Qiagen, Hilden, Germany). Individual genes assessed by qRT-PCR were selected from those differentially expressed genes identified earlier by microarray (Supplementary Table 1). RNA from the UGE samples designated for qRT-PCR were then used to validate 14 genes of interest as being differentially expressed between, Sox9 cKO and littermate control male UGEs on E16.75 using Taqman Gene Expression Assays (Life Technologies) and Applied Biosystems 7900 analysis.

More specifically, 10 μl of RNA (4.1 ng/μl) was reverse transcribed with random hexamers and Multiscribe MuLv from the High Capacity cDNA Reverse Transcription Kit (ThermoFisher Scientific) per manufacturer’s protocol. 10 μl of resulting cDNA was preamplified for genes selected through microarray analysis or other genes of interest using TaqMan Preamp Mastermix Kit (ThermoFisher Scientific) for 40 cycles in a 50 μl reaction volume. Probes used were Taqman gene expression assays (ThermoFisher Scientific) for Tns4 (Mm00553421_m1), Adamts9 (Mm00614433_m1), Plod2 (Mm00478767_m1), Col14a1 (Mm00805269_m1), Thbsl (Mm00449032_g1), Mia (Mm00444563_m1), Sbspon (Mm01237899_m1), Cxcr4 (Mm01996749_s1), Bmp3 (Mm00557790_m1), Hr (Mm00498963_m1), Cxcl14 (Mm00444699_m1), Shh (Mm00436528_m1), Epas1 (Mm01236112_m1), Sox9 (Mm00448840_m1), Actb1 (Mm02619580_g1) and Gapdh (Mm99999915_g1). Probes were chosen based on best coverage according to ThermoFisher database. Taqman gene expression assays are MIQE compliant and qRT-PCR was performed following MIQE. qRT-PCR reactions for the above genes were performed with Taqman Universal Master Mix (ThermoFisher Scientific) in a 20 μl reaction volume containing 2 μl of the preamplified cDNA. Thermal cycling parameters were carried out per manufacturer’s protocol. Reactions were done in triplicate. qRT-PCR analysis and calculations were performed in the Sequence Detection System v 2.4. All transcripts examined were normalized to Actbl through the comparative C_t (ΔΔ Ct) method (Livak and Schmittgen, 2001). Actb1 was unaltered by Sox9 cKO in both the microarray and qRT-PCR, and thus was suitable as the housekeeping gene. Differences in gene expression between Sox9 cKO and control UGEs of male and female fetuses, respectively, were determined by comparing the expression values for each gene relative to Actb1. Relative expression = 2^{−(Ct goi − Ct Actb1)} = 2^−ΔCt where Ct means threshold cycle, and goi and Actbl refer to gene of interest and beta actin. A one-tailed Student’s t-test was then used to identify significant differences between the relative expression of each analyzed gene (p ≤ 0.05) between male Sox9 cKO (n=4) and control (n=4) UGEs. The fold change in expression (2^−ΔΔCt) for each gene with respect to the Sox9 cKO condition was calculated by dividing the mean relative expression value of the Sox9 cKO UGEs by the mean relative expression value of the control UGEs (Supplementary Table 9).

Key resources Table

Reagent or resource	Source	Identifier
Antibodies
Rabbit Monoclonal Anti-SOX9	Abcam	Cat# ab185230, RRID:AB_2715497
Rabbit Monoclonal Anti-CDH1	Cell Signaling	Cat#: 3195, RRID:AB_2291471
Biotinylated Goat Anti-rabbit IgG	Vector Labs	Cat#: BA-1000, RRID:AB_2313606
Horse Anti-rabbit IgG, ImmPRESS VR Reagent	Vector Labs	Cat#: MP-6401, RRID:AB_2336529
Bacterial and Virus Strains
None
Biological Samples
Goat Serum	Millipore-Sigma	Cat#: G9023
Bovine Serum Albumin	Millipore-Sigma	Cat#: 29-102-5 GM
Trypsin	Difco	Cat#: 215240
Collagenase	Millipore-Sigma	Cat#: C9891
Chemicals, Peptides, and Recombinant Proteins
Bluo-gal	Gold Biotechnology	Cat#: B-673-250
Critical Commercial Assays
Agilent RNA 6000 Pico	Agilent Technologies	Cat#: 5067-1513
GeneChip Mouse Gene 1.0 ST Array	ThermoFisher Scientific	Cat#: 901171
Deposited Data
RNA Microarray Data	NCBI GEO Database	GSE113011; reviewer token mhizgciwxxcxbyl
Experimental Models: Cell Lines
None
Experimental Models: Organisms/Strains
Mouse: C57BL/6J	Jackson Laboratory	Stock#:000664, RRID: IMSR_JAX: 000664
Mouse: B6.129S4-Gt(ROSA)26sortm1Sor/J	Jackson Laboratory	Stock# 003474, RRID: IMSR_J AX: 003474
Mouse: B6. CgShhtm1(EGFP/cre)Cjt/J	Jackson Laboratory	Stock#005622; RRID: IMSR_J AX: 005622
Mouse: B6.129S7-Sox9tm2Crm/J	Jackson Laboratory	Stock#: 013106, RRID:IMSR_JAX:013106
Oligonucleotides
See Supplemental Table 11 for TaqMan Probe ID	ThermoFisher Scientific
Recombinant DNA
None
Software and Algorithms
Transcriptome Analysis Console	ThermoFisher Scientific
Ingenuity Pathway Analysis	QIAGEN
R	www.r-project.org
Sequence Detection System v 2.-4.	ThermoFisher Scientific
Other
ImmPACT DAB Staining Kit	Vector Labs	Cat#: SK-4105
Hematoxylin QS	Vector Labs	Cat#: H-3404
Peroxidase-conjugated Streptavidin	Vector Labs	Cat#: PK-6100
RNeasy Mini Kit	QIAGEN	Cat#: 74104
GeneChip WT Pico Reagent Kit	ThermoFisher Scientific	Cat#: 902623
High Capacity cDNA Reverse Transcription Kit	ThermoFisher Scientific	Cat#: 4368814
TaqMan Preamp Mastermix Kit	ThermoFisher Scientific	Cat#: 4384267

Abbreviations

A

anterior

AB

anterior bud

Array

microarray

BL

bladder

CDH1

Cadherin-1

cKO

conditional knockout

Con

control

Cre

creates re-combination

Cxcr4

C-X-C motif chemokine receptor 4

D

dorsal

DB

dorsal bud

DHT

5α-dihydrotestosterone

DLB

dorsolateral bud

E

embryonic day

ECM

extracellular matrix

ER

estrogen receptor

FGF

fibroblast growth factor

GO

gene ontology

GU

genitourinary

het

heterozygous

IHC

immunohistochemistry

IPA

ingenuity pathway analyses

Mia

melanoma inhibitory activity

MIAME

minimal information about a microarray experiment

MIQE

minimum information for the publication of qPCR experiments

MMP

matrix metalloproteinase

ns

no significant difference

P

postnatal day

PBST

phosphate buffered saline containing 0.05% Tween-20

SV

seminal vesicle

Shh

sonic hedgehog

qRT-PCR

quantitative real-time polymerase chain reaction

Sox9

sry-box 9

TAC

Transcriptome Analysis Console

UGS

urogenital sinus

UGE

urogenital sinus epithelium

UGM

urogenital sinus mesenchyme

UR

urethra

V

ventral

VB

ventral bud

WD

wolffian duct

WM

whole mount