Transcriptome Analysis of the Dihydrotestosterone-Exposed Fetal Rat Gubernaculum Identifies Common Androgen and Insulin-Like 3 Targets

ABSTRACT

Androgens and insulin-like 3 (INSL3) are required for development of the fetal gubernaculum and testicular descent. Previous studies suggested that the INSL3-exposed fetal gubernacular transcriptome is enriched for genes involved in neural pathways. In the present study, we profiled the transcriptome of fetal gubernaculum explants exposed to dihydrotestosterone (DHT) and compared this response to that with INSL3. We exposed fetal (Embryonic Day 17) rat gubernacula to DHT for 24 h (10 and 30 nM) or 6 h (1 and 10 nM) in organ culture and analyzed gene expression relative to that of vehicle-treated controls using Affymetrix arrays. Results were annotated using functional, pathway, and promoter analyses and independently validated for selected transcripts using quantitative RT-PCR (qRT-PCR). Transcripts were differentially expressed after 24 h but not 6 h. Most highly overrepresented functional categories included those related to gene expression, skeletal and muscular development and function, and Wnt signaling. Promoter response elements enriched in the DHT-specific transcriptome included consensus sequences for c-ETS1, ELK1, CREB, CRE-BP1/c-June, NRF2, and USF. We observed that 55% of DHT probe sets were also differentially expressed after INSL3 exposure and that the direction of change was the same in 96%. The qRT-PCR results confirmed that DHT increased expression of the INSL3-responsive genes Crlf1 and Chrdl2 but reduced expression of Wnt4. We also validated reduced Tgfb2 and Cxcl12 and increased Slit3 expression following DHT exposure. These data suggest a robust overlap in the DHT- and INSL3-regulated transcriptome that may be mediated in part by CREB signaling and a common Wnt pathway response for both hormones in the fetal gubernaculum.

Exposure of the developing fetal rat gubernaculum to dihydrotestosterone in vitro produces a transcriptome response characterized by overrepresentation of Wnt signaling pathway genes and strong overlap with the effects of INSL3.

INTRODUCTION

The gubernaculum testis is an intermediate mesoderm derivative that develops as a specialized fibromuscular ligament and orchestrates descent of the testis during mid to late gestation in mammals [1]. The fetal rodent gubernacular bulb contains an inner mesenchymal component that increases markedly in size during development and two peripheral layers of cremaster muscle [2]. Gubernacular swelling occurs due to cellular proliferation, extracellular matrix deposition, and enlargement of the muscle layer. The gubernaculum migrates into the scrotum around the time of birth, providing a space for testicular transit. Failure of normal development or function of the gubernaculum leads to cryptorchidism, or undescended testis, a common developmental abnormality.

Gubernacular development requires the Leydig cell hormones insulin-like 3 (INSL3), the ligand for the relaxin-insulin-like family peptide receptor 2 (RXFP2), and androgens (testosterone and its metabolite, dihydrotestosterone [DHT]), which activate transcription after binding to the androgen receptor (AR). Explant experiments identified cellular proliferation as a response to hormonal stimulation of the gubernacular bulb, with some studies suggesting synergistic effects of androgens and INSL3, but the target cells were not specifically defined [3, 4] The Rxfp2 transcript is expressed at high levels by Embryonic Day (E) 14.5 in mice [5], and both Rxfp2 and AR are primarily localized to mesenchymal cells within the central portion of the fetal gubernacular bulb, with minimal expression in the peripheral differentiated cremaster muscle [6, 7]. AR is localized to the nucleus in mesenchymal cells [8–10], suggesting that androgens signal via traditional genomic pathways in gubernaculum. However, AR activity in this tissue may be weaker than in other reproductive tissues [8, 11], with testicular descent being less responsive to antiandrogenic inhibition than other aspects of male reproductive development [12].

Targeted ablation of Rxfp2 or Ar in mesenchymal but not differentiated striated or smooth muscle cells interferes with gubernacular development and testicular descent, further implicating the hormone receptor-expressing mesenchyme as the target of hormonal action in the fetal gubernaculum [6, 7]. In gubernaculum-specific Rxfp2^−/− transgenic mice, desmin-positive (muscle-specific) cells are mislocalized within the developing mesenchymal core of the fetal gubernaculum, whereas the outer cremaster muscle layer fails to develop normally, suggesting that proliferation, migration, and/or differentiation of muscle precursors is dependent on INSL3/RXFP2 signaling [9]. Moreover, migration of AR-positive cells into the gubernaculum and proliferation of AR-positive cells within the gubernaculum fail to occur normally in conditional knockout Rxfp2 transgenic fetuses, suggesting interaction between RXFP2 and AR signaling pathways. Conditional deletion of Notch1 or Ctnnb1 in gubernacular mesenchyme also results in failed muscle patterning in the fetal gubernaculum, suggesting a role for these pathways downstream of RXFP2 signaling [9]. In contrast, cryptorchidism is present in the GU-ARKO mouse, with conditional deletion of Ar in the gubernacular mesenchyme, but the associated developmental phenotype is more subtle and limited to delayed and incomplete perinatal gubernacular migration and altered expression of muscle-specific genes in the postnatal cremaster [7].

In previous studies, we analyzed the fetal gubernaculum transcriptome following stimulation with INSL3 in vitro to define the downstream effects of RXFP2 signaling during fetal development and, most notably, observed altered expression of transcripts in neurodevelopmental pathways [13]. In the present study, we conducted similar experiments designed to elucidate the effects of androgenic stimulation on development of the gubernaculum. Our results suggest that the androgenic response of the fetal gubernaculum is less robust than the INSL3 response during this time frame. While the most prominent changes in gene expression are unique for each hormone, Wnt signaling is a common target, and significant overlap exists in gubernacular transcriptional response following exposure to DHT and INSL3.

MATERIALS AND METHODS

Animals

Long Evans rats (Charles River Laboratories) aged 2–3 mo were maintained at the Nemours Biomedical Research facility accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International, and animal care and use were approved by the Institutional Animal Care and Use Committee. Rats received food (Lab Diet Rat Chow 5021; PMI Nutrition International) and water ad libitum and were housed in polycarbonate cages with pine shaving bedding in a room with a 12L:12D photoperiod and controlled temperature (70°C ± 2°C) and humidity (35%–70%). After a period of adjustment following transport, female rats were placed with male rats and timed pregnancies generated. Vaginal smears were used to document presence of sperm, and the morning of the day of detection was designated E0. On E17 between 1200 and 1500 h, dams were euthanized by carbon dioxide asphyxiation and pups by decapitation.

Organ Culture

The culture system was modeled after that described previously by Emmen et al. [3]. Microdissected E17 paired gubernacular bulbs from a single animal were cultured on a Millicell CM membrane (catalog no. PICM03050; Millipore Corp.) in 6-well polystyrene culture plates (catalog no. 3506; Corning, Inc.) containing 1 ml of medium per well. All organ culture reagents were obtained from Invitrogen Corp. and included Dulbecco modified Eagle medium/Ham F-12 (catalog no. 21041-025), 2% charcoal stripped fetal bovine serum (catalog no. 12676-011), 1× insulin-transferrin-selenium-X supplement (catalog no. 51500-056), and 1× antibiotic-antimycotic (catalog no. 15240-096). Cultures were maintained at 37°C in a humidified chamber with 5% carbon dioxide. Gubernacula were cultured for 24 h in basal medium to provide a washout period and then for 6 or 24 h with either basal medium or medium containing DHT (1, 10, or 30 nM; catalog no. A8380; Sigma). The DHT concentrations were chosen based on previous in vitro experiments using fetal gubernacula [3, 4], myoblasts [14], or prostate cancer cells [15]. In addition, gubernacular bulbs were obtained from E17, E19, and E21 male fetuses derived from two or more litters for baseline Ar mRNA analysis.

Microarrays

Sample hybridization and data analysis were performed as described previously [13]. Total RNA (n = 5–6 replicates/group from ≥2 litters) was isolated using the RNeasy Micro Kit (Qiagen, Inc.) with DNase treatment, and RNA integrity numbers were greater than eight for all samples. For the 24-h exposure experiments, total RNA (50 ng) was labeled using the GeneChip 3′ IVT Express Kit (Affymetrix, Inc.), and biotin-labeled cRNA was hybridized to GeneChip Rat Genome 230 2.0 arrays (Affymetrix, Inc.). For the 6-h exposure studies, the Applause WT-AMP ST System (NuGen, Inc.) and Rat Gene 1.0 ST arrays (Affymetrix, Inc.) were used for labeling and hybridization, respectively. For the 24-h exposure studies, raw signal intensities for each probe set were background-corrected, normalized, and summarized using GC Robust Multiarray Analysis (GC-RMA) and the affylmGUI Bioconductor package within R (http://www.r-project.org/). For the 6-h exposure studies, raw signal intensities were background-corrected, normalized, and summarized using Robust Multiarray Analysis (RMA) and the oligo Bioconductor package within R. Array data are deposited in Gene Expression Omnibus of the National Center for Biotechnology Information under GEO accession number GSE19658 (www.ncbi.nlm.nih.gov/geo/).

Real-Time Quantitative RT-PCR

Genes of interest were analyzed in independent E17 gubernaculum samples (n = 11–14 per group from 3 separate litters) cultured for 24 h with 0, 10, or 30 nM DHT and in freshly harvested E17, E19, and E21 gubernacula (n = 12–20 per group). Total RNA purification, cDNA generation, TaqMan-based quantitative RT-PCR (qRT-PCR), and data analysis using the delta delta CT method were performed as described previously [13]. Expression of target genes was normalized to Gapdh expression and quantified relative to expression in a fetal rat embryo total RNA preparation (catalog no. AM7928; Applied Biosystems/Ambion). The following prevalidated TaqMan gene expression assays (Applied Biosystems, Inc.) were used: Adh1, Rn00570670_m1; Ar, Rn00560747_m1; Bmp4, Rn00432087_m1; Chrdl2, Rn01510694_m1; Crlf1, Rn01419973_m1; Cxcl12, Rn00573260_m1; Gapdh, Rn99999916_s1; Has2, Rn00565774_m1; Myh3, Rn00561539_m1; Myh7, Rn00568328_m1; Myog, Rn00567418_m1; Npy, Rn01410145_m1; Sfrp2, Rn01458837_m1; Slit3, Rn00580013_m1; Tgfb2, Rn00579674_m1; Wnt4, Rn00584577_m1; and Wnt5a, Rn00575260_m1.

Bioinformatics and Statistical Analysis

Filtering was applied to include probe sets with a log2 expression value greater than 3.5 in all replicates of at least one group. Using the LIMMA package in Bioconductor, we used a linear model approach with false-discovery rate (FDR) correction to identify differentially expressed probe sets, defined as those with an FDR-corrected P-value of less than 0.05 in comparisons of 1, 10, or 30 nM DHT exposure relative to the unexposed control group. The following lists of probe sets were analyzed: 1) all probe sets with significant expression changes after 10 or 30 nM DHT exposure, 2) all probe sets with significant expression changes after 10 nM DHT exposure, and 3) significant probe sets with an expression fold-change greater than or equal to an absolute value of 1.5. Pathway analysis was conducted using Ingenuity Pathway Analysis (IPA; Spring Release 2013; Ingenuity Systems, Inc.). Using IPA, functional annotations and canonical pathways enriched in the gene list with a Benjamini-Hochberg-corrected Fisher exact test P-value of less than 0.05 were considered to be significant. We also compared these lists/analyses with 1) those previously generated in experiments of fetal gubernaculum exposure to 10 or 100 nM INSL3 using the same experimental protocol [13]; 2) an updated list of genes linked to syndromes associated with cryptorchidism based on annotations accessed during June 2013 in Online Inheritance in Man (OMIM; http://omim.org), Mouse Genome Informatics (MGI; http://www.informatics.jax.org), and the literature [9, 16–19]; 3) relevant signaling pathway molecules annotated in NetPath (http://www.netpath.org) [20] and/or neXtProt (http://www.nextprot.org/) [21]; and 4) myoblast-specific AR-binding sites based on the data of Wyce et al. [14]. Quantitative RT-PCR data were analyzed after log transformation with IBM SPSS Statistics Version 19.0 using one-way ANOVA with Dunnett post-hoc tests, with P-values of less than 0.05 considered to be significant.

We analyzed the promoters of differentially expressed DHT-responsive genes for enrichment of transcription factor-binding sites using the Promoter Analysis and Interaction Network (PAINT; v4.0; www.dbi.tju.edu/dbi/tools/paint) [22] as previously described [13]. Gene IDs (n = 1555) retrieved by conversion of all differentially expressed DHT probe sets (n = 2533) were compared to all mouse promoters in the PAINT database. Overrepresented transcriptional response elements (TREs) contained in this data set with an FDR-corrected P-value of less than 0.05 are reported.

RESULTS

Global Gene Expression

In an initial analysis of E17 gubernacula exposed to 10 or 30 nM DHT for 24 h compared to non-hormone-exposed controls, we identified 2533 probe sets (mapped by IPA to 1648 unique genes) with significant expression changes. Of these, 1336 were upregulated, and 1197 were downregulated. The majority of differentially expressed transcripts responded to 10 nM DHT (2214/2533 [87%], representing 1448 genes), and of these, 286 showed altered expression after both 10 and 30 nM DHT exposure. Few transcripts (319/2533 [13%]) were exclusively responsive to 30 nM DHT. Significant changes in transcript levels following both 10 and 30 nM DHT exposure were always in the same direction, and when differences in transcript expression were significant for only one DHT exposure, the mean direction of change was concordant in 73%. However, 1307 additional transcripts were differentially expressed between 10 and 30 nM DHT exposure groups but not between either of the DHT groups and the control group, suggesting intrinsic physiologic differences in response to DHT at the concentrations used.

Transcripts representing annotated genes with prominent (≥2.5-fold) expression changes after both 10 and 30 nM DHT exposure (Table 1) include, most notably, Adh1, which was increased more than 25-fold and is a known androgen-regulated gene in kidney [23]. Other highly regulated genes include Gna14, encoding a guanine nucleotide binding protein; inhibitors of BMP signaling [24–26] Grem2, Smoc1, and Chrdl2; Vldlr, which participates in the reelin signaling pathway involved in neuronal migration [27]; the mesodermal gene Fibin [28]; and Shisa3, member of a family of Wnt and fibroblast growth factor antagonists [29]. In all, 1021 probe sets mapping to 724 unique genes showed at least a 1.5-fold difference in expression after DHT exposure (Supplemental Table S1, available online at www.biolreprod.org). In 6-h exposure experiments using DHT concentrations (1 and 10 nM) that were lower but overlapped those used in the initial experiment, we analyzed mRNA levels using whole-transcript gene expression arrays. This analysis failed to identify any significant differences between the 0, 1, and 10 nM exposure groups, suggesting a more complicated genomic AR response in this experimental model and/or a failure to adequately target AR-positive cells in nondissociated gubernaculum during this shorter window of exposure.

TABLE 1.

Differentially expressed genes with at least a 2.5-fold expression change relative to control after 10 nM and 30 nM DHT exposure.

^*Values represent log2 ratios of DHT nM:control and DHT 30 nM:control.

To better clarify the status of Ar mRNA expression as a variable that could impact response to DHT in this model, we measured baseline levels in freshly harvested fetal gubernacula and in E17 organs cultured with and without DHT. These data (Fig. 1) show that Ar mRNA levels do not change significantly between E17 and E21 in the gubernacular bulb and are 2.7-fold higher in organs cultured without hormone but similar to baseline levels in organs exposed to DHT. Additionally, Ar mRNA expression was 25% lower after 30 nM compared to 10 nM DHT exposure, although this difference was not statistically significant. These data suggest upregulation of Ar transcription in response to androgen withdrawal that is normalized with DHT replacement, which may in part explain the transcriptome differences that we observed following 10 and 30 nM DHT exposure.

FIG. 1.

Results of qRT-PCR for Ar. A) Ar mRNA levels in independent samples (n = 11–14 per group) of E17 gubernacula cultured without hormone for 24 h followed by exposure to 0, 10, or 30 nM DHT for 24 h. B) Ar mRNA levels in untreated fetal gubernacula harvested at E17, E19, and E21 (n = 12–20 per group). Values reflect the mean ± SEM. ***P < 0.001 by one-way ANOVA after log transformation.

Functional Annotation Analysis of Differentially Expressed Transcripts

After correction for multiple testing, overrepresented biological functions identified from analysis of the 286 probe sets (171 mapped genes) that were differentially expressed after exposure to both 10 and 30 nM DHT were limited to organization of extracellular matrix (P = 0.034) and categories related to vascular development (P = 0.042). Selected overrepresented functions and pathways (Table 2) were similar for analysis of transcripts significantly altered following exposure to 10 and/or 30 nM DHT (2533 probe sets) and to 10 nM DHT alone (2214 probe sets). These include general categories related to morphogenesis as well as those specific to development of the skeletal and muscular, neural, and connective tissue systems. When analyses were limited to probe sets with at least 1.5-fold expression changes after DHT exposure (1021 probe sets), the following functional categories were overrepresented: skeletal and muscular, connective, and cardiovascular tissue development; gene expression; and cellular movement; and cell morphology. The list of mapped genes with at least a 1.5-fold expression change (Supplemental Table S1) shows that levels for the majority (69%) were reduced following DHT exposure, but agreement between the 10 and 30 nM exposure levels and the degree of change was greatest for upregulated genes. However, when we analyzed all DHT-regulated probe sets based on direction of change, functional annotations for the downregulated group reflected those we observed for the group as a whole, but no categories were overrepresented for the upregulated group (data not shown). For the 1307 probe sets (824 mapped genes) showing differential expression in response to 10 versus 30 nM DHT but no significant difference relative to controls, IPA analysis results were limited to two canonical pathways (tRNA charging, P = 3.8E-05; prostate cancer signaling, P = 4.7E-03) and one functional category (processing of RNA, P = 0.018).

TABLE 2.

IPA analysis (P-values) of differentially expressed transcripts after DHT and/or INSL3 exposure.*

^*ns, not significant.

Results of the PAINT analysis indicate strong evidence for enrichment of several TREs in DHT-responsive genes, including c-ETS1, ELK1, CREB, CRE-BP1/c-June, NRF2, and USF (all P < 0.001) as well as STAT and v-MYB (P < 0.01) but, notably, no evidence for enrichment of AR response elements (AREs) in the promoters of differentially expressed genes. However, the majority of myoblast AREs are localized to enhancer, not proximal promoter, regions [14], and of the 1648 genes with altered expression after DHT exposure, 761 (46%) are also AR targets in myoblasts (see Supplemental Table S1 for notation of those with a ≥1.5-fold expression change). Wnt/β-catenin signaling emerged as a consistently overrepresented canonical pathway in IPA analyses of DHT-regulated genes (Table 2). Based on IPA and NetPath annotations, we identified a total of 46 Wnt pathway genes regulated by DHT (Table 3): 26 were downregulated (62% by at least ≥1.5-fold), whereas of only 2 of 20 upregulated Wnt genes were changed at least 1.5-fold. In contrast, the AR signaling pathway and a custom AR NetPath list were not significantly overrepresented when scored against the DHT-responsive gubernacular transcriptome.

TABLE 3.

Wnt pathway genes regulated by DHT (10 or 30 nM) and/or INSL3 (10 or 100 nM) in fetal gubernaculum.

^*Values represent log2 ratios of DHT:control or INSL3:control. Significantly increased and decreased expressions are in bold; nonsignificant differences are in italics.

^†Genes associated with undescended testis (UDT) in OMIM and/or MGI.

Overall, the results of our analysis suggest the response of the intact fetal gubernaculum to DHT results in altered transcription of developmental and, most notably, Wnt signaling pathway genes. Although we confirmed robust response of a few transcripts representing known androgen-regulated genes, most responses were modest, and their significance was most likely dependent on ligand and receptor levels in the target tissue. The observations that the 6-h DHT exposure was insufficient to produce statistically significant changes in the transcriptome and that, after 24 hours, transcripts are enriched in TREs that are involved in developmental signaling but do not include AREs suggest that androgen action in the fetal gubernaculum may require AR binding to enhancers, interaction with other transcription factors, and/or activation of additional downstream pathways.

Overlap of DHT and INSL3-Regulated Genes

We identified 1394 transcripts (representing 903 genes) that were present in both DHT (of 2533 transcripts [55%]) and prior INSL3 (of 3482 transcripts [40%]) differentially expressed probe set lists, and almost half of these were altered at least 1.5-fold by DHT (Fig. 2). Notably, the direction of change for 1335 (96%) of these transcripts was the same for DHT and INSL3. Functional analysis of transcripts regulated by both hormones showed strong overlap with the differentially expressed DHT gene lists, as expected (Table 2). Similarly, an updated analysis of gubernaculum transcripts altered by treatment with 10 and/or 100 nM INSL3 in IPA showed multiple highly significant developmental functions and pathways, including near-complete overlap (Table 2) with analyses of DHT-regulated genes. However, the significance levels for INSL3-regulated transcripts were much more robust for all categories. In particular, cellular growth and proliferation and cellular movement are strongly overrepresented categories that were not consistently overrepresented in DHT exposure lists. To better define distinct programs for each hormone, we functionally analyzed 2088 differentially expressed transcripts unique to INSL3 exposure. We identified several functional categories related to cellular growth, migration, and metabolism, including apoptosis (P = 3.62E-05), proliferation of muscle cells (P = 1.19E-03), and cell migration (P = 1.20E-03), as well as canonical pathways, including superpathway of inositol phosphate compounds (P = 3.45E-04), IGF-1 signaling (P = 4.68E-03), superpathway of cholesterol biosynthesis (P = 6.31E-03), and PTEN signaling (P = 7.24E-03). In contrast, a separate functional analysis of 1139 DHT-specific transcripts yielded no significant categories.

FIG. 2.

Venn diagram (created using VENNY [67]) showing overlap of differentially expressed probe sets following 24-h exposure of cultured fetal gubernaculum to INSL3 (all), DHT (all) and DHT (those with a ≥1.5-fold change in expression).

Wnt signaling remained a prominent theme in the results of DHT- and/or INSL3-regulated gene analyses (Table 2). Expression of 71 Wnt pathway genes was altered by DHT (n = 16), INSL3 (n = 26), or both (n = 29), with effects in the same direction in the majority (83%) of the latter group (Table 3). Notably, expression of Wnt2, Wnt4, Wnt5a, Wnt9a, and Wnt16 was increased after INSL3 exposure but decreased or unchanged after DHT exposure. In addition, we identified 53 hormone-regulated genes that have been identified as causative of syndromes associated with cryptorchidism in clinical syndromes and/or mice (Table 4), including eight Wnt signaling pathway genes (Table 3). Of these 53 genes, 18 showed significant changes in expression for both hormones, and in 16 (89%) of these cases, the direction of change was consistent for both. Transcripts representing five additional genes involved in Wnt signaling that are also associated with clinical cryptorchidism were not differentially expressed in response to DHT or INSL3 (Chd7 and Gpc3), were poorly expressed based on GeneChip Rat Genome 230 2.0 annotations but clearly present based on Rat Gene 1.0 ST chip data (Pax2 and Ror2), or were poorly expressed on both arrays (Wnt7a).

TABLE 4.

Genes associated with undescended testis and differentially expressed after DHT (10 or 30 nM) and/or INSL3 (10 or 100 nM) exposure.*

^*Excluding (Wnt) pathway genes.

^†Values represent log2 ratios of DHT:control or INSL3:control. Significantly increased and decreased expressions are in bold; nonsignificant differences are in italics.

In summary, comparison of the INSL3- and DHT-responsive fetal gubernacular transcriptome suggests more robust changes in gene expression after INSL3 exposure but an otherwise similar hormonal response. Not surprisingly, both hormones appear to target developmental genes, most consistently those involved in Wnt signaling, as well as specific genes associated with syndromic cryptorchidism.

Quantitative RT-PCR Validation of Gene Expression

We used qRT-PCR to selectively validate gene expression levels based on biological functions of interest, prior documentation of regulation by androgens, and/or prior documentation of differential expression in developing male gubernacula [30] as well as in vitro INSL3 exposure studies [13]. Although we observed a similar qualitative response to DHT in our analysis of these independent samples (Fig. 3), not all responses were robust or statistically significant. Whereas microarray data are considered to be generally reliable, differentially expressed transcripts with lower fold-changes may be less likely to be reproduced in qRT-PCR experiments [31] and/or qRT-PCR may interrogate different transcript isoforms, producing inconsistent results between the two assay methods.

FIG. 3.

Results of qRT-PCR in independent samples (n = 11–15 per group) of E17 gubernacula exposed to 10 or 30 nM DHT for 24 h. A) Genes analyzed include those regulated by androgens in other contexts (alcohol dehydrogenase 1 [Adh1], hyaluronan synthase 2 [Has2], and neuropeptide Y [Npy]) and developmental and muscle-specific genes (cytokine receptor-like factor 1 [Crlf1]; chordin-like 2 [Chrdl2]; wingless-type MMTV integration site family, member 4 [Wnt4]; bone morphogenetic protein 4 [Bmp4]; transforming growth factor, beta 2 [Tgfb2]; chemokine (C-X-C motif) ligand 12 [Cxcl12]; slit homolog 3 [Slit3]; and secreted frizzled-related protein 2 [Sfrp2]). B) Analysis of muscle-specific transcripts shows that DHT exposure is associated with reduced expression of myosin heavy chain 7, slow/cardiac (Myh7) but not myogenin (Myog) or myosin heavy chain 3, embryonic (Myh3). Values reflect the mean ± SEM (n = 11–14 samples/group). *P < 0.05, **P < 0.01, ***P < 0.001 by one-way ANOVA after log transformation.

Levels of Adh1 mRNA, documented as an androgen-sensitive gene [23], were increased by DHT at least 25- and 10-fold in the microarray and qRT-PCR experiments, respectively. The expression of Has2, known to be important for myogenic differentiation [32] and reduced in cremaster muscle of GU-ARKO mice with knockdown of Ar in the developing gubernacular mesenchyme [7], was increased 1.9-fold (array) and 2.5-fold (qRT-PCR) in response to 10 nM DHT exposure. Similarly, neuropeptide Y (NPY) expression in brain is correlated with androgen levels, and Npy expression is reduced in the GU-ARKO mouse. The Npy mRNA levels were increased 2.5-fold in the array experiments but only slightly increased by qRT-PCR in fetal gubernaculum after DHT exposure (Fig. 3).

We analyzed select developmental genes, including some belonging to Wnt and TGFB/BMP signaling pathways, and consistent with the array results, we observed marked upregulation of Crlf1 and Chrdl2 and reduced expression of Wnt4 and Bmp4 following DHT exposure. However, the 1.6-fold reduction of Wnt5a observed in the microarray experiment was not confirmed by a significant reduction in Wnt5a expression as determined by qRT-PCR (Fig. 3). The qRT-PCR data from our previous study [13] also showed marked upregulation of Crlf1 and Chrdl2 by INSL3, and in contrast to DHT experiments, Wnt4 and Wnt5a expression was also increased by INSL3 and mirrored the corresponding array data. The array data also show reduced Tgfb2 and Cxcl12 and increased Slit3 expression following exposure to either hormone; the present qRT-PCR data validate these results for DHT.

Although the array-based myosin isoform expression was not significantly changed, we also analyzed selected muscle-specific genes by qRT-PCR in view of a role for AR and RXFP2 in muscle development [7, 9], in altered muscle-specific genes in the cryptorchid ORL rat strain [30], and in cremaster muscle of boys with cryptorchidism [33]. We observed significantly reduced expression of cardiac/slow myosin (Myh7) after DHT exposure and reductions in the muscle differentiation gene Myog and embryonic myosin Myh3 that were not statistically significant (Fig. 3).

The array and qRT-PCR data show consistent results for the majority of genes, particularly those with the greatest magnitude of DHT response, and are consistent with prior data showing regulation of specific genes by androgens. The patterns of gene expression are also largely consistent with our previous analysis of untreated fetal gubernacula [30]. In those studies, we observed higher expression of Crlf1 (23.5-fold), Adh1 (5.8-fold), Wnt4 (4.5-fold), Bmp4 (2.4-fold), and Ar (2.4-fold) in male compared to female E18 gubernacula. In some cases, we also identified significant temporal changes in mRNA expression in male gubernacula between E18 and E19 that directly correlate with response to DHT: decreases in Wnt4 (5.3-fold) and Bmp4 (5.5-fold) and an increase in Adh1 (3.7-fold). Transcripts that are not upregulated in males until E20 (Has2, Tgfb2, Myh7, Myog, Myh3, and Sfrp2) were expressed at similar or higher levels in E18 females and showed a less robust or minimal response to DHT as determined by qRT-PCR (Fig. 3).

DISCUSSION

We performed the present gene expression experiments to complete studies that were designed to identify molecular patterns of hormonal signaling response during the rapid growth phase of the fetal gubernaculum. The present data suggest that androgens and INSL3 induce a qualitatively similar transcriptome response in the intact fetal gubernaculum in vitro when exposure occurs during its outgrowth phase, suggesting overlap in the developmental effects of these hormones. However, differentially expressed transcripts show a stronger link to specific developmental pathways and functional categories following INSL3 compared to DHT exposure. These findings are consistent with previous studies showing that both hormones induce cell division in the gubernaculum in vitro but that the proliferative effects of INSL3 appear more robust [3, 4]. However, our data suggest that even during this “androgen-sensitive programming window” [34], the gubernacular transcriptome remains more acutely responsive to stimulation by INSL3, and as circulating levels of this hormone peak in the male fetus during this time (Barthold et al., unpublished results), the demarcation between INSL3- and androgen-sensitive phases of testicular descent [35] may not be distinct. Nevertheless, because both Rxfp2^−/− and Ar^−/− mice are cryptorchid, these signaling pathways have unique roles in the developing gubernaculum despite the redundancy of hormonal response that we observe in the present studies at a single time point (E17). It is possible that the timing of peak responsiveness of the gubernaculum to each hormone differs in vivo.

Although we identified a robust transcriptome response of gubernacula to 24-h DHT exposure, no transcripts were differentially expressed after 6 h, which was unexpected based on existing studies on the timing of AR recruitment [36] and previous organ culture experiments that showed a similar response to INSL3 following 6- and 24-h exposures [13]. The exon array platform that we used for the 6-h experiment provided additional information regarding the gubernacular transcriptome, and we anticipated reproducibility between this and the 3′ IVT expression platform used for the 24-h INSL3 and DHT experiments [37]. We presume that the AR is present and functional in E17 gubernaculum for several reasons: 1) mRNA levels are easily measurable (Fig. 1); 2) AR protein is expressed primarily in mesenchymal cell nuclei at E16 to E19 [3, 6–8], coincident with the window of susceptibility to antiandrogen-induced cryptorchidism [34, 38–40]; and 3) targeted disruption of Ar in the developing fetal gubernaculum impairs its development [7]. Although a direct transcriptional response is expected by 6 h in androgen-responsive cell lines [36], factors that could account for the lack of statistically significant changes in gene expression at this time point may include reduced accessibility of AR-positive cells to DHT in this organ culture model and/or the possibility of an indirect transcriptional response to DHT. Cain et al. [40] reported that in vivo antiandrogen exposure at E15 to E17 is associated with reduced volume of both mesenchymal and muscular compartments of the gubernaculum but that these morphological effects do not occur until E20, suggesting a delayed secondary response. This type of indirect effect or “cascade” [41] may occur during androgen-regulated reproductive tract development via crosstalk with other gene regulatory networks or via autocrine or paracrine factors. Although the reason for the lack of response to 6-h DHT exposure cannot be determined from the present experiments, other data suggest the possibility that the transcriptional response to DHT is indirect. Our PAINT analysis did not show enrichment of AREs in the proximal promoter of transcripts responsive to DHT exposure in the fetal gubernaculum, yet these elements may exist further upstream [14] and may alter the timing of transcriptional response to AR activation [36]. Moreover, androgens may regulate gene expression via interaction with other transcription factors, as shown for serum response factor (SRF) in prostate cancer cells; these transcript expression changes required at least 8 h of DHT exposure to be measurable [42]. Notably, our PAINT analysis also showed enrichment of binding sites for ETS1 and ELK1, two transcription factors that also interact with AR to regulate expression of androgen-responsive genes in prostate cancer cells [43, 44] and could potentially function as coregulators in the gubernaculum.

Surprisingly, we observed statistically significant differences in expression (relative to the control group) for many more probe sets after 24-h exposure to 10 nM compared to 30 nM DHT, although for the majority of probe sets, the direction of change was consistent for both exposure levels. We also identified a large group of transcripts that showed a differential response to 10 and 30 nM DHT exposures but no significant difference between either DHT group and the unexposed control group. Our functional analysis identified overrepresentation of prostate cancer signaling (which in part involves AR) in this group. The reason for inconsistencies in the transcriptome response to DHT at the concentrations we employed remains unclear. However, our analysis of baseline and DHT-stimulated Ar mRNA levels (Fig. 1) shows marked upregulation of Ar mRNA expression in fetal gubernacula cultured without androgens. Additionally, Ar mRNA levels were reduced after DHT exposure (Fig. 1), whereas AR immunostaining was increased in fetal gubernacula exposed to R1881when cultured without a washout period [3]. These data suggest that AR mRNA and/or protein expression is androgen-sensitive in the fetal gubernaculum. Interestingly, the apparent association of Ar upregulation and increased sensitivity of cultured gubernacula to 10 nM DHT is in agreement with results of previous studies suggesting that overexpression of AR increases sensitivity to DHT in prostate cancer cells [15].

The overlap in transcriptome response to DHT and INSL3 in fetal rat gubernaculum and data obtained from molecular targeting approaches in mice support direct and/or indirect interactions between AR, INSL3/RXFP2, and Wnt signaling that promote muscle development and patterning within the enlarging fetal gubernaculum [3, 4, 7, 9, 45]. The overrepresentation of cAMP response element-binding protein (CREB) sites in our promoter analysis of DHT-responsive transcripts is of particular interest, because CREB-binding protein is involved in both AR and Wnt signaling [46, 47] and CREBBP mutations are associated with human cryptorchidism (Table 3). Because activation of the G protein-coupled receptor RXFP2 modulates cAMP levels in the gubernaculum, CREB signaling may link AR, INSL3/RXFP2, and Wnt pathways in the developing gubernaculum.

During normal development, both AR and RXFP2 are localized to the undifferentiated mesenchyme within the bulb, and levels decline as the cremaster muscle develops and the mesenchymal core regresses at and after birth [6, 7]. With knockdown of Rxfp2, Ctnnb1, or Notch1, the population of AR-positive cells is reduced, with evidence for failed migration of these cells into the gubernaculum, and normal gubernacular swelling and muscle patterning do not occur [9]. The fate of these AR-positive gubernacular cells remains unclear: they may migrate peripherally and contribute to the developing muscle layer, as suggested by some investigators [9, 48], and/or provide paracrine signals to the developing muscle, as suggested by Kaftanovskaya et al. [17] based on observations that the cremaster is thinner and relatively disorganized in the GU-ARKO mouse. Ectopic muscle within the mesenchymal core of the gubernaculum occurs in Rxfp2^−/− mice [9] and in fetal organs cultured with the synthetic androgen R1881 alone but not those cocultured with testis [3]. These observations suggest that INSL3/RXFP2 signaling stimulates migration of undifferentiated AR-positive cells into the gubernaculum and that their migration and/or functional activity produce normal patterning of the peripheral muscle layer. Our present functional analysis supports a role for INSL3-specific transcripts in cellular growth, metabolism, and migration. The available data therefore suggest both cooperative and unique aspects of hormonal signaling in the gubernaculum, with both hormones targeting developmental processes that culminate in muscle development. Whereas the phenotypic effects of loss of AR signaling are less severe, overexpression of INSL3 is unable to compensate for this loss [49].

We observed differential expression of many Wnt-associated genes after exposure to DHT and/or INSL3. The majority of transcripts were regulated similarly by DHT and INSL3 or were responsive to a single hormone, but expression levels of Wnt4, Wnt5a, and Wnt16—three of the most highly expressed Wnt ligands in the fetal gubernaculum—were increased and decreased after exposure to INSL3 and DHT, respectively, raising the possibility that these hormones target divergent Wnt pathways. The effects of Wnt signaling include regulation of cellular proliferation, differentiation, migration, and polarity and are dependent on ligand dosage and binding to coreceptors [50, 51]. Canonical and noncanonical Wnt pathways are not always distinct and may be active and/or competitive within cells [51]. Reduced β-catenin expression in gubernacular muscle of Rxfp2 hypomorphic fetal mice suggests that INSL3/RXFP2 regulates canonical Wnt signaling [9]. However, β-catenin may also mediate the effects of AR in muscle development [52] and masculinization of the external genitalia [53]. Classical noncanonical Wnt genes, including Wnt5a, its receptor Ror2, and Vangl2, are also associated with cryptorchidism and/or gubernacular maldevelopment [18], and INSL3 exposure is associated with increased Wnt5a expression [13]. However, the role of WNT5a/ROR2 in gubernacular development remains undefined. WNT5a/ROR2 mediate cytoskeletal organization and polarized cell migration via the Wnt/planar cell polarity signaling pathway, leading to developmental outgrowth of structures such as limbs and genital tubercle, but may also inhibit Wnt/β-catenin signaling [54]. Wnt5a mRNA is well expressed in the fetal gubernacular bulb based on our array and qRT-PCR data, but Chawengsaksophak et al. [55] reported absence of expression using whole-mount in situ hybridization at the same relative gestational age in mice [55]. However, Harisis et al. [56] recently confirmed that WNT5a and ROR2 proteins are expressed in the developing mesenchymal and muscle layers of the E17 gubernaculum and that ROR2 expression is unchanged in the rat E19 gubernaculum following exposure to the antiandrogen flutamide [56].

Combined deletion of Sfrp1 and Sfrp2, encoding two secreted frizzled-related proteins, also produces gubernacular maldevelopment, but the involved pathways remain unknown [18]. Secreted frizzled-related proteins were initially defined as Wnt inhibitors, but they may also positively regulate Wnt signaling by activating frizzled (Wnt) receptors [57] or by expanding the signaling domains of Wnt ligands [58]. Therefore, despite limited mechanistic insight, the available data suggest a role for both canonical and noncanonical Wnt signaling in gubernacular development.

Another potential role for Wnt, BMP, and/or AR signaling in the developing gubernaculum is regulation of muscle fiber type. The most abundant myosin heavy chain transcript in the fetal gubernaculum is the slow/cardiac isoform Myh7, and differentiated cremaster is composed primarily of type I (slow twitch) fibers [59, 60]. Evidence from chick embryos suggests that WNT4 positively regulates muscle proliferation and differentiation by antagonizing myostatin and also contributes to fast muscle fiber development, whereas WNT5a, β-catenin, and BMP4 are reported to promote a slow fiber type [61, 62].

Expression of several other transcripts involved in myogenesis is similarly increased (Crlf1 and Chrdl2) or reduced (Cxcl12 and Tgfb2) in fetal gubernaculum following in vitro exposure to DHT or INSL3 based on the present and prior [13] qRT-PCR and/or array data. Cxcl12 regulates migration and fusion of myogenic cells during development [63] and participates in ligand-independent activation of the AR in prostate cell lines [64]. Chrdl2 inhibits BMP activity, and although differentially expressed during myoblast differentiation, its role in that process remains unknown [65]. A corresponding increase in expression of the BMP antagonists Grem2 and Smoc1 in response to DHT exposure is consistent with a requirement for downregulated BMP activity during myogenesis [66].

In summary, our data suggest strong overlap in the in vitro DHT- and INSL3-responsive transcriptome of cultured fetal gubernaculum. We did not identify an acute transcriptional response to DHT, a finding not fully explored in the present experiments but consistent with other observations that suggest delayed and possibly indirect regulation of testicular descent by androgens. Wnt signaling is strongly overrepresented in the hormone-responsive transcriptome, but the most highly expressed Wnt ligands respond differently to INSL3 and DHT. These and previous data support a role for activation of more than one Wnt signaling pathway during gubernacular organogenesis. Unique patterns of gene expression may provide further insight regarding the process of testicular descent and may serve as useful markers in future studies that define the contribution of hormonal effects and genetic variants to gubernacular maldevelopment and susceptibility to cryptorchidism.