Hedgehog signaling regulates Wolffian duct development through the primary cilium

Abstract

Primary cilia play pivotal roles in embryonic patterning and organogenesis through transduction of the Hedgehog signaling pathway (Hh). Although mutations in Hh morphogens impair the development of the gonads and trigger male infertility, the contribution of Hh and primary cilia in the development of male reproductive ductules, including the epididymis, remains unknown. From a Pax2^Cre; IFT88^fl/fl knock-out mouse model, we found that primary cilia deletion is associated with imbalanced Hh signaling and morphometric changes in the Wolffian duct (WD), the embryonic precursor of the epididymis. Similar effects were observed following pharmacological blockade of primary cilia formation and Hh modulation on WD organotypic cultures. The expression of genes involved in extracellular matrix, mesenchymal-epithelial transition, canonical Hh and WD development was significantly altered after treatments. Altogether, we identified the primary cilia-dependent Hh signaling as a master regulator of genes involved in WD development. This provides new insights regarding the etiology of sexual differentiation and male infertility issues.

Modulation of primary-ciliogenesis and downstream Hedgehog signaling controls Wolffian duct development.

Introduction

The Wolffian duct (WD) originates from the mesonephros and is the embryonic precursor of the epididymis, a unique 6-m-long tubule in which spermatozoa acquire their motility and fertilizing capacity at the onset of puberty (reviewed by Hinton et al. 2011) [1]. In men, the proximal segment of the epididymis is mostly composed of efferent ductules, which are absorptive tubules indispensible for male fertility [2–4]. In mice, the most apical region of the WD forms the efferent ductules [5]. During embryonic development, the WD is subjected to profound morphological and molecular changes that convert a small linear duct into a long, convoluted and highly regionalized tubule [6]. Impairment of correct in utero WD morphogenesis is associated with structural and functional defects of the epididymis that eventually impede male fertility [7, 8]. As observed in other tubular structures, the elongation, coiling, and differentiation of the WD require a well-orchestrated interplay between epithelial and mesenchymal compartments [9, 10] mediated by several players, including androgens, inhibins (e.g. Inhba), transcription factors (e.g. Pax2 and Pax8), growth factors (e.g. Fgf8) and Wnt pathway components (e.g. Wnt5a and Wnt8b) (reviewed by Murashima et al. 2015) [7]. Although never explored, the importance of WD primary cilia to the transduction of these developmental pathways has been suggested based on the impairment of in utero WD coiling and development following epithelial-specific deletion of Pkd1 and Pkd2 ciliary components [11, 12].

The primary cilium is a non-motile microtubule-based organelle that extends from the surface of most cells [13], and is composed of a basal body to which the axonemal extension is anchored [14]. The formation of this sensory antenna is dynamic throughout the cell cycle and involves bidirectional intraflagellar transport (IFT) along the axonemal microtubule. While the membrane that surrounds the axoneme is continuous with the cell plasma membrane, it is composed of specific lipids and receptors that potentiate the response to extracellular factors, including to the morphogens Indian, Desert, and Sonic Hedgehog [15, 16] (IHH, DHH, and SHH, respectively) (reviewed by Garcia et al. 2018) [17]. In vertebrates, both activation of the Hedgehog (Hh) signaling pathway and its basal repression have been shown to depend on the primary cilium [18]. Although Hh signaling is essential for the development of diverse embryonic structures [19] and participates in epithelial–mesenchymal communication in fetal prostate and kidney [20, 21], SHH produced by the mesonephros is not required for correct WD development and differentiation [22]. However, a recent genetic study performed on a cohort of 430 male subjects indicated that mutations in the Dhh gene are associated with impaired development of the gonad, and male infertility issues [23] as well as with peritubar cell defects in the testis [24]. In addition, testicular tubule-like structures are formed in vitro under the control of the Hh signaling pathway [25]. These findings shed light on the possible involvement of other morphogens than SHH, i.e. IHH and/or DHH, in fetal male excurrent duct development.

Although primary cilia and the downstream Hh signaling pathway contribute to the homeostatic control of adult male reproductive tissues [26–28], their functions at early developmental stages of the male reproductive system remain poorly understood. Herein, to gain insights on mechanisms involved in higher mammal reproduction, we used mouse WD as a model system. By using a combination of in vivo/in vitro approaches, we tested the hypothesis that the primary cilia-dependent Hh pathway controls the regulation of WD/epididymis morphogenesis. Following the localization of primary cilia and Hh components in the developing WD from Arl13b-Cetn2 and Pax2^Cre; IFT88^fl/fl mice, ex vivo organotypic cultures of embryonic murine WD were treated with a blocker of primary ciliogenesis as well as Hh agonist and inhibitor to elucidate the role and the molecular response associated with primary cilia and Hh signaling during WD growth. Together, our results shed light on the role of WD primary cilia as mediators of the canonical Hh pathway, and as a master player in the control of WD mesenchymal–epithelial cell communication and development.

Material and methods

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Clémence Belleannée (Clemence.Belleannee@crchudequebec.ulaval.ca).

Materials availability

This study did not generate new unique reagents.

Data and code availability

Raw data are freely available from the Gene Expression Omnibus (GEO) repository (GSE145816). Any additional information required to re-analyze the data reported in this paper is available from the lead contact upon request.

Experimental model and subject details

Animals and isolation of WDs

All animal experiments were performed with approval from the ethics committee of the Institutional Review Board of the Centre Hospitalier Universitaire de Québec, and were conducted in compliance with the requirements defined by the Guide for the Care and Use of Laboratory Animals. Mice were bred and housed in the Specific Pathogen Free Animal Facility at the CHU de Quebec Research Center—Université Laval. Unless otherwise stated, the chemicals and reagents used in the study were purchased from Sigma-Aldrich (St Louis, Missouri, USA).

Embryos were collected after euthanasia of Tg (CAG-Arl13b/mCherry)1Kvand Tg (CAG- EGFP/CETN2) 3-4Jgg/KvandJ (Jackson Laboratory, stock #027967, referred to as Arl13b-Cetn2 in the manuscript), Pax2^Cre; IFT88^fl/fl (Generous gift from Dr Jinwoong Bok, Moon et al., 2020) [29], and CD-1 (Charles River, stock #022) pregnant mice at 16.5 days post-coitum (E16.5). After sexing the embryos, WD were isolated from E16.5 male embryos dissected free from testis, and fixed, snap-frozen, or cultured.

WDs from Arl13b-Cetn2 tg male embryos were used to localize primary cilia and Hh signaling components in the WD and to test the effect of ciliobrevin D (CilioD) on primary cilia formation in vitro. Wolffian ducts from Pax2^Cre; IFT88^fl/fl male embryo (IFT88 conditional knock-out (cKO)) and littermate controls (Ctrl) were used to assess in vivo effect of primary cilia deletion on WD features. Finally, WDs from CD-1 male embryos were used to assess the effect of Hh signaling on WD development using smoothened agonist (SAG) or cyclopamine (Cyclo). For all the models, we performed morphometric and molecular analysis (Supplementary Figure S1). For organotypic culture [1, 6], whole WD were dissected and placed on a 0.4-mm ThinCert^® tissue culture insert (Greiner Bio-One, Kremsmünster, Austria) in DMEM/F12 medium supplemented with 50 μg mL⁻¹ ampicillin, 1% ITS (insulin, transferrin, and selenium) and 10⁻⁸ M testosterone for 72 h at 37°C and 5% CO₂.

Methods details

Tissue fixation

WDs were washed three times in phosphate buffered saline (PBS; 137-mM NaCl, 3-mM KCl, 8-mM Na₂HPO₄ , and 1.5-mM KH₂PO₄) and fixed in periodate lysine paraformaldehyde 4% (PLP) for 2 h at room temperature while protected from light. Then, WD were washed three times in PBS and embedded in paraffin or cryoprotected overnight in 30% sucrose in PBS, placed in a plastic insert, embedded in Tissue-Tek O.C.T. Compound (Sakura Finetek, USA) and snap-frozen.

Immunofluorescence staining

For paraffin sectioning, 4-μm-thick WD sections were cut and a deparaffinization step in a xylene and ethanol bath was followed by an antigen retrieval in citrate buffer (10-mM sodium citrate, 0.05% Tween 20, pH 6) at 110°C for 10 min. For O.C.T.-embedded sections, 10-μm-thick WD sections were cut on a cryostat (Shandon Cryotome, ThermoFisher, USA) and collected on Superfrost-Plus slides (Superfrost Fisherbrand, ThermoFisher, USA). After rehydration, both paraffin and O.C.T.-embedded sections were treated with 1% (w/v) sodium dodecyl sulfate (SDS) and 0.1% (v/v) Triton X-100 in PBS for 4 min at room temperature and washed with PBS for 5 min. After blockade in 1% (w/v) bovine serum albumin in PBS for 15 min, the primary antibody was diluted in DAKO solution (DAKO Corporation, USA) and incubated overnight at 4°C. The primary antibodies were used as follows: alpha smooth muscle actin (αSMA; 1:100; ab5694, Abcam), Indian Hedgehog (IHH; 1:100; ab52919, Abcam), IFT88 (IFT88; 1:200; Rabbit; 13967-1-AP, Proteintech), gamma tubulin (γ-tub; 1:200; Mouse IgG1; ab27074, Abcam) and alpha acetylated tubulin (α-acetylated tubulin; 1:500; Mouse IgG2b; T7451, Sigma). Sections were washed twice in high-salt PBS (2.7% NaCl) for 5 min and immersed in PBS for 5 min. The secondary antibodies were then applied for 1–2 h at room temperature. The secondary antibodies used were as follows: goat anti-rabbit (GAR) and donkey anti-goat (DAG) in fluorescence 647 (1:1000; Jackson), goat anti-mouse (IgG1 488; 1:500; A21121, Invitrogen), and goat anti-mouse (IgG2b 568; 1:500; A21144, Invitrogen). For visualization of primary cilia in WD from Arl13b-Cetn2 mice, slides were directly mounted in Vectashield medium (DAPI; Vector Laboratories, Inc, Burlingame, CA) for imaging after rehydration for 15 min in PBS.

Confocal imaging

Image acquisitions were performed on an inverted Olympus IX80 confocal microscope operated with a WaveFX-Borealin-SC Yokagawa spinning disk (Quorum Technologies) and equipped with an Orca Flash 4.0 camera (Hamamatsu) controlled by MetaMorph software (Molecular Devices, Quorum WaveFX v7.8.4.0, Sunnyvale CA, USA). Immunostainings of WD from Pax2^Cre; IFT88^fl/fl embryos were aquired on an inverted LSM900 Zeiss confocal microscope operated with an Airyscan2 technology (Zeiss technologies) and controlled by Zen3.2 software. The images were obtained under 100× magnification using an optical Z-section for each channel, and were projected onto a single picture using the Z-project tool from ImageJ software.

WD 3D reconstruction

Detection of primary cilia was performed using 3D reconstruction. Wolffian ducts from E16.5 Arl13b-Cetn2 embryos were fixed in 4% PLP for 2 h at room temperature, and washed three times in PBS. WD was imaged with an Olympus IX81-FV1000 confocal microscope equipped with a UPLsAPO 30XS NA 1.05 lens. Fluorescence from GFP and mCherry was acquired in 3D (voxel size was 0.265 μm × 0.265 μm × 1.1 μm). Reconstruction of primary cilia and WD was performed in Mathworks Matlab^® 2018 using the image processing toolbox. Surface meshes were exported in object format, then animation and scene lightening was performed using Autodesk 3DS max 2019. Rendering was done with Mental Ray renderer.

Primary cilia measurements

Primary cilia length was measured from 20-μm Z-stack confocal projections using the straight-line tool from ImageJ software. Epithelial primary cilia density was determined by counting the number of primary cilia at the luminal edge of the WD and reporting it relatively to the circumference of the tubule, expressed as the number of primary cilia per μm of apical membrane. The same individual performed all measurements. Statistical Analysis System (SAS) software was used to perform Analysis of Variance (ANOVA) using the mixed procedure and Tukey’s test. The Shapiro–Wilk test was performed to evaluate the normality of data. Statistical significance was determined as P < 0.05.

Protein extraction and western blotting

Two pairs of snap-frozen WD were incubated in 100-μL RIPA buffer (150-mM NaCl, 50-mM Tris, 0.1% SDS, 1% Triton, 0.5% deoxycolate, 1-mM EDTA, pH 7.4) in the presence of a protease inhibitor cocktail (EDTA-free Protease Inhibitor Cocktail, Roche) and then ground with a plastic pestle in a 1.5-mL microtube. Debris were removed by centrifugation (10 000 g for 15 min at 4°C). The supernatant was transferred to a fresh tube and proteins were quantified by Bradford assay [30].

Protein extracts were denatured and reduced by boiling in Laemmli sample buffer containing 2% β-mercaptoethanol at 90°C for 5 min. A total of 15 μg of proteins isolated from WD were loaded onto a 10% polyacrylamide gel for SDS-PAGE. Protein transfer to nitrocellulose membranes (0.45 μm, BioRad) was performed using the Trans-Blot Turbo system (BioRad). Membranes were then blocked in 5% milk diluted in PBS containing 0.5% Tween 20 for 1 h at room temperature. Membranes were then incubated overnight at 4°C with primary antibody diluted in 5% milk in PBS (IHH, 1:1000; αSMA, 1:500; vimentin, 1:1000; β-actin, 1:5000, A5441, Sigma). In one set of experiments, a competitive assay was performed with a recombinant protein for IHH (ab218095, Abcam) pre-incubated with the anti-IHH antibody (10:1 ratio) at room temperature for 1 h prior to blotting the membrane.

Membranes were washed three times in PBS with 0.5% Tween 20 and incubated with the secondary antibody (GAR 1:5000; 111-035-045, Jackson Immunoresearch; GAM, 1:5000, 115-035-062, Jackson Immunoresearch) for 1 h at room temperature. After three washes in PBS with 0.5% Tween 20, antibody binding was detected by revealing the membrane with Clarity or Clarity Max Western ECL substrate (BioRad) on a ChemiDoc MP Imaging System (BioRad). Protein quantification was performed by measuring the band volume intensity with ImageLab system (BioRad) with normalization to β-actin expression. Experiments were performed on three biological replicates per treatment.

In situ RNAScope hybridization

RNAscope Multiplex Fluorescent Reagent Kit v2 was used to perform the experiments following the manufacturer’s protocol (Advanced Cell Diagnostics). Paraffin-embedded tissue sections of 5 μm were incubated with Protease Plus for 30 min at 40°C. After washes, Mm-ptch2 (435131), Mm-hhip (448441), 3-plex positive (320881) and negative control probes (320871) were added on the tissues and the slides were incubated in a humid incubator at 40°C for 2 h. Slides were washed and kept in a saline sodium citrate solution (0.75-M NaCl, 0.380-M sodium citrate, pH 7.0) overnight. After washes, slides were successively incubated with Amp 1, Amp 2, and Amp 3 solutions for 30, 30, and 15 min, respectively, at 40°C. Tissues were washed between each step. Each probe signal was revealed in its corresponding channel (C1 or C2 depending on the probe). Tissue sections were incubated with HRP-C1 or HRP-C2 for 15 min at 40°C, washed and incubated with the diluted fluorophore, Opal 520 or Opal 570, for 30 min at 40°C. Slides were washed and incubated with the HRP blocker for 15 min at 40°C. Signal development was performed the same way for the other probe. After several washes, the slides were mounted with DAPI, dried and stored at 4°C until imaging. Dots for each probe were counted both for epithelium and mesenchyme and reported on the number of cells counted. The proximal region of three WD were used in this analysis. The statistical analysis was performed by a two-tailed unpaired t-test and significance was determined as P < 0.05.

Organotypic cultures of WDs

For organotypic culture, whole WD were washed with PBS following dissection and placed on a 0.4-mm ThinCert^® tissue culture insert (Greiner Bio-One, Kremsmünster, Austria) in DMEM/F12 medium supplemented with 50 μg mL⁻¹ ampicillin, 1% ITS (insulin, transferrin, and selenium), and 10⁻⁸ M testosterone for 72 h at 37°C and 5% CO₂. Wolffian ducts were treated for 72 h with ciliobrevin D (CilioD), smoothened agonist (SAG), or cyclopamine (Cyclo). For each of these treatments, dose-response studies and wash-out recovery tests were performed to select drug concentrations that were reversible and not detrimental to WD tissues (Supplementary Figures S2, S3, S4, S5, S6, and S7).

CilioD is an inhibitor of the ATPase motor cytoplasmic dynein [31] and was used to assess the contribution of primary cilia to Hh signaling in the WD by preventing ciliogenesis. Organotypic WD cultures were treated with 20-μM CilioD diluted in DMSO (0.3% in the final medium) in the presence or absence of SAG to evaluate the contribution of primary cilia to Hh signaling in the WD. CilioD control received 0.3% of DMSO in the final medium. WD cultures received one of four different treatments, as follows: (1) control condition (no SAG/no CilioD in culture medium; n = 6), (2) 0.5-μM SAG (n = 6); (3) 20-μM CilioD (n = 6); and (4) 0.5-μM SAG and 20-μM CilioD (n = 6).

SAG acts as an activator of the G protein-coupled receptor Smoothened, and was used to activate the Hh signaling pathway by circumventing the initial step of Hh morphogens–PTCH1 receptor binding. The effect of SAG on WD development was investigated as follows: WD were isolated from E16.5 embryos and cultured in control medium (n = 14) or in the presence of 0.5-μM SAG (n = 17).

Cyclo is a steroid alkaloid that inhibits the Hh signaling pathway through direct binding to Smoothened. To assess the effect of Cyclo on WD development, WD were isolated from E16.5 embryos and cultured in control medium (n = 11) or in 25-μM Cyclo diluted in ethanol (0.2% in final medium) (n = 11). Cyclo control received 0.2% of ethanol in the final medium.

Images of WD were captured under 4× magnification on an inverted microscope (EVOS™ XL Core, Invitrogen, USA) before and after each culture session. At the end of the treatments, WD were either fixed for subsequent immunofluorescence staining or immunohistochemistry, or snap-frozen for RNA or protein extraction.

Measurement of WD morphological features

Wolffian ductal image acquisitions were analyzed with ImageJ [32] to determine WD morphological features, including elongation, number of coils, extraluminal thickness, and luminal area/length. After image calibration, elongation/length of WD was measured by drawing a freehand line in the WD lumen taking its origin from the extremity of the bladder up to the extremity of the testis. The number of coils was determined by counting the number of bends forming an angle of ≤60°, as previously described [33]. The luminal area was determined using polygon selection and fit spline tool. Lastly, the extraluminal thickness was determined by measuring the mean of the distance of five straight lines from the external border of the duct to the edge of the extraluminal space. For characterization of WD from Pax2^Cre; IFT88^fl/fl embryos, three and five WD for Control and cKO, respectively, were analyzed. For pharmacological treatments, measurements were taken individually on each side (right and left) of the WD and the mean was calculated for each WD pair in six replicates organotypic cultures sets. For both experiments, the same individual performed all measurements in a blind manner. ANOVA was performed using the mixed procedure of the SAS software. A Shapiro–Wilk test was performed to evaluate the normality of data. Statistical significance was determined as P < 0.05.

Time-lapse of developing WD

Time-lapse imaging of WD organotypic cultures was performed over a period of 46 h. Wolffian ducts from six CD-1 E16.5 embryos were cultured individually (n = 1) under different conditions as previously described for one of each of the organotypic culture treatments described above: (1) CilioD 20 μM, (2) SAG 0.5 μM, (3) SAG 0.5 μM and CilioD 20 μM, (4) Cyclo 25 μM, (5) Cyclo control, and (6) SAG 0.5 μM and CilioD 20 μM control. Individual images from each WD were obtained every 10 min over 46 h using a Zeiss Axio Observer Z1 equipped with a 2.5×/0.12 objective and a AxioCam MRm3 camera in brightfield illumination (Carl Zeiss Canada Ltd, Toronto, ON). Wolffian ducts were placed in a culture dish under a Pecon PMS1 stage top incubator inside a Pecon XL large incubator (also available from Zeiss) at 37°C and 5% CO₂. Images were treated using ImageJ to remove background, enhance contrast, and apply a gamma correction at 0.6.

Immunohistochemistry staining and image acquisitions

Following deparaffinization, tissue sections were subjected to antigen retrieval in citrate buffer (10 mM, pH 6) for 10 min at 110°C. Endogenous peroxidase activity was quenched in 3% H₂O₂ in methanol (v/v) for 10 min. Sections were washed for 5 min in 95% ethanol and 5 min in PBS. Blockade of non-specific binding sites was performed in 10% goat serum in PBS for 1 h, followed by overnight incubation at 4°C with the primary antibody diluted in DAKO solution (DAKO Corporation, USA). The primary antibodies used were as follows: αSMA (1:1000; ab5694, Abcam), vimentin (1:500, ab20346, Abcam), and KI-67 (1:200, SolA15, Biosciences). Sections were then incubated with biotinylated GAM (1:200, 115-065-062, Jackson Immunoresearch), GAR (1:200, 111-065-144, Jackson Immunoresearch), or rabbit anti-rat (RARat, 1:200, BA-4001, Vector) antibodies for 1 h and with ABC elite reagent (Vector Laboratories, Burlingame, CA, USA) for 30 min. Negative controls were performed in the absence of primary antibodies. Immunostaining was revealed with 3-amino-9-ethylcarbazole (AEC), and Mayer’s haematoxylin solution was used to counterstain the tissues.

Image acquisitions were performed using QCapture Pro (Qimaging Instruments, Teledyne Photometrics, Tucson, USA) on an epifluorescence microscope (Zeiss Axioskop2 Plus, Oberkochen, Germain). For KI-67, positive cells were quantified in ImageJ from the control condition (no SAG in culture medium; n = 4) and 0.5-μM SAG (n = 7), from three experimental replicates. The same individual performed all measurements in a blind manner. Proliferation in αSMA+ cells was assessed from serial sections. ANOVA was performed using the mixed procedure of the SAS software. A Shapiro–Wilk test was performed to evaluate the normality of data. Statistical significance was determined as P < 0.05.

RNA extraction, purification, and microarray analysis

One pair of snap-frozen WD was incubated in 200 μL of RLT lysis buffer (Qiagen) containing 10% β-mercaptoethanol and then ground with a plastic pestle in a 1.5 mL microtube. Total RNA was then purified by RNeasy Mini Kit (Qiagen) following the manufacturer’s protocol, with the addition of an RNase-free DNase (Qiagen) step to avoid genomic contamination. To quantify and assess the quality of the purified RNA, samples were analyzed on a NanoDrop 1000 and a Bioanalyzer. The RNA Integrity Number (RIN) obtained for the different samples ranged from 9.2 to 10.

Total RNA extracted and purified from WD collected on E16.5 WD cultured for three days (72 h) was subjected to microarray analysis and compared by ANOVA according to the treatment received: control (n = 3), SAG 0.5 μM (n = 4), and Cyclo 25 μM (n = 3). The analysis was performed on Affymetrix Mouse Clariom S arrays (ThermoFisher) following the Affymetrix standard protocol. In brief, a total quantity of 100-ng RNA was labeled using the GeneChip WT Plus Reagent kit, hybridized to the arrays with the hybridization cocktail, and incubated overnight at 45°C with rotation in a hybridization oven. The cocktail was then removed and the arrays were washed and stained in an Affymetrix GeneChip Fluidics Station. The arrays were scanned with Affymetrix GCS 3000 7G and Gene-Chip Command Console Software to produce the probe cell intensity data (CEL). Affymetrix Expression Console Software was used to perform quality control, background subtraction and normalization of probe set intensities using Robust Multiarray Analysis. The microarray analyses were performed by the Gene Expression Core facility of the Genomic platform of the Centre Hospitalier Universitaire de Quebec Research Center.

Quantitative PCR

After RNA purification and quantification, 1 μg of total RNA was reverse transcribed (RT-PCR) with the iScript Advanced cDNA Synthesis Kit (Biorad) using random primers following the manufacturer’s protocol. cDNA samples were used to perform real-time quantitative PCR (qPCR) using the SsoAdvanced Universal SYBR Green Supermix (BioRad) on a C1000 Thermal Cycler (Biorad). For qPCR, 0.5-μM specific forward and reverse primers for the Hhip, Gli1, Ptch2, and Inhba genes (Supplementary Table S1) and 2-μL cDNA samples were added to 10-μL wells. An RT-negative control and a no-template control were included. In brief, the reaction consisted of incubating the samples at 95°C for 5 min followed by 40 cycles of three amplification steps: 95°C for 15 s, the optimal primer-specific temperature (between 54 and 66°C) for 15 s and elongation at 72°C for 15 s. Samples were then heated from 65 to 95°C with a temperature change rate of 0.5°C per 0.5 s to generate a melting curve. The optimal annealing temperatures and primer efficiencies were determined for each set of primers (Supplementary Table S1). To confirm the specificity of amplification, the amplified products were resolved on a 2% agarose gel and sequenced. Each RT-qPCR reaction was performed as two technical replicates for each biological sample, and then normalized to the geometric mean of two reference genes: Gapdh and Vcl (Supplementary Table S1). Results were analyzed by calculating relative gene expression data while accounting for differences in primer efficiencies [34]. Experiments were performed on three biological replicates to validate differentially expressed genes according to the results of the microarray analysis.

Bioinformatics

Microarray analysis was performed on CEL archives with the TAC 4.0 analysis suite from Affymetrix. An RMA approach was used to normalize the data [35]. Values were log2-transformed to obtain a normal distribution. Differential expression analysis was subsequently carried out using the method implemented in LIMMA [36]. ANOVA analysis was performed by accounting for the contrasts among the three studied groups (Cyclo vs. Control; SAG vs. Control and SAG vs. Cyclo). A statistical significance threshold of a log2 fold change (FC) > 1.5 and a P-value corrected by false discovery rate (FDR) ≤ 0.03 was applied. The genes modulated in each contrast were systematically studied with five functional analysis suites (DAVID, STRING, GOrilla, Metscape, GSEA, and Blast2GO) [37–43].

In silico analysis of GLI-binding sites

The promoters of the differentially expressed genes found in the microarray were analyzed to determine the potential direct targets for GLI factors. Promoters were defined at ±3 kb from the transcription start site of the gene based on the mm10 version of Mus musculus genome (obtained from the TxDb.Mmusculus.UCSC.mm10.knownGene package (v1.4.0)). The two consensus sequences of the GLI transcription factors identified in the literature [44], i.e. 5’-TGGGTGGTC-3′ and 5’-GACCACCCA-3′, were searched exclusively in the promoter regions using the vcountPattern from the BSgenome package in R (v3.6.0).

Quantification and statistical analysis

All data are presented as the mean ± SEM. For tissues and structures mensurations, statistical analyses were performed first using a Shapiro–Wilk test to evaluate the normality of data. Afterwards, ANOVA was performed using the mixed procedure of the SAS software with Tukey post-hoc test when appropriate. For molecular analysis, statistical analysis was performed by ANOVA or t-test. Statistical significance was determined as P < 0.05.

Results

Primary cilia and the Hedgehog morphogen IHH are present in E16.5 WDs

To locate primary cilia in the different tissue compartments of the developing WD, we used an Arl13b-Cetn2 transgenic mouse model in which the centriolar Centrin2 and axonemal Arl13b proteins are associated with fluorescent tags [45]. Primary cilia were detected at the apical pole of epithelial cells (see Supplementary Video S1) and were found associated with mesenchymal cells from the proximal, median, and distal regions of WD from 16.5 days post-coitum (E16.5) of Arl13b-Cetn2 mouse embryos (Figure 1A–C). While shorter primary cilia were found in the distal WD (1.80 ± 0.06 μm) compared with the proximal (2.02 ± 0.07 μm) and median (2.00 ± 0.06 μm) segments (Figure 1D), the density of primary cilia located at the apical pole of epithelial cells tended (P = 0.06) to decrease from the proximal to the distal regions (Figure 1E). Furthermore, immunofluorescence staining for the αSMA protein, a marker of smooth muscle differentiation that is absent until E14.5 [21], indicated that primary cilia were ubiquitously found in mesenchymal stromal cells (MSC) and MSC-derived smooth muscle cells at E16.5 (Figure 1F–G).

Figure 1.

Primary cilia extend from epithelial and mesenchymal cells of the developing Wolffian ducts (WD) and display region-specific features. Detection of primary cilia was done in 16.5-days post-coitum embryonic (E16.5) WD and testis from Arl13b-Cetn2 transgenic mice. (A) Epifluorescence microscopy image of the embryonic testis, and proximal, median, and distal regions of the WD. (B) Confocal microscopy image of primary cilia components (axoneme and basal body) in mesenchymal cells of WD. Scale bar: 2 μm. (C) Confocal microscopy images of primary cilia located in epithelial (Ep) and mesenchymal cells (Mes) of the proximal, median, and distal WD, as well as in seminiferous cord (SC) and interstitial cells (In) of the testis. Lu: lumen. Scale bar: 10 μm. (D) Length of primary cilia (μm) in the proximal, median, and distal regions of the WD. (E) Number of primary cilia per μm of the apical membrane in the proximal, median, and distal regions of the WD. (F–G) Confocal microscopy images of αSMA⁺ (smooth muscle alpha-actin positive) and αSMA^−- (smooth muscle alpha-actin negative) mesenchymal cells in WD from Arl13b-Cetn2 transgenic mice. Scale bar: 10 μm. (g,g’,g”) High magnification sections showing primary cilia in epithelium, αSMA⁺, and αSMA⁻ mesenchyme compartments. Scale bar: 6 μm. Ep: epithelial cells. Mes: mesenchymal cells. Lu: lumen. Primary cilia are indicated by arrows. The negative control is shown in Supplementary Figure S11. ^* indicates a statistical significance of P < 0.05 between the regions (proximal vs. median vs. distal). ns indicates non-statistical significance. All quantitative data are presented as means and standard errors of the mean (SEM).

Since primary cilia mediate the canonical Hh signaling pathway in vertebrates in response to Hh morphogens (SHH, IHH, and DHH) [18, 46, 47], we investigated the presence of Hh signaling ligands in the WD from E16.5 Arl13b-Cetn2 and CD-1 mouse embryos. Considering that both IHH and SHH were detected in the adult epididymis [27, 28], we investigated the presence of these two Hh morphogens in the WD from E16.5 embryo. Although SHH was not detected at the protein level in our experimental conditions (not shown), IHH was detected and displayed a partial overlap with the Arl13b-positive axoneme of primary cilia from the proximal WD region (Figure 2A, and insets a, a’). Whereas IHH was detected in all segments of the E16.5 WD (Figure 2A–E), the signal was not overlapping with Arl13b-positive primary cilia in the median (Figure 2B, and insets b, b’) and distal (Figure 2C, and insets c, c’) segments. Although these observations indicate that IHH and primary cilia are predominant in the proximal region where they may be more potent, we have considered the WD as a whole to investigate the role of Hh signaling and primary cilia in its development.

Figure 2.

Indian Hedgehog (IHH) protein is detected in 16.5 days post-coitum embryonic (E16.5) WD and partial overlap the signal with Arl13b-positive primary cilia in the proximal segment. (A–D) Immunofluorescent staining of IHH in proximal, median and distal WD regions from Arl13b to Cetn2 transgenic mice. Ep: epithelial cells. Mes: mesenchymal cells. Lu: lumen. Scale bar: 10 μm. (a–c; a’–c’) High magnification section showing IHH in primary cilia in proximal, median, and distal WD of the squares from (A), (B), and (C). Primary cilia and IHH partial overlapping signals are indicated by arrows. Scale bar: 2 μm. (E) Western blot detection of IHH protein from E16.5 intestine (positive control), E16.5 WD and adult epididymis (Epi.) protein extracts. On the right, competitive antibody binding assay was performed in the absence (−IHH_rec) or presence (+IHH_rec) of IHH recombinant protein prior to Western blot on 10 weeks old epididymis extracts. The whole membranes and β-actin controls are shown in Supplementary Figure S12.

Impairment of primary cilia formation in vivo and in vitro has an impact on WD morphological features and Hedgehog signaling

To investigate the contribution of primary cilia organelles to WD development, we first examined WD morphological features from E16.5 Pax2^Cre; IFT88^fl/fl (IFT88 cKO) embryo in which ciliogenesis is impaired [29]. In this model, IFT88 anterograde transport component [48] has been conditionally invalidated in epithelial cells positive for Pax2 [49], whose expression occurs as early as E13.5 in the epithelium of the WD [12]. In E16.5 IFT88 cKO embryos, primary cilia were specifically deleted in the WD epithelium, but not in the mesenchyme (Figure 3A and B, and insets a, a’, b, b’). In addition, WD from IFT88 cKO embryo displayed enlarged lumens and thicker αSMA-positive (αSMA⁺) and αSMA⁻negative (αSMA) mesenchymal cell layers (Figure 3C). Although Ptch2 and Hhip expression was significantly upregulated in the mesenchyme of IFT88 cKO WD compared with controls, Hhip displayed a decreased expression level in the proximal epithelium of IFT88 cKO WD (Figure 3D and E), as shown by RNAscope studies. These results suggest that primary cilia found in WD epithelial cells, at least in the proximal WD, are required for normal Hh signaling transduction and tissue patterning in vivo.

Figure 3.

Primary ciliogenesis defect observed in Pax2^Cre-IFT88^fl/fl mice is associated with abnormal patterning and impaired Hedgehog signaling in 16.5-days post-coitum embryonic (E16.5) WD. (A–B) Immunofluorescent staining of α-acetylated tubulin and gamma tubulin in WD of control and IFT88 cKO mice. Ep: epithelial cells. Mes: mesenchymal cells. Lu: lumen. Scale bar: 10 μm. (a–b) High magnification section showing primary cilia in WD mesenchyme from (A) and (B). Primary cilia are indicated by asterisks. Scale bar: 6 μm. (a’–b’) High magnification section showing primary cilia in WD epithelium from (A) and (B). Primary cilia are indicated by asterisks. Scale bar: 6 μm. (C) Lumen area (arbitrary unit) and thickness (arbitrary unit) of αSMA⁺ (smooth muscle alpha-actin positive) and αSMA⁻ (smooth muscle alpha-actin negative) mesenchymal cells were measured in E16.5 WD from control and IFT88 cKO from serial immunohistochemistry stainings. (D) RNAscope in situ hybridization of Ptch2 and Hhip in WD of control and IFT88 cKO. Ep: epithelial cells. Mes: mesenchymal cells. Lu: lumen. Scale bar: 20 μm. (E) Relative intensity of Ptch2 and Hhip in epithelial and mesenchymal cells WD from control and IFT88 cKO mice. The negative control is shown in Supplementary Figure S13.

The contribution of primary cilia to WD development and signaling was further investigated in ex vivo WD organotypic cultures from Arl13b-Cetn2 E16.5 embryos (in vitro model). Following the pharmacological blockade of primary ciliogenesis with CilioD, a specific blocker of AAA+ ATPase motor cytoplasmic dynein [31], the number of primary cilia present at the apical pole of epithelial cells as well as the length of primary cilia found in all cells from the WD were significantly reduced (Figure 4A and B). This size reduction was also observed after treatment with CilioD in presence of the smoothened agonist SAG, an activator of the Hh signaling pathway. To determine if the presence of primary cilia was a requisite to proper Hh signaling in the developing WD, Hh activation was induced on CilioD-treated WD. CilioD pharmacological concentration was determined by drug response/recovery approach on WD ex vivo cultures (see Supplementary Figures S2 and S3).

Figure 4.

The presence of primary cilia is necessary to ensure WD responsiveness to Hedgehog agonist ex vivo. Detection of primary cilia and Hedgehog signaling factors was performed on organotypic cultures of 16.5 days post-coitum embryonic (E16.5) WD from Arl13b to Cetn2 transgenic mice. (A) Detection by confocal microscopy of primary cilia in epithelial (Ep) and mesenchymal cells (Mes) from WD cultured for 72 h in the presence of smoothened agonist (SAG, 0.5 μM), CilioD (20 μM), or both (SAG 0.5 μM + CilioD 20 μM). Lu: lumen. Scale bars: 10 and 2 μm. The squares signalized the primary cilia magnified on right panel. (B) Length of primary cilia (μm) and number of primary cilia per μm of the apical membrane measured in 20-μm-thick sections of E16.5 WD cultured for 72 h under Control, SAG, CilioD or SAG + CilioD conditions. (C) Light microscopy images of E16.5 WD cultured for 72 h under Control, SAG, CilioD or SAG + CilioD conditions. Scale bar: 1 mm. (D) Elongation (mm), extraluminal thickness (mm), and Gli1 and Hhip relative expression levels determined by qPCR from E16.5 WD cultured for 72 h under Control, SAG, CilioD, or SAG + CilioD conditions. Different letters indicate statistical differences of P < 0.05 between treatments. All quantitative data are presented as the mean and standard errors of the mean (SEM).

With regard to WD morphological features, both WD elongation and extraluminal thickness were significantly reduced when primary ciliogenesis was impaired following CilioD treatment (Figure 4C). The activation of Hh signaling by SAG did not conteract the impaired development observed in absence of primary cilia (CilioD). This suggests that, in line with the phenotype observed in vivo, the presence of primary cilia organelle is a requisite to proper WD development. Furthermore, although the expression of both Gli1 and Hhip Hh target genes was significantly increased following Hh signaling activation by SAG (Figures 4D), impairment of ciliogenesis by CilioD prevented the transcriptional activation of these two genes, suggesting that normal primary ciliogenesis is required for the proper transduction of Hh signaling in the developing WD. This ex vivo organotypic model thus mimics, at least in part, the responsiveness of the developing WD to primary cilia-dependent Hh signaling observed in vivo.

The Hh signaling pathway controls WD elongation, coiling, and mesenchymal space extension ex vivo

To further scrutinize the effect of Hh signaling on WD development, organotypic cultures of WD isolated from CD-1 mice were incubated with either SAG (Hh agonist), or cyclopamine (Cyclo; Hh inhibitor) for 72 h, with subsequent analyses at the morphological, biochemical, and molecular levels. Dose-response assays performed on WD with concentrations ranging from 0.5 to 2 μM (SAG) and 12.5 to 100 μM (Cyclo) indicated that 0.5-μM SAG and 25-μM Cyclo were sufficient to induce WD morphological changes (Supplementary Figures S4 and S5). Furthermore, recovery assays performed 72 h after these treatments indicated that WD development recovered following SAG or Cyclo removal, thereby demonstrating that the treatment conditions were reversible and not detrimental to WD tissues (Supplementary Figures S6 and S7).

WD elongation, the number of coils and intraluminal area were significantly reduced following 72 h of SAG treatment compared to control (Figure 5A and B), suggesting a contribution by Hh signaling to correct WD morphogenesis. These features were associated with significant increase of the extraluminal thickness (Figure 5A and B), possibly due to the increase in vimentin-positive mesenchymal cells following SAG treatment (Figure 5A and C). The equivalent αSMA levels detected following SAG and control conditions suggested that the αSMA-positive smooth muscle cells layer remained unchanged following the activation of the Hh pathway (Figure 5C). Contrasting effects were observed on WD features following Cyclo blockade of the Hh signaling pathway (see Supplementary Video S2), which triggered a decrease in WD elongation and extraluminal thickness in addition to an increase in the intraluminal area (Figure 5A and B). These results suggest that Hh signaling is involved in the control of WD development.

Figure 5.

Both activation and inhibition of Hedgehog signaling impairs ex vivo WD morphogenesis. Wolffian duct responsiveness to Hedgehog agonist (SAG) and antagonist (Cyclopamine, Cyclo) treatments was measured on E16.5 WD cultured for 72 h. (A) Light microscopy images of whole WD, and immunodetection of αSMA and vimentin on sections of WD cultured for 72 h under Control, smoothened agonist (SAG, 0.5 μM), and cyclopamine (Cyclo, 25 μM) conditions. All ducts in the IHC images are different parts of the whole WD. Scale bar: 1 mm for light microscopy images, and 0.05 mm for immunodetection images. The negative controls are shown in Supplementary Figure S14. (B) Elongation (mm), number of coils, luminal area/length (mm), and extraluminal thickness (mm) measured in E16.5 WD cultured for 72 h in SAG (0.5 μM) or Cyclo (25 μM), and compared with control condition. (C) Western blot detection of αSMA, vimentin, and β-actin in E16.5 WD cultured for 72 h under Control and SAG (0.5 μM) conditions. Relative expression levels of αSMA and vimentin were normalized to β-actin. ^* indicates a statistical significance of P < 0.05 between treatments. ns indicates non-statistical significance. All quantitative data are presented as means and standard errors of the mean (SEM).

The increase in extraluminal layer induced by SAG is not associated with a higher cell proliferation rate in mesenchyme

Since the activation of Hh signaling induced an increase in WD extraluminal thickness and vimentin expression, we hypothesized that the activation of Hh signaling might stimulate cell proliferation in the mesenchyme. As observed by immunohistochemistry on WD sections, the cell proliferation marker KI-67 was detected in epithelial cells as well as in αSMA-positive and αSMA-negative mesenchymal cells in the control (Figure 6A). Quantification of KI-67-positive cells in the whole WD section and in the different cellular populations, i.e. epithelial and mesenchymal cells and in αSMA-positive and αSMA-negative mesenchymal cells, was performed following SAG treatment. No significant change in KI-67 detection was observed in the whole WD (Figure 6B), nor in epithelial and αSMA-negative mesenchymal cell populations after SAG treatment (Figure 6C). In contrast, the proliferation rate of the αSMA-positive smooth muscle layer was significantly decreased following activation of the Hh pathway by SAG (Figure 6C). Thus, a higher cell proliferation rate in WD was not associated with the increase in extraluminal layer induced by SAG.

Figure 6.

Activation of Hedgehog signaling triggers a decrease of smooth muscle cell proliferation ex vivo. (A) Immunodetection of the proliferation marker KI-67 performed on 16.5 days post-coitum (E16.5) sections of WD cultured for 72 h under Control or SAG (0.5 μM) conditions. Scale bar: 0.05 mm. The percentage of KI-67-positive cells was quantified (B) in the whole WD, (C) in the epithelial cells, and in αSMA⁺ (smooth muscle cells) and αSMA⁻ mesenchymal cells of E16.5 WD cultured for 72 h under Control or SAG (0.5 μM) conditions. The area of αSMA+ and αSMA⁻ mesenchymal cells was delineated through IHC on serial sections. ^* indicates a statistical significance of P < 0.05 between treatments. ns indicates non-statistical significance. All quantitative data are presented as means and standard errors of the mean (SEM).

Activation and blockade of the Hh signaling pathway trigger major WD gene expression changes

As the activation and blockade of the Hh signaling pathway promoted opposing morphological changes that impaired WD development ex vivo, we scrutinized the molecular components and pathways involved in these responses. To this end, we performed complete RNA expression profiles of WD organotypic cultures treated with SAG (Hh activation) or Cyclo (Hh blockade) and compared them with the control conditions. Wolffian ducts were collected on E16.5 and cultured during 3 days (72 h) in the different conditions. Principal components analysis showed a clear distinction between the overall transcriptome profiles from samples subjected to the specific treatments (Figure 7A). ANOVA was conducted between the following groups: SAG vs. Control; Cyclo vs. Control; and SAG vs. Cyclo. Treatment with Cyclo showed a much greater effect on the transcriptomic profile than with SAG. Furthermore, since these two agents exert essentially opposing effects on the same pathway, the contrast between SAG and Cyclo was particularly enhanced (Figure 7B). For instance, with the same filtering stringency (i.e. statistical significance when FC > 1.5 and P-value corrected by FDR < 0.03), we identified 736 genes whose expression was significantly different between Cyclo vs. Control, 104 for SAG vs. Control and 1883 for SAG vs. Cyclo. The genes specific to each contrast were systematically studied with five functional analysis suites (DAVID, STRING, GOrilla, Metscape, GSEA, and Blast2GO). In this analysis, six biological processes were revealed in four of the analysis suites for SAG; these included morphogenesis and regulation of hormone levels (Supplementary Figure S8). For Cyclo, 29 biological processes consistently emerged in three of the analysis suites, including the regulation of cell adhesion, and cell surface receptor signaling pathway, with cell differentiation being the most represented biological process (Supplementary Figure S8). Finally, the comparison of SAG vs. Cyclo provided a total of 743 significantly enriched biological process terms, among which 22 were consistently found in four of the analysis suites, including kidney development, extracellular matrix (ECM) organization, branching morphogenesis of an epithelial tube and regulation of Wnt signaling pathway.

Figure 7.

Modulation of WD gene expression in response to Hedgehog agonist (SAG) and antagonist (Cyclopamine, Cyclo) particularly affects genes related to the maintenance of the extracellular matrix (ECM) and involved in WD development. Total RNA were extracted from 16.5 days post-coitum embryonic (E16.5) WD after 72 h of ex vivo treatment. (A) Principal component analysis of differentially expressed genes in E16.5 WD cultured for 72 h under Control (blue balls), SAG (red balls), and Cyclo (purple balls) conditions. (B) Volcano Plot of over-expressed genes in E16.5 WD cultured for 72 h in the presence of SAG (green points) or Cyclo (red points). (C) Non-supervised heat map of the differentially expressed Hedgehog signaling pathway related genes in E16.5 WD cultured for 72 h under Control (Ctrl), SAG, and Cyclo conditions (Schematic representation of effects of SAG and Cyclo under Hedgehog signaling pathway-related genes and qPCR: Supplementary Figure S15). (D) Non-supervised heat map of the differentially expressed ECM receptor related genes in E16.5 WD cultured for 72 h under Control (Ctrl), SAG, and Cyclo conditions. (E) Non-supervised heat map of the differentially expressed WD development-related genes in E16.5 WD cultured for 72 h under Control (Ctrl), SAG, and Cyclo conditions (qPCR: Supplementary Figure S16). In heat map: each number (_1, _2, _3, and _4) represents a biological replicate; Blue to red colors: lower to higher expression intensity levels. Threshold criteria: fold change > 1.5 and P-value corrected by false discovery rate (FDR) < 0.03.

The canonical Hedgehog pathway modulates genes and biological processes important to WD morphogenesis

The comprehensive repertoire of genes modulated in WD organotypic cultures after treatment with activator (SAG) or inhibitor (Cyclo) of Hh pathway showed that 17 genes belonging to well-described Hh canonical target genes were altered, including nine genes (Hhip, Gli1, Ptch1, Ptch2, Gpc3, Foxf1, Hhat, Gli2, and Boc) that were significantly up-regulated following SAG treatment and down-regulated following Cyclo treatment, and three genes (Fgf10, Shh, and Pax2) that displayed the opposite trend. In addition, five canonical Hh target genes (Bcl2, Arl13b, Fgfr2, Cdon, and Lrp2) were significantly modulated following Cyclo treatment only. Overall, all the main components of the canonical Hh pathway, including Hh receptors and co-receptors, GLI transcription factors, and Hh-interacting proteins, responded to WD treatment with an activator and blocker of this pathway as represented in Figure 7C. Genes belonging to different major pathways have been clustered downstream of this initial response. Out of these, 261 genes are likely to be regulated by GLI factors according to the presence of GLI consensus binding sites in their promoter regions (see Supplementary Table S2).

Among the pathways altered by the regulation of Hh signaling in the WD, ECM factors, and major players involved in WD development showed important and opposite effects following SAG and Cyclo treatments. Within the ECM cluster, the expression of 12 genes encoding collagen alpha chains, integrins Itga4 and Itgb8, and fibronectin 1 were significantly induced following Hh activation and conversely reduced following Hh inhibition. In contrast, the expression of integrin Itga3, syndecan-4 (Sdc4), and Fras1-related ECM gene 2 (Frem2) was strikingly induced following Hh signaling blockade (Figure 7D). Finally, with the identification of several genes that contribute to WD development in knockout mouse models [7], we investigated the expression of these genes in our model (Table 1). As androgens are also important to WD development, we also investigated the expression of Androgen Receptor gene in our data that demonstrate lower expression in SAG treated WD compared with control (Supplementary Figure S9). We found that many genes display significant transcriptional regulation under the control of Hh signaling, including Pax2, Pax8, and Gata3, whose expression were up-regulated following Cyclo treatment and, conversely, down-regulated following SAG treatment (Figure 7E). Altogether, these results indicate that proper WD development occurs under the control of a canonical Hh signaling through the regulation of developmental genes and ECM factors.

Table 1.

List of genes whose expression is significantly modulated by Hedgehog (Hh) signaling in the developing Wolffian duct (WD) following treatment with smoothened agonist (SAG, 0.5 μM) or cyclopamine (Cyclo, 25 μM) and that are associated with developmental defects according to in vivo animal models and clinical data

Genes	Expression during activation of Hedgehog in WD	Expression during blockade of Hedgehog in WD	Involvement in development based in vivo models
Hhip a	Up-regulated	Down-regulated	Muscle development [50]
Gli1 a	Up-regulated	Down-regulated	Hypospadia and early embryo urethral development [51]
Ptch1 a	Up-regulated	Down-regulated	Mammary gland morphogenesis [52]
Ptch2 a	Up-regulated	Down-regulated	Regulation of primordial germ cell migration [53]
Gpc3 a	Up-regulated	Down-regulated	Severe malformations and pre- and post-natal overgrowth associated with Simpson–Golabi–Behmel syndrome [54]
Foxf1 a	Up-regulated	Down-regulated	Mesenchymal and epithelial proliferation and differentiation in ureter development [55]
Hhat b	Up-regulated	Down-regulated	Testicular dysgenesis in a case of 46,XY disorder of sex development (DSD) [56]
Gli2 c	Up-regulated	Down-regulated	Placental development [57]
Boc d	Up-regulated	Down-regulated	Myelin formation [58]
Bcl2 c	Up-regulated	Down-regulated	Hearing development [59]
Arl13b d	Up-regulated	Down-regulated	Joubert syndrome [60]
Fgfr2 d	Down-regulated	Up-regulated	Kidney mesenchymal development [61]
Cdon d	Down-regulated	Up-regulated	Cell differentiation and determination during embryogenesis [62]
Lrp2 d	Down-regulated	Up-regulated	Epididymal post-natal development [63]
Fgf10 d	Down-regulated	Up-regulated	Prostate development [64]
Shh d	Down-regulated	Up-regulated	Bladder smooth muscle differentiation [65–67]
Pax2 c	Down-regulated	Up-regulated	Ductal and mesenchymal components dysgenesis during development of the urogenital system [68] Cystic development and epithelial neoplasms in adult epididymal tissues [69]
Pax8 c	Down-regulated	Up-regulated	Contributes to the mesonephros development [70–72] Cystic development and epithelial neoplasms in adult epididymal tissues [69,73]
Gata3 d	Down-regulated	Up-regulated	Morphogenesis of pro/mesonephros [70,72]
Emx2 d	Down-regulated	Up-regulated	Ureteric bud functions [74]
Wnt9b d	Down-regulated	Up-regulated	Mesenchymal-to-epithelial transition [75] Kidney morphogenesis and regeneration [76,77]
Fgfr2 d	Down-regulated	Up-regulated	Maintenance of the WD, mainly the caudal part [78]
Ros1 c	Down-regulated	Up-regulated	Post-natal development of the epididymis initial segment [79]
Dusp6 c	Down-regulated	Up-regulated	Major regulator of cell proliferation on epididymis caput and corpus [80]
Foxc2 b	Up-regulated	Down-regulated	Contributes to kidney and ureter morphogenesis [81]
Inhba c	Up-regulated	Down-regulated	Control of WD coiling [82]
Sfrp1 d	Up-regulated	Down-regulated	Defects in testes and epididymal development [83]
Wnt5a d	Up-regulated	Down-regulated	Ureteric buds mesoderm extension [84]
Pkd1 c	Up-regulated	Down-regulated	Mesonephric ducts cystic dilation [12]
Ar b	Down-regulated	Up-regulated	Degeneration of Wolffian ducts and testes [82,85]

^aGenes differentially expressed (fold change >1.5 and P-value corrected by FDR ≤ 0.03) between cyclopamine (Cyclo) vs. Control, smoothened agonist (SAG) vs. Control and Cyclo vs. SAG.

^bGenes differentially expressed (fold change >1.5 and P-value corrected by FDR ≤ 0.03) between SAG vs. Control and Cyclo vs. SAG.

^cGenes differentially expressed (fold change >1.5 and P-value corrected by FDR ≤ 0.03) between Cyclo vs. Control and Cyclo vs. SAG.

^dGenes differentially expressed (fold change >1.5 and P-value corrected by FDR ≤ 0.03) between Cyclo vs. SAG.

Discussion

Proper development, morphogenesis, and patterning of the embryonic WD are required to generate a fully functional epididymis in which sperm maturation occurs after puberty [6]. Although Hh signaling is a cell–cell communication system that regulates embryonic structures through the primary cilium, its contribution to epididymis development remains poorly characterized. Here, by using complementary in vivo and ex vivo models, we revealed that impairment of primary ciliogenesis and modulation of downstream Hh signaling result in defective WD coiling, elongation, and patterning. This occurs through the regulation of ECM components and key developmental genes, pointing to the maintenance of the Hh equilibrium as being of paramount importance during WD development (Figure 8).

Figure 8.

Schematic representation of WD responsiveness to Hedgehog (Hh) signaling. (A) Primary cilia are found at the surface of epithelial and mesenchymal cells from wild type WD where they transduce a canonical Hh signaling through SMO and Gli1/2, Ptch1/2, and Hhip target genes. The effects of primary cilia deletion as well as Hedgehog (Hh) signaling induction and blockade were assessed in vivo (B) as well as ex vivo (C) from organotypic cultures of developing WD. Under physiological conditions (middle panel), the balance existing between Hh morphogens and other signaling molecules ensures proper WD elongation and coiling, and maintenance of normal ECM integrity. Following the induction of the Hh signaling (right panel), the WD develops as a shorter tubule displaying a large extraluminal space. These morphological changes are associated with the alteration of ECM and urogenital tract developmental gene expression levels. Following blockade of Hh signaling (left panel), the WD display cyst-like intraluminal bulges and a smaller extraluminal space that is associated with modifications to the ECM and urogenital tract developmental gene expression levels.

IHH as a potential driver of WD development

The canonical Hh signaling pathway is transduced through the primary cilium in response to SHH, DHH, or IHH morphogens, depending on the extracellular context [86]. In men, mutations on Dhh are associated with gonadal dysgenesis and male infertility [23], highlighting the importance of this Hh morphogen and downstream cues on male reproductive tract development as well as on the etiology of certain human disorders of sexual differentiation. In mouse models, the contribution of Hh to WD development has been supported in a model where the ablation of Gli1 expressing cells impaired WD coiling and elongation [87]. Here, we also identify IHH as a potential driver of WD development through its probable association with epithelial primary cilia from the proximal segment of the embryonic tubule. Although this regionalized IHH feature in the embryo may reflect the differences existing between the rodent initial segment and the remainder of the epididymis at the adult stage, advance lineage tracing approach will be required to support this potential developmental link. This observation suggests that IHH binds to PTCH1 or PTCH2 ciliary receptors to control WD epithelial cell functions. Although WD development was shown to occur independently from mesonephro-derived SHH [22], the respective contributions of IHH and DHH morphogens to male reproductive tract development remain to be deciphered through complementary genetic studies in men and functional analyses from mouse models.

On/off Hh cues controls WD development through the regulation of the ECM and developmental genes

In the present study, imbalanced Hh signaling triggered disproportionate development of the WD mesenchymal/intraluminal compartment in vivo and ex vivo. This tissular phenotype was associated with reduced coiling/elongation of the WD, as well as marked changes to ECM components at the gene expression level. The contribution of ECM to WD development has already been suggested from a microarray study, in which changes in gene expression of ECM components were observed in different regions of the developing epididymis, and at different time points (E14.5–P1) [88]. This concept was further supported from studies on protein tyrosine kinase 7 (Ptk7) knockout WD that displayed an impaired elongation and changes in ECM integrity [33,89]. Given the opposing effects on modulation of ECM components in the presence of smoothened agonists and antagonists, it is likely that Hh-dependent ECM integrity is the major cause of impaired WD elongation observed in our system. For instance, the expression levels of collagen I-, II-, IV-, and VI-encoding genes are strongly reduced following Hh blockage, and is associated with both an increase size of the intraluminal compartment and a reduced WD elongation. These features mimic the cyst-like dilatation effect observed following collagenase treatment on WD organotypic cultures [89].

In addition, other genes involved in WD development respond to Hh signaling treatment, including genes known to participate in pro/mesonephros morphogenesis (Androgen receptor (Ar), Gata3) [70,82], maintenance of the WD structure (Emx2) [74], mesenchymal–epithelial transition (Wnt9b, Pax2, and Pax8) [68,71,90], WD/epididymal patterning (Fgfr2, Ros1, Dusp6) [78–80,91] as well as in coiling and elongation of the WD (Inhba, Pkd1) [12,82], cystic development and epithelial neoplasms in adult epididymal tissues (Pax2 and Pax8) [69,73] (Table 1). The overall modulation of key developmental genes therefore identifies Hh signaling as a master switch of WD development and patterning. Interestingly, corroborating the results on the ex vivo model as regard to Ar expression level, the anogenital distance of newborns Pax^cre; IFT88^fl/fl mice was lower (data not shown) compared with control mice, indicating that the production of testosterone were probably impaired in mutants [92]. This finding raises new questions regarding the potential contribution of primary cilia and downstream signalings in testicular steroid hormones production.

WD development is under the control of a canonical Hh signaling

Although “canonical” Hh signaling involves the contribution of the downstream Gli family of transcription factors and the regulation of target gene expression through GLI-consensus binding sites [86], some organ systems respond to Hh morphogens through a Gli-independent “noncanonical” Hh signaling [93–95]. By investigating the molecular response of the developing WD to Hh cues, we detected a change in expression level of Gli1 and Gli2 canonical Hh target genes, in addition to Ptch1, Ptch2, and Hhip that contain GLI consensus binding sites in their promoter region. Furthermore, in accordance with the recognized mechanisms that link Hh signaling with the primary cilium in vertebrates [18], the blockade of ciliogenesis in both in vivo (Pax^cre; IFT88^fl/fl mice) and ex vivo (CilioD) models triggered changes in the expression of Hh factors as well as WD morphometric features. Although primary cilia-dependent Hh signaling contribute to WD development, other signaling pathways also mediated by this organelle could be involved in that response, including Wnt, Notch, and Hippo [96]. For instance, in addition to Hh signaling factors Gli1 and Hhip that were used as a read-out of proper Hh signaling activation or inhibition, changes in Wnt (e.g. Wnt5a/b), Notch (e.g. Notch3), and Hippo (e.g. Wwc1) related genes were also observed in response to Hh treatments (Supplementary Figure S10), suggesting that Hh signaling is the nexus of diverse signaling pathways. The role of primary cilia in WD development aligns with previous studies where invalidation of Pkd1 or Pkd2 that encode the ciliary components polycystin1 (PC1) and 2 (PC2), respectively, displays impaired in utero epididymal coiling [11, 12] comparable to our ex vivo observations, following the pharmacological removal of primary cilia. The same developmental phenotype was observed in epithelial-specific Pdk1 or Pkd2 knockout mice, suggesting that the ciliary PC1–PC2 channels from epithelial cells are instrumental to proper WD elongation and coiling. In the kidney, PC1 and PC2 play a key role in the mechanical sensing of extracellular shear stress through the primary cilium [97]. It remains to be established whether mechanical shear stress or other signaling pathways such as FGF, Wnt, Notch, and Hippo [96] could potentiate/modulate the response of the developing WD to the Hh cue. Furthermore, Pax2 being expressed in the mesonephric ducts/efferent ductules, it is possible that an impaired luminal signaling coming from upstream the WD, for instance from the efferent ductules, could indirectly affect WD development in addition to the direct effect of disrupting primary cilium in the WD epithelium. For instance, an excretory function has been described across mouse mesonephric tubule epithelium as early as embryonic day E10.5 [98], and could potentially influence the luminal content and the downstream regulation of WD morphogenesis.

Conclusion

Ciliary dysfunctions underlie several diseases and disorders commonly referred to as ciliopathies. Most human ciliopathies exhibit altered ECM, leading to fibrosis [99–101], and some are associated with male infertility along with the presence of cysts in the epididymis [102–104]. However, the etiology of this ciliopathic-related male reproductive tract issue remains unknown. Our study provides evidence for the contribution of non-motile primary cilia and downstream Hh signaling pathways to the control of ECM integrity during WD development. These findings shed light on new molecular candidates that may be important for proper epididymis development that is instrumental to the control of male fertility.

Author contributions

MBRA BTH and CB contributed to conception and design of the work; MBRA, LG, CA, KHM, CLO, AB, DS, MET, JB, BTH contributed to acquisition and analysis of biological data; EC, CJB, AD, CR, and AB contributed to bioinformatics data analysis; MBRA and CB wrote the first draft of the manuscript; LG, CA, MET, BTH, CR contributed to manuscript revision.

Funding

MET lab is supported by the National Institutes of Health (grant R03HD101762).

Conflict of interest

None to declare.

Data availability statement

The data underlying this article are available from the GEO repository (accession number #GSE145816) and in its online supplementary material. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.