Identification of novel genes and pathways regulated by the orphan nuclear receptor COUP-TFII in mouse MA-10 Leydig cells

Abstract

In males, Leydig cells are the main producers of testosterone and insulin-like 3 (INSL3), two hormones essential for sex differentiation and reproductive functions. Chicken ovalbumin upstream promoter-transcription factors I (COUP-TFI/NR2F1) and COUP-TFII (NR2F2) belong to the steroid/thyroid hormone nuclear receptor superfamily of transcription factors. In the testis, COUP-TFII is expressed and plays a role in the differentiation of cells committed to give rise to fully functional steroidogenic adult Leydig cells. Steroid production has also been shown to be diminished in COUP-TFII-depleted Leydig cells, indicating an important functional role in steroidogenesis. Until now, only a handful of target genes have been identified for COUP-TFII in Leydig cells. To provide new information into the mechanism of action of COUP-TFII in Leydig cells, we performed microarray analyses of COUP-TFII-depleted MA-10 Leydig cells. We identified 262 differentially expressed genes in COUP-TFII-depleted MA-10 cells. Many of the differentially expressed genes are known to be involved in lipid biosynthesis, lipid metabolism, male gonad development, and steroidogenesis. We validated the microarray data for a subset of the modulated genes by RT-qPCR. Downregulated genes included hydroxy-delta-5-steroid dehydrogenase, 3 beta- and steroid delta-isomerase 1 (Hsd3b1), cytochrome P450, family 11, subfamily a, polypeptide 1 (Cyp11a1), prolactin receptor (Prlr), nuclear receptor subfamily 0, group B, member 2 (Shp/Nr0b2), ferredoxin 1 (Fdx1), scavenger receptor class B, member 1 (Scarb1), inhibin alpha (Inha), and glutathione S-transferase, alpha 3 (Gsta3). Finally, analysis of the Gsta3 and Inha gene promoters showed that at least two of the downregulated genes are potentially new direct targets for COUP-TFII. These data provide new evidence that further strengthens the important nature of COUP-TFII in steroidogenesis, androgen homeostasis, cellular defense, and differentiation in mouse Leydig cells.

An in-depth high-throughput transcriptomic analysis of COUP-TFII-depleted Leydig cells reveals novel gene pathways and networks regulated by this orphan nuclear receptor.

Graphical Abstract

Graphical Abstract.

Introduction

The chicken ovalbumin upstream promoter-transcription factor II (COUP-TFII/NR2F2) belongs to a subfamily of COUP-TF nuclear receptors. COUP-TFII is expressed in various cell types and tissues (reviewed in [1–4]). Global inactivation of Coup-tfii in mice causes severe hemorrhaging and ultimately death due to defects in angiogenesis and heart development [5]. Tissue-specific inactivation of Coup-tfii in stomach, uterus, limbs, skeletal muscle, and endothelial cells uncovered additional essential roles for COUP-TFII in cell differentiation, cell function, and organogenesis [6–11]. In humans, genetic mutations in COUP-TFII that disrupt COUP-TFII protein levels lead to forms of congenital heart abnormalities [12, 13] and testis formation in XX children [14].

COUP-TFII is detected in interstitial and peritubular cells of the human testis at the embryonic, neonatal, juvenile, and adult stages of development [15, 16]. In the mouse testis, COUP-TFII is also found in testicular interstitial cells throughout ontogeny [17]. Located in the interstitial space within the testis, Leydig cells produce the hormones insulin-like 3 (INSL3) and androgens, such as testosterone. Although INSL3 and androgens are essential for the development and function of the male reproductive system during both fetal and postnatal life, two distinct populations of Leydig cells are responsible for the production of these hormones during these periods: fetal Leydig cells (FLC) and adult Leydig cells (ALC) (reviewed in [18]). In mammals, ALC differentiation is characterized by four distinct stages: stem, progenitor, immature, and mature (reviewed in [19]). COUP-TFII is present in ALCs at all differentiation stages [17]. Using an inducible time-dependent Coup-tfii knockout model, Qin et al. found that in the absence of COUP-TFII, ALC differentiation is arrested at the progenitor stage leading to Leydig cell hypoplasia, decreased serum testosterone levels, and increased circulating luteinizing hormone (LH) [20].

Generally, COUP-TFII regulates gene expression by binding as monomer, homodimer, and heterodimer with RXR to a nuclear receptor response element (NRE) containing the core sequence AGGTCA (reviewed in [4]). In mouse Leydig cells, COUP-TFII contributes to steroidogenesis by activating Star gene expression through a direct repeat 1 (DR1) element in its promoter [17]. In mouse Leydig cells, COUP-TFII also activates transcription of the gene insulin-like 3 (Insl3) through a DR0 element [21], 3α-hydroxysteroid dehydrogenase enzyme type 1 (Hsd3a1, Akr1c14) via a DR7 element [22], and anti-Müllerian hormone receptor type 2 (Amhr2) via a GC-rich element [23].

To expand our knowledge of the mechanism of action of COUP-TFII in Leydig cells, we used small interfering RNAs (siRNAs) to knockdown COUP-TFII levels in MA-10 Leydig cells followed by transcriptomic analysis. Detailed microarray analyses of COUP-TFII-depleted MA-10 Leydig cells identified new COUP-TFII-dependent genes, which provide a better understanding of how COUP-TFII controls androgen production in Leydig cells.

Materials and methods

Expression and reporter plasmids

The mouse Amhr2 (−1486/+74 bp and −34/+74 bp) and Gsta3 (−2068/+38 bp and −76/+38 bp) reporter constructs were described previously [23, 24]. The −1923/+112 bp Hsd3b1 and −2108/+40 Inha promoter fragments were PCR amplified from mouse genomic DNA using the primer sets listed in Table 1. The PCR amplicons were gel extracted, enzyme digested (−1923/+112 bp Hsd3b1, Xho I/Kpn I; −2108/+40 Inha, Bgl II/Sal I), and cloned into digested pXP1 luciferase reporter vectors, respectively [25]. The mouse −21/+38 bp Gsta3 promoter construct was PCR amplified from the −76/+38 bp Gsta3 plasmid using PfuUltra High-Fidelity DNA Polymerase AD (Agilent Technologies, California, USA) following the manufacturer’s instructions with the primer set listed in Table 1. The mouse COUP-TFII expression vector was generated by subcloning the coding sequence into pcDNA3.1 (Invitrogen Canada, Ontario, Canada) as previously described [23]. The sequences of the constructed plasmids were confirmed by an on-site sequencing service.

Table 1.

Oligonucleotides used in this study

Purpose	Description	Template	Sequence	Tm °C
qPCR	Coup-tfii		F: TGGAGAAGCTCAAGGCACTG	62.6
			R: AAGAGCTTTCCGAACCGTGT
	Star		F: GTTCCTCGCTACGTTCAAGC	62.6
			R: GAAACACCTTGCCCACATCT
	Prlr		F: TCTCAGAGACACGCGGCTG	65
			R: TTCTGCTGGAGAGAAAAGTCTG
	Cyp11a1		F: CACCAGTATTATCAGAGGCCC	62.6
			R: GATGAAGTCCTGAGCTACACC
	Lhcgr		F: TGCCTTTGACAACCTCCTCA	62.6
			R: GAAACATCTGGGAGGGTCCG
	Gsta3		F: TGGCGGGGAAGCCAGTCCTT	62.6
			R: ACCTTGCCAGGTCATCCCGAGT
	Pde8a		F: GTGCAATTTGGCCCGATGA	62.6
			R: GATGTCATGGAGTTTGTCCTGG
	Fdx1		F: AAGAACCGAGATGGCGAGAC	62.6
			R: GACAAACTTGGCAGCCCAAC
	Scarb1		F: GCTGCTGTTTGCTGCGCTCG	62.6
			R: GGGTCCACGCTCCCGGACTA
	Inha		F: CGAACTTGTCCGGGAGCTCGT-	61
			R: TGGCTGGTCCTCACAGGTGGC
	Hsd3b1		F: GGC TGGATGGAGCTGCCTGG	62
			R: GCTCTCCTCAGGCAC TGGGC
	Nr0b2		F: TACCCAGGGTGCCCAGCCATC	62.6
			R: TGCAGGTGTGCGATGTGGCAG
	Amhr2		F: CCCTCTGCCCTCTGGGCCTT	68.8
			R: ACTGGCCATCCTGCCAACGC
	Rpl19		F: CTGAAGGTCAAAGGGAATGTG	62.6
			R: GGACAGAGTCTTGATGATCTC
	Pdgfra		F: CAATCCAAAGATGTCCAGGTC	62.6
			R: ACCAAGTCAGGTCCCATTTAC
	Insl3		F: TGGCTAGAGCAGAGACATC	62.6
			R: CCTGTGGTCCTTGCTTAC
Promoter constructs	−21/+38 bp Gsta3	−76/38 Gsta3 promoter	F: CTCATCAATGTAAGCTTGAGGGCGTATTCAAATTTA	61
			R: TAAATTTGAATACGCCCTCAAGCTTACATTGATGAG
	−1923/+112 bp Hsd3b1	Mouse gDNA	F: TAGCGctcgagTGTTCCCCTTCCTTGATCC	60
			R: GGggtaccGCTCAGTTCAGAATGTAG
	−2108/+40 bp Inha	Mouse gDNA	F: CGCagatctCTATTCCTTGGCTTAGTGGT	63.2
			R: CTTgtcgacGCTGCCCTGTGCCCTTTCTGT

Cell culture

Mouse MA-10 Leydig cells (ATCC, Cat# CRL-3050, RRID:CVCL_D789), donated by Dr. Ascoli (University of Iowa, USA), were grown in DMEM/F12 medium supplemented with 2.438 g/L sodium bicarbonate, 3.57 g/L HEPES, 15% horse serum, 50 mg/L penicillin and streptomycin sulfate on plates coated with 0.1% gelatin at 37°C, and 5% CO₂. MA-10 cells, generated from a mouse Leydig cell tumor, represent immature ALC [26]. Upon stimulation by luteinizing hormone/human chorionic gonadotrophin (Lhcgr), forskolin, or cAMP, MA-10 Leydig cells produce mainly progesterone due to a defect in 17α-hydroxylase/17,20 lyase (CYP17A1) activity [27].

siRNA-mediated depletion of COUP-TFII, RNA isolation, microarray screening, and RT-qPCR

Endogenous COUP-TFII was depleted in MA-10 Leydig cells using an siRNA approach as described previously [23]. Briefly, MA-10 cells were transfected with 150 nM of COUP-TFII targeting siRNA (Nr2f2-MSS235957 Thermo Fisher Scientific, Ontario, Canada) or with Stealth RNAi siRNA Negative Control, Med GC (siRNA Ctrl) (Thermo Fisher Scientific, Ontario, Canada) for 48 h using JetPRIME Transfection Reagent (PolyPlus-transfection, Illkirch, France), polyethylenimine hydrochloride (PEI) (Sigma-Aldrich Canada, Ontario, Canada), or Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher Scientific, Ontario, Canada). Next, the siRNA-treated cells were collected, and total RNA was isolated using either TRIZOL reagent (Thermo Fisher Scientific, Ontario, Canada) or the guanidinium thiocyanate procedure [28]. Total RNAs were isolated from three independent experiments, and 250 ng of each were submitted to the microarray facility at the Centre Hospitalier Universitaire de Québec Research Centre. Samples were prepared as described previously in [29]. DNA microarray experiments were carried out using Affymetrix Mouse Gene 1.0 ST arrays (Thermo Fisher Scientific, Ontario, Canada).

For microarray data validation, cDNA was synthesized using iScript cDNA Synthesis Kit (Bio-Rad Laboratories, Ontario, Canada) as previously described [23, 30, 31]. Generated cDNA was used as a template for qPCR experiments using FastStart DNA Master SYBR Green I (Roche Diagnostics, Québec, Canada) or SsoAdvanced Universal SYBR Green Supermix (Bio-Rad Laboratories, Ontario, Canada). The qPCR measurements were performed using a LightCycler 1.5 (Roche Diagnostics, Québec, Canada) or CFX96 (Bio-Rad Laboratories, Ontario, Canada) real-time PCR instrument. The sequence of primer sets for each gene and their melting temperatures are listed in Table 1. The mRNA levels were calculated using Pfaffl method [32] or as previously described [23, 30, 31]. Relative gene expression from COUP-TFII-depleted and Ctrl MA-10 cells was normalized to the expression of internal Rpl19 control and data plotted as a percent of the siRNA Ctrl-treated samples.

Progesterone quantification

MA-10 Leydig cells were cultured and transfected with siRNA Ctrl or siRNA against COUP-TFII as described above. Prior to sample collections, the cells were grown in cell culture media without serum and antibiotics for 4 h. Then, the samples were collected and stored at −80°C until the assay. On the day of the assay, the samples were diluted 1/20, and progesterone levels were quantified using enzyme-linked immunosorbent assays (ELISAs) according to the manufacturer’s instructions (Cayman Chemical Company, Michigan, USA; Cat# 582601). Data were analyzed using Microsoft Excel (Microsoft Corporation, USA; RRID:SCR_016137). Progesterone concentration for each sample was determined based on a standard curve generated using known progesterone concentrations. The results are plotted as fold change [Progesterone_{siRNA COUP-TFII}]/[Progesterone_{siRNA Ctrl}].

Microarray data processing

Microarray data were analyzed using Partek genomics suite 7.0 (Partek, Inc. St. Louis, MO, USA). To generate a list of differentially expressed genes, analysis of variance (ANOVA) was performed and genes with absolute difference greater than 1.3-fold and unadjusted P-value <0.01 compared to the siRNA Ctrl-treated samples were selected. Additionally, the microarray data from siRNA COUP-TFII-depleted cells were normalized to data from each siRNA Ctrl sample individually using the same parameters. From this, the coefficient of variation (CV) was calculated for selected genes and presented as a percent (%).

Gene ontology and motif prediction

First, the list of differentially expressed genes identified by Partek analysis was sorted in ascending order based on fold change. Enriched gene ontology biological processes (GO BP) and Reactome biological pathways were generated from the list of downregulated genes using the g:Profiler web-based tool with a threshold value set at 0.05 [33, 34]. The enriched networks were visualized using Cytoscape [35], a network visualization software (version 3.7.2), with Enrichment map plugin [36] and Reactome plugin [37]. To detect the presence of potential direct repeat (DR1) and nuclear receptor element (NRE) in a gene of interest, the proximal gene promoter regions (−1000 bp to +50 bp) were downloaded from the NCBI database and scanned for DR1 and NRE motifs [38] using the web-based motif discovery tool FIMO (version 5.1.0) [39]. The P-value and q-value are indicated in Table 6.

Table 6.

Results of the in silico promoter analyses

Gene symbol	Location of promoter (−1000/+50)	Strand	Start	End	P-value	q-value	Matched sequence RGGYCARAKRYMV	P-value	q-value	Matched sequence RGGYCA
Star	NC_000074.6 (25807474.0.25808523)	−	−75	−63	3.63e–05	0.352	GGGTCAAGGATGG
		−	−191	−179	8.72e–05	0.352	GGGGCAGAGGATA
		−	−392	−380	0.000774	0.467	CTGGCAGAGGCAA
		+	−470	−465				0.000195	0.373	GGGTCA
		−	−171	−166				0.000195	0.373	GGGTCA
		−	−68	−63				0.000195	0.373	GGGTCA
		+	−754	−749				0.000195	0.458	GGGTCA
		+	−864	−859				0.000432	0.458	AGGTCA
		−	−184	−179				0.000752	0.403	GGGGCA
Prlr	NC_000081.6 (10176238.0.10177287)	+	−662	−650	0.000236	0.418	AGGTCATGTGTGA
		+	−662	−657				0.000432	0.458	AGGTCA
		−	−455	−450				0.000592	0.377	GGGCCA
Inha*	NC_000067.6 (75482721.0.75483770)	−	−841	−829	7.16E–05	0.367	GGGTCCAGGGTAA
		−	−835	−823	0.000243	0.418	CGGCCAGGGTCCA
		−	−487	−475	0.000514	0.447	AGGGCAGGTGGTC
		+	−216	−211				0.000195	0.373	GGGTCA
		+	−211	−206				0.000432	0.377	AGGTCA
		+	−139	−134				0.000592	0.377	GGGCCA
		−	−856	−851				0.000946	0.607	CGGTCA
Amhr2	NC_00081.6 (102444390.0.102445439)	+	−281	−269	0.000638	0.447	AGGGCAGAAGGTC
		+	−253	−248				0.000432	0.377	AGGTCA
Lhcgr	NC_000083.6 (88792010.0.88793059, complement)	−	−468	−456	0.00012	0.352	AGGGCAAAATTCA
		+	−77	−65	0.000137	0.352	AGGTCAAGGAGAA
		+	−124	−112	0.0007	0.447	AGGGCAAGTCTTA
		+	+7	+19	0.000326	0.43	CGGGCAGAGGGTA
		−	−537	−525	0.000904	0.632	GGGTAAAATGGCC
		+	−203	−191	0.000961	0.51	AGTTCAAAGCCTC
		+	−77	−72				0.000432	0.377	AGGTCA
		−	−813	−808				0.000432	0.458	AGGTCA
Cyp11a1	NC_000075.6 (57997024.0.57998073)	+	−990	−978	0.00031	0.458	GGGCCATAGCTAA
		−	−155	−143	0.000356	0.43	GGTGCAGGTGTGA
		+	−224	−212	0.00052	0.447	GGGCCATGTGCTC
		+	−224	−219				0.000592	0.377	GGGCCA
		+	−990	−985				0.000592	0.458	GGGCCA
		+	−850	−845				0.000752	0.507	GGGGCA
Hsd3b1	NC_000069.6 (98859745.0.98860794, complement)	+	−552	−540	0.000435	0.508	GGGCCACATAATA
		−	+13	+25	0.000488	0.447	GGGGCAGCTTCAA
		−	−298	−286	0.000592	0.447	AGGTCCAGTGCTA
		+	−674	−662	0.000855	0.628	GGTTCCAGGATAC
		+	−552	−547				0.000592	0.458	GGGCCA
		−	+20	+25				0.000752	0.403	GGGGCA
Pde8a	NC_000073.6 (812122265.0.81213314)	−	−171	−159	0.000707	0.447	GGGGCACCTAGCC
		−	−258	−253				0.000432	0.377	AGGTCA
		+	−65	−60				0.000432	0.377	AGGTCA
		−	−164	−159				0.000752	0.403	GGGGCA
Scarb1	NC_000071.6 (125341045.0.125342094, complement)	−	−156	−144	0.000186	0.352	AGGGCAGAAGACC
		+	−329	−317	0.000244	0.402	AGTGCAAAGGGCC
		+	+35	+47	0.000911	0.51	GGGCCATGGCGCA
		−	−281	−276				0.000592	0.377	GGGCCA
		−	−269	−264				0.000592	0.377	GGGCCA
		+	+35	+40				0.000592	0.377	GGGCCA
Gsta3	NC_000067.6 (21239585.0.21240634)	+	−753	−741	1.26E–05	0.324	GGGCCACAGATCA
		+	−933	−921	9.70E–05	0.367	AGTTCAAAGATAA
		+	−994	−982	0.000143	0.367	AGGGCACATGCAA
		−	−267	−255	0.000943	0.51	CTGGCAGATGTTA
		+	−808	−803				0.000432	0.458	AGGTCA
		−	−549	−544				0.000432	0.458	AGGTCA
		−	−315	−310				0.000592	0.377	GGGCCA
		+	−753	−748				0.000592	0.458	GGGCCA
Fdx1	NC_000075.6 (519963553.0.51964602, complement)	+	−952	−940	0.000352	0.458	GGGCCAGGAGAAA
		+	+14	+26	0.000389	0.431	GGGTTATAGGACA
		−	−883	−871				0.000547	0.53	GGTTCAAGACTCA
		−	−660	−655				0.000592	0.458	GGGCCA
		+	−952	−947				0.000592	0.458	GGGCCA
		−	−394	−389				0.000946	0.488	CGGTCA
Pdgfra	NC_000071.6 (75150322.0.75151371)						NONE			NONE
Nr0b2	NC_000070.6 (133552376.0.133553425)	−	−548	−536	8.69E–05	0.367	GGGCCACAGGGAA
		−	−245	−233	0.000115	0.352	AGGGCACAGGGCC
		−	−253	−241	0.00018	0.352	GGGCCACCTGCCC
		+	−465	−453	0.000272	0.402	GGGTCATGTATTC
		+	−143	−131	0.000557	0.447	GGGTCAGCGTGTA
		+	−199	−187	0.000704	0.447	AGGCCACGTGGAG
		+	−465	−460				0.000195	0.373	GGGTCA
		−	−302	−297				0.000195	0.373	GGGTCA
		+	−143	−138				0.000195	0.458	GGGTCA
		+	−840	−835				0.000432	0.458	AGGTCA
		−	−541	−536				0.000592	0.458	GGGCCA
		−	−246	−241				0.000592	0.377	GGGCCA
		+	+9	+14				0.000592	0.377	GGGCCA
		−	−179	−174				0.000752	0.403	GGGGCA

^*Transcription start site according to [77].

Accession number

Microarray data have been deposited in the GEO database (GSE163283) and can be assessed using the following link https:// www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE163283.

Cell transfections and luciferase assays

To measure promoter activation, transient transfections of MA-10 Leydig cells were performed using polyethylenimine hydrochloride (PEI) (Sigma-Aldrich Canada, Ontario, Canada) as previously described [23]. Briefly, the cells were plated in 24-well plates 24 h prior to transfection. The next day, the cells were co-transfected with 400 ng of reporter vector along with 100 ng of expression vector (empty as control, or COUP-TFII expression vector) and PEI dissolved in OPTI-MEM medium (GIBCO by Life Technologies, Burlington, Ontario, Canada). Sixteen hours after transfection, the media was replaced, and the cells were grown for an additional 32 h. After the cells were lysed, the lysates were collected and luciferase measurements were performed using a Tecan Spark 10 M multimode plate reader (Tecan, North Carolina, USA) as previously described [31, 40].

Western blots

Extraction, quantification, electrophoresis, and Western blotting of total and nuclear proteins from MA-10 Leydig cells were performed as previously described [23, 31]. Total (20 μg) or nuclear (10 μg) proteins were used for immunodetection. The following antibodies were used: mouse monoclonal anti-COUP-TFII antibody (dilution 1:1000; R&D systems Inc, Minnesota, USA; Cat#PP-H7147–00, RRID:AB_2155627) [41], goat polyclonal anti-Lamin B antibody (dilution 1:1000; Santa Cruz Biotechnology, California, USA; Cat# sc-6216, RRID:AB_648156) [42], rabbit polyclonal anti-AMHR2 antibody (dilution 1:1000; Thermo Fisher Scientific, Ontario, Canada; catalog # PA5–112901; RRID:AB_2867635) [43], rabbit polyclonal anti-STAR antibody (dilution 1:1000; Santa Cruz Biotechnology, California, USA; Cat# sc-25806, RRID:AB_2115937) [44], mouse monoclonal anti-GAPDH antibody (dilution 1:5000; Santa Cruz Biotechnology, California, USA; Cat# sc-32233, RRID:AB_627679) [45], and rabbit monoclonal anti-GSTA3 (dilution 1:1000; Abcam, Massachusetts, USA; Cat# ab180928).

Statistical analyses

Statistical analyses between two groups were performed using Student t-test in Microsoft Excel (Microsoft Corporation, USA; RRID:SCR_016137) (RRID:SCR_016137 and https://scicrunch.org/resolver/SCR_016137). P-values from each comparison are indicated in the figure legend.

Results

Validation of siRNA-mediated reduction of COUP-TFII expression in MA-10 Leydig cells

Previous studies have established a role for COUP-TFII in Leydig cell differentiation and steroidogenesis [17, 20]. However, the mechanism of action of COUP-TFII in Leydig cells remains unclear as only a handful of COUP-TFII target genes have been identified in these cells [17, 21–23]. To gain additional insights into how COUP-TFII acts in Leydig cells, COUP-TFII was depleted in MA-10 Leydig cells using an siRNA-directed approach. Efficiency of COUP-TFII depletion was determined by Western blotting and RT-qPCR. As shown in Figure 1A, Coup-tfii mRNA levels were reduced by about 70% compared to the siRNA Ctrl-treated cells after 48 h. COUP-TFII protein levels were reduced by about 61% in COUP-TFII-depleted MA-10 Leydig cells (Figure 1B). Consistent with our previous report [17], progesterone levels were decreased by 26% in COUP-TFII-depleted MA-10 Leydig cells (Figure 1C). These data confirm that COUP-TFII has a significant role in MA-10 Leydig cell steroidogenesis.

Figure 1.

COUP-TFII was depleted in MA-10 Leydig cells using small interfering RNA (siRNA). The cells were transfected with either siRNA control (siRNA Ctrl) or siRNA targeting Coup-tfii (siRNA COUP-TFII) for 48 h prior to total RNA and nuclear protein extraction and progesterone measurements. (A) Coup-tfii mRNA levels were quantified by RT-qPCR and normalized to Rpl19 mRNA. (B) COUP-TFII protein levels are decreased in COUP-TFII-depleted MA-10 Leydig cells. Representative Western blot images from three independent experiments are shown. Ten micrograms of nuclear extract from control siRNA (lane 1) or siRNA targeting COUP-TFII (lane 2) transfected cells were loaded per lane. Lamin B (LMNB1) was used as a loading control. Western blot images were quantified to estimate COUP-TFII relative protein levels. The band intensities were measured using Image Lab 6.0 (Bio-Rad Laboratories). (Figure 1) continued COUP-TFII levels were normalized to Lamin B (LMNB1). (C) Progesterone concentrations were measured by ELISA and plotted as fold change. Prior to sample collections, the cells were incubated in serum-free medium for 4 h. Progesterone concentration ranged from 0.9 to 2.7 ng/ml in siRNA Ctrl-treated MA-10 cells. Data from three independent experiments are shown as the mean ± SEM (^**P < 0.01).

Microarray analysis

To identify differentially expressed genes in COUP-TFII-depleted MA-10 Leydig cells, total RNA was collected from three independent experiments and used for microarray studies. Microarray data were analyzed utilizing the Partek Genomics Suite software and are presented in Supplemental Table S1. The quality of the microarray data was visualized and evaluated using intuitive principal component analysis (PCA) method, which simplifies high-throughput data while retaining patterns. The 2D-PCA biplot produced two distinct, nonoverlapping clusters corresponding to the samples that were collected from cells treated with siRNA against COUP-TFII (green circles) or siRNA Ctrl (orange circles), indicating that COUP-TFII has an impact on the transcriptome of MA-10 Leydig cells (Figure 2A). A variation was observed within the cluster obtained from the siRNA Ctrl-treated cells (Figure 2A).

Figure 2.

Microarray analysis revealed differentially expressed genes in COUP-TFII-depleted MA-10 Leydig cells. The results from microarray screenings were analyzed using Partek Genomics Suite 7.0. (A) The 2D-PCA biplot illustrates two distinctly different gene set populations. The green circles represent data points from COUP-TFII-depleted cells, and orange circles from control cells. The percent variance (%) is indicated next to the principal component 1 (PC1), PC2, and overall principal component analysis (PCA). (B) The heatmap produced from the hierarchical clustering using genes with fold change of ±1.3 and unadjusted P-value <0.01. The number of differentially expressed genes is indicated below the heatmap. A relative intensity scale is shown on the right.

A heatmap showing differentially expressed genes was generated with a threshold fold change of ±1.3 and an unadjusted P-value of <0.01 and is presented in Figure 2B. Under these screening parameters, 262 genes were identified to be differentially expressed, of which 205 genes were downregulated (represented in red) and 57 upregulated (represented in blue). To ensure that the variation within the siRNA Ctrl cluster (Figure 2A) had no significant impact on data analysis, we performed additional analyses where the data set from the siRNA COUP-TFII-depleted cells (three samples) was normalized to each siRNA Ctrl-treated sample (siRNA Ctrl1, siRNA Ctrl2, and siRNA Ctrl3) individually using the same parameters. The top 50 downregulated genes are presented in Table 2, showing gene name, fold change, and P-value. A full list of differentially expressed genes is provided in Supplemental Table S2. The results from the normalization of siRNA COUP-TFII to siRNA Ctrl1, siRNA Ctrl2, and siRNA Ctrl3 as well as the coefficient of variation (CV) values are shown in Table 2. The CV values of the normalized fold changes (siRNA Ctrl1, siRNA Ctrl2, and siRNA Ctrl3) for the top 50 genes ranged from 2 to 26% with an average of 8%, confirming the absence of large variation between the fold changes of the selected genes in all three samples. The most downregulated gene was Pten with fold change of −2.27. As expected, the previously identified COUP-TFII Leydig cell target genes Hsd3b1 (−1.85 fold), Star (−1.71 fold), Akr1c14 (−1.66 fold), and Amhr2 (−1.41 fold) were picked up in the microarray screen (Table 2 and Supplemental Table S2) [17, 20, 22, 23]. Additionally, Coup-tfii was also downregulated by 1.56 fold (Table 2).

Table 2.

Top 50 downregulated genes

	Fold change	P-value	Fold change	P-value	Fold change	P-value	Fold change	P-value	CV (%)
Gene symbol	Normalized to siRNA Ctrl (1–3)		Normalized to siRNA Ctrl1		Normalized to siRNA Ctrl2		Normalized to siRNA Ctrl3
Pten	−2.27	3.53E–03	−2.10	2.45E–04	−2.141	2.02E–04	−2.53	5.24E–04	11%
Cd109	−2.25	2.24E–06	−2.28	7.71E–06	−2.345	2.42E–05	−2.28	1.07E–07	2%
Gsta3	−2.08	5.01E–05	−1.93	1.34E–06	−2.130	3.74E–06	−2.05	4.92E–05	5%
Plb1	−2.03	8.67E–03	−1.82	2.30E–04	−2.414	4.30E–04	n/s	n/s	n/a
Stk32a	−2.03	3.36E–03	−2.07	1.69E–03	−2.120	2.53E–04	−1.77	1.48E–03	9%
Vaultrc5	−2.00	4.09E–03	−2.13	1.27E–03	−2.229	9.06E–04	−1.73	2.59E–03	13%
Lgmn	−1.97	1.22E–04	−1.87	3.72E–05	−2.067	1.24E–05	−1.88	7.86E–05	6%
Pten	−2.27	3.00E–03	−2.10	2.45E–04	−2.141	2.02E–04	−2.53	5.24E–04	11%
Cav1	−1.90	5.93E–06	−1.78	7.88E–09	−1.851	3.65E–06	−1.88	2.19E–07	3%
Hsd3b1	−1.85	2.12E–03	−1.63	6.08E–05	−2.600	8.34E–05	−1.76	2.38E–04	26%
Tlcd2	−1.83	4.30E–04	−1.67	1.28E–07	−1.912	5.52E–06	−1.96	4.06E–05	8%
Pcx	−1.82	1.08E–03	−1.94	3.12E–05	−1.928	1.77E–04	−1.61	4.32E–04	10%
Sspn	−1.78	1.56E–04	−1.68	8.32E–06	−1.888	4.43E–05	−1.85	6.74E–06	6%
Hmox1	−1.77	1.42E–04	−1.67	2.35E–04	−1.765	4.87E–05	−1.86	2.31E–05	5%
Slc7a11	−1.76	1.32E–03	−1.52	2.55E–07	−1.888	4.30E–05	−1.97	3.46E–05	13%
Akr1c18	−1.75	1.82E–03	−1.63	3.23E–05	−1.938	5.85E–05	−1.60	2.59E–03	11%
Ehf	−1.75	6.40E–04	−1.62	4.45E–04	−1.577	5.56E–04	−2.05	4.83E–05	15%
Nrk	−1.74	9.71E–04	−1.75	1.16E–03	−1.963	7.20E–04	−1.72	1.87E–03	7%
Slc40a1	−1.74	6.45E–04	−1.79	6.08E–05	−1.592	2.06E–04	−1.94	9.73E–04	10%
Adam23	−1.73	3.80E–04	−1.69	5.49E–05	−1.743	1.17E–04	−1.64	4.57E–04	3%
Star	−1.71	2.12E–04	−1.59	4.68E–08	−1.794	4.06E–06	−1.70	5.23E–06	6%
Nqo1	−1.68	1.87E–03	−1.52	1.18E–05	−1.844	1.34E–05	−1.67	2.48E–05	10%
Txndc12	−1.67	4.12E–04	−1.56	5.81E–06	−1.869	4.69E–07	−1.67	1.26E–05	9%
Gm10639	−1.67	4.37E–04	−1.63	2.90E–05	−1.842	1.08E–05	−1.67	6.58E–06	7%
Akr1c14	−1.66	5.95E–04	−1.51	3.90E–05	−1.803	1.90E–05	−1.70	2.43E–05	9%
Prlr	−1.66	6.48E–04	−1.52	1.83E–04	−1.782	7.27E–08	−1.63	8.21E–05	8%
Cbr3	−1.65	1.29E–03	−1.53	2.05E–04	−1.846	1.50E–05	−1.62	8.77E–05	10%
Bik	−1.65	1.99E–03	−1.57	4.64E–04	−1.825	8.20E–05	−1.55	3.04E–03	9%
Gsta2	−1.64	2.15E–04	−1.60	1.15E–04	−1.800	3.29E–05	−1.63	1.64E–05	7%
Cdhr5	−1.63	2.32E–04	−1.57	2.49E–05	−1.804	7.58E–05	−1.72	2.00E–05	7%
Gm3776	−1.63	1.16E–03	−1.57	8.49E–05	−1.826	9.37E–06	−1.61	3.59E–06	8%
Htra1	−1.63	5.21E–03	−1.62	2.08E–03	−1.729	1.67E–03	−1.44	2.14E–03	9%
Gsta1	−1.63	8.57E–04	−1.52	4.98E–04	−1.806	1.72E–05	−1.62	1.29E–05	9%
Gsta4	−1.60	2.84E–03	−1.65	8.63E–04	−1.723	3.11E–04	−1.46	2.25E–03	8%
Aldh1a1	−1.59	3.62E–04	−1.52	2.37E–04	−1.670	2.30E–05	−1.61	1.44E–04	5%
Aldh1a7	−1.57	4.96E–04	−1.52	2.71E–04	−1.698	1.03E–03	−1.73	2.28E–04	7%
Coup-tfii	−1.56	1.71E–03	−1.48	2.10E–03	−1.495	1.27E–03	−1.74	1.08E–03	9%
Dram1	−1.56	9.29E–04	−1.46	9.50E–05	−1.661	7.12E–04	−1.79	4.46E–05	10%
Kdm7a	−1.56	1.86E–03	−1.41	6.96E–04	−1.624	3.18E–04	−1.63	4.77E–05	8%
Inha	−1.56	5.82E–04	−1.46	6.66E–04	−1.583	1.83E–05	−1.46	1.32E–03	5%
Prss35	−1.55	2.84E–04	−1.54	2.87E–06	−1.546	4.17E–07	−1.64	6.87E–06	3%
Tbata	−1.55	2.87E–05	−1.60	1.88E–05	−2.14	1.52E–04	−1.53	1.74E–05	2%
Frk	−1.54	1.04E–03	−1.43	1.07E–04	−2.34	8.10E–06	−1.45	5.83E–04	10%
Kdm7a	−1.56	6.36E–05	−1.41	6.96E–04	−2.13	3.18E–04	−1.63	4.77E–05	8%
Rxfp1	−1.54	4.80E–03	−1.42	1.69E–05	−2.41	4.44E–03	−1.48	4.14E–04	8%
Wnt6	−1.53	3.52E–04	−1.49	1.67E–04	−2.12	3.72E–07	−1.58	6.55E–05	5%
Slc44a3	−1.53	3.32E–05	−1.46	9.79E–06	−2.23	4.73E–05	−1.53	1.81E–05	3%
Hephl1	−1.52	1.45E–05	−1.46	2.30E–04	−2.07	3.28E–04	−1.55	1.09E–08	3%
Zcwpw1	−1.52	7.48E–03	n/s	n/s	−2.14	2.49E–04	−1.63	1.34E–03	n/a
Pi4k2b	−1.52	3.29E–04	−1.43	2.20E–05	−1.85	4.52E–04	−1.53	2.63E–04	5%

The genes are sorted by fold change normalized to siRNA Ctrl (Ctrl1, Ctrl2, and Ctrl3) and represented in increasing order. The genes shown in bold correspond to genes previously known to be affected by COUP-TFII depletion [17, 20, 22]. Coefficient of variation, CV; P-value <0.01. n/s: not significant. n/a: not available.

Biological processes affected by COUP-TFII

Gene Ontology (GO) computational analysis is a widely used and accepted approach to associate a gene product to a particular cellular, molecular, or biological process [46]. To identify biological processes (BPs) impaired by the depletion of COUP-TFII in MA-10 Leydig cells, the 205 downregulated genes were used in GO analysis. GO analysis produced 376 GO BP terms; the top 20 most affected biological processes are listed in Table 3 and the full list of BP terms is presented in Supplemental Table S3. Several functions were identified as impaired including biological, metabolic, cellular, cell proliferation, and developmental processes as well as their mode of regulation. This further supports the role of COUP-TFII in steroidogenesis in addition to identifying novel broader roles in Leydig cell function and differentiation.

Table 3.

Top 20 impaired biological processes

Description (GO ID)	Gene names	False discovery rate	Gene count
biological_process (GO:0008150)	Sdk1, Cib3, Smpdl3b, Slc25a33, Tbata, Lgmn, Nedd9, Psmg2, Blvrb, Ssu72, Zcwpw1, Nmrk1, Asah2, Cited4, Gm5862, Thra, Tprkb, Abcb1a, Pipox, Afmid, Gins1, Ribc1, Star, Fam13a, Acacb, Matn2, Casp7, Ecm1, Tmem53, Wdr34, Rhpn2, Tmem38b, Adam23, Lgals3, Gsta3, Gabrg1, Gucy1b2, Nuf2, Maoa, Pxk, Slc40a1, Slc11a1, Aldh1a1, Eno3, Gpt, Rasgrf2, Aox1, Gstt1, Akr1c18, Ehhadh, Wnt6, Slc6a15, Tap1, Hephl1, Pdgfd, Tns2, Serpinb6b, Aqp11, Ugt1a2, Gcnt4, Srxn1, Fbxo47, Cdkn1a, Car7, Bc025920, Avpi1, Slc7a11, Prlr, Gpcpd1, Dram1, Spta1, Tspan5, Inha, Hsd3b1, Nol3, Pi4k2b, Igsf11, Hao2, Cbr3, Htra1, Prss35, Tlcd2, Rxfp1, Syt12, Cdkn3, Rfk, Slc44a3, Kdm7a, Slc26a7, Rnf5, Serpinb1a, Myc, Crip1, A4galt, Cd274, Nqo1, Dhdh, Hmox1, Lrp8, Cd109, Klk1b21, Gsta4, C2, Grasp, Cdon, Fam135b, Gsta2, Entpd5, Gm10639, Gm3776, Tstd3, Rps19bp1, Pir, Tuba8, Cdhr5, Lactb2, Ppif, Txndc12, Psmb9, Zdhhc2, Aldh1a7, Pcx, Cmbl, 5031410i06rik, Pten, Ehf, Stk32a, Amhr2, Clcn5, Usp2, Triap1, Cav1, Hmgcs2, Plb1, Ero1l, Cep350, 5330417c22rik, Usp3, Slc25a34, Adh1, Pm20d1, Adam12, Erlin2, Paqr7, Gls2, Dysf, Fdx1, Tmem25, Bik, Slc47a1, Gm7347, Akr1c14, Plgrkt, Osmr, Ptgis, Nrk, E2f2, Pawr, Pkd2l2, Lynx1, Frk, Igfbp6, Gm5617, Trmt10a, Gsta1, Anxa9, Eepd1, Cdkl3, Klk1b22, 9130230l23rik, Nr0b2, Hagh, Lsr, Efemp1, Nr2f2	4.25E–53	175
metabolic process (GO:0008152)	Smpdl3b, Slc25a33, Lgmn, Nedd9, Blvrb, Ssu72, Nmrk1, Asah2, Cited4, Thra, Tprkb, Pipox, Afmid, Gins1, Star, Acacb, Casp7, Ecm1, Tmem38b, Adam23, Gsta3, Gucy1b2, Maoa, Pxk, Slc40a1, Slc11a1, Aldh1a1, Eno3, Gpt, Aox1, Gstt1, Akr1c18, Ehhadh, Wnt6, Hephl1, Pdgfd, Tns2, Serpinb6b, Ugt1a2, Srxn1, Gcnt4, Cdkn1a, Bc025920, Avpi1, Slc7a11, Prlr, Dram1, Gpcpd1, Spta1, Tspan5, Inha, Hsd3b1, Nol3, Pi4k2b, Hao2, Cbr3, Htra1, Prss35, Cdkn3, Rfk, Kdm7a, Rnf5, Serpinb1a, Crip1, Myc, A4galt, Nqo1, Dhdh, Hmox1, Lrp8, Cd109, Klk1b21, Gsta4, C2, Cdon, Fam135b, Gsta2, Entpd5, Gm10639, Gm3776, Pir, Lactb2, Ppif, Txndc12, Psmb9, Zdhhc2, Aldh1a7, Pcx, Pten, Ehf, Stk32a, Amhr2, Usp2, Triap1, Cav1, Hmgcs2, Plb1, Ero1l, 5330417c22rik, Usp3, Adh1, Pm20d1, Adam12, Erlin2, Gls2, Dysf, Fdx1, Akr1c14, Plgrkt, Ptgis, Nrk, E2f2, Pawr, Frk, Trmt10a, Gsta1, Anxa9, Eepd1, Cdkl3, Klk1b22, Nr0b2, Hagh, Lsr, Efemp1, Nr2f2	1.1E–42	125
cellular process (GO:0009987)	Sdk1, Smpdl3b, Slc25a33, Tbata, Lgmn, Nedd9, Psmg2, Blvrb, Ssu72, Nmrk1, Asah2, Cited4, Thra, Tprkb, Abcb1a, Pipox, Afmid, Gins1, Star, Acacb, Matn2, Casp7, Ecm1, Wdr34, Rhpn2, Tmem38b, Adam23, Lgals3, Gsta3, Gabrg1, Gucy1b2, Nuf2, Maoa, Pxk, Slc40a1, Slc11a1, Aldh1a1, Eno3, Gpt, Rasgrf2, Aox1, Gstt1, Akr1c18, Ehhadh, Wnt6, Pdgfd, Tns2, Serpinb6b, Aqp11, Ugt1a2, Gcnt4, Srxn1, Cdkn1a, Car7, Bc025920, Avpi1, Slc7a11, Prlr, Gpcpd1, Dram1, Spta1, Tspan5, Inha, Hsd3b1, Nol3, Pi4k2b, Igsf11, Hao2, Cbr3, Htra1, Tlcd2, Rxfp1, Syt12, Cdkn3, Rfk, Kdm7a, Rnf5, Serpinb1a, Myc, Crip1, A4galt, Cd274, Nqo1, Hmox1, Lrp8, Cd109, Gsta4, C2, Grasp, Cdon, Fam135b, Gsta2, Entpd5, Gm10639, Gm3776, Pir, Tuba8, Cdhr5, Lactb2, Ppif, Txndc12, Psmb9, Zdhhc2, Aldh1a7, Pcx, Pten, Ehf, Stk32a, Amhr2, Usp2, Triap1, Cav1, Hmgcs2, Plb1, Ero1l, Cep350, 5330417c22rik, Usp3, Adh1, Pm20d1, Erlin2, Paqr7, Gls2, Dysf, Fdx1, Bik, Akr1c14, Plgrkt, Osmr, Ptgis, Nrk, E2f2, Pawr, Lynx1, Frk, Igfbp6, Trmt10a, Gsta1, Anxa9, Eepd1, Cdkl3, Nr0b2, Hagh, Lsr, Efemp1, Nr2f2	2.75E–39	146
organic substance metabolic process (GO:0071704)	Smpdl3b, Slc25a33, Lgmn, Nedd9, Blvrb, Ssu72, Nmrk1, Asah2, Cited4, Thra, Tprkb, Pipox, Afmid, Gins1, Star, Acacb, Casp7, Ecm1, Tmem38b, Adam23, Gsta3, Gucy1b2, Maoa, Pxk, Slc40a1, Slc11a1, Aldh1a1, Eno3, Gpt, Gstt1, Akr1c18, Ehhadh, Wnt6, Pdgfd, Tns2, Serpinb6b, Ugt1a2, Gcnt4, Cdkn1a, Bc025920, Avpi1, Slc7a11, Prlr, Gpcpd1, Spta1, Tspan5, Inha, Hsd3b1, Nol3, Pi4k2b, Hao2, Cbr3, Htra1, Prss35, Cdkn3, Rfk, Kdm7a, Rnf5, Serpinb1a, Crip1, Myc, A4galt, Dhdh, Hmox1, Lrp8, Cd109, Klk1b21, Gsta4, C2, Cdon, Fam135b, Gsta2, Entpd5, Gm10639, Gm3776, Pir, Lactb2, Ppif, Psmb9, Zdhhc2, Aldh1a7, Pcx, Pten, Ehf, Stk32a, Amhr2, Usp2, Triap1, Cav1, Hmgcs2, Plb1, Ero1l, Usp3, Adh1, Pm20d1, Adam12, Erlin2, Gls2, Dysf, Fdx1, Akr1c14, Plgrkt, Ptgis, Nrk, E2f2, Pawr, Frk, Trmt10a, Gsta1, Anxa9, Eepd1, Cdkl3, Klk1b22, Nr0b2, Hagh, Lsr, Efemp1, Nr2f2	1.18E–38	118
cellular metabolic process (GO:0044237)	Smpdl3b, Slc25a33, Lgmn, Nedd9, Blvrb, Ssu72, Nmrk1, Asah2, Cited4, Thra, Tprkb, Pipox, Afmid, Gins1, Star, Acacb, Ecm1, Tmem38b, Gsta3, Gucy1b2, Maoa, Pxk, Slc40a1, Slc11a1, Aldh1a1, Eno3, Gpt, Aox1, Gstt1, Akr1c18, Ehhadh, Wnt6, Pdgfd, Tns2, Serpinb6b, Ugt1a2, Gcnt4, Cdkn1a, Bc025920, Avpi1, Slc7a11, Prlr, Dram1, Gpcpd1, Spta1, Inha, Hsd3b1, Nol3, Pi4k2b, Hao2, Cbr3, Htra1, Cdkn3, Rfk, Kdm7a, Rnf5, Serpinb1a, Myc, A4galt, Nqo1, Hmox1, Lrp8, Cd109, Gsta4, Cdon, Fam135b, Gsta2, Entpd5, Gm10639, Gm3776, Pir, Lactb2, Ppif, Psmb9, Zdhhc2, Aldh1a7, Pcx, Pten, Ehf, Stk32a, Amhr2, Usp2, Triap1, Cav1, Hmgcs2, Plb1, Ero1l, 5330417c22rik, Usp3, Adh1, Pm20d1, Erlin2, Gls2, Dysf, Fdx1, Akr1c14, Plgrkt, Ptgis, Nrk, E2f2, Pawr, Frk, Trmt10a, Gsta1, Anxa9, Eepd1, Cdkl3, Nr0b2, Hagh, Efemp1, Nr2f2	2.47E–34	111
primary metabolic process (GO:0044238)	Smpdl3b, Slc25a33, Lgmn, Nedd9, Ssu72, Nmrk1, Asah2, Cited4, Thra, Tprkb, Pipox, Afmid, Gins1, Star, Acacb, Casp7, Ecm1, Tmem38b, Adam23, Gucy1b2, Pxk, Slc40a1, Slc11a1, Aldh1a1, Eno3, Gpt, Gstt1, Akr1c18, Ehhadh, Wnt6, Pdgfd, Tns2, Serpinb6b, Ugt1a2, Gcnt4, Cdkn1a, Bc025920, Avpi1, Slc7a11, Prlr, Gpcpd1, Tspan5, Inha, Hsd3b1, Nol3, Pi4k2b, Hao2, Htra1, Prss35, Cdkn3, Rfk, Kdm7a, Rnf5, Serpinb1a, Myc, A4galt, Dhdh, Hmox1, Lrp8, Cd109, Klk1b21, C2, Cdon, Fam135b, Entpd5, Pir, Lactb2, Ppif, Psmb9, Zdhhc2, Pcx, Pten, Ehf, Stk32a, Amhr2, Usp2, Triap1, Cav1, Hmgcs2, Plb1, Ero1l, Usp3, Adh1, Pm20d1, Adam12, Erlin2, Gls2, Dysf, Fdx1, Akr1c14, Plgrkt, Ptgis, Nrk, E2f2, Pawr, Frk, Trmt10a, Gsta1, Anxa9, Eepd1, Cdkl3, Klk1b22, Nr0b2, Lsr, Efemp1, Nr2f2	1.65E–30	106
nitrogen compound metabolic process (GO:0006807)	Smpdl3b, Slc25a33, Lgmn, Nedd9, Blvrb, Ssu72, Nmrk1, Asah2, Cited4, Thra, Tprkb, Pipox, Afmid, Gins1, Acacb, Casp7, Ecm1, Adam23, Gsta3, Gucy1b2, Maoa, Pxk, Slc40a1, Slc11a1, Eno3, Gpt, Gstt1, Akr1c18, Ehhadh, Wnt6, Pdgfd, Tns2, Serpinb6b, Gcnt4, Cdkn1a, Bc025920, Avpi1, Slc7a11, Prlr, Spta1, Tspan5, Inha, Nol3, Hao2, Htra1, Prss35, Cdkn3, Rfk, Kdm7a, Rnf5, Serpinb1a, Myc, A4galt, Hmox1, Lrp8, Cd109, Klk1b21, Gsta4, C2, Cdon, Gsta2, Entpd5, Gm10639, Gm3776, Pir, Lactb2, Ppif, Psmb9, Zdhhc2, Pten, Ehf, Stk32a, Amhr2, Usp2, Triap1, Cav1, Hmgcs2, Ero1l, Usp3, Pm20d1, Adam12, Erlin2, Gls2, Dysf, Akr1c14, Plgrkt, Ptgis, Nrk, E2f2, Pawr, Frk, Trmt10a, Gsta1, Anxa9, Eepd1, Cdkl3, Klk1b22, Nr0b2, Hagh, Efemp1, Nr2f2	1.27E–28	101
organonitrogen compound metabolic process (GO:1901564)	Cdkn3, Rfk, Smpdl3b, Kdm7a, Rnf5, Serpinb1a, Myc, Lgmn, Nedd9, A4galt, Blvrb, Hmox1, Ssu72, Lrp8, Nmrk1, Cd109, Asah2, Klk1b21, Gsta4, C2, Cdon, Gsta2, Entpd5, Pipox, Afmid, Gm10639, Gm3776, Acacb, Casp7, Ecm1, Ppif, Adam23, Psmb9, Zdhhc2, Pten, Gsta3, Gucy1b2, Stk32a, Amhr2, Maoa, Usp2, Triap1, Cav1, Hmgcs2, Pxk, Ero1l, Slc11a1, Usp3, Eno3, Gpt, Pm20d1, Adam12, Erlin2, Gstt1, Ehhadh, Gls2, Dysf, Pdgfd, Tns2, Serpinb6b, Plgrkt, Gcnt4, Nrk, Cdkn1a, Pawr, Frk, Avpi1, Slc7a11, Prlr, Spta1, Gsta1, Tspan5, Inha, Anxa9, Nol3, Cdkl3, Klk1b22, Hagh, Hao2, Htra1, Prss35, Efemp1, Nr2f2	2.67E–28	83
response to stimulus (GO:0050896)	Sdk1, Smpdl3b, Slc25a33, Lgmn, Asah2, Cited4, Thra, Abcb1a, Star, Acacb, Matn2, Casp7, Ecm1, Rhpn2, Tmem38b, Adam23, Lgals3, Gsta3, Gabrg1, Gucy1b2, Maoa, Pxk, Slc40a1, Slc11a1, Aldh1a1, Eno3, Rasgrf2, Gstt1, Akr1c18, Wnt6, Tap1, Pdgfd, Tns2, Ugt1a2, Srxn1, Cdkn1a, Avpi1, Slc7a11, Prlr, Tspan5, Inha, Hsd3b1, Nol3, Igsf11, Htra1, Rxfp1, Syt12, Cdkn3, Rnf5, Serpinb1a, Crip1, Myc, Cd274, Nqo1, Hmox1, Lrp8, Cd109, Gsta4, C2, Grasp, Cdon, Gsta2, Entpd5, Gm10639, Gm3776, Ppif, Txndc12, Pten, Stk32a, Amhr2, Usp2, Triap1, Cav1, Ero1l, 5330417c22rik, Usp3, Adh1, Erlin2, Paqr7, Dysf, Fdx1, Bik, Slc47a1, Plgrkt, Osmr, Ptgis, Nrk, Lynx1, E2f2, Pawr, Pkd2l2, Frk, Igfbp6, Gsta1, Anxa9, Eepd1, Nr0b2, Efemp1, Nr2f2	9.28E–24	99
biological regulation (GO:0065007)	Sdk1, Smpdl3b, Slc25a33, Lgmn, Nedd9, Psmg2, Ssu72, Asah2, Cited4, Thra, Abcb1a, Star, Acacb, Ecm1, Rhpn2, Tmem38b, Lgals3, Gabrg1, Gucy1b2, Maoa, Pxk, Slc40a1, Slc11a1, Aldh1a1, Rasgrf2, Akr1c18, Wnt6, Tap1, Hephl1, Pdgfd, Tns2, Serpinb6b, Aqp11, Gcnt4, Cdkn1a, Car7, Bc025920, Avpi1, Slc7a11, Prlr, Dram1, Spta1, Tspan5, Inha, Hsd3b1, Nol3, Igsf11, Htra1, Tlcd2, Rxfp1, Syt12, Cdkn3, Rfk, Kdm7a, Rnf5, Serpinb1a, Myc, Crip1, Cd274, Nqo1, Hmox1, Lrp8, Cd109, Klk1b21, C2, Grasp, Cdon, Entpd5, Pir, Cdhr5, Ppif, Txndc12, Psmb9, Pten, Ehf, Stk32a, Amhr2, Usp2, Triap1, Cav1, Plb1, Ero1l, 5330417c22rik, Usp3, Adh1, Pm20d1, Adam12, Erlin2, Paqr7, Gls2, Dysf, Fdx1, Bik, Akr1c14, Plgrkt, Osmr, Ptgis, Nrk, E2f2, Pawr, Lynx1, Frk, Igfbp6, Anxa9, Cdkl3, Klk1b22, Nr0b2, Lsr, Efemp1, Nr2f2	7.06E–23	110
regulation of biological quality (GO:0065008)	Syt12, Slc25a33, Rnf5, Serpinb1a, Myc, Lgmn, Hmox1, Lrp8, Klk1b21, Grasp, Thra, Abcb1a, Star, Acacb, Cdhr5, Ppif, Tmem38b, Txndc12, Pten, Gabrg1, Maoa, Usp2, Triap1, Cav1, Pxk, Plb1, Ero1l, Slc40a1, Slc11a1, Usp3, Aldh1a1, Pm20d1, Adh1, Akr1c18, Fdx1, Dysf, Hephl1, Tns2, Akr1c14, Aqp11, Gcnt4, Ptgis, Pawr, Car7, Slc7a11, Prlr, Spta1, Inha, Hsd3b1, Anxa9, Nol3, Cdkl3, Klk1b22, Nr0b2, Igsf11, Lsr, Tlcd2, Nr2f2	1.1E–20	58
cellular response to chemical stimulus (GO:0070887)	Rxfp1, Syt12, Slc25a33, Rnf5, Crip1, Lgmn, Myc, Cd274, Nqo1, Hmox1, Lrp8, Cd109, Asah2, Gsta4, Gsta2, Thra, Gm10639, Star, Gm3776, Casp7, Ecm1, Ppif, Tmem38b, Adam23, Pten, Lgals3, Gsta3, Amhr2, Cav1, Ero1l, Slc40a1, Paqr7, Gstt1, Akr1c18, Fdx1, Dysf, Pdgfd, Osmr, Ugt1a2, Srxn1, Ptgis, Pawr, Slc7a11, Prlr, Gsta1, Nol3, Nr0b2, Htra1, Nr2f2	2.82E–20	49
cellular response to stimulus (GO:0051716)	Rxfp1, Syt12, Cdkn3, Smpdl3b, Slc25a33, Rnf5, Crip1, Myc, Lgmn, Cd274, Nqo1, Hmox1, Lrp8, Cd109, Asah2, Gsta4, Grasp, Cdon, Gsta2, Thra, Entpd5, Gm10639, Star, Gm3776, Matn2, Casp7, Ecm1, Rhpn2, Ppif, Tmem38b, Txndc12, Adam23, Pten, Lgals3, Gsta3, Gabrg1, Gucy1b2, Stk32a, Amhr2, Maoa, Triap1, Cav1, Ero1l, Slc40a1, Slc11a1, 5330417c22rik, Usp3, Rasgrf2, Erlin2, Paqr7, Gstt1, Akr1c18, Wnt6, Dysf, Fdx1, Pdgfd, Tns2, Bik, Osmr, Ugt1a2, Srxn1, Ptgis, Nrk, Cdkn1a, Lynx1, E2f2, Pawr, Frk, Igfbp6, Avpi1, Slc7a11, Prlr, Gsta1, Tspan5, Inha, Eepd1, Nol3, Nr0b2, Igsf11, Htra1, Efemp1, Nr2f2	6.73E–19	82
regulation of biological process (GO:0050789)	Sdk1, Smpdl3b, Slc25a33, Lgmn, Nedd9, Psmg2, Ssu72, Asah2, Cited4, Thra, Abcb1a, Star, Acacb, Ecm1, Rhpn2, Tmem38b, Lgals3, Gabrg1, Gucy1b2, Maoa, Pxk, Slc40a1, Slc11a1, Aldh1a1, Rasgrf2, Akr1c18, Wnt6, Tap1, Pdgfd, Tns2, Serpinb6b, Cdkn1a, Car7, Bc025920, Avpi1, Slc7a11, Prlr, Dram1, Spta1, Tspan5, Inha, Nol3, Igsf11, Htra1, Rxfp1, Syt12, Cdkn3, Rfk, Kdm7a, Rnf5, Serpinb1a, Crip1, Myc, Cd274, Nqo1, Hmox1, Lrp8, Cd109, C2, Grasp, Cdon, Entpd5, Pir, Cdhr5, Ppif, Txndc12, Psmb9, Pten, Ehf, Stk32a, Amhr2, Usp2, Triap1, Cav1, Plb1, Ero1l, 5330417c22rik, Usp3, Pm20d1, Adam12, Erlin2, Paqr7, Gls2, Dysf, Fdx1, Bik, Plgrkt, Osmr, Ptgis, Nrk, Lynx1, E2f2, Pawr, Frk, Igfbp6, Anxa9, Cdkl3, Nr0b2, Lsr, Efemp1, Nr2f2	1.86E–18	101
developmental process (GO:0032502)	Sdk1, Rxfp1, Kdm7a, Tbata, Myc, Lgmn, Nedd9, Cd274, Nqo1, Hmox1, Lrp8, Cd109, Cdon, Thra, Abcb1a, Gins1, Star, Pir, Acacb, Matn2, Casp7, Ecm1, Cdhr5, Tmem38b, Pten, Lgals3, Gsta3, Ehf, Amhr2, Usp2, Cav1, Ero1l, Slc40a1, Aldh1a1, Slc25a34, Eno3, Adam12, Paqr7, Wnt6, Dysf, Pdgfd, Tns2, Bik, Serpinb6b, Akr1c14, Aqp11, Gcnt4, Ptgis, Nrk, Cdkn1a, E2f2, Pawr, Frk, Slc7a11, Prlr, Gpcpd1, Spta1, Inha, Hsd3b1, Anxa9, Nol3, Cdkl3, Nr0b2, Lsr, Htra1, Efemp1, Nr2f2	5.91E–18	67
response to chemical (GO:0042221)	Sdk1, Rxfp1, Syt12, Slc25a33, Rnf5, Crip1, Myc, Lgmn, Cd274, Nqo1, Hmox1, Lrp8, Cd109, Asah2, Gsta4, C2, Gsta2, Thra, Abcb1a, Gm10639, Star, Gm3776, Acacb, Matn2, Casp7, Ecm1, Ppif, Tmem38b, Adam23, Pten, Lgals3, Gsta3, Amhr2, Cav1, Ero1l, Slc40a1, Slc11a1, Eno3, Aldh1a1, Adh1, Erlin2, Paqr7, Gstt1, Akr1c18, Fdx1, Dysf, Pdgfd, Slc47a1, Plgrkt, Osmr, Ugt1a2, Srxn1, Ptgis, Cdkn1a, Pawr, Slc7a11, Prlr, Gsta1, Hsd3b1, Nol3, Nr0b2, Htra1, Nr2f2	6.44E–18	63
animal organ development (GO:0048513)	Sdk1, Rxfp1, Kdm7a, Myc, Hmox1, Lrp8, Cd109, Cdon, Thra, Abcb1a, Star, Pir, Acacb, Matn2, Casp7, Ecm1, Tmem38b, Pten, Gsta3, Amhr2, Usp2, Cav1, Ero1l, Slc40a1, Aldh1a1, Wnt6, Pdgfd, Tns2, Bik, Serpinb6b, Akr1c14, Aqp11, Gcnt4, Ptgis, Nrk, Cdkn1a, Pawr, Slc7a11, Prlr, Gpcpd1, Spta1, Inha, Hsd3b1, Anxa9, Nr0b2, Lsr, Htra1, Efemp1, Nr2f2	9.67E–18	49
cofactor metabolic process (GO:0051186)	Pipox, Afmid, Gm10639, Gm3776, Acacb, Rfk, Slc7a11, Gstt1, Akr1c18, Ehhadh, Blvrb, Hmox1, Spta1, Gsta1, Nmrk1, Akr1c14, Gsta4, Hagh, Gsta3, Cbr3, Gsta2, Hmgcs2	2.71E–17	22
anatomical structure development (GO:0048856)	Sdk1, Rxfp1, Kdm7a, Tbata, Myc, Nedd9, Cd274, Hmox1, Lrp8, Cd109, Cdon, Thra, Abcb1a, Gins1, Star, Pir, Acacb, Matn2, Casp7, Ecm1, Tmem38b, Pten, Lgals3, Ehf, Gsta3, Amhr2, Usp2, Cav1, Ero1l, Slc40a1, Eno3, Aldh1a1, Slc25a34, Adam12, Paqr7, Wnt6, Dysf, Pdgfd, Tns2, Bik, Serpinb6b, Akr1c14, Aqp11, Gcnt4, Ptgis, Nrk, Cdkn1a, E2f2, Pawr, Slc7a11, Prlr, Gpcpd1, Spta1, Inha, Hsd3b1, Anxa9, Nol3, Cdkl3, Nr0b2, Lsr, Htra1, Efemp1, Nr2f2	9.01E–17	63
small molecule metabolic process (GO:0044281)	Aldh1a1, Eno3, Rfk, Adh1, Gpt, Pm20d1, Erlin2, Myc, Akr1c18, Ehhadh, Gls2, Dhdh, Fdx1, Dysf, Nmrk1, Asah2, Akr1c14, Ugt1a2, Entpd5, Ptgis, Pipox, Afmid, Star, Acacb, Slc7a11, Gsta1, Aldh1a7, Pcx, Pten, Hagh, Gucy1b2, Hao2, Cbr3, Cav1, Hmgcs2, Plb1, Ero1l	3.41E–16	37

P-value <0.0001. GO ID-Gene Ontology unique seven-digit identifier.

The uncovered BPs associate a gene to a particular process without demonstrating a relationship between the two. To better delineate the roles of COUP-TFII in Leydig cells, the list of 205 downregulated genes was further investigated using g:Profiler to identify which REACTOME pathways are affected (Figure 3 and Table 4) in COUP-TFII-depleted MA-10 cells. Use of REACTOME databases provides a highly detailed representation of the pathways at the molecular level. As shown in Table 4, 34 REACTOME pathways were enriched when the significance threshold level was set at 0.1. To illustrate the most probable pathways impaired in COUP-TFII-depleted MA-10 cells, the pathways were sorted based on the FDR values (the smaller the value, the higher the probability). Table 4 shows that several metabolic pathways were affected, including metabolism, lipid metabolism, steroid metabolism, biological oxidations, and signal transduction, all of which are consistent with a disruption in the steroidogenic function of the cell. Although the g:Profiler REACTOME approach identifies the pathways that are impaired in COUP-TFII-depleted Leydig cells, it does not represent the complex interrelationship between them. To get visual insights into pathway interactions, the Cytoscape plugin EnrichmentMap was used. As shown in Figure 3, COUP-TFII was found to be involved in general REACTOME pathways implicated in generic transcription pathway, RNA polymerase II transcription, signal transduction, and metabolism (Figure 3, blue ovals). Consistent with previous reports implicating COUP-TFII in Leydig cell function [17, 21–23], the REACTOME pathways involved in steroidogenesis were observed to be affected in COUP-TFII-depleted Leydig cells (Figure 3, green oval). These included metabolism of steroid hormones, metabolism of steroids, and metabolism of lipids, which are well linked to each other (Figure 3, green oval). The REACTOME pathways involved in small molecule membrane transport are grouped by a purple oval. Additional REACTOME pathways involved in cellular defense, which include bile acid metabolism (bile acid is produced from cholesterol) are identified by red ovals in Figure 3. Taken together, these data demonstrate a role for COUP-TFII in steroidogenesis and more generally lipid metabolism in Leydig cells.

Figure 3.

Depletion of COUP-TFII in MA-10 Leydig cells affects genes involved in biological pathways related to steroidogenesis. Enriched REACTOME pathways were obtained using g:Profiler from the list of downregulated genes with a significance threshold level (gSCS) set at 0.1. The interacting network graph was generated with the Cytoscape plugin EnrichmentMap using an FDR q-value cutoff of 0.1. The REACTOME root term node was removed from the network graph. The node size represents the number of genes grouped within a particular pathway.

Table 4.

Enriched pathways from differentially expressed genes

Description (REACTOME ID)	Genes	False discovery rate	Gene count
REACTOME root term (0000000)	Pten, Cd109, Gsta3, Plb1, Lgmn, Cav1, Hsd3b1, Hmox1, Slc7a11, Akr1c18, Slc40a1, Adam23, Star, Nqo1, Akr1c14, Prlr, Cbr3, Htra1, Gsta1, Aldh1a1, Kdm7a, Inha, Frk, Rxfp1, Wnt6, Slc44a3, Pi4k2b, Adam12, Ero1l, Serpinb1a, Igfbp6, Ecm1, Blvrb, Abcb1a, Cdon, Hao2, Ugt1a2, Adh1, Entpd5, Cmbl, Amhr2, Erlin2, Casp7, Ssu72, Tap1, Usp2, Spta1, Cited4, Zdhhc2, Slc26a7, Klk1b22, Aqp11, Rhpn2, Tuba8, Pir, Ids, Cd274, Klk1b21, Usp3, Car7, Gls2, Fam13a, Sdk1, Ptgis, Rnf5, Gcnt4, Clcn5, Asah2, Rps19bp1, Maoa, Dysf, Myc, Nr0b2, Acacb, Cdkn1a, Triap1, Fdx1, Gstt1, Slc11a1, Hmgcs2, Lrp8, C2, Gins1, Thra, Pdgfd, Nuf2, Pipox, Nmrk1, Rfk, Bc025920, Slc6a15	4.32E–28	96
Metabolism (R-MMU-1430728)	Blvrb, Gpt, Lrp8, Pipox, Cav1, Fdx1, Car7, Gsta1, Gstt1, Adh1, Nmral1, Aldh1a1, Eno3, Maoa, Ugt1a2, Akr1c18, Pten, Ehhadh, Hao2, Hsd3b1, Hmgcs2, Entpd5, Gls2, Asah2, Ptgis, Rfk, Plb1, Star, Ids, Cmbl, Slc44a3, Cbr3, Hmox1, Pi4k2b, Gsta3, Nqo1, Akr1c14, Nmrk1, Acacb	2.57E–20	39
Biological oxidations (R-MMU-211859)	Fdx1, Cbr3, Maoa, Ugt1a2, Gsta3, Gsta1, Ptgis, Gstt1, Adh1, Cmbl, Aldh1a1	2.61E–09	11
Metabolism of lipids (R-MMU-556833)	Asah2, Fdx1, Ptgis, Plb1, Star, Slc44a3, Pi4k2b, Akr1c18, Pten, Hao2, Ehhadh, Hsd3b1, Hmgcs2, Akr1c14, Acacb	1.71E–07	15
Phase I-Functionalization of compounds (R-MMU-211945)	Fdx1, Cbr3, Maoa, Ptgis, Adh1, Cmbl, Aldh1a1	4.37E–06	7
Metabolism of vitamins and cofactors (R-MMU-196854)	Lrp8, Akr1c18, Ptgis, Rfk, Plb1, Akr1c14, Nmrk1, Acacb	1.27E–05	8
Transport of small molecules (R-MMU-382551)	Slc47a1, Slc40a1, Slc6a15, Erlin2, Slc44a3, Slc11a1, Slc26a7, Abcb1a, Hmox1, Slc7a11, Clcn5, Aqp11, Rnf5	2.52E–05	13
Metabolism of steroids (R-MMU-8957322)	Fdx1, Akr1c18, Ptgis, Hsd3b1, Akr1c14, Star	4.67E–05	6
Retinoid metabolism and transport (R-MMU-975634)	Lrp8, Akr1c18, Plb1, Akr1c14	1.98E–04	4
Metabolism of fat-soluble vitamins (R-MMU-6806667)	Lrp8, Akr1c18, Plb1, Akr1c14	2.79E–04	4
Signal transduction (R-MMU-162582)	Lrp8, Cav1, Nuf2, Myc, Rhpn2, Plb1, Frk, Adh1, Rxfp1, Cdon, Fam13a, Aldh1a1, Tuba8, Pdgfd, Amhr2, Usp2, Akr1c18, Pten, Cdkn1a, Wnt6, Akr1c14, Spta1	3.04E–04	22
Transport of bile salts and organic acids, metal ions, and amine compounds (R-MMU-425366)	Slc11a1, Slc47a1, Slc40a1, Slc6a15, Slc44a3	0.001368	5
SLC-mediated transmembrane transport (R-MMU-425407)	Slc11a1, Slc26a7, Slc47a1, Slc40a1, Slc7a11, Slc6a15, Slc44a3	0.001925	7
Visual phototransduction (R-MMU-2187338)	Lrp8, Akr1c18, Plb1, Akr1c14	0.001931	4
Heme degradation (R-MMU-189483)	Blvrb, Hmox1	0.002755	2
Basigin interactions (R-MMU-210991)	Cav1, Slc7a11	0.003703	2
Phospholipid metabolism (R-MMU-1483257)	Pi4k2b, Pten, Plb1, Slc44a3	0.003775	4
Synthesis of bile acids and bile salts via 7alpha-hydroxycholesterol (R-MMU-193368)	Akr1c18, Ptgis, Akr1c14	0.006214	3
Glutathione conjugation (R-MMU-156590)	Gsta3, Gsta1, Gstt1	0.016312	3
Synthesis of bile acids and bile salts (R-MMU-192105)	Akr1c18, Ptgis, Akr1c14	0.017396	3
Ethanol oxidation (R-MMU-71384)	Adh1, Aldh1a1	0.017921	2
Metabolism of steroid hormones (R-MMU-196071)	Fdx1, Hsd3b1, Star	0.022533	3
Phase II-Conjugation of compounds (R-MMU-156580)	Ugt1a2, Gsta3, Gsta1, Gstt1	0.026377	4
Metabolism of porphyrins (R-MMU-189445)	Blvrb, Hmox1	0.028728	2
Bile acid and bile salt metabolism (R-MMU-194068)	Akr1c18, Ptgis, Akr1c14	0.030166	3
Regulation of PTEN stability and activity (R-MMU-8948751)	Pten, Frk	0.031781	2
Iron uptake and transport (R-MMU-917937)	Slc40a1, Hmox1	0.037622	2
Regulation of PTEN localization (R-MMU-8948747)	Pten	0.040824	1
RNA polymerase II transcription (R-MMU-73857)	Nr0b2, Bc025920, Usp2, Myc, Triap1, Cdkn1a, Thra, Ssu72, Cited4, Gls2	0.058265	10
ABC-family proteins–mediated transport (R-MMU-382556)	Abcb1a, Rnf5, Erlin2	0.066937	3
RA biosynthesis pathway (R-MMU-5365859)	Adh1, Aldh1a1	0.067602	2
Protein localization (R-MMU-9609507)	Pipox, Hmox1, Hao2, Ehhadh	0.076884	4
Metabolism of water-soluble vitamins and cofactors (R-MMU-196849)	Ptgis, Rfk, Nmrk1, Acacb	0.082467	4
Hemostasis (R-MMU-109582)	Tuba8, Lrp8, Cav1, Cd109, Slc7a11, Ecm1	0.083591	6
Generic transcription pathway (R-MMU-212436)	Nr0b2, Bc025920, Usp2, Myc, Triap1, Cdkn1a, Thra, Cited4, Gls2	0.086909	9

REACTOME ID-REACTOME pathway unique identifier.

Validation of microarray results via qPCR

In silico analyses of the microarray data provided significant evidence demonstrating that several genes and biological processes were impaired by depletion of COUP-TFII in MA-10 Leydig cells. To validate the variation in mRNA levels observed with the microarray data, a series of genes involved in steroidogenesis (Figure 4A), cholesterol transport (Figure 4B), metabolic process (Figure 4C), lipid metabolic process (Figure 4D), and male gonad development (Figure 4E) were selected for further investigation by RT-qPCR. The box and whiskers plots located above the RT-qPCR graphs represent the microarray fold change where the microarray data for the selected genes was normalized to each of the three control samples individually. As shown in Figure 4 (bottom graphs), mRNA levels for Star, Prlr, Cyp11a1, Lhcgr, Gsta3, Fdx1, Scarb1, Inha, Hsd3b1, Nr0b2, and Amhr2 were significantly reduced by 47%, 32%, 33%, 14%, 38%, 16%, 40%, 82%, 50%, 60%, and 55%, respectively, which is comparable to the calculated microarray fold changes obtained with the Partek analysis of the microarray data (Figure 4 box and whiskers plots above the graphs, and Table 5). Even though the transcript ID for Insl3 gene was not available on the gene chip used in our study, we confirmed that Insl3 mRNA levels are decreased by 25% in COUP-TFII-depleted MA-10 cells (Figure 4E), as expected [21]. From the list of upregulated genes from the microarray data (Table 5), we chose to validate Pdgfra that had a fold change +1.19. The results from RT-qPCR confirmed that the mRNA level of Pdgfra was indeed increased by 64% (Figure 4E). Next, we chose Pde8a with a fold change of −1.12, which was outside of our initial screening criteria (difference greater than absolute 1.3 fold). The mRNA level of Pde8a was found to be significantly reduced by 17% in the qPCR experiments (Figure 4A), therefore suggesting that genes with absolute fold changes as low as 1.12 may be considered.

Figure 4.

COUP-TFII regulates expression of genes involved in steroidogenesis (A), cholesterol transport (B), metabolic process (C), lipid metabolic process (D), and male gonad development (E). MA-10 Leydig cells were transfected with siRNA Ctrl or siRNA against COUP-TFII for 48 h and total RNA was isolated. For each selected gene, the fold change based on the microarray data from the siRNA COUP-TFII samples normalized to each of the three siRNA Ctrl individually is presented in the box and whiskers plots (top panels). The dash line represents the ±1.3-fold cutoff that was used in our analysis. Total RNA was reverse transcribed and selected genes quantified by qPCR as indicated (bottom panel). Relative mRNA levels were normalized to Rpl19 and data are plotted as mean ± SEM of 3–11 independent experiments (n). The black bars represent gene expression in MA-10 Leydig cells transfected with siRNA against COUP-TFII normalized to cells transfected with siRNA Ctrl (white bars). The genes shown in bold correspond to genes previously known to be affected by COUP-TFII depletion [17, 20, 21, 23]. The P-values and the calculated percent changes are indicated above the bars. (^***P < 0.001, ^**P < 0.01, ^*P < 0.05). n/a: not available.

Table 5.

List of genes chosen for further validation

Process	Gene symbol	Gene name	Microarray data		qPCR data
			Fold-change	P-value	Fold-change	P-value
Steroidogenesis	Star	Steroidogenic acute regulatory protein	−1.72	0.000	−1.89	0.010
	Prlr	Prolactin receptor	−1.65	0.001	−1.47	0.002
	Inha	Inhibin alpha	−1.48	0.001	−5.56	(n/a)
	Amhr2	Anti-Mullerian hormone type 2 receptor	−1.40	0.000	−2.22	0.009
	Lhcgr	Luteinizing hormone/choriogonadotropin receptor	−1.28	0.001	−1.16	0.001
	Cyp11a1	Cytochrome P450, family 11, subfamily a, polypeptide 1	−1.26	0.004	−1.49	0.003
	Hsd3b1	Hydroxy-delta-5-steroid dehydrogenase, 3 beta- and steroid delta-isomerase 1	−1.96	0.010	−2.00	(n/a)
	Pde8a	Phosphodiesterase 8A	−1.12	0.005	−1.21	0.043
Cholesterol transport	Scarb1	Scavenger receptor class B, member 1	−1.14	0.001	−1.67	0.000
Metabolic process	Gsta3	Glutathione S-transferase, alpha 3	−2.03	0.000	−1.61	0.000
Lipid metabolic process	Fdx1	Ferredoxin 1	−1.30	0.000	−1.19	0.003
Male gonad development	Pdgfra	Platelet-derived growth factor receptor, alpha polypeptide	1.19	0.074	1.64	0.016
	Nr0b2	Nuclear receptor subfamily 0, group B, member 2	−1.36	0.002	−2.50	0.015
	Insl3	Insulin-like 3	n/a	n/a	−1.33	0.02

n/a: not available.

GSTA3 and STAR protein levels are reduced in COUP-TFII-depleted MA-10 cells

We next performed Western blots to determine whether protein levels of potential COUP-TFII targets are affected in COUP-TFII-depleted MA-10 Leydig cells. Representative Western blots are shown in Figure 5A, while quantifications are presented in Figure 5B. STAR protein levels, which were previously shown to be reduced in COUP-TFII-depleted MA-10 cells [17], were used as a control and found to be reduced by 58% as expected. GSTA3 protein levels were also found to be significantly reduced by 34% (Figure 5). Although Amhr2 mRNA levels were reduced in COUP-TFII-depleted MA-10 cells (Figure 4A and [23]), there was no significant difference in AMHR2 protein levels (Figure 5).

Figure 5.

Depletion of COUP-TFII in MA-10 Leydig cells decreases GSTA3 and STAR protein levels. MA-10 Leydig cells were transfected with siRNA Ctrl or siRNA against COUP-TFII for 48 h and total proteins were isolated. Twenty micrograms of total protein extracts from cells transfected with either control siRNA (lane 1) or siRNA targeting COUP-TFII (lane 2) were loaded per lane. (A) Representative Western blot images from three independent experiments are shown for STAR, GSTA3, and AMHR2. GAPDH was used as a loading control. (B) Western blot images were analyzed using Image Lab 6.0 (Bio-Rad Laboratories) to quantify protein levels. Relative protein levels were normalized to GAPDH. STAR, known to be reduced in COUP-TFII-depleted MA-10 cells [17], was used as a control (shown in bold). The P-values are indicated above the bars (^**P < 0.01, ^*P < 0.05).

Motif discovery

COUP-TFII regulates gene expression by binding most often directly to nuclear response elements (AGGTCA, NRE) and direct repeats (AGGTCANAGGTCA, DR1), or their variants, that are present in the promoter region of target genes (reviewed in [2]). Therefore, it may be assumed that COUP-TFII can modulate the expression of the differentially expressed genes in COUP-TFII-depleted MA-10 Leydig cells through NRE, DR1, and their variants. We next determined if a subset of the modulated genes contain a DR1 or NRE by performing in silico analysis of the −1000/+50 bp region of their gene promoters. The promoter region of each gene analyzed, the matched sequences and their locations, and the P- and q-values are listed in Table 6. We screened the promoter regions for RGGYCARAKRYMV (DR1) and RGGYCA (NRE) motifs. As shown in Table 6, Pdgfra was the only gene that did not contain any potential DR1 or NRE in its proximal promoter region. Potential DR1 and NRE elements were detected, however, in the promoter of the Prlr, Inha, Amhr2, Lhcgr, Cyp11a1, Hsd3b1, Pde8a, Scarb1, Gsta3, Fdx1, and Nr0b2 genes (Table 6), in addition to the previously characterized DR1 in the Star promoter [17]. Based on the presence of DR1s or NREs, one can speculate that these genes are direct targets of COUP-TFII in MA-10 Leydig cells.

COUP-TFII activates the mouse Gsta3 and Inha promoters

To investigate whether COUP-TFII directly activates gene transcription via the predicted motifs mapped within a proximal promoter region, we performed functional promoter luciferase assays. We chose to investigate the Gsta3 gene, which is involved in metabolic process, as well as the Inha and Hsd3b1 genes that are involved in Leydig cell regulation and function. Since Amhr2 is a recently identified target for COUP-TFII in Leydig cells [23], the mouse Amhr2 promoter was used as a positive control. As expected, the −1486 bp Amhr2 promoter was activated by about 17-fold by COUP-TFII, while the minimal −34 bp reporter, devoid of any COUP-TFII response element, was not significantly activated (Figure 6A) [23].

Figure 6.

COUP-TFII activates the mouse Amhr2, Gsta3, and Inha promoters in MA-10 Leydig cells. MA-10 Leydig cells were cotransfected with either 100 ng of an empty expression vector (−, control, white bars) or 100 ng of a COUP-TFII expression vector (+, black bars), along with 400 ng of (A) mouse Amhr2 promoter constructs (−1486/+74 bp or −34/+74 bp), (B) mouse Gsta3 promoter constructs (−2062/+38 bp or −21/+38 bp), (C) mouse Inha promoter (−2108/+40 bp) or an empty Luc vector, and (D) mouse Hsd3b1 promoter (−1923/+112 bp). The gene shown in bold was previously known to be affected by COUP-TFII depletion [23]. Results are shown as the mean fold activation over control ± SEM from three to seven independent experiments as indicated. COUP-TFII response elements are indicated on each promoter. Gray hexagon: direct repeat 1 (RGGYCARAKRYMV); black ovals: nuclear receptor element (RGGYCA); hatched rectangle: GC-box. An asterisk (^*) represents a statistically significant difference from its control (empty expression vector, value set at 1, black bar over white bar). (^**P < 0.01, ^*P < 0.05).

In mice, members of the glutathione S-transferase A (GSTA) protein family (GSTA1, GSTA2, GSTA3, and GSTA4) are involved in detoxification and reduction of fatty acids and phospholipid hydroperoxides [47]. Besides its well-characterized role in detoxification, GSTA3 is also implicated in testosterone biosynthesis by isomerizing a double bond in Δ⁵-androstene-3,17-dione, a precursor of testosterone [48]. To test whether COUP-TFII activates the mouse Gsta3 promoter, transient transfections were performed in MA-10 Leydig cells using a −2062/+38 bp Gsta3 and a minimal −21/+38 bp Gsta3 reporter lacking putative COUP-TFII binding motifs. As shown in Figure 6B, a 2.6-fold activation was observed in the presence of COUP-TFII on the −2062/+38 bp reporter, while the −21/+38 bp construct was not significantly activated by COUP-TFII. These data are consistent with the fact that the mouse Gsta3 gene contains four predicted DR1 elements (−994/−982 bp, −933/−921, −753/−741 bp, and −267/−255 bp) and three NREs (−808/−803 bp, −753/−748 bp, −549/−544 bp, and −315/−310 bp) within the −1000/+50 bp region (Table 6). Since the −21/+38 bp Gsta3 construct lacks both DR1 and NRE, our data suggest that the DR1 elements and/or NREs may be involved in the COUP-TFII-dependent activation of the Gsta3 promoter.

Inhibin α (Inha) is expressed in Leydig cells and this hormone has been implicated in regulation of the HPG axis (reviewed in [49]). The first −1000 bp of the mouse Inha promoter contains three potential DR1 elements (−741/−729 bp, −735/−723 bp, and −387/−375 bp) and four potential NREs (−756/−751 bp, −116/−111 bp, −111/−106 bp, and −39/−34 bp) (Table 6). As shown in Figure 6C, a −2108/+40 bp fragment of the mouse Inha promoter was activated 3.2-fold in the presence of COUP-TFII, while a promoterless luciferase reporter, used as a negative control, was not affected by COUP-TFII. This result suggests that COUP-TFII activates this promoter construct via the potential DR1 elements and NREs.

In steroidogenic cells, the HSD3B1 enzyme is required for steroid hormone biosynthesis. In silico analysis revealed that the mouse Hsd3b1 gene contains four potential DR1 elements (−674/−662 bp, −552/−540 bp, −298/−286 bp, and +13/+25 bp) and two NREs (−552/−547 bp and + 20/+25 bp) (Table 6). Despite the presence of these potential COUP-TFII response elements, a −1923/+112 bp mouse Hsd3b1 luciferase reporter was not activated by COUP-TFII (Figure 6D).

Discussion

An inducible Coup-tfii global knockout mouse model has been used to show a role for COUP-TFII in Leydig cell differentiation and function [20]. In this model, Leydig cells failed to differentiate fully and were arrested at the progenitor stage [20]. Additionally, their steroid output was reduced [20]. Identifying the molecular pathways and specific target genes governed by COUP-TFII in Leydig cells, however, remain critical areas to completely define its mechanism of action in Leydig cells. Previous studies identified a handful of target genes directly regulated by COUP-TFII in Leydig cells [17, 21–23]. Each of these studies was focused on deciphering the mode of action on a single gene. Our present work provides a broader understanding of the genes regulated by COUP-TFII in Leydig cells. Using a transcriptomic analysis of COUP-TFII-depleted MA-10 Leydig cells, we identified 262 differentially expressed genes, including 205 downregulated genes implicated in steroidogenesis, androgen homeostasis, cell defense, and cell differentiation. An in-depth bioinformatic analysis of our data revealed that COUP-TFII orchestrates more than 20 biological processes and 35 distinct pathways in MA-10 Leydig cells.

COUP-TFII regulates transcription of multiple genes involved in Leydig cell steroidogenesis homeostasis of the hypothalamo-pituitary-gonadal axis

Our current results are consistent with previous findings where COUP-TFII was found to activate steroidogenesis by regulating the expression of the key steroidogenic gene Star in Leydig cells [17]. The results reported herein reveal additional genes encoding steroidogenic enzymes (Cyp11a1, Hsd3b1, Akr1c14, and Gsta3) and a supporting protein (Fdx1) that were reduced by depletion of COUP-TFII. The results are in agreement with findings from inducible Coup-tfii knockout mice where decreased expression of Hsd3b1 and Cyp11a1 was reported in Leydig cells [20].

COUP-TFII typically activates gene expression by binding directly to nuclear response elements (DRs and NREs) (reviewed in [1]) or GC-box elements [23] located within the promoter region of its target genes, and indirectly by interacting with DNA-bound transcription factors (reviewed in [50]). Although the mouse Inha, Hsd3b1, and Gsta3 proximal promoters contain multiple potential COUP-TFII response elements (DR1s and NREs) within the first −1000 bp region, only the Gsta3 and Inha promoters were activated by COUP-TFII. COUP-TFII likely regulates transcription of Gsta3 and Inha by acting through one or more COUP-TFII response elements found in these promoters. The fact that COUP-TFII failed to activate the Hsd3b1 promoter despite the presence of potential response elements is not unique to this transcription factor. It is known that the presence of a transcription factor–binding motif in a promoter sequence does not necessarily mean it is functional. The various members of the nuclear receptor subfamilies tend to have almost identical binding affinities for similar DNA response elements (reviewed in [51, 52]), which often results in competition for DNA binding. Leydig cells are known to express several nuclear receptors (reviewed in [53]). For instance, the nuclear receptors retinoid X receptor alpha (RXRα), expressed in Leydig cells [54, 55], and COUP-TFII exhibit identical affinity for binding to a DR1 element [56]. The element could also be occupied by another protein with a higher affinity for the element. Nuclear receptor SF1 (NR5A1, AD4BP) is highly expressed in Leydig cells (reviewed in [57]) where it regulates expression of some of the same genes as COUP-TFII, such as Star [17] and Amhr2 [58]. COUP-TFII and SF1 were found to compete for the same NRE present in regulatory regions of many genes, including Cyp17a1 [59] and Oxt [60]. Alternately, activation by COUP-TFII might require the binding of additional transcription factors to elements located outside of the promoter fragment used in our assays. Finally, it remains possible that the mouse Hsd3b1 promoter fragment (−1923/+112 bp) used in our assays may not contain an appropriate enhancer element required for COUP-TFII-dependent activation. Indeed, recent evidence has shown that recruitment of COUP-TFII to distal enhancers promotes DNA looping essential for the establishment of cell type–specific transcriptional programs [61]. Furthermore, COUP-TFII was found to co-localize with other pioneer factors including GATA, HNF4α, and estrogen receptor along with key cofactors to drive cell-specific activity [62]. A similar mode of action for COUP-TFII in the regulation of Hsd3b1 in Leydig cells is therefore possible. Additional experiments such as ChIP-Seq to determine the cistrome of COUP-TFII in Leydig cells will be instrumental for identifying genes directly regulated by this nuclear receptor.

Although COUP-TFII was not found to activate a fragment of the Cyp11a1 promoter [17], Cyp11a1 mRNA levels were nonetheless significantly reduced in COUP-TFII-depleted Leydig cells, implying a role for COUP-TFII. The mitochondrial CYP11A1 enzyme is responsible for the conversion of cholesterol into pregnenolone and marks the first step of testosterone synthesis. By donating electrons, FDX1 is considered a critical regulator of CYP11A1 enzymatic activity (reviewed in [63]). Fdx1 mRNA levels were also reduced in the absence of COUP-TFII. Under certain conditions, Leydig cells and Leydig cell lines rely on exogenously available cholesterol esters for steroidogenesis, which is regulated by the scavenger receptor class B type 1 (SRB1 or SR-BI) pathway, a receptor for cholesterol-carrying lipoproteins, encoded by the Scarb1 gene [64, 65]. Interestingly, we found that Scarb1 mRNA levels are significantly reduced in the absence of COUP-TFII, thus suggesting additional essential roles for COUP-TFII in the activation of steroidogenesis. Another important regulator of cholesterol bioavailability for steroidogenesis is the protein STAR and expression of the Star gene and STAR protein levels are significantly reduced in the absence of COUP-TFII (this work and [17]). Diminished SRB1 levels would lead to a reduction in intracellular cholesterol availability, while a decrease in STAR levels reduces the availability of CYP11A1 substrate, which ultimately contributes to a reduction of Cyp11a1 expression.

The nuclear receptor NR0B2 (small heterodimer partner, SHP) was found to be an important repressor of steroidogenesis [66]. NR0B2 acts by suppressing the expression of several steroidogenic genes including Star, Cyp11a1, Hsd3b1, and Cyp17a1 [66, 67]. Interestingly, several of these genes are also targets of COUP-TFII (this work and [8, 17]) raising the possibility that COUP-TFII and NR0B2 might act together in a regulatory pathway in Leydig cells. Supporting this is the fact that in COUP-TFII-depleted MA-10 Leydig cells, Nr0b2 mRNA levels are reduced by ~60%. Although the mechanism of COUP-TFII action in the expression of Nr0b2 remains to be fully elucidated, our data place COUP-TFII upstream of NR0B2 in the cascade of regulators of steroidogenesis. Altogether, these findings further strengthen the role of COUP-TFII as an essential activator of steroidogenesis in Leydig cells.

In addition to producing androgens, Leydig cells also contribute to the homeostasis of the hypothalamo-pituitary-gonadal axis by producing inhibin, a heterodimeric glycoprotein hormone composed of two α and β subunits encoded by the Inha and Inhba genes (reviewed in [49]). Our results have confirmed reduced mRNA levels for both Inha and Inhba genes in COUP-TFII-depleted Leydig cells. We also found that COUP-TFII directly activates the mouse Inha promoter (−2108/+40 bp) in Leydig cells, which most likely involves the predicted DR1 and/or NREs found in that promoter region. Additional work is required to elucidate the exact mechanism of COUP-TFII action in Inha expression and more globally in the homeostasis of the HPG axis.

COUP-TFII: a role in Leydig cell defense

Steroidogenesis produces a high number of reactive oxygen species, which are known to downregulate the expression of essential steroidogenic enzymes [68]. Members of the glutathione S-transferase (GST) superfamily of enzymes are mainly recognized for their roles in cellular detoxification by making electrophilic radical byproducts more suitable for cellular export (reviewed in [69]). Members of the GSTA subfamily, GSTA1, GSTA2, GSTA3, GSTA4, are detected in mouse and human Leydig cells [24, 70, 71]. Our microarray data revealed a significant reduction in the expression of several detoxification genes, including Gsta1, Gsta2, Gsta3, Gsta4, and ATP-binding cassette subfamily B member 1B (Abcb1b), in COUP-TFII-depleted MA-10 Leydig cells. We found that COUP-TFII is required for Gsta3 expression as both Gsta3 mRNA and protein levels are significantly reduced in COUP-TFII-depleted MA-10 Leydig cells. Consistent with this, COUP-TFII specifically activates the Gsta3 promoter in Leydig cells. Taken together, our data suggest that COUP-TFII may have a broader role in regulating the expression of several genes implicated in the oxidative stress response.

Implication of COUP-TFII in Leydig cell proliferation and differentiation

Prolactin, signaling through its receptor PRLR (encoded by the Prlr gene), has a biphasic effect on steroidogenesis in rats [72] as well as in the MA-10 Leydig cell line [73]. At low concentrations, PRL stimulates testosterone biosynthesis, while at high concentrations, it inhibits CYP17A1 activity, thus preventing the conversion of precursors into active androgens [74, 75]. In addition to their roles in steroidogenesis, pituitary hormones PRL and LH, acting via their receptors PRLR and LHCGR, are necessary for proper Leydig cell proliferation and differentiation [76]. In COUP-TFII-depleted Leydig cells, mRNA levels of both Prlr and Lhcgr were significantly reduced. Although the mechanism remains to be determined, these data along with data from the temporal Coup-tfii knockout mice [20] support the implication of COUP-TFII in Leydig cell proliferation and differentiation in addition to steroidogenesis.

In conclusion, our analysis of the transcriptional landscape of COUP-TFII-depleted Leydig cells revealed that this transcription factor is involved in numerous Leydig cell functions by regulating the expression of genes required for proper androgen production that is needed for overall men’s health.

Conflict of interest

Samir Mehanovic, Raifish E. Mendoza-Villarroel, Karine de Mattos, Philippe Talbot, Robert S. Viger, and Jacques J. Tremblay declare that they have no conflicts of interest.

Author contributions

SM performed the majority of the experiments along with REMV, KDM, and PT. JJT conceived the original idea and supervised the project. SM wrote the manuscript with support from JJT and RSV. All authors provided critical feedback and helped shape the research, analysis, and manuscript.

Data availability

All data generated or analyzed during this study are included in this published article or in data repositories.