Transcriptome analysis of the epididymis from Plag1 deficient mice suggests dysregulation of sperm maturation and extracellular matrix genes

The transcription factor pleomorphic adenoma gene 1 (PLAG1) is required for male fertility. Mice deficient in PLAG1 exhibit decreased sperm motility and abnormal epididymal tubule elongation and coiling, indicating impaired sperm maturation during epididymal transit. However, the downstream transcriptomic profile of the Plag1 knockout (KO; Plag1−/−) murine epididymis is currently unknown.


| INTRODUCTION
Pleomorphic adenoma gene 1 (PLAG1) is a highlyconserved zinc-finger transcription factor required for growth, 1 male fertility, 2 normal sperm production and motility, 3 and for normal epididymis development and morphology. 2 Plag1 knockout (KO; Plag1 −/− ) mice exhibit growth retardation during both embryonic and postnatal development, resulting in a 30% decreased size in adulthood. 1 Plag1 −/− mice also present with substantial fertility defects. Conception rates for male Plag1 −/− mice are reduced when paired with Plag1 +/+ females, 1 while we have previously shown in a 5 month breeding analysis that Plag1 −/− males are completely infertile, despite normal plugging behavior. 2 The number of motile spermatozoa present in the semen from the cauda of KO mice was reduced by 49% compared to their Plag1 +/+ counterparts and progressive motility was reduced by 80%. 3 As spermatozoa gain motility during epididymal transit, these results indicate that sperm maturation is substantially impaired in Plag1 −/− mice, however whether these effects are due to cell-autonomous defects in sperm, or nonautonomous defects due to epididymal dysfunction, is currently unknown.
The epididymis is divided into 10 transcriptionally unique segments in mice. [4][5][6] During epididymal sperm maturation, spermatozoa undergo a precise sequence of biochemical and structural modifications, which is facilitated by the tightly regulated transcriptomes of these segments. 4,7,8 The epididymides of Plag1 −/− mice display reduced relative weight, decreased tubule coiling and elongation, and the loss of morphology of the typical bulbous shape of the cauda. 2 It is therefore possible that the fertility defects in Plag1 −/− mice are due to loss of epididymal segment identity resulting in altered gene transcription in the epididymis, and ultimately leading to impaired sperm maturation as spermatozoa transit though the duct. Disruption to gene transcription in the epididymis is known to result in dysfunctional, immature or immotile sperm in various rodent models, 9 and has also been implicated in azoospermatic human male infertility. 10 We hypothesised that the underlying cause of the sperm motility defects observed in Plag1 −/− mice is due to aberrant transcriptional regulation in the Plag1 −/− epididymis. To that end, we sought to characterise the transcriptome of the Plag1 −/− epididymis in order to identify genes and pathways that are dysregulated following deletion of Plag1.
2 | RESULTS 2.1 | Transcriptome analysis of Plag1 −/− and Plag1 +/+ mice mouse epididymis RNA-Seq was used to analyse and compare the transcriptomes of epididymides from 7-week-old Plag1 −/− and Plag1 +/+ mice. Although the epididymis is often divided into 10, transcriptionally distinct segments in mice, the shortened, dysmorphic and aberrantly coiled epididymis in Plag1 −/− mice made it unfeasible to accurately dissect and sequence "match" transcriptionally-analogous segments or regions from Plag1 −/− and Plag1 +/+ epididymides. Rather, we reasoned that analysis of the transcriptome from whole Plag1 −/− epididymides would allow us to accurately determine the overall molecular profile of the Plag1 −/− epididymis. We found that 1728 genes were differentially expressed in the epididymis of Plag1 −/− compared to Plag1 +/+ mice; 1102 genes were upregulated and 626 genes were downregulated ( Figure 1). The top-10 up-and downregulated genes in Plag1 −/− compared to Plag1 +/+ are listed in Supporting Table 2. Additionally, we performed RNA-Seq comparing the transcriptome in epididymides from Plag1 +/− with that of Plag1 +/+ mice, but did not find any differentially regulated genes (Figure 1). For standard validation of the RNA-Seq, Mep1a, Spink14 and Timp2 were arbitrarily chosen from the list of dysregulated genes for confirmation of differential gene expression by quantitative polymerase chain reaction (qPCR), using samples from animals separate to those that were used in the RNA-Seq. Validation of Defb9 expression, which was not differentially expressed in the epididymis from Plag1 −/− mice in the RNA-Seq dataset, was also included. We validated the differential expression of Mep1a, Spink14 and Timp2 in the epididymides of Plag1 −/− mice; fold changes in mRNA as determined by qPCR are shown alongside fold changes as determined by RNA-Seq and mRNA expression levels determined by qPCR, confirming the differential expression of Mep1a, Spink14 and Timp2 (t-test, P < .01) are shown ( Figure 1). RNA-Seq indicated that Defb9 was not differentially expressed in the epididymides of Plag1 −/− mice, and this was confirmed by qPCR (P = .121; Figure 1). Together, these data validate our RNA-Seq analysis as a robust methodology to determine the PLAG1-dependent transcriptome within the murine epididymis.
DAVID functional annotation analysis was performed to identify genes associated with sperm maturation that are dysregulated in the epididymis of Plag1 −/− mice. This analysis revealed that a number of genes involved in sperm maturation, motility, capacitation, the acrosome reaction and the prevention of premature sperm capacitation were dysregulated in Plag1 −/− mice (Tables 1,2). Surprisingly, despite the fact that the epididymis of Plag1 −/− mice is underdeveloped, several cell proliferation genes were significantly upregulated in the epididymis from Plag1 −/− mice. Indeed, GO analysis revealed that the GO term "Regulation of cell proliferation" was overrepresented; 18 genes significantly upregulated in the epididymis from Plag1 −/− mice are categorised under this term (Table 3). Next, we performed GO analysis to identify pathways and processes the were affected in the epididymis of Plag1 −/− mice, which showed an overall downregulation of genes involved in cell migration, cell motility, metalloendopeptidase activity and metallopeptidase activity in Plag1 −/− mice. Additionally, GO analysis also showed an overall upregulation of genes associated with cell adhesion, molecular binding, and extracellular matrix (ECM) components in KOs ( Figure 2). Taken together, these data indicate a defective maintenance and integrity of the epididymis in Plag1 −/− mice. Enriched "biological process", "molecular function" and "cellular component" GO terms and dysregulated genes in the epididymis of Plag1 −/− mice categorised under the GO terms "Metalloendopeptidase activity" and "Extracellular matrix component" and are listed in supporting tables (supporting Tables 3-7). 2.2 | Cell proliferation in the epididymis from Plag1 −/− mice Ki67 staining and quantification was used to determine if the overall upregulation of cell proliferation genes F I G U R E 1 Significantly differentially expressed genes (DEGs) between Plag1 wild-type (WT), heterozygous (HET) and knockout (KO) mouse epididymis (FDR < 0.05) (n = 5 per genotype). A, Overlapping DEGs in the epididymides from Plag1 WT, HET, and KO mice. Red represents downregulated genes and black represents upregulated genes. B, Volcano plot showing fold changes of DEGs in the epididymis between WT and KO Plag1 mice. Red dots indicate significantly differentially expressed genes and blue dots represent nonsignificantly differentially expressed genes. C, Volcano plot showing fold changes of DEGs in the epididymis between KO and HET Plag1 mice. D, Heat map of all 21 910 genes identified from RNA sequencing and their expression across the three genotypes, compared to average expression. Each line represents one gene. Red indicates a positive fold change and blue indicates a negative fold change. E, Fold changes in mRNA expression of selected dysregulated genes in the epididymides from Plag1 knockout compared to wild-type mice (n = 5 per genotype). F, mRNA expression levels of selected dysregulated genes in the epididymides of Plag1 WT and KO mice, determined by qPCR (n = 5 per genotype). Values are normalised to Actb mRNA expression, and means were compared by t tests. **P < 0.01, Defb9, defensin beta 9; Mep1a, meprin A subunit alpha; Spink14, serine peptidase inhibitor Kazal type 14; Timp2, tissue inhibitor of metalloproteinase 2 T A B L E 1 Selection of significantly upregulated genes with functions related to sperm maturation in the epididymis of Plag1 • Can bind to murine sperm and increase motility 40,41 Abbreviations: FDR, false-discovery rate; HET, heterozygous; KO, knockout; LFC, log-fold change KO compared to WT; TPM, transcripts per kilobase million; WT, wild-type.
T A B L E 3 Upregulated genes categorised under the GO term 'Regulation of cell proliferation (GO: 0042127) in the epididymis from  (Table 3) correlates with increased cellular proliferation, in the epididymis of Plag1 −/− mice. The percentage of Ki67 + cells was significantly higher in the stroma of the cauda epididymis from Plag1 −/− compared to Plag1 +/+ mice (t-test, P = .0007), but not in the epithelium (P = .9889) ( Figure 3). There was no difference in the percentage of Ki67 + cells in the epithelium or stroma in the caput and corpus between genotypes (P > .05). There was a significantly higher percentage of stromal cells (relative to the total number of cells) in the caput (t test, P = .0033), corpus (P = .0028) and cauda (P = .0080) of Plag1 −/− mice compared to Plag1 +/+ (Figure 3). These data indicate that the stroma, albeit not the epididymal epithelium of Plag1 −/− mice, is hyperproliferative relative to WT controls.

| PLAG1 binding sites are not enriched in dysregulated genes
To determine if PLAG1 binding sites were enriched in the target genes determined by the RNA-Seq analysis, genes up-or downregulated in the epididymis from Plag1 −/− mice were analyzed for significant enrichment of PLAG1 binding sites using the PLAG1 motif from the JASPAR database and the binding sites defined by Madissoon et al. 42 This motif comprises two G-rich regions (a GRGGC Core and a G-Cluster) separated by 7 nonconserved nucleotides, and broadly takes the form "GA/GGGCNNNNNNNA/GGGG/T", where N is a random nucleotide and the underlined nucleotides are largely invariant. Of all annotated murine genes, 16 163 were found to have at least one PLAG1 binding site in their promoters, whereas 7390 did not have any PLAG1 binding sites, suggesting that the presence of a putative PLAG1 site is extremely commonplace, and that enrichment analysis may present more accurate indication of PLAG1 binding activity. The results of the Homer analysis, however, indicated that among the genes dysregulated in the epididymis from Plag1 −/− mice, there was no significant enrichment of PLAG1 binding sites. Furthermore, the motif scanning did not reveal significant enrichment of binding sites of any other transcription factors, suggesting that the PLAG1-dependent transcriptome in the epididymis does not comprise an over-represented hierarchical pathway of any known transcription factor families.

| DISCUSSION
The cause of male infertility is unidentified in 30% to 45% of clinical cases, 43 exemplifying the lack of knowledge pertaining to the underlying molecular and genetic mechanisms of sperm dysfunction. Several studies using transgenic mice have identified gene candidates involved in poor or abnormal sperm function, including Plag1, required for sperm motility, 3 normal epididymal morphology and male fertility. 2 However, the underlying cause of these defects is unknown and the genes that are controlled by PLAG1 in the epididymis have not yet been characterized. As PLAG1 is expressed in brain and pituitary (as well as elsewhere), we cannot rule out hormonal, metabolic, paracrine signaling or other, nonepididymal contributions to both male infertility and epididymal morphology defects we describe within our study. Nonetheless the abundant widespread expression of PLAG1 in the epithelium and stroma of the adult mouse epididymis 2 points to an important role in male fertility. This study aimed to characterise the epididymal target genes of PLAG1, and to identify processes and functions that may be affected in the epididymis when Plag1 is deficient. RNA-Seq results showed that there are no significantly differentially expressed genes in the epididymis from Plag1 +/− mice compared to Plag1 +/+ , indicating that one copy of the Plag1 gene is sufficient for normal gene expression. This is in agreement with previous findings that Plag1 +/− males have normal fertility, as numbers of litters born and litter size was not affected in the mice. 2 However, Plag1 +/− males do exhibit decreased epididymal tubule elongation and coiling in the caput and corpus epididymis, 2 suggesting that although gene expression does not appear to be significantly dysregulated, the morphology of the epididymis in these mice is altered. RNA-Seq revealed a number of dysregulated genes in the Plag1 −/− epididymis compared to Plag1 +/+ , that have known sperm maturation and storage-related functions (see refs. in Tables 1,2) and common functions are seen among these PLAG1 target genes. As spermatozoa were removed from the tissue samples, these genes identified by the RNA-Seq experiment are exclusively epididymal genes. Several genes, such as Txndc3, Txndc2, Dnajb13, Pebp1, and Rgn, are known to play a role in sperm motility or sperm flagellar development, or have been localized to the developing sperm flagellum, and several KO models of these genes exhibit defective sperm motility (see refs. in Tables 1,2). Txndc3 and Dnajb13 have been implicated in primary ciliary dyskinesia (PCD), a disease in humans resulting from dysfunction of motile cilia in which the majority of male sufferers are infertile. 44 The role of PLAG1 in PCD has not yet been studied, but may be a worthy genetic candidate for investigation, as the emerging diagnostic approach for this disease is genetic testing. 44 Several dysregulated genes in the Plag1 −/− epididymis play a role in regulating the acrosome reaction or sperm capacitation (see refs. in Tables 1,2). Interestingly, this includes the downregulation of two genes that encode decapacitation factors (Pebp1, Spinkl), which bind to the sperm head to prevent premature sperm capacitation (see refs. in Table 2). It is crucial for capacitation to be preserved for when the spermatozoon reaches the ovum; premature capacitation results in redundant spermatozoa that are incapable of fertilization. 45 In addition, previous research has shown that the most severe morphological abnormalities in the epididymis of Plag1 −/− mice are observed in the cauda region (lack of bulbous shape, reduced tubule elongation and coiling) and that in Plag1 +/+ mice, PLAG1 is potentially secreted into the lumen in this region. 2 Altogether, these findings may be indicative of dysfunction of the cauda epididymidis where mature spermatozoa are stored, resulting in premature capacitation and therefore infertility, and future studies should be targeted to examine differential gene expression specifically within this region.
The maintenance of epididymal integrity through the regulation of ion transport also appears to be affected; Atp1a4 and Rgn are dysregulated genes that are related to ion balance within the microenvironment of the epididymal lumen. Atp1a4 (upregulated in the Plag1 −/− epididymis) encodes the α4 subunit of the Na,K-ATPase and is required for sperm fertility due to its role in maintaining sperm ion balance, 23 which is crucial for sperm morphology and motility, and fertilization capacity. [46][47][48] The deletion of this gene in mice results in infertility in males, reduced sperm motility and abnormal sperm tail morphology, and spermatozoa are unable to fertilize eggs in vitro. 23 Rgn (downregulated in Plag1 −/− ) encodes a Ca 2 + -binding protein important for regulating Ca 2+ balance, and overexpression of this gene results in increased luminal Ca 2+ and decreased sperm motility and sperm counts. 36 Dysregulation of these genes may disrupt the movement of ions, sperm ion balance or the epididymal Ca 2+ gradient and contribute to the decreased sperm motility and infertility observed in Plag1 −/− mice, or the dysregulation of these genes in the epididymides from Plag1 −/− mice may be a compensatory mechanism. Interestingly, the expression of several genes such as Txndc3 and Dnajb13 have not been reported in the epididymis previously; Txndc3 expression was thought to be testisspecific and Dnajb13 expression has been detected on developing spermatids. However, as spermatozoa were rinsed from the epididymis tissues, it is unlikely that the expression detected was not epididymal expression.
GO analysis revealed several cell proliferation GO terms enriched in the dataset of significantly upregulated genes in the epididymis from Plag1 −/− mice, which is surprising given that the epididymides in these mice are reduced in weight, the tubule is less developed, 2 and PLAG1 is known to upregulate cell proliferation genes during early development. 49 Ki67 quantification showed that cell proliferation was significantly increased in the cauda stroma of the epididymis from Plag1 −/− mice, but not in the caput or corpus, or in the epithelium of any region. The proportion of stromal cells, out of total cells, was significantly increased in all regions of the epididymis from Plag1 −/− mice. These results indicate that cell proliferation is upregulated specifically in the cauda region, which correlates with the morphological defects being most severe in this region in Plag1 −/− mice 2 and may correlate with the dysregulation of ECM components shown by the RNA-Seq data, as the ECM regulates several cell behaviors including cell proliferation. 50,51 Although the GO term "positive regulation of epithelial cell proliferation" was also enriched in the upregulated genes in the epididymis from Plag1 −/− mice, the results showed that cell proliferation was normal in the epithelium. The upregulation of these genes may be a compensatory mechanism for the lack of tubule of elongation, which is a defect that persists into adulthood in Plag1 −/− mice. 2 GO analysis also revealed that within the list of downregulated genes in the epididymis from Plag1 −/− mice, there was significant enrichment of genes associated with the GO terms cell migration, cell motility, movement of cell or subcellular component and metalloendopeptidase/ metallopeptidase activity. The Mmp1b, Mep1a, Mme and Pappa genes encoding metallopeptidases or matrix metalloproteinases (MMPs) were downregulated in the Plag1 −/− epididymis, and are important in degrading proteins in the ECM, thereby playing a crucial role in tissue development, remodeling and repair. 52 MMPs are also involved in cell signaling, cell migration and the activation of growth factors, 53,54 and are partly regulated by tissue inhibitors of metalloproteinases (TIMPs). 52 The balance between MMPs and TIMPs is vital for ECM remodeling and maintenance, 52 and it has been shown that MMPs and TIMP2 are present in the epididymal fluid in the boar, ram and stallion, suggesting a role in epididymal sperm maturation. 55 Notably, Timp2, which encodes a TIMP protein, was upregulated in the Plag1 −/− epididymis. These results suggest that MMP activity may be decreased by both the downregulation of MMP genes and the upregulation of an MMP inhibitor. GO analysis also showed that among upregulated genes in the epididymides from Plag1 −/− mice, there was significant enrichment of genes mapped to the GO terms cell adhesion, glycosaminoglycan binding, ECM/region and proteinaceous ECM. Notably, a large number of collagen genes was upregulated in Plag1 −/− epididymides (Col6a3, Col12a1, Col4a6, Col15a1, Col6a1, Col23a1, Col1a2, Col6a5, Col4a3, Col4a5, Col11a1, Col3a1, Col6a2, Col5a2, Col6a6, Pcolce, Col1a1, Col9a1, Col5a3, Col5a1, Col4a1, Col14a1, Cogalt2 and Col4a4), and upregulated proteinaceous ECM genes included Spon1, Spon2, Lama2, Lamb1 and Ecm2. Overdeposition of ECM components can result in ECM stiffness or fibrosis in other tissues, leading to diseases such as cancer, cardiovascular disease, 56 pulmonary fibrosis 57 and kidney sclerosis. 58 ECM stiffness, however, has not yet been investigated in the epididymis. While ECM-related genes are known to be expressed in the Wolffian duct during embryonic development 59 and ECM remodeling is recognized as a crucial aspect of epididymis morphogenesis, 60 the role of ECM genes in the function of the adult epididymis is not well studied and the implication of ECM defects on sperm maturation are not known. As the ECM acts as a harbor for growth factors, 61 it is reasonable to speculate that defective ECM structure would have an impact on the release of growth factors or other proteins that are important for sperm maturation or epididymal tissue homeostasis. Taken together, the downregulation of cell motility and MMP genes, the upregulation of ECM component genes and Timp2, and the increased cell proliferation in the cauda epididymis stroma of Plag1 −/− mice suggest a defect in the maintenance and structure of the ECM, or perhaps disruption of normal tissue homeostasis in the epididymis of Plag1 −/− mice.
PLAG1 binding site analysis on dysregulated genes in the epididymides from Plag1 −/− mice did not show significant enrichment of PLAG1 binding sites. This could indicate either that PLAG1 control of these particular genes may be upstream or indirect, that PLAG1 may act in conjunction with other transcription factors, or perhaps that epididymal defects occurred earlier in development that we have analyzed during the present study. Moreover, motif scanning did not reveal enrichment of any other transcription factor motifs, providing further indication that the regulation of these target genes may require other transcription factors to act in concert. Some of the effects of PLAG1 deficiency on gene expression may be due to androgen imbalance; Hsd17b3 and Sultle1 (genes involved in the testosterone biosynthesis pathway) were downregulated in the testis from Plag1 −/− mice and the seminal vesicles from Plag1 −/− mice were disproportionately reduced in size, indicating decreased testosterone levels. 3 Androgenic control of many epididymal genes has been established, 62,63 however, some genes involved in sperm maturation are not controlled (or not entirely controlled) by androgens, suggesting that at the molecular level sperm maturation regulation is more complex than a requirement for normal androgen action. For example, Acrv1 mRNA expression in the caput mouse epididymidis spikes 8 hours after gonadectomy, but normalises to control levels 1 day post gonadectomy, indicating that Acrv1 is not exclusively regulated by androgens. 11 In conclusion, we investigated the genetic pathways that are regulated by PLAG1 in the mouse epididymis and identified dysregulated genes in Plag1 −/− mice that may be involved in the facilitation of proper sperm maturation, regulating the acrosome reaction and regulating sperm capacitation. Further, we show that there is an overall dysregulation of genes involved in the maintenance and structure of the ECM in Plag1 −/− mice, and that cell proliferation is significantly increased in the cauda epididymis from Plag1 −/− mice. However, since there was no significant enrichment of PLAG1 binding sites in the differentially expressed genes, PLAG1 control of these genes may be indirect and further investigation is required to further elucidate the precise direct targets that PLAG1 binds to and activate in order to influence proper sperm maturation. While the data presented in this study is suggestive of possible premature sperm capacitation and imbalanced ECM deposition and degradation in the epididymis from Plag1 −/− mice, further experimental validation is required. Future studies analyzing the functional capabilities of mature spermatozoa from Plag1 −/− mice such as the acrosome reaction, or characterization of the ECM structure in epididymides from Plag1 −/− mice may be insightful. Moreover, as global defects in Plag1 −/− mice may also contribute to generalized defects in tissue growth and development, targeted, tissue specific deletion of Plag1 may ultimately be necessary to determine epididymis-specific function. Although PLAG1 mutations have not yet been investigated in the context of fertility in human males; as the underlying causes of clinical male infertility are largely unidentified, we suggest Plag1 as a worthy candidate for investigation.  1 Genotypes were determined by PCR with genomic DNA isolated from ear clips as previously described. 2 RNA sequencing and data analysis Whole epididymides were used for RNA sequencing (RNA-Seq) in order to characterise the overall transcriptome of the Plag1 −/− mouse epididymis. Epididymides were harvested from 7-week-old Plag1 −/− , heterozygous (HET; Plag1 +/− ) and wild-type (WT; Plag1 +/+ ) mice (n = 5 per genotype) euthanized by CO 2 asphyxiation, and teased apart in cold phosphate-buffered saline (PBS). The tissues were rinsed in two changes of sterile PBS to remove spermatozoa. The samples were then homogenized using a mortar and pestle in TRIzol Reagent (Invitrogen, Carlsbad, California), according to manufacturer instructions. Total RNA was treated with DNase using the TURBO DNA-free kit (Invitrogen) and subsequently purified using the Zymo RNA Clean & Concentrator-25 kit (Zymo Research Corporation, Irvine, California) following the manufacturer's instructions. RNA concentration was then determined using a Nanodrop (Thermo Fisher, Waltham, Massachusetts). RNA-Seq was carried out at the La Trobe Genomics Platform (La Trobe University, Melbourne, Victoria, Australia). Fifteen libraries were constructed using the Illumina TruSeq Stranded mRNA kit (Illumina, San Diego, California) as per manufacturer instructions and the samples were sequenced by 150-bp, paired-end sequencing using the Illumina NextSeq 500 (Illumina). The raw RNA-Seq data were quality-assessed using fastQC (http://www.bioinformatics. babraham.ac.uk/projects/fastqc/). Adapter and low-quality fragments were removed using trimmimatic v0.32 64 and the clean reads were then mapped to the Mus musculus genome (assembly GRCm38.p6) by HISAT2 v2.0.5. 65 Alignment quality control was performed using RSeQC v2.6.4 66 and expression of each gene was determined by the featureCounts function in subread v1.4.6p5. 67 Genes with low count values were excluded from analysis; genes with valid expression were defined as genes with log counts per million >0.3 in at least three different replicates. In total, 20 505 genes were identified as meeting this criterion. Differential gene expression was then analyzed using edgeR v3.18.1 68 with false-discovery rate < 0.05 set as the cut-off. DAVID 69,70 was used for functional annotation analysis to identify genes associated with sperm-related gene ontology (GO) terms. The GOrilla analysis tool 71 was used to categorize differentially expressed genes into pathways and processes according to functional GO annotations. This allows enriched GO terms and common processes and pathways affected in the data sets to be identified. The RNA-Seq raw data have been deposited to NCBI's Gene Expression Omnibus (GEO) and are accessible through GEO Series accession number GSE140576.

| Quantitative PCR validation of RNA-Seq results
For quantitative PCR (qPCR) validation of the RNA-Seq results, epididymides were collected from different animals to those used for RNA-Seq (n = 5 for each of WT and KO). RNA was isolated as described above and cDNA was reverse-transcribed using MultiScribe reverse transcriptase (Life Technologies, Carlsbad, California), 10× M-MuLV buffer (50 mM Tris-HCl, 75 mM KCl, 3 mM MgCl 2 , 10 mM dithiothreitol, pH 8.3; New England BioLabs, Ipswich, Massachusetts), 10 mM dNTP mix (Bioline, London, UK), oligo(dT) 18 primer (Bioline) and 40 U/μl RiboSafe RNase Inhibitor (Bioline). Reactions were incubated for 10 minutes at 25 C, 2 hours at 42 C and then 5 minutes at 85 C in a thermocycler. qPCR was used to validate differential expression of the genes Defb9, Mep1a, Spink14 and Timp2 (primer sequences listed in supporting Table 1). Each qPCR amplification reaction mix contained 5 μL of cDNA, 300 nM of forward and reverse primer and 5 μL Fast SYBR Green Master Mix (Life Technologies). Reactions underwent the following qPCR protocol using the Biorad CFX96 Real Time System (Biorad, Hercules, California): 20 seconds at 95 C, 40 cycles of 3 seconds at 95 C and 30 seconds at 60 C. All samples were measured in triplicate. Actb was used as the reference gene, as expression levels of this gene was not different in the epididymides of Plag1 −/− compared to Plag1 +/+ mice according to the RNA-Seq data. qPCR data were processed and fold changes presented using LinRegPCR v.2017.0 software 72 as previously described. 3 Unpaired t-tests were performed using GraphPad Prism 7 (GraphPad Software, La Jolla, California) to compare average normalised mRNA expression levels in the epididymides of Plag1 −/− compared to Plag1 +/+ mice, as determined by qPCR.

| Ki67 immunofluorescence
Epididymides were collected from adult (7-week-old) mice (n = 5 for each of WT and KO), fixed for 2 hours in 4% (w/v) paraformaldehyde then cryoprotected for 2 hours in 10% (w/v) sucrose followed by 30% sucrose until the tissues sank. Tissues were stored at −80 C until sectioned into 5 μm cryosections, and subsequently stored at −80 C. Thawed sections were first treated with boiling 1 M trisodium citrate dihydrate (VWR, Radnor, Pennsylvania) for 40 minutes for antigen retrieval. Sections were then washed in phosphate-buffered saline (PBS), incubated for 2 hours in PBS containing 4% (w/v) bovine serum albumin (BSA), 0.1% (v/v) Triton X-100, 0.1 M lysine and 5% (v/v) normal goat serum (Antibodies Australia, Melbourne, Australia) (BSA-PBS-TX) at room temperature, then washed in PBS, before overnight incubation at 4 C with anti-Ki67 antibody (ab15580; Abcam, Cambridge, United Kingdom) diluted 1:500 in BSA-PBS-TX. Control slides were incubated in BSA-PBS-TX with the primary antibody omitted. Next, sections were washed in PBS, incubated for 2 hours at 4 C with goat anti-rabbit IgG Alexa Fluor 555 (A27039; Thermo Fisher, Waltham, Massachusetts) secondary antibody diluted 1:500 in BSA-PBS-TX. Sections were then washed in PBS with 0.1% Triton X-100 before application of DAPI diluted in PBS according to manufacturer instructions (Thermo Fisher). The sections were washed in PBS and mounted using aqueous mounting medium.

| Microscopy and image analysis
Sections were imaged using an Olympus BX41 microscope with an Olympus DP25 camera (Olympus Scientific Solutions Americas Inc., Waltham, Massachusetts). The 3-6 images were taken (20× magnification) of each region (caput, corpus and cauda) for analysis using Image J software (https://imagej.nih.gov/ij/). Percentages of Ki67 + cells in the epithelium and stroma, and the percentage of stromal cells (identified by cell size, shape and location) out of total cells, were determined using the "analyze particles" function. Epithelial and stromal areas were isolated using the freehand drawing tool to create "regions of interest", and sperm cells were excluded from the DAPI total cell counts by blanking out the luminal area using the freehand drawing tool and the "fill" function. Animal and genotype details were omitted from image files prior to the analysis to exclude bias. Unpaired t-tests were performed using GraphPad Prism 7 (GraphPad Software) to compare the mean percentages of Ki67+ or stromal cells between Plag1 −/− and Plag1 +/+ mice. Means were considered significantly different if P < .05.

| PLAG1 binding motif scanning
The TxDb.Mmusculus.UCSC.mm10.knownGene R library was used to obtain all the annotated transcriptional start sites (TSS) in mouse (mm10 genome version). The JASPAR database PLAG1 motif (JASPAR MA0163.1) and the binding sites defined by Madissoon et al were used for the analysis. 42 Two different promoter sizes were analyzed: the genomic regions spanning 300 bp upstream to 50 bp downstream of the TSS, and between 2 kb upstream to 100 bp downstream of the TSS. First we scanned selected target genes for PLAG1 sites using the Homer suite of sequence analysis tools 73 and performed χ-square test analysis to study differences in the occurrence of the sites compared to all other promoters. Then we extended the analyses to all differentially regulated genes and subjected the promoter regions to de novo motif discovery and general motif scanning using the findMotifs.pl program in Homer. The output of these analyses is a list of significantly enriched motifs in the promoters of the differentially regulated genes over all promoters in the genome, and their similarity to known motifs.