Peptidyl arginine deiminase 2 (Padi2) is expressed in Sertoli cells in a specific manner and regulated by SOX9 during testicular development

Peptidyl arginine deiminases (PADIs) are enzymes that change the charge of proteins through citrullination. We recently found Padi2 was expressed exclusively in fetal Sertoli cells. In this study, we analyzed the transcriptional regulation of Padi2 and the role of PADI2 in testicular development. We showed SOX9 positively regulated Padi2 transcription and FOXL2 antagonized it in TM3 cells, a model of Sertoli cells. The responsive region to SOX9 and FOXL2 was identified within the Padi2 sequence by reporter assay. In fetal testes from Sox9 knockout (AMH-Cre:Sox9flox/flox) mice, Padi2 expression was greatly reduced, indicating SOX9 regulates Padi2 in vivo. In vitro analysis using siRNA suggested PADI2 modified transcriptional regulation by SOX9. However, Padi2−/− XY mice were fertile and showed no apparent reproductive anomalies. Although, PADI2 is known as an epigenetic transcriptional regulator through H3 citrullination, no significant difference in H3 citrullination between wildtype and Padi2−/− XY gonads was observed. These results suggest Padi2 is a novel gene involved in testis development that is specifically expressed in Sertoli cells through the regulation by SOX9 and FOXL2 and PADI2 supports regulation of target genes by SOX9. Analysis of the Padi2−/− XY phenotype suggested a redundant factor compensated for PADI2 function in testicular development.

Recently, we identified a molecule, peptidyl arginine deiminase, protein-2 (PADI2), that is expressed specifically in Sertoli cells during testicular development. PADI2 has been reported to convert the amino acid arginine in a protein into the amino acid citrulline 10 . Citrullination can occur in any protein, including histone H3 proteins, suggesting that PADI2 mediates gene regulation through epigenetic mechanisms 11 . In order to identify the role of PADI2 in testicular development, we conducted in vitro and in vivo analyses, including generating Padi2 knock-out mice.

Results
Padi2 was exclusively expressed in Sertoli Cells in fetal developing testes. Based on previously reported microarray data, we found that Padi2 is expressed in a testes-specific manner over the prenatal period (Fig. 1a) and transcriptional profiling of somatic and germ cells suggested that testicular supporting cells expressed Padi2 (Fig. 1b,c) (data accessible at NCBI GEO database 12 , accession GSE27715 13 , GSE4818, and GSE5334; Gaido K, Lehmann K et al., 2006). In order to confirm these data, we employed whole mount in situ analysis and RT-PCR analysis of Kit Wv/Wv mouse gonads. Padi2 was expressed in the testicular cord (Fig. 1d), and its expression was limited to somatic cells because Padi2 expression was not reduced in Kit Wv/Wv mice, in which germ cells are absent (Fig. 1e). These results suggested that Padi2 is exclusively expressed in testicular supporting cells, i.e., Sertoli cells.
Antagonistic regulation of Padi2 by SOX9 and FOXL2 in vitro, but not in vivo. The temporal expression pattern of Padi2 suggested that the gene would be a target of the determinant of Sertoli cell differentiation, SOX9, and we carried out in vitro analyses using TM3 cells, a model of Sertoli cells 14 . Consistently, Sox9 introduction into TM3 cells transcriptionally upregulated endogenous Padi2 expression in a dose-dependent manner (Fig. 2a).
In order to investigate Padi2 regulation by SOX9 in vivo, we employed the XY gonads of AMH-Cre:Sox9 flox/ flox mice. The XY gonads of AMH-Cre:Sox9 flox/flox mice are phenotypically testes, and Sox9 expression is markedly reduced at 13.5 dpc, after the time of sex determination, and absent by 14.5 dpc, enabling a more specific analysis gonads. Padi2 expression was limited to testicular cord of male gonads. (e) Padi2 expression of gonad in WT and Kit Wv/Wv mice at 13.5dpc relative to ActB was quantified by qRT-PCR analysis. Padi2 expression in XY gonads was maintained in Kit Wv/Wv XY mice. Mean ± SD of three biologically independent experiments performed in triplicate was shown. Asterisks indicate level of statistical significance (*P < 0.05; NS: not significant). Statistical significance was determined using one-way ANOVA followed by a Tukey-Kramer post hoc test.  15 . In consistent with data from TM3 cells, in AMH-Cre:Sox9 flox/flox XY gonads at 13.5 dpc, Padi2 expression was dramatically suppressed, and its expression was almost abolished (Fig. 2b).
On the other hand, the positive regulation of Padi2 by SOX9 was attenuated by the ovarian molecule FOXL2 in TM3 cells (Fig. 2c). However, in Foxl2-null mice, Padi2 expression in XY gonads was not changed (Fig. 2d. In developing testes, Padi2 expression mostly depends on SOX9 (Fig. 2b) as observed in AMH-Cre:Sox9 flox/flox XY gonads, and suppression by FOXL2 is not required in vivo.

SOX9 and FOXL2 transcriptionally regulated Padi2 expression through a sequence on intron 1.
In order to elucidate the precise mechanisms of the regulation by SOX9 and FOXL2, we examined the sequence upstream of the 5ʹ-UTR and all introns of Padi2. Upstream of the gene, one possible biding sequence for FOXL2 (c.−94) was identified, while one possible binding sequence for FOXL2 (c.190 + 4486) and two for SOX9 (c.190 + 5806, c.190 + 8362) were identified (Fig. 3a) in the first intron. We employed three reporter constructs, #1, #2, and #3, that harbor possible binding sequences, as shown in Fig. 3a, and conducted a series of in vitro co-transfection reporter assays in HEK293T cells, which do not express endogenous Sox9 and Foxl2.
The reporters of construct #2, but not of constructs #1 and #3, were activated by Sox9 introduction in a dose-dependent manner (Fig. 3b), suggesting that c.190 + 8362 was responsible for the transactivation by SOX9. Consistently, by introducing mutations into the sequence, c.190 + 8362, the response of construct #2 was completely abolished (Fig. 3b).
Repression by FOXL2 was observed in constructs #2 and #3 but not in #1 (Fig. 3c). As shown in Fig. 2, FOXL2 cancelled the positive regulation of SOX9 for construct #2 (Fig. 3d). However, the suppressive effects were also observed in the reporter with a mutated possible biding site of FOXL2, construct #2 (mutant F) (Fig. 3d). On the other hand, construct #2 was suppressed by FOXL2 in the absence of SOX9 (Fig. 3c), suggesting that FOXL2 suppressed the transcription of Padi2 by binding to an unknown element of intron 1 rather than by interfering with the access of SOX9 to its binding element.
Transcriptional modification of the SOX9 target genes by PADI2 in vitro. Next, we examined the possible role of Padi2 in testicular development in vitro. PADI2 is a peptidyl arginine deiminase that converts the amino acid arginine into citrulline, and it has been suggested to modify transcriptional activity by an epigenetic mechanism, the citrullination of the histone protein H3 11 . Considering that Padi2 is downstream of SOX9, one possible explanation for the role of PADI2 is that it is a modulator of SOX9 action itself.
In order to verify this hypothesis, we knocked down Padi2 expression by siRNA and co-transfected SOX9 and then examined the expression levels of the target genes of SOX9. For the Ptgds gene, transcriptional activation by SOX9 was attenuated by knocking down Padi2 (Fig. 4a), while the inhibition of Padi2 expression additionally up-regulated the expression of the Cyp26b1 gene in TM3 cells (Fig. 4b), suggesting that PADI2 modifies SOX9 transcriptional regulation.
Padi2 −/− XY mice did not exhibit an obvious testicular phenotype, including fertility. To elucidate the role of PADI2 during testicular development in vivo, we generated Padi2 −/− mice using the CRISPR/Cas9 system. By targeting exon 1 (Fig. 5a), Padi2 −/− mice were successfully established (Fig. 5b). Padi2 −/− XY mice had normal male external genitalia (data not shown). Histological analyses revealed that Padi2 −/− XY gonads exhibited normal testicular cord formation. SOX9 was expressed in wildtype testis, and the expression of the ovarian marker FOXL2 was not detected (Fig. 5c). Padi2 −/− mice were fertile, and the number of offspring was identical Padi2 expression was not changed in Foxl2 null mice. The mean and SD of three biological replicates, except Foxl2 null mice, measured in triplicate was calculated. The data of Foxl2-null mice was from two biological replicate samples and mean value was calculated. Asterisks indicate level of statistical significance (*P < 0.05; **P < 0.01). Unpaired Student's t-test was used to demonstrate statistically significant difference between the given sample and the control. with that of the wildtype (Fig. 5d). No significant difference was observed in the expression of SOX9 and its target genes between wildtype and knock-out mice (Fig. 5e).
H3 citrullination in developing testes was not affected in Padi2 −/− mice. It has been reported that PADI2 is involved in an epigenetic mechanism, the citrullination of H3, and we examined the citrullination of (b-d) Reporter assay of the five constructs, construct #1~#3 and their mutated constructs, was performed by introducing Sox9 and/or Foxl2 into HEK293T cells. 50-75 ng of pSGSox9 and/or 25-75 ng of pcDN3Foxl2 was transfected to the 1.5 × 10 5 cells. The mean and SD of three biological replicates measured in triplicate was calculated. Asterisks indicate level of statistical significance (*P < 0.05; **P < 0.01, ns, not significant). Unpaired Student's t-test was used to demonstrate statistically significant difference between the given sample and the control for (a-c). For (d) one-way ANOVA followed by a Tukey-Kramer post hoc test was used for analysis. All data sets represent qRT-PCR analysis of Ptgds or Cyp26b1 mRNA expression relative to Gapdh (means ± SD of 3 biologically independent experiments performed in triplicate). Asterisks indicate level of statistical significance (*P < 0.05; **P < 0.01). Statistical significance was determined using one-way ANOVA followed by a Tukey-Kramer post hoc test. (b) In Padi2 −/− mice, the expression of PADI2 was abolished in adult mice brain. (c) Padi2 KO XY mice did not exhibit obvious testicular phenotype in immunofluorescence analysis. (d) The number of offspring was identical with that of the wild type from young adult (10-15 weeks of age) to matured adult mice (21-25 weeks of age). Three male mice for each group were mated with C57BL/6J wild type female mice and examined. The graphs indicate means and SD of each litter size. (e) mRNA levels of SOX9 target genes in Padi2 −/− mice at 13.5 dpc. All data sets represent qRT-PCR analysis of Sox9, Amh or Cyp26b1 mRNA expression relative to ActB (means ± SD of 3 biologically independent experiments performed in triplicate). Asterisks indicate level of statistical significance (*P < 0.05; **P < 0.01, ns, not significant). Unpaired Student's t-test was used to analyze significant difference for (d). One-way ANOVA followed by a Tukey-Kramer post hoc test was used for analysis of (e). Fulllength blots are presented in Supplementary Fig. S1a.
H3 in developing gonads. In wildtype XX and XY gonads, the level of citrullination, H3Cit2, 8, 17, was identical ( Fig. 6a,b), and in the XY gonads of Padi2 −/− mice, the level of citrullination, H3Cit2, 8, 17, was not attenuated compared to that of the wildtype (Fig. 6c). These findings suggested that in fetal developing testes, citrullination of H3 was not dependent on PADI2.

Discussion
Our present study revealed that Padi2 is specifically expressed in Sertoli cells and transcriptionally activated by the testicular determining factor SOX9. Previously, PADI2 function has been reported in other biological contexts, such as H3 citrullination during breast cancer oncogenesis [16][17][18] the estrus cycle in the uterus or in gonadotrope cells 11,19,20 peptidyl citrullination during colon cancer oncogenesis 21,22 and inflammation inducing multiple sclerosis and skin neoplasia [23][24][25] . Regarding embryonic development, other isoforms of PADI have been suggested to be involved: PADI4 was found to regulate pluripotency via H1 citrullination 26 and PADI6 was found to function during zygotic genome activation and be related to female infertility 27 . Our study is the first report to suggest a potential role of PADI2 in organ development.
One of the possible roles of PADI2 in testicular development is epigenetic function through the citrullination of histone H3. The citrulline amino acid residues are only produced through post-translational modifications catalyzed by PADI, which is encoded by the Padi gene 28 . Our study did not identify a significant difference in citrullinated H3 between wildtype mice and Padi2 −/− mice, suggesting that the citrullination of histone H3 is independent of PADI2 during gonadal development. To date, only two subtypes of PADI, PADI2 and PADI4, are considered to be involved in H3 citrullination. Although we did not examine the citrullination of H3 in the testes of Padi4 knock-out mice, Padi4 is not likely to contribute to the citrullination of H3 because of its low expression. Our data suggested the possibility that unknown mechanisms contributed to the citrullination of H3 during gonadal development.
Although the data from Padi2 KO mice suggested that the citrullination of histone H3 is independent of PADI2 during gonadal development, Padi2 affected the transcriptional regulation of SOX9 in TM3 cells. One possible explanation for this finding could be the citrullination of other proteins that affect SOX9 function. A homologue of PADI2, PADI4, has been reported to regulate co-factor interaction with transcriptional factors by citrullination, affecting transcriptional regulation 29,30 .
A specific high-affinity FOXL2 binding element (FLRE) has been reported, 5′-GT(C/G)AAGG-3′ 31 , and we identified two regions within the promoter and intron 1 in the Padi2 gene. However, our reporter assay revealed that the region was not responsible for the suppressive effect of FOXL2. The series of experiments designed to determine the role of FOXL2 in gonadotropin hormone regulation revealed that FOXL2 binds to sequences other than FLRE. In this regulation, the FOXL2 binding sequences do not seem to be strictly conserved and could depend on the biological context, including interaction with SMADs and other proteins 32,33 . Given these data, we speculate that FOXL2 binds to intron 1 of Padi2 through an unidentified sequence other than FLRE. It was intriguing that Padi2 expression was markedly repressed in AMH-Cre:SOX9 flox/flox mice. This result suggests that during testicular development, the transcriptional regulation of Padi2 would escape the redundant function of other SOXs, such as SOX8 and SOX10. Although testicular determination is solely SOX9 dependent, SOX9 positively regulates testicular genes, such as Amh and Dhh, together with SOX8 and SOX10 (two SOX-E subfamily members) after sex determination and maintains testicular development 15,[34][35][36] . We presume that the redundant regulation of testicular-specific genes by SOX E subfamily members could be balanced by the biological context. The essential genes for testicular development could be redundantly regulated by SOX E subfamily members because the testicular phenotype of AMH-Cre:SOX9 flox/flox mice is almost normal 15 . Indeed, the Padi2 −/− mice in our present study did not exhibit an obvious testicular phenotype.
In summary, we identified a novel testicular gene, Padi2, that is exclusively expressed in Sertoli cells and regulated by SOX9. Our in vitro data suggested that a possible function of PADI2 is to modify transcriptional regulation by SOX9. Further, during gonadal development, H3 is citrullinated in male and female gonads independent of Padi2. Although the role of the molecule in testicular development is likely limited because Padi2 −/− mice did not exhibit an obvious phenotype, our findings provide valuable insights that can be used to elucidate the molecular mechanisms of testicular development in mice.

Materials and Methods
Cell Culture. The murine testicular somatic cell line TM3 was obtained from the American Type Culture Collection (Manassas, VA, USA). HEK293T and TM3 cells were cultured in DMEM (Life Technologies, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA, USA) at 37 °C in 5% CO 2 .
Mouse strains. C57BL/6J, hybrid F1 of C57BL/6J × DBA/2 (BDF1), and ICR mice were purchased from Sankyo Lab (Tokyo, Japan) and CLEA Japan (Tokyo, Japan). C57BL/6J-Kit W-v /J mice were provided by the RIKEN BRC through the National Bio-Resource Project of the MEXT, Japan. Kit Wv/+ mice were intercrossed to generate Kit Wv/Wv embryos, which were depleted of germ cells. The generation of Padi2 −/− mice is explained in the next section.
Mouse embryos were collected from timed mating at noon of the day on which the mating plug was observed, designated as 0.5 d post-coitum (dpc). The genotype of the mouse embryos was determined by PCR assay. The generation of AMH-Cre:Sox9 flox/flox and Foxl2-null mice was reported previously 15,37 , and these mice were maintained in a mixed C57BL/6 background and C57B6/J/129/SVJ genetic background, respectively.
All animal protocols were approved by the Animal Care and Use Committee of the National Research Institute for Child Health and Development, Tokyo, Japan and the Center for Experimental Animals of Tokyo Medical and Dental University, Tokyo, Japan. All experiments were conducted in accordance with these approved animal protocols.
Generation of Padi2 −/− mice by the CRISPR/Cas9 system. To disrupt exon 1 of Padi2, we designed a guide RNA (gRNA) using the online tool [https://crispr.dbcls.jp/] 38 (Fig. 5a), and the selected sequence was 5′-CGGAATGCAGCCGCCTATACGG-3′. Construction of gRNA expression vectors, generation of template DNAs for in vitro transcription, and in vitro RNA synthesis were performed using a previously reported protocol 39 . The microinjection of mouse zygotes was performed as described previously 40,41 . Mouse zygotes were obtained by mating superovulated BDF1 females and BDF1 males. Genomic DNA samples from F0-generation offspring were extracted and tested for Padi2 mutations by direct sequencing. We selected the line that carried a frameshift mutation (Fig. 5a), and it was backcrossed with C57BL/6J mice. Plasmid Construction. The Padi2 5′-UTR and intron 1 fragments were generated by PCR using genomic C57BL/6J DNA as a template and cloned into pCR2.1 using a TOPO-TA cloning kit (Invitrogen, Carlsbad, CA, USA). The fragments were cloned into the multiple cloning site of the pGL4.12 luciferase reporter vector (Promega, Madison, WI, USA) and named construct #1, construct #2, and construct #3 as shown in Fig. 3a. Construct #2, which had a mutation in the putative SOX9 response element (ACAAT > CGCGG) and FOXL2 response element (GTCAAGG > TGACCTT), was generated using Q5 DNA polymerase (New England Biolabs, Inc, Ipswich, MA, USA). Two forms of this construct were named construct #2 (mutant S2) and construct #2 (mutant F).

Transfection and RNA interference(RNAi). TM3 and HEK293T cells were transfected with expression
vectors using Lipofectamine 2000 and 3000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. The total amount of DNA was standardized using the control vector, pcDNA3 (Invitrogen, Carlsbad, CA, USA).
TM3 cells were plated at 3.0 × 10 5 /well in 6-well plates 24 h before transfection. Two to four micrograms each of pSGSox9 42 and pcDNA3Foxl2 43 was added to each well and collected at 48 h post-transfection. For the RNA interference experiment, 4 µg of pSGSox9 and 50 µM Padi2 siRNA oligo (#: 1320003; Life technologies, Carlsbad, CA, USA) were added to transfect TM3 cells plated at 1.25 × 10 5 cells/well in a 6-well plate. The total amount of transfected siRNA oligo was standardized using the siRNA of LacZ.

RNA isolation and Quantitative real time PCR (qRT-PCR). Total RNA from TM3 cells was isolated
using Isogen (Nippongene, Tokyo, Japan). Two to three µg of RNA from the cells was used as a template for the synthesis of cDNA using SuperScript III (Invitrogen, Carlsbad, CA, USA) and random primers (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. cDNA samples were diluted 1:4, and 1 μl was used in each 20 μl of qRT-PCR reaction, which contained SYBR Green PCR Master Mix (Roche, Basel, Switzerland). Transcript levels were analyzed by a previously described protocol 44 . Glyceraldehyde-3-phosphate dehydrogenase (Gapdh) was used as the normalizing gene to standardize qRT-PCR data.
Embryonic gonads from Kit Wv/Wv mice and Padi2 −/− mice without mesonephros were dissected in PBS at the appropriate stages and immediately frozen in liquid nitrogen. Total RNA (60-300 ng) from embryonic gonads was isolated using Isogen (Nippongene) and was used for reverse transcription with SuperScript II (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. For quantitative PCR reactions, Power SYBR Master Mix (Applied Biosystems, Foster City, CA, USA) and SYBR Green PCR Master Mix (Roche, Basel, Switzerland) were used for Kit Wv/Wv mice and Padi2 −/− mice, respectively. For Kit Wv/Wv mice, transcript levels were analyzed on a 7900HT Fast Real-Time PCR System (Applied Biosystems) over 40 cycles of 95 °C for 10 s and 60 °C for 1 min, preceded by an initial 10-min step at 95 °C. For Padi2 −/− mice, an identical protocol as that used for TM3 cells was used for quantitative PCR reactions.
Gonadal samples from AMH-Cre:Sox9 flox/flox and Foxl2-null mice were analyzed by qRT-PCR as previously described 44 . All primers used were listed on Table 1.
Dual Luciferase assay. Transfected cells as described above were harvested and lysed 24 h after transfection, and both Firefly and Renilla luciferase activity was measured using a dual luciferase reporter assay system (#E1910, Promega). The firefly luminescence signal was normalized to that of Renilla, and the mean value of triplicated assay were calculated.
In situ hybridization. To generate the probe for Padi2, part of the Padi2 sequence was cloned into pCR2.1 (Invitrogen). PCR was performed with BIOTAQ (Bioline, London, UK), three sets of primers (Table 1), and the vector. Antisense digoxigenin-labeled RNA probes were generated by in vitro transcription using the PCR product, DIG RNA Labeling Mix (Roche), and ProbeQuant G-50 Micro Columns (GE Healthcare, Chicago, IL, USA) according to the manufacturer's instructions. A probe for the sense strand of Padi2 was used as a negative control.
Gonads and mesonephros from 13.5 dpc ICR mouse embryos were fixed overnight in 4% (w/v) paraformaldehyde with PBS at 4 °C and dehydrated with 25-75% methanol in PBS 0.1% (v/v) with Triton-X and embedded in 100% methanol. Subsequently, whole-mount in situ hybridization using the maleic acid buffer (MABT) method was performed as described previously 45 .
Statistical analysis. Student's t-test was used to compare the control and the other groups. For comparison of multiple groups, ANOVA with Tukey's method was employed.

Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. The sources of the datasets obtained from public repository are specified in each section above.