Androgen promotes squamous differentiation of atypical cells in cervical intraepithelial neoplasia via an ELF3-dependent pathway.

Takeo Matsumoto: Data curation (equal); formal analysis (equal); investigation (equal); resources (equal); writing – original draft (equal). Takuma Suzuki: Data curation (equal); formal analysis (equal); investigation (equal). Mitsuhiro Nakamura: Conceptualization (equal); resources (equal); supervision (equal). Megumi Yamamoto: Data curation (supporting); investigation (supporting). TAKASHI IIZUKA: Data curation (supporting); formal analysis (supporting); investigation (supporting). Masanori Ono: Resources (supporting). Kyosuke Kagami: Data curation (supporting); investigation (supporting). Haruki Kasama: Data curation (supporting); investigation (supporting). Tatsuhito Kanda: Data curation (supporting); investigation (supporting). Yuya Sakai: Data curation (supporting); investigation (supporting). Jyunpei Iwadare: Resources (supporting). Ayumi Matsuoka: Data curation (supporting); investigation (supporting). Kayo Kayahashi: Data curation (supporting); investigation (supporting). Kousho Wakae: Supervision (supporting). Masamichi Muramatsu: Supervision (supporting). Satoru Kyo: Supervision (supporting). Yasuhiko Yamamoto: Supervision (supporting). Yasunari Mizumoto: Conceptualization (equal); resources (equal); supervision (equal). Takiko Daikoku: Conceptualization (equal); funding acquisition (equal); supervision (equal).

All protocols were approved by the Medical Ethics Committee of Kanazawa University (approval number: 2015‐085). Human studies were conducted according to the Declaration of Helsinki principles, and written informed consent was received from participants prior to inclusion in the study.

This work was supported in part by a Grants‐in‐Aid for Scientific Research (nos. 17H04337, 19H01617, 19K22681, 20H03822, and 21H04837) and the Japan Agency for Medical Research and Development (no. 20ck0106549h0001).

Since AR expression was hardly detected in W12 cells, these cells were transfected with AR using a CAG promoter (Figure  1A ). RT‐PCR confirmed the mRNA expression of AR in AR‐transfected W12 cells (Figure  1B ). Immunocytochemical staining showed the intranuclear localization of AR and FOXP4 in AR‐transfected W12 cells (Figure  1C ). Protein expression of transfected AR was confirmed by Western blot analysis (Figure  1D ). LNCaP, an AR‐positive prostate cancer cell line, was used for positive control. DHT affects FOXP4 expression in AR‐transfected W12 cells. (A) Since AR expression was hardly detected in W12 cells, these cells were transfected with AR using a CAG promoter. (B) RT‐PCR confirmed the mRNA expression of AR in AR‐transfected W12 cells. (C) Immunocytochemical staining showed the intranuclear localization of AR and FOXP4 in AR‐transfected W12 cells. Scale bars show 50 μm. (D) Western blot analysis confirmed protein expression of transfected AR. LNCaP, an AR‐positive prostate cancer cell line, was used for positive control. (E) Western blot analysis showed that FOXP4 expression was not significantly changed by the transfection of the AR gene. DHT (10 nM) did not significantly change FOXP4 expression in CAG‐transfected W12 cells. (F, G) DHT treatment significantly reduced protein expression of FOXP4 in AR‐transfected W12 cells. (H) Extranuclear distribution of FOXP4 in W12 cells evaluated by ImageJ. Scale bars show 50 μm. (I) DHT (10 nM) treatment significantly promoted the FOXP4‐positive cytoplasmic area in AR‐transfected W12 cells. Five separate experiments were performed and bars represent SD. The data were analyzed by paired t ‐test and ANOVA followed by the Dunnett post hoc test. * p  < 0.05; ** p  < 0.01; N.S., not significant. Western blot analysis showed that FOXP4 expression was not significantly changed by transfection of the AR gene (Figure  1E ). The treatment of DHT (10 nM) did not significantly change the protein expression of FOXP4 in CAG‐transfected W12 cells (Figure  1E ), whereas it reduced FOXP4 expression in AR‐transfected W12 cells (Figure  1F,G ). This treatment promoted the extranuclear distribution of FOXP4, suggesting the inhibitory effects of androgen on FOXP4 functions (Figure  1H,I ). DHT treatment induced mature squamous phenotypes, showing a flattened appearance (arrow) and stratified growth (arrowhead) in places (Figure  2A ). This treatment significantly decreased the proliferation of W12 cells during 72‐h incubation (Figure  2B ). RT‐qPCR analysis demonstrated significant increases in mRNA expressions of squamous differentiation markers, KRT1 , KRT10 , and IVL , in AR‐transfected W12 cells (Figure  2C ). DHT inhibits proliferation and promotes squamous differentiation in AR‐transfected W12 cells. (A) DHT‐treatment induced mature squamous phenotypes in W12 cells, showing a flattened appearance (arrow) and stratified growth (arrowhead). Scale bars show 100 μm. (B) DHT decreased the proliferation of W12 cells during 72‐h incubation. The results are shown as means of individual experiments ( N  = 3) and bars represent SEM. The data were analyzed by paired t ‐test. (C) DHT increased mRNA expressions of KRT1 , KRT10 , and IVL . (D) DHT increased mRNA expressions of squamous differentiation‐related genes, TGM1 , SPRR1 , ELF3 , GRHL3 , HES1 , HES2 , HES5 , NOTCH2 , and NOTCH3 , but not NOTCH1 nor NOTCH4 in AR‐transfected W12 cells. Three separate experiments were performed and bars represent SD. The data were analyzed by ANOVA followed by the Dunnett post hoc test. * p  < 0.05; ** p  < 0.01; N.S., not significant. In a previous study, we reported that the downregulation of FOXP4 induced squamous differentiation‐ and NOTCH signal‐related gene expressions. 11 In this study, DHT administration significantly promoted mRNA expression of TGM1 , SPRR1 , ELF3 , GRHL3 , HES1 , HES2 , HES5 , NOTCH2 , and NOTCH3 , but not NOTCH1 nor NOTCH4 (Figure  2D ). Knockdown of ELF3 by siRNAs inhibited morphological changes in squamous differentiation in DHT‐treated W12 cells (Figure  3A ). This treatment also inhibited the expression of differentiation‐related genes induced by DHT (Figure  3B ). ELF3 , GRHL3 , and NOTCH3 are involved in androgen‐promoted squamous differentiation. (A) Knockdown of ELF3 by siRNAs inhibited morphological changes in squamous differentiation induced by DHT. Scale bars show 100 μm. (B) Knockdown of ELF3 also attenuated DHT‐induced gene expressions of the transcriptional factor GRHL3 and squamous differentiation markers TGM1 , SPRR1 , and IVL as well as NOTCH3 in DHT‐treated W12 cells. (C) Knockdown of GRHL3 inhibited gene induction of TGM1 , SPRR1 , NOTCH3 , and IVL in DHT‐treated W12 cells. (D) Knockdown of NOTCH3 in DHT‐treated W12 cells inhibited morphological changes in squamous differentiation. Scale bars show 100 μm. (E) Knockdown of NOTCH3 attenuated DHT‐induced gene expression of HES2 and IVL . Three separate experiments were performed and the bars represent SD. The data were analyzed by ANOVA followed by the Dunnett post hoc test. * p  < 0.05; ** p  < 0.01; N.S., not significant. Knockdown of GRHL3 inhibited gene induction of TGM1 , SPRR1 , NOTCH3 , and IVL in DHT‐treated W12 cells (Figure  3C ). Additional application of siRNAs targeting NOTCH3 showed inhibition of morphological changes in squamous differentiation induced by DHT (Figure  3D ) and attenuated gene induction of HES2 and IVL (Figure  3E ). These results suggest that GRHL3 and NOTCH3 are key molecules for DHT‐induced squamous differentiation. To examine whether ARs are generally involved in squamous differentiation, we investigated the effects of DHT on FOXP4 expression in the HaCaT cell line, one of the most used cell lines in the area of squamous differentiation research. Under the recommended Ca 2+ concentration (1.8 mM CaCl 2 ), intranuclear localization of AR in AR‐transfected HaCaT cells was verified by immunofluorescence staining (Figure  4A ). Protein expression of transfected AR was confirmed by Western blot analysis (Figure  4B ). RT‐PCR confirmed the mRNA expression of AR in AR‐transfected W12 cells (Figure  4C ). DHT (10 nM) reduced mRNA and protein expressions of FOXP4 during 72‐h culture (Figure  4D–F ). DHT inhibits FOXP4 expression in AR‐transfected HaCaT cells. (A) In the recommended maintaining culture condition (1.8 mM CaCl 2 ), intranuclear localization of AR in AR‐transfected HaCaT cells was verified by immunofluorescence staining. Scale bars show 50 μm. (B) Western blot analysis showed protein expression of transfected AR. (C) RT‐PCR confirmed the mRNA expression of AR in AR‐transfected W12 cells. (D) DHT (10 nM) reduced the mRNA expression of FOXP4 during a 72‐h culture. (E, F) Western blot analysis showed that DHT (10 nM) significantly decreased the protein expression of FOXP4. Three separate experiments were performed and bars represent SD. The data were analyzed by paired t ‐test. * p  < 0.05; ** p  < 0.01; N.S., not significant. An immunohistochemical study showed that AR was expressed in the basal to parabasal layers of the normal cervical epithelium (Figure  5A ‐a). In endocervical epithelial cells, AR expression was observed (arrowheads), but partially negative (arrows) (Figure  5A‐b ). In CIN1 and 2 lesions, AR was detected in the atypical squamous cells (Figure  5A‐c,d ), whereas AR expression had almost disappeared in the CIN3 lesion except for the basal layer (Figure  5A‐e , arrowheads), whose expression was partially abolished (Figure  5A‐e , arrows). AR expression was not detected in any cases of SCC (Figure  5A‐f ). The expression profiles of AR on normal tissues and CIN and SCC lesions by immunohistochemistry. (A) AR was expressed in the basal to parabasal layers of the normal cervical epithelium (a). In endocervical epithelial cells, AR expression was observed (arrowheads), but partially negative (arrows) (b). In CIN1 (c) and CIN2 (d) lesions, AR was detected in the atypical squamous cells. (e) AR expression had almost disappeared in the CIN3 lesion except for the basal layer (arrowheads), whose expression was partially abolished (arrows). (f) AR expression was not detected in SCC. Scale bars show 50 μm. (B) AR‐positive areas significantly increased from normal to CIN2 and decreased from CIN2 to CIN3. The data were analyzed by Kruskal–Wallis followed by Steel‐Dwass and shown as the median and interquartile range. * p  < 0.05; ** p  < 0.01; N.S., not significant. AR‐positive image analysis confirmed that AR‐positive areas significantly increased from normal to CIN2 and decreased from CIN2 to CIN3 (Figure  5B ).

It is well known that estrogens stimulate the proliferation of stratified squamous epithelium of the ectocervix. 22 In contrast, this study demonstrated that DHT attenuated cell proliferation using AR‐transfected W12 cells, suggesting that androgen acts antagonistically to estrogen in the stratified squamous cells of the cervix. We also showed that DHT promoted morphological changes that mimic squamous differentiation, and significantly increased the gene expressions of squamous differentiation markers, KRT1 , KRT10 , and IVL , suggesting that DHT can inhibit the proliferation and promote the differentiation of atypical squamous cells in CIN lesions. To support the different effects of sex steroid hormones on CIN, it was reported that prolonged exposure to estrogen induces carcinogenesis within the cervical and vaginal squamous epithelium using transgenic mice expressing the oncogenes of HPV16. 23 Estrogen receptor (ER) antagonists and selective ER modulators were also demonstrated to be candidates to protect cancer progression in both the cervix and the vagina. 24 By immunohistochemical examination, this study showed that AR was expressed in the basal to parabasal layers of the normal cervical epithelium. In CIN1 and CIN2 lesions, AR was detected in the atypical squamous cells, whereas AR expression had almost disappeared in the CIN3 lesion and was not detected in SCC. These expression profiles are compatible with previous reports 15 and suggest that the androgens do not act to promote squamous differentiation in the late stages of CIN. We recently reported that FOXP4 was expressed in atypical squamous cells in CIN and that downregulation of FOXP4 attenuated the proliferation of W12 cells, proposing that FOXP4 is a target molecule for differentiation therapy of CIN. 11 This study revealed that DHT attenuated the protein expression of FOXP4 and reduced its intranuclear distribution in AR‐transfected W12 cells. It was also demonstrated that DHT reduced the expression of FOXP4 in AR‐transfected HaCaT cells, supporting that androgen can regulate FOXP4 function. These findings also suggest that ARs are generally involved in squamous differentiation. Transcription factor ELF3 was reported to induce expression of NOTCH3 25 and the HES family. 26 Somatic mutation or reduced expression of ELF3 was reported in uterine cervical adenocarcinomas, 27 urothelial bladder carcinoma, 28 and oral SCC. 29 We previously showed that the downregulation of FOXP4 induced squamous differentiation in W12 cells via an ELF3‐dependent pathway. Consistent with it, this study showed that DHT induced the expression of squamous differentiation‐related genes in AR‐transfected W12 cells via an ELF3‐dependent pathway, indicating that ELF3 is a common molecular target for FOXP4‐ and androgen‐dependent differentiation therapy of CIN. Our preliminary study using microarray analyses showed several molecules were commonly downregulated by both knockdown of FOXP4 and administration of DHT (unpublished data). Consequently, these in vitro models may be useful to detect new candidates for target molecules to develop differentiation therapies of CIN. This study supports the previous concept that the sex hormone environment is involved in the progression of CIN. 30 , 31 In women, a high level of endogenous estradiol was reported to be a risk for HPV16‐induced onset of uterine cervix squamous carcinoma, 30 whereas the other study showed no significant relationship between CIN and endogenous sex steroid hormones such as estradiol, estrone, estrone‐sulfate, dehydroepiandrosterone sulfate, and progesterone. 32 However, there is little information about the pathophysiological effects of androgen on CIN progression. In this regard, this study provides a new viewpoint that androgen is another important hormonal factor for CIN differentiation. Since differentiation therapy by all‐ trans retinoic acid is effective for patients with acute promyelocytic leukemia, 33 , 34 several clinical approaches of differentiation therapy for CIN were tried using retinoids and showed some effect to block CIN2 progression. 35 , 36 However, they had no effects on CIN3 progression. 35 , 37 Two decades ago, based on the immune‐modulatory and tumor‐inhibitory activity of dehydroepiandrosterone, an adrenal androgen, against skin papillomas and carcinoma, 38 , 39 a pilot study was conducted for 12 women with low‐grade CIN and CIN regression was observed in 10 of the 12 women after intravaginal administration of dehydroepiandrosterone for 6 months. 40 Although it is unclear whether this regression rate is significant or not as compared with the spontaneous regression rate, this study reported that long‐term intravaginal administration of dehydroepiandrosterone had no side effects. Taken together, considering that AR is constantly expressed on the basal layer of the stratified squamous cells and CIN1 and CIN2 lesions, local intravaginal administration of androgen is a promising approach to the early stage of CIN from the new perspective of differentiation therapy. A major limitation of our study is the lack of animal experiments. Since animal cells are not natural hosts for HPV, the use of an animal model of CIN under HPV infection is difficult. Androgen effects should be investigated using transgenic mice expressing the oncogenes of HPV16 in the future. 23 Second, we should take into account the possibility that the forced expression of AR can cause artifacts in cell functions. In conclusion, as we summarize the novel findings in Figure  6 , this study showed that DHT attenuated proliferation and promoted the expression of squamous differentiation‐related genes in AR‐transfected W12 cells. It also indicated that ELF3 is a common molecular target for FOXP4‐ and androgen‐dependent differentiation therapy of CIN. This study supports that the sex hormone environment is involved in the progression of CIN and throws a new viewpoint that androgen is another important hormonal factor for CIN regulation. Based on these results, we propose that androgen is a novel factor that regulates CIN differentiation, providing a new strategy for nonsurgical and hormone‐induced differentiation therapy against CIN1 and CIN2. Estimated roles of androgen/AR in FOXP4‐related proliferation and squamous differentiation in the uterine cervix. We previously proposed that FOXP4 promotes proliferation and inhibits squamous differentiation through ELF3 and NOTCH3 in human uterine cervical epithelial cells, indicating that FOXP4 is a candidate target molecule for differentiation therapy of CIN. This study showed that DHT/AR promotes squamous differentiation and inhibits proliferation in W12 cells partially through regulating the FOXP4 function (shown in blue), suggesting the significant role of hormonal environment in CIN progression. These findings suggest that DHT/AR are new candidate molecules for differentiation therapy of CIN.

It is widely known that human papillomavirus (HPV) is the main cause of the onset and development of CIN and cervical squamous cell carcinoma (SCC). 1 Currently, prophylactic HPV vaccines have been demonstrated to be effective in the prevention of cervical cancers in the world. 2 However, since these vaccines do not eliminate preexisting infections, 3 the discovery of a new strategy to regulate HPV‐associated oncogenesis is warranted. 3 , 4 The initial pathological sign for CIN is impairment of squamous differentiation, which develops atypical squamous cell layers. 5 It was reported that squamous differentiation was inhibited by HPV16 oncoproteins, E6 and E7, via MAML1, NOTCH, and PTPN14. 6 , 7 , 8 , 9 , 10 However, since the production of E6 and E7 becomes dominant after DNA integration of HPV genes at the later stages of CIN, the main factor that impairs squamous differentiation at the early stages is still unknown. Recently, we found that FOXP4 was expressed in atypical squamous cells in CIN. 11 The FOXP family (FOXP1–4) was reported to regulate tissue development or cell differentiation. 12 , 13 In CIN1‐derived HPV16‐positive W12 cells, the downregulation of FOXP4 attenuated cell proliferation, whereas this treatment induced squamous differentiation. This FOXP4 downregulation‐induced squamous differentiation was regulated through an ELF3‐dependent pathway, suggesting that FOXP4 is a target molecule for differentiation therapy of CIN. 11 However, the factors that regulate FOXP4 expression in W12 cells remain unclear. Previously, it was reported that the androgen/AR complex can regulate mRNA and protein expressions of Foxp1 and Foxp2 in the rat brain, proposing that androgen is an upstream regulator of Foxp. 14 On the other hand, AR was reported to be expressed on the CIN lesions and AR expression becomes lost during CIN progression toward invasive SCC. 15 Thus, androgens are presumed to be regulators of CIN differentiation at the early stages, leading to the strategy that intravaginal administration of androgens may be a nonsurgical differentiation therapy for CIN. To test the above strategy, this study examined the effects of androgen on cell proliferation and squamous differentiation using AR‐transfected W12 cells. We also observed the role of androgen in FOXP4 function using AR‐transfected W12 cells and HaCaT cells, a human adult male skin‐derived keratinocyte cell line, 16 which were reported to exhibit Ca 2+ ‐ and FOXP4‐dependent squamous differentiation. 11 , 17 , 18

The authors declare no competing interests.

We obtained the human cervical dysplasia cell line W12 (RRID:CVCL_T290, clone 20,863), which contains HPV16 episomes, from Drs. Paul Lambert, Tomomi Nakahara, and Iwao Kukimoto. 19 This cell line was authenticated by STR profiling and was cultured as previously reported. 11 We obtained HaCaT (RRID:CVCL_0038), which is a spontaneously transformed immortal keratinocyte cell line derived from human adult male skin, 16 from Cosmo Bio (Tokyo, Japan) and cultured it as previously reported. 11 The human AR‐positive prostate cancer cell line LNCaP (RRID:CVCL_ 0395), which was authenticated by STR profiling, was gifted by Prof. Atsushi Mizokami, Department of Urology, Graduate School of Medical Sciences, Kanazawa University and was used for Western blot or RT‐PCR analysis as positive controls. Mycoplasma infections were detected regularly and all experiments were performed under mycoplasma‐free conditions. We obtained the uterine cervical tissues, normal ( n  = 5), CIN1 ( n  = 13), CIN2 ( n  = 14), CIN3 ( n  = 17), and SCC ( n  = 7), from 56 patients who had undergone hysterectomy or conization at Kanazawa University Hospital. These specimens were fixed with 20% formalin, embedded in paraffin, diagnosed, and used for the immunohistochemical study as previously reported. 11 Clinical information of each sample is shown in Table  1 . Clinical information of the patients included in this study. Normal‐#1 Normal‐#2 Normal‐#3 Normal‐#4 Normal‐#5 45 49 49 47 39 NILM NILM NILM NILM NILM Leiomyoma Leiomyoma Adenomyosis Adenomyosis Adenomyosis Hysterectomy Hysterectomy Hysterectomy Hysterectomy Hysterectomy — — — — — CIN1‐#1 CIN1‐#2 CIN1‐#3 CIN1‐#4 CIN1‐#5 CIN1‐#6 CIN1‐#7 CIN1‐#8 CIN1‐#9 CIN1‐#10 CIN1‐#11 CIN1‐#12 CIN1‐#13 41 64 39 40 48 48 35 40 34 33 48 30 59 LSIL AGC HSIL ASC‐US AGC/CIN1 HSIL HSIL HSIL HSIL LSIL HSIL ASC‐H LSIL CIN2 AGC CIN2 CIN2 AGC/CIN1 CIN2‐3 CIN3 CIN3 CIN3 CIN2‐3 CIN2 CIN2 CIN3 Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization CIN1 CIN1 CIN1 CIN1 CIN1 CIN1 CIN1/CIN3 CIN1/CIN3 CIN1/CIN3 CIN1 CIN1 CIN1 CIN1 CIN2‐#1 CIN2‐#2 CIN2‐#3 CIN2‐#4 CIN2‐#5 CIN2‐#6 CIN2‐#7 CIN2‐#8 CIN2‐#9 CIN2‐#10 CIN2‐#11 CIN2‐#12 CIN2‐#13 CIN2‐#14 40 35 42 46 28 42 40 40 49 55 32 27 38 31 HSIL HSIL ASC‐US HSIL AGC LSIL HSIL HSIL HSIL ASC‐H ASC‐US ASC‐H LSIL ASC‐US CIN3 CIN3 CIN3 CIN3 AIS/CIN1 CIN2‐3 CIN3 CIN3 CIN2 CIN2 CIN2‐3 CIN2‐3 CIN3 CIN2 Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization CIN2/CIN3 CIN2/CIN3 CIN2/CIN3 CIN2 AIS/CIN2 CIN2/CIN3 CIN2 CIN2 CIN2 CIN2 CIN2 CIN2 CIN2 CIN2 CIN3‐#1 CIN3‐#2 CIN3‐#3 CIN3‐#4 CIN3‐#5 CIN3‐#6 CIN3‐#7 CIN3‐#8 CIN3‐#9 CIN3‐#10 CIN3‐#11 CIN3‐#12 CIN3‐#13 CIN3‐#14 CIN3‐#15 CIN3‐#16 CIN3‐#17 32 36 30 35 41 31 34 46 29 38 43 35 48 47 49 63 26 HSIL HSIL SCC HSIL HSIL HSIL HSIL HSIL HSIL HSIL HSIL SCC HSIL HSIL HSIL HSIL‐SCC HSIL CIN3 CIN3 CIN3 CIN3 CIN2 CIN3 CIN3 CIN2 CIN3 CIN2 CIN3 CIN3 CIN3 CIN3 CIN3‐MIC CIN3‐MIC CIN2‐3 Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization Conization CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 CIN3 Abbreviations; IC, invasive cancer; MIC, microinvasive cancer. Bethesda System 2001. FIGO 2018. Immunohistochemical staining was performed using the avidin–biotin–peroxidase complex method according to the manufacturer's instructions (VECTASTAIN ABC Kit, Vector Laboratories, ) as described previously. 11 The slides were incubated overnight at 4°C with either one of primary antibodies, anti‐AR rabbit monoclonal antibody (1:200, clone SP107, RRID: AB_2537931, ab105225, Abcam) or anti‐p63 mouse monoclonal antibody (1:100 clone 4A4, RRID: AB_305870, ab735, Abcam). Control staining was performed by replacing the primary antibody with a normal rabbit or mouse serum solution. The W12 cells or HaCaT cells transfected with pCAGIPuro or pCAGIPuro‐AR were cultured on the glass chamber slide (Lab‐Tek®II, Thermo Fisher Scientific Inc.). The slides with 80% confluent cells were fixed in 4% paraformaldehyde and were incubated overnight at 4°C with anti‐FOXP4 rabbit polyclonal antibody (1:200, HPA007176, RRID: AB_1078911, Sigma‐Aldrich) or anti‐AR mouse monoclonal antibody (1:100, clone AR441, RRID: AB_11000751, Dako, Tokyo, Japan). Control staining was performed by normal rabbit or mouse serum solution. Secondary labelling was with Alexa Fluor™ 488‐labeled goat anti‐mouse IgG antibody (A‐11001, Invitrogen.) or Alexa Fluor™ 555‐labeled goat anti‐rabbit IgG antibody (A32732, Invitrogen), respectively. The AR‐positive rates in CIN lesions were analyzed and values were normalized using NIH ImageJ software as described previously. 11 , 20 Western blot analysis was performed as described previously. 11 Protein lysates were extracted with RIPA buffer (Cell Signaling Technology Inc.), and were electrophoresed on a 7.5% SDS‐PAGE gel and then transferred to a nitrocellulose membrane. The transferred membranes were incubated with primary antibody against FOXP4 (1:1000, RRID: AB_2262825, 16,772‐1‐AP, Proteintech Group Inc. ), AR (1:5000, the rabbit polyclonal antibody NH27, kindly provided by Prof. Atsushi Mizokami of Kanazawa University), or β‐Actin (1:5000, RRID: AB_630835, C‐11, Santa Cruz Biotechnology) overnight at 4°C. A secondary horse‐radish peroxidase conjugated antibody was applied for 1 h at room temperature. The blots were visualized as described previously. 11 RT‐qPCR was performed as described previously. 11 The cDNA was amplified using specific primers, forward (Fd) and reverse (Rs) primers (Table  2 ), and hypoxanthine phosphoribosyltransferase 1 (HPRT1) was used for the control. The results were shown as mean ± SD of three independent experiments. Primer sequences for RT‐PCR and quantitative real‐time PCR. The empty vector (pCAGIPuro) was created by excluding the flag‐Myc fragment from pCAGIPuro‐FlagmcMycWT with Xho I cutting and self‐ligation (Clone ID; RDB_14100, RIKEN DNA BANK). 21 To create an AR expression vector (pCAGIPuro‐AR), a 1.4‐kbp human AR fragment prepared from pEGFP‐AR (kindly provided by Prof. Atsushi Mizokami of Kanazawa University) with Bgl II and Bam HI was subcloned into pCAGIPuro. W12 cells or HaCaT cells were seeded in 6‐well plates at 5 × 10 5 cells per well. After culture overnight, the cells were transfected with pCAGIPuro or pCAGIPuro‐AR using GenomeOne‐GX (Ishihara Sangyo Kaisha, Ltd.) and KALA amphipathic peptide (Cosmo Bio Co., LTD) according to the manufacturer's protocol. After 48 h, 1 μg/μL puromycin (Thermo Fisher Scientific) was added to the cells for 7 days to establish stable cell lines. RNA interference for reduction of FOXP4 gene expression was performed using lentivirus particles of MISSION®small hairpin RNA (FOXP4‐shRNA1‐5: TRCN0000285257, TRCN0000274833, TRCN0000274834, TRCN0000274832, and TRCN0000274894, Sigma‐Aldrich) as described previously. 11 Transfection of siRNA into the W12 cells was performed as described previously. 11 siRNAs for FOXP4 (s41928 and s41929, Thermo Fisher Scientific), ELF3 (s4623 and s4624, Thermo Fisher Scientific), GRHL3 (s33752 and s33754, Thermo Fisher Scientific), NOTCH3 (s9640 and s9641, Thermo Fisher Scientific), and corresponding Silencer® Select Negative Control No.1 (AM4635, Thermo Fisher Scientific) were used. Cell proliferation assay was performed in triplicate using Cell Proliferation Reagent WST‐1 (Roche, Mannheim, Germany) as described previously. 11 After AR‐transfected W12 cells were treated with DHT (10 nM), FOXP4 was immunocytologically stained. The FOXP4‐positive extranuclear areas were detected by subtracting Hoechst‐positive nuclear regions and were calculated by NIH ImageJ software. In order to assess differences in the AR‐positive area ratio among CIN stages, Kruskal–Wallis followed by Steel‐Dwass was performed. The data are shown as the median and interquartile range. The differences in FOXP4‐positive extranuclear areas between DHT‐stimulated and nonstimulated W12 cells, and the differences in the expression of FOXP4, squamous differentiation‐related genes between the treated and non‐treated cells, and cell proliferation assay were evaluated by the paired t‐ test or ANOVA followed by the Dunnett post hoc test (SPSS Statistics version 25.0, IBM, USA). p ‐values of <0.05 were considered significant.