Oncogene (2013) 32, 15 --26 & 2013 Macmillan Publishers Limited All rights reserved 0950-9232/13 www.nature.com/onc

ORIGINAL ARTICLE Epigenetic silencing of PTPRR activates MAPK signaling, promotes metastasis and serves as a biomarker of invasive cervical cancer

P-H Su1,2, Y-W Lin3, R-L Huang4, Y-P Liao5, H-Y Lee4, H-C Wang4, T-K Chao6, C-K Chen5, MWY Chan7, T-Y Chu8, M-H Yu4 and H-C Lai1,2,4,5

Epigenetic modifications are a driving force in carcinogenesis. However, their role in cancer metastasis remains poorly understood. The present study investigated the role of DNA methylation in the cervical cancer metastasis. Here, we report evidence of the overexpression of DNA methyltransferases 3B (DNMT3B) in invasive cervical cancer and of the inhibition of metastasis by DNMT3B interference. Using methyl-DNA immunoprecipitation coupled with microarray analysis, we found that the protein tyrosine phosphatase receptor type R (PTPRR) was silenced through DNMT3B-mediated methylation in the cervical cancer. PTPRR inhibited p44/42 MAPK signaling, the expression of the transcription factor AP1, human papillomavirus (HPV) oncogenes E6/E7 and DNMTs. The methylation status of PTPRR increased in cervical scrapings (n ¼ 358) in accordance with disease severity, especially in invasive cancer. Methylation of the PTPRR promoter has an important role in the metastasis and may be a biomarker of invasive cervical cancer.

Oncogene (2013) 32, 15--26; doi:10.1038/onc.2012.29; published online 13 February 2012 Keywords: protein tyrosine phosphatase receptor type R; DNA methyltransferase; mitogen-activated protein kinase; metastasis; methylation; cervical cancer

INTRODUCTION changes, epigenetic modifications can be a driving force in 7 Cervical cancer remains one of the leading causes of death in early cancer initiation, promotion and progression. However, the women worldwide.1 The etiological role of the human papilloma- current literature provides only limited understanding of the virus (HPV) has led to the development of vaccines.2 The HPV mechanisms underlying epigenetic alterations and their relation- 8 oncoproteins E6 and E7 are the major viral proteins responsible for ship with metastasis. Our understanding of epigenetic modifica- cervical cancer initiation and progression. The p53 and retino- tions in cancer metastasis lags behind our knowledge regarding 8,9 blastoma proteins are well-characterized targets of HPV E6 and the genetics of this phenomenon. E7.3 From the premalignant to the malignant state, cervical Viral infections may cause cancers and induce epigenetic 10 lesions after HPV infection represent a spectrum of histological changes in host cells. After HPV infection, cervical epithelia abnormalities, from cervical intraepithelial neoplasia 1 (CIN1), must accumulate other genetic and epigenetic changes to 11 CIN2, CIN3 and carcinoma in situ (CIS) to invasive cancer. Although become an invasive cancer. The stepwise progression of most women are infected with HPV at some time, most of these precancerous lesions to in situ carcinomas and invasive cancers infections are transient and only very few progress to invasive is well defined in cervical cancer and this type of cancer is an ideal diseases.4 The driving force behind the stepwise progress from model for understanding the epigenetic alterations involved in precancerous lesions toward invasive cervical cancer remains cancer invasion and metastasis. Previous studies, including our largely unknown. It may take several years for progression from an own, revealed the presence of epigenetic alterations, such as DNA 12 --18 in situ neoplasm to invasive carcinoma. The mechanism that keeps methylation, in the cervical cancer. Using a methylomics pre-malignant cells in check and the colonial evolution that selects approach, we identified a LIM-homeobox gene, LMX1A,asa cells with invasive or metastatic potential also remain unidentified. metastasis suppressor that undergoes DNA-methylation-mediated Both genetic and epigenetic modifications contribute to the silencing, subsequently allowing the in situ cancer to invade and molecular basis of cancer. Genomics analysis identified defined metastasize via an incomplete epithelial--mesenchymal transition 13 genetic signatures in metastatic cells.5 Specific changes in gene (EMT). We hypothesize that de novo DNA methylation of some expression in metastatic cells could result from epigenetic unknown genes may constitute an important mechanism that is changes, such as DNA hypermethylation, which subsequently responsible for the failure to keep dysplastic cells in situ. The inactivate some metastasis suppressor genes. Increasing evidence present study investigated the role of DNA methylation in cervical indicates the importance of epigenetic regulation in response to cancer metastasis and discovered a tumor suppressor gene, environmental cues in cancer metastasis.6 Together with genetic phosphatase receptor type R (PTPRR), inhibiting MAPK signaling.

1Graduate Institute of Medical Sciences, National Defense Medical Center, Taipei, Taiwan; 2Laboratory of Epigenetics and Cancer Stem Cells, National Defense Medical Center, Taipei, Taiwan; 3Department of Microbiology and Immunology, National Defense Medical Center, Taipei, Taiwan; 4Department of Obstetrics and Gynecology, Tri-Service General Hospital, Taipei, Taiwan; 5Graduate Institute of Life Sciences, National Defense Medical Center, Taipei, Taiwan; 6Department of Pathology, Tri-Service General Hospital, Taipei, Taiwan; 7Department of Life Science, National Chung Cheng University, Min-Hsiung, Chia-Yi, Taiwan and 8Department of Obstetrics and Gynecology, Tzu Chi University, Hualien, Taiwan. Correspondence: Dr H-C Lai, Department of Obstetrics and Gynecology, Tri-Service General Hospital, National Defense Medical Center, 5F, 325, Sec 2, Cheng-Gong Road, Neihu District, Taipei City 114, Taiwan. E-mail: [email protected] Received 18 October 2011; revised 4 January 2012; accepted 8 January 2012; published online 13 February 2012 Epigenetic silencing of PTPRR P-H Su et al 16 RESULTS potential of HeLa3rd and CaSki cells was inhibited significantly Dysregulation of DNA methyltransferases 3B (DNMT3B) in invasive in vivo (Figures 2c--h). Downregulation of DNMT3B affected tumor cervical cancer growth and metastasis in cervical cancer in vivo. The copy number and mRNA level of DNMT3B appear to increase in cervical cancer.19 To explore the association between DNA methylation and cervical cancer invasion, the expression of DNMT DNMT3B targets the PTPRR promoter and mediates DNA methylation genes, including DNMT1, 3A and 3B, was assessed. The mRNA transcripts of DNMT3B were significantly overexpressed in a To identify the genes modulated by DNMT3B that are responsible cervical cancer cell line exhibiting a highly invasive phenotype, for the inhibition of metastasis, we used a methylomics approach HeLa3rd and CaSki (Figure 1a). The overexpression of DNMT3B using the methyl-DNA immunoprecipitation coupled with micro- was confirmed in cervical cancer tissues compared with tissue array analysis technology (Figure 3a). Differential methylation of from normal cervixes (Figure 1b). 255 genes was identified after DNMT3B knockdown (Supple- mentary Table 1). Online gene ontology data were checked to narrow the candidate list down to genes annotated with Downregulation of DNMT3B inhibits invasion and metastasis development; this approach identified 36 genes (Supplementary To validate further the role of DNMT3B in cervical cancer invasion, Table 2). From some aspects, cancer is a developmental disorder. we knocked down DNMT3B in two cervical cancer cell lines, Pathways important in development are commonly dysregulated 20 --22 HeLa3rd and CaSki. Knockdown efficiency was assessed using in cancers including cervical cancers. Our previous experi- quantitative real-time PCR (qRT--PCR) and western blot (Supple- ences in the discovery of genes methylated in cervical cancer also 12 --14,17,18 mentary Figure S1). Knockdown had no significant effects on cell led to developmental genes. In addition, DNMT3B has an 23 proliferation in HeLa3rd cells. Its effect in CaSki cells was important role of de novo methylation in development. There- significant (Figure 1c). Cell migration and invasion were inhibited fore, we focus on development-related genes in the present study. significantly in both cell lines in vitro (Figures 1d--g). Tumor Other potential genes may be further validated in the future. The formation was not affected in HeLa3rd cells, but was disrupted reexpression of 26 genes after DNMT3Bi was validated using significantly in CaSki cells (Figures 2a and b). The metastatic qRT--PCR (Figure 3b). Four genes were potential candidates:

** 20.0 * 1.2 1.6 HeLa CaSki+Scrambled RNAi * 18.0 HeLa3rd+Scrambled RNAi HeLa3rd CaSki+DNMT3B RNAi 16.0 1.0 HeLa3rd+DNMT3B RNAi 1.4 CaSki 14.0 * 1.2 12.0 0.8 4.0 1.0 * 0.6 0.8 0.6 2.0 0.4 0.4 Absorbance ( 492 nm ) 0.2 Absorbance ( 492 nm ) 0.2 Relative mRNA expression 0.0 0.0 0.0 DNMT1 DNMT3A DNMT3B 0 1 2 3 4 041 2 3 Days Days

HeLa3rd CaSki 1.2 Scrambled RNAi DNMT3B RNAi Scrambled RNAi DNMT3B RNAi Scrambled RNAi DNMT3B RNAi

1.0

0.8 * *

Relative cell migration 0.6

0.0 HeLa3rd CaSki

HeLa3rd CaSki Scrambled RNAi 1.2 Scrambled RNAi DNMT3B RNAiScrambled RNAi DNMT3B RNAi DNMT3B RNAi

1.0

* 0.8 *

0.6 Relative cell numbers 0.0 HeLa3rd CaSki Figure 1. Dysregulation of DNMT3B affects cell growth, migration and invasion in cervical cancer cells. DNMT3B was overexpressed in invasive cervical cancer cell lines (a) and cervical cancer tissues (b). The median of each group is indicated by a horizontal bar. (c) Growth curve of HeLa3rd and CaSki cells transfected with pSuper-Scrambled RNAi or pSuper-DNMT3B RNAi. Downregulation of DNMT3B affected cell proliferation in CaSki cells but not in HeLa3rd cells. The migration (d, e) and invasive (f, g) capabilities, assessed by wound healing and Matrigel invasion assays, of HeLa3rd and CaSki cells transfected with DNMT3B RNAi were inhibited significantly. The dotted line in (d) indicates the migration start site. Data in a, c, e and g are presented as mean±s.d. (n ¼ 4). *Po0.05, **Po0.01.

Oncogene (2013) 15 --26 & 2013 Macmillan Publishers Limited Epigenetic silencing of PTPRR P-H Su et al 17

HeLa3rd HeLa3rd Scrambled Scrambled RNAi RNAi DNMT3B DNMT3B RNAi RNAi 0.0 0.2 0.4 0.6 0.8 Tumor weight (g) CaSki CaSki Scrambled Scrambled RNAi RNAi

DNMT3B * DNMT3B RNAi RNAi 0.0 0.2 0.4 0.6 0.8 Tumor weight (g)

HeLa3rd i ii Scrambled RNAi HeLa3rd Scrambled RNAi

DNMT3B * RNAi DNMT3B 0 5 10 15 20 25 30 RNAi Metastatic nodules

CaSki Scrambled RNAi CaSki Scrambled RNAi

DNMT3B RNAi * DNMT3B 0 10 20 30 40 50 60 RNAi Metastatic nodules

HeLa3rd HeLa3rd CaSki Scrambled Scrambled Scrambled RNAi RNAi RNAi

DNMT3B DNMT3B DNMT3B RNAi RNAi RNAi

Figure 2. DNMT3B contributes to cell growth and metastasis in cervical cancer cells in vivo. (a) Photographs of the tumors excised from the mice on day 30. (b) Histogram of tumor weight at the end of the in vivo experiment. (c, d) Photographs of the lungs from mice that had been transplanted with tumor cells through tail vein injection. Two mice injected with HeLa3rd þ scrambled RNAi showed paralysis in the hind legs and were euthanized along with two mice injected with HeLa3rd þ DNMT3B RNAi 2 months after inoculation. Arrows indicated the metastatic nodules in the lungs of mice with HeLa3rd þ scrambled RNAi. No gross nodules were observed in the lungs of mice with HeLa3rd þ DNMT3B RNAi (ci). The other two mice were euthanized 3 months after inoculation when they developed dyspnea (cii). (e, f) Histogram of metastatic nodules at the end of the in vivo experiment shown in (c) and (d). Hematoxylin and eosin staining sections of the brains (g) from paralyzed mice and lungs (h) from the mice shown in (c) and (d). Circles represent metastatic nodules. The scale bar represents 1 mm. Data in b, e and f are presented as mean±s.d. (n ¼ 4). *Po0.05. dual-specificity tyrosine-(Y)-phosphorylation-regulated kinase 1A, promoter in HeLa3rd and HeLa3rd-DNMT3Bi is shown in Figure 3c. inhibitor of DNA-binding 1 dominant-negative helix--loop--helix The reexpression of its mRNA (Figure 3d) and the demethylation protein, phosphoglycerate dehydrogenase and PTPRR. Reexpres- of the PTPRR promoter after DNMT3B knockdown (Figure 3e and sion was highest for PTPRR. The probe intensity of the PTPRR Supplementary Figure S2). The gene silencing effect of DNA

& 2013 Macmillan Publishers Limited Oncogene (2013) 15 --26 Epigenetic silencing of PTPRR P-H Su et al 18

HeLa3rd HeLa3rd+DNMT3Bi PTPRR (protein tyrosine phosphatase receptor type R) mDIP-chip ↓ Genomics KS-Test, p<0.0001 position 255 genes ↓ Primary KS-Test, p<0.0001 transcripts + G. O. data 36 genes ↓ Control/DMT3Bi Validation Probe intensity Q-RT-PCR (log2 Ratio) ↓ Validation Q-MS-PCR CpGsites ↓ Scrambled RNAi 1.4 Scrambled RNAi DNMT3B RNAi DNMT3B RNAi PTPRR 2.0 1.2 (protein tyrosine phosphatase receptor type R) * HeLa3rd +Scrambled RNAi 1.0 1.5 0.8 * 0.6 1.0 0.4 HeLa3rd+DNMT3B RNAi Relative PTPRR

mRNA expression 0.2

ATP6V0C MTPN 0.0 Relative PTPRR methylation 0.0 HeLa3rd HeLa3rd B3GNT5 PHGDH CENTD2 PTCH1 Scrambled RNAi 1.4 CHML PTPRR 10 DNMT3B RNAi CISH SMA3 1.2 8 COL9A2 SMAP1 1.0 6 DYRK1A SRI 0.8 * * EEF1G STAB1 4 0.6 FHL3 THUMPD2 0.4 Relative PTPRR Relavive level of

mRNA expression 2 ID1 TLE1 0.2 0 PTPRR promoter binding L1CAM TRAPPC6B 0.0 μM μM μM μM DNMT3B DNMT1 0 5 MEST UTP11L 10 20 MGC22265 ZBTB7A 5-azadC conc.

-100% ~ 100%

Figure 3. Methylomic analysis of DNMT3B-mediated silencing of PTPRR in cervical cancer. (a) Flowchart of methyl-DNA immunoprecipitation coupled with microarray analysis (mDIP--chip) and validation procedures. (b) qRT--PCR analysis of 26 genes in HeLa3rd cells and HeLa3rd þ DNMT3B RNAi cells. The high- and low-expression levels are shown in red and green, respectively. (c) Relative probe intensity on the microarray at the PTPRR promoter in HeLa3rd cells compared with HeLa3rd þ DNMT3Bi cells. The promoter of PTPRR at chromosome 12 (genomics position) was shown. The black arrow indicates the position and direction of the transcription start site (primary transcript). The vertical red bars indicate the locations of probes on the genome and intensities of probes on the arrays (control/DMT3Bi probe intensity). One bar represents one probe. The positive bar means the probe intensity is higher in control cell and vice versa. (d) Reexpression of mRNA of PTPRR after DNMT3Bi. (e) qMS--PCR (left) and bisulfite sequencing (right) showed demethylation of the PTPRR promoter after DNMT3Bi. Filled circles represented methylated CpG sites and open circles represented unmethylated CpG sites. (f) PTPRR was reexpressed in HeLa3rd cells after treatment with 5-Aza-20-deoxycytidine. (g) The binding of DNMT3B and DNMT1 on the PTPRR promoter assessed by chromatin immunoprecipitation coupled with qPCR (ChIP--qPCR) decreased after DNMT3Bi. Data in d, e, f and g are presented as mean±s.d. (n ¼ 4). *Po0.05. A full colour version of this figure is available at the Oncogene journal online.

methylation on the expression of the PTPRR mRNA was confirmed investigate the role of PTPRR in cervical cancer, we generated using treatment with a demethylation agent, 5-Aza-20-deoxycyti- stable lines from HeLa3rd and CaSki cells that reexpressed PTPRR dine (Figure 3f). To confirm that methylation of the PTPRR (Supplementary Figure S3). PTPRR expression decreased the promoter was mediated by DNMT3B, chromatin immunoprecipi- proliferation of CaSki, but not HeLa3rd cells (Figure 4a). PTPRR tation coupled with qPCR analysis was used to test direct binding inhibited the migration and invasion capabilities of both cell lines of DNMT3B to PTPRR promoter. Binding of both DNMT3B and (Figure 4b-e). An inducible system (TREx-HeLa) expressing or DNMT1 decreased after DNMT3B knockdown (Figure 3g). Taken suppressing PTPRR was used to verify these results (Supplemen- together, these data suggest that methylation of the PTPRR tary Figure S4). promoter is mediated by DNMT3B in cooperation with DNMT1 in To understand further the mechanisms responsible for PTPRR- cervical cancer. related cancer invasion, we investigated the effects of PTPRR on the MAPK and AKT pathways. Phosphorylation of p44/42 MAPK, but not AKT, was inhibited by PTPRR (Figure 5a). The AP-1 family PTPRR expression phenocopies DNMT3B downregulation and members, including Fra-1, c-Fos and c-Jun, are the downstream inhibits MAPK signaling in cervical cancer targets of the MAPK pathway and are important in cervical cancer PTPRR removes a phosphate group from MAPK and affects neuron progression. Reexpression of PTPRR significantly suppressed the growth in the mouse,24,25 but its function in cancer is unknown. To expression of the Fra-1, c-Fos and c-Jun mRNAs (Figure 5b). As the

Oncogene (2013) 15 --26 & 2013 Macmillan Publishers Limited Epigenetic silencing of PTPRR P-H Su et al 19 0.7 1.0 HeLa3rd+Vector CaSki+Vector * 0.6 HeLa3rd+PTPRR CaSki+PTPRR 0.8 0.5 0.4 0.6

0.3 0.4 0.2 0.2 Absorbance (492 nm)

0.1 Absorbance (492 nm) 0.0 0.0 01234 01234 Days Days

HeLa3rd CaSki 1.2 Vector Vector PTPRRVector PTPRR PTPRR 1.0

0.8 * * 0.6 Relative cell migration

0.0 HeLa3rd CaSki

HeLa3rd CaSki Vector Vector PTPRRVector PTPRR 1.2 PTPRR 1.0 * 0.8

0.6 * 0.4 Relative cell numbers 0.0 HeLa3rd CaSki Figure 4. Effects of PTPRR on malignant phenotypes. (a) PTPRR expression affected cell growth in CaSki cells but not HeLa3rd cells. The migration (b, c) and invasion (d, e) capabilities were suppressed after PTPRR reexpression. The dotted line in (b) indicates the migration start site. Data in a, c and e are presented as mean±s.d. (n ¼ 4). *Po0.05. binding of AP-1 to the long control region of HPV increases the with U0126 caused dephosphorylation of p44/42 MAPK, expression of the E6 and E7 oncoproteins, which are an important suppression of AP-1, expression of HPV oncogenes and activation driving force in the development of cervical cancer and of CDH1, although there were some exceptions (Supplementary maintenance of malignant phenotypes, we propose that PTPRR Figure S7). limits the expression of HPV oncogenes by inhibiting AP-1. The expression of E6 and E7 was suppressed in cell lines reexpressing PTPRR inhibits tumor formation and metastasis in vivo PTPRR (Figure 5c). EMT is an important phenomenon in cancer Xenograft transplantation was used to analyze the effects of invasion and is related to MAPK pathway.26,27 Our results showed PTPRR on tumor formation in vivo. PTPRR significantly inhibited that PTPRR downregulated EMT markers ZEB1, SIP1 and fibro- tumor formation induced by CaSki cells, but not by HeLa3rd cells nectin, upregulated an epithelial marker, CDH1, in cervical cancer (Figures 6a and b). PTPRR inhibited lung metastasis associated cell lines (Figure 5d and Supplementary Figure S5). These results with both cell lines in mice (Figures 6c-f). Two mice injected with suggest that PTPRR affects the malignant phenotypes of cervical the HeLa3rd vector control had brain metastasis and were cancer by inhibiting the phosphorylation of p44/42 MAPK, which euthanized 2 months after implantation, together with two subsequently maintains the cell adhesion molecule, CDH1, and experimental mice (Figure 6ci). The other animals were euthanized inhibits the expression of HPV oncogenes. DNMT3B was reported 3 months after implantation (Figure 6cii). Pathological examina- recently as a downstream target of MAPK signaling.28 To test the tion confirmed the inhibitory effect of PTPRR on metastasis potential reciprocal regulation of DNMT and PTPRR via MAPK (Figures 6g and h). signaling, we assessed the inhibition of the expression of DNMTs (Figure 5e) and demethylation of the PTPRR promoter (Figure 5f) after reexpression of PTPRR; our results confirmed our hypothesis. Clinical relevance of PTPRR in cervical cancer Although PTPRR expression phenocopies DNMT3B downregula- To investigate the clinical relevance of PTPRR, its protein tion, the molecular changes, such as EMT patterns, were not expression levels and DNA methylation status were assessed in identical (Supplementary Figure S5 and S6), indicating that PTPRR human tissues. Immunohistochemical examination of samples is not the only gene mediated by DNMT3B in cervical cancer with cervical neoplasia showed that the expression of PTPRR was metastasis. minimal in normal squamous epithelium and was increased in CIN. An inhibitor of the MAPK pathway, U0126, was used to verify Expression was greatest in samples with in situ carcinoma, the suppressive effect of PTPRR on the MAPK pathway. Treatment whereas it was absent in cells that had become invasive and

& 2013 Macmillan Publishers Limited Oncogene (2013) 15 --26 Epigenetic silencing of PTPRR P-H Su et al 20

HeLa3rd+Vector 1.4 CaSki+Vector 1.2 HeLa3rd+PTPRR 1.2 CaSki+PTPRR 1.0 * 1.0 0.8 HeLa3rd CaSki 0.8 0.6 Vector ++ 0.6 * PTPRR ++0.4 0.4 * * * p-p44/42 MAPK 0.2 0.2 Relative mRNA expression Relative mRNA expression * 0.0 0.0 Fra-1 c-Fosc-Jun Fra-1c-Fos c-Jun p44/42 MAPK 1.2 HeLa3rd-Vector 1.2 CaSki+Vector CaSki+PTPRR pAKT HeLa3d-PTPRR 1.0 1.0

0.8 AKT 0.8 0.6 0.6 β-Actin 0.4 0.4 * * *

Relative mRNA epression 0.2 0.2 * Relative mRNA epression

0.0 0.0 E6 E7 E6 E7

8 7 1.2 CaSki+Vector HeLa3rd+Vector CaSki+Vector 1.4 HeLa3rd+Vector * CaSki+PTPRR HeLa3rd+PTPRR CaSki+PTPRR * HeLa3rd+PTPRR 6 1.2 1.0 6 1.0 * 5 0.8 2 2 0.8 * 0.6 0.6 0.4 1 0.4 * * * * * 0.2

* Relative mRNA expression 0.2 Relative mRNA expression Relative mRNA expression

Relative mRNA expression * * * * 0 0 0.0 0.0 ZEB1 SIP1fibronectin CDH1 ZEB1SIP1fibronectin CDH1 DNMT1 DNMT3A DNMT3B DNMT1DNMT3A DNMT3B

1.4 Vector PTPRR 1.2

1.0

0.8 * 0.6 * 0.4

0.2 Relavive PTPRR methylation 0.0 HeLa3rd CaSki Figure 5. Effects of PTPRR expression on MAPK signaling, HPV E6/E7 oncogenes, EMT, DNMTs expression and endogenous PTPRR promoter demethylation. (a) Immunoblotting showed that PTPRR inhibited the phosphorylation of p44/42 MAPK, but not AKT, in HeLa3rd and CaSki cells. PTPRR suppressed Fra-1, c-Fos and c-Jun (b), E6/E7 (c) and EMT markers ZEB1, SIP1 and fibronectin (d) in both HeLa3rd (left) and CaSki (right) cells. CDH1 was restored (d). The expression of DNMTs decreased (e) and the promoter of PTPRR was demethylated (f) in PTPRR- expressing cell lines. Data in b, c, d, e and f are presented as mean±s.d. (n ¼ 4). *Po0.05.

metastatic (Figure 7a). qMS--PCR analysis detected methylation of DISCUSSION the PTPRR promoter in most invasive cancer tissues. Cervical To date, the role of epigenetic modifications in cancer invasion scrapings are better suited to test potential clinical applications, and metastasis, which are the major characteristics of cancer, has compared with tissue samples. We analyzed cervical scrapings not been studied extensively. The expression of DNMTs has been from patients exhibiting the full spectrum of cervical lesions. reported in cancers;29 for example, DNMT3B is overexpressed in The methylation ratio of the PTPRR promoter increased with ovarian, breast, lymphoma, kidney, bladder and colon cancers.29 --33 disease severity, especially at the transition from in situ to invasive Genetically, a gain of DNMT3B copy number has been reported in carcinoma (Figure 7b). Bisulfite sequencing confirmed the HPV-immortalized keratinocytes and in cervical cancer tissues.19 hypermethylation patterns in samples with invasive cancer, but The functional relevance of DNMT3B expression in cancer not in samples of normal cervix (Figure 7c). Taken together, these phenotypes is unknown, even though one report showed that clinical results support our proposal that PTPRR is a tumor interference of DNMT3B inhibits cell migration (but not invasion) suppressor involved in cancer invasion and metastasis and in vitro in a prostate cancer cell line.34 Our study presented data suggest that PTPRR DNA methylation is a potential biomarker that show the unique role of DNMT3B in tumor growth and for cervical cancer screening. metastasis in a cervical cancer model. We also found that a

Oncogene (2013) 15 --26 & 2013 Macmillan Publishers Limited Epigenetic silencing of PTPRR P-H Su et al 21

HeLa3rd HeLa3rd Vector Vector

PTPRR PTPRR 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Tumor weight (g)

CaSki CaSki Vector Vector

PTPRR *

PTPRR 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Tumor weight (g)

HeLa3rd i ii Vector HeLa3rd Vector

PTPRR * PTPRR 0 2 4 6 8 10 12 14 16 18 Metastatic nodules

CaSki Vector CaSki Vector

PTPRR *

PTPRR 0 10 20 30 40 50 60 Metastatic nodules

HeLa3rd HeLa3rd CaSki Vector Vector Vector

PTPRR PTPRR PTPRR

Figure 6. Effects of PTPRR on tumorigenicity and metastasis in vivo. (a) Photographs of the tumors excised from the mice on day 30. (b) Tumor weights at the end of the in vivo experiment. (c, d) Photographs of the lungs from mice inoculated with tumors via tail vein injection. Arrows indicate the metastatic nodules in lungs. Arrows indicated the metastatic nodules in the lungs of mice with HeLa3rd þ vector. No gross nodules were seen in the lungs of mice with HeLa3rd þ PTPRR. (e, f) Histogram of metastatic nodules at the end of the in vivo experiment shown in (c, d). Hematoxylin and eosin staining of the brains (g) from paralyzed mice and lungs (h) from the mice shown in (c, d). Circles represent metastatic nodules. The scale bar represents 1 mm. Data in b, e and f are presented as mean±s.d. (n ¼ 4). *Po0.05. previously unidentified gene, PTPRR, is silenced by DNMT3B, which that DNMT1 can maintain the methylated genome by itself,35 leads to the activation of MAPK signaling and EMT. especially at high-density CpG sites or repeated sequences. A As DNMT3B is thought to have a role in de novo methylation, supportive role for DNMT3 in the maintenance of methylation has the knockdown of DNMT3B in cancer may not affect the been suggested (see review by Jones and Liang36). An updated methylation patterns already established. However, it is doubtful model of DNA methylation maintenance proposed that DNMT3A/

& 2013 Macmillan Publishers Limited Oncogene (2013) 15 --26 Epigenetic silencing of PTPRR P-H Su et al 22 Normal, n= 33

**** p for trend < 0.0001

CIN1, n = 15

CIN2, n = 7

CpG sites CIN3/CIS, n = 18 SCC tissue 1

SCC tissue 2 SCC, n = 53

Normal cervix 1

Meta SCC, n = 10 Normal cervix 2

Figure 7. The protein expression and promoter methylation status of PTPRR in samples across the full spectrum of cervical lesions. (a) Immunohistochemistry staining of PTPRR in tissue arrays comprising the full spectrum of disease severity of cervical cancer. The scale bar represents 100 mm. (b) Scatter plot showing the methylation level of PTPRR in cervical scrapings and tissues detected by TaqMan qMS--PCR. The PTPRR methylation can be detected in most invasive cervical cancer tissues, but rarely in normal cervix. ****Po0.0001. Significant trends were observed in the methylation ratios in clinical scrapings; P for trend o0.0001. The median for each group is indicated by a horizontal bar. (c) Bisulfite sequencing results in cervical cancer tissues and in normal cervix. Filled circles represented methylated CpG sites and open circles represented unmethylated CpG sites.

3B interacts with nucleosomes bearing specific modifications, and stages. We propose a model to explain this regulatory loop then cooperates with DNMT1 and other enzymes at the between the DNMT epigenetic machinery and MAPK oncogenic replication fork to maintain methylation patterns, especially at signaling. We speculate that, in oncogenic-responsive stresses, the repeats and imprinted genes. A later report suggested that the activation of PTPRR inhibits the phosphorylation of p44/42 MAPK, role of DNMT3B in the maintenance of DNA methylation is which inhibits AP-1 and the subsequent expression of E6 and E7, associated with H2AK119 monoubiquitination.37 Three repeated and keeps dysplastic cells in situ. However, when oncogenic sequences are close to the PTPRR promoter, at positions À32 to stresses overrule this checkpoint, DNMT3B expression may À72 (simple repeat), À847 to À1027 (LINE) and À1544 to À1800 cooperate with DNMT1 to promote de novo methylation of some (LINE), and may explain why PTPRR reexpresses after DNMT3B metastasis-suppressor genes. A recent report showing K-Ras- knockdown. induced DNMT3B binding to the CDH1 promoter, which resulted The expression of PTPRR in precancerous lesions in accordance in increased methylation and invasive migration, supports our with disease severity in tissue sections may fail to support the role speculation.28 De novo methylation of PTPRR may gradually inhibit of PTPRR as a tumor suppressor gene. However, its silencing in its expression, which results in the activation of MAPK signaling invasive cancer lesions and our in vitro and in vivo studies suggest and its downstream effectors. In return, the activated MAPK that PTPRR is a bona fide tumor suppressor, the expression of signaling upregulates DNMTs, which then help to consolidate the which increases in response to oncogenic stress in precancer methylation silencing of PTPRR (Supplementary Figure S8).

Oncogene (2013) 15 --26 & 2013 Macmillan Publishers Limited Epigenetic silencing of PTPRR P-H Su et al 23 Understanding what and how the stresses overrule the guardian extent to which our proposed model is amplified in relation to function of PTPRR may lead to new approaches to prevent and HPV oncogenes and HPV-related tumors, such as head and neck treat cancer. cancer, remains to be determined.63 Protein phosphorylation is important in cell signaling and in the EMT has been studied extensively as a mechanism of cancer regulation of cellular functions. The balance of protein tyrosine invasion.64,65 In the present study, PTPRR affected the invasion phosphorylation is a dynamic process that is controlled by and metastasis phenotypes of cervical cancers without changing tyrosine kinases and protein tyrosine phosphatases (PTPs). PTPs the complete set of EMT markers. Accumulating evidence, are classified by domain and fall into two subgroups: the receptor including our previous study, supports the role had by incomplete type (those with transmembrane domains and receptor-like EMT in cervical cancer progression.13,64 Changes in EMT markers extracellular domains) and the nonreceptor type (those without could begin as early as the immortalization of cells after transmembrane domains). In cancer development, the tumor- HPV infection.66,67 An incomplete EMT is sufficient to induce suppressive or oncogenic role of PTPs may depend on the metastasis.68 substrates. The expression of receptor-type PTPs, such as PTPRA, From the clinical point of view, DNA methylation has been PTPRE, PTPRF and PTPRH, is increased in tumors, but the effects of proposed as a potential biomarker for cervical cancer detection.16 this increase in cancers are not clear.38 The receptor-type PTPs, The feasibility of using methylation markers for cervical malig- such as PTPRD and PTPRO, are tumor suppressors; PTPRD inhibits nancy screening has been appraised critically.16,69,70 Various tumor growth in glioblastoma39 and PTPRO reduces proliferation genome-wide strategies have been used to discover genes and anchorage-independent growth in lung cancer.40 The PTPRR undergoing methylation silencing in cervical cancers and suggest knockout mouse has defects in motor coordination and balance.41 the potential for some newly discovered genes.17,71,72 Our present The roles of PTPRR in development and cancer are not clear. The study found that PTPRR is highly methylated in invasive cancer association between PTPRR and cancer was reported only recently tissues. Although protein expression in CIS is high, we detected for colon cancer.42 The methylation of PTPRR can be detected in the DNA methylation of PTPRR, indicating the molecular precancerous colorectal lesions, suggesting a role for this protein propensity of some in situ cells toward cancer invasion. This early in colon cancer development. However, the mechanistic methylation was detected in cervical scrapings and may provide a relevance of PTPRR in cancer biology has not been reported. specific biomarker for cervical cancer screening. The discrepancy PTPRR dephosphorylates p44/42 MAPK in response to some between PTPRR protein expression and DNA methylation may growth factors in cell lines.24,43,44 Molecules that inhibit the MAPK attribute to the difference of methodology. Protein expression signaling, such as nonmetastatic 23, Raf kinase inhibitor protein, revealed by Immunohistochemistry may not identify cells with mitogen-activated protein kinase kinase 4 and mitogen-activated DNA methylation, which can be detected by MS--qPCR. Therefore, protein kinase kinase 6, inhibit metastasis.45,46 We characterized both PTPRR protein expression and methylation can be detected the tumor-suppressor function of PTPRR, especially its role in in CIS lesions. Although the detection of PTPRR methylation in CIN metastasis via the MAPK pathway, in a cervical cancer model. The lesions is not satisfactory, the specificity for high-grade lesions is effects of PTPRR reexpression and the specific p44/42 MAPK very high, which may contribute to the potential gene panel for pathway inhibitor, U0126, were not the same, suggesting that cancer detection. Further standardization of the methylation assay PTPRR affects malignant phenotypes by dephosphorylating and use of validation tests in women with different geographic other substrates, such as ERK5 and p38 MPAK, as shown in other and ethnic backgrounds may help validate these results and species.43,47 The extent of kinases targeted by PTPRR in cancer translate this bench discovery to a better cervical cancer screening remains to be investigated. in the future. HPV is necessary for the development of cervical cancer, with the expression of the viral oncoproteins E6 and E7 being the driving force.2,48 The long control region of HPV is regulated by MATERIALS AND METHODS 49,50 many transcription factors, including AP-1. AP-1 transcription Please refer to the supporting information online for a detailed description factors, such as Fra-1, c-Fos and c-Jun, are the downstream targets of the experimental procedures. of the MAPK pathway and affect cancer cell growth and 51 --54 metastasis. In cervical cancer, both the AP-1 and MAPK Patients pathways are activated, although the mechanism of MAPK pathway activation is not clear.55 --57 Our study demonstrated that Cervical scrapings and tissues were from a hospital-based, retrospective, case-control study at the National Defense Medical Center, Taipei, Taiwan the DNA-methylation-mediated silencing of PTPRR contributed to 15 the activation of MAPK pathways in cervical cancer. The binding of as described previously. Informed consent was obtained for all patients AP-1 to the long control region increases promoter activity, and involved in this study and this study was approved by the Institutional the expression of E6/E7 correlates with cervical cancer progres- Review Board of the Tri-Service General Hospital, Taipei, Taiwan. Cervical sion.49 --50,57 Although the modulation of E6/E7 expression by AP-1 scraping DNA were collected on 338 patients, including women who had was known to affect cell growth rather than metastasis in cervical a normal uterine cervix (n ¼ 161), CIN1 (n ¼ 38), CIN2 (n ¼ 39), CIN3/CIS cancer, the ability of AP-1 in promoting metastasis through EMT in (n ¼ 60), squamous cell carcinoma (SCC) or adenocarcinoma (n ¼ 60). many cancers were reported.52,58,59 In addition, repression of E7 Cervical cancer tissue DNA were collected on 53 patients including women was reported to restore CDH1 expression in cervical cancer.60 who had SCC (n ¼ 34) and adenocarcinoma (n ¼ 19). Cervical tissue RNA The different effects of DNMT3B interference on phenotypes of were collected on 23 patients including women who had normal uterine HeLa3rd and CaSki are interesting. These differences may be HPV cervix (n ¼ 7) and SCC (n ¼ 16). Cytological, histological and clinical data for genotype or cell type dependent. The HPV E7 oncoprotein was all patients were reviewed by a panel of colposcopists, cytologists and recently reported to bind to, and to augment, DNMT1 enzyme pathologists. The uterine cervix diagnoses were according to histological activity, which causes the epigenetic suppression of CDH1.61,62 reports except normal uterine cervix, which were according to cytological However, the differential interactions between E7 from different reports. Normal cervixes in the tissue array and for RNA extraction were HPV genotypes and DNMT1 remain unknown. We speculate that obtained from patients with benign gynecological diseases. The diagnoses the expression of PTPRR may suppress E7 expression, inhibit the were made by histopathology. E7-DNMTs-mediated epigenetic effects and inhibit cancer pro- gression. However, the interaction between E6 or E7 and DNMT3B Cell lines, culture conditions and constructs has not been reported. The interactions between HPV oncopro- The human cervical cancer cell lines HeLa3rd and CaSki were used in this teins and DNMT3A/B also warrant further investigations. The study. The process of selecting subline (HeLa3rd) is described in

& 2013 Macmillan Publishers Limited Oncogene (2013) 15 --26 Epigenetic silencing of PTPRR P-H Su et al 24 Supplementary Information.18 Briefly, HeLa cells were seeded into the In vivo tumorigenicity and metastasis models transwell inserts (BD Bioscience, San Jose, CA, USA) that coated with Six-week-old CB-17 SCID mice were used in the tumorigenicity and Matrigel (BD Biosciences) and incubated for 72 h at 37 1C. The cells that metastasis analysis. All animal studies were approved by the Institutional migrated through the membrane and attached to the lower chamber were Animal Care and Use Committee of the National Defence Medical Centre, harvested and expanded for the second round of selection. The subline Taipei, Taiwan. In the tumorigenicity analysis, 106 cells from each stable of the third-round selection was designated as HeLa3rd. DNMT3B RNAi line were injected subcutaneously into both flanks of each mouse. The (50-GATCCCCAGATGACGGATGCCTAGAGTTCAAGAGACTCTAGGCATCCGTCA mice were killed at day 30. To investigate the metastatic properties, 106 TCTTTTTTGGAAA-30. Sequences in underlined are RNAi sequence)73 was cells were injected through the tail vein. Animals were killed at 12 weeks cloned into pSuper vector (OligoEngine, Seattle, WA, USA). PTPRR and examined for the presence of metastases. The brain and lung were (NM_130846) was constructed by inserting a full-length cDNA product then excised, stained with Bouin’s fixative and prepared for histopatho- into a pT-REx-DEST31 Gateway Vector (Invitrogen, Carlsbad, CA, USA). logical analysis.53

RNA extraction, cDNA synthesis and qRT--PCR Tissue microarray and immunohistochemistry Total RNA was isolated from each sample using a Qiagen RNeasy kit Paraffin-embedded cervical tissues of Chinese patients were retrieved from (Qiagen, Hilden, Germany). RNA was reverse transcribed to cDNA using the the Department of Pathology, National Defense Medical Center, Taiwan. SuperScript III first-strand synthesis system for RT--PCR (Invitrogen). The tissue microarrays comprized histologically normal squamous qRT--PCR was performed using RT2 SYBR Green qPCR Master Mixes epithelial samples (n ¼ 33), CIN1 (n ¼ 15), CIN2 (n ¼ 7), CIN3/CIS (n ¼ 18), (SABiosciences, Frederick, MD, USA) in an ABI 7500 Real-Time PCR system SCC (n ¼ 53) and metastatic SCC (n ¼ 10). A standard protocol for (Applied Biosystems, Carlsbad, CA, USA). The primers used in this study are immunohistochemistry was used with rabbit polyclonal anti-human PTPRR shown in Supplementary Table S3. antibody (ab12153, Abcam).

Demethylation treatment, bisulfite modification, qMS--PCR and Statistical analysis bisulfite sequencing SPSS (version 15) for Windows (IBM, Armonk, NY, USA) was used to analyze Cells were treated with DMSO or 5-Aza-20-deoxycytidine (Sigma, St Louis, the data. The Mann--Whitney U test was used to compare cell proliferation, MO, USA) for the demethylation and gene reexpression analysis. DNA and migration, invasion and relative RNA expression and promoter methylation RNA were extracted for further analysis after treatment. Bisulfite in the different stable transfectants or tissues. The significance of observed modification was performed using a CpGenome Fast DNA Modification trends in methylation data of clinical swabs was determined by linear Kit (Millipore, Bedford, MA, USA). qMS--PCR was performed in a TaqMan regression modeling by coding the diagnosis as an ordinal independent probe system using the LightCycler 480 Real-Time PCR System (Roche, variable. Indianapolis, IN, USA). PCR products for bisulfite sequencing were cloned into yT&A vector (Yeastern Biotech, Taipei, Taiwan) and then sequenced. CONFLICT OF INTEREST Methyl-DNA immunoprecipitation coupled with microarray The patent of using PTPRR DNA methylation as a cancer biomarker is pending. The analysis National Defense Medical Center owns the patent. H-C Lai is the inventor. All other authors declare no conflict of interest. Methyl-DNA immunoprecipitation was performed as described previously in the literature74 with anti-5-methyl cytosine antibody (ab1884; Abcam, Cambridge, UK). The immunoprecipitated DNA was amplified using a ACKNOWLEDGEMENTS whole-genome amplification kit (WGA2, Sigma). Amplified DNA was This work was supported by Grant NSC98-2314-B-016-030-MY3 from the National hybridized to HG18_promoter 2 array (Roche NimbleGen, Madison, Science Council, Taiwan, ROC (to H-CL); Grant TSGH-C100-010-014-S01 (to M-HY) and WI, USA). TSGH-C100-010-014-S02 (H-CL) from the Tri-Service General Hospital and Teh-Tzer Study Group for Human Medical Research Foundation (to H-CL). Chromatin immunoprecipitation coupled with qPCR Chromatin immunoprecipitation assays were performed according to the REFERENCES protocol from Millipore (EZ--Magna ChIP G Chromatin Immunoprecipita- 1 WHO/ICO Information Centre on HPV and Cervical Cancer. Human Papillomavirus tion Kit, Millipore). The antibodies used in the chromatin immunoprecipita- and Related Cancers in World. Summary Report 2010 2010. 2010/11/15. tion coupled with qPCR analysis were anti-DNMT3b (ab13604, Abcam), 2 zur Hausen H. Cervical carcinoma and human papillomavirus: on the road to anti-DNMT1 (ab16632, Abcam) and anti-mouse IgG (ab37355, Abcam). preventing a major human cancer. J Natl Cancer Inst 2001; 93: 252 --253. 3 Moody CA, Laimins LA. Human papillomavirus oncoproteins: pathways to Cell proliferation, migration and invasion assays transformation. Nat Rev Cancer 2010; 10: 550 --560. 4 Ostor AG. Natural history of cervical intraepithelial neoplasia: a critical review. Int J Cells were seeded in 96-well plates at a density of 2000 cells per well. On Gynecol Pathol 1993; 12: 186 --192. days 0, 1, 2, 3 and 4, cell viability was measured using an MTS assay 5 Nguyen DX, Massague J. Genetic determinants of cancer metastasis. Nat Rev (C CellTiter 96 AQueous Non-Radioactive Cell Proliferation Assay, Promega, Genet 2007; 8: 341 --352. Madison, WI, USA). Cell migration assays were performed by using an Oris 6 Jaenisch R, Bird A. Epigenetic regulation of gene expression: how the genome Universal Cell Migration Assembly kit, as instructed by the manufacturer integrates intrinsic and environmental signals. Nat Genet 2003; 33(Suppl): (Platypus Technologies, Madison, WI, USA). Cell invasion was measured in 245 --254. the Transwell system (BD Bioscience). The chamber membrane was coated 7 Esteller M. Epigenetics in cancer. N Engl J Med 2008; 358: 1148 --1159. 8 Rodenhiser DI. Epigenetic contributions to cancer metastasis. Clin Exp Metastasis with Matrigel (BD Bioscience). 2009; 26:5--18. 9 Sharma S, Kelly TK, Jones PA. Epigenetics in cancer. Carcinogenesis 2010; 31: Immunoblot analysis 27 --36. The antibodies used in the immunoblot analysis were anti-DNMT3B 10 Paschos K, Allday MJ. Epigenetic reprogramming of host genes in viral and microbial pathogenesis. Trends Microbiol 2010; 18: 439 --447. (ab16049, Abcam), anti-PTPRR (ab88598, Abcam), anti-phospho-p44/42 11 Woodman CB, Collins SI, Young LS. The natural history of cervical HPV infection: MAPK (Thr202/Tyr204) (9106, Cell Signaling, Danvers, MA, USA), anti-p44/ unresolved issues. Nat Rev Cancer 2007; 7: 11 --22. 42 MAPK (9102, Cell Signaling), anti-phospho-AKT (Ser473) (9271, Cell 12 Chung MT, Sytwu HK, Yan MD, Shih YL, Chang CC, Yu MH et al. Promoter Signaling), anti-AKT (9272, Cell Signaling) and anti-b-actin (ab8226, methylation of SFRPs gene family in cervical cancer. Gynecol Oncol 2009; 112: Abcam). 301 --306.

Oncogene (2013) 15 --26 & 2013 Macmillan Publishers Limited Epigenetic silencing of PTPRR P-H Su et al 25 13 Liu CY, Chao TK, Su PH, Lee HY, Shih YL, Su HY et al. Characterization of LMX-1A as 39 Veeriah S, Brennan C, Meng S, Singh B, Fagin JA, Solit DB et al. The tyrosine a metastasis suppressor in cervical cancer. J Pathol 2009; 219: 222 --231. phosphatase PTPRD is a tumor suppressor that is frequently inactivated and 14 Lin YW, Chung MT, Lai HC, De Yan M, Shih YL, Chang CC et al. Methylation analysis mutated in glioblastoma and other human cancers. Proc Natl Acad Sci USA 2009; of SFRP genes family in cervical adenocarcinoma. J Cancer Res Clin Oncol 2009; 106: 9435 --9440. 135: 1665 --1674. 40 Motiwala T, Kutay H, Ghoshal K, Bai S, Seimiya H, Tsuruo T et al. Protein tyrosine 15 Lai HC, Lin YW, Huang RL, Chung MT, Wang HC, Liao YP et al. Quantitative DNA phosphatase receptor-type O (PTPRO) exhibits characteristics of a candidate methylation analysis detects cervical intraepithelial neoplasms type 3 and worse. tumor suppressor in human lung cancer. Proc Natl Acad Sci USA 2004; 101: Cancer 2010; Jun:8. 13844 --13849. 16 Wentzensen N, Sherman ME, Schiffman M, Wang SS. Utility of methylation 41 Chirivi RG, Noordman YE, Van der Zee CE, Hendriks WJ. Altered MAP kinase markers in cervical cancer early detection: appraisal of the state-of-the-science. phosphorylation and impaired motor coordination in PTPRR deficient mice. Gynecol Oncol 2009; 112: 293 --299. J Neurochem 2007; 101: 829 --840. 17 Lai HC, Lin YW, Huang TH, Yan P, Huang RL, Wang HC et al. Identification of novel 42 Menigatti M, Cattaneo E, Sabates-Bellver J, Ilinsky VV, Went P, Buffoli F et al. The DNA methylation markers in cervical cancer. Int J Cancer 2008; 123: 161 --167. protein tyrosine phosphatase receptor type R gene is an early and frequent target 18 Chung MT, Lai HC, Sytwu HK, Yan MD, Shih YL, Chang CC et al. SFRP1 and SFRP2 of silencing in human colorectal tumorigenesis. Mol Cancer 2009; 8: 124. suppress the transformation and invasion abilities of cervical cancer cells through 43 Blanco-Aparicio C, Torres J, Pulido R. A novel regulatory mechanism of MAP Wnt signal pathway. Gynecol Oncol 2009; 112: 646 --653. kinases activation and nuclear translocation mediated by PKA and the PTP-SL 19 Wilting SM, Snijders PJ, Meijer GA, Ylstra B, van den Ijssel PR, Snijders AM et al. tyrosine phosphatase. J Cell Biol 1999; 147: 1129 --1136. Increased gene copy numbers at chromosome 20q are frequent in both 44 Pulido R, Zuniga A, Ullrich A. PTP-SL and STEP protein tyrosine phosphatases squamous cell carcinomas and adenocarcinomas of the cervix. J Pathol 2006; 209: regulate the activation of the extracellular signal-regulated kinases ERK1 220 --230. and ERK2 by association through a kinase interaction motif. EMBO J 1998; 17: 20 Talora C, Sgroi DC, Crum CP, Dotto GP. Specific down-modulation of 7337 --7350. Notch1 signaling in cervical cancer cells is required for sustained HPV-E6/E7 45 Berger JC, Vander Griend DJ, Robinson VL, Hickson JA, Rinker-Schaeffer CW. expression and late steps of malignant transformation. Genes Dev 2002; 16: Metastasis suppressor genes: from gene identification to protein function and 2252 --2263. regulation. Cancer Biol Ther 2005; 4: 805 --812. 21 Lichtig H, Gilboa DA, Jackman A, Gonen P, Levav-Cohen Y, Haupt Y et al. HPV16 46 Hickson JA, Huo D, Vander Griend DJ, Lin A, Rinker-Schaeffer CW, Yamada SD. The E6 augments Wnt signaling in an E6AP-dependent manner. Virology 2010; 396: p38 kinases MKK4 and MKK6 suppress metastatic colonization in human ovarian 47 --58. carcinoma. Cancer Res 2006; 66: 2264 --2270. 22 Gu W, Yeo E, McMillan N, Yu C. Silencing oncogene expression in cervical cancer 47 Buschbeck M, Eickhoff J, Sommer MN, Ullrich A. Phosphotyrosine-specific stem-like cells inhibits their cell growth and self-renewal ability. Cancer Gene Ther phosphatase PTP-SL regulates the ERK5 signaling pathway. J Biol Chem 2002; 2011; 18: 897 --905. 277: 29503 --29509. 23 Okano M, Bell DW, Haber DA, Li E. DNA methyltransferases Dnmt3a and Dnmt3b 48 Narisawa-Saito M, Kiyono T. Basic mechanisms of high-risk human papillomavirus- are essential for de novo methylation and mammalian development. Cell 1999; 99: induced carcinogenesis: roles of E6 and E7 proteins. Cancer Sci 2007; 98: 247 --257. 1505 --1511. 24 Noordman YE, Jansen PA, Hendriks WJ. Tyrosine-specific MAPK phosphatases and 49 Kikuchi K, Taniguchi A, Yasumoto S. Induction of the HPV16 enhancer activity by the control of ERK signaling in PC12 cells. J Mol Signal 2006; 1:4. Jun-B and c-Fos through cooperation of the promoter-proximal AP-1 site and the 25 Schmitt I, Bitoun E, Manto M. PTPRR, cerebellum, and motor coordination. epithelial cell type--specific regulatory element in fibroblasts. Virus Genes 1996; Cerebellum 2009; 8:71--73. 13:45--52. 26 Horiguchi K, Shirakihara T, Nakano A, Imamura T, Miyazono K, Saitoh M. Role of 50 Butz K, Hoppe-Seyler F. Transcriptional control of human papillomavirus (HPV) Ras signaling in the induction of snail by transforming growth factor-beta. J Biol oncogene expression: composition of the HPV type 18 upstream regulatory Chem 2009; 284: 245 --253. region. J Virol 1993; 67: 6476 --6486. 27 Strippoli R, Benedicto I, Perez Lozano ML, Cerezo A, Lopez-Cabrera M, 51 Chen H, Zhu G, Li Y, Padia RN, Dong Z, Pan ZK et al. Extracellular signal-regulated del Pozo MA. Epithelial-to-mesenchymal transition of peritoneal mesothelial kinase signaling pathway regulates breast cancer cell migration by maintaining cells is regulated by an ERK/NF-kappaB/Snail1 pathway. Dis Model Mech 2008; 1: slug expression. Cancer Res 2009; 69: 9228 --9235. 264 --274. 52 Eferl R, Wagner EF. AP-1: a double-edged sword in tumorigenesis. Nat Rev Cancer 28 Kwon O, Jeong SJ, Kim SO, He L, Lee HG, Jang KL et al. Modulation of E-cadherin 2003; 3: 859 --868. expression by K-Ras; involvement of DNA methyltransferase-3b. Carcinogenesis 53 Lemieux E, Bergeron S, Durand V, Asselin C, Saucier C, Rivard N. Constitutively 2010; Apr:7. active MEK1 is sufficient to induce epithelial-to-mesenchymal transition in 29 Robertson KD, Uzvolgyi E, Liang G, Talmadge C, Sumegi J, Gonzales FA et al. The intestinal epithelial cells and to promote tumor invasion and metastasis. Int J human DNA methyltransferases (DNMTs) 1, 3a and 3b: coordinate mRNA Cancer 2009; 125: 1575 --1586. expression in normal tissues and overexpression in tumors. Nucleic Acids Res 54 de Wilde J, De-Castro Arce J, Snijders PJ, Meijer CJ, Rosl F, Steenbergen RD. 1999; 27: 2291 --2298. Alterations in AP-1 and AP-1 regulatory genes during HPV-induced carcinogen- 30 Ahluwalia A, Hurteau JA, Bigsby RM, Nephew KP. DNA methylation in ovarian esis. Cell Oncol 2008; 30: 77 --87. cancer. II. Expression of DNA methyltransferases in ovarian cancer cell lines and 55 Prusty BK, Das BC. Constitutive activation of transcription factor AP-1 in cervical normal ovarian epithelial cells. Gynecol Oncol 2001; 82: 299 --304. cancer and suppression of human papillomavirus (HPV) transcription and AP-1 31 Veeck J, Esteller M. Breast cancer epigenetics: from DNA methylation to activity in HeLa cells by curcumin. Int J Cancer 2005; 113: 951 --960. microRNAs. J Mammary Gland Biol Neoplasia 2010; 15:5--17. 56 Branca M, Ciotti M, Santini D, Bonito LD, Benedetto A, Giorgi C et al. Activation of 32 Girault I, Tozlu S, Lidereau R, Bieche I. Expression analysis of DNA methyl- the ERK/MAP kinase pathway in cervical intraepithelial neoplasia is related to transferases 1, 3A, and 3B in sporadic breast carcinomas. Clin Cancer Res 2003; 9: grade of the lesion but not to high-risk human papillomavirus, virus clearance, or 4415 --4422. prognosis in cervical cancer. Am J Clin Pathol 2004; 122: 902 --911. 33 Amara K, Ziadi S, Hachana M, Soltani N, Korbi S, Trimeche M. DNA 57 Kyo S, Klumpp DJ, Inoue M, Kanaya T, Laimins LA. Expression of AP1 during methyltransferase DNMT3b protein overexpression as a prognostic factor in cellular differentiation determines human papillomavirus E6/E7 expression in patients with diffuse large B-cell lymphomas. Cancer Sci 2010; 101 (7): 1722 --1730. stratified epithelial cells. J Gen Virol 1997; 78(Part 2): 401 --411. 34 Yaqinuddin A, Qureshi SA, Qazi R, Abbas F. Down-regulation of DNMT3b in PC3 58 Fujisawa T, Joshi BH, Puri RK. IL-13 regulates cancer invasion and metastasis cells effects locus-specific DNA methylation, and represses cellular growth and through IL-13Ra2 via ERK/AP-1 pathway in mouse model of human ovarian migration. Cancer Cell Int 2008; 8: 13. cancer. Int J Cancer (e-pub ahead of print 19 August 2011). 35 Rhee I, Jair KW, Yen RW, Lengauer C, Herman JG, Kinzler KW et al. CpG 59 Sankpal NV, Mayfield JD, Willman MW, Fleming TP, Gillanders WE. Activator methylation is maintained in human cancer cells lacking DNMT1. Nature 2000; protein 1 (AP-1) contributes to EpCAM-dependent breast cancer invasion. Breast 404: 1003 --1007. Cancer Res 2011; 13: R124. 36 Jones PA, Liang G. Rethinking how DNA methylation patterns are maintained. Nat 60 Caberg JH, Hubert PM, Begon DY, Herfs MF, Roncarati PJ, Boniver JJ et al. Silencing Rev Genet 2009; 10: 805 --811. of E7 oncogene restores functional E-cadherin expression in human papilloma- 37 Jin B, Yao B, Li JL, Fields CR, Delmas AL, Liu C et al. DNMT1 and DNMT3B modulate virus 16-transformed keratinocytes. Carcinogenesis 2008; 29: 1441 --1447. distinct polycomb-mediated histone modifications in colon cancer. Cancer Res 61 Burgers WA, Blanchon L, Pradhan S, de Launoit Y, Kouzarides T, Fuks F. Viral 2009; 69: 7412 --7421. oncoproteins target the DNA methyltransferases. Oncogene 2007; 26: 1650 --1655. 38 Ostman A, Hellberg C, Bohmer FD. Protein-tyrosine phosphatases and cancer. Nat 62 Laurson J, Khan S, Chung R, Cross K, Raj K. Epigenetic repression of E-cadherin by Rev Cancer 2006; 6: 307 --320. human papillomavirus 16 E7 protein. Carcinogenesis 2010; 31: 918 --926.

& 2013 Macmillan Publishers Limited Oncogene (2013) 15 --26 Epigenetic silencing of PTPRR P-H Su et al 26 63 Leemans CR, Braakhuis BJ, Brakenhoff RH. The molecular biology of head and 69 Nephew KP. What will it take to obtain DNA methylation markers for early cervical neck cancer. Nat Rev Cancer 2011; 11:9--22. cancer detection? Gynecol Oncol 2009; 112: 291 --292. 64 Christiansen JJ, Rajasekaran AK. Reassessing epithelial to mesenchymal transition 70 Szalmas A, Konya J. Epigenetic alterations in cervical carcinogenesis. Semin Cancer as a prerequisite for carcinoma invasion and metastasis. Cancer Res 2006; 66: Biol 2009; 19: 144 --152. 8319 --8326. 71 Sova P, Feng Q, Geiss G, Wood T, Strauss R, Rudolf V et al. Discovery of novel 65 Thiery JP, Acloque H, Huang RY, Nieto MA. Epithelial-mesenchymal transitions in methylation biomarkers in cervical carcinoma by global demethylation and development and disease. Cell 2009; 139: 871 --890. microarray analysis. Cancer Epidemiol Biomarkers Prev 2006; 15: 114 --123. 66 Hellner K, Mar J, Fang F, Quackenbush J, Munger K. HPV16 E7 oncogene 72 Wang SS, Smiraglia DJ, Wu YZ, Ghosh S, Rader JS, Cho KR et al. Identification of expression in normal human epithelial cells causes molecular changes indicative novel methylation markers in cervical cancer using restriction landmark genomic of an epithelial to mesenchymal transition. Virology 2009; 391:57--63. scanning. Cancer Res 2008; 68: 2489 --2497. 67 Nussenzweig MC, Alt FW. Antibody diversity: one enzyme to rule them all. Nat 73 Leu YW, Rahmatpanah F, Shi H, Wei SH, Liu JC, Yan PS et al. Double RNA Med 2004; 10: 1304 --1305. interference of DNMT3b and DNMT1 enhances DNA demethylation and gene 68 Rodriguez-Sastre MA, Gonzalez-Maya L, Delgado R, Lizano M, Tsubaki G, reactivation. Cancer Res 2003; 63: 6110 --6115. Mohar A et al. Abnormal distribution of E-cadherin and beta-catenin in 74 Weng YI, Huang TH, Yan PS. Methylated DNA immunoprecipitation and different histologic types of cancer of the uterine cervix. Gynecol Oncol 2005; microarray-based analysis: detection of DNA methylation in breast cancer cell 97: 330 --336. lines. Methods Mol Biol 2009; 590: 165 --176.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene (2013) 15 --26 & 2013 Macmillan Publishers Limited