Dickkopf-1 expression is repressed by oncogenic human papillomaviruses (HPVs) and regulates the Cisplatin sensitivity of HPV-positive cancer cells in a JNK-dependent manner

Oncogenic human papillomavirus (HPV) types control the phenotype of cervical cancer cells through the sustained expression of the viral E6/E7 oncogenes. Here, we show that they strongly restrain expression of the putative tumor suppressor protein Dkk1 (Dickkopf-1) in HPV-positive cervical cancer cells through the restriction of p53 expression by the continuously expressed endogenous E6 oncoprotein. More-over, our study reveals that compromised Dkk1 expression is linked to increased resistance of HPV-positive cervical cancer cells toward the proapoptotic activity of Cisplatin. Although Dkk1 can act as a Wnt antagonist, the antiapoptotic effect result-ing from Dkk1 repression is not linked to an activation of this pathway. Rather, transcriptome and functional analyses uncover that Dkk1 repression leads to a strongly diminished stimulation of c-Jun N-terminal kinase (JNK) signaling which is required for efficient apoptosis induction by Cisplatin in cervical cancer cells. Further, we observed that Dkk1-depleted cervical cancer cells induce senescence under Cisplatin treatment instead of apoptosis, suggesting that Dkk1 levels can strongly influence the phenotypic response of these cells toward Cisplatin. Collectively, these results provide new insights into the virus/host cell crosstalk in cervical cancer cells by iden-tifying Dkk1 as a cellular target which is maintained under strong negative control by the continuous expression of the HPV oncogenes. Moreover, they identify Dkk1 as a critical determinant for the sensitivity of cervical cancer cells toward Cisplatin, showing that Dkk1 repression leads to increased Cisplatin resistance by impairing proapoptotic JNK signaling. Oncogenic human papillomavirus (HPV) types cause cervical cancer and control the phenotype of cervical cancer cells through the sustained expression of viral oncogenes. Here, the authors provide new insights into the virus/host cell crosstalk in HPV-transformed cancer cells, revealing that the viral oncogenes strongly restrict endogenous expression of the putative tumor suppressor protein Dkk1 (Dickkopf-1). Moreover, they uncover that Dkk1 is a critical determinant of the response of cervical cancer cells to Cisplatin treatment, showing that Dkk1 repression leads to increased Cisplatin resistance by impairing proapoptotic c-Jun N-terminal kinase (JNK) signaling.


What's new?
Oncogenic human papillomavirus (HPV) types cause cervical cancer and control the phenotype of cervical cancer cells through the sustained expression of viral oncogenes. Here, the authors provide new insights into the virus/host cell crosstalk in HPV-transformed cancer cells, revealing that the viral oncogenes strongly restrict endogenous expression of the putative tumor suppressor protein Dkk1 (Dickkopf-1). Moreover, they uncover that Dkk1 is a critical determinant of the response of cervical cancer cells to Cisplatin treatment, showing that Dkk1 repression leads to increased Cisplatin resistance by impairing proapoptotic c-Jun N-terminal kinase (JNK) signaling.

| INTRODUCTION
Infections with oncogenic human papillomaviruses (HPVs) are closely related to 5% of cancer cases worldwide, of which more than 70% are attributable to HPV types 16 or 18. 1 Cancer of the cervix uteri, which is almost exclusively caused by HPV infection, accounts for 13.3% of all cancers in females. 1,2 Cervical cancer is expected to remain a major health problem for decades to come. Albeit vaccines protecting against infections with the most important oncogenic HPV types can efficiently prevent the development of cervical preneoplasias and cancer, 3 worldwide vaccination rates are still low. 1 In addition, since the vaccines are prophylactic, individuals who are already persistently infected with an oncogenic HPV are not protected from tumor development, a process which typically takes decades. 4 Together with the fact that advanced stage or recurrent cervical carcinomas still exhibit poor survival rates, 5 this underlines the necessity to study in more detail the molecular mechanisms underlying HPV-induced carcinogenesis to pave the way for the development of novel and more efficient treatment options.
The HPV E6 and E7 oncoproteins are the key drivers of the malignant growth of HPV-positive cancers. While E7 accounts for the deregulation of cell cycle control by binding and inactivating the tumor suppressor retinoblastoma protein (pRb), 6 E6 leads to proteolytic degradation of p53. 4,7 Consequently, p53-dependent apoptotic responses, such as the proapoptotic p53/PUMA/Bax axis, 8 are impaired, which might also result in increased resistance towards cancer therapies.
Increasing evidence indicates an important role for the Dickkopf-1 (Dkk1) protein for human carcinogenesis; however, whether Dkk1 acts as a tumor suppressor or oncogenic factor seems to be highly dependent on the cancer type. [9][10][11] Dkk1 is a secreted protein and was originally identified as a head inducer during embryonic development in Xenopus, which was mechanistically linked to its Wnt-inhibitory activity. 12 Active, canonical Wnt signaling results in β-catenin stabilization and its translocation to the nucleus, where it activates the transcription of Wnt responsive genes, including those of Dkk1. 13,14 As a negative feedback regulator, Dkk1 binds to the Wnt specific LRP5/6 co-receptor and thereby prohibits Wnt activation. 15 In cervical cancer cells, it was reported that DKK1 is epigenetically silenced 16 and that it acts as a proapoptotic tumor suppressor whichupon ectopic overexpression-can repress the tumorigenicity of HeLa cells. 17 Interestingly, the chemotherapeutic agent Cisplatin, which belongs to the standard care for advanced and recurrent cervical cancers, 18 was described to induce Dkk1 expression in glioma and head and neck cancer cells, 19,20 and in the latter, Dkk1 repression was reported to contribute to Cisplatin resistance. While the authors of these two studies hypothesized that the proapoptotic activity of Dkk1 may be due to its Wnt-suppressive role, evidence is emerging that Dkk1's ability to induce apoptosis might be also mediated via β-catenin-independent mechanisms, 17,21 which potentially includes the involvement of noncanonical Wnt pathways. 22 In the present study, we aimed to investigate whether there is a crosstalk of the HPV oncogenes with Dkk1 and to gain insights into the potential role of Dkk1 for the efficiency of Cisplatin-based therapy in cervical cancer cells. We show that Dkk1 expression levels are highly restrained by the endogenous expression of the HPV oncogenes via a p53-dependent mechanism, indicating that HPV E6 is a key driver in this regulation. In addition, our results reveal that Dkk1 plays a major role for mediating the proapoptotic response of cervical cancer cells towards Cisplatin and, accordingly, Dkk1 repression significantly increases Cisplatin resistance. Further, we provide evidence that this latter effect is mechanistically not linked to an interference with canonical Wnt signaling, but to the induction of c-Jun N-terminal kinase (JNK) signaling. Overall, our findings indicate that the sustained expression of HPV E6/E7 oncogenes in cervical cancer cells contributes to their apoptosis resistance by continuously downregulating expression of Dkk1. Moreover, Dkk1 is a major determinant for the sensitivity of cervical cancer cells towards Cisplatin by modulating JNK signaling.

| Plasmids and siRNAs
All plasmids and siRNA sequences are described in detail in the Appendix S1. For RNA interference (RNAi) analyses, siRNAs were transfected at a final concentration of 10

| TOPflash/FOPflash luciferase reporter assays
For measuring canonical Wnt signaling activities, cells were transfected in 60 mm dishes with 2 μg TOP-or FOPflash plasmids in addition to 0.5 μg pCMV-β-galactosidase serving as an internal standard. 25 When indicated in the text, 0.5 μg of pcDNA3 or pcDNA3-Wnt3a were co-transfected. Twenty-four hours after transfection, cells were treated with Ctrl CM, Dkk1 CM or LiCl. After 16 hours cells were lysed and luciferase, as well as β-galactosidase activities were measured as previously described. 25

| Protein and RNA analyses
RNA extraction, qRT-PCR, protein extraction and immunoblot analyses were performed as previously described, 23 and qRT-PCR primer and antibodies are listed in the Appendix S1. For protein analyses in cell supernatants, the cell culture medium was collected from transfected cells, centrifuged for 5 minutes at 1000g to remove cell debris and boiled for 5 minutes at 95 C in SDS sample buffer. The volume loaded to the gel was determined relative to the respective protein concentrations in the lysates of cells from which the medium was taken. Immunoblots were visualized using the enhanced luminescence (ECL) reagent WesternBright Sirius (Advansta, San Jose, California) in the Fusion SL Detection System (Vilber Lourmat, Germany). Gene expression determined by qRT-PCR was normalized to the internal standard 18S rRNA and the comparative Ct ( 2ÀΔΔCt ) method was applied for relative quantification of transcript expression. 26

| Colony formation and senescence assays
For colony formations assays (CFAs) and senescence assays (SAs), cells were split after Cisplatin treatment and cultivated for the indicated time periods in standard medium. CFAs were fixed and stained with a crystal violet/formaldehyde solution. SA-β-gal staining was performed as described previously 23 and images were acquired using a Â10 magnification on the EVOScl Core Cell Imaging System (Thermo Fisher Scientific).  Table S1.

| Statistical analyses
All experiments were performed at least thrice with consistent results and data for qRT-PCR, TUNEL quantifications and luciferase assays is

| HPV oncogene expression restrains Dkk1 levels in cervical cancer cells
To investigate whether Dkk1 may be a downstream target of the HPV E6/E7 oncogenes, Dkk1 expression levels were analyzed after siRNA-mediated silencing of E6 or E6/E7 in HPV18-positive HeLa and HPV16-positive SiHa and CaSki cervical cancer cells ( Figure 1A, B). In all these cells, Dkk1 protein expression was strongly increased upon E6 or E6/E7 silencing, indicating that the sustained HPV oncogene expression efficiently restricts endogenous Dkk1 protein levels.
E6 inhibition alone was sufficient to increase Dkk1 concentrations in

siDkk1 #1+#2
F I G U R E 1 Dkk1 expression is restrained by HPV E6. (A) HeLa and SiHa cells were transfected with control siRNA (siCtrl) or siRNAs silencing HPV18 or HPV16 E6 and E6/E7 (siE6; siE6/E7), respectively, either alone or concomitantly with siRNA silencing p53 (sip53) expression. Dkk1, p53 and HPV18 (HeLa) or HPV16 (SiHa) E6/E7 protein levels were analyzed by immunoblots. β-actin: loading control. (B) Transfection of HPV16 E6 and E6/E7-specific siRNAs and corresponding immunoblot analyses in CaSki cells. (C) DKK1 mRNA expression was determined by qRT-PCR and is indicated by mean log 2 FC including standard deviations, relative to siCtrl-transfected cells (log 2 FC = 0). Statistical significance is indicated by asterisks (*P < .05; **P < .01; ***P < .001; n.s., not significant). (D) Secreted Dkk1 in the cell supernatant of different HPV-positive cancer cells after E6 or E6/E7 silencing compared to the protein levels in the corresponding lysates. HPV E7 is not secreted and served as an internal control that supernatants do not contain intracellular content. (E) Immunoblot analyses of Dkk1 and HPV18 or HPV16 E6 and E7 protein levels in HeLa and SiHa cells, respectively, transfected with two different siRNAs silencing Dkk1 expression, applied either alone (siDkk1 #1; siDkk1 #2) or as pool (siDkk1 #1 + #2). Vinculin: loading control parallel with p53 levels ( Figure 1A,B), which is interesting as DKK1 was previously reported to be transcriptionally activated by p53. 29 In order to explore the potential regulatory role for p53 in Dkk1 induction upon HPV oncogene repression we combined E6 or E6/E7 silencing with siRNA-mediated repression of p53. Concomitant silencing of E6 or E6/E7 and p53 efficiently blocked the increase in Dkk1 levels ( Figure 1A), indicating that E6 profoundly restricts Dkk1 expression via interference with p53-mediated DKK1 transactivation. Accordingly, E6-mediated repression of DKK1 is also observed at the transcript level, where DKK1 mRNA concentrations are significantly increased after E6 or E6/E7 silencing, which again was abolished when p53 was repressed in parallel ( Figure 1C). Further, intracellular Dkk1 induction after knockdown of the HPV oncogenes was accompanied by a strong increase of Dkk1 levels in the supernatant of all tested cervical cancer cells ( Figure 1D), indicating that also the amount of secreted Dkk1 underlies regulation by HPV E6. Collectively, these findings define Dkk1 as a novel indirect downstream target for the HPV oncogenes, which is repressed via E6-mediated interference with p53. On the other hand, silencing of Dkk1 did not affect HPV E6/E7 protein expression in HeLa or SiHa cells ( Figure 1E), indicating that endogenous Dkk1 does not act as an upstream regulator of the HPV oncogenes.   Figure 4A,B). Dkk1 was expressed in the cell lysate and supernatant of pCS2-hDkk1-Flag-, but not in vector control (pCS2)-transfected HeLa* cells ( Figure 4B) which were used to generate control conditioned medium (Ctrl CM). Both conditioned media were transferred to HeLa cells ( Figure 4A) and the TOP-/FOPflash luciferase reporter system was employed to measure canonical Wnt signaling activity. 32 In line with a previous report, 17    Cisplatin treatment, which may be a potential mechanism behind the increase in apoptosis resistance.

| Cisplatin-induced apoptosis is impaired upon suppression of JNK activity
In order to test whether JNK activation is indeed a decisive factor for the Dkk1-dependent apoptosis induction under Cisplatin treatment, we blocked JNK activity in HeLa cells, either by transfection of a well-characterized JNK1/2-targeting siRNA 36 or by treatment with JNK-IN-8, a selective pan-inhibitor for JNK1, JNK2 and JNK3. 37 We concentrated on the analysis of the ubiquitously expressed JNK1 and JNK2 isoforms, since JNK3 expression is restricted to heart, brain and testis. 35 As expected, siJNK1/2 downregulated total and phosphory- CaSki and SiHa cells ( Figure 6C). This was accompanied by a significant decrease of apoptotic SiHa cells in TUNEL assays ( Figure 6D).
Likewise, the induction of cl. PARP and cl. Caspase 9 by Cisplatin was impaired upon siRNA-mediated JNK silencing also in CaSki and SiHa cells ( Figure S3B). Overall, these results show that JNK activity plays a key role for Cisplatin-mediated apoptosis induction in cervical cancer cells and downregulation of Dkk1 provides these cells with increased Cisplatin resistance by interfering with JNK activation.

| DISCUSSION
The present work reveals that the expression of the putative tumor suppressor protein Dkk1 is continuously restrained in cervical cancer cells by the sustained expression of the HPV oncogenes. Furthermore, we provide evidence that Dkk1 plays an important role for the proapoptotic response of cervical cancer cells towards Cisplatin treatment.
We found that silencing of endogenous viral E6 oncogene expression alone is sufficient to relieve Dkk1 from the negative regulation by the HPV oncogenes, leading to strongly enhanced Dkk1 expression levels in cervical cancer cells. Further, RNAi analyses reveal that the induction of p53 upon E6 silencing is required for the increase of Dkk1 levels, indicating that E6 represses Dkk1 expression through its interference with p53 function. In line with this notion, DKK1 was reported to contain a putative p53 binding site in its promoter and to be a direct transcriptional target gene for p53, as was assessed by ectopic overexpression of wildtype p53. 29 Consistent with this data we observed that the induction of Dkk1 expression upon E6 silencing is detectable not only at the protein, but also at the transcript level. In this context, it is important to note that only a small subset of the over 3500 postulated p53 target genes 38 is significantly affected by silencing the HPV oncogenes in cervical cancer cells. 39,40 This may, for example, be due to the fact that a significant portion of the postulated p53 target genes have been identified by p53 overexpression studies and genes with low affinity p53 binding sites could . Each data point represents the log 2 FC of a single gene in relation to its Àlog 10 -transformed P-value (Àlog 10 P). Dashed lines indicate cutoff values for log 2 FC = À1/+1 and P-value = .05 which divide the datasets into four groups: nonsignificant (ns-gray), log 2 FC ≥ j1j and P > .05 (green), log 2 FC < j1j and P < .05 (blue), log 2 FC ≥ j1j and P < .05 (red). Data points marked by their gene names from the latter group were selected based on the strongest differential regulation in between comparison 2vs1 and 4vs3 (Δlog 2 FC) to exclude genes which are equally regulated in both parental and Dkk1 KO cells. (C) Gene Set Enrichment Analysis (GSEA) 27   Whereas the underlying molecular mechanism remains to be elucidated, it is noteworthy that interference with JNK activity has been reported to possess potential to promote cellular senescence. 50 This finding is not only interesting under mechanistic aspects concerning the antiproliferative effects of Cisplatin in cervical cancer cells, but could also be relevant for the clinic. On the one hand, therapy-induced senescence aims to target cancer cells by stopping their proliferation and inducing their clearance by the immune system.
On the other hand, emerging evidence shows that senescence can also have nonbeneficial consequences, such as the secretion of protumorigenic factors or enhanced chemotherapy resistance. 52 Consequently, apoptosis has been proposed to be a more beneficial therapeutic outcome than senescence. 53  In conclusion, the present study provides new insights into the