Cancer Research
Therapeutics, Targets, and Chemical Biology
STAT3/IRF1 Pathway Activation Sensitizes Cervical Cancer Cells to Chemotherapeutic Drugs € ckheim1, Jennifer Pahne-Zeppenfeld2, Jil Fischbach1, Barbara Walch-Ru € ttner5, Claudia Wickenhauser3, Lars Christian Horn4, Lars Tharun5, Reinhard Bu 6 7 8 8 € be9, Peter Mallmann , Peter Stern , Yoo-Jin Kim , Rainer Maria Bohle , Christian Ru 10 10 10 € € Russalina Stroder , Ingolf Juhasz-Boss , Erich-Franz Solomayer , and Sigrun Smola1
Abstract Neoadjuvant radio/chemotherapy regimens can markedly improve cervical cancer outcome in a subset of patients, while other patients show poor responses, but may encounter severe adverse effects. Thus, there is a strong need for predictive biomarkers to improve clinical management of cervical cancer patients. STAT3 is considered as a critical antiapoptotic factor in various malignancies. We therefore investigated STAT3 activation during cervical carcinogenesis and its impact on the response of cervical cancer cells to chemotherapeutic drugs. Tyr705-phosphorylated STAT3 increased from lowgrade cervical intraepithelial neoplasia (CIN1) to precancerous CIN3 lesions. Notably, pTyr705-STAT3 activation significantly declined from CIN3 to invasive cancer, also when compared in the same clinical biopsy. pTyr705-STAT3 was also low or absent in cultured human cervical cancer cell lines, consistent with the in vivo expression data. Unexpectedly,
IL6-type cytokine signaling inducing STAT3 activation rendered cervical cancer cells significantly more susceptible to chemotherapeutic drugs, that is, cisplatin or etoposide. This chemosensitization was STAT3-dependent and we identified IFN regulatory factor-1 (IRF1) as the STAT3-inducible mediator required for cell death enhancement. In line with these data, pTyr705-STAT3 significantly correlated with nuclear IRF1 expression in cervical cancer in vivo. Importantly, high IRF1 expression in pretreatment cervical cancer biopsy cells was associated with a significantly better response to neoadjuvant radio/chemotherapy of the patients. In summary, our study has identified a key role of the STAT3/IRF1 pathway for chemosensitization in cervical cancer. Our results suggest that pretherapeutic IRF1 expression should be evaluated as a novel predictive biomarker for neoadjuvant radio/chemotherapy responses. Cancer Res; 76(13); 3872–83. 2016 AACR.
Introduction
pithelial neoplasia (CIN1–3; ref. 1). This process takes years or decades and it is assumed that further changes within the (pre) neoplastic cells and their microenvironment critically influence the course of disease. Cervical cancer therapy is still a major clinical challenge. Responses to neoadjuvant radio/chemotherapy vary greatly in patients (2, 3). Intrinsic and acquired resistance of the neoplastic cells as well as substantial side effects from standard treatment including platinum-based chemotherapy limit the options for escalation (4). Identification of patients that may best benefit from chemotherapy would be useful for improved clinical management. This will require a better understanding of the mechanisms influencing the balance between sensitivity and resistance to cervical cancer cell death. The STAT3 transcription factor is commonly considered as a survival or progression factor in different cancer types. Constitutive STAT3 activation is documented in various human malignancies including head and neck, brain, breast, lung, pancreas, as well as prostate cancer and melanoma (summarized in ref. 5). STAT3 inhibition can affect tumor growth and enhance the response of certain tumors to chemotherapy directly or in an immune-dependent manner (6–8). Notably, STAT3 activation also has profound direct effects on the immune microenvironment (9). Our group has previously demonstrated pronounced tyrosine-phosphorylation of STAT3 in cervical high-grade lesions (10). Strongly activated STAT3 was detected within the inflammatory infiltrate of the lesions,
Cervical cancer represents the third most common cause of cancer-related death in women worldwide. Invasive cancer develops from persistent high-risk human papillomavirus (HPV) infection through well-defined stages of cervical intrae-
1
Institute of Virology, Saarland University, Homburg/Saar, Germany. Center for Molecular Medicine Cologne and Institute of Virology, University of Cologne, Cologne, Germany. 3Institute of Pathology, University of Halle, Halle, Germany. 4Institute of Pathology, University of Leipzig, Leipzig, Germany. 5Institute of Pathology, University of Cologne, Cologne, Germany. 6Department of Gynecology and Obstetrics, University of Cologne, Cologne, Germany. 7Institute of Cancer Sciences, University of Manchester, Manchester, United Kingdom. 8 Institute of Pathology, Saarland University, Homburg/Saar, Germany. 9 Department of Radiotherapy and Radiation Oncology, Saarland University, Homburg/Saar, Germany. 10Department of Gynecology and Obstetrics, Saarland University, Homburg/Saar, Germany. 2
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). €ckheim and J. Pahne-Zeppenfeld share first authorship of this B. Walch-Ru article. Corresponding Author: Sigrun Smola, Institute of Virology, Saarland University, Kirrbergerstrasse, Building 47, Homburg/Saar D-66421, Germany. Phone: 49– 6841–16–23931; Fax: 49–6841–16–23980; E-mail:
[email protected] doi: 10.1158/0008-5472.CAN-14-1306 2016 American Association for Cancer Research.
3872 Cancer Res; 76(13) July 1, 2016
STAT3/IRF1 Sensitizes Cervical Cancer to Chemotherapy
where it drives expression of the protumorigenic and vasculogenic matrix metalloprotease MMP-9, as well as in the premalignant epithelial cells (10). Major activators of the STAT3 pathway are members of the IL6-type cytokine family (11, 12). IL6 binds to the a-chain of the IL6R (IL6Ra, gp80), while the related cytokine oncostatin M (OSM), binds to the OSM receptor-b (OSMRb). Respective complexes then associate with the common receptor chain gp130, which recruits Janus kinases, leading to subsequent STAT3-phosphorylation at tyrosine 705 (13). Whereas gp130 is ubiquitously expressed, IL6 signaling is limited by the availability of transmembrane gp80 or its naturally occurring soluble form sgp80, which induces "trans-signaling" via gp130 (14). In human squamous cell carcinoma (SCC) of the cervix uteri, IL6 is strongly upregulated in situ and in vitro (15). IL6 has a negative impact on the prognosis of a patient (16). We have shown that it mainly acts in a paracrine manner and creates a protumorigenic and immunosuppressive microenvironment (10, 17, 18). Cervical cancer cells display only low responses to autocrine IL6 due to low gp80 expression levels (15, 19) but sgp80 can restore autocrine IL6 signaling and induce pronounced STAT3-binding activity (15). In this study, we investigated STAT3 activation during the different stages of cervical carcinogenesis in more detail. In biopsies comprising precancerous and cancerous lesions, we demonstrate that STAT3 tyrosine-phosphorylation is highest in CIN3 and significantly declines during progression to invasive cancer. Unexpectedly, when we forced STAT3 activation with IL6-type cytokines, cervical cancer cells were strongly sensitized to the cytotoxic effects of chemotherapeutic drugs. We identified IRF1 as the STAT3-inducible proapoptotic factor mediating chemosensitization. Notably, our data show that pretreatment IRF1 expression correlates with the response to radio/chemotherapy in cervical cancer patients in vivo.
Materials and Methods Immunohistochemical analysis Of note, 145 formalin-fixed paraffin-embedded (FFPE) anonymized lesions of the cervix uteri (22 CIN1/2, 29 CIN3, 94 SCC) were retrieved from local pathology archives of Cologne, Leipzig, and Saarland University Hospitals, Germany. These included a subset of 24 pretreatment SCC biopsies from patients that had been subjected to neoadjuvant chemo- or radio/chemotherapy. These tumors were pre- and post-therapeutically staged according to the International Federation of Gynecology and Obstetrics (FIGO) or TNM categories (Supplementary Table S1). Histologic stainings, diagnosis, and treatment responses were assessed by expert pathologists (C. Wickenhauser, L.C. Horn, L. Tharun, Y.-J. Kim, R.M. Bohle). Staining of 5-mm sections with Abs listed in Supplementary Table S2 was performed as described previously (10, 20) and classified using the Immunoreactive Score (IRS) according to Remmele and Stegner (21). Biopsies were evaluated with standardized settings with a DMI 6000B microscope (Leica) and Microsoft Image Composite Editor program. The retrospective study has been conducted according to Declaration of Helsinki principles and was approved by the local Ethics Committees of the Cologne, Leipzig, and Saarland Universities (at the Saar€ land-Arztekammer).
www.aacrjournals.org
Cells and cell culture Relatively low-passage cervical cancer cell lines 808 and 778 (22, 23) were last tested by short tandem repeat profiling in 2014. HPV18-positive cervical carcinoma cell lines SW756 (ATCC CRL10302), HeLa (ATCC CCL-2), and HPV16-positive SiHa (ATCC HTB-35) obtained from M. von Knebel Doeberitz (Department of Applied Tumor Biology, Institute of Pathology, University of Heidelberg, Heidelberg, Germany) before 2000 were last authenticated by HPV16/18-E6/E7-qRT-PCR and multiplex human cell line authentication test (Multiplexion) in April 2013 and cultured as described previously (17). Normal human exocervical keratinocytes (NECK) isolated from hysterectomy specimens according to ref. 24 were cultured in supplemented KBM-Gold (Lonza; approved by the local Ethics Committee of the Saarland Univer€ sity at the Saarland-Arztekammer). Written informed consent was provided by the study participants. Cell stimulation and Western blot analysis Carcinoma cells or NECK were seeded at a density of 1.5 106 cells in a 6-cm culture dish. Twenty-four hours later, they were incubated with medium, 10 ng/mL OSM, or 100 ng/mL IL6 (PeproTech) in the presence of 500 ng/mL sgp80 (R&D Systems; IL6/sgp80) for the indicated time intervals. Whole cell or nuclear extracts were prepared as described in ref. 25. Abs listed in Supplementary Table S2, secondary Abs (Sigma-Aldrich), and ECL reagent (Roche) were used for detection with ChemiDoc XRSþ Molecular Imager. All Western blots were performed under standardized conditions. Expression was quantified with the Quantity One analysis software (both Bio-Rad). Plasmids and transfections IRF1 and IRF2 cDNAs (26) were amplified with primers listed in Supplementary Table S3, cloned as NotI/SalI fragments in pCMV-Flag2 vector (Sigma-Aldrich) and sequences were verified. A total of 1.5 105 HeLa cells/6-well were transfected after 24 hours with 0.0 (mock control), 0.025 or 0.1 mg pCMV-Flag2-IRF1, or 0.6 or 0.8 mg pCMV-Flag2-IRF2 expression vector using Lipofectamine 2000 (Invitrogen). The total amount of DNA was adjusted to 1 mg with empty pCMV-Flag2 vector. Six or 10 pmol of indicated siRNAs (ON-TARGETplus Non-targeting siRNA #2, ON-TARGETplus siRNA #8 for human STAT3, siRNA #6 for IRF1, all from Thermo Scientific) were transfected with Lipofectamine RNAiMax (Invitrogen) as described previously (20). Cytotoxicity assays In cytotoxicity assays with cisplatin (Hexal) or etoposide (Sigma-Aldrich) cells were seeded in 96-well plates at a density of 1.0104 cells/well, stimulated with medium, OSM, or IL6/sgp80 for 8 hours and challenged for 48 hours with serial dilutions of chemotherapeutic drugs if not indicated otherwise. siRNA-transfected cells were pretreated for 2 hours and subsequently challenged with chemotherapeutic drugs for 20 hours. Cell viability was assessed by the neutral red uptake method as described previously (27). For sequential staining with Annexin-V–APC (BD Biosciences) and propidium iodide (PI; Sigma-Aldrich) assays were scaled up to 24-well plates and analyzed by flow cytometry (FACSCalibur, BD Biosciences). Quantitative real-time RT-PCR HeLa and SW756 cells were stimulated for 2 hours if not indicated otherwise. cDNA synthesis, qRT-PCR, and normalization
Cancer Res; 76(13) July 1, 2016
3873
€ckheim et al. Walch-Ru
Results
and 6 of 94 (6.4%) showed a moderate IRS (Fig. 1C). Positive pTyr705-STAT3 staining in SCCs was preferentially detected at the tumor borders adjacent to stroma. The pTyr705-STAT3 IRS did not correlate with FIGO stage (Supplementary Fig. S1). In negative SCCs, proper staining was ascertained by positively stained stromal infiltrating or endothelial cells. Notably, epithelial pTyr705-STAT3 staining was significantly weaker in SCCs than CIN3 (Fig. 1B and C; P < 0.0001) and this was also observed, when both SCC and CIN3 were present in the same biopsy (Fig. 1D). These results demonstrated that STAT3 is activated in human cervical epithelium during carcinogenesis peaking in CIN3. However, during progression from CIN3 to invasive cancer, STAT3 activation significantly declines or is lost.
Decline of STAT3 tyrosine-phosphorylation during malignant progression in human cervical carcinogenesis in situ In tissue sections from the different stages of human cervical carcinogenesis, we found activated STAT3 throughout all premalignant stages, however with different intensities. In lowgrade CIN, staining was mostly confined to the cells of the suprabasal layers (Fig. 1A). In contrast, in CIN3 pTyr705-STAT3 staining was detected in all epithelial layers and was strongest when a stromal inflammatory STAT3-positive infiltrate was present (Fig. 1A; ref. 10). The pTyr705-STAT3 staining score was significantly higher in CIN3 than in CIN1 or CIN2 (Fig. 1A and C; P < 0.0001), also in biopsies where both, low- and highgrade CIN, were present (Fig. 1D). Unexpectedly, 49 of 94 (52.1%) SCCs displayed only weak pTyr705-STAT3 staining, 39 of 94 (41.5%) were rated negative,
OSM and IL6/sgp80 signaling activate STAT3 in cervical carcinoma cells The same pTyr705-STAT3–specific Ab was used to analyze STAT3 phosphorylation in the established cervical SCC cell lines SW756 and SiHa, the adenocarcinoma cell line HeLa, as well as in the more recently generated cervical cancer cell lines 808 and 778 (22, 23). In medium controls, only faint constitutive STAT3 activation was observed. Stimulation with OSM or IL6/sgp80, led to a pronounced STAT3 phosphorylation in cervical cancer cells (Fig. 2), which was generally stronger than in NECK as shown in prolonged exposures of the Western blots (Supplementary Fig. S2). Thus, constitutive cell-autonomous STAT3 phosphorylation in cervical cancer cells is low or absent but strongly inducible by IL6-type cytokine signaling.
to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were performed as described previously (17, 25, 28). The 72-bp fragment of IRF1 was detected with primers listed in Supplementary Table S3 and probe no. 36 (Roche Universal Probe Library; Roche). Statistical analysis To evaluate the statistical differences between analyzed groups, two-sided t test was applied. Significances are indicated by asterisks ( , P < 0.05; , P < 0.01; , P < 0.001). Correlations of pTyr705-STAT3 IRS with IRF1 IRS or with FIGO stages of SCCs were investigated using Spearman rank correlation.
Figure 1. pTyr705-STAT3 expression in CIN1-3 and cervical SCCs. Human FFPE-sections were stained with anti-pTyr705-STAT3 Ab (brown). A, biopsy containing CIN1, CIN2, and CIN3. B, biopsy containing CIN3 and SCC (all 200). Bars, 100 mm. Asterisks in B indicate SCC. C, IRS of pTyr705-STAT3 staining in CIN1/2 (n ¼ 22), CIN3 (n ¼ 29), and SCC (n ¼ 94). Mean, blue line. D, IRS of pTyr705-STAT3 staining in a subgroup of biopsies containing two stages of cervical carcinogenesis (linked by lines), CIN1/2 and CIN3 (n ¼ 8), or CIN3 and SCC (n ¼ 19); statistical analysis: two-sided t test. , P < 0.01; , P < 0.001.
3874 Cancer Res; 76(13) July 1, 2016
Cancer Research
STAT3/IRF1 Sensitizes Cervical Cancer to Chemotherapy
Figure 2. OSM or IL6/sgp80 strongly activate STAT3 in cervical cancer cells. HeLa, SW756, SiHa, 808, 788 cells, and NECK were stimulated with medium, OSM, or IL6/sgp80 for 15 minutes. Whole cell extracts were analyzed by Western blot analysis using anti-pTyr705-STAT3-, STAT3-, or b-actin–specific Ab.
OSM and IL6/sgp80 signaling sensitize cervical cancer cells to chemotherapeutic drug–induced cell death As STAT3 protects various cell types from cell death (6, 29), we were interested whether OSM or IL6 trans-signaling influenced cell death induction by chemotherapeutic drugs in cervical cancer cells. Initially, HeLa cells were prestimulated with OSM for different time intervals, which did not significantly change cellular viability, and subsequently treated with cisplatin, the most common drug used in cervical cancer chemotherapy. Unexpectedly, OSM preactivation did not prevent but enhanced cisplatinmediated cell death in HeLa cells as judged by combined Annexin-V/PI flow cytometry (Fig. 3A). Two hours of OSM pretreatment were sufficient for significant sensitization to cisplatininduced cell death (P < 0.0001), which was further increased by 15% (P ¼ 0.001) after 8 hours of pretreatment (Fig. 3B). We therefore used an 8-hour cytokine pretreatment schedule for our series of cervical cancer cell lines. Stimulation with OSM or IL6/sgp80 alone neither altered cellular viability (Fig. 3 and 4) nor proliferation (Supplementary Fig. S3) under any condition tested. Cisplatin killed the individual cell lines at different concentrations. Notably, as a uniform response in all tested cancer cell lines, we observed a significant increase in cisplatin-induced cell death after pretreatment with IL6/sgp80 (26%–54% increase for the highest cisplatin dose, P values from 0.0002 to 0.0079) or OSM (19%–52% increase, P values 0.0003–0.0038; Fig. 4A). Similar observations were made for etoposide, another chemotherapeutic drug in clinical use for cervical cancer treatment (30). Etoposideinduced cell death increased by 64% or 55% in HeLa (P values 0.0001 or 0.003) and 72% or 69% (P values 0.0001 or 0.0004) in SW756 after pretreatment with OSM or IL6/sgp80, respectively (Fig. 4B). In contrast, NECK were not significantly sensitized (P > 0.3917) to either of the chemotherapeutic drugs by IL6/ sgp80 or OSM (Fig. 4A and B). Cervical cancer cells produce IL6 but their autocrine IL6 response is limited due to low expression of the receptor gp80 (15, 19). To restore autocrine signaling, HeLa and SW756 cells were pretreated with sgp80 alone without exogenous IL6. Nota-
www.aacrjournals.org
bly, sgp80 was sufficient to strongly sensitize both cell lines to cisplatin- (Fig. 4C; P ¼ 0.021 or 0.0015) or etoposide-induced cell death (Fig. 4D; P < 0.0001). These data demonstrated that reconstitution of autocrine IL6 signaling, IL6 trans-signaling, or OSM can sensitize cervical cancer cells but not normal human exocervical keratinocytes for chemotherapeutic drugs. STAT3 mediates sensitization for cell death by OSM and IL6/ sgp80 signaling in cervical cancer cells To investigate the involvement of STAT3 in chemosensitization in cervical cancer cells, STAT3-specific siRNA was used to knockdown STAT3 expression in HeLa (Fig. 5A, C, and D) and SW756 cells (Fig. 5B, E, and F). In both cell lines, STAT3 knockdown significantly reverted OSM- or IL6/sgp80-mediated sensitization to cell death induced by cisplatin (reversion up to 97%; P > 0.0021; Fig. 5C and E) or etoposide (reversion up to 96%, P > 0.0012; Fig. 5D and F), while control siRNA did not. Similar results were obtained with an independent STAT3 siRNA (Supplementary Fig. S4). To further substantiate this finding, HeLa cells were transiently transfected with a dominant-negative version of STAT3 interfering with phosphorylation at Tyr705 (dnSTAT3-Y705F, STAT3F), treated with OSM, and subsequently challenged with etoposide. Transfected cells were visualized by EGFP coexpression. STAT3F overexpression completely restored cellular viability further underlining that STAT3 activation is required for cell death sensitization (Supplementary Fig. S5). These data provided evidence that chemosensitization by IL6-type cytokines depends on the STAT3 pathway in cervical cancer cells. OSM and IL6/sgp80 signaling in cervical cancer cells strongly induce IRF1 in a STAT3-dependent manner To further elucidate the molecular mechanism underlying cell death sensitization, we performed mRNA gene expression analysis in SW756 cells after OSM or IL6/sgp80 stimulation. Strong
Cancer Res; 76(13) July 1, 2016
3875
€ckheim et al. Walch-Ru
Figure 3. OSM sensitizes HeLa cells for cisplatin-induced cell death in a time-dependent manner. A, HeLa cells were prestimulated with medium or OSM for 8 hours and treated with 25 mg/mL cisplatin for further 12 hours. Cells were stained with PI and Annexin-V and analyzed by flow cytometry. Left, one experiment out of n ¼ 3; right, three experiments. B, HeLa cells were prestimulated with medium (squares, black line) or OSM (circles, gray line) for 2, 4, or 8 hours and then treated with medium (open symbols) or 1.56 mg/mL cisplatin (filled symbols) for 48 hours. Cell viability was assessed by the neutral red uptake method, n ¼ 2, performed in triplicates; statistical analysis: two-sided t test. , P < 0.05; , P < 0.001. n.s., nonsignificant.
upregulation of CCL2 confirmed our previously published results (15) and validated the assay (data not shown). Besides CCL2, the transcription factor IRF1, which has proapoptotic activities (31) was significantly upregulated after IL6/sgp80 or OSM stimulation. qRT-PCR revealed a rapid (within 2 hours) and strong (10- to 12fold) induction of IRF1 mRNA after OSM or IL6/sgp80 stimulation in HeLa and SW756 cells (Fig. 6A; P
