KLF13 Regulates the Differentiation-Dependent Human Papillomavirus Life Cycle in Keratinocytes Through STAT5 and IL-8

ORIGINAL ARTICLE KLF13 regulates the differentiation-dependent human papillomavirus life cycle in keratinocytes through STAT5 and IL-8

W Zhang1,6, S Hong1,6, KP Maniar2, S Cheng1, C Jie3, AW Rademaker4, AM Krensky1,5 and C Clayberger1

High-risk strains of human papillomavirus (HPV) are the causative agents of cervical and anogenital cancers and are associated with 5% of all human cancers. Although prophylactic vaccines targeting a subset of HPV types are available, they are ineffective in HPV- infected individuals. Elucidation of the mechanisms controlling HPV replication may allow development of novel anti-HPV therapeutics. Infectious HPV virions are produced during terminal differentiation of host cells. The process of viral maturation requires synergistic interactions between viral and cellular proteins that leads to amplification of the viral genome and expression of late viral genes. Here we show that the transcription factor Kruppel-like factor 13 (KLF13) has a critical role in the HPV life cycle. KLF13 is overexpressed in HPV-positive keratinocytes and cervical cancer cell lines. Expression of KLF13 in normal cervical epithelium is low but increases significantly in cervical intraepithelial neoplasia and invasive squamous cervical cancer. After HPV infection, the E7 protein suppresses ubiquitin ligase FBW7 expression leading to an increase in KLF13 expression. Reduction of KLF13 with short hairpin RNA in differentiating HPV-positive cells resulted in diminished levels of viral gene expression and genome amplification. Knockdown of KLF13 also reduced the level of the transcription factor signal transducer and activator of transcription 5, which led to the downregulation of the ataxia-telangiectasia mutated DNA damage pathway and the chemokine interleukin-8 (IL-8). In addition, neutralization of IL-8 diminished viral genome amplification in differentiating HPV-positive cells. Thus, KLF13 is critical for the activation of the HPV productive life cycle and is likely involved in initiation and progression of cervical cancer.

Oncogene (2016) 35, 5565–5575; doi:10.1038/onc.2016.97; published online 4 April 2016

INTRODUCTION factors, including NF1, AP-1, OCT1, C/EBPβ, SP1, YY1 and KRF-1, Cervical cancer is the third most common cancer in women contribute either positively or negatively to the regulation of HPV worldwide and the seventh most common cancer overall.1 Human activity by directly binding to the viral genome in the upstream 10,11 papillomavirus (HPV) DNA is detected in 490% of cervical cancer regulatory region enhancer. In addition, viral proteins cases2 and in approximately 5% of all human cancers.3 HPVs are participate in numerous protein–protein interactions with cellular small DNA viruses that show a tropism for squamous epithelium. proteins through which they have important roles in modulating The Food and Drug Administration approved two prophylactic the host cells. Such protein–protein interactions most likely create HPV vaccines in 2006, but they do not protect against all HPV a cellular environment that is conducive to successive stages of subtypes that can lead to cervical cancer.4 Moreover, they are the viral life cycle. Despite our increased understanding of the HPV ineffective in HPV-infected individuals. Thus, there is an urgent life cycle, there is still a clear need to elucidate how HPV and host need to identify molecular mechanisms controlling HPV viral cell proteins determine the ultimate outcome of HPV infection. replication in order to rationally design and develop new agents The Kruppel-like family of transcription factors contain three with novel mechanisms of action against HPV. C2H2 zinc fingers at the C-terminus and bind to either a C(A/T)CCC HPV infects cells in the basal layer of stratified squamous element or GC-box consensus sequence, one of the most common epithelia. Following initial infection, the viral genome is replicated DNA elements in the promoters of numerous tissue-specific, as in conjunction with host DNA during S phase and maintained as well as housekeeping genes.12,13 Kruppel-like factors (KLFs) are stable episomes in infected undifferentiated basal cells.5–8 HPV involved in diverse aspects of cellular growth, differentiation and viral maturation requires a subset of host proteins that results in early embryonic development. Increasing evidence supports their high-level amplification of the viral genome and expression of the role and deregulation in a variety of human cancers.14 KLF13 was late viral genes. HPV viral gene expression is controlled through originally identified as an essential transcription factor controlling two major promoters and an enhancer located within the the late expression of chemokine CCL5 in human T lymphocytes.15 upstream regulatory region that contains multiple host cell In numerous studies, tumor-derived CCL5 correlates with the transcription factor binding sites.9 A number of transcription progression of many human cancers, including cervical cancer.16

1Department of Microbiology-Immunology, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA; 2Department of Pathology, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA; 3Department of Research Biostatistics, Des Moines University, Des Moines, IA, USA; 4Department of Preventive Medicine and the Robert H Lurie Comprehensive Cancer Center, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA and 5Department of Pediatrics, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA. Correspondence: Dr C Clayberger, Department of Microbiology-Immunology, Feinberg School of Medicine, Northwestern University, 300 East Superior Street, Chicago, IL 60611, USA. E-mail: [email protected] 6These authors contributed equally to this work. Received 28 July 2015; revised 4 January 2016; accepted 8 January 2016; published online 4 April 2016 KLF13 positively regulates HPV life cycle W Zhang et al 5566 In addition, KLF13 is overexpressed in oral cancer and can serve as HFK (HPV16), HPV18 stably infected HFK (HPV18), HPV31 stably an oncogene to positively regulate proliferation and survival of infected HFK (HPV31), HPV31-positive CIN612 cells21 and the cancer cells.17 Inhibition of KLF13 expression by small interfering HPV18-positive cervical cancer cell line HeLa. KLF13 protein is RNA (siRNA) in oral cancer cells resulted in a significant reduction increased in HPV types 16, 18 and 31 infected keratinocytes and in of cancer cell proliferation.17 HeLa cells (Figures 1a and b). As induction of late viral processes, Here, we show that KLF13 is upregulated in HPV-positive including genome amplification and activation of late viral genes, keratinocytes and cervical cancer cell lines and is frequently is dependent upon keratinocyte differentiation, we measured overexpressed in human cervical intraepithelial dysplasia and KLF13 in HPV31-positive and -negative keratinocytes before and squamous cell carcinoma. Increased expression of KLF13 results after differentiation in high calcium medium. The levels of SP1, the from a decrease in ubiquitin ligase FBW7-mediated ubiquitination of prototype of the KLF family, along with KLF5 were also examined KLF13 and its subsequent proteasomal degradation.18 Knockdown of because previous studies reported that these two transcription KLF13 in HPV-positive cells significantly reduces viral episomes, as factors may also have roles in cervical cancer.22,23 Keratinocyte well as decreases expression of viral genes. Reduction of KLF13 also growth and differentiation is associated with increased expression impairs the phosphorylation of signal transducer and activator of of epidermal differentiation markers such as involucrin and transcription 5 (STAT5) and the activation of the ataxia-telangiectasia keratin-10 (K10) (Figure 1c). Levels of SP1 and KLF5 were higher mutated (ATM) DNA damage pathway, an important regulator of to some degree in HPV-positive cells and these levels were HPV life cycle.19,20 In addition, we demonstrate that the proin- relatively stable upon differentiation (Figure 1c). Higher levels of flammatory chemokine interleukin-8 (IL-8) is upregulated in KLF5 were observed in HPV31-infected cells (HPV31 and CIN612) differentiation-dependent HPV amplification via KLF13 and STAT5. than in uninfected HFK (Supplementary Figure 1). Interestingly, Neutralization of IL-8 reduces viral genome amplification in KLF13 decreased during calcium-induced differentiation in all differentiated HPV-positive cells. Thus, KLF13 overexpression is an three cell types, but higher levels of KLF13 were detected in early event associated with HPV infection, and increased KLF13 levels HPV31-infected cells (HPV31 and CIN612) than in uninfected HFK positively regulate HPV viral gene expression and differentiation- (Figures 1c and d). dependent genome amplification. We next used immunohistochemistry to examine KLF13 expression in a series of human cervical sections to determine RESULTS whether aberrant upregulation of KLF13 in HPV-positive keratinocyte cell lines correlates with human cervical cancer initiation and KLF13 is upregulated in HPV-infected keratinocytes progression. All cases demonstrated morphologic features con- We first measured the level of KLF13 protein in normal human sistent with HPV-related intraepithelial neoplasia or HPV-related foreskin keratinocytes (HFK), as well as in HPV16 stably infected invasive carcinoma. In normal tissues, KLF13 expression is largely

Figure 1. KLF13 is increased in HPV-positive keratinocytes. (a) Western blot of KLF13 levels in HFK, HPV16-infected HFK (HPV16), HPV18-infected HFK (HPV18), HPV31-infected HFK (HPV31), HPV31-positive CIN612 cells and cervical cancer cell line HeLa. (b) The levels of KLF13 protein in the experiments represented in panel a were determined by ImageJ software (National Institutes of Health, Bethesda, MD, USA). The results are expressed as a relative ratio of KLF13 to GAPDH and are normalized to relative KLF13 expression in HFK. Quantitated data were from three experiments ± s.e.m. *Po0.05 vs HFK (unpaired t-test). (c) Western blot of SP1, KLF5 and KLF13 in HFK, HPV31-infected HFK and CIN612 cells differentiated in high calcium media for the indicated times. All results are representative of three similar experiments. (d)Quantification of KLF13 levels shown in panel c. The results are expressed as a relative ratio of KLF13 to GAPDH and are normalized to relative KLF13 expression in HFK at 0 h. Quantitated data are represented as mean ± s.e.m. of three experiments. Black bars, HFK; dark gray bars, HPV31; light gray bars, CIN612. The data were analyzed using a two-way analysis of variance. The KLF13 protein levels in HPV31 are significantly different from that of HFK across time (Po0.005), and the KLF13 protein levels in CIN612 are also significantly different from that of HFK across time (Po0.005).

Oncogene (2016) 5565 – 5575 © 2016 Macmillan Publishers Limited, part of Springer Nature. KLF13 positively regulates HPV life cycle W Zhang et al 5567 restricted to the nuclei of cells in the basal layer of the stratified ubiquitination-mediated protein degradation and microRNA.18,24,25 squamous epithelium, and KLF13 levels decrease as the cells To examine whether proteasome-mediated protein degradation is become terminally differentiated keratinocytes (Figure 2a). These involved in regulating levels of KLF13, HFK and CIN612 were observations are consistent with decreased KLF13 protein levels treated with dimethylsulfoxide or the proteasome inhibitor observed after calcium-induced in vitro keratinocyte differentia- MG-132 for 3 h. MG-132 treatment led to an approximate twofold tion (Figure 1c). Expression of KLF13 is significantly increased in increase in KLF13 protein levels in HFK cells but not in CIN612 cells areas of abnormal versus adjacent areas of normal cervix in (Figure 3b). In human T lymphocytes, KLF13 protein expression is sections from 22 patients (Figure 2b). When abnormal cervical regulated by the ubiquitin E3 ligase FBW7.18 The gene-encoding samples are re-classified as low-grade squamous intraepithelial FBW7 is located on chromosome 4 and encodes three transcripts lesion (LSIL/CIN1), high-grade squamous intraepithelial lesion (FBW7α, β and γ) that result from alternative splicing of their first (HSIL/CIN2-3) and invasive squamous cell carcinoma, there is a exons.26 The mRNA levels of FBW7α, FBW7β and FBW7γ were significant difference among these three diagnostic groups reduced in CIN612 cells compared with HFK (Figure 3c), consistent (Figure 2c), with greater mean expression of KLF13 in true with higher KLF13 protein expression in HPV-infected CIN612 cells. preneoplasia (HSIL) and invasive carcinoma compared with MiR-125a and miR-125b were previously shown to directly repress productive infection (LSIL). Together, these findings demonstrate KLF13 expression in human T cells and hematopoietic stem fi 24,25 that KLF13 protein is signi cantly increased in HPV-positive cells. To determine a potential role of microRNA in regulating keratinocytes both in vitro and in vivo, and this increase is KLF13 in keratinocytes, miR-125a and miR-125b were measured associated with the initiation and progression of cervical cancer. from total RNA using a standard TaqMan real-time PCR assay. Nucleolar RNU48 was used for miRNA normalization.27 No FBW7 regulates KLF13 expression in human keratinocytes difference in expression of either of these microRNAs was found Having established that HPV infection results in increased in HFK compared with CIN612 cells (Supplementary Figure 2), expression of KLF13 in keratinocytes, we sought to determine suggesting that these two microRNAs do not have a major role in the mechanism underlying this change. We first examined regulating levels of KLF13 in keratinocytes. KLF13 mRNA levels in HPV31-positive cells compared with HPV oncoproteins E6 and E7 have important roles in cell uninfected HFK. Reverse transcriptase–PCR (RT–PCR) revealed that transformation and immortalization.6 To determine whether E6 or KLF13 mRNA levels are similar in uninfected HFK, HPV31 stably E7 may be responsible for increased expression of KLF13 in infected HFK and CIN612 cells (Figure 3a). Along with others, keratinocytes, primary HFKs were infected with the retroviral we previously showed that the steady-state level of KLF13 vector pLXSN alone, retroviruses expressing HPV31 E6, E7 or protein is tightly regulated by several mechanisms, including both,19,28 and KLF13 was assayed by western blot analysis. KLF13

Figure 2. KLF13 is upregulated in CIN1, CIN2, CIN3 and invasive squamous cell carcinoma (ISCC) of uterine cervix. (a) Hematoxylin and eosin (upper panel) and immunohistochemical staining of KLF13 (lower panel) in parafﬁn-embedded sections of cervical dysplasia and carcinomas as indicated. Scale bar = 50 μm for all panels. (b) Scatter dot plot representation of KLF13 immunohistochemical (IHC) score in paired normal versus abnormal (CIN1/2/3+SCC) cervix. Values represent the mean of immunohistochemical score ± s.e.m. KLF13 IHC scores were compared between normal and abnormal cervix using the paired t-test. (c) Scatter dot plot representation of KLF13 IHC score in LSIL, HSIL and ISCC. Values represent the mean of immunohistochemical score ± s.e.m. Abnormal cervical sample scores were compared across LSIL, HSIL and ISCC classiﬁcations using one-way analysis of variance.

Figure 3. FBW7 regulates KLF13 in human keratinocytes. (a)RT–PCR analysis of KLF13 mRNA levels in HFK, HPV31 and CIN612 cells. Values represent the mean fold change ± s.e.m. for three experiments relative to GAPDH. (b) Western blot analysis of KLF13 expression in HFK and CIN612 cells treated with dimethylsulfoxide (DMSO; vehicle control) or MG-132 (5 μg/ml) for 3 h. Relative intensity was determined by ImageJ software. (c) The mRNA level of each FBW7 isoform in HFK and CIN612 cells was determined by RT–PCR and normalized to endogenous GAPDH. Black bars, HFK cells; gray bars, CIN612 cells. *Po0.05 vs HFK (unpaired t-test). Quantitated data are represented as mean ± s.e.m. of three experiments. (d) HPV E7 protein is responsible for the increased KLF13 expression. HFK cells retrovirally transduced with either empty control vector or pLXSN encoding HPV31 E6, E7, or E6 and E7 together were stably selected with G418 and analyzed by western blot for KLF13. Relative intensity was determined by ImageJ software. (e)RT–PCR analysis of KLF13 mRNA levels in HFK, HFK(E6), HFK(E7) and HFK(E6/7) cells. Values represent the mean fold change ± s.e.m. for three experiments relative to GAPDH. (f) The mRNA level of FBW7γ was determined by RT–PCR and normalized to endogenous GAPDH. Quantitated data are represented as mean ± s.e.m. of three experiments (*Po0.05, unpaired t-test). (g) Western blot analysis of KLF13 expression in HFK transfected with FBW7-speciﬁc or NTC siRNAs. (h) The levels of KLF13 protein in the experiments represented in panel g were determined by ImageJ software. Quantitated data were from three experiments ± s.e.m. *Po0.05 vs NTC (unpaired t-test). (i) The graph represents the mean of quantitated KLF13 protein levels from three independent experiments. Cells were treated with 100 μg/ml cycloheximide (CHX) for the times points indicated before protein harvesting, and the protein extracted were analyzed for KLF13 protein levels. KLF13 levels were compared between HFK and HPV31 cells using the three-way analysis of variance, and a signiﬁcant difference was observed (P = 0.008).

expression was similar in HFK and HFK-E6 cells, but was elevated these findings indicate that KLF13 expression in HPV-infected HFK in cells expressing E7 alone or both E6 and E7 (Figure 3d), whereas cells results from inhibition of FBW7-mediated protein degrada- KLF13 mRNA levels were not altered (Figure 3e). Expression of E7 tion regulated by HPV E7 protein. also resulted in a significant reduction in FBW7γ mRNA (Figure 3f). To examine the direct effect of FBW7 on KLF13 protein, HFK cells were transfected with siRNA for 48 h and the whole-cell Loss of KLF13 reduces viral gene expression and genome fi lysates were then analyzed by immunoblotting. Knockdown of ampli cation in HPV-infected cells FBW7 by siRNA increased the level of KLF13 protein in HFK cells We transduced CIN612 cells with lentivirus-expressing KLF13- (Figures 3g and h). To further determine whether KLF13 levels specific short hairpin RNAs (shRNAs) to determine whether KLF13 are affected by protein degradation, a cycloheximide chase affects the differentiation-dependent HPV life cycle. Two KLF13- experiment was carried out in HFK and paired HPV31 cells. As specific shRNAs that can significantly decrease KLF13 protein shown in Figure 3i, KLF13 protein is more stable in HPV31 cells levels were identified in CIN612 cells (Supplementary Figure 3). In compared with HFK cells from the same individual. Collectively, order to achieve optimal knockdown efficiency, we pooled these

Oncogene (2016) 5565 – 5575 © 2016 Macmillan Publishers Limited, part of Springer Nature. KLF13 positively regulates HPV life cycle W Zhang et al 5569 two lentiviruses to knockdown KLF13. HPV31-positive CIN612 cells subjected to Southern blot analysis. HPV31 viral amplification was were infected with non-target control (NTC) or KLF13-specific reduced in CIN612 cells lacking KLF13, and this reduction shRNA-expressing lentiviruses, cultured for 48 h, and then persisted after calcium-induced differentiation (Figure 4b). To test transferred to high calcium medium to initiate differentiation whether KLF13 affects genome amplification upon differentiation, (0 h). Cells were harvested at 0, 48 and 96 h and lysates were we calculated the ratio of episome copy number at T48 divided by prepared for western and Southern analyses. Knockdown of KLF13 that at T0. As shown in Figure 4c, loss of KLF13 resulted in a ratio was 490% at all three timepoints (Figure 4a). Moreover, o1, indicating that KLF13 has a role in regulating HPV31 genome expression of the differentiation marker involucrin was similar in amplification during cell differentiation. mock transduced and in NTC or KLF13-specific shRNA lentivirus- Having established that KLF13 affects HPV genome amplifica- transduced cells (Figure 4a). DNA from the mock and lentivirus- tion, we next asked whether KLF13 could also affect viral gene transduced cells treated as described above was extracted and transcription. HPV31-positive CIN612 cells were infected with NTC

Figure 4. HPV31 genome amplification and viral gene expression is reduced in KLF13 deficient cells. (a) Western blot analysis of KLF13 in CIN612 cells untreated or infected with lentiviruses expressing KLF13-specific or NTC shRNAs upon calcium-induced keratinocyte differentiation. GAPDH is used as a loading control; involucrin is a marker for keratinocyte differentiation. (b) Southern blot analysis for HPV31 episomes in monolayer cultures of CIN612 cells transduced with KLF13-specific shRNA or NTC shRNA-expressing lentivirus and cultured for the indicated times. (c) HPV31 episomes in the experiments represented in panel b at T0 (0 h) and T48 (48 h) were determined by ImageJ software. The fold increase of episome copy number was defined as the quantitated band at T48 divided by that at T0. Quantitated data were from three experiments ± s.e.m. *Po0.05 vs NTC (unpaired t-test). (d)RT–PCR analysis of HPV31 genes coding for non-structural E1, E2, E6 and E7 proteins in CIN612 cells transduced with KLF13 shRNA or NTC shRNA-expressing lentivirus for 48 h. Values represent the mean fold change ± s.e.m. for three experiments relative to GAPDH. Black bars, NTC; gray bars, KLF13 shRNA. *Po0.05 vs NTC shRNA (unpaired t-test). (e)RT–PCR analysis of HPV31 genes coding for non-structural E1, E2, E4, E5, E6 and E7 proteins in CIN612 cells transduced with KLF13 shRNA or NTC shRNA-expressing lentivirus and differentiated in high calcium medium for 96 h. Values represent the mean fold change ± s.e.m. for three experiments relative to GAPDH. Black bars, NTC; gray bars, KLF13 shRNA. *Po0.05 vs NTC shRNA (unpaired t-test). (f) Ratio of HPV31 E1, E2, E6 and E7 gene expression 48 h after calcium-induced keratinocyte differentiation versus undifferentiated cells. CIN612 cells were transduced with KLF13 shRNA or NTC shRNA-expressing lentivirus for 48 h. Values represent the mean ± s.e.m. for three experiments relative to GAPDH. Black bars, mock; dark gray bars, NTC; light gray bars, KLF13 shRNA. *Po0.05 vs NTC shRNA (unpaired t-test).

© 2016 Macmillan Publishers Limited, part of Springer Nature. Oncogene (2016) 5565 – 5575 KLF13 positively regulates HPV life cycle W Zhang et al 5570 or KLF13-speciﬁc shRNA-expressing lentivirus for 48 h, and then high calcium medium was measured by western blot. In mock or harvested or cultured for an additional 96 h in high calcium NTC-transduced cells, total STAT5 was similar and phosphorylated medium to induce cell differentiation. Total RNA was isolated and STAT5 increased upon differentiation. However, in high calcium assayed by quantitative RT–PCR (qRT–PCR) for the levels of HPV differentiated CIN612 cells lacking KLF13, total STAT5 protein was transcripts. Knockdown of KLF13 caused a decrease in early viral decreased (Figures 5a and b) and this decrease may come from gene transcription in undifferentiated HPV31-positive CIN612 cells reduction of STAT5B protein (Figures 5a and c) as previously (Figure 4d). However, viral gene transcription was much lower reported.7 Importantly, in the absence of KLF13, phosphorylated 96 h aft