A paradigm for post‐embryonic Oct4 re‐expression: E7‐induced hydroxymethylation regulates Oct4 expression in cervical cancer

The Octamer‐binding transcription factor‐4 (Oct4) is upregulated in different malignancies, yet a paradigm for mechanisms of Oct4 post‐embryonic re‐expression is inadequately understood. In cervical cancer, Oct4 expression is higher in human papillomavirus (HPV)‐related than HPV‐unrelated cervical cancers and this upregulation correlates with the expression of the E7 oncogene. We have reported that E7 affects the Oct4‐transcriptional output and Oct4‐related phenotypes in cervical cancer, however, the underlying mechanism remains elusive. Here, we characterize the Oct4‐protein interactions in cervical cancer cells via computational analyses and Mass Spectrometry and reveal that Methyl‐binding proteins (MBD2 and MBD3), are determinants of Oct4‐driven transcription. E7 triggers MBD2 downregulation and TET1 upregulation, thereby disrupting the methylation status of the Oct4 gene. This coincides with an increase in the total DNA hydroxymethylation leading to the re‐expression of Oct4 in cervical cancer and likely affecting broader transcriptional patterns. Our findings reveal a previously unreported mechanism by which the E7 oncogene can regulate Oct4 re‐expression and global transcriptional patterns by increasing DNA hydroxymethylation and lowering the barrier to cellular plasticity during carcinogenesis.

oncogene. 8Proteomic and genomic approaches have provided insights on the role of Oct4 in pluripotency and stemness, nevertheless these approaches remain a challenge for elucidating a role in cancer due to lower expression levels of Oct4.
The repressive nucleosome remodeling and deacetylase complex (NuRD), which is associated with many physiological functions such as regulating pluripotency and immune cell development, 9,10 and involved in diseases like cancer, 11 is reported to control Oct4 expression in ESCs and recruit Oct4 on tis target genes.This is done via the binding of MBD2 and MBD3 components of the NuRD complex on the Oct4 promoter. 12Nevertheless, the involvement of the NuRD complex and its individual components (MBD2 and MBD3) in the regulation of Oct4 in cervical cancer has not been previously described.Additional to the regulation of Oct4 in ESC, MBD2 and MBD3 are associated with global DNA methylation and hydroxymethylation patterns respectively.[15] For instance, in cervical and hepatocellular carcinomas, the oxidized 5-hydroxymethycytosine (5hmC) levels are elevated in premalignant lesions while during advanced malignancy 5hmC levels are low. 16On the contrary, metastatic prostate tissues demonstrated higher levels of 5hmC compared to the normal or precancerous state. 17The driving force behind the elevated hydroxymethylation in either premalignant or metastatic lesions is the activation or re-expression of the TET-eleven translocation enzymes (TET), which carry out the oxidation of 5mC to 5hmC. 14,18TET1 expression was previously reported to peak in premalignant lesions and decrease in advanced cervical malignancy. 19re, we describe the underlying mechanism by which Oct4 is re-expressed in cervical cancer by identifying the Oct4-protein −protein interaction (PPI) network in cervical cancer cells.We have found that the E7 oncogene expression leads to an interaction of Oct4 with a NuRD complex variant, characterized by MBD3 reader, rather than MBD2, and affecting transcriptional regulation.Additionally, E7 expression mimics aspects of MBD2 inhibition, suggesting that this activity may be a broader way in which E7 affects transcriptional regulation in cells.We thus propose a unique contribution of the viral oncogene E7 in regulating global DNA (de) methylation dynamics during certain aspects of carcinogenesis, and a potential paradigm of Oct4 re-expression in postembryonic cells.

| Cell lines and culture
CaSki and C33A cervical cancer cells were purchased from ATCC and maintained in DMEM and MEM (mixed with 1% L-glutamine), respectively.293T epithelia cells (ATCC) and HaCaT immortalized keratinocytes (CLS) were maintained in DMEM.All culture media were supplemented with 1% penicillin/streptomycin (P/S) and 10% fetal bovine serum.For the transfection experiments, C33A and HaCaT cells were placed at a density of 3 × 10 5 and 5 × 10, 5 respectively.Twenty-four-hour post-placing cells were transfected with Fugene transfection reagent.For the transduction experiments 293T cells were placed at a 1 × 10 6 density.Mammalian and retroviral plasmids were used as shown in Supporting Information S1: Table 3.
Fourty-eight-, 72,-and 96-hours post-transfection, the retrovirus was collected and applied in C33A, CaSki, and HaCaT cells mixed with 1 µg/mL Polybrene.Stable cells were selected with specific antibiotics as shown in Supporting Information S1: Table 3.

| RNA extraction, cDNA synthesis, and PCR
Cells used for RNA extraction were isolated with the Trizol method.
To remove DNA impurities from RNA extracts, the AMBION-DNA free kit was used and 300 ng of cDNA was synthesized with the iSCRIPT cDNA synthesis kit.For RT-PCR (reverse-transcriptionpolymerase chain reaction) the KapaTaq PCR kit was used whereas qPCR (real time-PCR) was achieved by KAPA SYBR FAST qPCR Master Mix (2X) kit according to the manufacturer's guidelines.The primer sequences used are cited in Supporting Information S1: Table 1.For each gene, the average C( t ) value was determined and was normalized to housekeeping (Gapdh or Actin) genes.Unpaired t-test (two-tailed) was used to calculate statistical significance.

| 3'-mRNA quant sequencing
C33A cells were transfected with (i) Oct4 + Neo Bam empty and (ii) Oct4 + HPV16 E7 vectors.Fourty-eight-hours post-transfection, the cells were collected for RNA extraction with The RNeasy mini kit by Qiagen and the expression of Oct4 and E7 was validated with RT-PCR.For the Quant-sequencing and analysis of the samples, we sent the RNA to BSRC Fleming Institute in Greece.The differentially expressed gene (DEG) analysis was performed with the PANDORA algorithm.Following normalization with the metaseqr2 pipeline and filtering (to remove artefacts) we selected and identified genes with a p < 0.05 value and a fold change greater than 2 (FC > 2). 48For the validation of the DEGs we used primers targeting some highly upregulated (Thsb1, C-myc, Rgs4, Ankrd1, Bzw2, Ptprc) and highly downregulated (Gls2, Chdr, Znf483) genes.

| Computational analysis to identify Oct4 and E7 PPIs
To identify Oct4 and E7 PPI we searched the Universal Protein Resource (UniProt) 49 using the queries <gene:pou5f1 OR gene:oct4 OR gene:oct3 OR name: "pou domain class 5 transcription factor 1> and <e7 organism:human AND gene:e7>."The query keywords for Oct4 retrieved 13 protein entries from nine different species, while E7 query keywords retrieved 66 protein entries from 66 different HPV strains.The identifiers gathered were used as query keywords for the collection of PPI data of Oct4 and E7 proteins.
Relevant data were downloaded from IntAct (version 4.2.16), 50ioGRID (release 3.5.186,25/05/2020) 24 and Agile Protein Interactomes Dataserver (APID) (version January 2019). 23The data sets downloaded from APID served as templates where additional PPI data were added by searching manually literature papers.All available PPI data sets as gathered from APID were manually examined thoroughly to minimize the possibility of having missing or inaccurate values.By using UniProt's Retrieve/ID mapping tool every protein interactor was mapped to its current UniProt identifier as well as its HGNC (HUGO Gene Nomenclature Committee) approved gene name.Regarding the experimental methods used for the identification of each interaction we used the Ontology Lookup Service (OLS). 25Specifically, Molecular Interactions Controlled Vocabulary Ontology terms were gathered from OLS to describe in more detail the entries regarding the experimental/computational methods that were used to identify the PPIs.Homology of Oct4 proteins and their interactors was evaluated by exploiting Phylogenomic databases such as KEGG Orthology (KO), 51 OrthoDB, 52 and InParanoid, 53 which provide information about the existence of orthologous proteins between different species.Taking into consideration the importance of E7 oncoprotein and its role in the viral life cycle all available E7 proteins of HPV strains were accounted as homologs.
For each PPI a score was calculated based on a simplified form of IntAct's scoring methodology (MIscore) as described in Equation 1. 54 For each PPI there is information about the method used for its identification, the type of that method and the publication from where the interaction was extracted.For each method and method type ontology term, a specific score is assigned.
2.4.1 | Equation 1: PPI scoring methodology S: interaction score, m_i: method score, mt_i: method type score, p_i: number of distinct publications, n: literature instances reporting the interaction scored.Method and method-type scores are summed up for each interaction and if there is more than one indication for that interaction the score is the sum of the individual observations.Then, to that score the number of distinct publications referring to the interaction is added, and the score is normalized based on all entries in the database.This formulation was used to score all interaction pairs for human Oct4 and HPV16 E7.
To, integrate Oct4/E7 interactors from other species/strains we use a slightly altered scoring methodology, which also takes into consideration the evolutionary distance as encoded in the sequence similarity to their human/HPV16 counterparts.In more details, when summing the method and method-type scores (as in Equation 1) we weight their contribution by the percentage identity to the source interactor (human Oct4 or HPV16 E7) as determined by BlastP (run with default settings against the "Nonredundant protein sequences (nr)" database at the NCBI; accessed November 2020) (Equation 2).This way interologs from more distantly related species are downweighed, which is a conservative approach to integrating this information to the human PPI network without introducing many false positives.

| Protein extraction and western blot analysis
Cells collected for western blot analysis were lysed with RIPA cell lysis buffer (150 mM NaCl, 5 mM EDTA, 50 mM Tris-HCL, 1% Triton X-100, 0.1% SDS, and 0.5% sodium deoxycholate) supplemented with protease/phosphatase inhibitors.The concentration of protein samples was quantified and normalized by Bradford.Eighty micrograms of protein extracts were used and loaded onto a SDS-PAGE gel for electrophoresis.Proteins were then transferred onto a nitrocellulose membrane by using the Wet transfer apparatus.The membranes were blocked with 5% BSA in TBS-Tween blocking buffer for 1 h at room temperature.The membranes were then incubated overnight at 4°C with primary antibodies as mentioned in Supporting Information S1: Table 4 and secondary antibodies conjugated with HRP were added and incubated at room temperature for 1 h.Protein expression was then viewed with ECL reagents by using G-box imager.

| Co-immunoprecipitation
C33A cells were transfected with Oct4, Oct4 + HPV16 E7, and Oct4 + E7 L67R vectors and 48 h posttransfection, cells were collected for immunoprecipitation.Protein extraction was achieved as mentioned above using ice-cold RIPA lysis buffer supplemented with protease inhibitors.Protein samples were precleared with 1:1 Sepharose beads in RIPA buffer for 1 h under low agitation.Following centrifugation at 120 000g for 30 s to remove the beads, the protein lysates were separated for the (a) input control sample, (b) IgG negative control, and (c) for immunoprecipitating Oct4.The primary antibodies were incubated with the protein lysates and sepharose beads (slurry) overnight at 4°C under low agitation.The slurry and antibodies were removed following boiling the samples at 95°C for 10 min and then the samples were loaded onto SDS-PAGE gels for electrophoresis.Then Oct4-protein interactions were visualized via western blot analysis.

| Mass spectrometry (MS)
C33A cells were used for the MS experiment to identify the Oct4 interactome.To avoid specificity issues and variability in detection limits with the MS due to the low abundance of Oct4 in C33A, we transfected the cells with an Oct4 vector (Supporting Information S1: Table 3).Following successful transfection, immunoprecipitation was performed using an Oct4 antibody following a preclearing step with 1:1 Sepharose beads diluted in RIPA buffer.The negative control used was an IgG antibody to ensure that the immunoprecipitation of Oct4 is specific to the Oct4 antibody.Additional control for the experiment was a stable Oct4-knockdown C33A cell-line which we already established. 8The MS was performed, and the raw data after initial analysis were provided by the EMBL proteomics facility.
Interactors were identified with the R-software and the duplicated interactors were removed.Data were normalized by using the vsn package. 55The peptide reads were classified as hits (when false discovery rate [FDR] < 0.05 and a FC of at least twofold) and candidates (FDR < 0.2 and a FC of at least 1.5-fold) with LIMMA analysis.The top interactors were validated by western blot analysis.

| Enrichment analysis
[58] 2.9 | DNA extraction and dot blotting Membranes were then blocked in 5% BSA mixed in TBS-Tween for 1 h at room temperature and then were incubated with primary antibodies (Supporting Information S1: Table 4) for 1 h at room temperature.HRP-conjugated antibodies were added to the membranes for 1 h at room temperature and then dots were visualized with ECL reagents at a G-box imager.
Permeabilisation was achieved with 0.2% Triton X-100 and 5% BSA blocking buffer was used on the coverslips for 30 min.Primary antibodies (Supporting Information S1: Table 4) were diluted in blocking buffer and added to the cells overnight at 4°C.FITC antirabbit and Alexa-594 anti-mouse secondary antibodies were added and incubated at room temperature for an hour.Cells were then viewed with a Fluorescent Zeiss microscope.Vectashield antifade mounting medium with DAPI (Vector H-1200) was used for mounting the slides and staining cells nuclei.Images were processed and analyzed by using the ImageJ software. 59

| Tissue microarrays (TMAs)
TMAs were acquired from Biomax.us (CR602a) containing tissues from adjacent uterine cervix (10 cases), chronic cervicitis (10 cases), premalignant (10 cases), and cancer (30 cases).TMAs were first baked at 60°C for 40 min to prevent detachment and then they were deparaffinized with xylene.The rehydration was done with a series of ethanol dilutions and 10 nM citrate buffer was used for antigen retrieval.Five percent BSA was then added to the tissues as the blocking buffer and incubated at room temperature for 60 min.
Primary antibodies (a-5hmC and a-TET1) were diluted in 5% BSA at concentrations given in Supporting Information S1: Table 4 and were incubated overnight at 4°C.FITC anti-rabbit secondary antibodies were added to the tissues for 60 min at room temperature following washes with 1XPBS.Mounting medium with DAPI was used for visualizing cell nuclei.The imaging was done with Zeiss Axio Observer A1 microscope.For the analysis and quantification of the images we used the ImageJ software [83] and we grouped together the adjacent uterine cervix and chronic cervicitis cases to represent the nonmalignant condition in our analysis.RNase A, and incubated overnight at 65°C.Proteinase K 20 mg/mL was added and incubated at 65°C for 1 h and DNA was purified with PCR purification Kit from Qiagen.Then, the purified DNA was used to perform qPCR by using primers (Supporting Information S1:

| NucGreen dead stain
Table 2) for specific loci on the Oct4 gene.
Sonication was achieved by using the Diagenode Bioruptor TM UCD-200 Sonicator to reach a chromatin length between 100 -300 bp.
Twenty-five micrograms of sheared chromatin was used for meDIP and hmeDIP assays (Abcam) to examine methylation and hydroxymethylation on the Oct4 gene, respectively.The guidelines from the manufacturer's company were followed and the eluted DNA was quantified using a nanodrop.Primers targeting different regions on the Oct4 gene were used for qPCR (Supporting Information S1: Table 2).

| Statistical analyses
Statistical analyses, plotting, and quantification were conducted with GraphPad Prism v.6.0.All the experiments were performed using at least three biological replicates.Statistical significance was calculated at p < 0.05.

| Oct4 interacts with the NuRD complex in cervical cancer
Currently, evidence regarding Oct4-PPI derives from studies in ESC and induced pluripotent stem cells (iPSC) [20][21][22] however, the Oct4 interaction network in cancer remains uncharacterized.To profile the PPI of Oct4 in C33A cervical cancer cells (HPV-negative) we performed antibody-mediated affinity purification coupled to MS (AP-MS).To enhance the capacity for antibody mediatedpurification, we transiently overexpressed Oct4 in cervical cancer cells (Supporting Information S1: Table 3).For negative controls, we used an IgG antibody for immunoprecipitation and the Oct4knockdown C33A transduced cells. 8For the analysis, we used a cut-off p-value less than 0.05 (p < 0.05) and a FC greater than 2 (FC > 2).As anticipated, Oct4 (POU5F1) was found significantly enriched in C33A transfected cells while we identified 1605 enriched hits as indicated by the volcano plot in Figure 1A.We have validated the AP-MS data by checking the interaction between Oct4 and four enriched hits (MCM7, PCNA, Vimentin, and p53) via immunoprecipitation followed by western blot analysis (Figure 1B).
To bolster our proteomic approach, we performed computational analyses on publicly available PPI data sets.We generated an Oct4-PPI network by mining and processing data from existing literature using three different databases (APID, IntAct, and BioGRID [23][24][25] ).554 proteins were found to interact with Oct4, while complementary to the Oct4-PPI analysis we generated an E7-PPI network to identify putative common interactors of both Oct4 and the HPV oncogene E7.Our analyses revealed 41 proteins as reported interactors of both Oct4 and E7 (Supporting Information S1: Figure 1A).To examine the molecular and functional interplay between the Oct4-E7 common interactors, functional enrichment analysis was achieved.Notably, common interactors of Oct4 and E7 exist in several complexes, the majority involved in chromatin regulation and transcriptional control.The NuRD complex emerges from this analysis as the one containing most common interactors between Oct4 and E7 (Supporting Information S1: Figure 1B).
To investigate pathways regulated by Oct4 based on the AP-MS experiment in C33A cells, we performed pathway analysis using the Enrichr software (Figure 1C).Additionally, we examined whether Oct4 interactors participate in protein complexes in cervical cancer cells and cross-referenced these complexes with our computational analyses.Interestingly, we detected the NuRD complex to be also regulated by Oct4 in cervical cancer cells (Figure 1C) while many components of the NuRD complex immunoprecipitated with Oct4 such as CHD4, GATAD2A, GATAD2B, HDAC2, MTA2, and RBBP7 (Figure 1D).We have previously reported an interaction between Oct4 and the viral oncogene E7 in cervical cancer cells.This interaction was not detected when the aminoacid 67 of E7 was mutated (L67R E7 mutant). 8To validate the Oct4-NuRD interaction in C33A cells and further examine whether this interaction is affected by the expression of the viral oncogene E7, we transfected C33A cells with Oct4, Oct4 + E7, and Oct4 + L67R vectors and following Oct4 immunoprecipitation, western blot analysis was carried out to validate the Oct4-NuRD interaction.Remarkably, most interactions of Oct4 with components of the NuRD complex do not vary in the presence of E7.However, we found the readers of the complex known for their Methylated DNA binding domain (MBD), to vary in the presence of wildtype E7.MBD2 interacts with Oct4 only when the wildtype E7 is not expressed in C33A cells whereas upon E7 expression MBD3 immunoprecipitates with Oct4 (Figure 1E).We have confirmed an interaction between Oct4 and MBD3 by using HPV-positive HeLa and CaSki cells which express the endogenous levels of E7 and Oct4 (Supporting Information S1: Figure 1G).From previous studies we know that MBD2 and MBD3 are mutually exclusive members of the NuRD complex, 26 consistent with our coimmunoprecipitation data.8][29][30][31] Hence, to investigate whether the E7 oncogene has an impact on pathways related to DNA methylation/demethylation in C33A cells, we isolated RNA from Oct4-expressing and Oct4 + E7-expressing C33A cells for Quant sequencing.We used a cut-off p-value less than 0.05 (p < 0.05) and we have identified 1134 DEGs (453 upregulated and 681 downregulated) (Supporting Information S1: Figure 1C).Certain genes (highly upregulated and downregulated) were selected for validating the sequencing by performing qRT-PCR (Supporting Information S1: Figure 1D) whereas RT-PCR was performed to validate the successful transfection of E7 (Supporting Information S1: Figure 1E).Pathway analysis revealed the mechanism of Base excision repair (BER) to be controlled by both Oct4 and E7 (Supporting Information S1: Figure 1F).The BER mechanism deletes epigenetic marks on chromatin structure 32,33 converting methylated cytosines into their unmethylated state.This conversion is part of the active demethylation process which converts 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC).5fC and 5caC are converted to unmethylated cytosine through this BER system. 32,33To check whether the increase in TET1 reflects an increase in 5hmC, genomic DNA from C33A cells was used at two different concentrations (100 and 10 ng) for dot blot analysis.At 100 ng we detected an increase in 5hmC both in the presence of wildtype and mutant E7 compared to the empty control (Figure 2D).Additionally, we used immunofluorescence to validate the elevated 5hmC in C33A when E7 was expressed.Both 5hmC and 5mC antibodies were used to stain the cells since an increase in 5hmC reflects a decrease in 5mC and vice versa.We observed a slight increase in 5hmC in HPV16 E7-C33A and E7 L67R-C33A cells compared to the empty-C33A cells, even though no major difference in 5mC intensity was reported (Figure 2E).
Similarly, we reported an increase in 5hmC upon the presence of wildtype E7 at 10 and 40 ng in HaCaT cells (Figure 2I).Immunofluorescence analysis validated the global increase in 5hmC (and decrease in 5mC) in E7-expressing HaCaT cells compared to the empty control and the mutant E7 (Figure 2J).The disparity between the two cell lines regarding the levels of 5hmC in the presence of the L67R mutant could be due to the distinct genetic and mutational background of the two cell lines and to other factors that influence global DNA methylation dynamics.
To assess the relevance of our findings during cervical carcinogenesis, TMAs including biopsies from non-malignant, premalignant, and malignant cervices were used to investigate TET1 expression and 5hmC.We reported high levels of both TET1 and 5hmC in premalignant cervical tissues compared to non-malignant and malignant ones.This is consistent with literature 34,35 and further validates our data since most of premalignant cases express HPV E7 (Supporting Information S1: Figure 2A-B).
We thus propose a novel function of E7 in modifying methylation-demethylation dynamics through the involvement of enzymes which catalyze the oxidation of 5mC to 5hmC (Supporting Information S1: Figure 3A).To further confirm the impact of E7 in demethylation, we analyzed the mRNA expression of certain genes participating in the demethylation process as revealed by the Quantseq data.When E7 was expressed, we noticed an upregulation of the TDG enzyme (thymine DNA glycolase) and its interactor GADD45A 36 which participate in the active demethylation cycle and take part in The Oct4 interaction with MBD proteins in cervical cancer is regulated by E7. (A) Volcano plot indicating 1605 proteins interacting with Oct4 in C33A cervical cancer cells as revealed by mass spectrometry (MS).IgG was used as the negative control of the experiment, and three independent replicates were used.(B) Validation of MS through the immunoprecipitation of Oct4 in C33A and C33A Oct4-knockdown cells followed by western blot analysis.IgG was used as the negative control and 5% input was used to verify the expression of these proteins in the cells.(E) C33A cells transfected with Oct4, Oct4 + E7, and Oct4 + L67R vectors were used for the immunoprecipitation experiment.Oct4 was pulled down with an Oct4 antibody and western blot analysis was performed to reveal Oct4 interactions with NuRD-associated proteins.IgG was used as the negative control of the experiment while 5% input verifies the expression of these proteins in cells.NuRD, nucleosome remodeling and deacetylase complex; Oct4, octamer-binding transcription factor-4.
the BER pathway (Supporting Information S1: Figure 3A, 3B).Apart from DEGs illustrated by the Quant-seq, we checked the profile of certain enzymes known for their methylating activities observing a significant downregulation in DNA methyltransferase enzymes (DNMT1, DNMT3A, and DNMT3B) (Supporting Information S1: Figure 3B).Thus, we propose the contribution of the viral oncogene E7 in the active demethylation cycle (Supporting Information S1: Figure 3A).

| E7 alters the methylation status of the human Oct4 gene
To address the impact of E7 on hydroxymethylation locally, we conducted DNA-methylation and DNA-hydroxymethylation Chromatin immunoprecipitation on the human Oct4 gene.For this reason, we transfected C33A cervical cancer cells with the Neo Bam empty and HPV16-E7 vectors while cells were fixed and collected for Chromatin shearing and for verifying successful transfection (Supporting Information S1: Figure 4A-4B).We observed a significant increase in the methylation state of 10 randomly selected loci on the Oct4 gene in Neo Bam empty-expressing C33A cells compared to the IgG control (Figure 3A).These 10 loci are spanning various regions on the Oct4 gene (including upstream and downstream the transcription start site [TSS]) (Supporting Information S1: Figure 4C).E7-expressing C33A cells displayed no significant change in the methylation of the 10 loci tested when compared to the IgG control (Figure 3A).
In hydroxymethylation-DNA immunoprecipitation experiments, we found an enrichment on certain loci on the Oct4 gene only in E7-expressing C33A cells.These enriched loci are found closer to TSS and downstream the TSS (very few loci were enriched upstream the TSS) (Figure 3B).

| MBD2 downregulation modifies global hydroxymethylation in cervical cancer cells
Unlike MBD2, MBD3 cannot make a strong contact with methylated cytosines, instead it binds unmethylated or hydroxymethylated cytosines. 31,37For this reason, we assessed MBD3 as a reader of E7-induced hydroxymethylation in cells.Surprisingly, MBD3 mRNA expression was unaffected by the presence of E7 (wildtype and mutant) in C33A cells but not in keratinocytes (Figure 4A).This finding led us hypothesize that the elevated 5hmC in E7-expressing cells could be secondary to a decrease in their methylation status.
Consequently, we checked MBD2 expression and found a significant reduction (mRNA and protein) in E7-expressing HaCaT and C33A cells (Figure 4B and Supporting Information S1: Figure 5A).The reduced MBD2 pattern detected in the L67R-expressing C33A cells (Figure 4B) possibly explains the increase in global 5hmC in L67Rtransfected C33A cells (Figure 2D).To further ensure that the reduced MBD2 expression led to an increase in global hydroxymethylation, we performed a dot blot analysis in two cervical cancer cells lines (C33A and CaSki) transduced with shRNA MBD2 knockdown and the shluciferase control.While validating the MBD2 knockdown, we noticed that shMBD2-2 in C33A cells could not yield a reduced expression of MBD2 (Supporting Information S1: Figure 5B).For this reason, we chose to omit these cells from further analyses.We observed an increase in 5hmC in shMBD2-1transduced C33A cells compared to the shluciferase-control. Similarly, in shMBD2-knockdown CaSki cells, there was an increase in 5hmC compared to the shluciferase control both at the 25 and 100 ng (Figure 4E).Lastly, we performed an immunofluorescence analysis to verify the elevated 5hmC observed by dot blots by using shMBD2-knockdown C33A and CaSki cells.Again, we found a slight increase in 5hmC in shMBD2 knockdown cells compared to the shluciferase control (Figure 4D,F) however, while quantifying changes in intensity the statistical trend was insignificant (Supporting Information S1: Figure 5C).

| E7 expression transcriptionally mimics MBD2-knockdown C33A cells
We hypothesized that the E7-MBD2 interplay during carcinogenesis could affect transcriptional traits of cells (not only the methylation/ hydroxymethylation status).For this reason, we assumed that E7 expression can mimic the transcriptional profile of MBD2 knockdown in C33A cells.To check this, we randomly selected genes from different molecular pathways such as the apoptotic, EMT, and stemness-related pathways (Figure 5) and compared their expression when E7 is present in C33A cells versus when MBD2 is knockeddown.
Remarkably, we noticed that 14 out of 15 genes that we To examine the mechanism behind the reduced viability in C33A cells upon treatment with KCC07, we asked whether these cells were subjected to cell death.For this reason, we used the NucGreen stain which gets incorporated in the nucleus of apoptotic cells due to the fractured cellular membranes.We quantified the number of green

C33A,
CaSki, and HaCaT cells were collected for the Dot blot experiments at a density of 6 × 10 6 cells.Cell pellets were treated as indicated by the DNAeasy Blood and Tissue kit by Qiagen.Eluted genomic DNA was quantified by nanodrop and boiled at 95°C for 10 min to allow denaturation and then a volume of 1.5 µL of the samples was added on a nitrocellulose membrane in the form of dots.

For
imaging and quantifying dead cells, the NucGreen Dead-488 Ready Probes Reagent from Invitrogen was used.C33A cells transfected with Oct4, HPV16 E7, and E7 L67R vectors were treated with 250 nM KCC07 MBD2 inhibitor or the DMSO control.At 24-, 48,-and 72-hours post-treatment, the cells were treated with the NucGreen Reagents as per the manufacturer's protocol and then visualized with a Fluorescent Zeiss Microscope and analyzed with ImageJ.2.13 | ChIP-qPCRCervical cancer C33A cells were transfected with (i) Neo-Bam empty, (ii) HPV16 E7, and (iii) E7 L67R vectors. 1 × 10 7 cells were used and DNA−protein complexes were cross-linked with 1% formaldehyde for 10 min and 125 mM glycine was added for 5 min for the quenching process.Cells were washed with ice-cold sterile 1xPBS and collected with centrifugation at 2000 rpm for 5 min at 4°C.ChIP lysis buffer (50 mM HEPES-KOH pH 7.5, 140 mM NaCl, 1 mM EDTA pH8, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, protease inhibitors added last) was added to the cell pellets for 10 min (on ice) and sonication to allow chromatin shearing was performed with the Diagenode Bioruptor TM UCD-200 Sonicator (to achieve a chromatin length of around 100−300 bp (sonication pulses 7 s on and 7 s off for a period of 10 min)).Fifteen micrograms of the sheared chromatin was used for the immunoprecipitation step with primary antibodies (Supporting Information S1: Table4) incubated for 1 h at 4°C and then 50 µL sepharose beads primed with 75 ng/µL herring sperm and 0.1 µg/µL BSA were added.The immunoprecipitated DNA was treated with 5 M NaCL, 10 mg/mL

3. 2 |
E7 elevates TET1 expression and global hydroxymethylation in C33A and HaCaT cells Cervical cancer cells (C33A) and human immortalized keratinocytes (HaCaT) transfected with an empty control (Neo Bam), HPV16-E7, and the mutant L67R were used to assess TET1 expression.C33A and HaCaT cells expressing the wildtype but not the mutant E7 exhibit higher TET1 mRNA and protein expression.(Figure 2A−C and 2F−H).
(C) Pathway analysis performed by the Enrichr Software demonstrates the pathways regulated by Oct4 interactors (dark blue color) while protein complexes involved in epigenetic regulation, which interact with Oct4, are shown in light blue color.(D) Components of the NuRD complex interact with Oct4 as shown by MS while the red bars indicate Oct4 immunoprecipitation and blue bars indicate IgG immunoprecipitation.
increases TET1 expression and global hydroxymethylation in C33A and HaCaT cells.C33A cells transfected with Neo Bam empty, HPV16 E7 and E7 L67R plasmids were used to check TET1 (A) mRNA and (B) protein levels.GAPDH was used as a control.(C) Bar chart illustrating the average relative protein expression of TET1 compared to GAPDH.(D) Dot blot experiment illustrates 5hmC levels in C33A cells transfected with Neo Bam empty, HPV16 E7 and E7 L67R vectors.Genomic DNA was used at concentrations 100 and 10 ng.Methylene blue was used as a loading control whereas BSA was used as a negative control of the experiment.The violin plot demonstrates the relative 5hmc levels compared to methylene blue stain.(E) Immunofluorescence analysis shows 5hmC and 5mC in C33A cells, while DAPI was used to stain cell nuclei (scale bars: 50 μm).HaCaT transfected with Neo Bam empty, HPV16 E7 and E7 L67R vectors were used while TET1 (F) mRNA and (G) protein levels were investigated.GAPDH was used as a control.(H) Bar chart illustrating the average relative protein expression of TET1 compared to GAPDH.(I) Dot blot experiment revealed 5hmC levels in HaCaT cells transfected with Neo Bam empty, HPV16 E7 and E7 L67R vectors.Genomic DNA was used at two concentrations (40 and 10 ng).The violin plot demonstrates the relative 5hmc levels compared to methylene blue stain.(J) Immunofluorescence shows 5hmC and 5mC in HaCaT cells.Three independent replicates were used and plotted values on graphs are the mean ± SEM.The statistical analysis was performed with two-tailed unpaired t-test with p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).HPV, human papillomavirus; ns, nonsignificant; TET1, TET-eleven translocation enzymes; 5hmC, 5-hydroxymethycytosine.

8 3. 6 |
have checked (Cyt-c, Tp53, Casp3, Casp9, Bax [Figure 5A], Ecad, Fn, Slug, Ncad [Figure 5B], Sox2, Klf4, Nanog, C-myc, Aldh1a [Figure 5C]) display similar trends both in the expression of E7 and in the downregulation of MBD2 in C33A cells.While examining the expression level of Oct4 in MBD2 knockdown-C33A cells we noticed a significant downregulation compared to the control cells (Supporting Information S1: Figure 6A) which this is not in line with the upregulation we report when E7 is present in C33A cells.MBD2 chemical inhibition impairs MBD2 binding on the Oct4 gene and attenuates cell viability To identify the reason behind the reduced mRNA expression of Oct4 upon MBD2 downregulation, we performed chromatin immunoprecipitation to determine whether MBD2 binds the Oct4 gene.We observed an enrichment of MBD2 on the Oct4 gene in C33A cells compared to IgG control.However, the MBD2 binding was restricted in the presence of wildtype E7 but not in the presence of the E7 mutant L67R (Supporting Information S1: Figure 6B).Hence, we decided to use an MBD2 chemical inhibitor (KCC07), (which impairs MBD2 function without affecting MBD2 expression) to check whether this restricted binding of MBD2 on the Oct4 gene is due to MBD2 lower expression or due to a displacement of MBD2 from Oct4.We used the KCC07 inhibitor at 250 nM for 24-, 48,-and F I G U R E 3 E7 expression in C33A cells deregulates the methylation status on the human Oct4 gene.C33A cells transfected with Neo Bam empty and HPV16 E7 vectors were fixed, and DNA was sheared for (A) methylation-DNA immunoprecipitation and (B) hydroxymethylation-DNA immunoprecipitation.IgG was used as the negative control of the experiment.qRT-PCR was performed to check 5mC and 5hmC enrichment by using primers targeting different loci on the Oct4 gene.Three independent replicates were used and plotted values on graphs are the mean ± SEM.The statistical analysis was performed with two-tailed unpaired t-test with p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).HPV, human papillomavirus; ns, nonsignificant; Oct4, octamer-binding transcription factor-4; RT-PCR, reverse-transcriptionpolymerase chain reaction; 5hmC, 5-hydroxymethycytosine.F I G U R E 4 MBD2 knockdown increases global 5hmC in cervical cancer cells.HaCaT and C33A cells were transfected with Neo Bam empty, HPV16 E7 and E7 L67R vectors to investigate the mRNA levels of (A) MBD3 and (B) MBD2.(C) Dot plot experiment was performed to assess 5hmC in C33A cells stably transduced with the shLuciferase control and the MBD2 knockdown.Genomic DNA was used at two concentrations (200 and 50 ng).Methylene blue was used as a loading control whereas BSA was used as a negative control.The relative 5hmc level compared to methylene blue stain was plotted on a violin chart.(D) Immunofluorescence validates the increase in 5hmC in MBD2-knockdown C33A cells.(E) The global 5hmC in CaSki cells transduced with the shLuciferase control and Mbd2 knockdown was investigated via a dot blot experiment.The bar chart illustrates the relative 5hmC level.(F) Immunofluorescence images verify the elevated 5hmC in Mbd2 knockdown CaSki cells.DAPI was used to stain cell nuclei and scale bars indicate 50 µm.Three independent replicates were used and plotted values on graphs are the mean ± SEM.The statistical analysis was performed with two-tailed unpaired t-test with p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).HPV, human papillomavirus; ns, nonsignificant; 5hmC, 5-hydroxymethycytosine. 72-hours on C33A cells and at 48 h post-treatment cells were collected for chromatin immunoprecipitation.We noticed that the MBD2 binding on the Oct4 gene was restricted in the presence of the inhibitor both in C33A cells expressing the Neo Bam empty control and the E7 mutant, conditions at which we found previously that MBD2 could bind the Oct4 gene (Figure 6A).To further characterize the impact of the KCC07 inhibitor on the viability of C33A cells expressing the Neo bam empty control and the wildtype and mutant E7, we treated the cells with 250 nM of the inhibitor and checked cell numbers at 24-, 48,-and 72-hours posttreatment.At 48-and 72-hours post-treatment there was a reduction in viability of empty-expressing and L67R-expressing C33A cells, F I G U R E 5 E7 expression mimics MBD2 knockdown in C33A cells at the transcriptional level.C33A cells either transduced with the MBD2 knockdown and the shLuciferase control or transfected with the Neo Bam empty and HPV16 E7 vectors were collected, and RNA was extracted for qRT-PCR analysis.Genes in the (A) apoptotic, (B) EMT, and (C) stemness-associated pathways were investigated.Three independent replicates were used and plotted values on graphs are the mean ± SEM.The statistical analysis was calculated with two-tailed unpaired t-test with p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).HPV, human papillomavirus; ns, nonsignificant; RT-PCR, reverse-transcriptionpolymerase chain reaction.however this is not the case in E7-expressing C33A cells which remained unresponsive to the inhibitor (Figure 6B).We also checked the viability of HaCaT keratinocytes and two HPV-positive cervical cancer cells lines (HeLa and CaSki) when treated with the MBD2 inhibitor.None of the cell lines had reduced viability when treated with the inhibitor compared to the DMSO control for the 72 h that we have tested (Supporting Information S1: Figure 7A) possibly indicating that normal cells and HPV-positive cancer cells are less sensitive to MBD2 inhibition.

F
I G U R E 6 MBD2 inhibition attenuates viability and increases apoptosis in C33A cells not expressing HPV16 E7. (A) The MBD2 inhibitor was used at 250 nM concentration to investigate the binding of MBD2 on the Oct4 gene.DMSO was used as a control for the experiment and 48 h post-application cells were collected for chromatin immunoprecipitation. (B) KCC07 MBD2 inhibitor was applied on HPV-negative C33A cervical cancer cells transfected with Neo Bam empty, HPV16 E7 and E7 L67R vectors.The inhibitor was applied at 250 nM and cell viability was measured for 24-, 48,-and 72-hours post-application.DMSO was used as the negative control.(C) Apoptosis was investigated by using the NucGREEN stain and apoptotic (green) cells were visualized under the microscope.Three independent replicates were used and plotted values on graphs are the mean ± SEM.The statistical analysis was calculated with two-tailed unpaired t-test with p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).HPV, human papillomavirus; ns, nonsignificant; Oct4, octamer-binding transcription factor-4.