Effects of Metformin on the virus/host cell crosstalk in human papillomavirus‐positive cancer cells

Oncogenic types of human papillomaviruses (HPVs) are major human carcinogens. The viral E6/E7 oncogenes maintain the malignant growth of HPV‐positive cancer cells. Targeted E6/E7 inhibition results in efficient induction of cellular senescence, which could be exploited for therapeutic purposes. Here we show that viral E6/E7 expression is strongly downregulated by Metformin in HPV‐positive cervical cancer and head and neck cancer cells, both at the transcript and protein level. Metformin‐induced E6/E7 repression is glucose and PI3K‐dependent but—other than E6/E7 repression under hypoxia—AKT‐independent. Proteome analyses reveal that Metformin‐induced HPV oncogene repression is linked to the downregulation of cellular factors associated with E6/E7 expression in HPV‐positive cancer biopsies. Notably, despite efficient E6/E7 repression, Metformin induces only a reversible proliferative stop in HPV‐positive cancer cells and enables them to evade senescence. Metformin also efficiently blocks senescence induction in HPV‐positive cancer cells in response to targeted E6/E7 inhibition by RNA interference. Moreover, Metformin treatment enables HPV‐positive cancer cells to escape from chemotherapy‐induced senescence. These findings uncover profound effects of Metformin on the virus/host cell interactions and the phenotype of HPV‐positive cancer cells with implications for therapy‐induced senescence, for attempts to repurpose Metformin as an anticancer agent and for the development of E6/E7‐inhibitory therapeutic strategies.


| INTRODUCTION
Oncogenic types of human papillomaviruses (HPVs), such as HPV16 or HPV18, are closely linked to the development of prevalent anogenital and head and neck squamous cell carcinomas (HNSCCs), accounting for approximately 5% of the total cancer incidence in humans. 1 Best characterized is the causative role of HPVs for the development of cervical cancer, which each year is diagnosed in approximately 570 000 females and results in over 300 000 cancer deaths. 2 Cervical cancer cells are virtually always HPV-positive and their proliferation depends on the continuous expression of the multifunctional viral E6/E7 oncoproteins. 3 Inhibition of E6/E7 expression in HPV-positive cancer cells leads to rapid and efficient induction of cellular senescence, 4,5 classically considered to be an irreversible growth arrest. 6 Collectively, these findings indicate that HPV-positive cancer cells are "oncogene addicted" and must maintain viral E6/E7 oncogene expression in order to avoid induction of senescence.
The biguanide Metformin is a widely used drug for the first-line therapy of Type 2 diabetes (DT2) where it primarily acts through reducing hepatic gluconeogenesis. 7 In recent years, Metformin has also raised much interest in the field of oncology. A number of clinical studies indicate beneficial preventive and/or therapeutic effects of Metformin at least on some cancer entities in DT2 patients, and possibly also in nondiabetic patients. [8][9][10] However, other reports challenge these findings and question the advantageous effects of Metformin for cancer treatment. 11 The reasons for these discrepancies are still largely elusive and might be explained, for example, by differences in the study setup or by tumor type-specific responses. Conflicting data has also been published for cervical cancer. Metformin was reported to reduce the risk of cervical cancer development in DT2 patients 12 and to decrease cervical cancer specific and overall mortality in cervical cancer patients with DT2 13 whereas no effect of Metformin on the course of the disease was described in another study. 14 Both indirect and direct effects have been proposed to contribute to the antitumorigenic potential of Metformin. As indirect effects, the reduction of serum glucose and insulin concentrations could negatively affect tumor cell growth. Additionally, immunological and antiinflammatory effects may play a role. 7 On the other hand, there are a large number of studies indicating direct antitumorigenic potential, showing that Metformin can act antiproliferative in a broad spectrum of tumor cells in vitro. 7 These also include cervical cancer cells for which different potential mechanisms for mediating Metformin-induced growth suppression were reported, such as repressing Wnt signaling via DVL3 inhibition, 15 decreasing AMPK O-GlnNAcetylation, 16 or modulating the expression of HMGA2, 17 DEC1 18 or FOXM1. 19 Interestingly, however, none of these studies addressed the question whether Metformin affects the HPV oncogenes or the virus/host cell crosstalk in HPVpositive cancer cells.
In the present study, we show that Metformin leads to a strong inhibition of viral E6/E7 oncogene expression. At the same time,  RNAi experiments were performed utilizing siRNA pools of three different siRNAs each (si16E6/E7; si18E6/E7, si18E6), which efficiently and specifically block HPV16 or HPV18 E6 or E6/E7 oncogene expression, respectively, as described in detail elsewhere. [22][23][24] 2.2 | RNA and protein analyses Protein extractions and immunoblot analyses were performed as previously described, 22 with the exception that cells were lysed in CSK-1 buffer (10 nM Pipes pH 6.8, 300 mM NaCl, 1 mM EDTA, 300 mM RNA extractions and quantitative real-time PCRs were performed as described. 22 For E6/E7 mRNA measurements, primers were employed which recognize all three transcript classes coding for HPV16 or HPV18 E6/E7. 22 Relative quantification was performed by the comparative Ct (2 ΔΔCt ) method. 25  To investigate possible modulations of p53 targets, genes that are regulated by p53 with high confidence 28 were used to create a gmx file for subsequent GSEA. This gene set "P53 targets high confidence" was loaded into the GSEA software and its enrichment was examined for the proteome data of Metformin-treated SiHa cells (Table S1). For investigating the enrichment of this gene set after siRNA-mediated E6/E7 repression in cervical cancer cells, genes that were significantly differentially expressed in a corresponding transcriptome analysis of HeLa cells 29 were analyzed by GSEA.

| Proliferation, senescence and colony formation assays
Real-time analyses of cellular proliferation rates were performed using the IncuCyte S3 live-cell imaging system (Essen BioScience, Hertfordshire, UK) as described. 30 HeLa and SiHa cells were labeled with nuclear restricted mKATE2 fluorescent protein after the protocol provided by the supplier (Essen BioScience), treated with different concentrations of Metformin, as specified in the text, and labeled nuclei were counted over time. Four images per well were acquired every 4 hours at a magnification of 10× and analyzed using the IncuCyte S3 2019B software. Experiments were performed at least thrice in triplicates with consistent results.
For senescence detection, cells were stained for SA-β-Gal activity as described 30 and visualized by the EVOSxl Core Cell Imaging System (Thermo Fisher Scientific, Dreieich, Germany) with 20-fold magnification.
For senescence assays in Figure 5, cells were transfected with E6/E7-inhibitory siRNAs or control siRNA (siContr-1), cultivated in the absence or presence of Metformin, and stained for SA-β-Gal activities 5 days after transfection. All senescence assays were performed independently at least thrice with consistent results.
Colony formation assays (CFAs) were performed as previously described. 23 Briefly, cells were cultured for 24 hours under Metformin or under hypoxia and subsequently treated with 10 μM Etoposide or, depending on the cell line, with 0.1 μM or 0.2 μM Doxorubicin (see experimental schemes in Figure 6B and Figure S6B). The cells were grown for further 48 hours under Metformin or hypoxia, splitted and cultured for 7 to 12 days under normoxia in drug-free medium. Colonies were fixed and stained with formaldehyde-crystal violet. Control cells were treated accordingly, but consistently kept under normoxia without Metformin.
CFAs were performed independently at least thrice with consistent results.   Figure 1C). Repression of E6/E7 is further observed upon treatment with another biguanide, Phenformin, which may also have the potential to be repurposed for cancer treatment 31 ( Figure S1). The downregulation of E6/E7 by Metformin is not limited to cervical cancer cells, but is also detected in HPV16-positive HNSCC cells in a dose-( Figure S2A) and time-dependent ( Figure S2B) fashion, and is also observed at the transcript level ( Figure S2C).

| Statistical analyses
Release kinetics reveal that E6/E7 repression by Metformin is reversible in that a change to cell culture medium devoid of Metformin restores E6/E7 expression in a time-dependent manner, both in HPVpositive cervical cancer cells ( Figure 1D) and HPV-positive HNSCC cells ( Figure S2D).
We recently found that hypoxia leads to a strong downregulation of viral E6/E7 oncoprotein expression through a mechanism that can be counteracted by unphysiologically high glucose supply (25 mM)  Figure 2C). Thus, active PI3K is a key determinant for Metformin-induced HPV oncogene expression whereas AKT seems not to be involved in this process-in contrast to its critical role for hypoxia-induced E6/E7 repression.

| Modulation of the proteome of HPV-positive cancer cells by Metformin
To assess the effects of Metformin on the proteome composition of HPV-positive cancer cells, the relative changes in protein abundances were assessed in a mass spectrometry (MS)-based proteomics analysis 26 in Metformin-treated HPV16-positive SiHa cells (Table S1). Interestingly, employing preranked GSEA 33   p53 degradation. 43 Notably, in strong contrast, GSEA of the proteome data under Metformin-induced E6/E7 repression (Table S1) reveals that there is no significant enrichment for p53-induced factors ( Figure S3B). These findings raise the possibility that there is a differential regulation of p53, a critical factor for senescence induction in HPV-positive cancer cells, 4,5,44 under the two treatment conditions.
Indeed, immunoblot analyses reveal that-in contrast to the strong p53 induction upon siRNA-mediated E6/E7 repression-there is no p53 induction under Metformin in both SiHa and HeLa cells and p53 levels even fall below those of untreated control cells ( Figure 4D).
Finally, we also observed that particularly under prolonged treatment with higher Metformin doses, the cell counts of HeLa and SiHa cells fall below those at the starting point of the kinetic ( Figure 4A).
As Metformin has also the potential to act proapoptotic under certain conditions (see Section 4), we analyzed the expression of the apopto-

| Metformin interferes with prosenescent chemotherapy
Next, we tested whether Metformin may also interfere with pro-  Figure 6B). These effects are also observed upon cotreatment with Etoposide and Metformin in the absence of glucose ( Figure S5A,B).
Further, corresponding results are obtained for Doxorubicin, also showing that senescence is impaired ( Figure S6A) and colony formation capacity is substantially increased upon cotreatment of Doxorubicin with Metformin ( Figure S6B), corroborating that Metformin interferes with CT-induced senescence in HPV-positive cancer cells.

| DISCUSSION
Metformin is currently under intense discussion to be repurposed for cancer therapy and a large number of clinical trials examining the antitumorigenic potential of Metformin are ongoing. 47 The results of the repression in HPV-positive cancer cells. 5,44 Moreover, it is notable that active mTORC1 signaling can play an important role for efficient senescence induction in several cell models, 48 including in cervical cancer cells in response to E6/E7 repression or prosenescent CT, 23 and both hypoxia 23  only through inducing tumor cell death but also through causing senescence ("therapy-induced senescence, TIS"). 50,51 The here observed potential of Metformin to interfere with TIS is remarkable in regard of the fact that a large fraction of the clinical trials testing Metformin for its anticancer activity is performed by combining Metformin with CT or RT. 47 However, whereas some researchers consider TIS beneficial for cancer therapy, 52 others point at the possibility that, rarely, tumor cells can escape from TIS and eventually can exhibit a more aggressive growth behavior. 51 It also should be noted that Metformin not only can block cell growth but also has the potential to induce apoptosis at higher doses, a response that has been reported to be enhanced under low glucose levels. 53

DATA AVAILABILITY STATEMENT
The data and other items supporting the results of the study will be made available upon reasonable request. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD011095.