Covalent Inhibitors of Human Papillomavirus Type 16 E6 Protein Restore p53 Function and Suppress Growth of HPV Driven Tumors in Vivo

Abstract

High-risk human papillomaviruses (HPVs) promote malignant progression through sustained expression of the viral oncoprotein E6, which drives degradation of the tumor suppressor p53 and creates an oncogenic dependency in HPV-positive cancers. Here, we identify a genotype-defined therapeutic vulnerability by selectively and irreversibly inactivating HPV-16 E6 through covalent targeting a cysteine proximal to its E6AP-binding interface. Pharmacologic inhibition of E6 restored p53 protein stability and transcriptional activity in HPV-16–positive cancer cells, inducing apoptosis and senescence while sparing HPV-negative epithelial cells. A CRISPR-engineered E6 cysteine-to-serine knock-in abolished compound activity in vitro and in vivo, establishing on-target mechanism. Transcriptomic profiling confirmed activation of p53-dependent tumor suppressor programs following E6 inactivation. In xenograft models of cervical and oropharyngeal cancer, irreversible inhibition of E6 suppressed growth of established tumors with minimal toxicity and no evidence of acquired resistance. These findings support covalent inactivation of HPV-16 E6 by a small molecule as a novel therapeutic strategy for HPV-associated malignancies.

Introduction

There are no antiviral-specific therapeutics for the millions of individuals infected with human papillomavirus (HPV). Although infections of the cervical, anal, genital, and oropharyngeal epithelia are highly prevalent, most are benign and resolve spontaneously. Persistent infection with specific high-risk (HR) HPV genotypes can progress to locally invasive and metastatic cancers. HPVs account for ~ 5% of all cancers worldwide (1, 2). These epithelial malignancies typically evolve from benign precursor lesions to advanced disease over years to decades, providing a prolonged window for therapeutic intervention.

The E6 oncoprotein is required for both stable viral genome maintenance and HR HPV–mediated transformation of keratinocytes (3–5). E6 is a multifunctional protein that interacts with numerous cellular partners, among which the HECT-domain ubiquitin ligase E6AP (UBE3A) and the tumor suppressor p53 are the best characterized. E6 does not bind p53 in the absence of E6AP, nor does E6AP interact with p53 independently; instead, E6 serves as an adaptor that recruits E6AP to p53, leading to ubiquitination and proteasomal degradation of p53 (6). Both HPV-16 E6 and E7 are consistently expressed in HPV-driven tumors. This is particularly relevant to our approach, as HR HPV E7 alone stabilizes and activates p53 (7, 8), and our chemical strategy to inactivate E6 does not interfere with E7 expression. HPV-positive tumor cells are therefore dependent on sustained expression of both viral oncoproteins, and genetic depletion of E6 by RNAi or CRISPR-mediated knockout restores wild-type p53 levels and activity, resulting in growth arrest or apoptosis in cervical cancer cells (9–15).

Structural studies have provided critical insight into the molecular basis of E6 function. Travé et al. solved the crystal structure of HPV-16 E6 in complex with a maltose-binding protein (MBP)–fused E6AP peptide containing an alpha-helical LxxLL motif (PDB 4GIZ) (16). This motif docks into a well-defined binding pocket of E6 (16–21). Subsequent structural analysis of the trimeric HPV-16 E6•E6AP•p53 complex (PDB 4XR8) revealed that E6AP binding induces conformational stabilization of E6, exposing an extended interaction surface that enables p53 recruitment (20, 22). NMR studies further demonstrated that free E6 is conformationally flexible, with its two zinc-binding domains adopting variable orientations, whereas LxxLL peptide binding rigidifies the connecting helix and organizes the protein into a single functional conformation (16). In this stabilized state, both zinc-binding domains contribute to the formation of a large p53 interaction interface. The E6AP-binding pocket also serves as a molecular “hot spot” for interactions with additional cellular proteins containing LxxLL motifs [reviewed in (16, 23, 24)].

Here we describe the biological consequences of small molecules designed to irreversibly bind and inactivate HPV-16 E6. By covalently targeting E6, these compounds restore p53 stability and transcriptional activity, with the expectation that additional E6-dependent oncogenic functions are simultaneously disrupted.

Results

Characterization of compounds that specifically bind to HPV-16 E6 and block p53 degradation

Screening and biochemical characterization of a large series of compounds that bind to and form a covalent bond with cysteine at position 51 (Cys-51) proximal to the E6AP binding pocket of the HPV-16 E6 protein will be reported elsewhere. Some publications refer to this as cysteine 58 due to an upstream ATG codon that adds 7 amino acids onto the E6 reading frame. Using the profile of molecules we previously identified (13), derivatives with an acrylamide warhead predicted to approximate cysteine 51 were synthesized. In addition, a small molecule library of warhead containing molecules scanned for docking to E6 were screened. Here we describe the biological effects of two compounds: KTI-218 and KTI-240 (see insets Figure 1A,B,C). To screen for activity in live cells, we constructed a luciferase based ‘gain of signal’ reporter assay to measure inhibition of E6•E6AP-mediated p53 degradation. Wild-type p53 was cloned in frame with renilla luciferase (R-Luc). The E6•E6AP complex binds the p53 moiety of the fusion protein and induces degradation of the chimeric protein (25, 26). These DNA constructs were transfected into HPV-16 expressing human cervical cancer derived cell line SiHa, which was selected because levels of E6 and E7 are directed by the native HPV-16 promoter. RPE-1 cells (telomerase [hTERT] immortalized HPV-negative epithelial cells expressing wild-type p53) expressing the identical p53-R-Luc reporters were also isolated. The E6 mediated pathway for degradation of p53 in the cervical cell lines was validated using siRNAs to inhibit expression of E6, E6+E7, and E6AP, which resulted in increased p53-R-Luc signal, whereas scrambled control siRNA did not (Supplementary Figure 1A). Following incubation of SiHa cells with KTI-218 or KTI-240, p53-luc increased by 8- and 14-fold with an EC50 of 4.2±0.1 μM and 2.3±0.1 μM, respectively, without changing p53-luc activity in the control RPE-1 reporter cells (Figure 1A, B). To investigate the contribution of the covalent group, KTI-239, a non-covalent analog of KTI-218 was synthesized. KTI-239 induced p53-luc activity at a much lesser extent than KTI-218 in SiHa-luc cells without affecting Luc activity in RPE-1 cells (Figures 1C,D).

Figure 1. HPV16 E6 inhibition selectively increases p53 protein and requires cysteine 51.

A clone of HPV-16 E6/E7 expressing human cervical cancer SiHa cells with a p53-Luciferase (p53-Luc) reporter and a HPV-negative RPE-1 cell clone expressing the p53-Luc reporter were incubated with 0.02% (v/v) DMSO or increasing concentrations of compounds. SiHa-p53-Luc cells (red) and RPE-1 p53-luc cells (blue) were treated for 24h with (A) KTI-218 or (B) KTI-240. KTI-239 is an analog of KTI-218 that lacks the acrylamide warhead incubated with SiHa-p53-Luc (C) or RPE-p53-luc cells (D). Luc signal was normalized to cell viability and p53-Luc induction expressed as fold-change over DMSO. p53 immunoblots of SiHa (E) and RPE-1 (F) cells cultured in KTI-218, KTI-239 and KTI-240 for 24 hours. (G) KTI-218 increased p53 within 6 hours in SiHa cells but not in RPE-1 or SiHa-C51S cells. (H) siRNA knockdown of E6AP and 16E6E7 result in comparable p53 inductions in both SiHa and SiHa-C51S cells. SCR–scrambled RNAi control. UT–untreated cells. (I) SiHa C51S cells were treated with KTI-218, −239, and −240 for 24 hours and did not induce p53 protein levels. ETO–etoposide. D1, D2–DMSO controls. E6 inhibition increases p53 in HPV-16 E6 expressing human oropharyngeal cancer cells SCC-47 (J) and SCC-104 (K) cultured with KTI-218, KTI-240, etoposide (ETO, 25 μM), or DMSO (D1, D2) for 24 hours, harvested and p53 and p21 proteins analyzed by immunoblot. (L) p53 in human cervical dysplasia W12-E cells cultured in KTI-218, KTI-240 or DMSO (D1, D2) for 24 hours. Protein expression was normalized to GAPDH and expressed as fold change. Data expressed as S.E.M and was analyzed using one-way ANOVA with Dunnett’s or Bonferroni post hoc analysis. Each experiment was completed at least three independent times, * p<0.05.

E6 inhibition was confirmed by evaluating cellular p53 protein levels following exposure to compounds in culture media. Western blots showed robust p53 elevation in HPV16-positive SiHa cells (Figure 1E) but not in p53 wild-type RPE-1 cells (Figure 1F), implying this induction is not due to off-target genotoxic stress. The positive control etoposide induced p53 in both cell lines, confirming an intact p53 pathway. KTI-239 lacks the acrylamide warhead of KTI-218 and did not produce a rise in p53 levels. Increased levels of p53 protein in SiHa cells were detectable by 6 h in culture with KTI-218 (Figure 1G).

To provide evidence for the requirement of covalent bonding to Cys51 for E6 inhibition, CRISPR mediated mutagenesis was used to convert the codon for cysteine 51 in HPV-16 E6 to serine. This was possible because unlike most human-derived HPV cancer cells that have tens to hundreds of copies HPV16, SiHa cells have only one or two integrated copies. We isolated a C51S clone and confirmed that it was derived from the parental SiHa cell line by autosomal STR profiling and that the DNA sequences of the E6 region were otherwise preserved (Supplementary Figures 2A,B). The C51S SiHa cells proliferated in culture and levels of p53 were comparable to the parental SiHa line (Supplementary Figure 2C), which was expected, for while high-risk HPV E6 proteins bind to E6AP, the targeted cysteine is exclusive to HR HPV types 16, 35, 45, and 68 and is not present in other oncogenic HPV genotypes. RNAi mediated depletion of E6AP or HPV-16 E6 resulted in increased p53 levels similar to wild-type SiHa cells (Figure 1H), demonstrating the E6•E6AP pathway for p53 degradation is intact in the CRISPR-edited SiHa cells and that cysteine 51 can be functionally replaced by serine (27). Importantly, incubation of C51S SiHa cells with KTI-218 or KTI-240 did not increase p53 levels, while positive control etoposide did (Figure 1I). These results infer that the observed p53 induction depends on covalent binding of the compound to cysteine 51 in HPV-16 E6 and is not due to a consequence of a covalent association with a cellular protein.

Levels of p53 protein also increased in HPV-16 E6 expressing human oropharyngeal cancer (OPC) cancer derived cell lines such as UM-SCC-47 and UM-SCC-104 (Figures 1J,K) following compound exposure. Elevated levels of the p53 transcription target p21 protein were also detected (Figures 1J,K; Supplementary Figures 3A–D). We next investigated whether these compounds could inhibit E6 activity in human keratinocytes that maintain episomal HPV-16 genomes and express all early viral proteins. W12-E cells, derived from a woman with a premalignant cervical lesion, showed robust induction of p53 levels following incubation in media containing KTI-218 and KTI-240 (Figure 1L). Etoposide was included as a control to induce DNA damage and p53; however, in most of our experiments with HPV-16 E6 expressing cells and as shown for the OPC line SCC-104, etoposide did not induce p53, presumably due to the overriding effect of E6 mediated p53 degradation.

KTI-240 requires cysteine 51 in HPV16 E6 to inhibit E6AP binding

To further determine whether these compounds require cysteine 51 in HPV16 E6 to disrupt E6AP binding and restore p53 signaling, we performed co-immunoprecipitation (co-IP) assays (Fig. 2A–C). SiHa cells were transfected with eGFP, eGFP–HPV16 E6 wild-type, or the cysteine mutant eGFP–HPV16 E6 C51S, followed by treatment with vehicle (DMSO) or KTI-240 for 16 hours. Cell lysates were immunoprecipitated using a GFP antibody and analyzed by immunoblotting for E6AP, GFP, and E6. Both wild-type and C51S mutant E6 proteins co-immunoprecipitated E6AP under vehicle-treated conditions. Notably, KTI-240 treatment disrupted the interaction between E6AP and wild-type HPV16 E6, as evidenced by loss of E6AP co-ip. In contrast, KTI-240 had no effect on the interaction between E6AP and the C51S mutant (Fig. 2A–C).

Figure 2. Specific covalent inhibition of HPV16 E6 at cysteine 51 disrupts E6AP binding and reactivates p53 signaling.

SiHa cells transfected with eGFP, eGFP-16E6 WT or eGFP-16E6 C51S and treated with KTI-240 for 16 hours. Cells were lysed, lysates were split in half and proteins were immunoprecipitated with GFP antibody, proteins were separated by SDS-Page. Both Co-IPs were blotted for E6AP (A, B) and either for E6 (#1006, (A) or GFP (B). (C) Co-IP of E6AP with GFP antibody was quantified and normalized to E6 band intensity. (D) HeLa and CaSki cells were cultured with KTI-240, etoposide (ETO, 25 μM), or DMSO for 24 hours, cells were harvested and p53 protein analyzed by immunoblot and normalized to GAPDH protein expression. (F-K) SiHa, SiHa-C51S, and RPE-1 cells were cultured in KTI-240 for 16 hours and RNA was extracted and sequenced from three independent experiments. (E) Parallel cell cultures were analyzed for p53 protein. Differentially expressed genes identified using DESeq2 analysis with a FDR<0.01 and FC>2. Volcano plots show significant transcriptional increases (right, red) and decreases (left, blue) comparing vehicle vs. KTI-240 treated cells. HPV-16 E6 and E7 mRNAs were detected and did not differ between WT and C51S-SiHa cells. (I, K) TP53 mRNA was unchanged following KTI-240 treatment. Increased expression of p53 regulated genes occurred only in wild-type SiHa but not C51S or RPE-1 cells (I, J, K) p53 protein expression normalized to GAPDH and expressed as FC. Data expressed as S.E.M and analyzed with one-way ANOVA with Dunnett’s post hoc analysis (* p<0.05; n=3).

To assess the specificity of this effect, we examined p53 induction in other cell lines harboring HPV types. HeLa cells (HPV18-positive), which express an E6 protein lacking an analogous cysteine residue at its E6AP binding surface, and human cervical cancer cells CaSki. These were cultured with increasing concentrations of KTI-240 (Figure 2D; Supplementary Figure 3F). KTI-240 failed to induce p53 accumulation in HeLa cells, whereas it induced p53 in a dose-dependent manner in CaSki cells. Collectively, these data demonstrate that KTI-240 disrupts HPV16 E6•E6AP binding and reactivates p53 in a cysteine 51-dependent manner, supporting a mechanism in which these covalent inhibitors selectively target HPV16 E6.

Transcriptomic profiling following E6 inhibition

We next characterized the transcriptional changes in HPV E6/E7 expressing tumor cells following E6 inhibition. Elevated levels of p53 protein are predicted to alter expression of its transcriptionally regulated target genes. While C51S mutated SiHa cells did not show endogenous p53 induction following incubation in media containing KTI-218 and KTI-240, it is possible that some functions of E6, especially those pathways mediated by other binding partners of E6, could manifest and be differentiated from downstream effects of p53. The C-terminus of E6 binds to several PDZ proteins, and although this domain is distinct from the E6 region targeted by our compounds, its interactions might also be affected. RNA from compound treated RPE-1 cells was included in these transcriptome studies to disclose potential off-target consequences in the absence of HPV E6. In this set of experiments, SiHa, SiHa C51S, and RPE-1 cells were cultured in 5 μM KTI-240 or DMSO for 16 hours. This time point was selected to avoid accumulation of non-viable cells, and we confirmed p21 protein levels were elevated within this time frame (Figure 4E). Total RNA was isolated from three independent experiments and was subjected to quality control analysis. Each sample had a RIN of >9 and mRNA was sequenced with an average depth of ~50×10⁶ reads per sample. Parallel lysates were analyzed by Western blot and revealed a robust 2.2±0.2 fold induction of p53 (Figure 2E; Supplementary Figure 4A).

Figure 4. Validation key pathways in HPV oncogenesis.

SiHa, SiHa-C51S, and RPE-1 cells were cultured in KTI-240, DMSO or etoposide (ETO) for 16 hours, and proteins subjected to Western blot. A) KTI-240 increased p21 levels without affecting p16. (B) E6AP and MDM2 protein levels were not changed after KTI-240 treatment. p53 included as a positive control. (C) KTI-240 decreased Foxm1 and Rb protein in SiHa and C51S cells but not RPE-1, while (D) Puma increased only in SiHa cells. (E, F) Heatmaps summarize detected changes at the mRNA and protein level, respectively. (G, H) E6 protein levels were significantly decreased in SiHa cells, while E7 was unchanged in both HPV+ cell lines. I) HPV+ cervical cancer cell lines, SiHa and CaSki, oral cancer derived UM-SCC-47 and UM-SCC-104, and RPE-1, human foreskin keratinocytes (HFK) were incubated with KTI-218 or DMSO for 24 h with cell viability analyzed by Calcein-AM assay. (J) RPE-1, SiHa and C51S cells analyzed for viability after KTI-218 treatment. (L, M) cell viability of SiHa, CaSki, SCC-47, SCC-104, SiHa-C51S, and RPE-1, HFK and normal oral keratinocytes assessed after KTI-240 or DMSO for 48 hours. Viability expressed as a percent change vs. DMSO. (K) HPV+ SiHa cells, C51S and RPE-1 cells incubated with DMSO, 5 μM KTI-240 or 25 μM etoposide for 48 hours and analyzed for cleaved Caspase 3 protein levels. (N, O) Cells were treated for 48 h with DMSO, KTI-240 (3 and 10 μM) or etoposide (2 μM) and apoptosis quantified. (P) Senescence was analyzed by a SA-β-gal assay after 48 h of KTI-240 (3 μM), etoposide (2 μM) or DMSO treatment. Annexin-FITC and β-Gal staining intensities were normalized to DMSO. Data expressed as S.E.M and analyzed with one-way ANOVA with Dunnett’s, Bonferroni post hoc analysis (K, O), and unpaired Student’s t-test (N) * p<0.05; n≥3.

The sequences of ~28,000 mRNAs were identified in SiHa cells. E6, E7 and p53 transcripts remained unchanged following exposure to KTI-240 (Figure 2I; Supplementary Table 1). With a false discovery rate (FDR) of <0.01, fold change (FC) of ≥2.0 and mean reads >5, expression of 197 genes increased, and 221 genes decreased in SiHa cells following exposure to KTI-240. Volcano plots are shown in Figure 2F and selected relevant genes are indicated in Figure 2I. Supplementary Table 1 lists the differentially expressed genes (DEGs). Expression of several p53 target genes such as p21(CDKN1A), MDM2, and Sestrin1 (SESN1) increased after KTI-240 treatment (Figure 2I). p53 transcription targets that mediate apoptosis including FAS, Trail-R2 (TNFRSF10B), and PUMA (BBC3) were upregulated. Some p53 responsive genes were modestly altered. For instance, the splicing factor Zmat3 increased with a FC of 1.5 at an FDR of <0.05 (28). The FOXM1 transcript decreased following KTI-240 treatment. FoxM1 is a pro-proliferative transcription factor that is repressed by p53 during the DNA damage response and is often over-expressed in cancers (29).

The basal RNA transcriptome in edited SiHa C51S was compared to that in SiHa. Most transcripts including E6, E7, and p53 were present at similar levels in both cell lines. There were transcriptional differences with SiHa cells as depicted in the volcano plot in Supplementary Figure 4B. Together, these amount to less than 0.08% of all captured mRNAs. Following incubation of C51S cells with KTI-240, 64 mRNAs increased and 109 decreased (Figure 2H; Supplementary Table 2). These 173 mRNAs may include changes resulting from the inhibition of p53-independent activities of E6. In contrast to parental SiHa, incubation with KTI-240 did not induce changes in p53 responsive transcripts (Figure 2K). This further confirms the compound’s reliance on Cys-51 for E6•E6AP mediated p53 degradation.

RNA transcriptome changes in RPE-1 after incubation with KTI-240 were also characterized (Figure 2G, Supplementary Table 3). E6 and E7 mRNAs were not detected and low levels of p53 transcripts were observed as in the SiHa cells (Figure 2J). Only 25 genes were dysregulated following exposure to KTI-240 and six DEGs (SLC7A11, VSIR, NMRAL2P, OSGIN1, DHRS3, GCLM) overlapped with the changes in SiHa and C51S cells (Figure 3A). While these genes could be responding to KTI-240, none were associated with p53 induction. These results demonstrate that KTI-240 induced very limited HPV independent effects.

Figure 3. KTI-240 affects cell cycle and p53 pathways limited to wild-type E6 expressing cells.

(A) Venn diagram of DEGs in SiHa, SiHa-C51S, and RPE-1 cells. 25 DEGs were identified after KTI-240 treatment in RPE-1 cells, of which six overlapped with SiHa and SiHa-C51S cells. SiHa had the most DEGs in response to KTI-240. (B, C) RNA seq data analyzed using Qiagen’s IPA. SiHa cells displayed the most affected pathways while no pathways were significantly altered in RPE-1 cells. B) Bubble map showing affected pathways are cell cycle associated and inhibited but not in C51S cells (C).

Signaling pathway analysis using Qiagen’s IPA software revealed reductions in cell cycle checkpoints, kinetochore metaphase signaling, mitotic prometaphase and metaphase pathways in SiHa cells post KTI-240 (Figures 3B,4B). The identified network highlights the role of CDKN1A (p21) in inhibiting cell cycle progression by decreasing entry into interphase and reducing cell viability, leading to increased cell death and necrosis (Supplementary Figure 5A,B). Similarly, while TP53 transcript levels were unchanged, IPA predicted a central role for p53 in this E6 inhibition network, influencing multiple pathways by increasing CDKN1A, IFNG, IL1A, IL1B, and TNF, while decreasing EP400 and RABL6. The observed increase in IL1A and IL1B in response to KTI-240 is consistent with the reported repression of these genes by HPV-16 E6 and E7 during various stages of HPV pathology including late stages of carcinogenesis (30). Altered expression of these inflammatory cytokines, which can increase apoptosis and cell death in tumor cell lines, likely contributes to apoptosis in SiHa cells. The network also revealed connections between necrosis and apoptosis.

In contrast, cell cycle and p53 pathway transcripts were unchanged in HPV-negative RPE-1 cells incubated with KTI-240 (Supplementary Figures 5A,B,C), and IPA did not detect a single significant pathway alteration. In contrast, a limited number of pathways were affected in C51S cells by KTI-240 exposure. The cell cycle checkpoint pathway was similarly inhibited as in wild-type SiHa cells (Figures 3C and 4D) albeit to a lesser degree, with fewer genes being affected, and p53 signaling pathway transcripts were not dysregulated in KTI-240-treated C51S cells (Supplementary Figure 6B,C), and neither p53 nor p21 was identified as major responders in the IPA analysis (Supplementary Figure 5C,D). A few pathways overlapped with low z-scores and involved fewer genes than in SiHa cells (Supplementary Figure 6A). We speculate that the transcriptional changes shared by wild-type and C51S SiHa cells are mediated by E6 interactions independent of the E6•E6AP•p53 complex. One such candidate is transcription factor ATF3 (31).

We compared levels of representative proteins involved in HPV-16 E6/E7 mediated carcinogenesis in SiHa, RPE-1, and C51S cells after 16 h in KTI-240 supplemented media. CDKN1A (p21), which increased by ~7-fold at the mRNA level (Figure 4E), showed a 3-fold elevation at the protein level in parental SiHa cells (Figures 5A,F; Supplementary Figure 7A) and remained unchanged at the mRNA or protein level in RPE-1 cells after KTI-240 treatment (Figures 5A,E,F). We also confirmed both E6 binding compounds increase levels of p21 in HPV16+ OPC cells (Figures 1J, K). In C51S cells, KTI-240 did not increase p21 protein levels (Figures 5A,F), although this mRNA was modestly elevated by 1.5 FC with an FDR of <0.05. In addition, p16 (CDKN2A) was unchanged at both the mRNA and protein level. Similarly, E6AP mRNA and protein expression remained at baseline after KTI-240 treatment in both SiHa cell lines (Figures 5B,E,F). This finding is similar to (32) but appears to differ from previous studies that were mechanistically and methodologically different (33). p53 activation stimulates MDM2 transcription which in turn induces p53 protein degradation in a negative feedback loop. Interestingly, although there was a 4-fold induction in MDM2 transcript levels in SiHa cells by KTI-240 (Figure 5E), this increase did not correspond with its protein levels (Figures 5B,F). FOXM1 mRNA was significantly downregulated in SiHa and C51S cells with a corresponding decrease at the protein level of ~50% but remained unchanged in RPE-1 cells (Figures 5C,E; Supplementary Figure 7E). Rb1 transcript levels were unaltered whereas Rb protein decreased in SiHa and C51S cells following KTI-240 exposure, but this was not observed in RPE-1 (Figures 5C,E,F). Both BBC3 (PUMA) mRNA and protein increased by 2.7 and 2.8-fold respectively in SiHa but were unchanged in RPE-1 and C51S cells (Figures 5D,E,F; Supplementary Figure 7G). HR HPV E6 induces hTERT expression and increase telomerase activity (34). KTI-240 decreased hTERT mRNA by 1.9-fold (FDR 0.01) in SiHa cells, whereas this was unchanged in C51S and RPE-1 cells. The expression levels of several miRNAs were altered in KTI-240 treated SiHa cells. For example, the long non-coding RNA MIR34AHG is a p53 regulated transcript that is dysregulated in HPV pathologies (35, 36) and showed a FC of 2.4 (FDR 0.01) only in the SiHa cells.

Figure 5. KTI-218 and KTI-240 inhibit growth of HPV-16 expressing human cervical and oropharyngeal tumor cells in vivo.

(A, B) Nu/Nu mice were injected subcutaneously with HPV-16+ SCC-UM-47 cancer cells. Intraperitoneal (IP) injections of 50 mg/kg KTI-218 (n=8, green) or vehicle (VH, n=7 black) began when tumors were ~50–100 mm³. Tumor size was measured by calipers and expressed as S.E.M. and analyzed using two-way ANOVA (p<0.01). (B) Representative mice and extracted tumors. (C) Nu/Nu mice were injected s.c. with HPV-16+ SiHa cancer cells stably expressing a luciferase construct and treated intraperitoneally with 50 mg/kg KTI-218 (n=5, green) or vehicle (VH, n=4 black). (D) Bioluminescence was measured by IVIS SpectrumCT optical imaging. (E) Representative tumors and tumor weights. (F) Nu/Nu mice injected with HPV-negative cervical cancer cell line C33a or (G) SiHa-C51S cells. IP injections of 50 mg/kg KTI-218. (H) Nu/Nu mice injected SC with SCC-UM-47 cells. IP injection of 50 mg/kg KTI-240 (n=10, green) or vehicle (VH, n=9 black). All injections initiated when tumors measured ~50–100 mm³. Data are expressed as S.E.M and were analyzed using two-way ANOVA; *p<0.05.

The stability of E6 protein depends on association with E6AP (33). We questioned whether compound binding to E6 would alter its levels, and indeed Western blots showed a significant reduction in HPV-16 E6 protein levels in SiHa cells but not in C51S cells after 16 h culture in KTI-240 (Figure 4G, Supplementary Figure 7H). While E6AP has also been reported to bind to and stabilize the E7 protein (37), E7 protein levels were unaltered after KTI-240 exposure (Figure 4H; Supplementary Figure 7I).

E6 binding compounds specifically reduce HPV cell viability

Two human cervical HPV16 cancer cell lines, SiHa and CaSki (ATCC^® CRL-1550^™), and two human HPV16 oropharyngeal cancer derived lines, UM-SCC-47 and UM-SCC-104, were cultured with increasing concentrations of KTI-218. These HPV16+ cells showed decreased viability with IC50 values per cell line ranging from 1–3 μM after 24 h of KTI-218 exposure, in contrast to primary human foreskin keratinocytes (HFK) and RPE-1, which showed a slight decrease in cell numbers at the highest concentrations of the E6 inhibitor (Figure 4I). Similarly, KTI-239, the non-covalent analog of KTI-218, did not reduce SiHa cell viability (Supplementary Figure 8A), supporting the requirement for covalent binding to E6 to drive cell death.

After cells were exposed to KTI-240 for 24 h, a significant but modest reduction in viability was observed in SiHa cells (Supplementary Figure 8B), while RPE-1 cells were unaffected. After 48 h of culture in KTI-240, the viability of cervical and oral HPV+ cells decreased with IC50s between 1–3 μM for SiHa, CaSki and SCC-104, while SCC-47 was slightly less sensitive with an IC50 of 7.4 μM. KTI-240 did not decrease viability of RPE-1 cells and slightly decreased viability of HFKs and normal oral keratinocytes (NOKs) at the highest concentrations (Figure 4L). Given the high induction of p53 in W12-E cells after KTI-218 and KTI-240 treatment, there was a corresponding reduction in cell viability (Supplementary Figure 8C), implying that inhibition of HPV-16 E6 in cells with episomal HPV genomes causes cell death. The transcriptomic data predicted the inhibition of cell cycle death pathways in C51S cells albeit to a lesser extent than in SiHa cells in response to KTI-240. Indeed, there was a small reduction in C51S cell viability in KTI-240 or KTI-218 supplemented media to a much lesser extent than in parental SiHa cells (Figures 5J,M).

RNA-seq data documented activation of the apoptotic cell death pathway in SiHa cells. KTI-240 specifically increased cleaved caspase 3 by 3.5-fold in SiHa cells but not in RPE-1 cells, consistent with the cell viability studies (Figures 5K). C51S SiHa showed a 1.6-fold increase in the number of apoptotic cells. To further investigate this finding, we measured cell apoptosis in the absence of drug. There was an ~15% rate of basal apoptosis in C51S cells compared to 5% in parental SiHa cells (Figure 4N; Supplementary Figure 9A). KTI-240 (3, 10 μM) treatment for 48 h significantly increased apoptosis in SiHa cells but only modestly in C51S cells (Figure 4O). Next, we investigated whether KTI-240 induces senescence in these three cell lines. Basal β-galactosidase staining was negligible in SiHa and RPE-1 cells, while SiHa C51S cells had significantly higher basal β-gal staining intensity than parental SiHa cells (Supplementary Figure 9B). Expression of SA-β-gal was induced in etoposide treated SiHa and C51S cell lines. KTI-240 increased this senescence marker in SiHa cells but not in RPE-1 and only modestly in C51S cells (Figure 4P; Supplementary Figure 9B,C).

Activity in vivo tumor xenotransplantation experiments

HPVs are species-specific and do not infect other animals including rodents. We decided to use authentic human cancer cell lines that express E6 and E7 from the native HPV promoter and which are known to be dependent on maintaining these viral proteins (38). Prior to this series of ‘proof of concept’ experiments, in vivo bioavailability of KTI-218 after intraperitoneal (IP) injection into C57BL/6 mice was measured. KTI-218 was administered at a dose of 50 mg/kg and plasma levels were measured over 24 h (Supplementary Figure 10A,B). KTI-218 showed an acceptable pharmacokinetic profile for in vivo studies. Initial testing was performed in a xenograft model using the oral HPV+ cancer cell line UM-SCC-47. Treatment of male Nu/Nu mice with 50 mg/kg KTI-218 or vehicle per day began ~ 7 days post-implantation when the tumors were 50–100 mm³. Mice were treated once daily for six days a week. Administration of KTI-218 significantly slowed tumor growth without affecting body weight (Figure 5A,B; Supplementary Figure 10C). To visualize viable tumor cells, we injected a subclone of the cervical HPV+ SiHa cell line that constitutively expresses luciferase (Luc) and exhibits growth properties like those of parental cells. Luciferase expression reflects the presence of viable tumor cells and was quantified by in vivo detection using an IVIS SpectrumCT scanner. This experiment demonstrated that some mice did not exhibit measurable luciferase activity at day 39, while others showed significantly less activity than vehicle treated animals (Figures 6C,D). Overall, the measured Luc activity correlated well with the caliper-measured tumor sizes. Body weight was not affected by the administration of KTI-218 (Supplementary Figure 10F). When animals reached endpoints, the mice were sacrificed, and the remaining nodules were extracted and weighed (Figure 5E). Tumors of drug-treated animals were significantly smaller and weighed less than those of vehicle-treated animals. To test whether any remaining cells developed resistance to the drug, we isolated cells from two tumors and propagated these in culture. These cells were treated with KTI-218 and found to be as sensitive as the injected SiHa cells, implying that persistence of tumor cells is not due to development of drug resistance. DNA from a small residual UM-SCC-47 tumor was sequenced and confirmed the E6 cysteine 51 codon was present. We also tested the efficacy of KTI-218 in male Nu/Nu mice transplanted with SiHa cells. Mice were treated with 50 mg/kg KTI-218 or vehicle per day beginning ~14 days post-implantation when the tumors were 50–100 mm³. KTI-218 also significantly slowed tumor growth in this model (Supplementary Figure 10D,E). In contrast, KTI-218 did not affect the tumor growth of HPV negative cervical cancer C33a cells transplanted into female nude mice (Figure 5F). Next, we tested the efficacy of KTI-240 on tumor growth of SCC-47 xenografts. When administered once daily at 25 mg/kg, inhibition of SCC-47 tumor growth was not observed. However, by increasing the dose of KTI-240 to 50 mg/kg/day, reduced tumor growth of SCC-47 xenografts was achieved (Figure 5H). 2 h after the final dose, mice were sacrificed and tumors analyzed for drug levels. KTI-240 was detectable in the tumor tissue with an average concentration of 1.5 μg/gm. We then inoculated nude mice with SiHa C51S cells, which formed tumors with growth like that of wild-type SiHa. C51S SiHa tumors enlarged comparably in mice treated with KTI-218 by daily IP injection or with vehicle alone (Figure 5G). DNA sequencing confirmed the presence of the serine codon in the tumor tissue. These results further support the importance of covalent bond formation with Cys51 in mediating inhibition of E6 activity in vivo.

Discussion

The HPV E6 oncoprotein is essential for maintenance of episomal viral replication in keratinocytes and, together with E7, is uniformly expressed in HPV-associated malignancies. Genetic depletion of E6 by siRNA or CRISPR-mediated targeting induces apoptosis and senescence in HPV-positive tumor cells, underscoring its requirement for tumor maintenance (39, 40). In contrast, expression of high-risk HPV E7 alone stabilizes and activates p53 (41–43), highlighting the necessity of E6-mediated p53 suppression in HPV-driven cancers. These observations establish E6 as an attractive antiviral and anticancer target that can selectively eliminate HPV-infected cells while minimizing the likelihood of resistance (44). Despite this strong rationale, prior efforts have failed to identify drug-like inhibitors capable of blocking E6 function in vivo (13, 45–50).

To develop the present strategy, we focused on a conserved cysteine residue positioned adjacent to the E6AP-binding pocket of HPV-16 E6, which engages alpha-helical LxxLL motif–containing proteins (16–21). This residue offers a unique opportunity for irreversible covalent targeting to disrupt E6 interactions with its cellular partners. Although many high-risk and other HPV types bind E6AP, only HPV-16 and a subset of oncogenic genotypes (HPV-35, −45, and −68) encode an analogously positioned cysteine. Consistent with this, KTI-240 did not induce p53 accumulation in HPV18-positive HeLa cells. While sequence variation among HPV-16 E6 isolates has been reported in benign and malignant lesions (51), such variants are rare and largely uncharacterized. To our knowledge, substitution at cysteine 51 has not been observed.

E6AP associates with E6 primarily through a helical LxxLL motif, but additional contacts contribute to its overall binding affinity (52), which has been reported to reach picomolar levels in vitro (53). Cryo-EM structures of HPV-16 E6 in complex with full-length E6AP and p53 reveal that much of the E6 surface was occluded (54–57), raising concerns about whether the LxxLL-binding pocket would be accessible to small molecules or amenable to competitive inhibition. By incorporating a reactive warhead designed to covalently modify cysteine 51, our compounds circumvent these limitations. A recent study described a peptide engineered to covalently bind this cysteine could disrupt E6–E6AP interactions in vitro (53). Our data show that KTI-240 inhibits complex formation between wild-type E6 and E6AP but not with the E6 C51S mutant protein. These results demonstrate that covalent inhibition of E6 is feasible in cells and produces robust antitumor activity in vivo.

Several non-mutually exclusive mechanisms might underlie this inhibition. Newly synthesized E6 may be particularly vulnerable to covalent modification prior to E6AP engagement. Alternatively, compound binding may impair LxxLL motif recognition while allowing partial E6AP association through secondary interfaces, yielding a complex that is structurally incompetent for p53 ubiquitination. Covalent modification could also interfere with E6 conformational rearrangements required for p53 binding (20, 22) or directly obstruct p53 recruitment or ubiquitin transfer. Although future studies are required to distinguish between these possibilities, our data clearly demonstrate that KTI-218 and KTI-240 selectively restore p53 stability and tumor suppressor activity in HPV-16 E6–expressing cells.

Consistent with prior reports that E6AP stabilizes E6 (33), we observed a reduction in E6 protein levels following compound treatment, without changes in E6 or E7 mRNA abundance or E6AP expression. These findings suggest that covalent modification destabilizes E6, potentially by disrupting productive E6–E6AP interactions. Although no atomic structure of E6 bound to a small-molecule inhibitor has been reported, our successful co-crystallization studies will be published separately.

Functionally, E6 inhibition resulted in robust restoration of p53 protein levels and transcriptional activity. Canonical p53 targets mediating cell cycle arrest and senescence, including CDKN1A/p21, were strongly induced, as were pro-apoptotic genes such as PUMA and FAS. Reduced cell viability in HPV-16–positive cells may also reflect suppression of telomerase activity, as E6-mediated induction of hTERT (34) was significantly diminished following treatment. Prior siRNA-based studies suggested that E6–E6AP interactions are required to prevent p21-driven senescence (32). Our covalent inhibitors appear to be more effective by simultaneously activating p53-dependent pathways of apoptosis and senescence.

KTI-240 induced minimal transcriptional changes in HPV-negative RPE-1 cells expressing wild-type p53. Moreover, CRISPR-engineered SiHa cells harboring a cysteine-to-serine substitution at position 51 retained intact E6•E6AP-mediated p53 degradation, failed to activate p53 transcriptional programs upon compound exposure, and exhibited only modest growth inhibition. These residual effects likely reflect the p53-independent functions of E6 mediated through LxxLL motif–containing partners or other interactions that contribute to cellular proliferation (58, 59). Structural considerations suggest that the compounds do not interfere with the C-terminal PDZ-binding motif of E6 (60, 61). The differences observed in the basal transcriptome of C51S cells likely reflect clonal variation rather than functional reprogramming. Importantly, a comparison with published datasets of E6 and E6/E7 overexpression revealed an inverse pattern of pathway regulation (62), further supporting the specificity of E6 inhibition (Supplementary Figure 11).

For in vivo studies, we used authentic human HPV-positive cancer cell lines that remain dependent on E6 and E7 expression 38, necessitating the use of immunodeficient mouse hosts. This limitation precludes the evaluation of immune-mediated effects, including potential restoration of innate immune signaling suppressed by E6 interactions with IRF-3 (63, 64). Although inducible transgenic mouse models expressing HPV oncogenes exist (65–67), these require prolonged latency and additional oncogenic events and may not accurately reproduce human E6–host protein interactions. In our xenograft models, tumor suppression was restricted to wild-type E6–expressing tumors and was completely absent in C51S derivatives. Residual tumor persistence likely reflects suboptimal drug delivery rather than resistance, as recovered cells remained drug-sensitive and retained the wild-type E6 sequence.

Although established cancer cell lines may not fully represent the genetic diversity of naturally arising tumors, our findings demonstrate that restoration of p53 function, loss of cell viability, and suppression of tumor growth are strictly dependent on the covalent modification of HPV-16 E6 at cysteine 51. Mutation of this single residue abolished all downstream p53 responses, providing definitive genetic validation of target engagement and excluding alternative mechanisms such as proteasome inhibition, DNA damage responses, or stress-kinase activation that would occur independently of E6 residue identity.

HPV-16 is the most prevalent high-risk genotype and accounts for approximately half of all cervical cancers worldwide, as well as the majority of HPV-associated oropharyngeal, anal, and genital malignancies (68). Although prophylactic vaccination is highly effective, it does not benefit individuals with established infections, and global vaccine coverage remains insufficient. Projections have estimated tens of millions of cervical cancer cases over the coming decades (69), underscoring the urgent need for therapeutic interventions. E6 inhibitors may be applicable across multiple disease contexts, including advanced cancers with poor outcomes, treatment de-escalation strategies for HPV-positive oropharyngeal cancer, and combination therapies that leverage restored p53 signaling to enhance sensitivity to radiation or chemotherapy. The robust induction of p53 in premalignant cervical dysplasia derived cells suggests that E6 inhibition may prevent malignant progression. Since persistence of HPV in basal keratinocytes depends on E6 expression, abrogation of this function would also be expected to interfere with viral maintenance (70, 71).

Materials and Methods

Cell Culture

SiHa (ATCC, HTB-35), C51S, CaSki (ATCC, CRL-1550), UM-SCC-47, UM-SCC-104, HeLa, C33a and RPE-1 (ATCC, CRL-4000) cells were cultured in Dulbecco’s modified Eagle Medium (DMEM) (Corning#10–013-CV), supplemented with 10% fetal bovine serum (FBS;) and 1% penicillin-streptomycin. N-TERT immortalized normal oral keratinocytes (NOKs) and de-identified human foreskin keratinocytes (HFK) were maintained in Keratinocyte-SFM (Gibco 2993836) with 1% penicillin-streptomycin. W12-E cells were grown in F-medium with J2–3T3 fibroblast feeders. Cells were regularly tested for mycoplasma (Lonza’s MycoAlert^® assay#LT07–118).

Generation of HPV16 E6 C51S missense mutation in SiHa cells

SiHa cells were engineered using CRISPR-Cas9 (72). using the guide sequence (HPV16_E6_C51S_G1:TATGTATAGTATATAGAGAT;HPV16_E6_C51S_G2:TATGCATAGTATATAGAGAT) and homology directed repair template (GCTGCAAACAACTATACATGATATAATA TTAGAATGTGTGTACTGCAAGCAACAGTTACTGCGACGTGAGGTATATGACTTTGCTTTTCGGGATTTATCAATTGTATACAGAGACGGGAATCCATATGCTGTATGTGATAAATGTTTAAAGTTTTATTCTAAAATTAGTGAGTA) (73). SiHa cells were electroporated with a plasmid encoding Cas9-P2A-GFP and a single guide RNA targeting E6 together with a protected homology-directed repair template. Correctly targeted clones were identified by PCR amplification and Sanger sequencing.

Luciferase and viability multiplex assay

SiHa and RPE-1 cells were transfected with p53-renilla-luciferase using polyethylenimine (PEI). Stable SiHa p53-Luc) and RPE-1 p53-Luc) cell lines were established by geneticin selection. These cells were seeded in complete DMEM in white plates for drug exposure studies. Media was replaced to DMEM with 1% FBS containing varying concentrations of KTI compounds or DMSO. The DMSO concentration was kept constant at 0.02% (v/v). Studies on HFKs, NOKs, and W12-E cells were performed in Keratinocyte-SFM supplemented with 1% FBS. After 24 or 48h, media was removed and 50 μl of PBS containing 2 μM Calcein-AM (calcein-acetoxymethyl ester; AAT Bioquest 22003) was added per well and incubated for 30 min at 37°C. Cell viability was determined by measuring fluorescence at excitation/emission wavelengths of 488 nm/520 nm. PBS containing Calcein-AM was removed and Renilla luciferase activities were measured using Dual-Glo^® Luciferase Assay (Promega E2920). Renilla luciferase activity was normalized to cell viability and data expressed as percent change over DMSO. Assays were performed at least in triplicate and repeated at least three independent times.

Co-immunoprecipitation assays

SiHa cells were transfected with eGFP-N1 plasmids encoding eGFP, eGFP-HPV16 E6 wild type, or eGFP-HPV16 E6 C51S mutant. 24 h post-transfection, cells were cultured in 5 μM KTI-240 or DMSO for 16h, lysed (50 mM HEPES pH 7.2, 300 mM NaCl, 0.5% NP-40, and 1 mM DTT, 1X SigmaFast EDTA free), and proteins were co-immunoprecipitated with anti-GFP antibody (Santa Cruz, #sc-9996) and Protein G–Sepharose beads. Beads were washed with lysis buffer and bound proteins were analyzed by Western blotting.

Western Blots

Harvested cells were lysed (10 mM Tris-HCl pH 8, 150 mM NaCl, 2% SDS, 1X protease inhibitor cocktail). Unless noted, 20 μg protein was loaded per lane (40 μg for cleaved caspase-3, 100 μg for HPV16 E6, 80 μg for HPV16 E7). Samples were run on 4–12% or 4–20% gels, transferred to PVDF, blocked in 5% milk, and probed with antibodies against Rb (Cell Signaling Technology, Inc (CST), 9309T), FoxM1 (CST#20459T), PUMA (CST#98672T), FasL (CST#68405T), p21 Waf1/Cip1 (CST#2947T), Cleaved Caspase-3 (CST#9661S), ZMAT3 (Proteintech 10504–1-AP), MDM2 (Proteintech 27883–1-AP), p53 (DO-1) (Santa Cruz sc-126), GAPDH (Santa Cruz sc-47724), HPV16 E6 (Arbor Vita #1006), HPV16 E7 (GeneTex GTX133411), E6AP (Bethyl A300–351A), or GFP tag Rabbit PolyAb (Proteintech#50430–2-AP) and HRP-conjugated secondary antibodies: anti-Rabbit IgG (Jackson#711-035-152), and anti-Mouse IgG (Jackson#115-035-174;#715-035-150). Proteins were detected using SuperSignal^™ West Femto (low expression; ThermoFisher Scientific#1859022) or ECL (high expression, including GAPDH; Cytiva #RPN2232). Signals were imaged on a ChemiDoc system and quantified with ImageJ. Experiments were performed ≥3 times, with GAPDH as a loading control.

siRNA-mediated knockdown of E6AP and E6/E7

Stable SiHa and RPE-1 luciferase cells were seeded in white 96-well plates, followed by transfection with 10 nM siRNA (control, E6–1, E6–2, E6AP, and E6/E7) and RNAiMax. Luc activity wasmeasured 48 h and 72 h after transfection. Parental SiHa and C51S cells were seeded at a density of 1.25×10⁵ cells/well in 12-well plates containing complete DMEM supplemented with 10% FBS. After 24 h, cells were transfected with siRNA targeting E6AP (AmbionSilencerSelect, siRNA#IDs14604), HPV16 E6E7 (74), E6–1 (5′-AGCAAAGACAUCUGGACAA-3′), E6–2 (5′-AAAAGCAAAGUCAUAUACCUC-3′) or scrambled siRNA (Ambion SilencerSelect Negative Control #1) at a final concentration of 15 nM using Lipofectamine RNAiMax (Life Sciences Technologies#13778030). After 72h, cells were harvested in lysis buffer.

Senescence and Apoptosis Assays

Cells (40,000/well, 24-well plates) were treated with varying doses of KTI-240 or etoposide (2 μM; Sigma E1383) for 48 h in DMEM + 1% FBS, with DMSO (0.02%) as control. Senescence was assessed by β-Gal staining (overnight, 37°C, APExBIO#K2185) and apoptosis using the Biotia kit (#30018). Images were captured at 20× using Olympus IX73 and ECHO fluorescence microscopes. Senescent and apoptotic cells were quantified in ImageJ (8-bit conversion, thresholding, binarization, watershed, and particle analysis) (75). Images were tinted yellow for visualization. Experiments were performed in three independent biological replicates with technical duplicates.

RNA isolation and transcriptome analysis

SiHa, SiHa C51S, and RPE-1 cells (400,000/well, 6-well plates) were treated with DMSO (0.02%) or 5 μM KTI-240 in 1% FBS for 16 h. Cells were collected for RNA (RNeasy Mini Kit (Qiagen#74104), DNase (Qiagen#79254)) and protein (western blot) analyses. Experiments were performed in triplicate. RNA quality (Agilent TapeStation) showed RIN 9.3–10.

Total RNA (100ng) was used for library preparation using the Illumina Stranded mRNA Prep, Ligation kit (Illumina), following the manufacturer’s instruction. Each resulting uniquely dual-indexed library was quantified and quality assessed using Qubit and Agilent TapeStation, and multiple libraries were pooled in equal molarity. The pooled libraries were sequenced with 2×100bp paired-end configuration on an Illumina NovaSeq X PLUS sequencer.

Sequencing reads were quality checked using FastQC (v.0.11.5, Babraham Bioinformatics, Cambridge, UK). The sequence data were then mapped to the human reference genome hg38 and thuman papillomavirus type 16 (NC_001526.4) using the RNA-seq aligner STAR (v.2.7.10a, (76)) with the following parameter: “--outSAMmapqUnique 60”. To evaluate the quality of RNA-seq data, the number of reads that fell into different annotated regions (exonic, intronic, splicing junction, intergenic, promoter, UTR, etc.) of the reference genome was assessed using bamutils (from ngsutils v.0.4.17, (77)). Uniquely mapped reads were used to quantify gene expression using featureCounts (subread v.2.0.3 (78); with the following parameters: “-s 2 -Q 10”. Data were TMM(trimmed mean of M values)-normalized and analyzed in Partek Flow; features ≤ 5.0 were excluded. Differential expression used DESeq2 (median ratio), with significance defined as FC≥2, FDR <0.01 and mean expression ≥5. Pathway analysis was performed using Qiagen Ingenuity Pathway Analysis (IPA) platform with the same cutoffs. Dysregulated signaling pathways were analyzed using the. Data are available under GEO accession: GSE308563.

Xenograft studies in mice

All animal experiments were performed following the approval of Indiana University IACUC and animal care policies. Six-week-old athymic nude mice were from Envigo (Inotiv Hsd:Athymic Nude-Foxn1nu (069)). Mice were bred in-house and homozygous nude mice were used in the xenograft studies. 6–9-week-old mice were injected subcutaneously in the flank with SiHa cells (2×10^6 cells/mouse), SiHa-luc (1×10^6 cells/mouse), SCC-47 (2×10^6 cells/mouse), SiHa C51S (1×10^6 cells/mouse), or C33a (2×10^6 cells/mouse). Tumors were measured twice a week with calipers and body weight was recorded. Tumor volume was calculated using the formula V = 0.5 × length × width². Unless otherwise stated, drug administration began at a size of 50–100 mm³. Mice bearing established tumors were randomized into treatment groups when tumor volumes were comparable across groups. KTI-218 (50 mg/kg) was administered once daily via IP administration in 5% (v/v) NMP, 20% (v/v) Kolliphor RH40, 20% (v/v) PEG400, 55% (v/v) MilliQ-H₂O. This vehicle was used as a control for all KTI-218 studies. KTI-240 (25 mg/kg) was administered once or twice daily via IP in 5% (v/v) NMP, 20% (v/v) Kolliphor HS15, 75% (v/v) MilliQ-H₂O, which was used as the vehicle control for KTI-240 experiments. Mice were treated six days a week. Bioluminescence (BLI) of SiHa-luc xenografts was measured on days 8 and 39 after cell injection. D-Luciferin (VivoGlo^™ Luciferin, P1042) in PBS (200mg/10ml) was administered subcutaneously (150 mg/kg). Animals were anesthetized using isoflurane. Images were acquired over 30 min with the IVIS SpectrumCT optical imaging system with 15 images taken within 2 min delay and a 2 min exposure. After background subtraction, the area under the curve (AUC) and Cmax were calculated.

Pharmacokinetic study

Bioavailability of KTI-218 after a single IP dose of KTI-218 in plasma was investigated in male non-fasting C57BL/6 mice. Mice were euthanized and plasma was collected from three animals each at 0.08, 0.25, 0.5, 1, 2, 4, 8, 12, 16, 24h. KTI-218 levels in plasma were analyzed by HPLC-MS/MS.

Statistical analysis

Densitometric protein analysis (ImageJ) was normalized to GAPDH intensities. Fluorescence and β-Gal staining intensities were normalized to DMSO and expressed as fold-change. Data were analyzed using one-way ANOVA (Dunnett’s/Bonferroni) or unpaired t-test; dose–response curves used nonlinear regression, and tumor growth used two-way ANOVA. Experiments were performed ≥3 independent times (n≥3; * P<0.05) and reported as SEM.

Significance Statement.

Human papillomaviruses cause cervical, oropharyngeal (OPC), and anogenital cancers. Viral replication and oncogenic transformation are dependent on the E6 protein, which binds to multiple host proteins including the ubiquitin ligase E6AP (UBA3) and the tumor suppressor p53. This leads to p53 degradation resulting in very low levels in tumor cells. We identify and characterize novel compounds that covalently and irreversibly bind to and inactivate E6, thereby restoring p53 and activating its downstream transcriptional programs to induce cell death and senescence. Treatment of mice transplanted with human derived cervical and OPC HPV cancers suppresses tumor growth in vivo. This work presents a promising strategy to overcome a longstanding challenge in targeting viral oncoproteins.