Activating Mutations in Pik3ca Contribute to Anal Carcinogenesis in the Presence or Absence of HPV-16 Oncogenes

Anal cancer is a relatively rare cancer in humans accounting for approximately 0.5% of all new cancer cases in United States. Anal cancer incidence and deaths have been increasing over the last decade (1). The most common etiologic factor in anal cancer is high-risk human papillomavirus (HPV), implicated in other anogenital cancers and a growing fraction of head and neck cancers. HPV is detected in 95% of human anal cancers with HPV16 accounting for 89% of HPV positivity (2). Previously, we established a mouse model for human anal cancer using HPV16 E6/E7 transgenic mice treated topically with DMBA (7,12-Dimethylbenz(a)anthracene; ref. 3). Studies using this animal model led to the discoveries that HPV16 E7 is the more potent oncogene in anal carcinogenesis (4) and that the mTOR (“mechanistic target of rapamycin” or “mammalian target of rapamycin”) signaling pathway is activated in anal squamous cell carcinoma arising in HPV16 transgenic mice as well as in human anal cancers (5). mTOR is a serine/threonine protein kinase commonly activated downstream of the PI3K signaling pathway (6). mTOR forms two distinct complexes, mTORC1 and mTORC2, that regulate multiple cellular processes required for cell growth and metabolism (7). mTOR also acts as a downstream regulator of the PI3K–AKT signaling pathway; therefore mTOR signaling and its downstream pathway are also regulated by the PI3K/AKT signaling pathway (8). Alterations in PI3K/AKT/mTOR signaling caused by hyperactivation of PI3K are thought to contribute to cancers, including HPV-associated cancers (9–13). Recently, mutations that lead to increased PI3K signaling were found in 36% of 12 types of human cancers (14). Of these mutations causing activation of PI3K signaling, the most prevalent mutations occurred in the Pik3ca gene that encodes p110α, the catalytic subunit of PI3K. Three hot-spot mutations in the Pik3ca gene, E542K and E545K (exon 9) in the helical domain and H1047R/L (exon 20) in the kinase domain of p110α, lead to the constitutive activation of PI3K signaling independent of the PI3K regulatory subunit p85. Such mutations in Pik3ca arise in approximately 20% of human anal cancers, 90% of which are in exon 9 (15). The importance of these mutations in anal cancer has not been determined.

Here, we asked if such activating mutations in Pik3ca contribute to anal carcinogenesis in the context of mice expressing or lacking HPV16 oncogenes, E6 and E7. We generated mice expressing a mutant form of Pik3ca, either Pik3ca^H1047R or Pik3ca^E545K, ± the HPV16 oncogenes, E6 and E7, and treated them topically in the anus with the chemical carcinogen DMBA. We found these activating mutations in Pik3ca to drive anal carcinogenesis, and their oncogenic potential was synergistic with HPV16 E6 and E7, even in the absence of DMBA. We also found that tumor spheres and tumor grafts derived from these anal tumors were inhibited in their growth by TAK-228 (also known as sapanisertib, INK-128 and MLN0128), an investigational drug that inhibits both TORC1 and TORC2 (16). We conclude that activating mutations in Pik3ca contribute to anal carcinogenesis and that this pathway represents an important new target for therapeutic intervention.

K14E6, K14E7, KRT14-cre/Esr1 (K14CreERtm) mice have been described previously (17, 18). R26-Pik3ca^H1047R (heretofore referred to as Pik3ca^H1047R) and R26-Pik3ca^E545K (heretofore referred to as Pik3ca^E454K) mice were kindly provided by Dr. Dustin A Deming and described previously (The Jackson Laboratory; Stock Number 016977; ref. 19). Details on the breeding schemes used to generate the various strains of mice used in this study, an ethics statement, and methods used to induce expression of the Pik3ca mutants and treat mice with DMBA are provided in Supplementary Information.

Anal tumors were resected, rinsed with sterile PBS, and placed in a chelation buffer on ice (20) for one hour. The tumor tissue was then digested with collagenase and dispase at 37°C. Cells were pelleted and the supernatant discarded. The pellet was then re-suspended in advanced DMEM/F12 (ADF, Invitrogen) and the resulting cell suspension was combined with Matrigel at a 1:1 ratio. The suspension was plated by placing 50 μL droplets into wells of a 24-well culture plate and incubated for 15 minutes at 37°C. The droplets were then covered with feeding medium consisting of ADF supplemented with murine epidermal growth factor (EGF, Invitrogen) to a final concentration of 50 ng/mL. Spheroids were passaged at least once before therapeutic investigations. Spheroids beyond 9 passages were not utilized. Therapeutic investigations were performed by exchanging feeding media containing the desired concentration of the agent over the spheroids suspended in Matrigel. An investigational oral dual TORC1/2 inhibitor, TAK-228 (#I-3344 LC Labs) was dissolved in dimethyl sulfoxide to make a 10 mmol/L stock. TAK-228 was then diluted 1:1,000 in culture medium before diluting to the indicated final concentration. Spheroids were treated for 48 hours. To measure changes in cellular proliferation, spheroid diameters were measured using 4X bright-field microscopy at baseline and post-treatment. For western blot analysis to assess the efficacy of the drug on inhibiting the PI3K/AKT/mTOR signaling pathway, spheroids were recovered 24 hours post-treatment of TAK-228. Details on the staining of spheroids with hematoxylin and eosin (H&E) are provided in Supplementary Information.

Anal tumors from 4-OHT treated K14E6/E7/Pik3ca^E545K: K14CreERtm mice were rubbed with 10% providone-iodine (CareFusion) and/or isopropyl alcohol, rinsed with sterile phosphate buffered saline (PBS), and excised with sterile instruments. Tumor tissue was minced in sterile PBS and then combined with reduced growth factor Matrigel (BD Matrigel Basement Membrane Matrix, Growth Factor Reduced (GFR) Catalog no. 354230) at a 1:1 ratio. 200 μL of minced tumor mixed with Matrigel was injected subcutaneously into immunodeficient (NOD scid gamma) mice in their flanks (initial graft set as passage zero or P0). Tumors were passaged by injecting 200 μL of minced tumor pieces in PBS mixed 1:1 with Matrigel. Details on treatment of tumor grafts are provided in Supplementary Information.

Primary tissue and tumor grafts were harvested, snap frozen, briefly sonicated in cold RIPA buffer with protease and phosphatase inhibitors, agitated 30 minutes at 4°C, and spun for 30 minutes at 12,000 rpm in a microfuge. Supernatant was collected, and protein concentration determined using the Bradford assay. Spheroid cultures were collected following 24 hours culture in the presence or absence of TAK-228, snap-frozen, protein extracted, and immunoblotting performed as previously described (21). Details on the choice of antibodies used are provided in Supplementary Information.

Immunohistochemistry was performed as previously described (17, 22). Details on the choice of antibodies/reagents used are provided in Supplementary Information.

A Log-rank test was used to determine the significance of differences in the onset of anal tumors between groups. A two-sided Fisher's exact test was used to determine the significance of differences in the incidence of anal cancer between groups. To determine the significance of differences in the severity of disease, DNA synthesis level, growth change in spheroids and growth rate in tumor graft between groups, two-sided Wilcoxon rank sum test was performed. All statistical analyses were performed using MSTAT statistical software version 6.1.4 (http://www.mcardle.wisc.edu/mstat).

To determine whether activating mutations in Pik3ca found in human anal cancers contribute to anal carcinogenesis, we generated mice that conditionally express mutant forms of Pik3ca in anal, stratified squamous epithelium. We first used R26-Pik3ca^H1047R (Pik3ca^H1047R) mice, carrying a ROSA 26 locus knock-in allele of Pik3ca^H1047R positioned downstream of a floxed transcriptional STOP cassette upstream (23). To generate mice expressing this mutant form of Pik3ca in anal epithelium, K14CreERtm transgenic mice expressing the tamoxifen (or 4-hydroxy-tamoxifen: 4-OHT)-inducible CreERtm from the keratin 14 (K14) promoter were bred to the Pik3ca^H1047R mice. To assess the efficiency of expression of the mutant form of Pik3ca, p110α^H1047R, through inducible Cre-recombination in anal epithelium from Pik3ca^H1047R: K14CreERtm and K14E6/E7/Pik3ca^H1047R:K14CreERtm mice, we performed western blot analysis to monitor levels of phospho S6 Ribosomal Protein (p-S6) in lysates from the anoderm of these mice. These mice had been treated with 4-OHT 5 consecutive days to activate Cre and then anal tissues were collected at 6 weeks post-treatment. S6 Ribosomal Protein (S6) is a component of the 40s ribosomal subunit and also is known to be phosphorylated by activation of the PI3K–AKT–mTOR signaling pathway. As previously described, activating mutations of Pik3ca lead to an increase in phosphorylation of AKT and subsequent phosphorylation of S6 and 4E-BP1 (24, 25). We have previously demonstrated that the HPV16 E6 and E7 oncogenes cause increased pS6 in the squamous cell carcinomas of mouse anus (5). In lysates from both Pik3ca^H1047R:K14CreERtm and K14E6/E7/Pik3ca^H1047R, phosphorylation of S6 was increased over that seen in Pik3ca^H1047R mice (Fig. 1A and B). Not surprisingly, phosphorylation of S6 was more significantly increased in K14E6/E7/Pik3ca^H1047R:K14CreERtm mice compared with the other groups, Pik3ca^H1047R:K14CreERtm and K14E6/E7/Pik3ca^H1047R. A similar finding was observed when we analyzed levels of pS6 in these same tissues by immunofluorescence (Fig. 1C).

We then assessed the susceptibility of these mice to anal carcinogenesis using our previously established model, in which mice are treated with DMBA (0.12 μmol/week) for 20 weeks and then held an additional 8 weeks before sacrifice (5). Using this experimental design, we observed that the activating mutation in Pik3ca present in the Pik3ca^H1047R:K14CreERtm mice caused anal cancers (Fig. 2A). Tamoxifen was first administered by intraperitoneal injection to systemically induce Cre-recombination thereby turning on expression of the mutant Pik3ca gene. Then we administered DMBA to the anal canal once a week for 20 weeks. All (n = 14) of the Pik3ca^H1047R:K14CreERtm mice developed overt anal tumors with an average onset of 10.3 weeks after initial DMBA treatment, whereas none of the nontransgenic mice developed any overt tumors at this time point (Fig. 2A). These observations indicate that the mutant form of Pik3ca, p110α^H1047R, leads to increased susceptibility to anal tumors.

Next, we monitored the tumorigenic potential of the mutants of Pik3ca on anal carcinogenesis in context of HPV16 E6/E7 transgenic mice. Here, we used both Pik3ca^E545K mice as well as Pik3ca^H1047R mice. K14CreERtm mice were bred to either Pik3ca^H1047R or Pik3ca^E545K mice and the resulting Pik3ca^H1047R: K14CreERtm or Pik3ca^E545K:K14CreERtm mice were crossed to K14E6 and K14E7 mice. We generated 7 different groups of mice with the following genotypes: Pik3ca^H1047R, Pik3ca^E545K, Pik3ca^H1047R:K14CreERtm, Pik3ca^E545K:K14CreERtm, K14E6/E7/Pik3ca^H1047R, K14E6/E7/Pik3ca^H1047R:K14CreERtm, and K14E6/E7/Pik3ca^E545K:K14CreERtm.

To assess the tumorigenic contributions of expressing p110α^H1047R or p110α^E545K in combination with HPV16 E6/E7 oncoproteins, we modified our prior protocol in terms of timing and method of delivery of tamoxifen to induce Cre-recombination and the duration of administration with DMBA. Using our original protocol (5) with systemically induced expression of p110α^H1047R, we observed not only the early onset of tumors (Fig. 2A), but also morbidity issues (tumors arising in other epithelial tissues, edema) in Pik3ca^H1047R:K14CreERtm (data not shown). Therefore, instead of intraperitoneal injection with tamoxifen, the anal canal was topically treated with 4-hydroxy-tamoxifen (4-OHT) to avoid those morbidities. Also, we applied DMBA for only 5 weeks, anticipating that the onset of tumorigenesis would be accelerated compared to that seen in the absence of the expression of the HPV16 oncogenes (Fig. 2B). None of Pik3ca^H1047R or Pik3ca^E545K mice developed overt anal tumors reflective of the short duration of DMBA treatment (Fig. 2B). In the K14E6/E7/Pik3ca^H1047R group, 11 out of 17 mice (64.7%) had overt tumors and these tumors arose between 11.5 and 26 weeks after initial administration with DMBA. 17 out of 21 Pik3ca^H1047R:K14CreERtm mice (80.8%) had overt anal tumors with onset of anal tumors between 7 and 20.5 weeks. The onset of overt tumors in Pik3ca^H1047R:K14CreERtm mice was significantly earlier compared with that observed in K14E6/E7/Pik3ca^H1047R mice [Fig. 2B, log-rank sum test P (one-sided) = 0.0231], though the incidence of overt anal tumor was not significantly different between the two groups. In Pik3ca^E545K:K14CreERtm mice, 8 out of 12 mice (66.7%) developed overt anal tumors with onset of tumors between 8.5 and 23 weeks. These results further demonstrate the oncogenic potential of these activating mutations in Pik3ca contributes to anal carcinogenesis in the absence of HPV oncogenes, even when the duration of treatment with DMBA was reduced. Interestingly, both K14E6/E7/Pik3ca^H1047R:K14CreERtm and K14E6/E7/Pik3ca^E545K:K14CreERtm mice developed overt tumors significantly earlier compared with that seen in Pik3ca^H1047R:K14CreERtm [Fig. 2B, log-rank sum test P(two-sided) = 5.83 × 10⁻⁶], Pik3ca^E545K:K14CreERtm [P(two-sided) = 5.42 × 10⁻⁶] mice or K14E6/E7/Pik3ca^H1047R mice [P (two-sided) = 6.21 × 10⁻⁷, 5.92 × 10⁻⁷, respectively]. The overt tumors in the K14E6/E7/Pik3ca^H1047R:K14CreERtm group arose between 5.5 and 9.5 weeks with an average age of onset of 6.7 weeks after DMBA treatment began. Similarly, K14E6/E7/Pik3ca^E545K:K14CreERtm mice developed overt tumors between 5.5 and 11 weeks with an average onset of 7.0 weeks after DMBA treatment began. These observations indicate that both activating mutations in Pik3ca strongly drive anal tumorigenesis in the presence of HPV16 E6/E7, and that they show synergistic effects with the viral oncogenes on anal tumorigenesis.

To learn temporally how an activating mutation in Pik3ca contributes to anal carcinogenesis, we monitored the oncogenic potential of delayed expression of Pik3ca^H1047R in the context of HPV16 E6/E7 mice. Both Pik3ca^H1047R:K14CreERtm and K14E6/E7:Pik3ca^H1047R:K14CreERtm mice were administered DMBA to their anus for 5 weeks. 20 weeks post-administration of DMBA, 4-OHT was topically applied to the anus for 5 consecutive days. All of K14E6/E7/Pik3ca^H1047R:K14CreERtm mice developed anal tumors before 24 weeks (Fig. 2C); whereas only 4 out of 12 Pik3ca^H1047R:K14CreERtm mice (33.4%) developed overt anal tumors. K14E6/E7/Pik3ca^H1047R:K14CreERtm mice developed anal tumors significantly earlier compared with K14E6/E7/Pik3ca^H1047R mice or Pik3ca^H1047R:K14CreERtm mice [Fig. 2C, log-rank sum test P (two-sided) = 0.045, 8.06 × 10⁻⁶, respectively]. These observations indicate that an activating mutation in Pik3ca accelerates anal carcinogenesis in the presence of HPV16 oncogenes, even when onset of its expression is delayed.

HPV16 E6/E7 mice do not develop anal cancers without administration of DMBA (4). We asked if an activating mutation in Pik3ca, in the context of expression of these HPV16 oncogenes, drives anal carcinogenesis without treatment with DMBA. Pik3ca^H1047R:K14CreERtm and K14E6/E7/Pik3ca^H1047R:K14CreERtm mice were treated topically with 4-OHT on the anus for 5 consecutive days, but not treated with DMBA. After 28 weeks, anal tissues were recovered and histologically evaluated to determine the extent of neoplasia. Strikingly, 5 out of 10 K14E6/E7/Pik3ca^H1047R:K14CreERtm mice developed overt anal tumors without DMBA treatment; whereas, none of the Pik3ca^H1047R:K14CreERtm mice developed tumors (Fig. 2D). As seen previously with K14E6/E7 mice (4), the K14E6/E7/Pik3ca^H1047R mice did not develop anal tumors in the absence of DMBA treatment. The onset of overt tumors in K14E6/E7/Pik3ca^H1047R:K14CreERtm mice was significantly different than that observed in Pik3ca^H1047R:K14CreERtm (Fig. 2D, log-rank sum test P (two-sided) = 0.0056) or in K14E6/E7/Pik3ca^H1047R mice. These observations indicate that expression of an activating mutant form of Pik3ca in the presence of HPV16 oncogenes is sufficient to drive anal tumorigenesis in mice.

To determine whether the mutant forms of Pik3ca cause an increased severity of neoplastic disease in the anal region, we performed detailed histopathological analysis of the tissue harvested from 4-OHT and DMBA-treated mice. Tissues from the mice of each genotype were harvested, fixed, paraffin-embedded, and sectioned, and then the sections stained with H&E and subjected to detailed histopathologic analysis (Table 1). None of Pik3ca^H1047R mice developed a malignant disease, but did develop low-grade dysplasia. In contrast, 7 out of 17 K14E6/E7/Pik3ca^H1047R mice (41.1%) had malignant anal tumors. The severity of disease (in which we give a score from 1 to 7 for the worst grade of disease in each mouse, with normal histology given a score of 1, low-grade dysplasia a score of 2, high-grade dysplasia a score of 3, verrucous carcinoma a score of 4, and grades 1–3 squamous cell carcinoma scores of 5–7, respectively) in this genotype was significantly worse than in Pik3ca^H1047R mice. In the Pik3ca^H1047R:K14CreERtm group, 14 out of 21 mice (66.7%) developed anal cancer. The incidence of cancer and severity of disease were significantly greater than for the Pik3ca^H1047R mice, but were not significantly different from the K14E6/E7/Pik3ca^H1047R mice. Interestingly, most of the cancers (12 out of 14) in Pik3ca^H1047R:K14CreERtm mice were verrucous carcinoma, which is an uncommon variant of squamous cell carcinoma known as a very well-differentiated squamous malignancy with minimal atypia. In Pik3ca^E545K:K14CreERtm mice, 4 out of 9 mice (44.4%) developed anal cancer, with 2 out of these 4 mice having verrucous carcinoma. Consistent with the observation of overt tumors in Fig. 2B, all K14E6/E7/Pik3ca^H1047R:K14CreERtm mice had anal cancers, with half developing poorly differentiated invasive squamous cell carcinomas. Incidence of cancer and severity of disease in K14E6/E7/Pik3ca^H1047R:K14CreERtm mice were significantly greater than in the Pik3ca^H1047R:K14CreERtm and K14E6/E7/Pik3ca^H1047R mice. Similarly, 12 out of 13 K14E6/E7/Pik3ca^E545K:K14CreERtm mice developed malignant tumors. These data indicate that activating mutations in Pik3ca and expression of HPV16 E6/E7 synergize to increase the severity of anal carcinogenesis.

We observed anal cancers from five genotypes: Pik3ca^H1047R:K14CreERtm, Pik3ca^E545K:K14CreERtm, K14E6/E7/Pik3ca^H1047R, K14E6/E7/Pik3ca^H1047R:K14CreERtm and K14E6/E7/Pik3ca^E545K:K14CreERtm. As shown in Table 1, the majority of anal cancers in Pik3ca^H1047R:K14CreERtm mice were verrucous carcinomas (12 out of 14 mice, 85.7%). Most verrucous carcinomas arose around the anus and were observed as overt, exophytic tumors. By histopathological analyses, 15 out of 16 tumors (93.8%) developed below the anorectal junction in Pik3ca^H1047R:K14CreERtm mice (Supplementary Fig. S1). In biomarker studies, MCM7 was strongly expressed in cancer from Pik3ca^H1047R:K14CreERtm mice, but its expression was restricted to the basal and parabasal layer cells (Supplementary Fig. S2, third). As expected, p-S6 was robustly detected in the cytoplasm of anal cancer cells. Similarly, half of the tumors (2 of 4) in Pik3ca^E545K:K14CreERtm mice were verrucous carcinoma and all of these tumors developed from below the recto-anal junction but still within the anus. In contrast, 3 out of 7 K14E6/E7/Pik3ca^H1047R mice (42.9%) were scored to have squamous cell carcinoma, with 4 out of 9 tumors (44.5%) arising at the anorectal junction or above the junction, where HPV-associated anal cancer in human patients is commonly found (Supplementary Fig. S1). In K14E6/E7/Pik3ca^H1047R mice, the expression of MCM7 was highly upregulated in suprabasal layer cells as well as basal cells (Supplementary Fig. S2, third). Consistent with the observation in epithelium (Fig. 1C), p-S6 was detected in the cytoplasm of cancer cells (Supplementary Fig. S2, fourth). The cancer phenotypes from K14E6/E7/Pik3ca^H1047R:K14CreERtm and K14E6/E7/Pik3ca^E545K:K14CreERtm mice were more similar to that observed in the K14E6/E7/Pik3ca^H1047R than in the Pik3ca^H1047R:K14CreERtm mice. Unlike the anal cancers in Pik3ca^H1047R:K14CreERtm mice, 12 out of 14 K14E6/E7/Pik3ca^H1047R:K14CreERtm mice (85.7%) developed invasive squamous cell carcinoma, and about half of the cancers (18 out of 35, 51.4%) originated at the anorectal junction or above the junction (Supplementary Fig. S1). Similarly, 11 out of 12 K14E6/E7/Pik3ca^E545K:K14CreERtm mice (91.7.%) developed invasive squamous cell carcinoma and 5 out of 13 tumors were found at the anorectal junction (Supplementary Fig. S1).

HPV16 E6/E7 induce DNA synthesis in the suprabasal compartment of anal epithelium as well as the skin (26), cervix (27, 28) and lingual/esophageal epithelium (29). To determine whether activating mutations in Pik3ca induce DNA synthesis in anal epithelium, we analyzed by immunohistochemistry the frequency of bromodeoxyuridine (BrdUrd)-positive basal and suprabasal cells in sections of tissue from mice injected with this nucleoside analog 1 hour before sacrifice (27). Expression of p110α^H1047R as well as p110α^E545K significantly induced DNA synthesis in basal cells of the anal epithelium, and also to a lesser degree in suprabasal cells (Fig. 3). In both 4-OHT–treated Pik3ca^H1047R:K14CreERtm mice and Pik3ca^E545K:K14CreERtm mice, the frequency of basal cells supporting DNA synthesis was significantly higher than in the control group (Pik3ca^H1047R:K14CreERtm vs. Pik3ca^H1047R and Pik3ca^E545K:K14CreERtm mice vs. Pik3ca^E545K, P = 0.02857 and P = 0.05714, respectively) or K14E6/E7/Pik3ca mutant mice (Pik3ca^H1047R:K14CreERtm or Pik3ca^E545K:K14CreERtm vs. K14E6/E7/Pik3ca^H1047R, P = 0.03 for both comparisons). There was no significant difference in the frequency of BrdUrd positivity in basal cells between the Pik3ca^H1047R: K14CreERtm and Pik3ca^E545K:K14CreERtm mice. Suprabasal DNA synthesis in Pik3ca^H1047R:K14CreERtm mice as well as Pik3ca^E545K:K14CreERtm mice also was significantly increased compared with the control group (Pik3ca^H1047R and Pik3ca^E545K, P = 0.01972 and P = 0.05714, respectively). However, this induction of suprabasal DNA synthesis was significantly lower than that observed in K14E6/E7/Pik3ca mutant mice (Pik3ca^H1047R:K14CreERtm vs. K14E6/E7/Pik3ca^H1047R, Pik3ca^E545K:K14CreERtm vs. K14E6/E7/Pik3ca^E545K, P = 0.02857, respectively).

To assess the effect of inhibiting the mTOR pathway on the growth of mouse anal cancers, we tested an investigational dual TORC1/2 inhibitor, TAK-228, on spheroid cultured tumor cells. After isolating primary mouse anal tumors from K14E6/E7, K14E6/E7/Pik3ca^H1047R:K14CreERtm, and K14E6/E7/Pik3ca^E545K:K14CreERtm mice, tumor cells were combined with Matrigel and plated in 24-well plates, allowing us to culture these tumors as spheroids (Fig. 4A). Before treatment with TAK-228, we measured the diameter of each spheroid in the experimental and control groups. Spheroids in the experimental group were cultured in medium containing 10 μmol/L of TAK-228. After 48 hours, the diameter of spheroids from both groups were again measured (Fig. 4B). All tumor spheroids treated with TAK-228 showed a significant reduction in growth compared with the control group not treated with the drug. Interestingly, spheroids from the K14E6/E7 mice treated with TAK-228 also showed significantly reduced cell growth, consistent with our prior observations (5) and those shown above (Fig. 1; Supplementary Fig. S2) that mTOR pathway is activated in anal tumors arising in K14E6/E7 mice. However, TAK-228 caused a greater reduction in the growth of cells expressing a mutant form of Pik3ca. To verify inhibition of the mTOR pathway in TAK-228–treated spheroids, we performed western blot analysis on protein lysates from spheroids using three biomarkers of the PI3K–AKT–mTOR signaling pathway (Fig. 4C). In contrast with the control group, phosphorylation of AKT, S6 and 4-EBP1 was diminished in all drug-treated groups, consistent with TAK-228 causing the inhibition of growth in tumor spheroids through the inhibition of the PI3K–AKT–mTOR signaling pathway. These data confirmed that TAK-228 efficiently restricts the activation of both the TORC1 (phosphorylation of both S6 and 4-EBP1) and the TORC2 (phosphorylation of AKT) pathways.

To examine the effect of TAK-228 treatment in vivo, we generated tumor grafts in immunodeficient mice (NSG mice). Here, we were interested in testing the efficacy of TAK-228 on anal squamous cell carcinoma, as opposed to benign tumors. For this reason, tumor grafts from anal tumors arising in K14E6/E7/Pik3ca^E545K:K14CreERtm mice were first established in NSG mice and verified by histology to be squamous cell carcinomas. Tumor grafts (passage 3) that were verified to be squamous cell carcinoma were then injected subcutaneously into the flanks of 12 NSG mice. In the treatment arm, 6 randomized mice received 1 mg/kg of TAK-228 dissolved in vehicle, whereas the other 6 mice in the control arm were provided vehicle only, in both cases delivered by oral gavage daily for 8 to 9 days (Fig. 5A). The growth rate of tumors in treatment group was significantly decreased compared with that seen in the control group [P (two-sided) = 1.08 × 10⁻⁵, Fig. 5B]. This observation is similar to the result obtained with the spheroids, indicating that TAK-228 was effective in controlling the growth of tumors arising in mice expressing an activating mutation in Pik3ca and the HPV16 oncogenes. To assess inhibition of the PI3K/AKT/mTOR pathway in TAK-228–treated tumor grafts, we performed western blot analysis on protein lysates from tumor grafts harvested at 0, 3, and 24 hours post-treatment with TAK-228 (Fig. 5C). TAK-228 efficiently inhibited the phosphorylation of AKT, S6 and 4-EBP1 at 3 hours post-treatment but not at 24 hours post-treatment, consistent with the known pharmacokinetics of the drug (16).

TAK-228 inhibits cell proliferation and induces apoptosis (30, 31). To determine whether these phenotypes are affected by treatment with TAK-228 in our hands, we analyzed sections of paraffin-embedded tissues from animals in both the control and treatment arms. To determine whether proliferation is changed after TAK-228 treatment, we carried out Ki67 immunohistochemistry (Fig. 5D). In TAK-228–treated tumor grafts, the frequency of Ki67-positive cells was decreased compared with vehicle-treated tumor grafts. Using the TUNEL assay, we observed a large number of cells with fragmented DNA indicative of apoptosis in the TAK-228–treated tumor grafts, whereas such cells were rarely detected in the vehicle-treated tumor grafts (Supplementary Fig. S3). We also performed immunofluorescence for K10, a marker of the spinous layer, and K14, a marker of basal cells, to examine any changes in the epithelial differentiation/maturation status following treatment with TAK-228 (Supplementary Fig. S4). Compared with vehicle-treated tissue, the expression of K10 was slightly increased in TAK-228–treated tumor grafts, whereas expression of K14 was decreased. These observations indicate that TAK-228 might also induce differentiation/maturation of anal tumor cells.

In this study, we demonstrate the oncogenic potential of Pik3ca mutations in driving anal carcinogenesis in the presence or absence of HPV16 oncogenes, and show that an investigational dual TORC1/2 inhibitor now in clinical trials is effective at inhibiting the growth of tumor spheroids in vitro and tumor grafts in vivo. These data establish the importance of the PIK/AKT/mTOR pathway in anal carcinogenesis, and the potential therapeutic value of targeting this pathway in treating human anal cancers.

Activating mutations in Pik3ca arise in 20%∼25% of human anal cancers (15, 32), and 40%∼55% of HPV-positive human head and neck cancers (11). The most frequent mutations are found in exon 9 (E542K, E545K) and are consistent with the hypothesis that HPV16 drives APOBEC (“apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like”)-mediated mutagenesis by cytosine deamination. Cytidine deaminase activity of APOBEC family members is argued to cause oncogenic mutations in many types of cancers (33–36). APOBEC-mediated cytosine deamination of the PIK3CA helical domain (Exon 9) has been linked to HPV-associated carcinogenesis (37). HPV oncogenes E6 and E7 can induce the expression of APOBEC3 family members (38, 39), which can explain how HPV oncogenes drive APOBEC-mediated DNA mutagenesis. A mutational signature of the APOBEC family is frequently detected in human anal cancers (32). Thus HPV may drive the accumulation of certain activating mutations in Pik3ca through activation of APOBECs. The Pik3ca^H1047R mutation is more oncogenic than the Pik3ca^E545K mutation in anal carcinogenesis in the absence of HPV oncogenes (Fig. 2B; Table 1), as seen previously in a mouse mammary cancer model (19). However, in the presence of HPV16 oncogenes, this difference in potency was lost (Fig. 2B; Table 1), suggesting that HPV16 oncogenes potentiate the oncogenic activities of these two Pik3ca-activating mutations similarly. This together with the fact that HPV drives APOBEC activity may explain why H1047R mutations are less common in HPV-positive human cancers.

HPV16 E6 oncogene activates the PI3K–AKT–mTOR signaling pathway in tissue culture (40), consistent with our prior (5) and current (Fig. 1) data that this pathway is activated in the anal epithelium of HPV16 transgenic mice in the absence of expression of activating mutations in Pik3ca. Nevertheless, activating mutations in HPV-associated human cancers are frequently observed, which led us to the current study assessing the importance of such mutations in anal carcinogenesis. Our data clearly indicate that such mutations help drive anal carcinogenesis even in the presence of HPV16 oncogenes. The PI3K–AKT–mTOR signaling pathway was clearly enhanced when both an activating mutation in Pik3ca and HPV16 oncogenes were expressed (Fig. 1) providing insight into why such activating mutations in Pik3ca contribute to HPV-associated anal cancers.

Activating mutations in Pik3ca play an important role in Ras-driven carcinogenesis in mice. Disruption of the interaction of p110α with oncogenic Ras causes a defect in AKT activation and reduced transformation efficiency by H-Ras (41) In this study, we observed that early onset expression of an activating mutant form of p110α together with HPV16 oncogenes drove anal carcinogenesis even in the absence of DMBA, a carcinogen known to drive initiation of carcinogenesis (Fig. 1). This leads us to hypothesize that activating mutations in Pik3ca contribute to the initiation phase of carcinogenesis, with HPV16 oncogenes contributing to the promotion phase, as we have clearly demonstrated previously (42). Consistent with this concept, activating mutations in Pik3ca have been shown to promote tumor initiation in breast and lung carcinogenesis (43, 44) and drive rapid onset of colon cancers (24). Even when an activating mutation in Pik3ca is expressed late in the context of expression of HPV oncogenes, it led to rapid onset of anal carcinogenesis (Fig. 2C), which is further evidence that such mutations are strong drivers of anal carcinogenesis regardless of when they arise.

Currently, concurrent chemotherapy and radiotherapy (CRT) is the standard therapy for patients with anal cancers and is associated with a high degree of morbidity (45). Treatment of recurrent anal cancer, which is common, remains a challenge. In this study, the dual TORC1/2 inhibitor, TAK-228, proved highly effective in treating mouse anal tumor grafts as well as tumor spheroids suggesting that this or similar drugs might be effective new options, alone or in combination with other therapies, in the treatment of primary or recurrent anal cancers in humans.

D.S. Meyer is an employee of Idorsia Pharmaceuticals Ltd. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M.-K. Shin, A. Bilger, D.A. Deming, P.F. Lambert

Development of methodology: M.-K. Shin, S. Payne, A. Bilger, M. Bentires-Alj, D.A. Deming

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.-K. Shin, S. Payne, A. Bilger, K.A. Matkowskyj, E. Carchman, D.A. Deming

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.-K. Shin, S. Payne, A. Bilger, K.A. Matkowskyj, D.A. Deming, P.F. Lambert

Writing, review, and/or revision of the manuscript: M.-K. Shin, S. Payne, A. Bilger, K.A. Matkowskyj, E. Carchman, M. Bentires-Alj, D.A. Deming, P.F. Lambert

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Bilger, D.S. Meyer

Study supervision: P.F. Lambert

This study was supported by grants from the National Cancer Institute (to P.F. Lambert; CA022443, CA210807, CA171873) and (to D.A. Deming; CA226526). Research in the laboratory of M. Bentires-Alj is supported by the European Research Council, the Swiss National Science Foundation, the Krebsliga Beider Basel, the Swiss Cancer League, the Department of Surgery of the University Hospital of Basel and the Swiss Initiative for Systems Biology (SystemsX.ch). We thank Tobias Eichlisberger (FMI) for technical advice and shipping of the animals.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.