Blockade of the PGE2 Pathway Inhibits the Growth of PTEN-Deficient HNSCC Tumors

Abstract

Increased PI3K signaling as a result of PIK3CA mutation or amplification or decreased expression of phosphatase and tensin homolog deleted on chromosome 10 (PTEN) is one of the most common alterations in head and neck squamous cell carcinoma (HNSCC). PTEN negatively regulates PI3K signaling and its downstream effectors including COX2. COX2 mediates the synthesis of prostaglandin E2 (PGE2) which contributes to immunosuppression in the tumor microenvironment. PGE2 also binds to one or more EP receptors (EP1-EP4) and promotes the growth of tumor cells via activation of EP2 and EP4. However, the role of PGE2 in PTEN-deficient HNSCC is incompletely understood. In this study, we assessed PGE2 signaling in PTEN-deficient HNSCC and evaluated the effect of aspirin or TPST-1495, a dual EP2/EP4 antagonist, on the growth of PTEN knockout and PIK3CA-altered HNSCC tumors in immunocompetent mice. Our results demonstrated that aspirin selectively inhibits the growth of PTEN knockout HNSCC tumors. TPST-1495 inhibited tumor growth and substantially increased the antitumor activity of the immune checkpoint inhibitor anti-PD1. To date, there are no FDA-approved therapies for PI3K pathway-altered HNSCC. Our findings suggest that NSAIDs demonstrate antitumor activity in PTEN-deficient or PI3K-altered tumors whereas EP2/EP4 targeting may augment FDA-approved anti-PD1 therapy in HNSCC.

Figure 1:. Loss of PTEN activates PI3K pathway in HNSCC.

(A) PTEN protein expression in PTEN knockout (PTEN KO) Cal27 and FaDu human cell lines was significantly reduced compared to parental cells in conjunction with increased expression of phosphorylated AKT (pAKT). (B) PTEN protein expression in PTEN KO MOC1 or MOC22 murine cells was significantly reduced compared to parental cells in conjunction with increased expression of phosphorylated AKT (pAKT). (C) PTEN mRNA expression is significantly reduced in PTEN KO compared to parental MOC1 and MOC22 cells (p<0.0001, as determined by the student’s t-test). The data presented is a mean of triplicate measurements and SD of each group. The experiment was conducted twice with representative data from one experiment is shown. (D) Protein expression of FLAG-tagged proteins in amplified WT PIK3CA, PIK3CA-E545K, or PIK3CA-H1047R MOC1 tumors confirmed the presence of genetically engineered expression constructs. PI3K activation, as measured by pAKT expression, consistently increased in engineered MOC1 tumors compared to parental tumors (p<0.0001 for all groups). Each lane represents a distinct xenograft tumor derived from the engineered cell lines as indicated. The mean densitometry measurement and SEM of pAKT level is shown. *** p< 0.001 determined by the Dunnett’s method.

Figure 2:. Elevated COX2 expression and PGE2 production in PTEN-deficient HNSCC cells.

(A) COX2 mRNA expression was upregulated in human FaDu PTEN KO cells compared to parental cells (p=0.0003). (B) COX2 mRNA expression was upregulated in murine MOC1 PTEN KO cells compared to parental cells (p=0.0003). (C) COX2 mRNA expression was upregulated in MOC22 PTEN KO cells compared to parental cells (p=0.0003). (D) Increased secreted PGE2 levels, as measured by ELISA, were observed in FaDu PTEN KO cells (p=0.022). (E) Increased secreted PGE2 levels, as measured by ELISA, were observed in MOC1 PTEN KO cells (p=0.0003). (F) Increased secreted PGE2 levels, as measured by ELISA, were observed in MOC22 PTEN KO cells (p=0.0003). The data presented is the mean of triplicate measurements and SD of each group. The experiments were performed at least twice and representative data from one experiment is shown. * p< 0.05 *** p< 0.001 as determined by the Dunnett’s or student’s t-test.

Figure 3:. Aspirin inhibits tumor growth in murine HNSCC tumors with PTEN loss.

MOC1 parental or PIK3CA mutant (H1047R or E545K) or PTEN KO cells were inoculated into the flanks of C57BL/6 mice and grown until tumors were palpable (about 100 mm³; approximately 10 days). Mice were then treated with aspirin (60 mg/kg, three times per week) or vehicle (PBS) for two weeks (n=5/group). Aspirin did not affect the growth of tumors derived from MOC1 parental cells, whereas tumors derived from MOC1 cells engineered for PIK3CA mutations H1047R (p=0.0036) or E545K (p=0.012) or PTEN KO (p=0.002) were significantly inhibited by aspirin treatment. Data shown are the mean and SEM of each group. * p< 0.05 and ** p<0.01 determined by a Mixed Effects Analysis of Rate of Tumor Growth.

Figure 4:. Increased EP2 and EP4 receptor expression in PIK3CA-altered or PTEN-deficient murine HNSCC cells.

(A) PGE2 receptor EP1 mRNA expression was significantly decreased in PIK3CA-H1047R MOC1 cells (p=0.018), EP3 mRNA expression was significantly decreased in PIK3CA-altered cells (p=0.0003 for PIK3CA-E545K, p=0.0064 for PIK3CA-H1047R) and PTEN KO MOC1 cells (p<0.0001). On the other hand, EP2 and EP4 mRNA expression were significantly increased in PIK3CA-altered and PTEN KO MOC1 cells (p<0.0001 for all models). (B) Similar expression profiles of EP2 (p=0.013 for PIK3CA-E545K, p<0.0001 for PIK3CA-H1047R, and p=0.0013 for PTEN KO) and EP4 (p<0.0001 for PIK3CA-E545K and PIK3CA-H1047R, and p=0.001 for PTEN KO) mRNA were seen in MOC22 PIK3CA-altered and PTEN KO cells. However, unlike in MOC1 cells, EP3 mRNA were slightly increased PIK3CA-H1047R MOC22 cells (p=0.039) while EP1 mRNA were decreased in PIK3CA-H1047R (p=0.0001) MOC22 cells. Data shown is a mean of triplicate measurements from each group and SD. Experiments were performed at least twice and representative data from one experiment is shown. * p< 0.05, **p< 0.01, and *** p< 0.001 is determined by Dunnett’s method.

Figure 5:. Dual EP2/EP4 inhibition decreases HNSCC cell proliferation and tumor growth.

(A) MOC1 parental cells or cells engineered for expression of WT PIK3CA, PIK3CA-E545K, PIK3CA-H1047R, or PTEN KO were treated with TPST-1495 (100 μM) or vehicle (DMSO) for 72 hours before determining viable cell counts. TPST-1495 significantly suppressed cell proliferation in PIK3CA-E545K (p=0.0062) and PTEN KO (p<0.0001) MOC1 cells. The data presented is a mean of duplicate wells and SD for each group. Experiments were conducted at least twice and representative data is shown. ** p< 0.01and *** p< 0.001 determined by Sidak’s method. MOC1 parental cells (B), MOC1 PIK3CA-E545K cells (C), (D) MOC1 PIK3CA-H1047R cells, (E) or MOC1 PTEN KO cells were inoculated subcutaneously in C57BL/6 mice and tumors were allowed to grow until palpable (about 100 mm³; approximately 10 days). TPST-1495 reduced tumor growth in MOC1 tumors independent of PIK3CA or PTEN status (p=0.025 for parental MOC1, p=0.024 for PIK3CA-E545K, p=0.0004 for PIK3CA-H1047R, and p=0.049 for PTEN KO). MOC22 parental cells (F) or PTEN KO (G) cells were similarly implanted into the flanks of C57BL/6 mice and grown until palpable (about 100 mm³; approximately 10 days). TPST-1495 selectively reduced tumor growth in PTEN KO xenografts [(G) p=0.065] compared with parental control tumors [(F) p=0.22]. Box plots are drawn for each group with minimum and maximum values. * p< 0.05 and *** p< 0.001 as determined by the Sidak’s method.

Figure 6:. Dual EP2/EP4 inhibition enhances anti-PD1 anti-tumor activity.

MOC1 parental cells (2x10⁶ cell/mouse) were implanted subcutaneously into flanks of immunocompetent C57BL/6 mice. Once tumors were palpable (about 100 mm³; approximately 10 days), mice were treated with vehicle (methylcellulose bi-daily and/or IgG), TPST-1495 (150 mg/kg, bi-daily), anti-PD1 (0.2 mg per mouse on days 1, 4, and 7 after palpable tumors were established), or the combination of TPST-1495 and anti-PD1 (n=5 per group). (A) Concomitant administration of TPST-1495 and anti-PD1 showed an additive effect on tumor growth inhibition compared to TPST-1495 (p=0.011) or anti-PD1 (p=0.007) alone. Shown is the mean measurement and SEM of each group. * p<0.05 and ** p<0.01 determined by Dunnett’s method. (B) Immune profiling was performed to determine CD8+/CD4+ T cell ratios. Tumors from combinatory treatment (TPST-1495 and anti-PD1) groups demonstrated an increase in CD8/CD4 ratio compared to vehicle treated tumors (p=0.022). Shown are the individual ratios for each replicate with the statistical significance determined by the ad hoc pairwise comparisons. Dunnett’s method was used to account for the multiplicity of the ad hoc comparisons.