Dual Blockade of EP2 and EP4 Signaling is Required for Optimal Immune Activation and Antitumor Activity Against Prostaglandin-Expressing Tumors

While the role of prostaglandin E2 (PGE2) in promoting malignant progression is well established, how to optimally block the activity of PGE2 signaling remains to be demonstrated. Clinical trials with prostaglandin pathway targeted agents have shown activity but without sufficient significance or dose-limiting toxicities that have prevented approval. PGE2 signals through four receptors (EP1–4) to modulate tumor progression. EP2 and EP4 signaling exacerbates tumor pathology and is immunosuppressive through potentiating cAMP production. EP1 and EP3 signaling has the opposite effect through increasing IP3 and decreasing cAMP. Using available small-molecule antagonists of single EP receptors, the cyclooxygenase-2 (COX-2) inhibitor celecoxib, or a novel dual EP2/EP4 antagonist generated in this investigation, we tested which approach to block PGE2 signaling optimally restored immunologic activity in mouse and human immune cells and antitumor activity in syngeneic, spontaneous, and xenograft tumor models. We found that dual antagonism of EP2 and EP4 together significantly enhanced the activation of PGE2-suppressed mouse and human monocytes and CD8+ T cells in vitro as compared with single EP antagonists. CD8+ T-cell activation was dampened by single EP1 and EP3 antagonists. Dual EP2/EP4 PGE2 receptor antagonists increased tumor microenvironment lymphocyte infiltration and significantly reduced disease burden in multiple tumor models, including in the adenomatous polyposis coli (APC)min+/− spontaneous colorectal tumor model, compared with celecoxib. These results support a hypothesis that redundancy of EP2 and EP4 receptor signaling necessitates a therapeutic strategy of dual blockade of EP2 and EP4. Here we describe TPST-1495, a first-in-class orally available small-molecule dual EP2/EP4 antagonist. Significance: Prostaglandin (PGE2) drives tumor progression but the pathway has not been effectively drugged. We demonstrate significantly enhanced immunologic potency and antitumor activity through blockade of EP2 and EP4 PGE2 receptor signaling together with a single molecule.

Despite the well-established linkage between PGE2 production and tumor progression, there are no FDA-approved cancer therapeutics targeting the prostaglandin pathway. Broad inhibition of PGE2 signaling pathways with NSAIDs that block both COX-1 and COX-2 (e.g., naproxen), or those that block only COX-2 (e.g., celecoxib), results in antitumor effects in mouse syngeneic tumor models. Epidemiologic and prospective studies indicate that inhibition of the PGE2 pathway can translate to clinical benefit (27)(28)(29). Indeed, regular use of NSAIDs after diagnosis with colorectal cancer and head and neck cancer has been associated with improved outcomes in PIK3CA-mutated cancers (30,31). However, gastrointestinal, renal, and cardiac toxicities are associated with chronic usage of COX inhibitors due to imbalances in prostacyclin and thromboxane levels limiting both tolerable dose levels and their long-term use for treatment of advanced cancers.
Here we describe the in vitro and in vivo activity of TPST-1495, a first-in-class dual EP2/EP4 PGE2 receptor antagonist. TPST-1495 is an orally bioavailable small molecule with potent and selective activity against the EP2 and EP4 receptors while sparing the structurally homologous, yet differentially active, EP1 and EP3 counterparts. We show that TPST-1495 was significantly more potent than single EP receptor antagonists or COX-2 inhibition at inducing immune activation, antitumor immunity, and tumor clearance in multiple models of human malignancies, highlighting the redundancy of EP2 and EP4 signaling in the tumor and the necessity of dual receptor blockade. These results support the scientific rationale for a new cancer therapeutic approach based on the targeted combined antagonism of EP2 and EP4.

TPST-1495 Selection and IC 50 Determination
Cell lines stably expressing EP1, 2, 3, or 4 receptors were purchased from Eurofins Panlabs, Inc (catalog nos. HTS142L, HTS185L, HTS099C, and HTS092L). EP2 and EP4 Cell lines were cultured in T-150 media for up to 50 passages. Cells were passaged every 3-5 days when less than 80% confluent. EP1 and EP3 cell lines were thawed immediately prior to use per manufacturer's recommendations. To facilitate calcium flux assay readout, cloned EP2 and EP4 receptor-expressing ChemiBrite cells were made by stable transfection of HEK293 cells with ChemiBrite clytin, EP2 or EP4 receptor linked genetically to a promiscuous G protein to couple EP receptors to the calcium signaling pathway. Cells were plated 18-24 hours prior to assay initiation in basal media per manufacturer's recommendations. For EP2 and EP4 stable cell lines, 2 × 10 4 cells were plated per well. For EP1 and EP3, cells were thawed and plated into two 96-well plates. Hank's Balanced Salt Solution (HBSS) + 20 mmol/L HEPES pH 7.4 was used as dilution assay buffer. A total of 10 μmol/L Coelenterazine (Invitrogen) assay buffer was used as loading buffer and prepared the day of conducting the assay. Plates were incubated in dark for 2-3 hours. PGE2 (Sigma P5640-5 mg) was used at a final concentration of 10 nmol/L for EP1, EP2, and EP4 assay. For EP3 assay, PGE2 was used at a final concentration of 200 nmol/L. Plates were incubated in dark for 2-3 hours. Serial 11-point half log dilutions of TPST-1495 were prepared beginning at 30 μmol/L final concentration for EP1 and EP3 and 1 μmol/L for EP2 and EP4. Antagonist was added at 25 μL per well and incubated for 20 minutes prior to being read on a Flexstation (Molecular Devices). TPST-1495 wells were run in duplicate. Control wells were run in quadruplicate.

PGE2 Blockade Whole Blood Assay
The PGE2 Blockade whole blood assay was performed using mouse and human whole blood. Human whole blood assays were performed with fresh, same day drawn healthy donor whole blood (catalog no. 70507.2, StemCell Technologies) and plated at 75 μL per well. Mouse whole blood assays were performed with whole blood drawn from terminal cardiac puncture of up to 15 female Balb/c mice and plated at 75 μL per well. Whole blood was then treated with dilutions of TPST-1495 and incubated for 30 minutes at 37°C, followed by 10 nmol/L or 500 nmol/L PGE2 (catalog no. P0409, Sigma-Aldrich) and incubated for 30 minutes at 37°C. Lipopolysaccharide (LPS) was then added to a final concentration of 0.5 mg/mL and the combination of these reagents was incubated at 37°C for 16-20 hours, at which point plates were isolated to separate supernatant for collection. TNFα was measured by murine (catalog no. BMS6073, Thermo Fisher Scientific) or human (catalog no. BMS223, Thermo Fisher Scientific) ELISA.

PGE2 Blockade Assay with T-cell Activation
Human T-cell assays were performed with peripheral blood mononuclear cell (PBMC) isolated from a fresh leukopak (catalog no. 70500.2, StemCell Technologies). Human PBMC were thawed in warm RPMI+ Benzonase, enriched for T cells using magnetic separation (catalog no. 130-096-535, Miltenyi Biotec), then plated at 2 × 10 6 cells/mL in 100 μL RPMI supplemented with 10% FBS (catalog no. 10438026, Gibco) and penicillin-streptomycin (catalog no. 15140163, Gibco). Mouse T-cell activation assays were performed with unenriched single-cell suspensions isolated after red blood cell (RBC) lysis (catalog no. A1049201, Gibco) of pooled lymph nodes and spleens from OT-1 Mice [Jax, C57BL/6-Tg(TcraTcrb)1100 Mjb/J] and plated at 0.5-1 × 10 6 cells per well. In both human and mouse assays, cells were then preincubated with EP inhibitors for 20 minutes at 37°C, followed by PGE2 (catalog no. P0409, Sigma-Aldrich) for 20 minutes at 37°C, exposed to varying concentrations of PGE2 (10-1,000 nmol/L) for 30 minutes at 37°C. After these preincubation steps, CEF peptides (catalog no. 3616-1, Mabtech) or SIINFEKL peptide (catalog no. S7951, Sigma-Aldrich) were added, and all components were incubated for 6 additional hours at 37°C. At that time, supernatants were collected, and cytokines were measured by Luminex (catalog no. HCYTA-60K-PX48, Millipore). For APC min/+ in vivo tumor experiments, mice were obtained from JAX at 6-8 weeks and aged in-house until they were 12-13 weeks old. Mice were then administered a regimen of 100 mg/kg TPST-1495 orally every day or twice a day, E7046 (MedChem Express, HY-103088), PF-04419848 (SelleckChem, catalog no. S7211), or celecoxib (MedChem Express, Hy-14398) for 3 weeks, at which time tumors were resected and prepared for histology and RNA sequencing. In some experiments, mice were administered drugs for 6 weeks then followed for humane endpoints (lethargy, body weight) were reached for survival analysis, as described in the relevant Figure legends.

Cell
The dose, route, and regimen of celecoxib used for our studies was based on literature describing the activity and toxicity of COX inhibitors in vivo. The 60 mg/kg maximum dose used was well tolerated without cage-side observations and was the maximally effective dose for in vivo studies (32).

Prostaglandin Pathway Inhibitor Drugs
PGE2 pathway inhibitors used are listed in Table 1A. Each inhibitor was resuspended in DMSO (catalog no. BP231-100, Thermo Fisher Scientific) at 50 mmol/L and kept frozen at −20°C until use.

Flow Cytometry
Tumors were processed for flow cytometry by maceration with a scissors followed by dissociation in Collagenase type IV (catalog no. LS004189, Worthington Biochemical Corporation) and Benzonase (catalog no. MFCD00131010,

TPST-1495 Inhibition of Tumor Cell Growth In Vitro
Tumor cells were plated at 5,000 cells per well in a 96-well plate and incubated overnight to adhere. Cells were then exposed to increasing concentrations of TPST-1495 for 30 minutes before exposure to increasing concentrations of PGE2. Cells with drug + PGE2 were incubated for 24 or 48 hours before an 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (catalog no. M6494, Invitrogen) was performed to assess relative cell viability and abundance.

TPST-1495 Inhibition of Tumor Cell Growth In Vitro with Serum Starvation
Tumor cells were plated at 5,000 or 10,000 cells per well in a 96-well plate and incubated overnight in complete media to adhere. Complete media was aspirated, cells were washed once with PBS (CAT #14190-250, Gibco), before media with varying concentrations of FBS including 0%, 1%, 5%, and 10% FBS was added to the plate. Cells equilibrated in the FBS starved media conditions for 30 minutes at 37°C before adding increasing concentrations of TPST-1495 using an autodispenser (TECAN). Cells were incubated in drug and FBS-restricted media for 48 hours before a CyQuant XTT Cell Viability Assay (catalog no. X12223, Invitrogen) was performed following the manufacturer's protocol to measure cell viability and abundance.

IHC
Histology of APC min/+ tumors was performed at CrownBio. Antibodies that were used for staining are listed in Table 1B. Histology of CT26 tumors was performed by Histowiz, LLC. Antibodies that were used for staining are listed in Table 1C.

Statistical Analyses
Graphs and statistical analyses were made using GraphPad Prism 9.4.1 software (GraphPad Software). Student t test and ANOVA were utilized to perform analyses between two groups or more than two groups as necessary. Gibbs tests for outliers were used to exclude outlier data points as necessary. Statistical significance in Kaplan-Meier curves was determined with log-rank and Mantel-Cox analysis. All values from statistical analyses were reported as follows.

Data Availability
The sequencing data generated in this study are publicly available in Gene Expression Omnibus at series record GSE229806.

Characterization of TPST-1495, a First-in-class Dual Antagonist of the EP2 and EP4
Analysis of The Cancer Genome Atlas (TCGA) using Cbio Portal revealed that EP2 was found to be expressed in diverse malignancies together with EP4 ( Fig. 1A Fig. S1). In the PGE2 competition assay conducted with HEK293-EP1 or HEK293-EP3 cell lines, TPST-1495 did not achieve 50% inhibition of EP1 or EP3 receptors at concentrations up to 30 μmol/L, and the IC 50 values were calculated to be 134,200 nmol/L and 108,800 nmol/L for EP1 and EP3, respectively ( Supplementary Fig. S1A). These results demonstrate that TPST-1495 is an exquisitely selective and specific dual antagonist of human EP2 and EP4 ( Supplementary Fig. S1B). TPST 1495 demonstrated oral bioavailability and good exposure in animals, as did E7046 (Eisai LTD, co.), a single EP4 receptor antagonist in clinical development and used extensively in this investigation as a comparator ( Supplementary Fig. S2). In addition, structural homology between murine and human EP receptors enabled the use of murine and human model systems to interrogate the activity of TPST-1495 herein.

Dual Blockade of EP2 and EP4 PGE2 Receptors Overcomes PGE2-mediated Immune Suppression in Mouse and Human Monocytes In Vitro
To define the response of dual EP2 and EP4 antagonism in comparison to single antagonism of either receptor, we exploited the capacity of PGE2 to suppress downstream TNFα production from monocytes following stimulation with LPS ( Fig. 2A; refs. 9, 38). We exposed mouse whole blood to a low concentration of PGE2 (10 nmol/L) that would effectively engage the EP4 but not the EP2 receptor, or with 500 nmol/L PGE2, a concentration that would ensure PGE2 signaling through both EP2 and EP4 receptors. The affinity of murine PGE2 for receptors EP2 and EP4 is 12 nmol/L and 1.9 nmol/L, respectively. TPST-1495 treatment rescued production of TNFα by treated monocytes at both low and high PGE2 concentrations with IC 50 values of 1 nmol/L and 382 nmol/L, respectively (Fig. 2B). In contrast, E7046, a specific EP4 antagonist, rescued production of TNFα only in the presence of low PGE2 (Fig. 2C). PF04419848, a specific EP2 antagonist, was unable to induce TNFα at the PGE2 concentrations tested (Fig. 2D). TPST-1495 rescued >80% of TNFα production from PBMC at both 10 nmol/L PGE2 (Fig. 2E) and 500 nmol/L PGE2 (Fig. 2F), whereas E7046 and PF04419848 had less than 50% recovery at both concentrations of PGE2. We then profiled the activity of TPST-1495 in OT-1 CD8 + T cells to determine whether dual EP2/EP4 inhibition was required to overcome PGE2  We then tested whether our observations in mouse monocyte cultures would extend to human immune cells. PBMCs from healthy adult human donors were treated with TPST-1495 or E7046 antagonists in the presence of low (10 nmol/L) or high (500 nmol/L) PGE2. Consistent with our observations in mouse whole blood, treatment with TPST-1495 led to recovery of TNFα production at all concentrations of PGE2 tested (Fig. 3A). However, treatment with single EP4 blockade restored TNFα production only under low PGE2 conditions (Fig.   3B). We hypothesized that dual blockade of EP2 and EP4 receptors would be similarly required to overcome PGE2 suppression of human Ag-specific CD8 + T cells. Indeed, TPST-1495 treatment restored IFNγ production in PGE2 concentrations up to 1,000 nmol/L during stimulation with a peptide pool of common CD8 + T-cell epitopes from cytomegalovirus, Epstein-Barr virus, and influenza together (CEF; Fig. 3C). In contrast, the single EP4 antagonist E7046 did not rescue IFNγ production from CD8 + T cells in CEF-stimulated cultures at PGE2 levels above 10 nmol/L (Fig. 3D) and the single EP2 antagonist PF04419848 failed to rescue IFNγ production above 333 nmol/L PGE2 (Fig. 3E).
To determine whether EP1 and EP3 inhibition was detrimental to T-cell activation, we conducted the same T-cell activation assay adding combined EP1 and EP3 antagonists, either individually or in combination with TPST-1495. In the presence of 333 nmol/L PGE2, EP1, and EP3 single antagonists dampened IFNγ production in a dose-dependent manner (Fig. 3F). However, the addition of EP1 and EP3 antagonists to TPST-1495 did not consistently result in a decrease in IFNγ rescue as compared with TPST-1495 treatment alone (Fig. 3G).
Collectively, our results demonstrate that dual inhibition of both EP2 and EP4 receptors in human PBMC cultures reverses PGE2-mediated immune suppression over a wide concentration range significantly more effectively than either single EP2 or EP4 receptor antagonists and that specific combined inhibition of EP2 and EP4 may be necessary to potentiate full CD8 + T-cell activity.

PGE2-Producing Tumors
Having shown the potency of dual blockade of EP2 and EP4 receptors in human and mouse immune cells in vitro, we then tested whether these findings translated to enhanced antitumor activity in vivo. Several tumor models were selected on the basis of a wide range of COX2 pathway activity, determined by measuring PGE2 levels in the supernatant of cultured tumor cells and PGEM levels in urine of tumor-bearing mice ( Fig. 4A and B). TPST-1495 therapy at 100 mg/kg twice a day was initiated when tumors were 80-100 mm 3 , a dose to 333 nmol/L PGE2 before stimulation with CEF peptides. Each datapoint represents the average percent recovery from one experiment, with four total experiments shown. G, Percent IFNγ production rescue, as normalized to control with 0 nmol/L PGE2 and 0 nmol/L TPST-1495. Cells obtained as in C-F were treated with increasing TPST-1495 in the presence or absence of 10 mmol/L EP1 inhibitor and 10 mmol/L EP3 inhibitor. Each datapoint represents the average percent recovery from one experiment, with four total experiments shown. Mean and SD are depicted in all graphs and the Student two-sided t test was used to assess significance between individual groups. ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05.
level resulting in plasma exposures that exceeded the protein adjusted EC 50 for rescue of TNFα production. TPST-1495 inhibited tumor growth in mice bearing colon carcinoma (CT26), triple-negative breast cancer (4T1), and mammary adenocarcinoma (TS/A), but not in those mice bearing LLC or B16F10 melanoma tumors (Fig. 4C). TPST-1495 exhibited a maximum tumor growth inhibition (TGI) of 54% in the CT26 model, 47% in the TS/A model, 37% in the 4T1 model, and 18% in the B16F10 model. In contrast, TGI in celecoxib-treated mice did not exceed 31% in any model tested (Fig. 4D). While TGI was generally higher among cohorts treated with TPST-1495 as compared with the single EP4 antagonist, E7046, these differences were not significant (Fig. 4D).

Immune-Modulatory Effects of TPST-1495
Because combined inhibition of PGE2 signaling through EP2 and EP4 receptors significantly reversed PGE2-mediated immune suppression and due to previous work suggesting the importance of COX2 expression and immune inhibition (19), we interrogated components of the immune response to characterize mechanisms associated with TPST-1495 therapeutic benefit. BALB/c mice bearing established CT26 tumors were treated with TPST-1495 for 14 days at 100 mg/kg twice a day and the number of TILs were quantified by flow cytometry. TPST-1495 induced significant infiltration of tetramer-positive AH1-specific CD8 + T cells and CD4 + FoxP3 − cells into the tumor and moderately increased CD49b + natural killer (NK)-cell infiltration and the ratio of tumor CD8 + to Treg + T cells (Fig. 4E). Infiltration of CD8 + T cells in CT26 tumors was additionally measured by IHC staining (Fig. 4F), confirming values measured in Fig. 4E. In addition, we utilized the ovalbumin-expressing EL-4 tumor cell line EG-7 and confirmed that tumor-infiltrating T cells from animals treated with TPST-1495 were significantly more effective at producing cytokines after ex vivo stimulation with phorbol 12-myistate 13-acetate and ionomycin ( Supplementary Fig. S3A-S3D). To further understand the role of T cells, tumor outgrowth was measured in CT26 tumor-bearing mice that were depleted of CD8α + cells 2 days prior to initiating TPST-1495 therapy. In this context, after 14 days of TPST-1495 treatment therapy, tumor outgrowth was inhibited by 33% inhibition when T cells were depleted and 52% without T-cell depletion, suggesting that the observed anti-CT26 tumor response resulted from both T cell-dependent and T cell-independent mechanisms ( Supplementary Fig. S4).
In addition to promoting increased lymphocytic infiltration into the TME, TPST-1495 treatment also resulted in an increase in the numbers of XCR1 + DC1 cross-presenting dendritic cells, as a proportion of total cells within the tumor (Fig. 4G). While macrophage cell numbers in the TME were largely unchanged in TPST-1495-treated mice, we found that macrophages were significantly shifted to an M1 effector phenotype, as indicated by an increase in MHCII + CD206 − macrophages as compared with vehicle-treated mice, which displayed a higher prevalence of the MHCII low CD206 + M2 macrophage phenotype (Fig. 4H). Notably, TPST-1495 significantly enhanced PD-1 therapeutic efficacy in mice bearing CT26 flank tumors, suggesting that primed T-cell responses were potentiated by the combination of PD-1 and PGE2 blockade ( Supplementary Fig. S5). Collectively, these results demonstrate that treatment of mice bearing high PGE2-producing CT26 tumors with TPST-1495 remodeled the TME as shown by a significant increase of innate and adaptive effector immune cell populations, combined with the reduction of immune suppressive innate and adaptive immune cell populations and significant therapeutic benefit when combined with immune checkpoint inhibitors.

TPST-1495 Exhibits Immune-Independent Therapeutic Antitumor Activity
In addition to its immunosuppressive capacity, other tumor-extrinsic properties of PGE2 include its support of the TME by promoting angiogenesis as well as tumor-intrinsic properties of driving tumor cell survival (11,(39)(40)(41).
The diverse tumor-promoting properties of PGE2 prompted us to test whether inhibition of immune-independent mechanisms might also play a role in the observed antitumor activity of TPST-1495. To assess this possibility, we evaluated the antitumor activity of TPST-1495 in RAG2 −/− and NSG genetically defined immune-deficient mice. TPST-1495 treatment inhibited the outgrowth of CT26 flank tumors in both RAG2 −/− and wild-type (WT) mice, though to a lesser degree in the absence of adaptive T-cell immunity ( Fig. 5A and B). As it has been shown previously that tumor-infiltrating NK cells play a critical role in initiating antitumor immunity in response to COX-2 pathway inhibition, we tested whether NK depletion would reduce the therapeutic effect of TPST-1495.
We found that while NK-cell depletion increased tumor outgrowth, TPST-1495 therapy was similarly efficacious in WT and RAG −/− tumor-bearing mice that were depleted of NK cells, demonstrating that the mechanism of TPST-1495 antitumor activity was both NK cell and T cell independent (Fig. 5C-E). To further test this possibility, we utilized LS174T human colon tumor cells, which harbor a PIK3CA mutation that enhances COX-2 expression and signal transduction and tumor proliferation through autocrine signaling of PGE2 through EP2 and EP4 receptors. LS174T tumor cells were orthotopically implanted into the cecum of NSG mice (Fig. 5F), which, due to IL2rg deficiency along with NOD and SCID mutations, lack T, B, and NK cells in addition to hyporesponsive innate immune cells. TPST-1495 treatment did not significantly inhibit primary tumor growth in the cecum, but tumor metastasis to both the lung and liver was significantly inhibited, as indicated by both tumor number and tissue weight as compared with the nontreated mice (Fig. 5G). We have not observed significant antiproliferative properties of TPST-1495 when added to cultured tumor cells, tested at drug levels sustained in the periphery of animal experiments ( Supplementary Fig. S6). (Continued) PF04418948 (100 mg/kg every day, orally), or Celecoxib (60 mg/kg every day, orally) and followed for outgrowth until vehicle mice reached humane endpoints. Results are representative of two experiments with N = 5 mice per group. Mean and SD are depicted in all graphs and the Student two-sided t test was used to assess significance between individual groups at the last timepoint before mice were sacrificed. **, P < 0.01; *, P < 0.05. Flow cytometry (E) or IHC (F) from CT26 tumors after 14 days of treatment. A total of 1 × 10 6 CT26 cells were implanted in the flanks of animals and followed until the average was 80-100 mm 3 . Mice were then treated for 14 days at which time tumors were resected and stained. Results depict one experiment with 8 mice, representative of more than three similar experiments. G and H, Flow cytometry from CT26 tumors resected after 17 days of treatment. Tumors were implanted as in E, treated for 17 days with TPST-1495, and then time tumors were resected and stained for DC1 + dendritic cells (G) or macrophages (H). Results shown in G represent one experiment with 10 mice, representative of multiple similar experiments with an average of 2-fold increase over vehicle, but data depicted are the only replicate that met P < 0.05 statistical cutoff. One datapoint in G met the Gibbs criteria for outliers and was excluded. Results in H represent one experiment of two similar experiments with at least 8 mice. Mean and SD are depicted in all graphs and the Student two-sided t test was used to assess significance between individual groups. ****, P < 0.0001; **, P < 0.01; *, P < 0.05. TPST-1495 every day and twice a day groups are concatenated because of limited tumor tissue in these groups. Mean and SD are depicted in all graphs and the Student two-sided t test was used to assess significance between individual groups. ****, P < 0.0001; ***, P < 0.001; **, P < 0.01; *, P < 0.05. BID, twice a day; QD, every day.

TPST-S1495 Effectively Increases Survival and Modulates the TME in APC min/+ Mice
transduction through the β-catenin pathway and e-cadherin, recapitulating human disease. Previous research demonstrated that PGE2 administration accelerates disease progression in APC min/+ mice by promoting tumor cell growth, vascular formation, and immune suppression (39,(42)(43)(44). We tested the antitumor efficacy of TPST-1495 in 12 to 13 weeks old APC min/+ mice compared with other prostaglandin pathway inhibitors. Untreated mice at this age contained multiple tumors largely confined to the small intestine that measured >1 mm in diameter. After a 3-week course of therapy, we observed that mice given TPST-1495 had significantly lower adenoma burden in the small intestine as compared with mice treated with other PGE2 pathway inhibitors tested or control mice (Fig. 6A). We then sought to expand the therapeutic regimen in the APC min/+ mouse model and compare the longitudinal effects of an extended treatment regimen among the various prostaglandin pathway inhibitors. APC min/+ mice were given a 6-week course of therapy and monitored for survival. The extended treatment regimen significantly increased the survival of mice treated with TPST-1495 given every day or a twice a day dosing regimen.
In contrast, treatment of APC min/+ mice with celecoxib or with single EP2 or EP4 PGE2 receptor antagonists did not significantly increase survival (Fig. 6B).
To ensure that our dosing strategy for EP2 and EP4 single antagonists were effective at blocking their respective receptors, we combined single EP2 and EP4 receptor antagonists and observed that their in vivo activity was similar to that of TPST-1495. To further these observations, we tested the same treat-  Fig. S3E and S3F).
To determine whether our observations of differing efficacy in APC min/+ mice among the prostaglandin pathway inhibitors we tested aligned with differences in TME immune infiltrates, we measured lymphocyte populations in APC min/+ mice by performing IHC on resected small intestinal areas with tumors or hypertrophy after 3 weeks of prostaglandin pathway inhibitor treatment. Strikingly, at this timepoint, no adenomatous or carcinomatous tissue could be detected in TPST-1495-treated animals by blinded independent pathology assessment ( Supplementary Fig. S7). In contrast, either no or a minimal decrease in the number of adenomas and carcinomas was observed in the small intestine of APC min/+ mice treated with COX-2 or single EP2 or EP4 PGE2 receptor inhibitors. Following the 3-week course of therapy, IHC staining revealed a significant increase in tissue-infiltrating CD8 + T cells in hypertrophic areas of the small intestine in mice treated with TPST-1495, E7046, and celecoxib, though not PF04418948 (Fig. 6C). While the level of CD8 + T-cell infiltration in TPST-1495-treated mice was greater than with other agents, these differences were nonsignificant. However, CD4 + T-cell infiltration was significantly higher in TPST-1495-treated mice. Interestingly, we observed that FoxP3 staining of CD4 + T cells increased uniquely in celecoxib-treated mice, suggesting that the increase of CD4 + T cells in TPST-1495-treated animals did not include an abundance of regulatory T cells.

Effective Antagonism of PGE2 Signaling Modulates Significant Transcriptional Shifts in the TME
We show here that redundancy between EP2 and EP4 signaling makes single EP receptor blockade necessary but not sufficient to prevent PGE2 mediated immune suppression over a broad concentration range of prostaglandin. To compare the transcriptional profile of tumors treated with TPST-1495 to those treated with single EP antagonists or COX2 inhibitors, we performed RNA sequencing on hyperplasias resected from 16-week-old APC min/+ mice that were given a 3-week treatment course of TPST-1495, E7046, PF04419848, or celecoxib. We then curated a gene set by concatenating genes found to be significantly perturbed in publications describing transcriptional changes after genetic or pharmacologic inhibition of the PGE2 pathway [ Fig. 7A (19,21,26)]. We found that TPST-1495 treatment upregulated or downregulated several genes from this curated set that was similar to published literature, most notably CXCL9, CXCL10, IFNγ, CCL20, and CCL5. However, E7046, PF04419848, and celecoxib treatment led to fewer and lower magnitude perturbations in this gene set, along with additional genes (Supplementary Fig. S8), demonstrating that EP2 and EP4 combined antagonism more effectively blocked inhibitory effects of PGE2. This conclusion was further supported by gene set enrichment analysis (GSEA), in which the IFNγ response signature was only significantly enriched in TPST-1495-treated samples (Fig. 7B).

Discussion
In this investigation, we characterized the activity and mechanism of TPST-1495, a first-in-class dual antagonist which specifically inhibits signaling through the tumor-promoting EP2 and EP4 PGE2 receptors and allows for continued of PGE2 signaling through the immune-promoting EP1 and EP3 receptors. We show that this specific dual antagonism results in reversal of PGE2-mediated immune suppression and significantly increased immune activation of both mouse and human immune cells in vitro over broad concentrations of PGE2 as compared with COX2 or single EP2 or EP4 receptor inhibitors.
To evaluate the capacity of targeted and selective EP receptor blockade to overcome PGE2-mediated immune suppression, we developed innate and adaptive immune cell systems to measure recovery of immune activation using effector cytokine readouts. Production of TNFα by monocytes in response to LPS stimulation could be completely suppressed with PGE2, which increases intracellular cAMP levels and in turn inhibits NFκB activation and TNFα production. Addition of TPST-1495 recovered production of TNFα in human and mouse monocytes stimulated with LPS at both low (10 nmol/L) or high (500 nmol/L) PGE2 concentrations. Strikingly, recovery of TNFα production with a single EP4 receptor antagonist was observed with low levels of 10 nmol/L PGE2, but not at the high 500 nmol/L PGE2 concentration tested. PGE2 has a relatively low affinity for EP2 (K d = 12 nmol/L in mouse, 13 nmol/L in humans) compared with EP4 (K d = 1.9 nmol/L in mouse, 0.6 nmol/L in humans).

FIGURE 7
Effective antagonism of PGE2 signaling modulates significant transcriptional shifts in the TME. A, Volcano plots of selected genes. Blue coloration represents significantly (P adj < 0.05) downregulated genes while red correlation represents significantly upregulated genes. Genes in gray did not meet the significance cutoff. B, Heat map depicting significantly upregulated gene sets from GSEA performed on RNA sequencing analysis above and representative GSEA plot from IFNγ signature. All results represent data from one experiment with 4 animals per group. *** indicates FDR < 0.25 for that gene set compared with vehicle-treated animals.
While there are several investigations exploring the activity of prostaglandin targeted therapies in immunogenic and immune checkpoint blockade responsive tumor cells, there are fewer examples testing the activity of this agent class in tumor settings that are not T cell-inflamed and not responsive to immune checkpoint therapy. We chose to study the possible differentiated therapeutic activity of TPST-1495 in the spontaneous APC min/+ model of polyposis and colorectal cancer for several reasons, which we posit underline the importance of dual EP2 and EP4 receptor blockade: (i) prostaglandin signaling is known to be a significant driver of colorectal cancer; (ii) the APC min/+ model is spontaneous and less inflammatory than conventional implanted syngeneic tumor cells; and (iii) early clinical development involves studies in advanced patients that are refractory to or become refractory to immune checkpoint inhibition. We observed striking and significantly better potency of TPST-1495 compared with all prostaglandin pathway inhibitors tested in tumor-bearing APC min/+ mice. At 13 weeks, APC min/+ mice have established large and small intestine adenomas. TPST-1495 treatment for 3 weeks in 13-week-old APC min/+ mice led to near complete regression of gut disease determined by gross dissection and by blinded histopathology analysis ( Fig. 6; Supplementary Fig. S7). In comparison, single EP receptor antagonists or celecoxib led only to partial tumor reduction. These results differed somewhat from those using subcutaneously implanted tumor models (Fig. 4C), in which EP4 inhibition by E7046 was not statistically distinguishable from dual EP2/4 inhibition by TPST-1495. APC min/+ and subcutaneous models used here differ significantly by multiple parameters including etiology, immune infiltration, and time on treatment. Further studies will be required to define treatment differences in these models.
Transcriptome analysis of APC min/+ tumors reinforced our understanding of the mechanism of action (MOA) (as defined by in vitro stimulation and in vivo depletion and GEMM studies) differences between TPST-1495 and other PGE2 blockade approaches. While transcriptional changes induced by TPST-1495 treatment were similar to those induced by single EP4 blockade, the changes were higher in magnitude. This observation suggests that redundant signaling of PGE2 through the EP2 receptor reduces the effect of single EP4 receptor blockade, for example, EP4 blockade is necessary but insufficient, and supports our proposed rheostat control model. The EP2 single antagonist had very little effect on immune cell activity, tumor regression, and the tumor transcriptome, reflecting the relatively low expression and affinity for PGE2 of this receptor compared with EP4.
Unique to our study, we demonstrated that TPST-1495 significantly inhibited metastasis of PIK3CA-mutant human LS147T tumors in immune-deficient NSG mice (Fig. 5). We chose this model due to observations that the PIK3CA mutation enhances COX-2 expression and promotes tumor proliferation via autocrine signaling, possibly informing eventual selection of patient populations that may be responsive to TPST-1495 therapy (22). While we did not define the mechanism underlying this result, we hypothesize the TPST-1495 either directly inhibited tumor proliferation required for metastasis and/or inhibited CXCL1dependent vascularization of tumors, a known tumor-extrinsic tumor pathway promoted by PGE2 (39). LS174T cells undergo reduced apoptosis in serum-free conditions when exposed to increasing amounts of PGE2, but to our knowledge there are no analogous studies performed using ex vivo monoculture of isolated APC min/+ neoplasms to test the sensitivity of these adenomas and carcinomas to PGE2. Murine syngeneic transplantable models we tested did not reveal changes in proliferation or cytotoxicity of TPST-1495 treatment over a range of serum conditions ( Supplementary Fig. S6). However, we have not ruled out the possibility that the requirement of PGE2 for tumor cell growth and survival is not well modeled in vitro, and that other cellular and environmental pressures necessitate PGE2 signaling in vivo, but not in vitro. Collectively, the experimental results support our assertion that the antitumor activity of TPST-1495 is due both to inhibition of tumor-intrinsic PGE2 signaling and inhibition of tumor-extrinsic PGE2-mediated suppression of the immune response.
TPST-1495 represents a unique approach to inhibit PGE2-driven malignancies and possibly increase the efficacy of immune checkpoint inhibitors. Supported by the experimental results presented here, we are evaluating the safety, tolerability, pharmacokinetics/pharmacodynamics, and possible antitumor activity in a dose escalation, optimization and scheduling clinical phase Ia/Ib study, both as monotherapy and in combination with pembrolizumab, in patients with advanced solid tumors who have failed available standard-of-care therapies (NCT04344795). The results shown in this investigation also support the rationale for testing TPST-1495 in familial adenomatous polyposis coli, in which individuals carrying the APC min/+ germline mutation have nearly a 100% lifetime risk for developing colorectal cancer.