The Wnt Pathway Inhibitor RXC004 Blocks Tumor Growth and Reverses Immune Evasion in Wnt Ligand–dependent Cancer Models

Wnt signaling is implicated in the etiology of gastrointestinal tract cancers. Targeting Wnt signaling is challenging due to on-target toxicity concerns and lack of druggable pathway components. We describe the discovery and characterization of RXC004, a potent and selective inhibitor of the membrane-bound o-acyl transferase Porcupine, essential for Wnt ligand secretion. Absorption, distribution, metabolism, and excretion and safety pharmacology studies were conducted with RXC004 in vitro, and pharmacokinetic exposure assessed in vivo. RXC004 effects on proliferation and tumor metabolism were explored in genetically defined colorectal and pancreatic cancer models in vitro and in vivo. RXC004 effects on immune evasion were assessed in B16F10 immune “cold” and CT26 immune “hot” murine syngeneic models, and in human cell cocultures. RXC004 showed a promising pharmacokinetic profile, inhibited Wnt ligand palmitoylation, secretion, and pathway activation, and demonstrated potent antiproliferative effects in Wnt ligand–dependent (RNF43-mutant or RSPO3-fusion) colorectal and pancreatic cell lines. Reduced tumor growth and increased cancer cell differentiation were observed in SNU-1411 (RSPO3-fusion), AsPC1 and HPAF-II (both RNF43-mutant) xenograft models, with a therapeutic window versus Wnt homeostatic functions. Additional effects of RXC004 on tumor cell metabolism were confirmed in vitro and in vivo by glucose uptake and 18fluorodeoxyglucose-PET, respectively. RXC004 stimulated host tumor immunity; reducing resident myeloid-derived suppressor cells within B16F10 tumors and synergizing with anti-programmed cell death protein-1 (PD-1) to increase CD8+/regulatory T cell ratios within CT26 tumors. Moreover, RXC004 reversed the immunosuppressive effects of HPAF-II cells cocultured with human peripheral blood mononuclear cells, confirming the multiple anticancer mechanisms of this compound, which has progressed into phase II clinical trials. Significance: Wnt pathway dysregulation drives many gastrointestinal cancers; however, there are no approved therapies that target the pathway. RXC004 has demonstrated the potential to block both tumor growth and tumor immune evasion in a genetically defined, clinically actionable subpopulation of Wnt ligand–dependent gastrointestinal cancers. The clinical utility of RXC004, and other Porcupine inhibitors, in such Wnt ligand–dependent cancers is currently being assessed in patient trials.


Wnt Palmitoylation Assay
Palmitoylation of Wnt was performed by the Burris group, under a fee-forservice agreement, and was detected as described in Galli and colleagues (16).
Briefly, HEK293T cells were transfected with GFP-Wnt1-Fc constructs, followed by overnight incubation with alkyne palmitate (Alk-C16) in the presence of DMSO or 100 nmol/L RXC004. Cells were lysed and A/G agarose beads used to pull down GFP-Wnt1-Fc, which was then subjected to click chemistry to covalently attach biotin azide to any attached alkyne palmitate. Following SDS-PAGE and Western blotting, palmitoylation of Wnt was detected using Streptavidin IR 800 dye. GFP detection served as a transfection and/or loading control.

In Vitro/In Vivo Drug Metabolism and Pharmacokinetics
Aqueous solubility: A total of 1 mL of buffer was added to 1 mg of compound and incubated for 24 hours (Bioshake iQ, 650 rpm, 25°C). Following filtration under positive pressure, compound concentration in solution was assessed by high-performance liquid chromatography coupled to an ultraviolet detector (LC-UV).
Lipophilicity (logD 7.4 ): A total of 30 μL of a 20 mmol/L stock solution of RXC004 was added to a vial containing 1.5 mL of pH7.4 phosphate buffer saturated octanol and 1.5 mL of octanol-saturated pH7.4 phosphate buffer, shaken and centrifuged to ensure phase separation. logD 7.4 was calculated as the logarithm of the ratio of the concentration of RXC004 (measured by LC-UV in octanol vs. buffer phase). pK a s were determined using the fast UV (spectrometric) technique. Samples were titrated under methanol-water mixture between pH2.0 and pH12.0 at 30-15 μmol/L at 25°C. The ionic strength was 0.192 mol/L (KCl). Aqueous pK a s were determined per Yasuda-Shedlovsky (17).
Plasma protein binding: Mouse, rat, dog, and human plasmas were obtained from TCS Biosciences Ltd and Sera Laboratories Ltd. A total of 20 μmol/L RXC004 in 10% plasma diluted with phosphate buffer was dialyzed against phosphate buffer for 18 hours at 37°C. The concentration of RXC004 each side of the dialysis plate was determined by LC/MS-MS. Results were corrected to 100% plasma.
Cytochrome P450 (CYP) assays were performed using human liver microsomes incubated with P450 specific substrates as recommended by the FDA (19) at a concentration equivalent to K m in the presence of 0.03-30 μmol/L RXC004, or solvent control. Concentration of the metabolites was analyzed by LC/MS-MS.
Permeability of RXC004 was assessed in MDR1-MDCKII (Netherlands Cancer Institute) and Caco-2 cells (Merck Life Sciences). RXC004 (10 μmol/L) was added to either apical or basolateral chamber for 2 hours at 37°C, then an aliquot from the opposite side of the monolayer was removed and quantified by LC/MS-MS. Apparent permeability (P app ) was calculated according to standard practice (20).
In vivo pharmacokinetics: RXC004 was dosed to male Sprague Dawley rats, male CD1 mice or dog (beagle) by intravenous and oral routes. Bioanalysis of blood and plasma samples was performed by LC/MS-MS using a Waters TQS or TQMS mass spectrometer. Pharmacokinetic information was derived using WinNonlin software (Certara).

In Vitro Profiling Across Genetically Defined Cancer Cell Lines
Cells were treated with a concentration-series of RXC004 for 5 days and proliferation measured using ATPLite

In Vitro Glucose Uptake
SNU-1411, HPAF-II, and HCT116 cells were treated in vitro for 48 hours with RXC004 (100 nmol/L), Latrunculin A (Abcam #ab144290; 500 nmol/L), or DMSO prior to analysis using the Glucose Uptake Glo Assay Kit (Promega #J1341). The ATPlite assay was used to control for cell viability, confirming RXC004 has no effect on proliferation at this 48 hours timepoint.
Mice were imaged prior to dosing, day 3 and day 7 after dosing. A 10 MBq injection of 18 FDG (PETNET), with a 45-minute washout was followed by a 20-minute PET scan using a Siemens Inveon PET scanner. Tissues removed on day 7 for biodistribution analysis were assessed using a PerkinElmer Wallac Wizard Gamma counter.

In Vitro Human Peripheral Blood Mononuclear Cell Assays
HPAF-II cells were incubated for 3 days with RXC004 (100 nmol/L) or DMSO control, washed and detached using Accutase and equal number of viable cells (as assessed by Trypan Blue exclusion) per treatment were reseeded in 96-well plates. Frozen human peripheral blood mononuclear cells (PBMC), originally collected by density gradient separation using Histopaque from four healthy donors, were thawed and cultured in complete media. After approximately 40 hours, nonadherent PBMCs were added to coculture with the HPAF-II cells, at a PBMC:HPAF-II ratio of 5:1. T-cell proliferation and cytokine production was stimulated using Immunocult CD3/CD28-tetramer (Stemcell Technologies #10971). For monoculture nonadherent PBMC experiments, T cells were stimulated using Immunocult in the presence or absence of recombinant Wnt3a (Biotechne #5036-WN-010) and/or recombinant Wnt5a (Biotechne #645-WN-010), both at 100 ng/mL. After 5 days, supernatants from monoculture or coculture experiments were removed for analysis, whereas suspension cells were fixed, stained with anti-Ki67 (BD #563462), anti-CD4 (BioLegend #317420), and anti-CD8 (BD #564116) antibodies, before flow cytometry was performed on a Novocyte flow cytometer. Cytokines and chemokines in the supernatant were quantified using a 30-plex panel on Magpix (Thermo Fisher Scientific #LHC6003M) or ELISA for IFNγ (R&D Systems # DY285B-05).

Statistical Analysis
Unless stated otherwise, drug metabolism and pharmacokinetic (DMPK) data analysis was performed using Excel (Microsoft). All other statistical analyses were performed using Prism software (GraphPad Prism). For each analysis, residual plots were examined and where necessary a log transformation was applied prior to analysis. Where indicated, IC 50 values were determined using 4-parameter curve fitting within Prism software.

Data Availability Statement
The data generated in this study are available upon request from the corresponding author.

Structure and In Vitro Characterization of RXC004
A rational design approach was used to identify potent novel Porcupine inhibitors. Routine in vitro screening allowed for the elucidation of structure activity relationships which ultimately resulted in the discovery pmol/L. Luciferase activity was restored by the addition of recombinant Wnt3a, demonstrating that RXC004 had no effect on downstream Wnt signaling (Fig. 1F).

In Vitro and In Vivo DMPK
The physicochemical profile of RXC004 is shown in Table 1. The solubility of RXC004 was 56 μmol/L in pH 7.4 buffer, with a logD (pH7.4) of 2.4. pK a s were 3.8 (most basic) and 11.2 (most acidic), meaning that the predominant species at pH 7.4 is the neutral, nonionized form of the molecule. The average percentage free of RXC004 in plasma across species ranged from 2.5% to 7.5%. Microsomal CL int values ranged from 3.9 to 31.6 μL · minute −1 · mg −1 , with mouse having the lowest and dog the highest predicted clearances. Hepatocyte CL int values ranged from < 3 to 12.5 μL · minute −1 · 10 6 cells, with dog displaying moderate clearance while rodents and humans display low clearance. RXC004 has good intrinsic permeability, showing some evidence of efflux in MDR1-MDCKII cells but not in Caco-2 cells.
RXC004 displays in vitro absorption, distribution, metabolism, and excretion properties which are predictive of the low metabolic clearance and good oral bioavailability, as successfully observed in vivo in mouse, rat, and dog (Supplementary Table S1). Volume of distribution in vivo was low, resulting in a short in vivo half-life in preclinical species (Supplementary Table S1). Low concentrations of an acidic metabolite and hydroxylated versions of RXC004 have been identified from microsomes and hepatocytes of mouse, rat, dog, and human. In recombinant CYP enzyme assays, RXC004 was exclusively metabolized by CYP3A4. RXC004 does not inhibit any of the five major CYPs at concentrations of biological relevance (Table 1).
RXC004 inhibited proliferation of cell lines with RNF mutations and RSPO fusions in a concentration-dependent manner but had no effect on cell lines with mutations in downstream Wnt signaling (Fig. 2D). Effects of RXC004 on proliferation were mirrored by a concentration-dependent downregulation of c-Myc mRNA ( Supplementary Fig. S1E). Further analysis demonstrated that RXC004 reduced the proportion of cells in S-phase, and strongly inhibited expression of the mitosis marker phospho-histone-H3 in cells with upstream aberrations in Wnt pathway components (Fig. 2E-H), indicative of cell-cycle arrest. RXC004 did not induce cell death, as confirmed by lack of significant effect on Caspase 3/7 activity in SNU-1411 and HPAF-II cell lines ( Supplementary  Fig. S1F and S1G), and as evidenced by the delayed inhibition of cell growth by IncuCyte growth curve analysis ( Supplementary Fig. S1H and S1I).

Effects of RXC004 on Tumor Growth In Vivo
In vivo xenografts models of SNU-1411, AsPC1, HPAF-II, and HCT116 cell lines were used to test a single dose of RXC004 (1.5 mg/kg twice daily). A reduction in tumor growth, and inhibition of Wnt-responsive gene expression including c-Myc, was observed in the Wnt ligand-dependent SNU-1411, AsPC1, and HPAF-II models (Supplementary Fig. S2). Moreover, mRNA for MUC5AC and MUC4 were upregulated, consistent with increased epithelial differentiation. No effect of RXC004 on tumor growth was seen in the Wnt ligand-independent HCT116 xenograft model (Supplementary Fig. S2P).
Further efficacious dose schedules and pharmacokinetic/pharmacodynamic relationships were explored in the colorectal cancer RSPO-fusion model SNU-1411. Significant reduction in tumor volume was observed at various doses of RXC004, including 1.5 and 5 mg/kg continuous once daily dosing, and 1.5 mg/kg twice daily dosing either continuous or on and a 5-day on/2-day off dosing schedule (Fig. 3A). In a separate 7-day study, pharmacokinetic analysis showed good dose proportionality of RXC004 levels in the plasma and tumor following oral administration of 1.5 or 5 mg/kg RXC004 once daily (Supplementary Fig. S3). Inhibition of c-Myc gene expression in tumor was observed at both doses ( Fig. 3B and C), with more prolonged inhibition at the 5 mg/kg dose, which maintained free plasma concentrations above that required for approximately 10 × IC 50 cover over 24 hours (Fig. 3C). Similar effects were observed on Axin2, RNF43, and CD44, while MUC4 and MUC5A mRNA increased in a dose-dependent manner ( Supplementary Fig. S3). Consistent with this, increased tumor differentiation and increased mucin levels (as detected by Alcian-blue/PAS staining) were observed within tumor sections of RXC004treated tumors, with intermediate effects at 1.5 mg/kg once daily and stronger effects at 5 mg/kg once daily ( Fig. 3D; Supplementary Fig. S4A and S4B). Ki67positive cells were reduced by RXC004 in terms of the total tumor area, and more significantly in areas of differentiated tumor ( Supplementary Fig. S4C).
Reimplantation of SNU-1411 tumors from RXC004-treated mice into naïve, untreated animals demonstrated a prolonged impact on tumor growth of the prior RXC004 treatment (5 mg/kg once daily) compared with vehicle pretreated controls (Fig. 3E).
To establish a therapeutic window with RXC004, mice were dosed for 28 consecutive days with RXC004 at 1.5 mg/kg twice daily or 5 mg/kg (once daily or twice daily). Disruption of villi architecture (villi clubbing, blunting, or fusion) was observed only in some animals dosed at 5 mg/kg twice daily. No effects were seen at either the 1.5 mg/kg twice daily or the 5 mg/kg once daily doses previously demonstrated to inhibit tumor growth and Wnt-responsive gene expression (Fig. 3F). Moreover, there was no effect on the Ki67 proliferation marker in sections of intestinal crypts at 1.5 mg/kg twice daily, but a clear decrease at 5 mg/kg twice daily which was associated with a loss of body weight ( Supplementary Fig. S4D-S4G).

RXC004 Suppression of Tumor Metabolism in Genetically Selected Cancer Models
In vitro, RXC004 inhibited glucose uptake by over 50% in the Wnt liganddependent cell models SNU-1411 and HPAF-II cells, but had no effect in the AACRJournals.org Cancer Res Commun; 2(9) September 2022 Wnt ligand-independent HCT116 cell line (Fig. 4A-C). At this relatively early 48-hour timepoint, RXC004 had no effect on cell numbers as assessed by ATPlite or by IncuCyte growth curve analysis (Supplementary Fig. S1H and S1I). The positive control Latrunculin A blocked glucose uptake in all three cell lines.
In vivo 18

RXC004 Effects on Immune Evasion in the B16F10 "Cold" Tumor Model
The B16F10/C57BL/6 melanoma syngeneic model is immune "cold" and nonresponsive to immune checkpoint therapy. We confirmed anti-PD-1 treatment had no effect on tumor volume, whereas significant tumor growth inhibition versus control animals was seen in response to RXC004 as either a mg/kg twice daily, 5 mg/kg once daily, or 5 mg/kg twice daily. Loss of villi architecture observed at 5 mg/kg twice daily only. Statistical comparison; ordinary one-way ANOVA (A) or unpaired t test (E). P < 0.05 denoted by "*," P < 0.01 denoted by "**." monotherapy or in combination with anti-PD-1 ( Fig. 5A and B, bodyweight data for the SNU1411 model and the B16F10 models are shown in Supplementary  Fig. S5A and S5B). Similar findings were seen in combination with anti-CTLA-4 treatment in this model ( Supplementary Fig. S6). RXC004 monotherapy also significantly increased survival in this B16F10/C57BL/6 model (Supplementary Fig. S5C). Moreover, RXC004 monotherapy efficacy was shown to be dose dependent and also achievable with a 5-day on/2-day off dosing schedule (Supplementary Fig. S6). When B16F10 cells were implanted into SCID-Beige mice, (an immune compromised strain), RXC004 treatment did not reduce tumor growth (Fig. 5C). This indicates that RXC004 exerts its antitumor effects in this model via suppressing immune evasion.
To confirm this proposed immune-related mechanism of action, flow cytometry studies were conducted in both B16F10 tumors and blood from tumor-bearing mice. RXC004 treatment significantly reduced tumorassociated MDSCs, with a concomitant increase in circulating MDSC in AACRJournals.org Cancer Res Commun; 2(9) September 2022 is indicated alongside individual datapoints. Statistics by unpaired t test. P < 0.05 denoted by "*" and P < 0.01 denoted by "**." B16F10 tumor-bearing mice ( Fig. 5D and E). Nanostring nCounter IO 360 analysis (23) indicated that RXC004 significantly increased the overall tumorassociated CD45 + immune cells, including dendritic cells, and increased tumor expression of the chemokine CXCL9 in this B16F10 model (Fig. 5F-H).

RXC004 Effects on the Anti-PD-1 Immune Response in the CT26 "Hot" Model
The CT26/BALBc colorectal syngeneic model is an immune "hot" model and, as expected, significant tumor growth inhibition was seen in response to anti-PD-1 treatment as a single agent (Fig. 6A and B). RXC004 monotherapy had limited effect on tumor growth in this model. Bodyweight data are shown in Supplementary Fig. S5D. However, analysis of CT26 tumor-infiltrating immune cells demonstrated a significant decrease in regulatory T cells (CD3 + ; CD4 + ; FOXP3 + ) when RXC004 was combined with anti-PD-1 treatment over control or either monotherapy. While a change in infiltrating CD8 + T cells was not observed, a significant increase in the CD8 + /regulatory T cell ratio was seen for the RXC004 plus anti-PD-1 combination versus control or either monotherapy ( Fig. 6C-E).

RXC004 Treatment of HPAF-II Cells Abrogates Their Immune Suppression In Vitro
In vitro coculture of control HPAF-II cancer cells with nonadherent human PBMCs significantly suppressed the CD3/CD28-induced proliferation of both CD8 and CD4 T cells (Fig. 7A and B), and also reduced levels of IFNγ and GMCSF ( Fig. 7C; Supplementary Fig. S7A), cytokines important in generating an active antitumor immune response (24,25). This ability of HPAF-II cells to inhibit T-cell proliferation and IFNγ/GMCSF release was fully reversed upon pretreatment of the cancer cells with RXC004 ( Fig. 7A-C;   Supplementary Fig. S7A). Furthermore, RXC004 pretreatment potentiated the HPAF-II-dependent release of the chemokine CXCL9 (Fig. 7D), while reducing the HPAF-II-dependent secretion of the angiogenesis promoting growth factor VEGF ( Supplementary Fig. S7B). In nonadherent PBMC monoculture, recombinant Wnt3a or Wnt5a, when treated alone or in combination, had no significant effect on CD4 or CD8 T-cell proliferation or cytokine release (Supplementary Fig. S7C-S7F), suggesting that RXC004 effects in the coculture were unlikely to be solely via a direct reduction of Wnt levels. We therefore hypothesize that RXC004 treatment of HPAF-II cancer cells, and their circulating MDSCs (E) in B16F10/C57BL/6 mice at day 19 of treatment with vehicle or RXC004 at 5 mg/kg once daily. Mean ± SEM is indicated alongside individual datapoints, n = 4-6. Effects of 5 mg/kg once daily RXC004 versus vehicle on CD45 + score (F), dendritic cell (DC) score (G), or CXCL9 mRNA (H) in B16F10 tumors from C57BL/6 mice by Nanostring. Data are mean + SEM, n = 4-8. Statistical comparison; ordinary one-way ANOVA (B) or unpaired t test (D-H). P < 0.05 denoted by "*", P < 0.01 denoted by "**" and P < 0.0001 as "****." DC: dendritic cells, MDSC; myeloid-derived suppressor cells.
resulting differentiation, has a secondary effect to reduce their immunosuppressive nature.

Discussion
Dysregulation of canonical and noncanonical Wnt pathways plays a pivotal role in many gastrointestinal cancers (26). To date, attempts to target these pathways clinically have been limited to biological approaches, including the anti-Frizzled-7 antibody vantictumab and the decoy Frizzled receptor ipafricept, or small-molecule compounds with poorly defined mechanisms of action.
Efficacy of the Frizzled-directed biologicals may be limited by redundancy in extracellular signaling, as these biologicals block only a subset of Wnt-Frizzled interactions. Furthermore, these clinical trials were performed in broad cancer patient populations, with no exploration of whether efficacy might be improved by preselection of patients with Wnt ligand-dependent disease. These biologicals have not performed well in the clinic to date, reporting doselimiting toxicities, which may be due in part to their use in combination with standard-of-care chemotherapy regimens in many of these trials (7).
RXC004 is a potent small-molecule inhibitor of Porcupine that blocks Wnt palmitoylation, secretion, and downstream signaling, having a range of AACRJournals.org Cancer Res Commun; 2(9) September 2022 Statistical comparisons by ordinary one-way ANOVA. P < 0.05 denoted by "*", P < 0.01 denoted by "**," P < 0.001 denoted by "***," and P < 0.0001 as "****." antitumor effects, as outlined in Fig. 7E. Direct tumor cell targeting was demonstrated by cell-cycle arrest, leading to reduced proliferation in vitro and inhibition of tumor growth in xenograft models. As predicted, antiproliferative effects were specific to models which retain a dependency on Wnt secretion, namely those driven by upstream Wnt pathway aberrations such as LoF mutations in RNF or activating fusions in RSPO/. This is consistent with findings with other Porcupine inhibitors (27)(28)(29).
To explore the mechanism behind antiproliferative effects, we showed that RXC004 induced epithelial differentiation in genetically defined cancer cell lines with upstream Wnt pathway aberrations, as indicated by upregulation of MUC4 and MUC5AC mRNA and increased levels of mucin protein within cancer xenografts in vivo. This is consistent with findings with ETC-159, another Porcupine inhibitor, which extensively remodeled the transcriptome of colorectal cancer xenografts with RSPO translocations, decreasing stem cell markers and promoting differentiation from adenocarcinoma to mucinous phenotype (30). Moreover, we demonstrated that, alongside increased differentiation, RXC004 reduced the glucose uptake in Wnt ligand-dependent cancer models, both in vitro and in vivo using 18 FDG tracer. This suggests an opportunity to assess clinical activity of RXC004 in patients with genetically selected cancer by monitoring FDG-PET imaging in the clinic.
RXC004 effects were long lasting; RXC004-treated tumors reimplanted into naïve mice showed reduced tumor growth even in the absence of drug. This is consistent with findings with ETC-159, where tumor regrowth was suppressed for up to 6 weeks after withdrawal of treatment (30) and might be explained by the prodifferentiation effects of porcupine inhibition in these models. Longlasting effects have also been observed in other situations were upstream Wnt signaling is blocked. Reengraftment of colorectal patient xenografts with RSPO fusions from mice treated with anti-RSPO3 was significantly reduced, and regrowth inhibited even in the absence of continued anti-RSPO3 treatment (31). These findings of prolonged benefit after treatments have ceased could allow for more flexible dosing schedules for patients.
Reduced T-cell infiltration is associated with poor prognosis in colorectal and pancreatic cancers (32,33) and canonical Wnt signaling is associated with downregulation of T-cell inflammation within the tumor microenvironment (34). Moreover, noncanonical Wnt5a has been shown to skew dendritic cells into a more tolerogenic state (35). We showed that, in coculture with human PBMCs, control RNF-mutant HPAF-II pancreatic adenocarcinoma cells inhibit the stimulation of GMCSF and IFNγ production as well as CD4 and CD8 T-cell proliferation, effects that were reversed by preincubation of the HPAF-II cells with RXC004. RXC004 also potentiated release of CXCL9 from these cocultures, and reduced HPAF-II-dependent VEGF production. RXC004 also reduced MDSCs, while increasing overall immune infiltrate into the "cold tumor" B16F10 syngeneic model, which was confirmed to be nonresponsive to anti-PD-1 monotherapy, demonstrating a profound ability of RXC004 to reset aberrant immune responses, consistent with another recent study (36). In addition, CXCL9 mRNA expression was increased by RXC004 within the tumor microenvironment of this model. CXCL9 is a T-cell chemoattractant and its expression in tumors is linked to improved response to immune checkpoint inhibition in patients with cancer (37) and has also been associated with better prognosis in colorectal cancer (38).
The ability of RXC004 to inhibit tumor growth and enhance survival in the B16F10 model was dependent on a functional immune system, because neither effect was seen in B16F10 explants grown in immune-compromised SCID-Beige mice.
In the immune competent "hot tumor" CT26/BALBc syngeneic model, RXC004 synergized with anti-PD-1 treatment to decrease immunosuppressive regulatory T cells, and increase the ratio of cytotoxic CD8 T cells to regulatory T cells within the tumor. Higher CD8/regulatory T cell ratios in tumors have been shown to predict clinical benefit with anti-PD-1 therapy in patients with nonsmall cell lung cancer (39), hence our data support a rationale for combining RXC004 with anti-PD-1 therapies in the clinic.
Inducible ablation of β-catenin in the intestines of adult mice demonstrates the importance of canonical Wnt signaling in maintaining intestinal integrity (40). Given the pivotal role of Wnts in stem-cell renewal and homeostasis, the potential side effects of Porcupine inhibitors warrant investigation. High doses of Porcupine inhibitors have previously shown disruptive effects on villi architecture (41). We observed similar villi defects in mice at high doses of RXC004 but, importantly, profound effects of RXC004 on tumor growth and immune infiltrate were seen at lower doses that had no impact on intestinal homeostasis, confirming previous observations that a therapeutic window exists for this class of inhibitors (42). This window may be due to several factors; the high expression of drug exporters in stromal cells of the lower crypt, the ability of different populations of crypt stem cells to compensate for each other, and/or the existence of a Wnt gradient potentially protecting basal crypt stem cells from the effects of Porcupine inhibitors (43)(44)(45). Moreover, the ability of the intestinal epithelium to recover from insult has been well established; for example, restoration of functional APC in animal colorectal cancer models led to resumption of normal crypt-villus homeostasis, with aberrantly proliferating cells reacquiring self-renewal and multilineage differentiation capability (46).
Several Porcupine inhibitors, including RXC004, are currently being investigated in the clinic. These include WNT974 (previously LGK974), CGX1321, ETC-159, and XNW7201. Encouragingly, there has been a lack of widespread intestinal adverse events published to date. The most consistent side effects include loss of bone density and associated fractures, and dysgeusia (47), the latter likely due the role of Wnts in the renewal of taste buds (48). Bone effects of Porcupine inhibitors can be mitigated in animals by prophylactic treatment with RANKL inhibitors or bisphosphonates (49), an approach that has also proved effective in the clinic (50). An advantage of small-molecule inhibitors such as RXC004 compared with biologicals such as ipafricept and vantictumab is the ability to explore subtly different dosing strategies. It is interesting to note the differences in dosing regimens currently being explored with Porcupine inhibitors, ranging from priming dose only for WNT974 (when in combination with an anti-PD-1 therapy), to once daily (RXC004), or once every 2 days (ETC-159; refs. 51-53). The impact of these different dosing schedules on efficacy and safety is yet to be determined.
Our observations that Porcupine inhibitors are effective in a subset of genetically defined cancer models suggest that prospective selection of patients with Wnt ligand-dependent cancers will be required. Examples include patients with colorectal or pancreatic cancer with RSPO/ fusions and LoF RNF mutations. Because of these genetically selected patient tumors likely being dependent on Wnt ligand for survival and growth, it follows that high Wnt ligand levels in the tumor microenvironment could also be responsible for increased immunosuppression. This may increase susceptibility of such patient tumors to the positive immune effects of Porcupine inhibitors such as RXC004. The synergistic effect between RXC004 and anti-PD-1 on immune infiltrate in the "immune hot" CT26 model and the reversal of immune suppression in the "immune-cold" B16F10 model supports investigating the clinical activity of Porcupine inhibitors in combination with anti-PD-1 therapy, a strategy currently being explored (47). Investigation of Porcupine inhibitors in other gastrointestinal cancers may also be of interest. For example, in biliary cancer, Wnt ligands are reported to be highly expressed and associated with poor prognosis. In addition there are reports of hypermethylation-induced silencing of extracellular Wnt pathway antagonists (54,55).
In summary, the Porcupine inhibitor RXC004 shows potent, antiproliferative, prodifferentiation, and immune-stimulating effects in colorectal, pancreatic, and syngeneic cancer models at doses that have no effect on normal intestinal integrity. RXC004 is currently undergoing clinical trials in patients with cancer as a monotherapy and in combination with anti-PD-1 therapy.