The protein tyrosine phosphatase SHP2 binds to phosphorylated signaling motifs on regulatory immunoreceptors including PD-1, but its functional role in tumor immunity is unclear. Using preclinical models, we show that RMC-4550, an allosteric inhibitor of SHP2, induces antitumor immunity, with effects equivalent to or greater than those resulting from checkpoint blockade. In the tumor microenvironment, inhibition of SHP2 modulated T-cell infiltrates similar to checkpoint blockade. In addition, RMC-4550 drove direct, selective depletion of protumorigenic M2 macrophages via attenuation of CSF1 receptor signaling and increased M1 macrophages via a mechanism independent of CD8+ T cells or IFNγ. These dramatic shifts in polarized macrophage populations in favor of antitumor immunity were not seen with checkpoint blockade. Consistent with a pleiotropic mechanism of action, RMC-4550 in combination with either checkpoint or CSF1R blockade caused additive antitumor activity with complete tumor regressions in some mice; tumors intrinsically sensitive to SHP2 inhibition or checkpoint blockade were particularly susceptible. Our preclinical findings demonstrate that SHP2 thus plays a multifaceted role in inducing immune suppression in the tumor microenvironment, through both targeted inhibition of RAS pathway–dependent tumor growth and liberation of antitumor immune responses. Furthermore, these data suggest that inhibition of SHP2 is a promising investigational therapeutic approach.
Inhibition of SHP2 causes direct and selective depletion of protumorigenic M2 macrophages and promotes antitumor immunity, highlighting an investigational therapeutic approach for some RAS pathway–driven cancers.
Allosteric inhibition of the protein tyrosine phosphatase SHP2 (encoded by PTPN11), an established signaling node in the RAS–MAPK growth and survival pathway, is a novel, investigational therapeutic strategy for patients bearing tumors with specific oncogenic mutations in this pathway (1–4). SHP2 is a positive transducer of receptor tyrosine kinase (RTK) signaling [see Frankson and colleagues for a recent (5) review], but the molecular mechanism is still unclear. We and others have shown that SHP2 acts upstream of RAS and promotes RTK-mediated RAS nucleotide exchange and activation, likely through a scaffolding interaction with SOS1 (1, 6, 7). Clinical studies with investigational SHP2 inhibitors are ongoing, and preliminary signs of clinical activity in patients with non–small cell lung cancer (NSCLC) harboring KRAS mutations, particularly KRASG12C, have been reported (8). SHP2 is also widely expressed in hematopoietic cells, including both lymphoid and myeloid cells, and there is emerging evidence to support a role in antitumor immunity. The majority of reported studies have focused on establishing a role for SHP2 in the regulation of T-cell function (9–11), although recently myeloid-restricted deletion of SHP2 in mice was shown to suppress melanoma growth (12). Tumor-associated myeloid cell infiltration is associated with clinical resistance to immunotherapies (13) and correlates with a negative prognosis for most tumor types (14–22). Identification of therapeutic strategies that can modulate the recruitment, survival, and/or reprograming of tumor-associated macrophages (TAM) and improve the clinical response to currently available immunotherapies is critical (23). Building a comprehensive understanding of the impact, if any, of allosteric inhibition of SHP2 on innate and adaptive immunity, and how this can influence the clinical response to checkpoint blockade, is fundamental to realizing the full potential of this molecular targeted therapeutic strategy.
SHP2 may also be an important signaling node downstream of inhibitory receptors in immune cells. SHP2 binds to tandem phosphorylated immunoreceptor tyrosine-based inhibition motif (ITIM) and immunoreceptor tyrosine-based switch motif (ITSM) domains on regulatory receptors in immune cells, including inhibitory immune receptors like PD-1 and BTLA (24–26), and multiple reports have demonstrated a SHP2/PD-1 physical interaction in vitro (25, 27–33). Regulation of T-cell receptor signaling in vitro by SHP2 association with CTLA4 has also been reported (34), although canonical ITIM/ITSM domains are not present in CTLA4, so the significance of these reported associations is unclear (35). More recently, through the application of cell-free biochemical experiments, it has been proposed that SHP2 transduces the PD-1–inhibitory checkpoint signal by direct dephosphorylation of the costimulatory molecules CD28 and CD226 and, consequently, limits T-cell activation (28, 36). Collectively, these studies have pointed to a role for SHP2 in regulation of T cells. However, using a T-cell–specific SHP2-deficient mouse model, Rota and colleagues concluded that SHP2 is dispensable for PD-1 signaling in T cells in vivo, as well as for the global induction of T-cell exhaustion (11), a process that PD-1 has been implicated in controlling. Furthermore, the control of immunogenic tumors was not improved in these T-cell SHP2–deficient mice, and the response to anti–PD-1 checkpoint blockade therapy was not affected (11). One plausible explanation for the apparent discrepancy between these observations is that redundant mechanisms, such as the related tyrosine phosphatase SHP1, can mediate PD-1–inhibitory signaling in the setting of SHP2 deficiency (37). The emergence of these types of compensatory signaling mechanisms highlights the limitations of using genetically engineered mouse models to interrogate the in vivo mechanism(s) of action of SHP2. Moreover, the selective deletion of SHP2 protein from only a subset of immune cells obscures the clinical implications of the findings thus far, as it does not appropriately model the effects of pharmacologic inhibition of SHP2 broadly in multiple immune cell types in addition to tumor cells.
The recent availability of selective, orally-bioavailable small-molecule allosteric inhibitors of SHP2 provides an opportunity to interrogate the immunomodulatory mechanism(s) of action of SHP2 in vivo using pharmacologic tools that circumvent the various limitations imposed by genetic approaches. Accordingly, Zhao and colleagues have reported that a selective, but low potency, small-molecule inhibitor of SHP2 decreases tumor burden by augmenting cytotoxic T-cell–mediated antitumor immunity (9). However, in this study no evidence was provided to substantiate a direct effect of the SHP2 inhibitor on T cells in vivo, and further, the impact of SHP2 inhibition on the myeloid compartment was not evaluated (9).
In this study, we used the previously described potent and selective allosteric inhibitor of SHP2, RMC-4550 (1), to generate an in-depth understanding of the integrated effects of SHP2 inhibition in vivo in the tumor microenvironment. Using syngeneic mouse tumor models, we reveal an unanticipated impact of SHP2 on tumor immunity through modulation of both innate and adaptive immune cells. Similar to immune checkpoint blockade, RMC-4550 caused an increase in CD8+ T-cell tumor infiltrates. RMC-4550 also produced a direct and selective depletion of protumorigenic M2 macrophages through attenuation of CSF1 receptor (CSF1R) signaling. The antitumor effects of RMC-4550 were additive with either immune checkpoint inhibitors or an anti–CSF1R antibody, consistent with a pleiotropic role for SHP2 in the tumor microenvironment. Tumors that are intrinsically sensitive to SHP2 inhibition and also sensitive to checkpoint blockade were particularly susceptible to RMC-4550 alone or the combination treatment.
Collectively, these findings highlight that SHP2 inhibition is a promising molecular therapeutic strategy in cancer with potential dual activity: targeted suppression of tumor-intrinsic RAS/MAPK-dependent growth and promotion of antitumor immune responses through transformation of the suppressive tumor immune microenvironment. Translation of the preclinical combination advantages of a SHP2 inhibitor and checkpoint blockade into the clinical setting would be a significant advance for patients bearing oncogenic RAS pathway alterations and for whom current therapeutic options and benefits are limited.
Materials and Methods
Cell lines and reagents
All cell lines were obtained from the ATCC except for MC38 (NTCCChina). Cells were grown in RPMI (CT26.WT, A20, and 4T1) or DMEM (MC38, EMT6, B16-F10) supplemented with 10% heat-inactivated FBS and 1% penicillin/streptomycin (Gibco). Cells were maintained at 37°C in a humidified incubator at 5% CO2. All cells were mycoplasma free, and identity was confirmed by short tandem repeat profile. Antibodies used for in vivo treatment were from BioXcell: anti–PD-L1 (10F.9G2), rat IgG2b (LFT-2), anti-PD1 (RMP1-14), rat IgG2a (2A3), anti–CTLA4 (9D9), mIgG2b (MPC11); anti–CSF1R (AFS98), rat IgG2a (2A3), anti-CD4 (GK1.5), and anti-CD8 (2.43).
In vivo tumor challenge
All studies were compliant with all relevant ethical regulations regarding animal research in accordance with approved Institutional Animal Care and Use Committee protocols at MI Bioresearch, Inc., WuXi Apptec, and HD Biosciences. Female (6 to 8 weeks old) immunocompetent mice were implanted subcutaneously with 5E+05 CT26.WT cells, 5E+05 A20 cells, or 5E+05 EMT6 cells (BALB/c, Envigo or BALB/c Rag2 ko/ko, Taconic), and 2E+05 MC38 or 5E+04 B16-F10 cells (C57BL/6J, SLAC Laboratory Animal Co., LTD.); or injected in the mammary fat pad with 5E+05 4T1 cells (BALB/c, Shanghai Lingchang Biological Technology Co., LTD.). Once tumors reached an average size of 48 to 90 mm3, administration of RMC-4550 (30 mg/kg, by daily oral administration) or vehicle (2% HPMC in 50 mmol/L sodium citrate buffer), anti–PD-1, anti–PD-L1 or anti–CTLA4 (10 mg/kg, by intraperitoneal administration every 3 days), or anti–CSF1R (2 mg per mouse on day 1 followed by 0.2 mg per mouse 16 days after cell implantation, by intraperitoneal administration), was initiated. Experiments with A20 or CT26 where anti–PD-L1 was investigated were not staged, and treatment started 3 days after cell implantation. The selective depletion of tissue macrophages by anti–CSF1R administration was confirmed in liver by flow cytometry.
In vivo immune cell depletion experiments
Treatment began on day 7 at an overall mean tumor burden of 72 mm3. Anti-CD4, anti-CD8, or combination were administered intraperitoneally (0.5 mg per mouse on days 7, 8, and 9 followed by 0.2 mg per mouse on days 13 and 17). RMC-4550 was administered orally (30 mg/kg daily during 21 days starting at day 9). Depletion of immune cells in blood was confirmed by flow cytometry.
Immune phenotyping studies in tumors
Treatment with anti–PD-L1 (10 mg/kg intraperitoneal on days 3, 6, 10, and 13), RMC-4550 (30 mg/kg, by oral daily on days 3 to 15), or combination started on day 3, and tumors were processed for analysis on day 16 after cell implantation. Treatment with anti–CTLA4 (10 mg/kg, intraperitoneal on days 7, 10, and 14), RMC-4550 (30 mg/kg, daily oral on days 7 to 15), or the combination started on day 7 (79 mm3 tumors), and tumors were processed for analysis on day 16. Tumors were dissociated into single-cell suspension (GentleMACS C tubes and tumor dissociations Kit from Milteny Biotec). Antibodies used included CD3 (145-2C11, Biolegend), CD4 (RM4-5, BD Biosciences), CD8a (53-6.7, BD Biosciences), CD45 (30-F11, Biolegend), CD25 (PC61, Biolegend), PD-1 (29F.1A12, Biolegend), FoxP3 (3G3, ThermoFisher) and MHC Class I (34-1-25, Biolegend), CD11b (M1/70, BD Biosciences), Ly6C (HK1.4, Biolegend), F4/80 (BM8, Biolegend), MHC Class II (proprietary from MI Bioresearch), CD45 (30-F11, BD Biosciences), CD206 (CO68C2, Biolegend), CD11c (N418, ThermoFisher), Ly6G (1A8, BD Biosciences), CD19 (1D3, BD Biosciences) and PD-L1 (B7H1, Biolegend), Ki67 (Biolegend, 16A8), and AH1 Dextramer (Immudex JG3294). ACK Lysing Buffer (Biolegend), Zombie Viability Dye (Biolegend), Fc blocking agent (anti-CD16/32, Biolegend), FoxP3 Fix/Perm kit (eBiosciences), AbC Total compensation (ThermoFisher), and cell staining buffer (BD Biosciences) were used. Samples were run in an Attune NxT flow cytometer.
IHC detection for CD8a, F4/80, in mouse paraffin-embedded tumors
Anti-mouse CD8a (Cell Signaling Technology, 98941, 1.6 μg/mL) or anti-F4/80 (Cell Signaling Technology, 70076, 1.4 μg/mL) rabbit monoclonal antibodies were used with citrate-based pH 6.2 Heat-Induced Epitope Retrieval. Sections (5 μm) were stained on the Biocare intelliPATH platform using the manufacturer's recommended settings. Antibody binding was detected with MACH4 HRP-polymer Detection System followed by IntelliPATH FLX DAB chromogen and IntelliPATH Hematoxylin kits. All reagents were from Biocare Medical. TissueScope LE whole slide scanner (Huron Digital Pathology), Huron Viewer software, and HALO Image Analysis software from Indica labs were used for analysis.
PD-1 NFAT Luciferase Reporter Assay
Engineered CHO-K1 cells (BPS Bioscience, 60536) were incubated overnight in RPMI medium supplemented with 10% heat-inactivated FBS and 1% penicillin/streptomycin (Gibco). Engineered Jurkat cells (BPS Bioscience, 60535) were preincubated with RMC-4550 or anti–PD-1 for 30 minutes and added to CHO-K1 cells. After 16 hours, One-Step Luciferase Assay system (BPS Bioscience, 60690) was added according to manufacturer's instructions, and luminescence was measured on an EnVision Multilabel Plate Reader (Perkin Elmer).
Staphylococcus aureus enterotoxin B superantigen T-cell stimulation assay
Human buffy coat was obtained from San Diego Blood Bank. Peripheral blood mononuclear cells (PBMC) were isolated using SpMate (Stemcells) and treated with anti–PD-1 (S228P, Invivogen), isotype control (Human IgG4, Invivogen), RMC-4550, or vehicle (0.2% DMSO) at concentrations indicated. Thirty minutes to 1 hour after compound treatment, cells were stimulated with staphylococcal enterotoxin B (SEB; 0.1 μg/mL, Toxin Technologies), followed by incubation in presence of SEB and compounds at 37°C with 5% CO2 incubator for 3 days. IL2 content was analyzed in supernatant by standard ELISA (Abcam) with Perkin Elmer Envision Microplate Reader.
Mixed lymphocyte reaction
Monocytes were isolated from fresh PBMC from healthy donors (EasySep monocyte enrichment kit, Stemcell). Monocytes were differentiated (3 days) and matured (3 days) into monocyte-derived dendritic cells (Mo-DC) by using Milteny Biotec reagents. Cells were immunophenotyped with CD14, CD209, and CD83, and purity was confirmed to be >90%. Responder CD3+ T cells were prepared from a different donor using a negative selection kit (Stemcell) to obtain untouched T cells. Cells were cocultured at a final ratio of T cells to Mo-DCs of 10:1. Anti–PD-1 (S228P, Invivogen), isotype control (Human IgG4, Invivogen), RMC-4550, or vehicle was incubated for 5 days, and supernatants were assessed for IFNγ by ELISA (Invitrogen).
In vitro studies with murine bone marrow–derived macrophages
Culture of bone marrow–derived macrophages (BMDM): BM was isolated from the femurs of BALB/c mice. BM cells were plated in complete Alpha-MEM media (Gibco) containing 10% heat-inactivated FBS (VWR) and 1% Pen-Strep (Corning) and supplemented with CSF1 (Peprotech) at 10 ng/mL or GMCSF (Peprotech) at 25 ng/mL. Growth inhibition and apoptosis assays: After 7 days of culture, BM cells were dissociated (Gibco Cell Dissociation Buffer) and plated with media containing CSF1 at 10 ng/mL or GMCSF at 25 ng/mL. After 3 hours, cells were treated with either RMC-4550 or BLZ-945 (Selleckchem). Cell proliferation was measured 72 hours after compound treatment using the CellTiterGlo reagent (Promega). Caspase activity was measured using the Caspase-Glo 3/7 kit (Promega) 48 hours after compound treatment. Polarization assay: BM cells were isolated as above and cultured in 10 ng/mL CSF1 for 6 days. The appropriate cytokine (R&D Systems) was added (M1: IFNγ: 20 ng/mL, LPS: 100 ng/mL M2: IL4: 20 ng/mL), and cells were cultured for an additional 24 hours. Polarized BMDMs were treated with compound in the presence of CSF1 and the appropriate cytokine and analyzed for cell proliferation and caspase-3/7 activity as described. Phosphorylated ERK (pERK) assay: BM cells were isolated as above and cultured in either CSF1 or GMCSF for 7 days. Growth factor was removed overnight, and cells were treated with compound for 1 hour. Cells were acutely stimulated with CSF1 at 100 ng/mL or GMCSF at 50 ng/mL for 10 minutes. pERK was analyzed using the AlphaLisa SureFire p-ERK1/2 Thr202/Tyr204 kit (PerkinElmer) according to the manufacturer's instructions. pAKT was assessed using the pAKT (Ser473) MSD kit (Meso Scale Diagnostics) according to the manufacturer's instructions.
Quantitative data are presented as the mean ± SD or the SEM, as specified in the figure legends. Statistical tests were performed using GraphPad Prism 7.0. Two-sided Student t tests were used for comparisons of the means of data between two groups, and one-way ANOVA with post hoc Tukey test was used for comparisons among multiple independent groups, unless otherwise specified. For animal studies, animals were randomized before treatments, and all animals treated were included for the analyses. P value < 0.05 was considered significant.
SHP2 inhibition induces antitumor immunity in vivo in checkpoint blockade–sensitive tumors
We first examined the antitumor efficacy of the SHP2 inhibitor RMC-4550 in three syngeneic tumor models that are partially sensitive to checkpoint blockade: A20 B-cell lymphoma and both MC38 and CT26 colon carcinomas. RMC-4550 had a modest effect on growth of A20 cells in 3D in vitro culture but did not reduce the viability of MC38 or CT26 cancer cells at concentrations achievable in vivo (IC50 = 2 μmol/L, >10 μmol/L, and 10 μmol/L, respectively; Supplementary Fig. S1A; ref. 1). RMC-4550 did inhibit RAS–MAPK signaling, as measured by pERK levels, in A20 and MC38 cells (IC50 of 4 μmol/L and 22 nmol/L, respectively, Supplementary Fig. S1A) but not in CT26 cells (IC50 > 10 μmol/L; Supplementary Fig. S1A). Repeated oral daily dosing of RMC-4550 at 30 mg/kg significantly slowed tumor growth in each of these models (Fig. 1A). No effect of RMC-4550 on CT26 tumor growth was observed in vivo when tumors were established in RAG2–deficient mice, which lack T and B lymphocytes and are thus immunocompromised (Fig. 1B). These data, together with the lack of in vitro effect on both viability and RAS–MAPK signaling in CT26 cells (Supplementary Fig. S1A), provided confidence that the observed efficacy in vivo was a function of SHP2-mediated effects on immune cells in the tumor microenvironment. To corroborate these findings, we demonstrated that RMC-4550 did not inhibit tumor growth in immunocompetent mice when both CD4+ and CD8+ T cells had been functionally depleted in vivo with blocking antibodies (Fig. 1C). CD8+ T-cell depletion alone completely abrogated the efficacy of RMC-4550, indicating that these immune effector cells are essential for SHP2 inhibitor–mediated antitumor immunity (Fig. 1C). Depletion of CD4+ T cells inhibited tumor growth in vehicle-treated mice, and no further inhibition of growth was apparent with RMC-4550 (Supplementary Fig. S1B).
SHP2 inhibition is additive in combination with a checkpoint inhibitor
RMC-4550 induced significant tumor growth inhibition of CT26 tumors that was superior to anti–PD-L1 (Fig. 1D) and comparable to anti–CTLA4 (Fig. 1E). The combination of RMC-4550 with anti–PD-L1 demonstrated robust antitumor benefit as evidenced by a significant increase in the time to reach endpoint tumor burden and by tumor regressions in 4 of 10 mice (Fig. 1D). In contrast, RMC-4550 or anti–PD-L1 treatment alone did not result in any tumor-free animals. Tumor-free survivors remained tumor-free for at least 40 days and, importantly, were resistant to tumor reimplantation, suggesting long-lasting adaptive immunity (Supplementary Fig. S1C). All treatments, including the combination, were well tolerated (Supplementary Fig. S1D). Similar effects were observed with anti–CTLA4, although the combination regimen was less well tolerated (Fig. 1E; Supplementary Fig. S1E).
SHP2 inhibition does not confer sensitivity to checkpoint blockade in PD-1 refractory models
We tested the hypothesis that combination treatment with a SHP2 inhibitor could confer sensitivity to checkpoint blockade using two syngeneic models that are refractory to anti–PD-1 treatment, 4T1 and B16-F10. 4T1 breast carcinoma is a RAS/MAPK-dependent syngeneic line sensitive to both MEK (trametinib, IC50 3D-growth = 5 nmol/L) and SHP2 (RMC-4550, IC50 3D-growth = 33 nmol/L, Supplementary Fig S1A) inhibition in vitro, whereas B16-F10 melanoma cells are insensitive to SHP2 inhibition in vitro (Supplementary Fig. S1A). RMC-4550 did not increase sensitivity to anti–PD-1 in vivo in either of these models, irrespective of whether the cells were sensitive (4T1, Supplementary Fig. S1F) or insensitive (B16-F10, Supplementary Fig. S1G) to the tumor-intrinsic effects of SHP2 inhibition on RAS/MAPK signaling.
SHP2 inhibition stimulates adaptive immunity
Analysis of the immune landscape of CT26 tumors demonstrated that RMC-4550 treatment increased the percentage of CD3+ T cells by 2-fold, from a baseline of 8% ± 3% of CD45+ tumor-infiltrating leukocytes (TIL; Supplementary Fig. S2A and S2B). CD8+ T-cell frequency was increased in tumors from RMC-4550–treated mice, whereas there was no change in CD4+ T or T regulatory (Treg) cell frequency (Fig. 2A and B). Furthermore, the CD8+ T cells expressed less of the inhibitory molecule PD-1 (Fig. 2B). The increase in CD8+ T-cell frequency was comparable with that observed with checkpoint blockade, and the combination of RMC-4550 with either anti–PD-L1 or anti–CTLA4 exhibited additivity (Fig. 2A and B). The combinatorial effect with anti–CTLA4 on CD8+ T-cell frequency was statistically significant (Fig. 2B). RMC-4550 and anti–CTLA4 treatment–evoked increases in CD8+ T cells were mostly localized to the border of the tumor, and although an increase relative to vehicle control was apparent for each of the single-agent treatment regimens, only in the case of the combination of RMC-4550 with checkpoint blockade did the increase reach statistical significance (Fig. 2C).
Following RMC-4550 treatment, a higher frequency of the CD8+ tumor-infiltrating T cells were specific for the tumor-associated antigen AH1 (analyzed by dextramer staining; Supplementary Fig. S2C) and exhibited an activated profile, as they were proliferative (Ki67 staining) and expressed the cytotoxic cytokine IFNγ (intracellular staining; Supplementary Fig. S2C). These changes did not reach statistical significance but collectively are consistent with functional activation of the CD8+ tumor-infiltrating T cells.
RMC-4550 also increased the expression of class I MHC molecules and PD-L1 in CD45-negative tumor cells, similarly to anti–CTLA4 (Fig. 2D). These effects were dependent on IFNγ and CD8+ T cells, as they were abrogated by depletion with the corresponding antibodies in vivo (Fig. 2D). Consistent with the lack of intrinsic effects of RMC-4550 on proliferation of CT26 cells in vitro, the proliferation of CT26 cells in vivo was not affected, as measured by Ki67 staining (Fig S2D). Finally, consistent with previous report (38), SHP2 inhibition did not affect CT26 tumor vascularization as analyzed by CD31 tissue staining (Supplementary Fig. S2E).
SHP2 inhibition does not phenocopy the effect of anti–PD-1 in T cells
The direct effects, if any, of SHP2 inhibition on T cells were explored in vitro and in vivo and were compared with those of immune checkpoint blockade. Focusing on the proposed role of SHP2 as a downstream transducer of PD-L1/PD-1 signaling (25, 28), we obtained robust biochemical evidence that the tandem phosphorylated ITIM and ITSM in PD-1 can activate the autoinhibited form of SHP2. Titration of purified full-length SHP2 with a synthetic peptide that mimics the PD-1 tandem phosphorylated ITIM/ITSM increased enzyme activity by 270-fold (EC50 of 3.2 nmol/L; Supplementary Fig. S2F). The PD-1 peptide (10 nmol/L) induced activation of the autoinhibited form of SHP2 was blocked by RMC-4550 with an IC50 of 7.1 nmol/L. To monitor PD-L1/PD-1 signal transduction in a cellular context, we used a bioluminescent reporter assay in Jurkat T cells. These Jurkat cells were engineered to express human PD-1 and a luciferase reporter driven by an NFAT response element, and were cocultured with a variant of CHO cells that can serve as antigen-presenting cells (APC). These APCs are CHO-K1 cells expressing human PD-L1 and an engineered cell surface protein designed to activate cognate T-cell receptors in an antigen-independent manner. RMC-4550 caused a concentration-dependent activation of the NFAT luciferase reporter with an apparent potency (EC50 = 3.5 nmol/L) consistent with an on-mechanism effect for the SHP2 inhibitor. However, the maximal signal induction was approximately 4-fold lower than that observed with anti–PD-1 (Fig. 2E). In human PBMC cultures, both RMC-4550 and anti–PD-1 enhanced IL2 secretion in response to the superantigen SEB, but the response to RMC-4550 was not prominent compared with that of anti–PD-1 (Fig. 2F, top). Furthermore, RMC-4550 (up to a test concentration of 5 μmol/L) did not elicit a response in human T cells during a mixed lymphocyte reaction, whereas anti–PD-1 produced a robust increase in IFNγ release (Fig. 2F, bottom).
Given the equivocal findings in vitro, we elected to use an in vivo model to investigate the role of SHP2 in PD-1 signaling. Checkpoint blockade has been shown to reduce CD8+ T-cell exhaustion in different systems including the lymphocytic choriomeningitis virus (LCMV) infection mouse model (39). A role for CD28 costimulation in CD8+ T-cell rescue in the LCMV model has been confirmed (40), and we used this model to determine whether SHP2 inhibition mimics the effects of anti–PD-1 on T-cell exhaustion and viral load reductions in vivo. In our study, mice were challenged with LCMV clone 13 to establish a chronic infection, followed by administration of RMC-4550 or anti–PD-L1. RMC-4550 induced a significant increase in the frequency of CD8+ T cells in the spleen, but this was not accompanied by a significant increase in antigen-specific CD8+ T cells. Ultimately, RMC-4550 failed to decrease viral titers in peripheral organs (Supplementary Fig. S2G). In contrast, anti–PD-L1 treatment did effectively increase antigenic CD8+ T cells, correlating with higher viral control in various organs (Supplementary Fig. S2G).
In summary, although we cannot rule out a role for SHP2 downstream of PD-1 signaling, we have demonstrated, using various model systems in vitro and in vivo, that SHP2 inhibition and PD-1 blockade are not equivalent with respect to direct modulation of T-cell function. Rather, it seems likely that SHP2 inhibition can restrain PD-1 signaling to some extent, but that the full downstream effects are blunted, potentially due to the recruitment of redundant signaling effector molecules.
SHP2 inhibition modulates innate immunity, an effect not seen with checkpoint blockade
To explore additional mechanisms of SHP2 inhibitor action in vivo beyond transduction of checkpoint signals, we focused on myeloid cells in the tumor microenvironment. CT26 tumors are rich in myeloid cells; CD11b+ cells constitute 79%±2% of CD45+ TILs and 64% ± 2% of those are F4/80+ TAMs. RMC-4550 treatment had a striking impact on tumor myeloid infiltrates, in particular macrophages, inducing a 3-fold decrease in the frequency of F4/80+ cells among CD45+ TILs (Fig. 3A–C; Supplementary Fig. S3A). This finding was confirmed by IHC staining (Fig. 3D); the decrease in macrophages was most evident in the core of the tumor (Fig. 3D; Supplementary Fig. S3B).
TAMs are highly plastic and can acquire different phenotypes in the tumor microenvironment ranging from proinflammatory M1 TAMs (MHCIIhigh and CD206negative/low) to protumorigenic M2 TAMs (MHCIIlow and CD206high; ref. 41). RMC-4550 induced a significant decrease in the frequency of M2, the predominant population in CT26 tumors (>90% of TAMs, Fig. 3A–C), and an increase in the frequency of M1 among CD45+ TILs; by extension, the M2/M1 ratio was dramatically reduced (Fig. 3B and C). Checkpoint blockade elicited only a modest effect on TAM frequencies (Fig. 3B and C), but the combination of checkpoint blockade and RMC-4550 drove an even deeper modulation of TAMs (Fig. 3B and C). Checkpoint blockade previously has been shown to modulate TAM frequencies indirectly, via modulation of CD8+ T-cell frequency and IFNγ secretion in the tumor microenvironment (42, 43). In contrast, RMC-4550–mediated modulation of M2-TAM frequencies was unchanged by depletion of effector cells or IFNγ cytokine (Fig. 3E). As expected, IFNγ or CD8+ T-cell depletion decreased the overall frequency of M1-TAMs; however, a significant RMC-4550–mediated increase was still apparent (Fig. 3E). The expression of MHCI in M1 and M2-TAM was significantly increased with RMC-4550 treatment, and this effect was dependent of IFNγ and CD8+ T cells (Fig. 3F). The expression of PD-L1 in tumor-associated macrophages was not changed by RMC-4550 treatment (Supplementary Fig. S3E).
Granulocytic myeloid-derived suppressor cells (gMDSC) and monocytic MDSC (mMDSC) accounted for 7% ± 1% and 14% ± 1% of CD45+ TILs, respectively. Treatment with RMC-4550 increased the frequency of mMDSCs but had no effect on gMDSCs (Fig. 3G and H). The expression of MHCI or PD-L1 in MDSC was not changed upon RMC-4550 treatment (Supplementary Fig. S3E). To explore potential functional consequences of a SHP2 inhibitor–mediated increase in mMDSC, we used an in vitro suppression assay. Coculture of human MDSC with T cells induced suppression of T-cell proliferation and IFNγ release (Supplementary Fig. S3F). RMC-4550 alone had no effect on T-cell proliferation or cytokine release (Supplementary Fig. S3F) but was able to block the antiproliferative effects of MDSCs on CD8+ T cells (Supplementary Fig. S3F). A concomitant concentration-dependent increase in IFNγ release was also observed (Supplementary Fig. S3F). The viability of MDSCs in vitro was not affected by RMC-4550 (92.5%–93.5% viable compared with 93.3% viable in DMSO-treated MDSCs, determined by flow cytometry).
The frequency of myeloid cells in spleen or peripheral blood of tumor-bearing mice was unchanged with RMC-4550 treatment, suggesting that myelopoiesis was not affected at this timepoint (Supplementary Fig. S3C and S3D).
In summary, SHP2 inhibition produces a marked shift in polarized macrophage populations in the tumor microenvironment in favor of antitumor immunity, an effect that was not observed upon checkpoint blockade. This selective effect of RMC-4550 on myeloid cells may underlie the combination benefit of a SHP2 inhibitor and checkpoint blockade on tumor growth inhibition (Fig. 1D and E).
SHP2 inhibition suppresses CSF1R signaling and selectively affects viability of M2 macrophages
The prominent reduction in macrophage frequency observed in vivo following administration of RMC-4550, and the apparent lack of dependence on effector lymphocytes or cytokines, is consistent with a direct effect of SHP2 inhibition on macrophage viability. To evaluate this possibility, BM cells from BALB/c mice were differentiated with CSF1 or GMCSF in vitro. CSF1 differentiated BMDMs represent a population of F4/80+, MHCIIlow, CD11c- macrophages, whereas GMCSF–differentiated BM cells are MHCIIhigh, CD11c+, and likely represent a mixture of macrophages and DCs (44). RMC-4550 potently inhibited the growth of CSF1–differentiated (IC50 = 13 nmol/L) but not GMCSF–differentiated (IC50 > 1 μmol/L) BM cells (Fig. 4A). In addition, SHP2 inhibition selectively induced caspase-3/7 activation, as a marker of apoptosis, in CSF1–differentiated BMDMs (EC50 = 2.8 nmol/L, Fig. 4B).
The CSF1R is an RTK that controls the survival and proliferation of macrophages (45) and is the target of several therapeutic agents in clinical development for cancer (46). The selective CSF1R kinase inhibitor BLZ945 (47) also showed selective growth inhibition and induction of apoptosis in CSF1–differentiated, but not GMCSF–differentiated, BMDMs (Supplementary Fig. S4A and S4B). The time course of growth inhibition by BLZ945 or RMC-4550 was similar, and comparable to that caused by CSF1 deprivation (Supplementary Fig. S4C). Given these observations, together with the well-established role of SHP2 as a positive signal transducer downstream of many RTKs, we hypothesized that SHP2 inhibition suppresses CSF1R signaling. Indeed, we observed strong inhibition of ERK 1/2 phosphorylation by RMC-4550 after acute stimulation of BMDMs with CSF1 (IC50 = 3 nmol/L, Fig. 4C). These results were recapitulated using a recombinant cell line that reports on CSF1R activation and signaling (Supplementary Fig. S4D). SHP2 inhibition also decreased GMCSF–induced ERK 1/2 phosphorylation, albeit to a lesser extent (IC50 = 93 nmol/L, Fig. 4C), an effect which was not observed with BLZ-945 (Supplementary Fig. S4E) and likely accounts for the moderate growth-inhibitory effect of RMC-4550 in these cells. Importantly, these in vitro results were recapitulated in monocytes purified from human PBMCs, with SHP2 inhibition resulting in decreased ERK 1/2 phosphorylation and potent inhibition of growth (IC50 = 35 nmol/L; Fig. 4D; Supplementary Fig. S4F). Moderate suppression of AKT phosphorylation, another important signaling node for survival downstream of CSF1R, was observed with RMC-4550 in human monocytes but not in murine BMDMs (Supplementary Fig. S4G–S4I).
BMDMs were polarized to either an M1 (IFNγ, LPS) or M2 (IL4) phenotype to explore the contribution of a selective intrinsic effect of SHP2 inhibition on M2 macrophages over M1 in vitro. The M1-polarized macrophages expressed higher levels of inducible nitric oxide synthase, whereas M2 polarization resulted in increased levels of CD206 and arginase (Supplementary Fig. S4J). M2 macrophage viability was sensitive to RMC-4550 (IC50 = 19 nmol/L), but M1-polarized macrophages remained almost entirely refractory to drug treatment (IC50 > 1 μmol/L; Fig. 4E). Similarly, SHP2 inhibition selectively induced caspase-3/7 activation in M2 but not M1 macrophages (Fig. 4F), which likely accounts for the dramatic decrease in M2 frequency observed in vivo (Fig. 3B).
We were unable to determine the impact of SHP2 inhibition on macrophage differentiation per se because RMC-4550 produced a significant decrease in monocyte viability when present during the differentiation, precluding robust phenotypic characterization of the differentiated cells.
Recent data have suggested that increased levels of IFNγ and TNFα in the tumor microenvironment, caused by infiltration of CD8+ T cells, can trigger an adaptive response of CSF1 production by certain cancer cells (48). This in turn can promote recruitment and proliferation of immunosuppressive TAMs, hampering the antitumor immune response to checkpoint inhibitors. Treatment of CT26 cells in vitro with IFNγ and TNFα did increase production of CSF1 mRNA (Supplementary Fig. S5A). However, we propose that the ability of RMC-4550 to inhibit CSF1R signaling and decrease immunosuppressive TAM populations, as shown herein, would negate any inhibitory effects of CSF1 release by tumor cells.
SHP2 inhibition exhibits greater antitumor activity relative to CSF1R inhibition in vivo
The contribution of SHP2-mediated blockade of the CSF1R signaling pathway to the antitumor efficacy of RMC-4550 in the CT26 model was examined by comparing the response with that of CSF1R blockade. Anti–CSF1R treatment, in contrast to RMC-4550, did not induce any significant tumor growth delay (Fig. 4G and H). These findings provide evidence that the in vivo antitumor immunomodulatory effects of a SHP2 inhibitor reflect more than modulation of the myeloid compartment alone.
The combination of anti–CSF1R and RMC-4550 showed additive antitumor effects in the CT26 model (Fig. 4G and H). Although unexpected, this result may reflect the differential mechanisms of inhibition of CSF1R signaling by these two agents. Activation of parallel signaling pathways downstream of CSF1R (e.g., PI3K/AKT) is insensitive to SHP2 blockade (Supplementary Fig. S4I), whereas direct receptor inhibition likely suppresses additional prosurvival signaling pathways. Given the role of both SHP2 and CSF1R as key signaling nodes in multiple cell types and the complexity of the tumor microenvironment in vivo, further studies are required to elucidate the precise mechanism(s) underlying this combinatorial effect.
SHP2 inhibition is additive in combination with checkpoint blockade in a SHP2 inhibitor–sensitive syngeneic model
The combined tumor-intrinsic and immune-mediated antitumor effects of SHP2 inhibition have not been reported. EMT6 breast carcinoma is a RAS/MAPK-dependent syngeneic line sensitive to both MEK (trametinib, IC50 3D-growth = 47 nmol/L) and SHP2 (RMC-4550, IC50 3D-growth = 100 nmol/L, Supplementary Fig. S1A) inhibition in vitro. RMC-4550 alone significantly inhibited growth of established EMT6 tumors in immunocompetent mice in vivo, an effect superior to that of anti–PD-1 (Fig. 5A). The combination of RMC-4550 and anti–PD-1 resulted in sustained tumor growth inhibition that greatly increased the time to reach endpoint (Fig. 5B). This treatment also led to tumor regressions in 20% of mice, which were resistant to tumor reimplantation, suggestive of long-lasting adaptive immunity (Supplementary Fig. S5B). Treatment of EMT6 tumor–bearing mice with RMC-4550 also induced a significant reduction in tumor cell proliferation, as measured by Ki67 staining, analyzed 9 days after treatment (Supplementary Fig. S5C). These data corroborate the findings of cell-intrinsic effects of RMC-4550 on proliferation of EMT6 cells in vitro (Supplementary Fig. S1A).
Based on the collective observations presented here, we propose a model in which the pleiotropic effects of SHP2 inhibition on both innate and adaptive immunity cooperate to enhance tumor cell elimination (Fig. 5C). This study reveals a direct role for SHP2 in supporting an immunosuppressive tumor microenvironment in addition to an impact on proinflammatory macrophages, although the mechanism underlying the effect on M1 macrophages is unclear. We have demonstrated that CD8+ T cells are obligatory for the antitumor activity of SHP2 inhibition; however, the underlying mechanistic driver(s) of the augmented adaptive immune response remains to be determined.
In the present study, we demonstrate that SHP2 inhibition promotes antitumor immunity by modulating both innate and adaptive immune cells. We propose that, although induction of antitumor immunity by SHP2 inhibition is T-cell–dependent, a major driver of the response is modulation of the macrophage compartment rather than a direct effect on T-cell signaling, thus differentiating SHP2 inhibition from checkpoint blockade. Our data support a model in which SHP2 inhibition has a direct impact on the viability of TAMs, thereby promoting a less immunosuppressive tumor microenvironment. An appreciation of the tumor-extrinsic immune-modulatory mechanisms of SHP2 should be instructive to the clinical evaluation of SHP2 allosteric inhibitors as a novel molecular therapeutic strategy in patients with cancer.
Consistent with the proposed role of SHP2 as a downstream transducer of PD-1 checkpoint signaling in T cells (25, 27–33, 36, 49), we have observed similarities between the in vivo responses to SHP2 inhibition and immune checkpoint blockade in the tumor immune microenvironment. We and others (9) have shown that an increase in tumor-infiltrating CD8+ T cells is essential for SHP2 inhibitor–mediated control of established tumor growth and that these T cells express less PD-1, suggesting that they are less exhausted in response to chronic antigen exposure. However, although we have found a general concordance between the responses to anti–PD-1 and SHP2 inhibition in various in vitro readouts of T-cell function, we have been unable to demonstrate that pharmacologic inhibition of SHP2 is equivalent to PD-1 blockade. In particular, the disparate magnitude of the responses suggests that SHP2 is not the sole effector of inhibitory PD-1 signaling in these model systems, as has been proposed previously (11, 37). The failure of RMC-4550 to phenocopy the effects of anti–PD-1 in the LCMV T-cell exhaustion model in vivo also points to a greater complexity in PD-1 signaling than has perhaps been appreciated thus far. In summary, although the present observations are consistent with a role for SHP2 in PD-1 signal transduction and T-cell biology, the precise role for SHP2 in this pathway vis a vis other redundant mechanisms has yet to be elucidated.
More striking is the enhancement of tumor growth inhibition that we and others (9) observe with the combination of global SHP2 inhibition and checkpoint blockade; this is indicative of additional functions for SHP2 beyond checkpoint transduction in T cells. Significantly, we found that SHP2 inhibition had a profound impact on the survival and function of suppressive monocytic immune cells such as TAMs and MDSCs. Here, we demonstrate using a pharmacologic approach that SHP2 is a positive regulatory of ERK signaling downstream of CSF1R in human monocytes and murine BMDMs, which is in agreement with previous studies using genetic deletion of PTPN11 (50). The inhibition of CSF1R prosurvival signaling likely accounts for the selective effects on M2 macrophage populations, as has been observed previously with CSF1R inhibitors (51–54), and is supported by the in vitro experiments in the present study. The selective depletion of M2 macrophages in the tumor microenvironment after SHP2 inhibition, without major effects on the M1 population, has important translational implications. We did not observe effects of SHP2 inhibition on GMCSF–differentiated macrophages in vitro, suggesting that SHP2 does not play a role downstream of this receptor. The GMCSF receptor transduces prosurvival signals in M1 macrophages, which may be an explanation of why SHP2 inhibition spares M1 cells and instead has a selective effect on M2s. In addition to its role as a positive regulator of the RAS pathway, SHP2 has also been proposed to negatively regulate STAT1 activation downstream of IFNγ signaling (55–57). As IFNγ/STAT1 signaling is important in M1 macrophage activation (58), inhibition of SHP2 may be supporting a feed-forward loop for M1 macrophage polarization and survival, which encompasses not only macrophage-intrinsic effects on signaling, but is influenced by the infiltration of IFNγ-producing CD8+ T cells into the tumor. Consistent with this hypothesis, IFNγ or CD8+ T-cell depletion induced an overall decrease in M1 frequency, although a SHP2 inhibitor–mediated increase was still apparent. A role for SHP2 downstream of PD-1 in myeloid cells may also be possible. PD-1 signaling in myeloid cells can dampen antitumor immunity by regulating lineage fate commitment and function of myeloid cells (59). Myeloid-specific deletion of PD-1 in tumor-bearing mice resulted in a diminished accumulation of immature immunosuppressive cells and an enhanced output of differentiated, inflammatory effector mMDSCs, and phagocytic macrophages, a phenotype similar to that of SHP2 inhibition.
Adaptive responses to signals in the tumor microenvironment are not restricted to the immune compartment. There is compelling evidence that the infiltration of CD8+ T cells can induce production of CSF1 by melanoma cells and other cancers by secretion of IFNγ and TNF-α (48, 60), an effect we also observed in vitro in the colon CT26 model. Increased levels of CSF1 promote an increase in immunosuppressive M2 macrophages, via CSF1R activation, and a negative correlation with overall patient survival (48). The opposing effects of CD8+ T-cell infiltration induced by checkpoint blockade could be counteracted by combination with anti–CSF1R therapies in a murine melanoma model (48). Intriguingly, our results suggest that SHP2 inhibition has the potential both to induce CD8+ T-cell infiltration and simultaneously to counteract its negative consequences by suppressing CSF1R signaling and therefore contract the immunosuppressive macrophage population in the tumor microenvironment. This mechanism of action may contribute to the enhanced antitumor activity we observed with RMC-4550 in combination with checkpoint blockade. Correspondingly, it may account for the superior tumor growth inhibition observed with the SHP2 inhibitor relative to anti–CSF1R.
The potential for SHP2 inhibitors to provide therapeutic benefit in solid tumors bearing SHP2-sensitive oncoproteins, in particular in NSCLC, is the focus of intensive clinical investigation. Multiple, rational combination strategies for a SHP2 inhibitor with agents that target alternate nodes in the RAS–MAPK pathway [e.g., MEK (6), KRASG12C (61), or RTK (62) inhibitors] or extraproliferative functions of RAS (e.g., CDK4/6; ref. 63) have also been proposed. The present data provide a strong rationale for a clinical combination strategy with a SHP2 inhibitor and agents that target the immune system directly, such as anti–PD-1 and anti–CSF1R. Patients bearing tumors that harbor oncogenic driver mutations sensitive to SHP2, and with established clinical sensitivity to checkpoint inhibitors, for example KRASG12C-mutant NSCLC patients, could be particularly susceptible to this combination therapy. On the other hand, the present preclinical findings suggest that SHP2 inhibition seems unlikely to increase sensitivity to an immune checkpoint inhibitor in checkpoint resistant tumors.
In summary, we have shown using preclinical models that SHP2 plays a central role in inducing immune suppression in the tumor microenvironment both by inhibiting T cells and supporting the viability of protumorigenic macrophages. SHP2 inhibition is an attractive investigational therapeutic strategy with potential dual activity: targeted inhibition of RAS–MAPK-dependent tumor growth and liberation of antitumor immune responses by transformation of the tumor microenvironment.
Disclosure of Potential Conflicts of Interest
E. Quintana is Director at Revolution Medicines. C.J. Schulze is Senior Scientist at Revolution Medicines. D.R. Myers is Scientist at Revolution Medicines. T.J. Choy is Research Associate II at Revolution Medicines. K. Mordec is Senior Research Associate Scientist at Revolution Medicines. D. Wildes is Senior Director/Principal Scientist at Revolution Medicines. N. Tobvis Shifrin is Sr. Scientist at Revolution Medicines. A. Belwafa is Senior Research Associate at Revolution Medicines. E.S. Koltun is Senior Director Medicinal Chemistry at and has an ownership interest (including patents) in Revolution Medicines. A.L. Gill is Senior Vice President, Chemistry, at Revolution Medicines. M. Singh is Sr. Director at and has an ownership interest (including patents) in Revolution Medicines. S. Kelsey is President, Research and Development, at and has an ownership interest (including patents) in Revolution Medicines. M.A. Goldsmith is CEO at Revolution Medicines. R. Nichols is Director, Biology, at and has an ownership interest (including patents) in Revolution Medicines. J.A.M. Smith is Sr. Vice President, Biology, at Revolution Medicines. No potential conflicts of interest were disclosed by the other authors.
Conception and design: E. Quintana, C.J. Schulze, A.L. Gill, M. Singh, S. Kelsey, M.A. Goldsmith, J.A.M. Smith
Development of methodology: D.R. Myers, D. Wildes, A.L. Gill, J.A.M. Smith
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.J. Schulze, T.J. Choy, K. Mordec, D. Wildes, N. Tobvis Shifrin, A. Belwafa, E.S. Koltun, M. Singh, J.A.M. Smith
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Quintana, C.J. Schulze, D.R. Myers, T.J. Choy, K. Mordec, D. Wildes, A. Belwafa, S. Kelsey, M.A. Goldsmith, J.A.M. Smith
Writing, review, and/or revision of the manuscript: E. Quintana, C.J. Schulze, D.R. Myers, T.J. Choy, D. Wildes, N. Tobvis Shifrin, M. Singh, S. Kelsey, M.A. Goldsmith, R. Nichols, J.A.M. Smith
Study supervision: E. Quintana, J.A.M. Smith
Other (key tool compound design and synthesis): E.S. Koltun
We would like to thank Dylan Daniel, Art Weiss, and Cliff Lowell for providing expert advice during the course of this work. We would also like to thank the respective research teams at the following contract research organizations for the conduct of in vitro and in vivo studies: MI Bioresearch, HDB, WuXI Apptec, Ensigna, HistoTox Labs, and PAIRimmune, Inc. This work was supported in part by Sanofi.
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.