Overcoming Resistance to Checkpoint Blockade Therapy by Targeting PI3K-γ in Myeloid Cells

Abstract

Recent clinical trials using immunotherapy demonstrate its potential to control cancer by disinhibiting the immune system. Immune checkpoint blocking (ICB) antibodies such as anti-cytotoxic T-lymphocyte-associated protein 4 (anti-CTLA-4) or anti-Programmed cell death protein 1/anti-Programmed death-ligand 1 (anti-PD-1/anti-PD-L1)¹ have demonstrated durable clinical responses in various cancers. Although these new immunotherapies have significant impact on cancer treatment, multiple mechanisms of immune resistance exist in tumors. Among the key mechanisms, myeloid cells play a major role in limiting effective tumor immunity. ^2–4 Growing evidence suggests that high infiltration of immune-suppressive myeloid cells correlates with poor prognosis and ICB resistance. ^5,6 These observations suggest a need for a precision medicine approach where the design of the immunotherapeutic combinations are tailored based on tumor immune landscape to overcome such resistance mechanisms. Herein we employ a preclinical model system and show that resistance to ICB is directly mediated by the suppressive activity of infiltrating myeloid cells in various tumors. Furthermore, selective pharmacologic targeting of the gamma isoform of phosphoinositide 3-kinase (PI3K-γ), highly expressed in myeloid cells, restores sensitivity to ICB. We demonstrate that targeting PI3K–γ, with a selective inhibitor, currently being evaluated in a phase 1 clinical trial (NCT02637531), can reshape the tumor immune microenvironment and promote cytotoxic T cell-mediated tumor regression without targeting cancer cells directly. Our results introduce opportunities for new combination strategies using a selective small molecule PI3K-γ inhibitor, such as IPI-549, to overcome resistance to ICB in patients with high levels of suppressive myeloid cell infiltration in tumors.

Tumor-associated myeloid cells (TAMCs) constitute a major component of the tumor microenvironment. Although some controversies persist in the precise description of the distinct subsets of this heterogeneous population, it is accepted that these cells promote tumor immune-suppression.⁷ Recent studies support their contribution to the suppression of T cell function, which is not abolished by the use of ICB.^8–11 To understand the association between resistance to ICB and myeloid cell infiltration, we compared multiple mouse tumor models treated with ICB. We show that mice bearing 4T1 breast carcinoma are resistant to anti-PD-1 or anti-CTLA-4 therapy (Fig. 1a and Extended data Fig. 1). We observe that myeloid cells (CD11b⁺), constitute the majority of CD45⁺ tumor-infiltrating leukocytes (TILs) in this model (Fig. 1b). This correlates with reduced CD8⁺ T cell infiltration and cytolytic function (Fig. 1b,c). In contrast, B16-F10 melanoma tumors, which are more responsive to ICB (Fig. 1a and Extended data Fig. 1), are less infiltrated with myeloid cells but contain more activated CD8⁺ T cells (Fig. 1b). Additionally, CD8⁺ T cells express more granzyme B in the B16-F10 model. They also express higher levels of PD-1 and CTLA-4 (Fig. 1a–c, data not shown), which might explain their sensitivity to ICB. Furthermore, myeloid cells from 4T1 tumors or spleens suppress proliferation of T cells to a greater extent compared to myeloid cells from B16-F10 (Fig. 1d and Extended data Fig 1b). These data suggest that TAMCs have varying phenotypes and are more suppressive in ICB resistant tumors. Tumor-derived soluble factors such as granulocyte-macrophage colony-stimulating factor (GM-CSF) help shape the tumor microenvironment by promoting myelopoiesis and recruitment of suppressive myeloid cells.^12,13 To directly assess the ability of suppressive myeloid cells to induce resistance to ICB in the B16 melanoma, we used B16-F10 cells transduced with a GM-CSF expression construct (B16-GM-CSF) ¹⁴. Tumors in this B16 model (B16-GM-CSF) become infiltrated by suppressive TAMCs and lose sensitivity to ICB compared to B16-F10 controls (Fig. 1a–d and Extended data Fig. 1), indicating the critical role suppressive myeloid cells play in ICB resistance.

Figure 1. Resistance to checkpoint blockade is associated with suppressive Myeloid cells infiltration in tumor microenvironment.

a. Mean tumor volume of subcutaneous (4T1) or intradermal (B16, B16-GMCSF) implants in anti-PD-1, anti-CTLA4 or control treated mice (n=10), upper panel. Survival of 4T1, B16 or B16-GMCSF tumor bearing mice treated with anti-PD1 or anti-CTLA4 compared to control (vehicle treated only) (n=10), lower panel. b. Representative flow cytometric analysis and quantification of CD11b⁺, CD8⁺, CD4⁺ and Tregs (CD4⁺ Foxp3⁺) cell populations in 4T1, B16, B16-GMCSF tumors at 14 days post implants (n=5). c. Representative flow cytometric analysis and quantification of Granzyme B, PD-1 expression on CD8⁺ T cell populations in 4T1, B16, B16-GMCSF tumors at 7 days post implants (n=5). d. In vitro suppressive activity of tumor-infiltrating CD11b⁺ cells purified from 4T1, B16, B16-GMCSF tumor-bearing mice. Representative histograms of CD8⁺ T cell proliferation at corresponding CD11b⁺ to CD8⁺ T cell ratio (left panel) and quantification of CD8⁺T cell proliferation using CFSE dilution (right panel) (n=3). Data represent analysis of n (shown above for each experiment) mice per group, mean ± s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (non-parametric Mann–Whitney t-test, Log-rank (Mantel-Cox) test survival comparison).

Kaneda et al¹⁷ and previous studies^15,16, have shown that PI3K-γ is highly expressed in myeloid cells and promotes migration and production of inflammatory mediators. Mice deficient in p110γ sub-unit of PI3K-γ show reduced tumor growth and immunosuppressive TAMC.¹⁸ We reasoned that pharmacological PI3K-γ inhibition might lead to effective anti-tumor immune response and subvert ICB resistance caused by suppressive myeloid cells. To test this hypothesis, we used IPI-549, a selective PI3K-γ inhibitor in multiple tumor models¹⁹ (Extended data Fig. 2a). IPI-549 treatment alone led to tumor growth inhibition in 4T1, B16-GMCSF, MC38, CT26, and LLC tumor models (Fig. 2a and Extended data Fig. 2b). Lack of activity in tumors with low suppressive TAMCs (B16-F10, Fig. 2a) suggests PI3K-γ inhibition affects myeloid cells and has no direct effect on tumor cells or other TILs. We also observe a significant reduction in lung metastasis after IPI-549-treatment (Fig. 2b). TIL quantification shows no consistent differences in the total myeloid (CD11b⁺) or macrophage (CD11b⁺/F480⁺) cell populations in either IPI-549- or vehicle-treated 4T1 or B16-GMCSF tumors (Fig. 2c). The effects of IPI-549 on TAMC subsets were comparable to those seen in p110γ −/− mice^17,18. PI3K-γ inhibition switches the activation of macrophages from an immunosuppressive M2-like (CD11b⁺F4/80⁺CD206⁺) phenotype to a more inflammatory M1-like (CD11b⁺F4/80⁺MHCII⁺) state (Fig. 2c, Extended data Fig. 2c). We further tested RNA expression of M1 and M2 markers in 4T1 and B16-GMCSF tumors after IPI-549. The expression of prototypic M2-markers (TGF-β, Arg-1, IDO) are reduced, while M1-markers (IL-12, INOS) are higher in IPI-549-treated tumors (Fig. 2d). Given that CD11b⁺F4/80⁺ macrophages constitute only a portion of suppressive TAMCs, we further subdivided myeloid cells into granulocytes (CD11b⁺Ly6G⁺), monocytes (Ly6C^high/MHCII^low = Mono-Lo), immature macrophages (LyC6^high/MHCII^high = Mono-Hi), M1 macrophages (Ly6C^low/MHCII^high = TAM-M1) or M2 macrophages (Ly6C^low/MHCII^low = TAM-M2)^20,21. When we analyzed the myeloid cells with the above gating strategy in 4T1 TILs, we also observed that IPI-549 shifts myeloid cells toward the TAM-M1 population (Extended data Fig. 3a). Moreover, relative mRNA expression of M1 and M2 markers correlate with a less suppressive function of these cells (Extended data Fig. 3b). We subsequently tested the suppressive function of myeloid cells derived from IPI-549 treated B16-GMCSF tumor bearing mice on naïve CD8 T cells proliferation (Fig. 2e). We show that suppression of CD8⁺ T cells is abolished in IPI-549-treated mice or when IPI-549 is added to the media. A similar functional observation was made in human myeloid suppressor cells in PBMCs (Extended data Fig. 4b) confirming its potential clinical use. In addition, pharmacodynamic evaluation of whole blood confirms inhibition of PI3K-γ in monocytes after activation with a PI3K-γ stimulus in human volunteers (Extended data Fig. 4a.). To confirm that inhibition of PI3K-γ in myeloid cells is required to delay tumor progression, we depleted myeloid cells (anti-CD11b) prior to implanting LLC-Brei tumors in mice. Treatment with IPI-549 did not delay tumor growth in the absence of TAMCs (Extended data Fig. 2e). Taken together these findings suggest that PI3K-γ inhibition using IPI-549 is mainly effective in a tumor landscape rich in suppressive myeloid cells and allows for more precise delineation of patients in which it will potentially yield the greatest activity.

Figure 2. Selective targeting of PI3K-γ reduces tumor growth and metastasis in various checkpoint blockade-resistant tumor models associated with high myeloid cell infiltrates.

a. Therapy regimen (upper panel) and mean tumor volume of subcutaneous (4T1) or intradermal (B16-GMCSF) implants in vehicle vs IPI-549 (15 mg/kg administered orally, daily) treated mice (n=10), lower panel. b. Representative pictures and hematoxylin and eosin-stained sections of lung from B16-GMCSF tumor bearing mice at day 7 on treatment with IPI-549 or vehicle (left panels). Quantification of lung metastasis at day 14 and 21 from tumor challenge (right panel). c. Flow cytometric analysis and quantification of CD11b⁺F4/80⁺ (TAM) cell populations in 4T1 tumors at day 7 post implants (n=5), expression of CD206 (M2) and MHCII (M1) in CD11b⁺F4/80⁺ cell populations. The histogram bars show the percent change in each cell population (CD11b+, TAM, M2 and M1) in IPI-549 treated group in comparison with vehicle-treated control d. mRNA expression of selected M1 and M2 markers in 4T1 and B16-GMCSF tumors as determined by RT-PCR (n=3), data were relative to GAPDH expression and normalized versus the mean of vehicle treated tumors. e. In vitro suppressive activity of tumor-infiltrating CD11b+ cells purified from 4T1, tumor-bearing mice at day 7 post treatment with IPI-549 or Vehicle. Representative histograms of CD8⁺ T cell proliferation in CD11b+ to CD8+ T cell ratio 1:1 (left panel) and percent CD8⁺ T cell proliferation (right panel) (n=3). Data represent analysis of 5–10 mice per group, mean ± s.e.m. *P<0.05, **P<0.01, ***P<0.001 (non-parametric Mann–Whitney t-test).

The IPI-549-driven polarization of myeloid cells to a less immunosuppressive phenotype correlates with increased CD8⁺ T cell infiltrates and a higher CD8/Treg ratio at day 14 in both 4T1 and B16-GMCSF tumors (Fig. 3a). CD8⁺ T cells expressed more granzyme B and proliferated more (Ki67⁺) in IPI-549 treated groups (Fig. 3a Extended data Fig. 5a). Along with these effector mechanisms, the expression of PD-1 and CTLA-4 were enhanced in T cells (Fig. 3a and Extended data Fig. 5a). Furthermore, in mice lacking T cells (RAG KO, Nu/Nu or after anti-CD8 T cell depletion) the impact of IPI-549 was abrogated (Fig. 3b, Extended data Fig. 5b,c). These findings confirm that targeting PI3K-γ enhances the T cell-mediated anti-tumor activity by modifying myeloid cells suppressive fucntion. We were concerned that targeting a key pathway in myeloid cells may affect their antigen presentation. Reassuringly, we found that IPI-549 did not reduce, but rather enhanced, the activation of tumor antigen-specific T cells. This was demonstrated by the transfer of tumor antigen specific cells (Pmel-1) in B16-GMCSF bearing mice (Fig. 3c) and by ELISpot using CD8⁺ T cells isolated from CT26 tumor-bearing mice (Extended data Fig. 5d). Additionally, the proportion of effector memory cells (CD44⁺CD62L⁻) in CD8+ and CD4+ populations was not affected (Extended data Fig. 6), suggesting that IPI-549 does not affect the ability of antigen-presenting cells to activate antigen-specific T cells. Taken together, these results demonstrate that selective PI3K-γ inhibition reduces the immune-suppressive function of myeloid cells, ultimately promoting T-effector activation and T-cell mediated cytotoxicity.

Figure 3. Reduction of myeloid suppressive phenotype correlates with higher anti-tumor T cell activity.

a. Quantification by Flow cytometry of ratio of M1/M2 in CD11b⁺F4/80⁺ cell population, CD8⁺ T cells in CD45⁺ TILs and ratio of CD8⁺ T cell/Treg (CD4⁺FoxP3⁺) in CD45⁺ TILs in 4T1 and B16-GMCSF tumors at day 7 and 14 on treatment (n=5), (left panels), Quantification of Granzyme B and PD1 expression in CD8⁺ TILs in 4T1 and B16-GMCSF tumors at day 7 and 14 on treatment (right panels). b. Mean tumor volume of intradermal B16-GMCSF tumors in IPI-549 vs vehicle treated wild type C57Bl6 (left) or Rag^−/− mice (right) (n=10). c. Flow cytometric analysis and quantification of gp-100 antigen specific T cells transferred into IPI-549 or vehicle-treated B16-GMCSF tumor bearing mice (n=5). Data represent analysis of 5–10 mice per group, mean ± s.e.m. *P<0.05, **P<0.01 (non-parametric Mann–Whitney t-test).

We found that treatment of myeloid-enriched tumors with IPI-549 leads to upregulation of PD-1 and CTLA-4 expression on CD8⁺ T cells (Fig. 3a. and Extended data Figs 2d and 5a). This strongly suggests that combining PI3K-γ inhibition with ICB therapy could yield additional anti-tumor activity. In fact, in the ICB resistant 4T1 or B16-GMCSF models, the combination of either anti-CTLA-4 or anti-PD-1 with IPI-549 significantly delayed tumor growth when compared to ICB alone (Fig 4a,b). We further confirmed this observation in the ICB-resistant LLC tumor model (Extended data Fig. 7). Analysis of TILs in the combination regimen of IPI-549 with PD-1 blockade shows that the addition of IPI-549 reduces immune suppression by increasing the M1/M2 ratio, resulting in improved T cell effector function in 4T1 and B16-GMCSF tumors (Extended data Fig. 8). These results are similar to those reported by Kaneda et al where PD-1 blockade was combined with a dual PI3K inhibitor (γ/δ)¹⁷; however, given the important effect of PI3K-δ in T cell function and more specifically regulatory T cells (Tregs), it is unclear whether the observed antitumor drug effect is myeloid and/or T cell dependent. As CTLA-4 and PD-1 have distinct immunologic mechanisms of action, greater therapeutic efficacy is achieved when both therapies are combined²². We tested the anti-PD-1 and anti-CTLA4 combination with IPI-549 in the 4T1 and B16-GMCSF models (Fig. 4a). Double checkpoint blockade therapy alone did not result in any complete tumor regressions in 4T1-bearing mice and only 20% of the mice bearing B16-GMCSF benefited from the therapy, further confirming multiple checkpoint resistance in these models. Notably, the addition of IPI-549 to the combination of anti-CTLA-4 + anti-PD-1 was associated with complete remissions in 30% of 4T1 and 80% of B16-GMCSF tumor-bearing mice. Importantly, tumor-free survivors were resistant to tumor re-implantation (Extended data Figure 9), indicating long lasting adaptive immunity.

Figure 4. Resistance to checkpoint blockade therapy is overcome when combined with selective PI3K-γ inhibition.

a. Therapy regimen; b. Above panels: mean tumor volume of subcutaneous 4T1 tumor in control, anti-CTLA4 or anti-PD-1 treated mice in combination with IPI-549 or vehicle (n=10). Individual tumor volumes of subcutaneous 4T1 implants in mice treated with anti-CTLA4 and anti-PD1 in combination with IPI-549 or vehicle (n=10), dotted red line corresponds to end of treatment day post tumor implant (day21). Survival at 150 days of 4T1 tumor bearing mice treated with anti-CTLA4 and anti-PD1 in combination with IPI-549 or vehicle compared to control (vehicle treated only) (n=10); c. Mean tumor volume of intradermal B16-GMCSF implants in anti-PD-1, anti-CTLA4 or control treated mice in combination with IPI-549 or vehicle (n=10). Individual tumor volumes of intradermal B16-GMCSF implants in mice treated with anti-CTLA4 and anti-PD1 in combination with IPI-549 or vehicle (n=10). Survival at 150 days of B16-GMSCF tumor-bearing mice treated with anti-CTLA4 and anti-PD1 in combination with IPI-549 or vehicle compared to control (vehicle treated only) (n=10). Data represent analysis of 5–10 mice per group, mean ± s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (non-parametric Mann–Whitney t-test or Log-Rank test).

Taken together, our data indicate that high infiltration by suppressive TAMCs promotes resistance to ICB, further confirming the complex, multi-factorial mechanisms of immune suppression in the tumor microenvironment. We show that modulating the suppressive phenotype of these cells toward a more inflammatory one can be achieved by selective pharmacologic targeting of the myeloid PI3K-γ isoform. By doing so, sensitivity to ICB can be effectively restored in TAMC rich tumors. These findings provide a strong rationale to consider exploring PI3K-γ inhibition to overcome resistance to ICB in clinical trials based on a precision medicine-type assessment of the tumor immune landscape. This hypothesis is currently being tested in the IPI-549 phase 1/1b trial, as both a monotherapy and in combination with PD-1 blockade in solid tumors (clinicaltrials.gov NCT02637531).

Methods summary

Cell Lines

The murine cancer cell lines for melanoma (B16F10, referred to as B16), breast cancer (4T1) and colon cancer (CT26) were obtained from ATCC. The colon cancer cell line MC38 was obtained from the NCI and the lung cancer (LLC-Brei) from Caliper Life Sciences. Cells were maintained in RPMI medium supplemented with 10% fetal calf serum (FCS) and penicillin with streptomycin (complete RPMI media). The GMCSF-secreting B16 cell line (referred as to B16-GMCSF) has been reported and was used to increase the number of myeloid cells recruited to the tumor.² The cell lines have been mycoplasma tested.

Quantitative RT-PCR

cDNA was prepared using 1μg RNA with the qScript cDNA Synthesis Kit (Quanta Biosciences). Sybr green-based qPCR was performed using murine primers to Arg1, Ifng, Il10, Il12p40, Il1b, Il6, Ccl2, Gapdh, Nos2, Tgfb1, Tnfa, IL4ra, Indo, Ctla4, Pd-l1, CD86, CxCr2, Fizz1, Ymd1 (Qiagen QuantiTect Primer Assay). mRNA levels were normalized to Gapdh (dCt = Ct gene of interest – Ct Gapdh) and reported as relative mRNA expression (ddCt = 2^-(dCt sample – dCt control)) or fold change.

Cytokine Analysis

Tumors were excised, snap frozen in liquid nitrogen, and pulverized using a tissue grinder. Tumor protein lysates were prepared in MSD Tris Lysis Buffer (Meso Scale Discovery, Rockville Maryland) containing 2X Halt Protease and Phosphatase Inhibitor Cocktail (Fisher Scientific). Total protein concentration was normalized to 4 mg/ml and cytokines were quantified using the MILLIPLEX MAP Mouse Cytokine/Chemokine Magnetic Bead 32 Plex Panel and MILLIPLEX MAP Mouse CD8⁺ T Cell Magnetic Bead Panel kits according to the manufacturer’s instructions (Millipore, Billerica, MA).

RNAseq

Mice bearing CT26 tumors were treated with vehicle or IPI-549 (15 mg/kg/day, PO) for 6 or 9 days. Tumors were isolated, and frozen until needed; tumors then thawed and RNA was extracted from all cells. RNAseq was done at Expression Analysis (Q2 Solutions). Sequence reads were aligned to the mouse B38 reference genome using OmicSoft ArrayStudio and the UCSC gene model. Log₂(FPKM) was calculated for each gene, and data were mean centered for display in heat maps. The analysis focused on a compilation of about 4,200 mouse genes related to cancer immunology and PI3K pathway signaling compiled from numerous sources including BioCarta pathways, GO gene ontologies, KEGG pathways, WikiPathways, and literature.⁵

Mice

C57BL/6J and Balb/c mice (6–8 weeks old) were purchased from Jackson Laboratory. Pmel-1 TCR transgenic mice have been previously reported¹ and were provided by N. Restifo (National Cancer Institute, Bethesda, MD). All mice were maintained in micro isolator cages and treated in accordance with the NIH and American Association of Laboratory Animal Care regulations. All mouse procedures and experiments for this study were approved by the Memorial Sloan Kettering Cancer Center Institutional Animal Care and Use Committee. Ten to fifteen mice per treatment strategy were used to allow 90% power, and a 5% significance level, and detect differences in tumor-free survival from 10% to 80%. Typically, tumors grow in 100% of control animals. An additional 5 mice per group were for tissue harvest at day 7 and day 14 on treatment. Mice cage, treatment allowance were at random at day 7 after tumor implants.

Tumor Challenge and Treatment Experiments

On day 0 of the experiments, tumor cells were injected intradermally (i.d.) in the right flank. For the B16 model, 2.5×10⁵ B16-WT or B16-GMCSF cells were injected into C57BL/6J mice. For 4T1 model and for the CT26 model, 5×10⁵ cells were used subcutaneously in Balb/c mice. For studies in immune compromised mice, the CT26 study was done in the Balb/c nu/nu strain and the B16-GMCSF in C57Bl.6 rag1^−/− mice. Mice were obtained from Jackson Labs and Charles River Labs. Treatments were given as single agents or in combinations with the following regimen for each drug. The PI3K-γ inhibitor drug IPI-549 was dissolved at 5% 1-Methyl-2-pyrrolidinone in Polyethylene Glycol 400 and administered by oral gavage once a day at 15 mg/kg. Treatment was initiated on day 7 ending on day 21 post tumor implant. Control groups received vehicle (5% NMP, 95% PEG) without the active product. Anti-CTLA-4 antibody (100 μg/mouse, clone 9H10, Bio X cell) and anti-PD-1 antibody (250 μg/mouse, clone RPM1- 14, Bio X cell) were injected intraperitoneally (i.p.) on days 7, 10, 13 and 16 for the B16, B16-GMCSF and 4T1 models and on days 10, 13 and 16 for the CT26 model. Tumors were measured every second or third day with a caliper, and the volume (length × width × height) was calculated. Mice that had no visible and palpable tumors that could be measured on consecutive measurement days were considered complete regressions. Animals were euthanized for signs of distress or when the total tumor volume reached 2500 mm³. For the re-challenge study: mice with complete responses in the anti-PD-1 treatment group and the anti-PD-1 and IPI-549 combination group were re-challenged with 2.5 × 10⁵ CT26.WT tumor cells (on day 106 of the study since original tumor implant). Additional mice with complete responses from an additional IPI-549 and anti-PD-1 group were implanted with 1×10⁵ 4T1 tumor cells.

Cell depletion studies

Female BALB/c mice were depleted of CD8⁺ T cells by intraperitoneal (IP) injection of a CD8 depleting antibody beginning 3 days before tumor implantation and continuing every 3 days (Q3D) for the duration of the study. Control animals were injected IP with an isotype control antibody according to the same dosing schedule. CD8⁺ T cell depletion was verified by flow cytometry of splenic cells from a separate cohort of mice at the time of tumor implant. Anti-CD8 antibody (BioXCell, In Vivo MAb Rat IgG2b anti-mouse CD8, clone 169.4) and isotype control antibody (BioXCell, In Vivo MAb Rat IgG2b, κ isotype control, clone LTF-2) were used. CD8⁺ T-cell-depleted and isotype control treated mice were implanted with 2.5×10⁵ CT26.WT cells subcutaneously to the dorsal right flank. Eight days after implant, tumor-bearing animals with average tumor volumes of 50 to 60 mm³ began treatment. Animals were dosed daily with vehicle or IPI-549 (15/mg/kg, PO) and dosed every three days with 100 μg antibody, either isotope control or anti-CD8. Tumor measurements were taken every second or third day during the 13 days of dose administration. For CD11b⁺ cell depletion, murine Lewis Lung Carcinoma tumor brei (here referred to as LLC) was propagated into ten C57BL/6 Albino male mice. When tumors reached an average of 1500 mm³, tissue was collected and made into a single cell suspension. Tissue was dissociated in a glass dunce, filtered through a 100 micron filter and washed with cold buffer of PBS pH 7.2, 0.5% BSA and 2mM EDTA. An aliquot of cells was separated, and placed on ice until the time of re-transplant. Cells were counted and adjusted to 2×10⁷ total cells. To deplete the CD11b cells from the tumor homogenate, CD11b micro beads (MACS #130-049-601) were added to the sample and incubated on ice for 15 min. Cells were washed and re-suspended with buffer. A negative depletion was preformed following the Miltenyi autoMACS protocol. The positive selected cells and starting tumor inoculum were counted and cell numbers adjusted to the tumor inoculum used for historical studies. For each cell condition 30 C57BL/Albino male mice were inoculated. Treatment of IPI-549 or vehicle started when tumors for each cell condition reached an average of 200 mm³ regardless of what day post implant this occurred.

Isolation of Tumor-infiltrating Cells and Lymphoid Tissue Cells

Mouse tumor samples were minced with scissors prior to incubation with 1.67 U/ml Liberase (Roche) and 0.2mg/ml DNase (Roche) in RPMI for 30 min at 37 °C. Tumor samples were homogenized by repeated pipetting and filtered through a 100-μm nylon filter (BD Biosciences) in RPMI supplemented with 7.5% FCS to generate single-cell suspensions. Cell suspensions were washed once with complete RPMI and purified on a Ficoll gradient to eliminate dead cells. Cells from mouse spleens were isolated by grinding spleens through 40-μm filters. After red blood cell (RBC) lysis (ACK Lysing Buffer, Lonza) when required, all samples were washed and re-suspended in FACS buffer (PBS/0.5%albumine) or RPMI depending on further use..

Flow Cytometry and Morphology Analysis

Cells isolated from mouse tumors and spleens were pre-incubated (15 min, 4 °C) with anti-CD16/32 monoclonal antibody (Fc block, clone 2.4G, BD Biosciences) to block nonspecific binding and then stained (30 min, 4 °C) with appropriate dilutions of various combinations of the following fluorochrome-conjugated antibodies: anti-CD45-AF 700 (clone 30-F11), anti-CD11b-APC-Cy7 (clone M1/70), anti-CD11b-PE-TR(M1/70.15), anti-Ly6G-APC (clone 1A8), anti- F4/80-PercP-Cy5.5 (clone BM8), anti-Ly6C-PE-Cy7 (clone AL-21), anti-MHC Class II-eFluor 450 (clone M5/114.15.2), Anti-CD206-PE (clone 19.2), anti-CD8- Percp-Cy5.5 (clone 53-6.7), anti-CD8- PE Texas Red (clone 5H10), anti-CD4-PE-Cy7 (clone RM4-5), anti-CD4-Pacific Blue (clone RM4-5), anti-Foxp3-APC (clone FJK-16s), anti-Foxp3-PE-Cy7 (clone FJK-16s), anti-CD25- APC-Cy7 (clone PC61), anti-CD44-PE-Cy7 (clone IM7), anti-CD62L-PE (clone MEL-14), anti-Ki67 (clone B56), anti-CTLA-4-APC (clone), anti-Granzyme B-PE-TR (clone GB11), antibodies, all purchased from BD Biosciences, eBioscience or invitrogen. For intracellular stain, cells were further permeabilized using a FoxP3 Fixation and Permeabilization Kit (eBioscience) and stained for Foxp3 (clone FJK-16s, Alexa-Fluor-700-conjugated, eBioscience), Ki67 (clone SolA15, eFluor-450-conjugated, eBioscience) or CTLA-4. The stained cells were acquired on a LSRII Flow Cytometer using BD FACSDiva software (BD Biosciences) and the data were processed using FlowJo software (Treestar). Dead cells and doublets were excluded on the basis of forward and side scatter and Fixable Viability Dye eFluor 506.

Purification of Myeloid or CD8 T cells from tumors or spleen

Mouse tumor and spleen single-cell suspensions were generated as described in the previous section. Tumor cells were subsequently separated from debris over a Ficoll gradient (Sigma-Aldrich). B cells were depleted from splenocytes using CD19 microbeads and LD columns according to the manufacturer’s instructions (Miltenyi Biotec) to enrich the myeloid fractions. Cells were stained with anti-CD45.2-Alexa- Fluor-700, anti-CD11b-APC-Cy7, anti CD8-PE antibodies for flow sorting on a FACSAria^™ II Cell Sorter (BD Biosciences). Dead cells were excluded using DAPI (Invitrogen). Purity of flow-sorted populations was above 90%.

Isolation of Pmel Lymphocytes and Adoptive Transfer

Spleens and lymph nodes from pmel-1 TCR transgenic mice were isolated and ground through 100-μm filters. After RBC lysis, CD8⁺ T cells were purified by positive selection using Miltenyi magnetic beads. The isolated cells were loaded with CellTrace^™ Far Red DDAO-SE (Thermo Fisher Scientific) and injected into recipient animals via tail vein at indicated numbers. Activated CD8+ T cells were generated by culturing splenocytes with soluble anti-CD3 (1 μg/ml, clone 145-2C11, eBioscience) and anti-CD28 (2 μg/ml, clone 37.51, eBioscience) for 72 h. Recombinant mouse IL-2 (30 U/ml, Chiron) was added for the final 24 h of culture. CD8⁺ T cells were subsequently positively selected with Miltenyi magnetic beads prior to injection via tail vein, as described above. The frequency and proliferation of pmel cells were measured 2 weeks after tumor challenge and 7 days after adoptive transfer of 1×10⁶ in vitro activated CD8⁺pmel T cells using Thy1.1 antibody and by assessing CellTrace^™ Far Red DDAO-SE dilution by flow cytometry, respectively.

T cell Suppression Assay

Spleens from naïve mice were isolated and ground through 40-μm filters to generate a single cell suspension. After RBC lysis, CD8⁺ cells were purified using anti-CD8 (Ly-2) microbeads (Miltenyi Biotech) according to manufacturer’s protocol and labeled with 1 mM CFSE (Invitrogen) in pre-warmed PBS for 10 min at 37°C. The CFSE-labeled CD8⁺ T cells were then plated in complete RPMI media supplemented with 0.05 M β-mercaptoethanol onto round bottom 96-well plates (25 × 10E3 cells per well) coated with 1 μg/ml anti- CD3 (clone 1454-2C11) and 5 μg/ml anti-CD28 (clone 37N) antibodies. Purified myeloid cells were added in indicated ratios and plates were incubated at 37 °C. After 48 h, cells were harvested and CFSE signal in the gated CD8⁺ T cells was measured by flow cytometry (LSRII Flow Cytometer, BD Biosciences). For the human MDSC suppression assay PBMCs were isolated using Lymphoprep^™ from donor blood. T cells were isolated by CD3⁺ selection (Easysep, Stem Cell Technologies) and frozen in Sigma freezing media for later use. The remaining PBMCs minus T cells were incubated with 20 ng/ml GMCSF and 20 ng/ml IL6 for 6 days to differentiate the myeloid cells and were incubated with or without added IPI-549. After 6 days, the MDSC cells were isolated by CD33⁺ selection (Easysep, Stem Cell Technologies). These MDSC cells were then mixed with the autologous T cells (at a 1:4 ratio) that had been pre-stained with cell trace violet and activated with anti-CD3 and anti-CD28 beads (Dynal). The T cell proliferation is determined by Cell Trace Violet dye dilution measured by flow cytometry after 72 hours.

IFN-γ ELISPOT assay

Blood was collected from Vehicle and IPI-549 (15mg/kg) treated CT26 tumor-bearing mice after 10 days of treatment. PBMC were isolated using Lymphoprep density gradient media (Stem Cell Technologies, Vancouver, BC). IFN-γ producing cells were quantified using the CTL Mouse Immunospot IFN-γ Single Color ELISPOT kit (CTL, Shaker Heights, OH) according to the manufacturer’s instructions. For in vitro re-stimulation, 1×10⁵ PBMC were co-cultured with 1×10⁵ irradiated (2000 rad) CT26 colon carcinoma target cells in CTL test media (CTL) for 16 hours. Irradiated (2000 rad) 4T1 mammary carcinoma target cells were used as negative control targets to assess specificity. IFN-γ spots were quantified using a CTL Immunospot S6 Micro Analyzer and Immunospot Professional Software (CTL).

Macrophage polarization assay

Bone marrow derived macrophages were prepared from C57Bl/6 mice femur and tibias. Red blood cells were lysed and then the remaining cells were plated in bone macrophage media (BMM) consisting of DMEM, 20% FBS plus pen/Strep and 50 ng/ml M-CSF and incubated for 6 days. Cells were polarized towards an M2 phenotype with the addition of 20 ng/ml IL4 and 50 ng/ml M-MCSF (both from R&D Systems) with or without added IPI-549. Cells were incubated for 48 hours and then RNA was harvested from the cells (Qiagen RNeasy). qRT-PCR was performed using primers for mouse ARG1 (Mm00475988_m1, Life Technologies NY) and mouse B-actin (Mm00607939_s1 Life Technologies, NY).

Biochemical and cell based assays

Biochemical and Cell based assays for the Class I and Class II PI3K isoforms were run as previously described.^3,4

PI3K-gamma-specific whole blood PD assay

Whole blood from 6 healthy donors was pretreated with a dose titration of IPI-549 and then stimulated for 2 minutes with 2.3 ug/ml CXCL12. Cells were lysed, fixed and stained. Human blood samples have been collected after ICF approval. Response to stimulation was determined by measuring phosphorylation of AKT S473 in monocytes (CD14+) by flow cytometry and comparing the value to that of untreated controls. IC₅₀ values for IPI-549 were calculated for each donor, measuring compound potency against PI3K-γ and in whole blood.

Statistics

Where indicated, data were analyzed for statistical significance and reported as P values. Data were analyzed by Mann-Withney test when comparing means of two independent groups and two-way ANOVA when comparing more than two groups. P<0.05 was considered statistically significant (*P<0.05, **P< 0.01, ***P<0.001, ****P< 0.0001). Evaluation of survival patterns in tumor-bearing mice was performed by the Kaplan–Meier method, and results were ranked according to the Mantel–Cox log-rank test. P<0.05was considered statistically significant (*P<0.05, **P<0.01, ***P<0.001, ****P<0.0001). Survival was defined as mice with tumors <2500 cm³.

Data availability

Source data for the main figures are provided with the paper.

Extended Data

Extended Data Figure 1. Impact of suppressive myeloid TILs in response to Checkpoint blockade.

a. Individual tumor growth of subcutaneous (4T1) or intradermal (B16, B16-GMCSF) implants in anti-PD-1, anti-CTLA4 or control treated mice (n=10). b. In vitro suppressive activity of tumor-infiltrating CD11b⁺ cells purified from spleen of 4T1, B16, B16-GMCSF tumor-bearing mice. Representative histograms of CD8⁺ T cell proliferation at corresponding CD11b⁺ to CD8⁺ T cell ratio (left panel) and quantification of CD8⁺ T cell proliferation (right panel) (n=3), mean ± s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (non-parametric Mann–Whitney t-test).

Extended Data Figure 2. Impact of PI3K-γ selective inhibition on tumor growth and Myeloid TILs.

a. Binding affinities (Kd) and cellular IC50 inhibition of pAKT by IPI-549 for class I PI3K isoforms (Left table). Percent of inhibition of ARG-1 expression on bone marrow derived macrophages polarized with M-CSF and IL-4, (right panel). b. Percent of tumor growth inhibition in LLC, MC38, 4T1, CT26, B16-GMCSF tumor bearing mice treated with IPI-549 (table). c. Quantification of CD11b, CD206, NOS2 and PD-L1 expression in CD11b+ tumor infiltrating leukocytes from IPI-549 vs vehicle treated CT26 tumor bearing mice. d. RNAseq of co-stimulatory and checkpoint molecules on whole tumors from CT26 tumor-bearing mice treated for 6 or 9 days with IPI-549 compared to vehicle. e. Mean tumor volume of subcutaneous LLC-Brei implants in IPI-549 vs Vehicle treated mice without or after CD11b⁺ cell depletion. Data represent analysis of 5–10 mice per group, mean ± s.e.m. *P<0.05, ***P<0.001 (non-parametric Mann–Whitney t-test).

Extended Data Figure 3. Impact of PI3K-γ selective inhibition on subsets of CD11b myeloid cells.

a. Representative flow cytometry analysis and quantification of Ly6C, MHC class II expression in CD11b⁺Ly6G⁻ cells infiltrating 4T1 tumors. b. mRNA expression of selected M1 and M2 markers in sorted Ly6C low-MHCII Low (TAM-M2) compared to Ly6C low-MHCII High (TAM-M1) population from 4T1 tumor, data were relative to GAPDH expression and normalized versus the mean of TAM-M1 population. Mean ± s.e.m. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001 (non-parametric Mann–Whitney t-test).

Extended Data Figure 4. Impact of PI3K-γ selective inhibition on suppressive PBMC derived human myeloid cells.

a. Inhibition of CXCL-12 activation of PI3K-γ in monocytes as measured by pAKT (S473) in human whole blood. b. Representative histograms and quantification of Human CD8 T cells proliferation after 72 hours of co-culture with or without autologous MDSCs generated from the T-cell-depleted PBMCs ± IPI-549.

Extended Data Figure 5. Impact of PI3K-γ selective inhibition on function of tumor specific T cell responses.

a. Quantification of KI67, and CTLA4 expression in CD8+ T cells in TILs of 4T1 or B16-GMCSF tumors at IPI-549 compared to vehicle treatment day 7 and 14. b. Mean tumor volume of subcutaneous 4T1 tumor in IPI-549 vs vehicle treated BALB/c nu/nu mice (n=10). c. Mean tumor volume of subcutaneous CT-26 tumor in IPI-549 vs vehicle treated BALB/c mice with or without CD8+ T cell depletion by anti-CD8 antibody (n=10). d. Quantification and representative pictures of CT26 tumor specific immune responses in PBMCs from CT26 tumor-bearing mice treated with IPI-549 in comparison to vehicle by ELISPOT. PBMC were collected from tumor-bearing animals after 10 days of vehicle or IPI-549 treatment and restimulated overnight with irradiated CT26 or 4T1 stimulator cells.

Extended Data figure 6. Impact of PI3K-γ selective inhibition on the differentiation of T cells in tumors.

a. Representative Flow cytometry analysis and quantification of CD62L and CD44 expression in CD8 and CD4 T cell infiltrates in tumor, LN and spleen of 4T1 tumor bearing mice treated with IPI-549 compared to Vehicle. Data represent analysis of 5 mice per group, mean ± s.e.m. NS = non statistical significance (non-parametric Mann–Whitney t-test).

Extended Data Figure 7. Impact of combination of PI3K-γ selective inhibitor with checkpoint blockade on various tumors.

a. Survival to 2000 mm³ tumor volume of LLC Brei tumor in IPI-549 or vehicle treated mice in combination or not with anti-CTLA4 (Vehicle and IPI-549 groups, n=14; anti-CTLA4, n=13; IPI-549 and anti-CTLA4 combination, n=10). b. Mean tumor volume of CT26 tumor in IPI-549 or vehicle treated mice in combination or not with anti-PD-L1 (n=15 for all groups except vehicle, n = 13).

Extended Data Figure 8. Impact of combination of PI3K-γ selective inhibitor with checkpoint blockade on TILs.

a. Mean tumor volume of subcutaneous 4T1 tumor bearing mice treated with IPI-549, Vehicle or anti-PD-1 in combination with IPI-549 or vehicle (n=10). b. Representative flow cytometry analysis of CD206 and MHCII labeling in CD11b+ F4/80+ cell populations in the different treatment groups of 4T1 tumor bearing mice. c. Quantification of CD11b⁺ F4/80⁺, M1/M2 ratio, CD8/Treg in TILs and Granzyme B expression in CD8 T cells from 4T1 tumors in the different treatment groups. d. Quantification of CD11b⁺ F4/80⁺, M1/M2 ratio, CD8/Treg in TILs and Granzyme B expression in CD8⁺ T cells from B16-GMCSF tumors in the different treatment groups, mean ± s.e.m. *P<0.05, **P<0.01 (non-parametric Mann–Whitney t-test).

Extended Data Figure 9. Impact of combination of PI3K-γ selective inhibitor with checkpoint blockade on acquisition of anti-tumor memory.

a. Tumor rechallenge at 100 days (from first tumor implant) following primary tumor complete response in B16-GMCSF tumor bearing mice treated with Vehicle (blue) or IPI-549 (red) in combination with both anti-PD1 and anti-CTLA4, b. CT26 tumor-bearing mice with complete responses in the anti-PD-1 treatment group and the IPI-549 + anti-PD-1 combination treatment group were rechallenged with CT26 tumor implant. Additional mice with complete responses from the IPI-549 + anti-PD-1 combination were implanted with 4T1 tumors. There was a low or no tumor take with CT26 rechallenge, while all 4T1 tumors grew, indicating specific anti-tumor memory.