Targeted therapies prime oncogene-driven lung cancers for macrophage-mediated destruction

Abstract

Macrophage immune checkpoint inhibitors, such as anti-CD47 antibodies, show promise in clinical trials for solid and hematologic malignancies. However, the best strategies to use these therapies remain unknown and ongoing studies suggest they may be most effective when used in combination with other anticancer agents. Here, we developed a novel screening platform to identify drugs that render lung cancer cells more vulnerable to macrophage attack, and we identified therapeutic synergy exists between genotype-directed therapies and anti-CD47 antibodies. In validation studies, we found the combination of genotype-directed therapies and CD47 blockade elicited robust phagocytosis and eliminated persister cells in vitro and maximized anti-tumor responses in vivo. Importantly, these findings broadly applied to lung cancers with various RTK/MAPK pathway alterations-including EGFR mutations, ALK fusions, or KRAS^G12C mutations. We observed downregulation of β2-microglobulin and CD73 as molecular mechanisms contributing to enhanced sensitivity to macrophage attack. Our findings demonstrate that dual inhibition of the RTK/MAPK pathway and the CD47/SIRPa axis is a promising immunotherapeutic strategy. Our study provides strong rationale for testing this therapeutic combination in patients with lung cancers bearing driver mutations.

Figure 1:. An unbiased compound library screen identifies cooperation between targeted therapy and macrophage-directed immunotherapy for EGFR mutant lung cancer.

(A) Experimental setup of an unbiased functional screen to identify drugs that synergize with macrophage-directed immunotherapy. Primary human macrophages were co-cultured in 384-well plates with GFP+ PC9 cancer cells (a human EGFR mutant NSCLC cell line). The wells were treated with 10 ug/mL anti-CD47 antibody, and then a drug compound library (n = 800 FDA-approved drugs) was overlayed at a concentration of 5 uM. The cells were incubated for 3–5 days and GFP+ area was quantified by automated microscopy and image analysis. As controls, GFP+ PC9 cells were cultured with each individual drug of the library alone for comparison. (B) Representative images of whole-well microscopy showing GFP+ area (purple) as quantified by automated image analysis from wells treated with drugs found to enhance (erlotinib, gefitinib) or inhibit (dexamethasone) macrophage-dependent cytotoxicity of PC9 cells. Scale bar, 800 um. (C) Volcano plot summarizing the results of the drug library screen. Each point represents the mean of each individual drug treatment condition from n = 5 experimental trials. The phenotypic effect size (x-axis) is depicted as log₂ fold-change of GFP+ area in the macrophage+anti-CD47 condition relative to PC9 cells alone. Values were normalized to account for variation due to well position. Dashed lines represent 2-fold change in effect size (x-axis) or p<0.05 by t test (y-axis). Gefitinib and erlotinib (blue) were identified as the top enhancers of macrophage-dependent cytotoxicity of PC9 cells, whereas drugs depicted in red inhibited macrophage dependent cytotoxicity or were drugs that macrophages protected against. (D) Representative curves showing macrophage-dependent cytotoxicity over time as represented by decreases in GFP+ area of macrophage+anti-CD47 condition relative to the control condition. Curves from one representative plate showing gefitinib and erlotinib enhance macrophage-dependent cytotoxicity within ~48 hours. Dashed lines indicate the empirical 95% tolerance interval. (E) Box and whisker plot of drug classes included in the screen as ranked by normalized log₂ fold-change of GFP+ area in macrophage versus PC9 control condition. Each box indicates the median, interquartile range, maxima and minima (excluding outliers) for the indicated drug class. Drug classes that significantly increased relative GFP+ area are depicted in red, whereas EGFR TKIs (blue) were identified as the only drug class that significantly decreased relative GFP+ area. Each class of drugs was compared with controls (DMSO and empty wells) using a t test (**FDR<0.01, ***FDR<0.001).

Figure 2:. Combined targeting of EGFR and CD47 enhances macrophage phagocytosis in vitro.

(A) Histograms showing CD47 expression on the surface of NSCLC cell lines containing the indicated driver mutations as assessed by flow cytometry. (B) Flow cytometric analysis of macrophage immune checkpoint molecules on the surface of established and patient-derived NSCLC cell lines containing the indicated driver mutations. Geometric mean fluorescence intensity (Geo. MFI) for each antigen was compared to CD47. (C) CD47 expression on EpCam+ cancer cells from malignant pleural effusion specimens of patients with NSCLC containing the indicated driver mutations. Left, percent of CD47-positive cells. Right, CD47 geometric MFI. Data represent mean ± SD from 3 technical replicates of n = 10 independent patient specimens. (D) Representative examples of phagocytosis assays using primary human macrophages and GFP+ PC9 cells. The PC9 cells were exposed to vehicle control (PBS) or 1 uM EGFR TKI (erlotinib, gefitinib, or osimertinib) for 24 hours. The cells were then collected and co-cultured with primary human macrophages ± an anti-CD47 antibody for 2 hours. Phagocytosis was measured by flow cytometry as the percentage of macrophages containing engulfed GFP+ PC9 cells as indicated in plots. (E) Quantification of phagocytosis assays using the indicated EGFR TKIs at 1 uM concentration. Phagocytosis was normalized to the maximal response for each independent donor. Data depict mean ± SD from n = 9 independent blood donors combined from 3 independent experiments using CFSE+ or GFP+ PC9 cells. (F) Phagocytosis assays performed using GFP+ PC9 cells exposed to 1 uM osimertinib for varying amounts of time prior to co-culture with primary human macrophages (n = 4 independent donors). The cells were collected and analyzed for phagocytosis as in (F). (G) Phagocytosis assays performed using GFP+ PC9 cells exposed to varying concentrations of osimertinib for 24 hours prior to co-culture with human macrophages. Data represent mean ± SD from of 2 independent experiments using a total of n = 8 individual macrophage donors with 3 technical replicates per donor. (B, E-G) ns, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 by one-way (B, E) or two-way (F-G) ANOVA with Holm-Sidak multiple comparison test.

Figure 3:. Combining TKIs with anti-CD47 antibodies eliminates EGFR mutant persister cells in long-term co-cultures assays with human macrophages.

(A) Representative images showing quantification of GFP+ fluorescence from co-culture assays on day 6.5. GFP+ PC9 cells were co-cultured in 384-well plates with primary human macrophages alone or with anti-CD47 antibodies (10 ug/mL) and the indicated EGFR TKIs (1 uM). Whole-well microscopy and automated image analysis was performed to quantify GFP+ area (purple) per well over time. Scale bar, 800 um. (B) Representative plot showing growth curves of GFP+ PC9 cells in co-culture with primary human macrophages over time. The cells were treated with 1 uM osimertinib and/or 10 ug/mL anti-CD47 antibody as indicated. Data represent mean ± SEM of three technical replicates per donor using n = 9 independent macrophage donors. Data are combined from two independent experiments with statistical analysis on day 14 of co-culture. (C) Growth of GFP+ PC9 cells in co-culture with primary human macrophages in the absence or presence of anti-CD47 antibodies (10 ug/mL) and the indicated EGFR TKIs (1 uM). Points represent individual replicates, bars represent mean. Data represent three technical replicates per donor using n = 9 independent macrophage donors combined from two independent experiments. (D) Growth of GFP+ MGH119-1 patient-derived cells in co-culture with primary human macrophages in the absence or presence of anti-CD47 antibodies (10 ug/mL) and the indicated EGFR TKIs (1 uM). Points represent individual replicates, bars represent mean. Data represent three technical replicates per donor using n = 3 independent macrophage donors. (E) Growth of GFP+ MGH134-1 patient-derived cells in co-culture with primary human macrophages in the absence or presence of anti-CD47 antibodies (10 ug/mL) and the indicated EGFR TKIs (1 uM). MGH134-1 cells are resistant to first-generation EGFR TKIs (erlotinib, gefitinib) but sensitive to third-generation TKIs (osimertinib). Points represent individual replicates, bars represent mean. Data represent three technical replicates per donor using n = 4 independent macrophage donors. (F) Growth of GFP+ PC9 cells in co-culture assays with primary human macrophages and the indicated macrophage immune checkpoint inhibitors (10 ug/mL). Cells were cultured with antibodies alone or in combination with 100 nM osimertinib. Points represent individual replicates, bars represent means. Data represent GFP+ area on day 6.5 with 3–4 technical replicates per donor and n = 4–8 independent macrophage donors. (B-F) ns, not significant, *p<0.05, **p<0.01, ***p<0. 001, ****p<0.0001 by one-way (B-E) or two-way (F) ANOVA with Holm-Sidak’s multiple comparisons test.

Figure 4:. Targeted inhibition of the MAPK pathway primes NSCLC cells for macrophage-mediated destruction.

(A) Growth of GFP+ NCI-H3122 (a human ALK-positive NSCLC cell line) in co-culture with primary human macrophages in the absence or presence of anti-CD47 antibodies (10 ug/mL) and the indicated ALK-specific TKIs (1 uM). Points represent individual replicates, bars represent mean. Data represent three technical replicates per donor using n = 4 independent macrophage donors. (B) Growth of GFP+ NCI-H3122 cells in co-culture with primary human macrophages in the absence or presence of anti-CD47 antibodies (10 ug/mL) and varying concentrations of the ALK-specific TKI lorlatinib. Data represent mean ± SD of three replicates each from n = 4 independent macrophage donors on day 6.5 of co-culture. IC₅₀ of lorlatinib alone (PBS) = 10.29 nM (95% CI [8.665, 12.22]) versus IC₅₀ of lorlatinib+anti-CD47 = 2.135 nM (95% CI [0.6934, 6.261]). (C) Growth of GFP+ NCI-H358 (a human KRAS^G12C mutant NSCLC cell line) in co-culture with primary human macrophages in the absence or presence of anti-CD47 antibodies (10 ug/mL) and the indicated KRAS^G12C-specific inhibitors (1 uM). Data represent mean ± SEM from three technical replicates per donor using n = 4 independent macrophage donors. (D) Growth of GFP+ NCI-H358 cells in co-culture with primary human macrophages in the presence or absence of anti-CD47 antibodies (10 ug/mL) and varying concentrations of the KRAS^G12C-specific inhibitor sotorasib. Data represent mean ± SD of three replicates each from n = 4 independent macrophage donors on day 6.5 of co-culture. IC₅₀ of sotorasib alone (PBS) = 896.5 nM (95% CI [558.6, 1697]) versus IC₅₀ of sotorasib+anti-CD47 = 10.30 nM (95% CI [2.949, 40.48]). (E) Diagram depicting EGFR-RAS-MAPK signaling pathway. KRAS activation leads to bifurcation of signaling via downstream MAPK pathway elements or the PI3K-AKT pathway. Drugs (red boxes) indicate specific inhibitors of pathway elements used in this study. (F) Growth of GFP+ PC9 cells in co-culture with primary human macrophages in the absence or presence of anti-CD47 antibodies (10 ug/mL) and varying EGFR-RAS-MAPK or PI3K-AKT pathway inhibitors as indicated in (E). Points represent individual replicates, bars represent mean. Data represent three technical replicates per donor using n = 4 independent macrophage donors on day 6.5 of co-culture. (G) Growth of GFP+ NCI-H358 cells in co-culture with primary human macrophages in the presence or absence of anti-CD47 antibodies (10 ug/mL) and varying EGFR-RAS-MAPK or PI3K-AKT pathway inhibitors as indicated in (E). Points represent individual replicates, bars represent mean. Data represent three technical replicates per donor using n = 4 independent macrophage donors on day 6.5 of co-culture. (A, B, F, G) ns, not significant, *p<0.05, **p<0.01, ***p<0. 001, ****p<0.0001 by one-way ANOVA with Holm-Sidak multiple comparisons test.

Figure 5:. The combination of targeted therapy and CD47 blockade enhances anti-tumor responses in mouse tumor models.

(A) EGFR mutant NSCLC xenograft model of PC9 cells engrafted into NSG mice. Tumors were allowed to grow to approximately 500 mm³ and then mice were randomized to treatment with vehicle control, anti-CD47 antibodies (250 ug three times weekly), osimertinib (5 mg/kg five times weekly), or the combination of anti-CD47 plus osimertinib. Tumor volumes were measured over time. Data depict mean tumor volume ± SEM (left), growth curves of individual mice (middle), or percent change in tumor volume from baseline (right). Complete responses were observed in 4/10 mice (40%) in the combination cohort. Data represent n = 9–11 mice per cohort combined from two independent experiments. (B) EGFR mutant NSCLC xenograft model of MGH134-1 patient-derived cells engrafted into NSG mice and treated as in (A). Data represent percent change in tumor volume from baseline with mean ± SEM of n = 4 mice per cohort. (C) ALK-positive xenograft model of NCI-H3122 cells engrafted into NSG mice and treated with vehicle control, anti-CD47 antibodies (250 ug three times weekly), lorlatinib (6 mg/kg five times weekly), or the combination of anti-CD47 antibodies and lorlatinib. Data represent percent change in tumor volume from baseline with mean ± SEM of n = 4 mice per cohort. (D) KRAS^G12C mutant xenograft model of NCI-H358 cells engrafted into NSG mice and treated with vehicle control, anti-CD47 antibodies (250 ug three times weekly), sotorasib (100 mg/kg five times weekly), or the combination of anti-CD47 antibodies and sotorasib. Data represent percent change in tumor volume from baseline with mean ± SEM of n = 4 mice per cohort. (E) Syngeneic tumor model of KRAS^G12C mutant lung cancer using wild-type 3LL ΔNRAS cells or a CD47-knockout variant engrafted into C57BL/6 mice. The mice were treated with vehicle control or sotorasib (30 mg/kg five times weekly) starting on day 7 post-engraftment. Points indicate individual tumor volumes, bars represent median with n = 9–10 mice per treatment cohort. ns, not significant, *p<0.05 by paired t test for the indicated comparisons. (A-D) **p<0.01 by unpaired t test for combo versus targeted therapy.

Figure 6:. Targeted therapies induce cross-sensitization to anti-CD47 therapy and downregulate B2M and CD73.

(A) Diagram showing generation of GFP+ cell lines that are resistant to targeted therapies. For each parental cell line (PC9, NCI-H3122, or NCI-H358), cells were cultured in the presence of 1.0 uM of appropriate targeted therapy for prolonged duration until resistant cells emerged and proliferated in culture. (B-D) Long-term co-culture assays using GFP+ PC9 cells (B), GFP+ NCI-H3122 cells (C), or GFP+ NCI-H358 cells (D) that are resistant to the indicated targeted therapies. In each case, anti-CD47 therapy resulted in significant enhancement in macrophage-mediated cytotoxicity relative to the naïve parental lines. Experiments performed once with 4 independent donors (erlotinib, gefitinib, osimertinib, alectinib, crizotinib resistant lines) or twice with a total of 8 independent donors (lorlatinib, sotorasib resistant lines) with 3 technical replicates per donor. Bars represent means from analysis at 6.5 days of co-culture. (E) Scatter plot showing results of comprehensive surface immunophenotyping of parental NCI-H358 cells versus a GFP+ sotorasib-resistant variant. Each dot represents the normalized mean fluorescence intensity (nMFI) of an individual surface antigen from a total of 354 specificities tested in one experiment. Antigens that exceed the 95% predicted interval for expression on the parental line (red) or resistant line (blue) are indicated. (F) Treatment of parental NCI-H358 cells with the indicated targeted therapies causes downregulation of B2M (top) and CD73 (bottom) over time as measured by flow cytometry. ****p<0.0001 for each drug treatment condition compared to time = 0 h by one-way ANOVA with Tukey’s multiple comparison test. (G) Evaluation of wild-type versus B2M KO lung cancer cell lines in long-term co-culture assays with human macrophages. (H) Evaluation of wild-type versus CD73 KO PC9 cells in long-term co-culture assays with human macrophages. (I) Treatment of PC9 cells with a CD73-blocking antibody alone or in combination with anti-CD47 in long-term co-culture assays with human macrophages. (G-I) Data represent at least two independent experiments performed with 6–12 independent macrophage donors. (B-D, G-I) ns, not significant, *p<0.05, **p<0.001, ***p<0.001, ****p<0.0001 by two-way ANOVA with Holm-Sidak multiple comparisons test; #GFP+ area underrepresented due to high confluency of wells and was not visually different by phase microscopy.