Immune checkpoint blockade (ICB) has revolutionized cancer therapy. However, the response of patients to ICB is difficult to predict. Here, we examined 81 patients with lung cancer under ICB treatment and found that patients with MET amplification were resistant to ICB and had a poor progression-free survival. Tumors with MET amplifications had significantly decreased STING levels and antitumor T-cell infiltration. Furthermore, we performed deep single-cell RNA sequencing on more than 20,000 single immune cells and identified an immunosuppressive signature with increased subsets of XIST- and CD96-positive exhausted natural killer (NK) cells and decreased CD8+ T-cell and NK-cell populations in patients with MET amplification. Mechanistically, we found that oncogenic MET signaling induces phosphorylation of UPF1 and downregulates tumor cell STING expression via modulation of the 3′-UTR length of STING by UPF1. Decreased efficiency of ICB by MET amplification can be overcome by inhibiting MET.

Significance:

We suggest that the combination of MET inhibitor together with ICB will overcome ICB resistance induced by MET amplification. Our report reveals much-needed information that will benefit the treatment of patients with primary MET amplification or EGFR–tyrosine kinase inhibitor resistant-related MET amplification.

This article is highlighted in the In This Issue feature, p. 2659

Immune checkpoint blockade (ICB) immunotherapy targeting immune checkpoint pathways and costimulatory pathways has shown clinical and preclinical benefits in cancer treatment (1). However, ICB in non–small cell lung cancer (NSCLC) centered on the PD-1/PD-L1 axis has variable therapeutic benefit. Only a portion of patients with NSCLC show a response to ICB (1–4). For the biomarkers of ICB resistance, current clinical research suggests that some patients with NSCLC with canonical driver genes are not sensitive to immunotherapy. This includes tumors with EGFR mutations and STK11 mutation together with KRAS mutant [STK11 mutations are associated with resistance to ICB in KRAS mutation but not KRAS wild-type (WT) NSCLC; ref. 5], even though tumors with EGFR mutations always have high PD-L1 expression (6). This may be related to tumor immunogenicity and the tumor microenvironment. Therefore, it is important to understand the biomarkers and mechanisms for immunotherapy response of these patients with NSCLC.

Hepatocyte growth factor receptor (MET) activation has been implicated as an oncogenic driver in NSCLC and can mediate primary and secondary resistance to EGFR tyrosine kinase inhibitors (TKI; refs. 7, 8). However, the role of MET in modulating the antitumor immune response remains unclear. Understanding the immune microenvironment and how MET gene amplification affects immunotherapy is extremely important for the treatment of patients with primary MET amplification or with tumors resistant to EGFR-TKI.

High MET Copy Number Was Related to a Low STING Signal and Predicted Poor Survival and Response to ICB

We present case reports of two patients with lung cancer with WT EGFR and negative ALK. A 57-year-old man (patient 1) was diagnosed 9 months earlier with poorly differentiated adenocarcinoma in the right upper lobe of the lung, with two tumors in the right lung, invasion of the right chest wall and right hilar structure, and enlargement of the supraclavicular lymph node (cT4N3M0; stage IIIC). A 28-year-old man (patient 2) had been diagnosed 6 months earlier with right lung adenocarcinoma (size 5–6 cm), invasion of the right hilum structure, enlargement of the supraclavicular lymph node, and metastasis to the liver (cT3N3M1; stage IV). Sequential chemotherapies (pemetrexed and cisplatin) had been scheduled for these patients after diagnosis; however, the disease progressed, and a series of chemotherapy regimens with cisplatin and paclitaxel were used in the last 4 to 5 months. Despite these therapies, the disease continued to progress. Because these patient tumors show WT EGFR and negative ALK, we further examined MET, which is another driver for lung cancer. Tissues were assessed with hematoxylin and eosin staining, MET IHC, and MET FISH, and MET amplification was identified in both patient tumors (Supplementary Fig. S1A and S1B). Both showed PD-L1 positivity (Supplementary Fig. S1C), a tumor proportion score of 55% (patient 1) or 75% (patient 2), and tumor mutation burden (TMB) of 3.54 (patient 2; Supplementary Fig. S1D).

Because of their progression on chemotherapy, the patients underwent immune therapy with nivolumab 3  mg/kg every 2 weeks for four cycles. Both patients were found to be nonresponsive to anti–PD-1 therapy (Supplementary Fig. S1E–S1H). Brain metastasis developed in one patient (Supplementary Fig. S1G), whereas the other patient developed bone metastasis (Supplementary Fig. S1H). Our follow-up revealed that these patients had shortened progression-free survival (PFS; patient 1, 73 days; patient 2, 55 days; Fig. 1A). Collectively, these two patients were resistant to ICB treatment despite positive PD-L1 expression. Because of the important role of IFN and STING in the efficacy of ICB, we analyzed IFN and STING levels in patient samples using IHC and ELISA and found that the STING and IFNβ levels in these patient samples were extremely low (Supplementary Fig. S1I and S1J). These findings suggest that the resistance to ICB in patients with MET amplification may be related to STING signaling.

Figure 1.

High MET copy number was related to low STING signal and predicted poor survival in ICB. A, The survival and immunotherapy response of the two patients with MET amplification. B, Representative IHC of MET, STING, cGAS, CD8, CCL5, and granzyme B for patients with MET WT or AMP. Scale bar, 100 μm. C, Correlations of MET, STING, cGAS, CD8, CCL5, and granzyme B expression levels with MET WT/AMP status (χ2 test; P values are indicated). D, The treatment plan for patients with lung cancer treated with anti–PD-1. E, Evaluation of the patient's anti–PD-1 response based on the criteria of RECIST1.1. F and G, Kaplan–Meier curve for PFS between patients with MET copy <3 and MET copies 3 to 5 (F) or between patients with MET copy <5 and MET copy >5 (G) after anti–PD-1 therapy in this cohort (log-rank test; P value is indicated). CR, complete response; PD, progressive disease; PR, partial response; SD, stable disease.

Figure 1.

High MET copy number was related to low STING signal and predicted poor survival in ICB. A, The survival and immunotherapy response of the two patients with MET amplification. B, Representative IHC of MET, STING, cGAS, CD8, CCL5, and granzyme B for patients with MET WT or AMP. Scale bar, 100 μm. C, Correlations of MET, STING, cGAS, CD8, CCL5, and granzyme B expression levels with MET WT/AMP status (χ2 test; P values are indicated). D, The treatment plan for patients with lung cancer treated with anti–PD-1. E, Evaluation of the patient's anti–PD-1 response based on the criteria of RECIST1.1. F and G, Kaplan–Meier curve for PFS between patients with MET copy <3 and MET copies 3 to 5 (F) or between patients with MET copy <5 and MET copy >5 (G) after anti–PD-1 therapy in this cohort (log-rank test; P value is indicated). CR, complete response; PD, progressive disease; PR, partial response; SD, stable disease.

Close modal

To elucidate the clinical relevance of MET, STING, cGAS, CCL5, tumor-infiltrating CD8+ T cells, and granzyme B expression in patients with lung cancer with MET amplification (MET copy number >5; n = 11) and MET WT (n = 12), we measured their expression in 23 NSCLC samples by IHC (Fig. 1B; Supplementary Fig. S1K). All these patients were negative for EGFR, RAS, and ALK mutations. We found that the expression level of MET was negatively associated with the status of STING (P = 0.0028), CCL5 (P = 0.0006), CD8 (P = 0.0391), and GZMB (P = 0.0033) but was not associated with cGAS (P = 0.8548; Fig. 1C). Collectively, these data support the notion that MET amplification inhibits anticancer immunity, allowing MET-amplified tumors to escape T-cell killing.

We next interrogated MET copy numbers in cohorts of previously diagnosed patients with NSCLC with updated clinical data in our hospital. We included 229 patients with NSCLC, after excluding 121 patients with no anti–PD-1 or combination ICB therapy and 27 patients with EGFR, RAS, or MET mutations and/or ALK-positive patients, and analyzed the clinical data of the remaining 81 patients (EGFR WT and ALK) who were treated with anti–PD-1 monotherapy (Fig. 1D; Supplementary Table S1). We observed that clinical nonresponders were associated with high (>5) MET copy numbers and related low IFNβ (Fig. 1E; Supplementary Fig. S1L and S1M); conversely, patients with better PFS had relatively low MET copy numbers in this cohort (MET copy number >5 accounted for 10%–15% of NSCLCs, whereas MET/CEP7 >2.2 accounted for about 3%–5%; ref. 9). To better characterize the copy-number variation (CNV) for this association, we compared the PFS of the following groups: MET CNV <3 versus CNV 3–5 (P = 0.23; HR, 1.3; 95% CI, 0.83–2.2), MET CNV <5 versus CNV >5 (P < 0.00017; HR, 3.4; 95% CI, 1.7–6.6), MET CNV <3 versus CNV >5 (P = 0.00014; HR, 4.1; 95% CI, 1.9–9.1), and MET CNV 3–5 versus CNV >5 (P = 0.0028; HR, 2.8; 95% CI, 1.4–5.8; Fig. 1F and G; Supplementary Fig. S1N).

Clinical data from two Memorial Sloan Kettering Cancer Center (MSKCC) cohorts (MSKCC JCO2018, n = 240 and MSKCC Cancer Discovery 2017, n = 860) within NSCLC public cohorts involving immunotherapeutic patients were also analyzed (10, 11). Only four of the patients in the MSKCC JCO2018 cohort had MET-amplified tumors (no STK11 mutant identified) and were treated with anti–PD-1/PD-L1 treatment, and none of them had any durable clinical benefit (DCB; Supplementary Fig. S1O). Twenty-five patients in the MSKCC Cancer Discovery 2017 cohort had MET amplification, six of whom were treated with anti–PD-1/PD-L1 (no STK11 mutant) plus chemotherapy and 19 with the same chemotherapy regimens alone (Supplementary Fig. S1P and S1Q). Strikingly, the mortality rate was 66.7% for the immunotherapy plus chemotherapy group and 26.3% for the group receiving chemotherapy alone (Supplementary Fig. S1R and S1S). We also examined all the major genomic alternations (genes with >5% alternation) and found significant negative correlations between MET amplification and DCB in the MSKCC cohort (DCB: 0/6). Furthermore, we performed multivariate Cox regression analyses on the MSKCC cohort to evaluate the MET AMP as well as other clinicopathologic variables (age, smoke, and TMB). We found that MET amplification was an independent prognostic factor for poor immunotherapy response for both the MSKCC (Supplementary Fig. S2A and S2B) and Hubei cohort (Supplementary Fig. S2C). These results show that patient tumors harboring MET amplification were nonresponsive to ICB and that a high MET copy number (>5) is a new predictor of poor anti–PD-1 immunotherapy efficacy.

We further analyzed the correlation of MET amplification/expression and STING expression and examined baseline immune activity as measured by the cytolytic (CYT) score in several cancers in The Cancer Genome Atlas (12, 13). These data showed that MET amplification or high expression was correlated with low STING levels (Supplementary Fig. S2D–S2F) and CYT scores (Supplementary Fig. S2G and S2H). Together, these findings suggest that MET amplification might be a clinically actionable biomarker and play a role in determining anti–PD-1 responsiveness in lung tumors.

Single-cell RNA Sequencing Reveals the Distinctive Functional Composition of Immune Cells in MET Amplification Patients

Next, we used single-cell RNA sequencing (RNA-seq) to identify which cell populations best described the differences between MET WT responders and MET amplification nonresponders in terms of immune cell frequency and function. We performed the initial analysis with four peripheral blood mononuclear cell (PBMC) samples (two EGFR and KRAS WT, ALK-negative, MET WT responders and two EGFR and KRAS WT, ALK-negative, MET-amplified nonresponders) from patients with NSCLC prior to anti–PD-1 immunotherapy (Fig. 2A). We found decreased CD8+ T cells and natural killer (NK) cells in patients with MET amplification compared with those with MET WT (Fig. 2BE; Supplementary Fig. S3A). Genes associated with the activation and effector function of CD8+ T cells and NK cells were decreased in patients with MET amplification relative to the controls (Fig. 2F). On the other hand, there was a marked increase in the expression of genes associated with T-cell and NK-cell exhaustion in patients with MET amplification, especially CD96 and TIGIT, indicating a suppressive phenotype (Fig. 2G). With deep analysis of T cells and NK cells, we found a new class of exhausted NK cells in patients with MET amplification, a cluster of XTIST/CD96/KLRG1 triple-positive cells with high expression of a six-gene signature (Fig. 2H). We further analyzed the novel CD96+ NK cell separately and found that the phenotype was CD56+CD16. CD96 is a recently described inhibitory immune checkpoint receptor on NK cells and CD8+ T cells and is a new target for cancer immunotherapy to regulate CD8+ T- and NK-cell effector function. Blocking CD96 in combination with PD-1/PD-L1 inhibitors is a new strategy to enhance antitumor activity and suppress tumor growth (14, 15). To confirm these findings from human patients, we used Lewis lung tumor cells and overexpressed MET in Lewis lines and planted them into mice. We also found more CD96+ NK cells in the tumor tissues of the MET overexpression (OE) tumor model (Supplementary Fig. S3B).

Figure 2.

Single-cell RNA-seq analysis reveals the distinctive functional composition of immune cells. A, The schematics for single-cell RNA-seq experiments for PBMCs of patients. B, Whole immune landscape of PBMCs from four patients with lung cancer. The t-distributed stochastic neighbor embedding (t-SNE) projection of immune cells from PBMCs of patients with lung cancer showed the formation of 17 main clusters shown in different colors. C, Whole immune landscape of patients with lung cancer with MET WT (n = 2) and amplification (n = 2). D, The t-SNE projection of immune cells of CD4+ T, CD8+ T, and NK cells. E, The t-SNE projection of CD4+ T, CD8+ T, and NK cells from patients with MET WT and MET-amplified lung cancer. F and G, Unsupervised hierarchical clustering and correspondingly colored t-SNE plot at high resolution (granularity 1.2) of the same data set partition 14 clusters, of which one (boxed area) comprises a marker of cytotoxic and exhausted T and NK cells. H, The phenotype of novel NK-cell signature gene expression related to E.

Figure 2.

Single-cell RNA-seq analysis reveals the distinctive functional composition of immune cells. A, The schematics for single-cell RNA-seq experiments for PBMCs of patients. B, Whole immune landscape of PBMCs from four patients with lung cancer. The t-distributed stochastic neighbor embedding (t-SNE) projection of immune cells from PBMCs of patients with lung cancer showed the formation of 17 main clusters shown in different colors. C, Whole immune landscape of patients with lung cancer with MET WT (n = 2) and amplification (n = 2). D, The t-SNE projection of immune cells of CD4+ T, CD8+ T, and NK cells. E, The t-SNE projection of CD4+ T, CD8+ T, and NK cells from patients with MET WT and MET-amplified lung cancer. F and G, Unsupervised hierarchical clustering and correspondingly colored t-SNE plot at high resolution (granularity 1.2) of the same data set partition 14 clusters, of which one (boxed area) comprises a marker of cytotoxic and exhausted T and NK cells. H, The phenotype of novel NK-cell signature gene expression related to E.

Close modal

We also found a decrease in the expression of genes representing IFNβ and IFNγ responses and a decrease in CD8+ cells, CD4+ T cells (Supplementary Fig. S3C–S3E), and NK cells (Supplementary Fig. S3F–S3H) from PBMCs of patients with MET amplification. MET amplification, therefore, is associated with a global reshaping of the immune compartment and decreased IFN response.

MET Mediates Immune Evasion and Modulates the Efficiency of Anti–PD-1/PD-L1 Immunotherapy via STING Signaling

The clinical studies above established an association between MET amplification and impaired antitumor immunity. However, whether MET amplification directly regulates the ICB response is unclear. To determine whether MET regulates immunity, we performed xenograft studies using animal models. We subcutaneously inoculated MET-overexpressing Lewis lung cancer cells (MET OE) or parental cells into immunodeficient nude mice and found that MET overexpression had no significant effect on tumor growth after 2 weeks (P = 0.079; Fig. 3A). We then subcutaneously inoculated parental or MET OE Lewis lung cancer cells into immunocompetent C57BL/6 mice. We treated mice with IgG or anti–PD-1 antibody ± tivantinib (MET inhibitor) as indicated (Fig. 3B). All the treatment groups and tumor responses are shown in Fig. 3C. Specifically, we found that MET overexpression caused increased tumor growth in untreated mice (Fig. 3D). Consistent with clinical observations, monotherapy with anti–PD-1 was completely ineffective in the MET OE model(P < 0.01; Fig. 3D and E), whereas control mice showed a partial response to anti–PD-1 therapy. The expression of PD-L1 was high in the MET OE models, and this is consistent with our findings above (Supplementary Fig. S4A) and previous reports (16, 17). Patients with lung cancer with high PD-L1 expression are usually sensitive to ICB (18, 19). Therefore, MET amplification- or overexpression-mediated resistance to ICB may not be due to changes in PD-L1 expression. MET OE models were more sensitive than control to treatment with MET inhibitor alone (P < 0.05; Fig. 3D and F). We hypothesized that if MET amplification or overexpression causes resistance to ICB treatment, then the combination of MET inhibitors with ICB would help overcome this resistance. Indeed, the combination of tivantinib and anti–PD-1 antibody greatly improved tumor growth inhibition in the MET OE model (Fig. 3C and G). These results suggest that overexpression of MET induces resistance to ICB, and this can be overcome by METi treatment. To confirm this, we also used the anti–PD-1-sensitive murine lung cancer model CMT167. We found that tumors in the MET OE group were resistant to anti–PD-1 immunotherapy, whereas METi combined with anti–PD-1 could reverse such immunotherapy resistance (Fig. 3H). These results support the combined use of METi and anti–PD-1 in tumors with MET amplification.

Figure 3.

MET mediates immune evasion and modulates the efficiency of anti–PD-1/PD-L1 immunotherapy via STING. A, The growth curve of Lewis cells with MET overexpression and parental Lewis cells in nude mice following reagent intervention with IgG + PBS [100 μL/mouse, intraperitoneally (i.p.); two-sided unpaired t test; P value is indicated]. B, The schematic of the drug intervention protocol, related to C–M. Tumors were measured at the indicated time points. C, The growth curve of parental or MET overexpression Lewis tumors in tivantinib and/or PD-1 antibody-treated C57BL/6 mice. Tumors were measured at the indicated time points (n = 5/group). Combined the curve with different treatments (two-sided unpaired t test; *, P < 0.05; **, P < 0.01). D–G, The growth curve of parental or MET overexpression Lewis tumors in tivantinib and/or PD-1 antibody-treated C57BL/6 mice. Tumors were measured at the indicated time points (n = 5/group). DG, Presentation of different pictures of the same batch of experiments (two-sided unpaired t test;*, P < 0.05; **, P < 0.01). H, The growth curve of MET overexpression CMT167 tumors in tivantinib and/or PD-1 antibody–treated C57BL/6 mouse model. Tumors were measured at the indicated time points (n = 5/group). The drug intervention model is the same as B (two-sided unpaired t test; *, P < 0.05; **, P < 0.01). I, The efficiency of mouse MET overexpression and STING1 knockdown in the Lewis cell line, related to J–M. J–M, Knockdown of STING1 reverses the antitumor effect of combined MET inhibitor and PD-1 blockade in vivo. Tumor growth curves from vehicle, tivantinib alone (100 mg/kg, 4 of 7 days, q.d.), anti–PD-1 alone (100 μg, 3 of 7 days), and anti–PD-1 + tivantitinib treatment groups in C57BL/6 mice with different genotypes (n = 5/group). JM is a presentation of different pictures of the same batch of experiments (two-sided unpaired t test; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant). N, The efficiency of mouse STING overexpression in the Lewis cell line, related to O. O, STING overexpression rescued responsiveness to immunotherapy in MET OE tumors in vivo. Tumor growth curves from vehicle and anti–PD-1 alone (100 μg, 3 of 7 days) treatment groups in C57BL/6 mice with vector or MET overexpression (MET OE) ± STING overexpression (STING OE) groups (n = 5/group; two-sided unpaired t test; ***, P < 0.001).

Figure 3.

MET mediates immune evasion and modulates the efficiency of anti–PD-1/PD-L1 immunotherapy via STING. A, The growth curve of Lewis cells with MET overexpression and parental Lewis cells in nude mice following reagent intervention with IgG + PBS [100 μL/mouse, intraperitoneally (i.p.); two-sided unpaired t test; P value is indicated]. B, The schematic of the drug intervention protocol, related to C–M. Tumors were measured at the indicated time points. C, The growth curve of parental or MET overexpression Lewis tumors in tivantinib and/or PD-1 antibody-treated C57BL/6 mice. Tumors were measured at the indicated time points (n = 5/group). Combined the curve with different treatments (two-sided unpaired t test; *, P < 0.05; **, P < 0.01). D–G, The growth curve of parental or MET overexpression Lewis tumors in tivantinib and/or PD-1 antibody-treated C57BL/6 mice. Tumors were measured at the indicated time points (n = 5/group). DG, Presentation of different pictures of the same batch of experiments (two-sided unpaired t test;*, P < 0.05; **, P < 0.01). H, The growth curve of MET overexpression CMT167 tumors in tivantinib and/or PD-1 antibody–treated C57BL/6 mouse model. Tumors were measured at the indicated time points (n = 5/group). The drug intervention model is the same as B (two-sided unpaired t test; *, P < 0.05; **, P < 0.01). I, The efficiency of mouse MET overexpression and STING1 knockdown in the Lewis cell line, related to J–M. J–M, Knockdown of STING1 reverses the antitumor effect of combined MET inhibitor and PD-1 blockade in vivo. Tumor growth curves from vehicle, tivantinib alone (100 mg/kg, 4 of 7 days, q.d.), anti–PD-1 alone (100 μg, 3 of 7 days), and anti–PD-1 + tivantitinib treatment groups in C57BL/6 mice with different genotypes (n = 5/group). JM is a presentation of different pictures of the same batch of experiments (two-sided unpaired t test; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant). N, The efficiency of mouse STING overexpression in the Lewis cell line, related to O. O, STING overexpression rescued responsiveness to immunotherapy in MET OE tumors in vivo. Tumor growth curves from vehicle and anti–PD-1 alone (100 μg, 3 of 7 days) treatment groups in C57BL/6 mice with vector or MET overexpression (MET OE) ± STING overexpression (STING OE) groups (n = 5/group; two-sided unpaired t test; ***, P < 0.001).

Close modal

Because we observed that MET amplification/overexpression is associated with decreased STING signaling, we next determined the extent to which the resistance to anti–PD-1 blockade is mediated through STING downregulation. For this purpose, we knocked down STING in MET OE or parental Lewis lung cancer cells. The STING-depleted tumor cells were then injected into C57BL/6 mice (Fig. 3IM). STING knockdown significantly increased tumor growth in the control tumor models but had no significant effect in MET OE models (Fig. 3J). Similarly, in mouse models treated with anti–PD-1, the control mice were no longer sensitive to anti–PD-1 treatment when STING was depleted; however, STING depletion had no additional effect on anti–PD-1 response in the MET OE arms (Fig. 3K), suggesting that MET functions through STING. MET OE models were more sensitive to METi or METi + anti–PD-1 treatment than the controls, as we showed before (Fig. 3L and M). However, STING depletion rendered both MET OE and control models resistant to these treatments. These results suggest that MET mediates immune evasion largely by inhibiting STING. To further confirm this, we overexpressed STING in MET OE models (Fig. 3N and O). We found that immunotherapy resistance induced by MET OE could be overcome by STING OE (Fig. 3O), again suggesting that STING is a key effector regulated MET. In vitro, we used the MET inhibitor tivantinib to treat the MET-amplified human lung cancer cell line H1993 for 48 hours and then cocultured it with PBMCs for 7 days with or without anti–PD-1. Consistent with the in vivo models, anti–PD-1 treatment was not effective against H1993. However, combining METi with anti–PD-1 had a strong inhibitory effect against this cell line (Supplementary Fig. S4B). In addition, we collected biopsy tumor tissues from a patient with MET amplification and autogenous PBMCs of the same patient. MET-amplified patient tumor cells were then cocultured with STING WT or STING kinase-dead (KD) autogenous PBMCs under different treatment groups (IgG, MET inhibitor, anti–PD-1, MET inhibitor + anti–PD-1). Interestingly, we found that MET inhibitor improved the efficacy of immunotherapy in patients with MET amplification/overexpression in vitro independent of host (PBMC) STING (Supplementary Fig. S4C). To further investigate the effects of MET and anti–PD-1 combination in the lung cancer microenvironment, we performed immune profiling on xenograft tumor models (Supplementary Fig. S5A). MET OE tumors had immune infiltrates containing lower levels of CD8+ cytotoxic T cells and NK cells than control tumors in the IgG or anti–PD-1 arm (Supplementary Fig. S5B–S5F). Both CD8+ T cells and NK cells are critical for controlling tumor growth (20, 21). METi treatment significantly boosted CD8+ cytotoxic T cells and NK infiltration in MET OE tumors, whereas only a moderate boost was observed in control tumors (Supplementary Fig. S5B–S5D). However, there was no significant difference in the number of CD4+ T cells (Supplementary Fig. S5G). All these observations suggest that MET mediates immune evasion and modulates the efficiency of anti–PD-1/PD-L1 immunotherapy via downregulating STING signaling in tumor, and this can be overcome by MET inhibitors or STING overexpression.

Intratumoral STING Signaling Is Required for T-cell and NK-cell Recruitment in MET Amplified/Overexpressing Tumors Treated with METi

To confirm that the STING pathway is affected by MET overexpression, we examined the expression of factors in the STING pathway in Lewis tumors. Consistent with what we found in human samples, we found that the expression of STING was decreased in MET OE models, as were the downstream signals pTBK1 and pIRF3 (Supplementary Fig. S6A). We also examined STING and MET levels in MET OE animal models (as in Fig. 3) in the course of therapy at different time points. These results suggest that the STING pathway is inhibited by MET overexpression in vivo (Supplementary Fig. S6B and S6C). When we treated MET OE cells with METi, we found that METi treatment increased STING protein expression and pTBK1 and pIRF3 signals (Supplementary Fig. S6A). The mRNA levels of STING1 and IFNB1 were also increased by METi (Supplementary Fig. S6D and S6E), suggesting that METi regulates STING through affecting mRNA levels. The STING pathway has been reported to regulate the expression of chemokines such as CCL5 and CXCL10. These two chemokines are key mediators for the chemotaxis of CD8+ T lymphocytes, and CXCL10 and CCL5 overexpression is associated with the presence of CD8+ T lymphocytes in lung cancer, melanoma, gastric cancer, and colorectal cancer (22, 23). MET inhibitor treatment also caused significant upregulation of CXCL10 (P < 0.05) and CCL5(P < 0.001) mRNA expression in the MET OE cell line, consistent with enhanced STING signaling (Supplementary Fig. S6F and S6G). The upregulation of IFNβ, CXCL10, and CCL5 by METi was blocked by short hairpin RNA (shRNA)–mediated knockdown of STING1 (Supplementary Fig. S6D–S6G), suggesting that METi enhances antitumor immune signaling through STING. Furthermore, we used MET inhibitor to treat the H1993 cell line for 48 hours and, after washing the MET inhibitor, cocultured H1993 cells with PBMCs for 4 days. MET inhibitor treatment caused significant upregulation of CD107+CD8+ T cells and CD107+CD56+ NK cells, and this upregulation could be blocked by STING knockdown (Supplementary Fig. S6H and S6I). Overall, these results suggest that MET inhibitor–induced enhanced T-cell and NK-cell recruitment and effector immune response require STING.

MET Regulates STING Expression by UPF1

To systematically evaluate the regulation of STING by MET, we generated STING1 knockdown cell lines in the context of two MET-amplified cell lines (H1993 and H820) and treated cells with c-GAMP and a MET inhibitor. We found that METi upregulated STING and downstream signaling (Supplementary Fig. S7A and S7B). The MET inhibitor also potentiated STING-dependent activation of endogenous TBK1 and IRF3, which was abolished in STING-knockdown cells (Supplementary Fig. S7A and S7B). Next, we examined the potential mechanism of oncogenic MET signaling in the regulation of STING expression. We found that STING1 mRNA increased rapidly following MET knockdown. Reexpression of MET WT but not a MET KD reversed the upregulation of STING (Supplementary Fig. S7C). These results suggest that MET kinase activity regulates STING mRNA levels. We investigated the possibility that oncogenic MET regulates the stability of STING1 mRNA using actinomycin D. We found that STING1 mRNA stability was decreased by MET activation, and this decrease could be blocked by MET inhibitor, suggesting that MET signaling regulates STING1 mRNA stability (Supplementary Fig. S7D).

We speculated that RNA binding proteins might mediate the regulation of STING expression downstream of MET. Therefore, we performed a selected shRNA screen of candidate genes—ELAV, HNRNPD, and UPF1 (shRNA efficiency shown in Fig. 4A). These factors are able to bind and target ARE-containing mRNAs for rapid degradation or stabilization (24, 25). shRNA-mediated knockdown of UPF1 most consistently increased STING1 mRNA levels across the cell line panel, whereas knockdown of ELAV and HNRNPD did not significantly affect STING1 mRNA levels (Fig. 4B; Supplementary Fig. S7E). UPF1 has been shown to bind mRNAs in a 3′ untranslated region (UTR) length-dependent manner and potentiate mRNA decay. To investigate whether MET regulates STING1 mRNA via UPF1, we silenced UPF1 expression using shRNA in the context of MET activation using HGF stimulation or MET overexpression. UPF1 knockdown was partially able to rescue the decrease in STING expression caused by MET activation (Supplementary Fig. S7F). Crucially, we found that MET-regulated STING1 mRNA stability could be restored by UPF1 knockdown (Supplementary Fig. S7G).

Figure 4.

MET regulates STING expression through UPF1. A, The efficiency of shRNAs targeting the indicated RNA binding proteins in H820 cells and H1993 cells. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01; ***, P < 0.001). B, qPCR analysis of STING mRNA level after transfection with indicated shRNAs in H820 cells. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01; ns, not significant). C, qPCR analysis of UPF1 mRNA levels in H1993 and H820 cells after treatment with tivantinib (200 nmol/L) for 48 hours. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; ns, not significant). D, qPCR analysis of STING1 mRNA level in H1993 and H820 cells after overexpression of UPF1. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01). E, Western blotting analysis of UPF1 tyrosine phosphorylation after immunoprecipitations from H1993 cells transfected with indicated constructs. F, Western blotting analysis of UPF1 tyrosine phosphorylation in H1993 and H820 cells expressing the indicated constructs 48 hours posttransfection. G, qPCR analysis of the 3′ UTR length of STING1 in cells transfected with control shRNA or shUPF1 cells treated with or without METi. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01; ns, not significant). H, qPCR analysis of RNA immunoprecipitation (RNA-IP) in H1993 and H820 cells using UPF1 antibody. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01). I, qPCR analysis of the STING1 mRNA level in UPF1 depletion H1993 cells reexpressing UPF1 WT or UPF1 Y818F mutant treated with or without METi. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01). J and K, UPF1-depleted H1993 cells reexpressing WT UPF1 or UPF1Y818F mutant were treated with METi (200 nmol/L for 48 hours) and cocultured with PBMCs (tumor/PBMCs = 1:10) for 96 hours after washing away excess tivantinib. FACS analysis of effector CD8+CD107+ T cells and CD56+CD107+ NK cells from cocultured PBMCs. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01). L, Lewis tumor cells expressing vector or MET were transfected with constructs encoding UPF1 WT and UPF1 Y818 mutant and implanted into mice. Mice were treated with IgG or anti–PD-1 (100 μg, 3 of 7 days) and tumor growth was measured (n = 5/group; two-sided unpaired t test; ***, P < 0.001).

Figure 4.

MET regulates STING expression through UPF1. A, The efficiency of shRNAs targeting the indicated RNA binding proteins in H820 cells and H1993 cells. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01; ***, P < 0.001). B, qPCR analysis of STING mRNA level after transfection with indicated shRNAs in H820 cells. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01; ns, not significant). C, qPCR analysis of UPF1 mRNA levels in H1993 and H820 cells after treatment with tivantinib (200 nmol/L) for 48 hours. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; ns, not significant). D, qPCR analysis of STING1 mRNA level in H1993 and H820 cells after overexpression of UPF1. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01). E, Western blotting analysis of UPF1 tyrosine phosphorylation after immunoprecipitations from H1993 cells transfected with indicated constructs. F, Western blotting analysis of UPF1 tyrosine phosphorylation in H1993 and H820 cells expressing the indicated constructs 48 hours posttransfection. G, qPCR analysis of the 3′ UTR length of STING1 in cells transfected with control shRNA or shUPF1 cells treated with or without METi. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01; ns, not significant). H, qPCR analysis of RNA immunoprecipitation (RNA-IP) in H1993 and H820 cells using UPF1 antibody. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01). I, qPCR analysis of the STING1 mRNA level in UPF1 depletion H1993 cells reexpressing UPF1 WT or UPF1 Y818F mutant treated with or without METi. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01). J and K, UPF1-depleted H1993 cells reexpressing WT UPF1 or UPF1Y818F mutant were treated with METi (200 nmol/L for 48 hours) and cocultured with PBMCs (tumor/PBMCs = 1:10) for 96 hours after washing away excess tivantinib. FACS analysis of effector CD8+CD107+ T cells and CD56+CD107+ NK cells from cocultured PBMCs. Data are presented as mean ± SEM from three independent experiments (two-sided unpaired t test; **, P < 0.01). L, Lewis tumor cells expressing vector or MET were transfected with constructs encoding UPF1 WT and UPF1 Y818 mutant and implanted into mice. Mice were treated with IgG or anti–PD-1 (100 μg, 3 of 7 days) and tumor growth was measured (n = 5/group; two-sided unpaired t test; ***, P < 0.001).

Close modal

MET inhibition did not affect the mRNA level of UPF1 (Fig. 4C), and UPF1 overexpression was sufficient to significantly decrease STING expression (Fig. 4D). Next, we tried to determine how MET signaling regulates UPF1. We found that UPF1 can be phosphorylated at tyrosine residues in a MET kinase-dependent manner (Fig. 4E). We carried out experiments that aimed at mapping phosphorylation sites on UPF1 in MET WT and MET knockout cells (Supplementary Fig. S8A). Mass spectrometry showed UPF1 to be phosphorylated at Y818 by MET (Supplementary Fig. S8B and S8C). To confirm that MET can induce UPF1 phosphorylation at Y818, we mutated UPF1 and found that mutation of Y818 abolished UPF1 tyrosine phosphorylation in cells with MET amplification (Fig. 4F). Furthermore, inhibiting MET rapidly increased the 3′ UTR length of STING1 mRNA; UPF1 knockdown also increased the 3′ UTR length of STING1 mRNA, and no additional effect was observed when METi was used (Fig. 4G). This collectively demonstrates that MET signaling decreases 3′ UTR length of STING1 through UPF1. Crucially, we found that endogenous UPF1 precipitated with STING mRNA in RNA immunoprecipitation reactions from MET-amplified lung cancer cells, H1993 and H820 cells (Fig. 4H). Furthermore, we used anti-UPF1 antibody to pull down UPF1 binding RNA in H1993 cells treated with or without METi and performed RIP-seq to analyze the changes of the associated mRNA. We found that STING is the most important immune-related pathway associated with UPF1 (Supplementary Fig. S8D and S8E), and METi decreased the binding of UPF1 to STING1 mRNA (Supplementary Fig. S8D). In addition, MET inhibition could increase the expression of STING in cells expressing WT UPF1 (Fig. 4I). Cells expressing the Y818F mutant UPF1 also showed increased STING expression (Supplementary Fig. S8F), whereas no additional increase of STING expression was observed in these cells treated with METi (Fig. 4I). These results demonstrate that MET signaling decreases STING mRNA level specifically by phosphorylating UPF1 at Y818. Moreover, we treated cells expressing UPF1 WT and UPF1 Y818F mutant with a MET inhibitor for 48 hours, and after washing cells were cocultured with PBMCs for 4 days. MET inhibitor treatment caused significant upregulation of CD107+CD8+ T cells and CD107+CD56+ NK cells in cells expressing WT UPF1 (Fig. 4J and K); cells expressing UPF1 Y818F had upregulation of CD107+CD8+ T cells and CD107+CD56+ NK cells in untreated cells, and METi could not further activate CD107+CD8+ T cells and CD107+CD56+ NK cells in this culture system (Fig. 4J and K), further supporting the important role of Y818 phosphorylation in MET-regulated immune function. We further tested the role of Y818 in antitumor immune response in vivo using the MET OE models. We found that immunotherapy resistance induced by MET OE could be largely blocked by the expression of UPF1 Y818F mutant in vivo (Fig. 4L). Collectively, these data demonstrate the functional importance of the MET–UPF1 pathway in tumor-immune evasion by directly regulating STING1 mRNA degradation (Supplementary Fig. S8G).

We have shown that MET amplification in lung tumor results in weakened IFN response mediated by STING and less tumor-infiltrating CD8 T and NK cells, leading to decreased tumor immunogenicity and resistance to ICB therapy. These effects can be reversed by MET inhibition. These findings may lead to a new therapeutic strategy by combining MET inhibitor and ICB to overcome resistance to immunotherapy for MET-amplified NSCLC.

Tumors lacking T-cell and NK-cell infiltration and low immunogenicity often fail to respond to ICB. Strategies to inflame the tumor microenvironment are a high therapeutic priority. These include increased delivery and epigenetic derepression of innate immune regulators. Recently, some reports have shown that STING agonists can enhance NK-cell activation, cytotoxicity, and antitumor effects in part by inducing type I IFN (26–28). Our data suggest that inhibition of MET signaling in MET-amplified lung cancer cells can induce the expression of STING to increase antitumor immunity and overcome resistance to checkpoint blockade.

Mechanistically, we found MET signaling reduced STING mRNA stability through phosphorylation of UPF1 at Y818, which enhances UPF1 binding to STING1 mRNA and promotes a decrease of STING1 mRNA, consistent with a previous report that UPF1 senses 3′ UTR length to potentiate mRNA decay (29). UPF1 is a multifunctional RNA and DNA helicase best known for its essential role in nonsense-mediated mRNA decay (NMD), which is an mRNA turnover pathway that targets mRNAs with premature translation termination codons for rapid degradation (30–32). Tumor-targeted NMD inhibition is a new approach to stimulate protective antitumor immunity (33, 34). Our studies suggest that inhibition of the MET–UPF1 pathway rescues antitumor immunity. This opens up new directions to stimulate antitumor innate immunity in patients with MET amplification. How MET-induced UPF1 phosphorylation at Y818 regulates its binding to STING mRNA and other mRNAs is not clear. MET seems to enhance UPF1′s binding to some mRNAs (e.g., STING1) while decreasing UPF1′s binding to other mRNAs (e.g., TMEM221). Future studies are needed to address the underlying mechanism.

Our findings for the first time suggest that MET copy number is a key factor in the response to ICB in patients with lung cancer. Our findings may be readily translated into the clinic, and MET copy number as a biomarker may be applied to the design of novel therapeutic regimens for MET amplification in patients with lung cancer. These findings should be further tested in a prospective clinical trial to confirm the function of MET copy-number prediction in ICB.

Study Population

We reviewed the medical records of 229 patients with NSCLC diagnosed with MET FISH at the Hubei Cancer Hospital between June 2018 and July 2020. EGFR-activating mutations were defined as mutations associated with exon 19 deletion, G719X, L858R, and L861Q. We excluded 27 patients with EGFR, RAS, or MET mutations and/or ALK-positive patients and 121 patients who did not receive anti–PD-1/PD-L1 or received combination chemotherapy alongside anti–PD-1/PD-L1 treatment. Finally, 81 EGFR WT, ALK-negative patients with anti–PD-1 treatment (nivolumab 3  mg/kg every 2 weeks and pembrolizumab 2 mg/kg every 3 weeks) for NSCLC were included in this study (baseline characteristics of patients are shown in Supplementary Table S1). ICB was administered throughout the study until patients had no evidence of continued clinical benefit, unacceptable toxicity, or withdrawal of consent. Tumor responses were assessed by a clinical doctor based on RECIST V.1.1. Patients who achieved partial response or complete response (CR) were included in this study as responders, compared with patients with progressive disease on treatment as nonresponders. No CR was observed in this patient cohort. All the patients' data and specimens were obtained from two cancer centers. We and our colleagues obtained written informed consent from patients, and this study was approved by the Institutional Review Board of Hubei Cancer Hospital and Cancer Center, Union Hospital, Tongji Medical College, and informed consent was obtained from all patients before the use of specimens for analysis.

MET FISH and ALK IHC

To perform FISH hybridization, the fluorescence probes were reconstituted with hybridization buffer, overlaid on the tissue microarray slides, and covered with coverslips. The hybridization reaction was performed using the Dako Hybridizer by first heating to 90°C for 5 minutes, followed by gradual cooling to 37°C and incubation for 14 to 20 hours. Nonspecific binding was removed by washing three times with diluted stringent wash buffer at 60°C. The sections were dehydrated, mounted in medium with DAPI (Invitrogen), allowed to stand at room temperature for 24 hours in the dark, and then stored at 4°C. FISH signals were observed and counted using a fluorescence microscope (Axio Scope A1; Zeiss). The level of MET amplification was evaluated in deparaffinized 4-mm sections using dual-color FISH and a MET/CEN7q Dual Color FISH Probe (Vysis; Abbott Laboratories). Amplification was considered present in cases with a MET/CEN7 ratio of greater than 2.0 and an average MET gene copy number greater than 5.0 per cell.

Cell Lines and Cell Culture

NSCLC cell lines A549, H1993, H820, and Lewis were purchased from ATCC between 2015 and 2018. CMT-167 was provided by Dr. Alan Fields in 2020. Human lung cancer cell lines were cultured with RPMI 1640 complete medium supplemented with 10% FBS and antibiotics (100 U/mL streptomycin and 100 U/mL penicillin). The murine lung cancer cell line Lewis was propagated in DMEM complete media. PBMCs derived from patients with MET amplification lung cancer were grown in RPMI, and the medium was supplemented with 50 mmol/L 2-mercaptoethanol. All the cell lines were routinely tested for Mycoplasma infections. All cell lines were Mycoplasma tested every 3 months using the MycoProbe Mycoplasma Detection Kit (R&D Systems). The length of time between cell line thawing and use in experiment did not exceed 1 month (two or more passages). All cell lines were authenticated by short tandem repeat DNA fingerprinting every 1 to 2 years and by visual inspection prior to experiments and freezing of cell line stocks in a central cell bank.

Animal Studies

Six- to eight-week-old female C57BL/6 mice and nude female mice were purchased from The Jackson Laboratory. Animal studies were approved by the Institutional Animal Care and Use Committee at Tongji Medical College, Huazhong University of Science and Technology.

Tumorigenicity assays were performed using mouse subcutaneous Lewis lung cancer models. For the subcutaneous tumor model, Lewis cells (5 × 106) were subcutaneously injected into the right inguinal fold regions of nude mice and C57BL/6 mice. Mice were randomly assigned to groups according to mean tumor volume. For antibody-based drug intervention, 100 μg anti-mouse PD-1 antibody (RMP1-14; Bio X Cell) or rat IgG (control; BioXCell) was injected intraperitoneally every 3 days starting day 3 after tumor cell inoculation. For drug-based intervention, mice were given daily oral doses of 100 mg/kg tivantinib formulated in tocopherol polyethylene glycol 1000 succinate (BioXtra, water-soluble vitamin E conjugate). Subcutaneous tumors were measured using a caliper, and tumors were evaluated using high-frequency ultrasound (VevoTumor sizes were measured by calipers, and the tumor volume was calculated using the following formula: W2 × L × 0.5, where L is the longest dimension and W is the corresponding perpendicular dimension). At the experimental endpoint, mice were killed using CO2 exposure, and tumors were excised for subsequent histologic analysis or processed immediately for Western blot and flow cytometric analyses.

Tumor-Infiltrating Immune Cell Characterization by FACS

Under anesthesia, the tumor mass was isolated from mice. Tumor tissues were minced into small pieces and digested for 1 hour with collagenase type V (320 μg/mL), hyaluronidase (500 μg/mL), and DNase I (5 μg/mL; Sigma-Aldrich) under rapid shaking at 37°C. After incubation, cells were treated with red blood cell lysis buffer and filtered through a 70-μm cell strainer to remove debris. Then, the tumor-infiltrating immune cells were enriched by Percoll gradient (GE Healthcare). The cells were collected and used for flow cytometry analyses. Antibodies are listed in Supplementary Table S2.

Statistical Analysis

Data in bar and line graphs are presented as mean ± SEM of three independent experiments. Statistical analyses were performed with the Student t test. Statistical significance is represented in figures by *, P < 0.05; **, P < 0.01; ***, P < 0.001.

No disclosures were reported.

Y. Zhang: Resources, supervision, methodology, writing–original draft, project administration. Q. Yang: Resources, data curation.X. Zeng: Resources, methodology. M. Wang: Resources, formal analysis, investigation. S. Dong: Resources, data curation, investigation. B. Yang: Resources, data curation. X. Tu: Data curation, software.T. Wei: Software, formal analysis. W. Xie: Resources, data curation. C. Zhang: Resources, data curation. Q. Guo: Software, methodology. J.A. Kloeber: Writing–original draft. Y. Cao: Investigation. G. Guo: Data curation, validation. Q. Zhou: Data curation. F. Zhao: Data curation. J. Huang: Data curation, investigation. L. Liu: Supervision, funding acquisition. K. Zhang: Software, visualization. M. Wang: Data curation, methodology. P. Yin: Validation, investigation.K. Luo: Data curation. M. Deng: Formal analysis. W. Kim: Validation. J. Hou: Investigation. Y. Shi: Software. Q. Zhu: Resources, validation. L. Chen: Formal analysis. S. Hu: Supervision, validation. J. Yue: Supervision, visualization. G. Pi: Resources, supervision, investigation. Z. Lou: Supervision, project administration, writing–review and editing.

The MET kinase dead and WT plasmid were from Professor Dihua Yu at the University of Texas MD Anderson Cancer Center (Houston, TX).

This work was supported by grants from the Mayo Foundation, the National Natural Science Foundation of China (no. 81773056, 81602501, 81972308), and two grants from the Natural Science Foundations of Hubei Province (no. 2016CFB217, 2018CFC846).

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.

1.
Reck
M
,
Rodriguez-Abreu
D
,
Robinson
AG
,
Hui
R
,
Csoszi
T
,
Fulop
A
, et al
Pembrolizumab versus chemotherapy for PD-L1-positive non-small-cell lung cancer
.
N Engl J Med
2016
;
375
:
1823
33
.
2.
Mok
TSK
,
Wu
YL
,
Kudaba
I
,
Kowalski
DM
,
Cho
BC
,
Turna
HZ
, et al
Pembrolizumab versus chemotherapy for previously untreated, PD-L1-expressing, locally advanced or metastatic non-small-cell lung cancer (KEYNOTE-042): a randomised, open-label, controlled, phase 3 trial
.
Lancet
2019
;
393
:
1819
30
.
3.
Topalian
SL
,
Hodi
FS
,
Brahmer
JR
,
Gettinger
SN
,
Smith
DC
,
McDermott
DF
, et al
Safety, activity, and immune correlates of anti-PD-1 antibody in cancer
.
N Engl J Med
2012
;
366
:
2443
54
.
4.
Hamid
O
,
Robert
C
,
Daud
A
,
Hodi
FS
,
Hwu
WJ
,
Kefford
R
, et al
Safety and tumor responses with lambrolizumab (anti-PD-1) in melanoma
.
N Engl J Med
2013
;
369
:
134
44
.
5.
Ricciuti
B
,
Arbour
KC
,
Lin
JJ
,
Vokes
N
,
Hoojghan
AV
,
Li
YY
, et al
Effect of STK11 mutations on efficacy of PD-1 inhibition in non-small cell lung cancer (NSCLC) and dependence on KRAS mutation status
.
J Clin Oncol
2020
;
38
:
e15113
.
6.
Lisberg
A
,
Cummings
A
,
Goldman
JW
,
Bornazyan
K
,
Reese
N
,
Wang
T
, et al
A phase II study of pembrolizumab in EGFR-mutant, PD-L1+, tyrosine kinase inhibitor naive patients with advanced NSCLC
.
J Thorac Oncol
2018
;
13
:
1138
45
.
7.
Engelman
JA
,
Zejnullahu
K
,
Mitsudomi
T
,
Song
Y
,
Hyland
C
,
Park
JO
, et al
MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling
.
Science
2007
;
316
:
1039
43
.
8.
Turke
AB
,
Zejnullahu
K
,
Wu
YL
,
Song
Y
,
Dias-Santagata
D
,
Lifshits
E
, et al
Preexistence and clonal selection of MET amplification in EGFR mutant NSCLC
.
Cancer Cell
2010
;
17
:
77
88
.
9.
Finocchiaro
G
,
Toschi
L
,
Gianoncelli
L
,
Baretti
M
,
Santoro
A
. 
Prognostic and predictive value of MET deregulation in non-small cell lung cancer
.
Ann Transl Med
2015
;
3
:
83
.
10.
Rizvi
H
,
Sanchez-Vega
F
,
La
K
,
Chatila
W
,
Jonsson
P
,
Halpenny
D
, et al
Molecular determinants of response to anti-programmed cell death (PD)-1 and anti-programmed death-ligand 1 (PD-L1) blockade in patients with non-small-cell lung cancer profiled with targeted next-generation sequencing
.
J Clin Oncol
2018
;
36
:
633
41
.
11.
Jordan
EJ
,
Kim
HR
,
Arcila
ME
,
Barron
D
,
Chakravarty
D
,
Gao
J
, et al
Prospective comprehensive molecular characterization of lung adenocarcinomas for efficient patient matching to approved and emerging therapies
.
Cancer Discov
2017
;
7
:
596
609
.
12.
Mandal
R
,
Samstein
RM
,
Lee
KW
,
Havel
JJ
,
Wang
H
,
Krishna
C
, et al
Genetic diversity of tumors with mismatch repair deficiency influences anti-PD-1 immunotherapy response
.
Science
2019
;
364
:
485
91
.
13.
Rooney
MS
,
Shukla
SA
,
Wu
CJ
,
Getz
G
,
Hacohen
N
. 
Molecular and genetic properties of tumors associated with local immune cytolytic activity
.
Cell
2015
;
160
:
48
61
.
14.
Blake
SJ
,
Dougall
WC
,
Miles
JJ
,
Teng
MW
,
Smyth
MJ
. 
Molecular pathways: targeting CD96 and TIGIT for cancer immunotherapy
.
Clin Cancer Res
2016
;
22
:
5183
8
.
15.
Chan
CJ
,
Martinet
L
,
Gilfillan
S
,
Souza-Fonseca-Guimaraes
F
,
Chow
MT
,
Town
L
, et al
The receptors CD96 and CD226 oppose each other in the regulation of natural killer cell functions
.
Nat Immunol
2014
;
15
:
431
8
.
16.
Martin
V
,
Chiriaco
C
,
Modica
C
,
Acquadro
A
,
Cortese
M
,
Galimi
F
, et al
Met inhibition revokes IFNgamma-induction of PD-1 ligands in MET-amplified tumours
.
Br J Cancer
2019
;
120
:
527
36
.
17.
Saigi
M
,
Alburquerque-Bejar
JJ
,
Mc Leer-Florin
A
,
Pereira
C
,
Pros
E
,
Romero
OA
, et al
MET-oncogenic and JAK2-inactivating alterations are independent factors that affect regulation of PD-L1 expression in lung cancer
.
Clin Cancer Res
2018
;
24
:
4579
87
.
18.
Gibney
GT
,
Weiner
LM
,
Atkins
MB
. 
Predictive biomarkers for checkpoint inhibitor-based immunotherapy
.
Lancet Oncol
2016
;
17
:
542
51
.
19.
Cristescu
R
,
Mogg
R
,
Ayers
M
,
Albright
A
,
Murphy
E
,
Yearley
J
, et al
Pan-tumor genomic biomarkers for PD-1 checkpoint blockade-based immunotherapy
.
Science
2018
;
362
:
3593
.
20.
Ferrari de Andrade
L
,
Tay
RE
,
Pan
D
,
Luoma
AM
,
Ito
Y
,
Badrinath
S
, et al
Antibody-mediated inhibition of MICA and MICB shedding promotes NK cell-driven tumor immunity
.
Science
2018
;
359
:
1537
42
.
21.
Li
H
,
van der Leun
AM
,
Yofe
I
,
Lubling
Y
,
Gelbard-Solodkin
D
,
van Akkooi
ACJ
, et al
Dysfunctional CD8 T cells form a proliferative, dynamically regulated compartment within human melanoma
.
Cell
2019
;
176
:
775
89
.
22.
Sceneay
J
,
Goreczny
GJ
,
Wilson
K
,
Morrow
S
,
DeCristo
MJ
,
Ubellacker
JM
, et al
Interferon signaling is diminished with age and is associated with immune checkpoint blockade efficacy in triple-negative breast cancer
.
Cancer Discov
2019
;
9
:
1208
27
.
23.
Dangaj
D
,
Bruand
M
,
Grimm
AJ
,
Ronet
C
,
Barras
D
,
Duttagupta
PA
, et al
Cooperation between constitutive and inducible chemokines enables T cell engraftment and immune attack in solid tumors
.
Cancer Cell
2019
;
35
:
885
900
.
24.
Kim
YK
,
Furic
L
,
Desgroseillers
L
,
Maquat
LE
. 
Mammalian Staufen1 recruits Upf1 to specific mRNA 3′UTRs so as to elicit mRNA decay
.
Cell
2005
;
120
:
195
208
.
25.
Peng
SS
,
Chen
CY
,
Xu
N
,
Shyu
AB
. 
RNA stabilization by the AU-rich element binding protein, HuR, an ELAV protein
.
EMBO J
1998
;
17
:
3461
70
.
26.
Kitajima
S
,
Ivanova
E
,
Guo
S
,
Yoshida
R
,
Campisi
M
,
Sundararaman
SK
, et al
Suppression of STING associated with LKB1 loss in KRAS-driven lung cancer
.
Cancer Discov
2019
;
9
:
34
45
.
27.
Deng
L
,
Liang
H
,
Xu
M
,
Yang
X
,
Burnette
B
,
Arina
A
, et al
STING-dependent cytosolic DNA sensing promotes radiation-induced type I interferon-dependent antitumor immunity in immunogenic tumors
.
Immunity
2014
;
41
:
843
52
.
28.
Nicolai
CJ
,
Wolf
N
,
Chang
IC
,
Kirn
G
,
Marcus
A
,
Ndubaku
CO
, et al
NK cells mediate clearance of CD8(+) T cell-resistant tumors in response to STING agonists
.
Sci Immunol
2020
;
5
:
2738
.
29.
Hogg
JR
,
Goff
SP
. 
Upf1 senses 3′UTR length to potentiate mRNA decay
.
Cell
2010
;
143
:
379
89
.
30.
Popp
MW
,
Maquat
LE
. 
Organizing principles of mammalian nonsense-mediated mRNA decay
.
Annu Rev Genet
2013
;
47
:
139
65
.
31.
He
F
,
Jacobson
A
. 
Nonsense-mediated mRNA decay: degradation of defective transcripts is only part of the story
.
Annu Rev Genet
2015
;
49
:
339
66
.
32.
Lykke-Andersen
S
,
Jensen
TH
. 
Nonsense-mediated mRNA decay: an intricate machinery that shapes transcriptomes
.
Nat Rev Mol Cell Biol
2015
;
16
:
665
77
.
33.
Pastor
F
,
Kolonias
D
,
Giangrande
PH
,
Gilboa
E
. 
Induction of tumour immunity by targeted inhibition of nonsense-mediated mRNA decay
.
Nature
2010
;
465
:
227
30
.
34.
Johnson
JL
,
Stoica
L
,
Liu
Y
,
Zhu
PJ
,
Bhattacharya
A
,
Buffington
SA
, et al
Inhibition of Upf2-dependent nonsense-mediated decay leads to behavioral and neurophysiological abnormalities by activating the immune response
.
Neuron
2019
;
104
:
665
79
.