ATM inhibition activates cGAS/STING pathway in CRC cells and has synergistic effect with radiation
In this study, we used ATM knockout CT26 cells and ATM knockdown HCT116 cells to investigate the effect of ATM gene deficiency on CRC cells. We performed RNA-seq on vector control CT26 cell line and ATM-KO CT26 cell line to investigate the transcription changes caused by ATM deficiency. Consistent with previous studies(30), ATM-deficient cells showed a significant enrichment of genes associated with immune response compared to vector control cells (Extended Data Figure. 1A-1C). In addition, our RNA-seq results showed an enrichment of genes related to the CYTOSOLIC DNA SENSING PATHWAY in ATM-deficient CRC cells (Figure. 1A), suggesting that ATM-deficiency activates cGAS/STING pathway in CRC cells. From our results, ATM-deficient CRC cells showed upregulated expression of cGAS/STING pathway related proteins and upregulated transcription of CXCL10 and CCL5, which are downstream of the cGAS/STING pathway (Figure. 1B and 1C). Furthermore, we picked KU60019, an ATM inhibitor in clinical trials, to study the impact of drug targeting ATM on CRC cells. We found that KU60019 activates cGAS /STING pathway in CRC cells dose-dependently (Figure. 1D and 1E). These results suggest that gene deletion or chemical inhibition of ATM could activate the cGAS /STING pathway in CRC cells.
To determine the feasibility of targeting ATM to improve radiosensitivity of CRC cells, we first performed a series of in vitro experiments. First, after irradiated vector control CT26 cells and ATM-KO CT26 cells with a gradient from 0Gy to 6Gy, we used colony formation assay to analyze cell proliferation after radiation (Extended Data Figure. 2A). Second, we compared the effect of ATM deficiency on apoptosis after radiation using Annexin V and PI (Extended Data Figure. 2B and 2C). Results showed that cell growth inhibition and cell death caused by radiation were more severe in ATM-deficient CRC cells. Radiation induced cell death is not only caused by direct DNA damage, but also by the production of reactive oxygen species (ROS), which amplify signals to generate irreversible DNA damage(9, 31). We further detected ROS production and the expression level of DNA damage marker phosphorylated H2AX (γH2AX) after radiation in ATM deficient cells(32, 33). As shown in (Figure. 2A and Extended Data Figure. 2D), ATM-deficient cells produced more ROS and stronger DSB after radiation. In addition, the results showed that KU60019 has a synergistic effect with radiation in inhibiting cell proliferation and promoting DNA damage in CRC cells (Extended Data Figure. 2D-2F). Taken together, these results suggest that targeted inhibition of ATM can aggravate DNA damage and cell death caused by radiation.
We know that γH2AX is an early marker of DNA damage, and cGAS is a cytosolic DNA and micronuclei sensor that activates the type I interferon pathway(11, 12, 13). We demonstrated that cGAS activation was more pronounced in ATM-deficient cells as the radiation dose increased (Figure. 2A). To assess DNA damage in detail, we examined whether cytosolic dsDNA is associated with cGAS. As shown in (Figure. 2B and 2C), ATM-deficient cells produced more micronuclei after radiation. Besides, cGAS is colocalized with micronuclei. Western blot analysis showed that the levels of phosphorylated STING and TBK1 were higher in ATM-deficient cells (Figure. 2D) with or without radiation. The qPCR results showed that CCL5, CXCL10 and IFN-β were remarkably upregulated after radiation in ATM-deficient cells (Figure. 2E), demonstrating that ATM-deficient cell activates type I interferon pathway after radiation by increasing micronuclei. Besides, KU60019 has a synergistic effect with radiotherapy in activating cGAS /STING pathway (Figure. 2F and 2G). Moreover, the Elisa results showed that the upregulation of CXCL10 and IFNβ in the cells was more evident after ATM inhibition combined with radiation (Figure. 2H, 2I). We concluded that radiation enhances cellular DNA damage and cGAS/STING pathway activation caused by ATM inhibition.
The MHC class I expression in CRC cells was upregulated by ATM inhibition and augmented by radiation
From our RNA-seq on vector control and ATM-KO CT26 cell line, we found that multiple immune factors (H2-K1, H2-D1, β2M, IRF1, NLRC5) closely related to MHC class I were upregulated by ATM knockout (Figure. 3A and 3B), which were verified by qRT-PCR. Western blot and flow cytometry showed that the total b2M protein level and the membrane MHC class I level were upregulated in ATM-deficient cells and KU60019 treated cells (Figure. 3C-3G).
At MHC class I promoters, NF-κB interacts with the NF-κB-binding sites of enhancer A, the ISRE binds to IRF family members such as IRF1, and NLRC5 forms an enhanceosome with transcription factors (such as the RFX complex) at the MHC class I promoter to induce MHC class I gene expression(34). qRT-PCR results showed that inhibition of ATM could promote IRF1 and NLRC5 transcription, and MHC class I upregulation may occur at the transcriptional level (Figure. 3H and 3I). Moreover, ATM inhibitor induced the transcription of MHC class I and NLRC5 in APCs isolated from mice but inhibited IRF1 transcription (Extended Data Figure. 3A-3C). These results suggest that the mechanism by which ATM inhibition augments the expression of MHC class I may differ between cancer cell lines and APCs, but both involve NLRC5.
To explore whether radiation could amplify the effect of ATM on MHC class I expression, we detected the expression level of MHC class I in the membrane of d-ATM and p-ATM cell lines after radiation and IFN-γ stimulation. After IFN-γ stimulation and radiation, the expression of MHC class I was upregulated in the two cell lines, especially in d-ATM tumor cell lines (Extended Data Figure. 4A-4C). The expression level of MHC class II and PD-L1 in d-ATM cell lines and p-ATM cell lines were upregulated to some extent after IFN-γ stimulation and radiation. Still, the difference was not significant between p-ATM cell lines and d-ATM cell lines (Extended Data Figure. 4D-4G). Here, we found that ATM inhibition promotes the expression of MHC class I in CRC cell lines after radiation, suggesting that the upregulation of MHC class I after DSBs may be related to ATM in CRC cell lines.
NF-κB transcriptional pathway is involved in the upregulation of MHC class I, IRF1 and NLRC5 caused by ATM inhibition
To clarify whether ATM inhibition induces MHC class I expression through increased IRF1 and NLRC5 expression, we used transient transfection with small interfering RNA (siRNA) to knockdown IRF1 and NLRC5. We found that IRF1 downregulation not only significantly reduced MHC class I upregulation caused by ATM knockout, but also reduced NLRC5 upregulation caused by ATM knockout (Figure. 4A-4C). However, NLRC5 downregulation only reduced ATM deficiency induced MHC class I upregulation, and had no significant effect on IRF1 levels (Figure. 4D-4F). These results suggest that ATM inhibition induces the expression of NLRC5 partially through the induction of IRF1. To determine whether STING plays a role in this process, we used short hairpin RNA to knockdown STING (Figure. 4I). To our surprise, we found that ATM silencing induced MHC class I upregulation did not depend on STING expression (Figure. 4G and 4H).
Canonical NF-κB pathway involves activation of the IκB kinase (IKK) complex, leading to phosphorylation of IκB proteins and subsequent ubiquitinylation and degradation by the proteasome. Non-canonical NF-κB pathway involves the NF-κB inducing kinase (NIK) that activates IKKα, thereby leading to the phosphorylation and proteasome-dependent processing of p100. We used MG132, which inhibits proteasome-mediated IκB and P100 degradation processes and NF-κB pathway activation, to determine whether NF-κB plays a role in this process. The results showed that MG132 could reverse the upregulation of MHC class I in ATM-deficient cell lines (Figure. 5A and 5B). In addition, MG132 suppressed the transcription of IRF1 and NLRC5 in ATM-deficient cell lines (Figure. 5C), indicate NF-κB pathway is involved in the upregulation of MHC class I, IRF1 and NLRC5 in ATM-deficient cell lines. We next determined whether the canonical and non-canonical NF-κB pathways are activated in basal conditions in the gene deletion or chemical inhibition of ATM. We found a significant enrichment of genes associated with the NF KAPPA B SIGNALING PATHWAY in ATM-deficient CRC cells (Figure. 5D). By isolating cytoplasmic and nucleosomal proteins, we found that basal relocation of P65 and P52 was stronger in ATM-deficient cells and ATM inhibitor treated cells (Figure. 5E). In addition, we found that inhibition of canonical and non-canonical NF-κB pathways caused by MG132 significantly reversed the up-regulation of IRF1 and NLRC5 (Figure. 5F). We further determined that the basic expression level and nuclear relocalization of P65 and P52 in ATM-deficient cells were stronger than in WT cells, and this trend could be amplified by radiation (Figure. 5G). Taken together, these results show that the inhibition of ATM results in NF-κB transcriptional pathway activation, thereby upregulating the expression of IRF1, NLRC5 and MHC class I.
ATM deficient tumors have better immunogenicity and response to radiotherapy and immunotherapy in vivo
BALB/C mice were injected with vector control CT26 cells and ATM-KO CT26 cells to record tumor growth rates after different treatments. We performed in vivo experiments, as shown in Fig. 3A, to observe the difference in efficacy of immunotherapy and radiotherapy between CT26-WT tumors and CT26-ATM−/− tumors. As can be seen from tumor growth curves we recorded, CT26-WT tumors showed a difference only in the combination group (Figure. 6B). However, for CT26-ATM−/− tumors, not only the radiotherapy group but also the anti-PD-L1 group showed significant tumor growth delays, and the combination group had the best efficacy (Figure. 6C). Furthermore, the treatment groups of CT26-ATM−/− tumors showed prolonged overall survival compared to CT26-WT tumors (Fig. 6D and 6E).
Immunohistochemistry was used to determine whether the number of CD8+ cells in the tumor center differed in each group. As shown in positive cell count analysis, CT26-ATM−/− tumors had more CD8+ cell infiltration in both untreated and treated groups compared with CT26-WT tumors (Figure. 6F and 6G).
Tumor infiltrating T cells are the main effector cells in the tumor microenvironment that kill tumor cells. Since activation of CD8 positive T cells depends on MHC class I-TCR stimulation signal, we next explored whether ATM inhibition could further activate tumor infiltrating T cells. We used flow cytometry to measure Ki67 and perforin to determine the proliferation and cytotoxic ability of tumor infiltrating T cells (Extended Data Figure. 5A). Flow cytometry results showed that anti-PD-L1 and radiotherapy both enhanced proliferation ability and cytotoxic ability of tumor infiltrating T cells in CT26-ATM−/− tumors (Figure. 7F-7K), while this effect was not observed in CT26-WT tumors (Figure. 7A-7D).
In addition to recognizing of antigen-specific signals by TCR, T cell activation also requires CD86/CD80 on APCs to provide costimulatory signals. We analyzed the CD86 expression level of CD11c+ MHC class II+ cells in tumor-draining lymph nodes to determine the activation of APCs (Extended Data Figure. 5B). As shown in (Figure. 7E, 7L and Extended Data Figure. 5C), APCs activation was higher in CT26-ATM−/− tumors than in CT26-WT tumors in all treatment groups.
These results suggest that T cells and APCs in d-ATM tumors are more functional. ATM-deficient tumors respond better to immunotherapy and radiotherapy compared to WT tumors.