CTLA-4 depletion induces senescence in melanoma cells To investigate the role of CTLA-4 in cancer, we silenced CTLA-4 either alone or in combination with doxorubicin (Dox) (30, 31) (Fig. 1a). Human melanoma cell line A375 showed higher expression of CTLA-4 than other cells (Supplementary Fig. 1a-d)(17, 32). Unexpectedly, CTLA-4 silencing resulted in senescence phenotype, including an increase in cell size (Fig. 1a, b), SA-β-Gal activity proven by senescent green probe (Fig. 1b, c), and decreased cell viability (Fig. 1d) and colony forming assays (Fig. 1e). In addition, western blot (WB) analysis showed increased expression of senescence markers such as p21 and p16, and the heterochromatin marker, H3K9me3, compared to the control cells (Fig. 1f). Moreover, confocal imaging showed that p21 expression inversely correlated with that of CTLA-4 (Fig. 1g). These effects were synergistic in both the siCTLA-4 and Dox-treated group compared to the Dox(only)-treated group. Additionally, results of fractionation assay confirmed that CTLA-4 was predominantly expressed in the cytosol (Supplementary Fig. 1e, f).
We confirmed this with B16-F10 mouse melanoma cells which expressing CTLA-4 well (Supplementary Fig. 1b). We silenced CTLA-4 alone or in combination with Dox in B16-F10 cells. Depletion of CTLA-4 alone resulted in increased cell size (Fig. 2a, b), SA-β-Gal activity (Fig. 2b), and decreased cell viability (Fig. 2c). Furthermore, p21, p16, and H3K9me3 protein levels increased, as shown by WB (Fig. 2d). Moreover, p21 expression was inversely correlated with that of CTLA-4 (Fig. 2e, f). Notably, all these effects were more significant in the siCTLA-4 and Dox combination group than in the Dox(only)-treated group. We also repeated these experiments using a different CTLA-4 specific siRNA (siCTLA-4*) (Supplementary Fig. 2) and the results were similar to those obtained from the siRNA sequence used in A375 cells (Fig. 1f). Taken together, our data suggest that targeting CTLA-4 in cancer cells induces senescence and halts cancer cell proliferation, and Dox treatment enhances these outcomes synergistically.
AKT pathway is required to induce CTLA-4-deficiency-induced-senescence As a well-known component of CTLA-4 related signaling, the AKT pathway is activated through the surface CTLA-4 signaling pathway(17, 33). Therefore, we investigated the status of the AKT pathway and its role in CTLA-4 deficiency-induced senescence. Ironically, phospho(p)-AKT was also upregulated along with other senescence markers, including p53, p21, p27, p16, and H3K9me3, in both CTLA-4 alone silenced and Dox combination groups (Fig. 3a). AKT pathway-mediated senescence is known as oncogene-induced senescence (OIS)(34, 35). In addition, confocal imaging showed increased expression of p-AKT and p-mTOR, molecules acting downstream of AKT, in siCTLA-4 groups (Fig. 3b, c). These experiments were performed in mouse melanoma cells and similar results were obtained (Fig. 3d-f). A recent study reported an 80 kDa splicing mTOR isoform, mTORβ, which is considered an active protein kinase beyond full-length mTOR (mTORα)(36). It showed an expression pattern similar to that of p-AKT in both A375 and B16-F10 cells (Supplementary Fig. 3a, b). However, CTLA-4 overexpression blocked Dox treatment -induced senescence phenotype, including morphological changes and p21, p-AKT, and p-mTOR expression in B16-F10 as well as A375 cells (Supplementary Fig. 4a-d). Taken together, we conclude that the AKT pathway is required for CTLA-4-deficiency-induced senescence.
CTLA-4 deficiency causes DNA damage response DNA damage is one of the dominant causes or consequences of senescence(2, 11, 37). So, we examined the DNA damage marker γ-H2AX, and found it was upregulated and inverse correlation with CTLA-4 expression in CTLA-4-silenced B16-F10 cells (Fig. 4a). In addition, other DNA damage markers H3K9me3 and p53 (Fig. 1f, 2d, 3a, and 3d) were elevated along with other senescence markers. γ-H2AX expression was validated by confocal imaging in B16-F10 and A375 cells (Fig. 4b-d). These results were confirmed in CTLA-4 knockout (KO) B16-F10 cells (Supplementary Fig. 5a). CTLA-4 KO cells showed higher sensitivity to the anticancer drugs cisplatin and Dox (Supplementary Fig. 5b, c).
Micronuclei are tiny nuclear DNA, and often observed in cancer and senescence. Senescence often appears with aneuploidy induced by chromosome missegregation, which triggers the formation of cytoplasmic chromatin fragments (CCFs), resulting in and they become micronuclei(38). We observed micronuclei following CTLA-4 silencing, which is colocalized with γ-H2AX (Fig. 4d: enlarged part of Fig. 4c) and their content was much higher in the Dox-co treated group in mouse (Fig. 4e, f) as well as human melanoma cells (Fig. 4g, h). Interestingly, micronuclei co-localized with γ-H2AX in the CTLA-4 silenced cells, confirming that micronuclei are a byproduct of DNA damage caused by genome instability (Fig. 4d). In addition, Aurora B, which prevents micronuclei formation and whose downregulation induces senescence(39, 40), was downregulated by siCTLA-4 treatment (Fig. 4i-l, Supplementary Fig. 5d ) and this effect was reverted by CTLA-4 overexpression (Fig. 4m). These results imply that CTLA-4 depletion induced genomic instability and DNA damage response.
STING signaling regulates CTLA-4-depletion-induced senescence via modulating AKT signaling pathway Micronuclei present in the cytosol, trigger cGAS-STING pathway(41). cGAS binds to dsDNA in micronuclei and triggers cyclic GMP-AMP (cGAMP) synthesis. Subsequently, cGAMP attaches to STING and recruits TBK1, which in turn phosphorylates STING. Phosphorylated STING recruits interferon regulatory factor-3 (IRF3), which is then phosphorylated by TBK1 and dimerizes, followed by its nuclear translocation. Eventually, IRF3 transcribes genes encoding proteins including interferons and cytokines in the nucleus(42).
Therefore, we examined the effects of micronuclei on the STING pathway in the CTLA-4 silenced group. p-STING and p-IRF3 increased in a time-dependent manner, along with p-AKT and p-p53 (S15) (Fig. 5a). Furthermore, γ-H2AX expression correlated with CTLA-4 depletion time points in both WB and confocal imaging assay (Fig. 5a, b). However, these effects were reversed by siSTING treatment. Namely, activated AKT signaling as well as the elevation of γ-H2AX by CTLA-4 silencing, were blocked by siSTING treatment (Fig. 5c). Notably, p-AKT expression was also observed in the confocal imaging (Fig. 5d) and these findings were confirmed in B16-F10 cells by WB and confocal imaging (Supplementary Fig. 6a-d). Overall, our results indicated that CTLA-4 deficiency potentiates DNA damage-induced STING signaling mediated via AKT signaling.
DNA-PKcs intervenes with CTLA-4-depletion-induced senescence DNA-PKcs belongs to the phosphatidylinositol 3-kinase family. DNA-PKcs forms the active DNA-PK holoenzyme with the Ku80/Ku70 heterodimer to regulate DDR following DSB. DNA-PKcs rapidly moves to the damaged sites and is activated. It then sends damage signals via p53, ultimately leading to cell cycle arrest, aging, or apoptosis. Aside from its role in the cell cycle(43), DNA-PKcs keeps genome integrity(44). Furthermore, DNA-PKcs recognizes cytosolic DNA and activates the cGAS-STING pathway (29).
Next, we examined whether DNA-PKcs plays a role in CTLA-4-depletion-induced senescence. DNA-PKcs was upregulated together with activated STING pathway (p-STING and p-IRF3), AKT signaling markers (p-AKT and p-mTOR), and p-p53 (S15) in the CTLA-4-depleted and Dox-combined group of B16-F10 cells compared to the control and Dox-(only)-treated group (Fig. 6a, b). Additionally, DNA-PKcs was activated (p-DNA-PKcs) by siCTLA-4 in A375 cells (Supplementary Fig. 7a, b).
Surprisingly, all activated signals, including those of the STING-AKT pathway (p-STING and p-mTOR), CDK inhibitor (p21), and DNA damage marker (γ-H2AX) following siCTLA-4 treatment of B16-F10 cells were abolished by siDNA-PKcs treatment (Fig. 6c), and these results were confirmed in A375 cells (Fig. 6d-g). DNA-PKcs expression was elevated together with those of p-IRF3, p-STING, and p-AKT in both CTLA-4 silenced and Dox-cotreated with CTLA-4 silenced groups compared to control and only Dox-treated groups, respectively (Fig. 6d). However, the STING pathway (p-STING and p-TBK1), AKT signaling factors (p-AKT and p-mTOR), cell cycle inhibitors (p21 and p27), and the DNA damage marker γ-H2AX activated by siCTLA-4 were blocked by siDNA-PKcs treatment (Fig. 6f) or the specific DNA-PKcs inhibitor Nu7441 (Fig. 6g). Interestingly, targeting only DNA-PKcs via silencing or inhibition showed a different manner compared to that observed following co-treatment with the CTLA-4 silencing group (Fig. 6c, f, and g). Namely, targeting DNA-PKcs alone elevated AKT and p21 levels as well as STING pathway (2nd lane of 6C, 3rd lane of 6F, and 2nd -3rd lane of 6G), which may be attributed to the lack of DNA repair following DNA-PKcs silencing or inhibition(45). DNA-PKcs abrogation downregulates the protein expression induced by CTLA-4 silencing, which may be due to the absence of a DNA-PKcs signaling in the STING pathway (46). Indeed, siDNA-PKcs almost abrogated CTLA-4 depletion-led senescence in A375 cells, evidenced by morphological changes (Fig. 6h) and the degree of senescence (Fig. 6i, j), indicating that DNA-PKcs is essential for CTLA-4-depletion induced senescence.
To confirm the selectivity of DNA-PKcs for CTLA-4 depletion-induced senescence, we compared DNA-PKcs with ataxia telangiectasia mutated (ATM), which has also been implicated in DSBs(47). The ATM-specific inhibitor KU-60019 showed no influence CTLA-4 depletion effect; however, DNA-PKcs inhibition by NU7441 or siRNA knockdown abrogated the CTLA-4 depletion effects (Fig. 6g, Supplementary Fig. 7a, b). These results suggest that the effects of CTLA-4 are selectively linked with DNA-PKcs. Furthermore, immunoprecipitation (IP) assay revealed an interaction between CTLA-4 and DNA-PKcs after CTLA-4-overexpression in A375 cells (Fig. 6k). In fact, CTLA-4 was predominantly located in the cytosol (Fig. 1g, 2e, Supplementary Fig. 1e, f), along with DNA-PKcs (Fig. 6b and e), which can facilitate their interaction to detect cytosolic nucleic acids (Fig. 6k). To further investigate the correlation of CTLA-4 with DNA-PKcs, we used The Cancer Genome Atlas (TCGA) data, and observed that the mRNA expression of PRKDC, which encodes DNA-PKcs, was negatively correlated with that of CTLA-4 (Fig. S7c). Furthermore, patients were categorized into high- (top 1/3) and low- (bottom 1/3) CTLA-4 groups based on the CTLA-4 expression level and showed an inverse correlation with non-homologous end joining pathway components (NHEJ), including DNA-PKcs, in patients with skin cutaneous melanoma (Supplementary Fig. 7d). Additionally, these features were observed in lung, cervical, head, and neck squamous cell carcinomas (Supplementary Fig. 7e). Taken together, DNA-PKcs plays an indispensable role and orchestrates CTLA-4-depletion-induced senescence of cancer cells via modulating the STING-AKT pathway axis.
CTLA-4-depletion impedes tumor growth via senescence induction Finally, we performed experiments in mice to verify whether CTLA-4-depletion affected tumor growth. We generated a CTLA-4 knockout (KO) B16-F10 cell line using the CRISPR-Cas9 system. Next, C57BL/6 mice were subcutaneously injected with 5x105 B16-F10WT and B16-F10CTLA − 4 KO cells followed by intraperitoneal injections of 9 mg/kg Dox. The tumors were collected on day 16 (Fig. 7a).
Size, weight, and volume of tumors derived from mice injected with CTLA-4 KO cells, were much smaller than those of tumors derived from mice injected with wild-type (WT) cells in both DMSO- and Dox-treated groups (Fig. 7b-d). Furthermore, SA-β-Gal staining of B16-F10CTLA − 4 KO derived tumors was much stronger than that of B16-F10WT derived tumors both with and without Dox treatment, confirming the effect of CTLA-4 on tumor cell senescence (Fig. 7e). B16-F10CTLA − 4 KO derived tumors showed higher expression of γ-H2AX and p-STING, than that in B16-F10WT derived tumors in both the absence and presence of Dox treatment groups by WB analysis (Fig. 7f), which was later confirmed by IHC analysis (Fig. 7g-i). In conclusion, once CTLA-4 depletion occurs in cancers, it causes cellular senescence via DNA damage-DNA-PKcs-STING-AKT-p21 pathway, and eventually leads to tumor suppression.