Hepa 1–6 cells (a C57/L murine liver cancer cell line) and H22 cells (a BALB/c murine liver cancer cell line) were obtained from the Shanghai Institute of Biochemistry and Cell Biology. Hepa 1–6 cells were cultured in DMEM (dulbecco's modified eagle medium) with 10% FBS (fetal bovine serum). H22 cells were cultured in RPMI-1640 (Roswell Park Memorial Institute-1640) medium with 10% FBS.

Female C57BL/6 mice (4–6 weeks old) or BALB/c mice (4–6 weeks old) were provided by the Southern Medical University Animal Center. Hepa 1–6 cells (1×10⁷) were implanted subcutaneously in the right hind flank of the C57BL/6 mice. H22 cells (5×10⁶) were implanted subcutaneously in the right hind flank of the BALB/c mice. Tumor volumes were calculated using a standard formula: length×width²/2. Radiation was administered when tumors reached approximately 200 mm³. Tumor volume was measured every 5 days. For ethical considerations, mice were sacrificed when the tumor volumes reached 1000 mm³. Endpoint day was designated as the 60th day after commencement of treatment.

C57BL/6 mice (4–6 weeks old) were anesthetized by intraperitoneal injection with 2% sodium pentobarbital solution. After iodophor disinfection, liver was exposed, and 1×10⁶ Hepa 1–6 cells were slowly injected into the liver, followed by closure of the abdominal cavity. MRI was performed to visualize progression of the orthotopic HCC tumor, with treatment administered when the maximum diameter of the tumor reached 3–5 mm. After 40 days of treatment, the mice were sacrificed, and the livers were removed.

Mice received three fractions of 6 Gy on days 1, 3, and 5 from the start of treatment. Imaging and radiation were performed using a small animal radiation research platform (SARRP, Xstrahl). Mice were intragastrically administered AZD6738 (S7693, Selleck; 75 mg/kg) on days 1, 3, and 5 from the start of treatment. For the PD-L1 blockade experiment, 200 µg anti–PD-L1 (Clone:10F.9G2, Catalog #: BE0101) was administered intraperitoneally on days 1, 4, and 7 from the start of treatment. Mice were intraperitoneally injected with C-176 (S6575, Selleck; 5 mg/kg/day) daily from 1 week before radiation until the complete of radiation.

Peripheral blood samples were collected from the retro-orbital sinus on days 1, 4, and 7 after treatment initiation. Alanine aminotransferase (ALT) and serum creatinine (Scr) were detected using an animal biochemical analyzer.

Statistical comparisons between three or more groups were performed using one-way analysis of variance (ANOVA), followed by Tukey’s multiple comparison test. For mouse subcutaneous tumor volume experiments, statistical analysis was performed using mixed-effects model, followed by Tukey’s multiple comparison test. Survival curves for different groups of mice were generated using the Kaplan-Meier method. Survival data were compared using the log-rank Mantel-Cox test. P<0.05 was considered statistically significant. All graphs were plotted using GraphPad Prism V.7.

A previous report showed that inhibition of the DDR pathway reverts radiation-induced PD-L1 upregulation, suggesting that a DDR inhibitor might attenuate a radiation-induced immune microenvironment. We first tested the effect of AZD6738 on the expression level of PD-L1 in HCC cells treated with AZD6738, radiation, or the combination. We found that at 24 and 48 hours, radiation alone significantly upregulated PD-L1 expression (fold-change in PD-L1 median fluorescence intensity±SD, radiation vs vehicle: 24 hours, 2.02±0.15-fold, p<0.0001; 48 hours, 1.58±0.20-fold, p<0.0001). And the addition of AZD6738 to radiation decreased the radiation-induced PD-L1 upregulation (radiation vs AZD6738 plus radiation: 24 hours, 2.02±0.15-fold vs 1.05±0.02-fold, p<0.0001; 48 hours, 1.58±0.20-fold vs 0.92±0.07-fold, p<0.0001). At 72 hours, no difference was observed between radiation and AZD6738 with radiation (online supplementary figure 2).

Next, we explored the role of AZD6738 on immune microenvironment in vivo. Hepa 1–6 cells were implanted on the flanks of C57BL/6 mice. When the subcutaneous tumors grew to 200 mm³, we treated the mice with AZD6738, radiation, or the combination. At day 8, radiation alone increased the percentage of tumor-infiltrating lymphocyte (TIL) Tregs (0.099%±0.036% radiation vs 0.041%±0.024% vehicle, p=0.0188). However, AZD6738 plus radiation reduced the percentage of Tregs compared with radiation alone (0.033%±0.015% AZD6738 plus radiation vs radiation, p=0.0072). No differences were observed on day 14 (figure 1A). Moreover, addition of AZD6738 to radiation significantly increased the proportion of CD8⁺ T cells compared with vehicle on day 14 (65.7%±7.6% AZD6738 plus radiation vs 40.7%± 2.5% vehicle, p＜0.0001; figure 1B). At days 8 and 14, no differences in the proportion of CD8⁺ T cells were observed between vehicle and radiation.

Next, we determined the activation status of tumor-infiltrating CD8⁺ T cells. We investigated expression levels of PD-1 and Tim3 in CD8⁺ T cells. At day 8, we found that administration of AZD6738 plus radiation significantly downregulated PD-1 expression (fold-change in median fluorescence intensity±SD, AZD6738 plus radiation vs vehicle: 0.53±0.23-fold; p=0.0381) but showed no effect on Tim3 expression. At day 14, radiation treatment upregulated Tim3 expression (fold-change in median fluorescence intensity±SD, radiation vs vehicle: 1.43±0.37-fold; p=0.0112), and administration of AZD6738 plus radiation significantly downregulated Tim3 expression (AZD6738 plus radiation: 0.65±0.39-fold vs radiation; p=0.0413), whereas no differences were observed in PD-1 expression (figure 1C). At day 8, AZD6738 with radiation reduced the percentage of TIL CD8⁺PD-1⁺ T cells in comparison with radiation alone (11.7%±1.2% AZD6738 plus radiation vs 25.7%±3.2% radiation alone, p=0.0227). No differences were observed at day 14 (figure 1D).

At day 8, AZD6738 plus radiotherapy increased the percentage of TIL CD8⁺ T cells that produced IFN-γ compared with radiation alone (13.9%±3.7% AZD6738 with radiation vs 2.6%±1.3% radiation, p=0.0151). At day 14, compared with vehicle, both radiation alone and AZD6738 plus radiation increased the percentage of TIL CD8⁺ T cells that produced IFN-γ (figure 1E). In summary, AZD6738 potentiated the effect of radiation on reducing the number of exhausted T cells and increasing the cell-killing activity of CD8⁺ T cells, meanwhile, reverted radiation-induced Tregs infiltration.

Because AZD6738 improved the radiation-induced immunosuppressive microenvironment, it is possible that its use could also enhance the antitumor effect of radiotherapy combined with ICIs. To determine the treatment efficacy of AZD6738, the Hepa 1–6 and H22 tumor-bearing mice were treated with varied treatment regimens (figure 2A). In Hepa 1–6 tumor-bearing mice models, for vehicle and AZD6738 alone group, the designated tumor volume endpoint (1000 mm³) was reached at days 15 and 20, respectively. Both radiation alone and anti-PD-L1 alone groups reached the endpoint on day 25. AZD6738 plus radiation group reached the endpoint at day 30. Until day 35, anti-PD-L1 plus radiation (radioimmunotherapy) group and radiotherapy combined with anti-PD-L1 and AZD6738 (triple therapy) group still did not reach the endpoint yet. Furthermore, compared with radioimmunotherapy, triple therapy significantly inhibited tumor growth (mean tumor volume±SEM: 81±59 mm³ triple therapy vs 369±147 mm³ radioimmunotherapy, p=0.0172; figure 2B,C). Similarly, in H22-tumor-bearing mouse models, triple therapy significantly inhibited tumor growth relative to radioimmunotherapy (mean tumor volume±SEM: 438±61 mm³ triple therapy vs 771±72 mm³ radioimmunotherapy; p=0.0301) (online supplementary figure 3). In Hepa 1–6 tumor-bearing mice models, the average survival time of vehicle group was 15 days, and these of the AZD6738 alone, anti-PD-L1 alone, and radiation alone groups were 25, 32, and 42 days. Compared with radiation alone, the addition of AZD6738 or anti-PD-L1 to radiation did not further increase the mice survival. Meanwhile, triple therapy significantly prolonged survival compared with radiation alone or radioimmunotherapy. In addition, triple therapy showed higher complete response rate than radioimmunotherapy (83% triple therapy vs 58% radioimmunotherapy; figure 2D). Therefore, the addition of AZD6738 to radioimmunotherapy significantly inhibited tumor growth, increased complete response rate, and prolonged the survival time of mice.

Importantly, to better replicate a clinical scenario, we treated orthotopic HCC-bearing mice with various regimens, subsequently observing that triple therapy exhibited better antitumor efficacy than radioimmunotherapy (mean tumor volume±SEM: 16.7±9.4 mm³ triple therapy vs 32.9±17.0 mm³ radioimmunotherapy, p=0.0421; figure 2E,F).

More importantly, the addition of AZD6738 was well tolerated without abnormal levels of ALT and Scr in mice blood samples. These data suggested that AZD6738 increased the antitumor effect of radioimmunotherapy without impairing liver and kidney functions in mice (online supplementary figure 4).

As the antitumor effect of the triple therapy is better than that of radioimmunotherapy, we need to identify the effect of AZD6738 on the alterations in tumor immune microenvironment. Lymphocytes from the tumors and spleen were collected at days 8 and 14 (figure 3A). At day 14, both radioimmunotherapy and triple therapy significantly increased the number of TIL CD8⁺ T cells compared with radiation alone (30.2±15.2 radiation alone vs 238.8±124.3 radioimmunotherapy, p=0.0060; vs 350.6±90.8 triple therapy, p<0.0001; figure 3B). AZD6738 increased the numbers of TIL CD8⁺ T cells in triple therapy compared with that of in radioimmunotherapy, but the difference did not reach statistical significance. At day 14, anti-PD-L1 with radiation significantly increased the ratio of CD8⁺/CD3⁺ T cells compared with radiation alone (50.0%±3.0% radiation alone vs 64.2%±7.3% anti-PD-L1 with radiation, p=0.0239; figure 3C). Triple therapy further increased the ratio of CD8⁺/CD3⁺ T cells compared with anti-PD-L1 with radiation at days 8 and 14 (day 8, 61.2%±3.0% triple therapy vs 50.0%±5.5% radioimmunotherapy, p=0.0384; day 14, 82.0%±8.4% triple therapy vs 64.2%±7.3% radioimmunotherapy, p=0.0046; figure 3C).

Meanwhile, we examined the effect of AZD6738 on the number of TIL Tregs. On day 8, AZD6738 plus radiation, radioimmunotherapy, and triple therapy significantly inhibited the radiation-induced TIL Tregs infiltration (4.4±2.1 AZD6738 plus radiation vs 8.9±2.7 radiation, p=0.0111; 2.2±1.4 radioimmunotherapy vs radiation, p=0.0003; 1.1±1.3 triple therapy vs radiation, p<0.0001; figure 3D). At days 8 and 14, compared with AZD6738 plus radiation and anti-PD-L1 plus radiation, triple therapy further reduced the number of TIL Tregs, but the difference did not reach statistical significance (figure 3D). Although no difference in the number of TIL Tregs was observed between triple therapy and radioimmunotherapy, triple therapy significantly increased the CD8⁺/Treg ratio at day 14 compared with radioimmunotherapy (191.5±51.8 triple therapy vs 95.7±69.6 radioimmunotherapy, p=0.0222; figure 3E). From day 8 to day 14, the CD8⁺/Treg ratio in the spleen increased in triple therapy. No differences in CD8⁺/Treg ratio among different groups were observed on day 8. At day 14, only triple therapy significantly upregulated CD8⁺/Treg ratio than that of radiation alone group in spleen (264.1±223.2 triple therapy vs 10.2±5.8 radiation, p=0.0286; figure 3F).

Next, we examined the proliferating (Ki67⁺) splenic and TIL T cell population from varied treatments. At day 8, compared with radiation alone, only triple therapy significantly increased the percentage of TIL CD8⁺ Ki67⁺ T cells (66.6%±10.5% triple therapy vs 47.5%±6.3% radiation p<0.0001; vs 47.5%±6.3% AZD6738 plus radiation, p=0.0136; vs 37.3%±9.3% radioimmunotherapy, p=0.0003; figure 4A). At day 14, the addition of AZD6738 to radioimmunotherapy did not impact the percentage of TIL CD8⁺ Ki67⁺T cells compared with that of radioimmunotherapy group (figure 4C). Similar phenomena were seen in spleens. Only triple therapy significantly increased the percentage of splenic CD8⁺ Ki67⁺ T cell at day 8 compared with radiation alone (60.0%±22.2% triple therapy vs 20.7%±18.3% vehicle, p=0.0253). No differences were observed at day 14 (figure 4B and E).

With respect to the percentage of TIL effector (Eff) T cells, at day 8, radioimmunotherapy and triple therapy significantly increased the percentage of TIL CD4⁺Ki67⁺ Eff T cells compared with radiation alone (20.9%±6.5% radiation vs 46.6%±9.1% radioimmunotherapy, p=0.0006; vs 42.7%±10.9% triple therapy, p<0.0001; figure 4D). Triple therapy further increased this percentage compared with that of radioimmunotherapy group (46.6%±9.1% radioimmunotherapy vs 42.7%±10.9% triple therapy, p=0.0282; figure 4D). No differences were observed between radioimmunotherapy and triple therapy at day 14 (figure 4D). Furthermore, at day 8, no differences in the percentage of splenic Ki67⁺CD4⁺ Eff T cells were observed between radioimmunotherapy and triple therapy (figure 4F). At day 14, the percentage of CD4⁺Ki67⁺ Eff T cells in triple therapy group increased compared with that of radioimmunotherapy, but the difference did not reach statistical significance (figure 4F).

Coexpression of T cell immune exhaustion markers, including PD-1, Tim-3, and LAG-3, can inhibit T cell antitumor function.18–21 Therefore, we evaluated the expression of T cell surface immune exhaustion markers in all treatment regimens. At day 8, compared with radiation alone, radioimmunotherapy did not impact the percentage of TIL CD8⁺PD-1⁺LAG3⁺ T cells, whereas triple therapy significantly reduced the percentage of these immunosuppressive T cells (2.5%±1.6% triple therapy vs 23.4%±17.6% radiation, p=0.0122; figure 5A and C). At day 14, the percentage of TIL CD8⁺PD-1⁺LAG3⁺ T cells was lower in triple therapy than that in AZD6738 plus radiation group or radioimmunotherapy group, although statistical differences were not achieved (figure 5C).

At day 8, we observed no differences in PD-1 or Tim3 expression. At day 14, we found that triple therapy significantly downregulated Tim3 expression (fold-change in median fluorescence intensity±SD, triple therapy vs radiation: 0.43±0.18-fold; p=0.103), whereas no differences were observed in PD-1 expression (online supplementary figure 5A,B. Triple therapy maintained a lower percentage of TIL CD8⁺PD-1⁺Tim3⁺T cells compared with AZD6738 plus radiation and radioimmunotherapy, but the difference did not reach statistical significance (figure 5B,D). There were no differences in the percentage of CD8⁺PD-1⁺LAG3⁺ and CD8⁺PD-1⁺Tim3⁺ T cells between triple therapy and radioimmunotherapy in the spleens (online supplementary figure 5C and D). In summary, compared with radioimmunotherapy, triple therapy can further reduce the coexpression of CD8⁺ T cell exhaustion markers in mice xenograft tumors and spleens.

To confirm whether the addition of AZD6738 to radioimmunotherapy can enhance effector function after reducing coexpression exhaustion markers of CD8⁺ T cells, we used phorbol 12-myristate 13-acetate (PMA) and ionomycin to stimulate IFN-γ production of CD8⁺ T cells. At day 8, adding either AZD6738 or anti-PD-L1 to radiation significantly increased the percentage of TIL CD8⁺ T cells that produced IFN-γ, and triple therapy further increases this percentage (20.5%±3.5% triple therapy vs 13.6%±2.6% AZD6738 plus radiation, p=0.0024; vs 15.6%±2.1% radioimmunotherapy, p=0.0317; figure 5E,G). At day 14, the percentage of TIL CD8⁺ T cells that produced IFN-γ remained the same as that of day 8 in mice treated with radiation alone, AZD6738 with radiation, and radioimmunotherapy; however, that of triple therapy was further increased, and it was significantly higher than that of the other treatment groups (50.3%±24.3% triple therapy vs 6.4%±1.8% radiation, p=0.0004; vs 12.8%±5.0% AZD6738 with radiation, p=0.0019 vs 16.3%±8.7% radioimmunotherapy, p=0.0044; figure 5G).

With respect to the percentage of CD8⁺ T cells that produced IFN-γ in mice spleens, no differences were observed among all treatments at day 8 (figure 5F,H). At day 14, the percentage in triple therapy was significantly higher than that in the other treatment groups (20.8%±10.2% triple therapy vs 9.2%±4.4% radiation, p=0.0308 vs 9.7%±2.9% AZD6738 plus radiation, p=0.0392; vs 9.6%±2.7% radioimmunotherapy, p=0.0369; figure 5H).

We additionally examined the percentage of TIL CD4⁺ T cells that produced IFN-γ. No differences were observed between radioimmunotherapy and triple therapy at day 8 (online supplementary figure 6A). At day 14, the percentage of TIL CD4⁺ T cells that produced IFN-γ in triple therapy was significantly higher than that in each treatment groups (51.0%±22.9% triple therapy vs 10.0%±2.2% radiation, p=0.0003; vs 16.0%±4.8% AZD6738 plus radiation, p=0.0012; vs 13.9%±0.9% radioimmunotherapy, p=0.0007; online supplementary figure 6A). Meanwhile, no differences in the percentage of splenic CD4⁺ T cells that produced IFN-γ were observed between radioimmunotherapy and triple therapy at any time points (online supplementary figure 6B). In summary, the addition of AZD6738 to radioimmunotherapy significantly increased the percentage of CD8⁺ and CD4⁺ T cells that produced IFN-γ in tumors and the spleen.

We then explored the mechanism underlying the synergistic effect of AZD6738 and radioimmunotherapy. Previous studies showed that radiation-induced DNA damage activates cGAS/STING signaling; therefore, we hypothesized that AZD6738, as a DDR inhibitor, might play a role in enhancing cGAS/STING activation. Analysis of levels of several key factors involved in cGAS/STING signaling in Hepa 1–6 subcutaneous tumors from mice treated with radioimmunotherapy or triple therapy revealed that AZD6738 plus radioimmunotherapy treatment increased levels of cGAS, p-STING, and p-TBK1, suggesting that AZD6738 increased activation of the cGAS/STING pathway (figure 6A). Furthermore, treatment of Hepa 1–6 tumor-bearing mice with the STING inhibitor C-176 impaired the antitumor efficacy of triple therapy (482.4±72.9 (triple therapy plus STING inhibitor) vs 80.5±31.4 (triple therapy), p<0.0001; vs 265.2±27.4 (radioimmunotherapy), p=0.0003) (figure 6B). These data indicate that AZD6738 exerted a synergistic antitumor effect with radioimmunotherapy by activating cGAS/STING signaling.

We analyzed the memory status of CD8⁺ and CD4⁺ Eff T cells in mice tumors and spleens. At day 8, compared with radioimmunotherapy, triple therapy increased the percentage of TIL CD8⁺ central memory T (T_CM) cells (25.5%±3.9% triple therapy vs 17.4%±5.6% radioimmunotherapy, p=0.0287; figure 7A,B). In addition, triple therapy reduced the percentage of TIL CD8⁺ effector memory T (T_EM) cells (63.5%±4.8% triple therapy vs 75.4%±6.9% radioimmunotherapy, p=0.0433). At day 14, no differences in the percentage of TIL CD8⁺ T_CM were observed between radioimmunotherapy group and triple therapy group (figure 7B). The percentage of TIL CD8⁺ T_EM cells in triple therapy was significantly elevated compared with that in radioimmunotherapy at day 14 (83.7%±3.8% triple therapy vs 72.4%±6.6% radioimmunotherapy, p=0.0169; figure 7C). No significant differences in activation or memory status of TIL CD4⁺ T cells were observed between radioimmunotherapy group and triple therapy group at any time points (online supplementary figure 7).

At day 14, we found there was no differences in the percentage of splenic CD8⁺ T_CM cells between radioimmunotherapy group and triple therapy group (online supplementary figure 8A). Meanwhile, the percentage of splenic CD8⁺ T_EM cells was higher in triple therapy group compared with that in radioimmunotherapy group (43.6%±16.2% triple therapy vs 22.0%±2.7% radioimmunotherapy, p=0.0065; online supplementary figure 8B). Similar effects were seen in the percentage of splenic CD4⁺ T_CM. The percentage of splenic CD4⁺ T_CM cells in triple therapy increased over time, and it was significantly higher in triple therapy than that in the other treatments at day 14 (online supplementary figure 8C). With respect to the percentage of splenic CD4⁺ T_EM cells, it was significantly higher in triple therapy group than that in radioimmunotherapy group at day 8 (55.7%±7.3% triple therapy vs 20.7%±2.7% radioimmunotherapy, p<0.0001; online supplementary figure 8D). Meanwhile, no differences in the percentage of splenic CD4⁺ T_EM were observed between radioimmunotherapy and triple therapy at day 14 (online supplementary figure 8D).

As the activation and memory status of CD8⁺ and CD4⁺ Eff T cells is closely related to tumor control, next we determined the effect of AZD6738 on tumor recurrence. We re-challenged the treatment completed tumor-bearing mice with Hepa 1–6 cells in the contralateral flank on the 20th day, and then monitored the tumor volumes of these mice treated with different regimens. We found that, compared with radioimmunotherapy, triple therapy showed a more potent inhibitory effect on the secondary tumor growth (figure 7D). These findings indicated that triple therapy enhanced immunologic memory and prevented recurrence of HCC by boosting T_CM and T_EM cells in mice tumors and spleens.