Increasing evidence indicates that local hypofractionated radiotherapy (LRT) can elicit both immunogenic and immunosuppressive local and systemic immune responses. We thus hypothesized that blockade of LRT-induced immunosuppressive responses could augment the antitumor effects and induce an abscopal response. In this study, we found that the upregulation of Foxp3+ regulatory T cells (Tregs) in the mesothelioma tumor microenvironment after nonablative oligofractionated irradiation significantly limited the success of irradiation. Using DEREG mice, which allow conditional and efficient depletion of Foxp3+ Tregs by diphtheria toxin injection, we observed that transient Foxp3+ Treg depletion immediately after nonablative oligofractionated irradiation provided synergistic local control and biased the T cell repertoire toward central and effector memory T cells, resulting in long-term cure. Furthermore, this combination therapy showed significant abscopal effect on the nonirradiated tumors in a concomitant model of mesothelioma through systemic activation of cytotoxic T cells and enhanced production of IFN-γ and granzyme B. Although local control was preserved with one fraction of nonablative irradiation, three fractions were required to generate the abscopal effect. PD-1 and CTLA-4 were upregulated on tumor-infiltrating CD4+ and CD8+ T cells in irradiated and nonirradiated tumors, suggesting that immune checkpoint inhibitors could be beneficial after LRT and Foxp3+ Treg depletion. Our findings are applicable to the strategy of immuno-radiotherapy for generating optimal antitumor immune responses in the clinical setting. Targeting Tregs immediately after a short course of irradiation could have a major impact on the local response to irradiation and its abscopal effect.
Malignant pleural mesothelioma (MPM) is an aggressive malignancy with dismal prognosis and a median survival of <20 mo despite aggressive treatment with chemotherapy, surgery, and radiation (1–3). Immunotherapy provides new potential opportunities, but the results with checkpoint inhibitors have had limited success so far (4, 5). Preclinical models suggest that combining immunotherapy with conventional therapy provides the best approach (6). Hence, better understanding of the mesothelioma tumor immune microenvironment and the dynamics of the immune response to the conventional therapies is critical to developing new effective combination treatment strategies and defining potential biomarkers of response.
We developed a new and unique, to our knowledge, multimodality approach for MPM using a short course of oligofractionated hemithoracic irradiation targeting the whole tumor volume before surgical resection (7, 8). The results of this surgery for mesothelioma after radiation therapy (SMART) approach have been encouraging, with an overall median survival of 28.3 mo as an intention-to-treat analysis (3). An analysis of the tumor microenvironment in our initial cohort of 69 consecutive patients undergoing the SMART approach showed that an increased infiltration of CD8+ T cells at the time of surgery after irradiation was an independent predictor of improved survival along with epithelioid histology and the absence of metastatic nodal disease (9). Our previous mouse study demonstrated that nonablative oligofractionated irradiation induces a specific activation of the immune system against mesothelioma with the development of an in situ vaccination, which is maintained through memory T cells directed against the tumor (10). The SMART approach may thus provide an ideal platform to introduce immunotherapy as part of the multimodality therapy (5, 9).
Oligofractionated irradiation is a short course (2–5 fractions) of hypofractionated radiation (characterized by dose of 3 Gy or more per fraction). Similar doses of radiation are used in preclinical mice models and in the clinical setting, as the absorb dose of radiation is the same in mice and humans. Recent evidence has shown that oligofractionated irradiation not only has a direct cytotoxic effect but also activates a specific immune response against the tumor and contributes to the benefit of irradiation therapy (10–12). Irradiation promotes the release of danger signals, tumor-associated Ags, and cytokines recruiting inflammatory cells and APC into the tumor microenvironment and the draining lymph nodes that activate cytotoxic T cell function. Irradiation can also attract immunosuppressive cells into the tumor microenvironment (11). Activation of immunosuppressive responses after radiotherapy promotes radioresistance and tumor recurrence (13). However, responses from the immune system to irradiated tumor microenvironment are complex. The impact of different radiation fraction sizes and patterns on the tumor immune microenvironment as well as optimal immunotherapeutic targets in mesothelioma remain largely unknown.
In this study, we investigated the kinetics of immune cells in the tumor microenvironment after radiotherapy using murine mesothelioma model. The AB12 mesothelioma cell line was chosen, as this model is well established in our laboratories and others, originates from asbestos exposure, and presents histological, immunological, and genetic alterations that are similar to human mesothelioma (14). We demonstrated that the rapid accumulation of regulatory T cells (Tregs) in the tumor microenvironment after irradiation had an important role in the failure of nonablative oligofractionated irradiation. To test the therapeutic potential of Tregs blockade, we used transgenic DEREG mice, which harbor a diphtheria toxin (DT) receptor under the control of the Foxp3 promoter and enable selective depletion of Tregs by DT injection (15). We observed synergistic antitumor effects of combination therapy with nonablative oligofractionated irradiation, followed by transient Treg depletion in both irradiated and nonirradiated tumor sites through systemic activation of T cells and enhanced production of IFN-γ and granzyme B. Although one fraction of radiation was sufficient to generate a synergistic local response, three fractions were required to achieve an abscopal effect on the nonirradiated tumor sites. These findings are applicable to the strategy of immuno-radiotherapy for the generation of optimal antitumor immune responses.
Materials and Methods
Mice and tumor cell line
BALB/c mice at 6–8 wk of age and DEREG [C.B6-Tg (Foxp3-DTR/EGFP) 23.2Spar/Mmjax] mice (15) were purchased from The Jackson Laboratory (Bar Harbor, ME). DEREG mice were bred in a facility at the University Health Network, University of Toronto. Animal care and experiments were performed in accordance with institutional and Canadian Institute of Health guidelines.
AB12, murine mesothelioma cell lines, were kindly provided by Dr. J. Kolls (University of Pittsburgh). They were developed in asbestos-exposed BALB/c mice by Davis et al. (16). Tumor cells were cultured in RPMI 1640 medium containing 10% of FCS, l-glutamate (Invitrogen, Carlsbad, CA), and penicillin–streptomycin. Tumor cells were maintained at 37°C in a humidified 5% CO2 atmosphere.
Tumor growth and treatments
Mice were injected s.c. in the left flank with 2.0 × 106 AB12 cells in 100 μl of PBS at day 0. On day 2, the right flank was injected s.c. with 2.0 × 106 AB12 cells. Tumor dimensions were measured using microcalipers. Tumor size was expressed as tumor area in squared millimeters using the longest length and the perpendicular width (length × width). Tumor growth was monitored every 2–3 d. On days 10–12, when the longest length of primary tumors reached >5 mm, animals were randomly assigned to different treatment groups. Mice were sacrificed when total tumor dimension reached 150 mm2 or showed signs of ulceration as per institutional ethics protocols.
For intrathoracic tumor model of malignant mesothelioma, the procedure was previously described in detail (17, 18). Briefly, mice were anesthetized and injected with 0.5 × 106 of AB12 cells suspended in 200 μl of PBS into the right pleural cavity through an intercostal space using 24-G Angiocath (BD Medical, Sandy, UT). The surgical procedure was not associated with mortality or morbidity. Survival study was carried out in lieu of tumor burden assessment to compare treatment groups. The animals were sacrificed when they met the predetermined criteria (i.e., tumor growth compromising food and water intake or respiration) established to minimize pain and suffering and were scored as death.
Tregs were depleted at the indicated time points after tumor cell inoculation by i.p. injection of 1 μg of DT (Sigma-Aldrich Canada) on two consecutive days if not stated otherwise.
Local radiotherapy (LRT) was delivered to the tumor as previously described (10). Radiation was given using the X-Rad 225Cx small-animal image-guided irradiator (Precision X-Ray). The irradiator has a 225-kVp x-ray tube (Varian Associates) and a flat-panel silicon detector mounted on a 360° rotation C-arm gantry. The automated stage is movable on the x-, y-, and z-axes. The mean targeting displacement error is ≤0.1 mm in the x–y–z-planes. Radiation was given to mice under isoflurane anesthesia. To initially visualize the animal, the tumor fluoroscopic mode was used. To precisely target the tumor, a scout cone-beam computed tomography was created at a 40-kVp tube potential and 0.5-mA current. The tomography was then reconstructed at a 0.4-mm voxel size. The beam source was collimated to either a 1.5- or 2-cm diameter circular field. To confirm the area to be irradiated, the tumor was then visualized under fluoroscopic imaging with the collimator in place, immediately prior to delivery of the treatment. Radiation was delivered at a tube potential of 225 kVP and a 13-mA current for a dosage rate of 3.02 Gy/min. The daily dose was given from two angles, half from above (180°), and half from below (0°). Total dose was given in divided fractions over 3 d according to treatment protocols. After irradiation, mice were placed back in their cages and housing facilities.
Peripheral blood was drawn from the inferior vena cava of mice immediately after euthanized by inhalation of CO2, and spleen, draining lymph nodes, and tumors were collected from mice. Tumors were minced and transferred to gentleMACS C Tubes (Miltenyi Biotec) containing DNAse (10104159001; Roche Diagnostics) and Liberase (Roche Diagnostics) and were dissociated into single-cell suspension with a gentleMACS Octo Dissociator with Heaters (Miltenyi Biotec). Homogenized spleen, lymph node, and tumor were passed through the 70-μm cell strainer to achieve single cells. Ammonium–chloride–potassium lysis buffer (Invitrogen) was added to lyse RBCs. After washing thrice with FACS buffer and Fc receptor blocking with purified rat anti-mouse CD16/CD32 (BD Biosciences), cells were stained with surface markers. Dead cells were discriminated with Fixable Viability Dye eFluor 780 (Thermo Fisher Scientific). Nuclear or cytoplasmic staining was performed with the Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific) or the Intracellular Fixation and Permeabilization Buffer Set (Thermo Fisher Scientific) according to the instructions of the manufacturer. The stained cells were analyzed with LSR II Flow Cytometer (BD Biosciences). The data analysis was performed using the FlowJo software (Tree Star). The following Abs were used for the flow cytometry analysis: PE-Cy7–anti-CD3e (clone 145-2C11; Thermo Fisher Scientific), APC-eF780–anti-CD4 (clone GK1.5; Thermo Fisher Scientific), FITC–anti-CD4 (clone GK1.5; Thermo Fisher Scientific), PE–anti-CD4 (clone GK1.5; Thermo Fisher Scientific), PerCP-Cy5.5–anti-CD4 (clone GK1.5; Thermo Fisher Scientific), APC–anti-CD8a (clone 53-6.7; Thermo Fisher Scientific), FITC–anti-CD11b (clone M1/70; BioLegend), APC–anti-CD25 (clone PC61.5; Thermo Fisher Scientific), PerCP-Cy5.5–anti-CD44 (clone IM7; Thermo Fisher Scientific), APC-eF780–anti-CD45 (clone 30-F11; Thermo Fisher Scientific), APC-eF780–anti-CD62L (clone MEL-14; Thermo Fisher Scientific), PE–anti-CD68 (clone FA-11; Thermo Fisher Scientific), PE–anti-CD69 (clone H1.2F3; Thermo Fisher Scientific), APC–anti-CD206 (clone C068C2; BioLegend), PE-anti–CTLA-4 (clone UC10-4B9; Thermo Fisher Scientific), PE–anti-Foxp3 (clone FJK-16s; Thermo Fisher Scientific), FITC–anti-F4/80 (clone BM8; BioLegend), PE-anti–Ly-6C (clone HK1.4; BioLegend), APC-anti–Ly-6G (clone 1A8; BioLegend), and PE-anti–PD-1 (clone J43; Thermo Fisher Scientific).
Cytokine production analysis
For intracellular flow cytometric analysis of cytokine production, splenocytes were cultured with RPMI 1640 containing 10% of FBS, Cell Stimulation Cocktail (Thermo Fisher Scientific), and Protein Transport Inhibitor (BD GolgiStop) for 5 h. After staining for the surface markers and dead cells, the cells were fixed and permeabilized using the Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific) or the Intracellular Fixation and Permeabilization Buffer Set (Thermo Fisher Scientific) according to the instructions of the manufacturer and then stained with PE-anti–IFN-γ (clone XMG1.2; Thermo Fisher Scientific), PE–anti-perforin (clone eBioOMAK-D; Thermo Fisher Scientific), and PE-anti–granzyme B (clone NGZB; Thermo Fisher Scientific). The frequency of cytokine producing cells within the CD8+ T cell population was determined by flow cytometry.
At 10 d after completion of treatment, tumors were dissected from mice of each group and embedded into OCT and snap frozen in liquid nitrogen. The tumor blocks were kept in −80°C freezer until sectioning was performed. Sections (5 μm) were fixed with paraformaldehyde 2% and stained with polyclonal anti-CD3 Ab (Dako) and anti-Foxp3 Ab (clone FJK-16s; Thermo Fisher Scientific) and then with Alexa Fluor 488 anti-rabbit Ab and Alexa Fluor 555 anti-rat Ab. Nuclei were stained with DAPI (Sigma-Aldrich). The fluorescence images of whole slides were captured by Wave FX Yokogawa Spinning-Disk Confocal microscope (Yokogawa Electric) with an automated stage and Velocity software (Mountain View, CA).
Total RNA was extracted from tumors and spleens using Rneasy Microarray Tissue Mini Kit (QIAGEN). cDNA was synthesized with High-Capacity cDNA Reverse Transcription Kits (Thermo Fisher Scientific) on a C1000 Touch Thermal Cycler (Bio-Rad Laboratories) following the manufacturer’s protocols. Quantitative real-time PCR was performed on the CFX384 Touch Real-Time PCR Detection System (Bio-Rad Laboratories) using 5 μl of SsoAdvanced Universal SYBR Green Supermix (Bio-Rad Laboratories), 2 μl of cDNA (500 ng/μl), and 300 nM of gene primers in a total volume of 10 μl. Conditions used for quantitative RT-PCR included an initial step of 30 s at 96°C, followed by 40 cycles of 5 s at 95°C and 20 s at 60°C. Primers of housekeeping and all target genes were designed by using ABI Prism Primer Express software Version 2.0. Results were normalized to Gapdh, and relative expression levels were calculated using the 2-ΔΔCT method. The following primers were used for real-time PCR experiments (from 5′ to 3′): Arg1 forward, 5′-TTATCGGAGCGCCTTTCTCA-3′, and reverse, 5′-GGAGCTGTCATTAGGGACATCAA-3′; Cd80 forward, 5′-GCTGCTGATTCGTCTTTCACAA-3′, and reverse, 5′-CGGCAAGGCAGCAATACCT-3′; Cd86 forward, 5′-ACTTACGGAAGCACCCACGAT-3′, and reverse, 5′-CCACGGAAACAGCATCTGAGA-3′; Ctla4 forward, 5′-GCCTTTTGTAGCCCTGCTCA-3′, and reverse, 5′-CCACTGAAGGTTGGGTCACC-3′; Fasl forward, 5′-TGCAGAAGGAACTGGCAGAAC-3′, and reverse, 5′-GGCTCTTTTTTTTCAGAGGGTGTAC-3′; Gapdh forward, 5′-GAACGGATTTGGCCGTATTG-3′, and reverse, 5′-TTGGCTCCACCCTTCAAGTG-3′; Havcr2 forward, 5′-ACCACGGAGAGAAATGGTTCAG-3′, and reverse, 5′-TTTCATCAGCCCATGTGGAAA-3′; Icos forward, 5′-GGAACCTTAGTGGAGGATATTTGC-3′, and reverse, 5′-CTATTAGGGTCATGCACACTGGAT-3′; Ifng forward, 5′-CAACAGCAAGGCGAAAAAGG-3′, and reverse, 5′-TGGTGGACCACTCGGATGA-3′; Il2 forward, 5′-CAGGATGGAGAATTACAGGAACCT-3′, and reverse, 5′-TTGAGATGATGCTTTGACAGAAGG-3′; Mrc1 forward, 5′-GCTTCTCCTGCTTCTGGCTTT-3′, and reverse, 5′-GCAGCGCTTGTGATCTTCATT-3′; Nos2 forward, 5′-ACTGGCCTCCCTCTGGAAAG-3′, and reverse, 5′-CATGAAGGACTCTGAGGCTGTGT-3′; Pdcd1 forward, 5′-GGAGCAGAGCTCGTGGTAACA-3′, and reverse, 5′-AGGGACAGGTGCTGCTGAAG-3′; Prf1 forward, 5′-AGAGTGTCGCATGTACAGTTTTCG-3′, and reverse, 5′-TGATAAAGTGCGTGCCATAGGA-3′; Tgfb1 forward, 5′-GAGCCCTGGATACCAACTATTGC-3′, and reverse, 5′-CTTCCAACCCAGGTCCTTCC-3′; Tnfa forward, 5′-CCACGCTCTTCTGTCTACTGA-3′, and reverse, 5′-TAGTTGGTTGTCTTTGAGATCCATG-3′; Tnfrsf18 forward, 5′-AGAACGGAAGTGGCAACAACA-3′, and reverse, 5′-CACCGGAAGCCAAACACAAT-3′; and Tnfsf18 forward, 5′-AAGGGCAGAGAGGTGCAAGAA-3′, and reverse, 5′-TGCAGGACTCGATGGCAGTT-3′.
Statistically significant differences between two groups were assessed using unpaired t test. Comparisons between more than two groups were carried out by ANOVA with Tukey multiple comparisons test. Differences were considered statistically significant at a p value <0.05. Kaplan–Meier survival curves were used to estimate survival rates for different treatment groups and were compared with the log-rank test. All statistical analyses were performed using GraphPad Prism 8 software. No statistical method was used to predetermine the sample size.
All animal experiments were approved by the Animal Research Ethics Board at the Toronto General Research Institute, University of Toronto. This study was conducted in accordance with institutional and Canadian Institute of Health guidelines.
Tregs accumulate in the tumor microenvironment after nonablative oligofractionated irradiation
We previously developed a murine mesothelioma model that mimicked the clinical setting in which mice received nonablative oligofractionated irradiation with a radiation dose of 15 Gy in three fractions (5 Gy for three consecutive days) (10). An s.c. model was chosen to facilitate the safe delivery of radiation to the tumor. A radiation dose of 15 Gy in three fractions cannot be delivered to the intrathoracic mesothelioma model, which we had developed previously (17, 18), because of toxicity. The s.c. model and intrathoracic model showed upregulation of CD8+ T cell and Tregs infiltration in the tumor microenvironment (Supplemental Fig. 1A, 1B), suggesting that the s.c. model is an adequate surrogate of the intrathoracic model, although less aggressive. The kinetics of accumulation of immune cells and the proportion of Tregs and CD8+ T cells infiltration, however, present some differences that may be due to speed of tumor growth in the intrathoracic model. In the s.c. model, LRT effectively delays the tumor growth for ∼7–10 d before tumor growth resumes (10). Using this platform, we first examined the infiltration of T cells and myeloid cells in the tumor, spleen, and blood at different time points after nonablative oligofractionated irradiation to better characterize the potential local and systemic impact of LRT. Samples were collected 2 d (early phase after LRT), 5 d (irradiation-responsive phase), 7 d (tumor stabilization), and 12 d (irradiation failure phase) after completing LRT. The proportion of total CD3+ T cells, CD4+ T cells (CD3+CD4+), CD8+ T cells (CD3+CD8+), Tregs (CD4+CD25+Foxp3+), monocytic myeloid-derived suppressor cells (M-MDSCs; CD11b+Ly-6ChighLy-6G−), polymorphonuclear myeloid-derived suppressor cells (PMN-MDSCs; CD11b+Ly-6ClowLy-6G+), M2 macrophages (F4/80+CD68+CD206+), monocytes (CD11b+SSClowLy-6G−), Ly-6Chigh classical monocytes, and Ly-6Clow nonclassical monocytes were analyzed by flow cytometry.
The proportion of total CD3+ T cells, CD8+ T cells, and Tregs of total CD45+ cells in tumor were not different between irradiated and untreated mice on days 2 and 5 after LRT (Fig. 1A–D). However, on days 7 and 12, both CD8+ T cells and Tregs in the tumor microenvironment significantly increased in mice treated with LRT compared with untreated tumors (Fig. 1A–D, Supplemental Fig. 2A–D). The proportion of CD8+ T cells in spleen 7 d after LRT was lower compared with untreated control (Supplemental Fig. 3A, 3B), whereas the proportion of CD8+ T cells in blood 5 and 7 d after LRT was higher than untreated control (Supplemental Fig. 3C, 3D). Although the levels of M-MDSCs in spleen and blood were not different between irradiated and untreated mice over time (Supplemental Fig. 3G, 3I), M-MDSC population was significantly higher in the tumor microenvironment in irradiated mice on days 2, 5, and 7 after LRT and dramatically decreased on day 12 (Fig. 1E, Supplemental Fig. 2E). In contrast, the levels of PMN-MDSCs in tumor, spleen, and blood were lower after LRT compared with control (Fig. 1F, Supplemental Fig. 2E, Supplemental Fig. 3H, 3J). The proportion of M2 macrophages in irradiated mice significantly decreased compared with untreated mice on days 2, 5, and 7 (Fig. 1G, Supplemental Fig. 2F). Over 90% of the macrophages were M2 phenotype in untreated mice. Although the ratio of M2 macrophages in the tumor decreased to ∼50% after irradiation, the proportion reversed back to over 90% by day 12 (Fig. 1H, Supplemental Fig. 2G). Although the proportion of classical monocytes in the blood was similar between irradiated and nonirradiated mice, the proportion of nonclassical monocytes were significantly higher on days 5, 7, and 12 after LRT (Supplemental Fig. 3K–O).
Taken together, these data demonstrate that LRT induces an inflammatory response with an immediate rise in M-MDSC and M1-like macrophages in the tumor microenvironment, correlating with the tumor response. The T cell response occurs in a second phase during the period of stability and is characterized by an upregulation of Foxp3+ Tregs and CD8+ T cells in the tumor microenvironment. In a third phase, the inflammatory response resolves, and the tumor relapses. These observations suggest that Foxp3+ Tregs could lead to the resolution of the inflammatory response, generating a switch from M1 to M2 macrophages associated with irradiation failure and tumor progression.
Selective depletion of Foxp3+ Tregs reduces tumor growth in mesothelioma
To directly investigate the role of Foxp3+ Tregs in failure of nonablative oligofractionated irradiation, we took advantage of a well-validated DEREG mouse model (15). Consistent with previous reports (15, 19, 20), DT injection to DEREG mice led to marked depletion (more than 90%) of Foxp3+ Tregs in the spleen, blood, and lymph nodes (Fig. 2A, 2B). To determine the effect of selective Foxp3+ Treg depletion using DT on tumor growth, we used the s.c. mesothelioma model and intrathoracic mesothelioma model. DEREG mice treated with DT demonstrated significantly longer survival than DT-injected wild-type BALB/c (WT) mice in intrathoracic mesothelioma model (Fig. 2C). We next measured tumor growth in s.c. mesothelioma model. After the longest diameter of the tumor reached at least 5 mm in each mouse, DT was injected on days 10 and 11 after tumor cell injection. The tumor growth curves in WT + PBS, WT + DT, and DEREG + PBS groups were similar. However, tumor growth in DT-injected DEREG mice was significantly slower compared with the three other groups (Fig. 2D). These data demonstrate that selective Foxp3+ Treg depletion leads to reduction in tumor growth but does not cure mesothelioma-bearing mice. Because tumor growth in WT + PBS, WT + DT, and DEREG + PBS groups were similar, we used only DEREG mice for the following experiments.
Synergistic antitumor effects of LRT and transient depletion of Tregs
We next investigated whether transient Foxp3+ Treg depletion after nonablative oligofractionated irradiation could generate a synergistic effect on tumor growth. AB12 were injected s.c. into DEREG mice, and these mice were randomly assigned to one of the following treatment groups on day 10 after tumor cell injection: 1) no treatment (no radiation [NoRTx] + PBS), 2) DT injection alone (NoRTx + DT), 3) LRT alone (LRT + PBS), or 4) combination of LRT and DT injection (LRT + DT) (Fig. 3A). As shown in Fig. 3C, transient Foxp3+ Treg depletion by DT injection after LRT showed the best tumor growth delay compared with LRT alone (p = 0.009) and DT injection alone (p < 0.0001). Moreover, three out of five mice (60% of mice) in LRT + DT group showed complete responses in the long term (10 wk after tumor cell injection; Fig. 3B).
To evaluate the involvement of tumor-infiltrating T cells and Tregs in each treatment group, we stained tumor sections with Abs against CD3 and Foxp3. Immunofluorescent staining of the tumor demonstrated high infiltration of Foxp3+ cells after LRT compared with no treatment and high infiltration of CD3+ T cells with very few Foxp3+ cells in tumors treated with LRT + DT (Fig. 3D). Quantitative RT-PCR analysis was performed to examine gene profiles of the immune-associated cytokines, cytolytic enzymes, costimulatory molecules, and coinhibitory molecules. The mRNA expression levels of three cytokines IFN-γ, IL-2, and TNF-α and two cytolytic enzymes perforin 1 and Fas ligand (Fasl) were significantly higher in the tumor treated with LRT + DT (Fig. 4A–E). Both coinhibitory molecules, such as PD-1 (Pdcd1) and TIM3 (Havcr2), and costimulatory molecules, such as ICOS, GITR (Tnfrsf18), CD80, and CD86, were significantly upregulated in the tumor treated with LRT + DT (Fig. 4F–L). Treg-related gene TGFb1 expression was not significantly different between the four groups (Fig. 4M). Furthermore, expression levels of M-MDSC and macrophage-related genes (Mrc1, Arg1, and Nos2) were mostly similar across all four groups at this time point (Fig. 4N–P). The mRNA expression level of GITR ligand (GITRL; Tnfsf18) was upregulated in the tumor treated with LRT + DT (Fig. 4Q).
LRT combined with transient depletion of Tregs generates an in situ vaccination with immune memory T cells protective against tumor rechallenge
Because 60% of mice treated with nonablative oligofractionated irradiation followed by selective Treg depletion demonstrated complete responses in the long term, we next investigated the role of this combination therapy to generate a protective immunologic memory response. Two DEREG mice that were injected s.c. with AB12 and cured with LRT and DT injection (Fig. 3A, 3B) were rechallenged at 6 wk after complete responses, with the same tumor delivered to the opposite flank. Both mice completely rejected the new tumor (Fig. 5A). We then examined the memory T cell subsets in these tumor-rejected DEREG mice using flow cytometry. AB12 were again inoculated to the tumor-rejected DEREG mice (second rechallenge), and spleen, skin-draining lymph nodes, and blood were collected 3 d after the second rechallenge. CD44+CD62L+ (central memory) CD4+ and CD44+CD62L+ (central memory) CD8+ T cells increased in spleen, skin-draining lymph nodes, and blood of the rechallenged tumor-rejected mice compared with naive mice and tumor-bearing mice (Fig. 5B). CD44+CD62L− (effector memory) CD4+ also increased in spleen and skin-draining lymph nodes of rechallenged tumor-rejected mice compared with naive mice and tumor-bearing mice (Fig. 5B). CD44+CD62L− effector memory CD8+ T cells increased in spleen of rechallenged tumor-rejected mice compared with naive mice and tumor-bearing mice (Fig. 5B). These findings demonstrate that nonablative oligofractionated irradiation followed by transient Foxp3+ Treg depletion promotes an in situ vaccination with the development and long-term maintenance of memory T cell subsets, which could remain poised to rapidly recall effector functions upon Ag re-exposure.
Transient Treg depletion after LRT induces abscopal effects
Radiotherapy can induce not only direct cancer cell death but also systemic antitumor effect outside the irradiated field (abscopal effect). To investigate this effect, AB12 were injected s.c. into DEREG mice at two separate sites, defined as the primary site that was irradiated and the secondary site outside the irradiation field with nonablative oligofractionated irradiation and selective Foxp3+ Treg depletion. These mice were randomly assigned to one of the following groups before starting treatments: 1) no treatment (NoRTx + PBS), 2) DT injection alone (NoRTx + DT), 3) LRT alone (LRT + PBS), or 4) combination of LRT and DT injection (LRT + DT). We first examined the antitumor effects with one dose (5 Gy) of radiation (Fig. 6A). Mice treated with LRT (5 Gy) + DT showed significantly smaller primary tumor size than mice treated with LRT alone (p = 0.007) and DT alone (p = 0.03). In total, three out of five mice (60% of mice) in the LRT (5 Gy) + DT group showed complete responses at the primary tumor site (Fig. 6B). However, the secondary (nonirradiated) tumor size in LRT (5 Gy) + DT group was not significantly different compared with NoRTx and DT alone (Fig. 6C). We then examined the abscopal effect with 15 Gy in three fractions using three daily doses of 5 Gy of LRT (Fig. 6D). Mice treated with LRT (5 Gy × 3) followed by Treg depletion showed significantly smaller primary tumor size than the mice treated with LRT alone (p = 0.02) and DT injection alone (p < 0.0001) (Fig. 6E). Furthermore, the size of secondary tumor in LRT (5 Gy × 3) + DT group was significantly smaller than the size in LRT + PBS group (p < 0.0001) and NoRTx + DT group (p = 0.02) (Fig. 6F). These results indicate that transient Foxp3+ Treg depletion after LRT enhances local control even with one fraction of nonablative irradiation. However, despite improved local control, one fraction of radiation is not sufficient to generate an abscopal effect. In contrast, three fractions of 5 Gy, which is known to generate an immunogenic cell death (21), was sufficient to generate significant abscopal effects when Foxp3+ Tregs were depleted after LRT.
Transient Treg depletion after LRT enhances immune cytotoxicity
Finally, we investigated the mechanism of this abscopal effect induced by nonablative oligofractionated irradiation, followed by transient Treg depletion. DEREG mice were injected s.c. with AB12 bilaterally and received treatments as shown in Fig. 6D. Tissue samples were collected 10 d after the last treatment and examined by flow cytometry. We only assessed spleen and secondary tumor because the primary tumor size in LRT + DT group 10 d after the last DT injection was very small, and we could not obtain enough number of single cells to perform flow cytometry accurately. The proportion of GFP+Foxp3+ Tregs of total CD3+CD4+ T cells in spleen and secondary tumor decreased significantly in mice treated with LRT + DT compared with those treated with LRT alone and those without treatment (Fig. 7A, 7B). Interestingly, the proportion of GFP+Foxp3+ Tregs of total CD3+CD4+ T cells in secondary nonirradiated tumors increased significantly in mice treated with LRT alone compared with those without treatment (Fig. 7B). PD-1 and CTLA-4 expression on CD8+ T cells in secondary tumors were significantly upregulated in LRT + DT group compared with LRT alone and no treatment groups (Fig. 7D, 7F). In contrast, PD-1 and CTLA-4 expression on CD4+ T cells in secondary tumors were not different between LRT + DT and LRT-alone groups (Fig. 7C, 7E). The proportions of total CD4+ T cells and total CD8+ T cells in spleen did not show any significant difference (Fig. 7G, 7H); however, CD69+-activated CD4+ and CD8+ T cells increased in the group treated with LRT + DT (Fig. 7I, 7J). Furthermore, the secretion of IFN-γ, perforin, and granzyme B was evaluated by intracellular staining using flow cytometry in CD8+ splenocytes. CD8+ T cells in spleen demonstrated increased intracellular IFN-γ, perforin, and granzyme B production in mice treated with a combination of LRT + DT (Fig. 7K–M, Supplemental Fig. 4A–C). These results indicate that transient Foxp3+ Treg depletion after nonablative oligofractionated irradiation induces systemic activation of T cells and enhances the production of antitumor cytokines and cytolytic enzymes.
This study demonstrated that a major limitation of nonablative oligofractionated irradiation is the upregulation of Foxp3+ Tregs in the tumor microenvironment. A transient depletion of Foxp3+ Tregs at the end of irradiation provided cure in over 50% of the mice even with one fraction of nonablative irradiation, whereas Foxp3+ Treg depletion alone had limited impact on tumor growth. Foxp3+ Treg depletion after three fractions of nonablative irradiation was also associated with an abscopal effect and long-term immune memory response. One fraction of nonablative irradiation, however, was not sufficient to generate the abscopal effect.
Translational works in preclinical animal models performed by our group and others have shown that the benefit of oligofractionated irradiation is, to a large part, mediated by the development of a specific activation of the immune system against the tumor (10, 11, 13, 21, 22). In this study, we demonstrated that LRT induced both local and systemic immune responses in a time-dependent manner after LRT. We observed three phases after LRT. In the first phase, irradiation generated an inflammatory response with upregulation of inflammatory monocytes and M1-like macrophages. T cells were then upregulated in the tumor microenvironment in a second phase, with high proportion of Foxp3+ Tregs and CD8+ T cells correlating with tumor stability. The third phase demonstrated M2 macrophages repopulating the tumor immune microenvironment in association with tumor relapse.
This time course demonstrates that Foxp3+ Tregs are critical in the switch from the inflammatory and specific antitumoral immune response generated by the short course of irradiation back toward an immune protective microenvironment. This observation correlates with recent observation, demonstrating that irradiation failure is due to Tregs even in the presence of checkpoint inhibitors (23). Oligofractionated irradiation can render poorly immunogenic tumors sensitive to immune checkpoint blockades, but the response is transient unless Tregs are depleted (23). CTLA-4 inhibitors can deplete Tregs and generate a systemic response with potential cure in combination with oligofractionated irradiation in preclinical models (24). However, the ability to deplete Tregs with CTLA-4 Ab rely on the Ab-dependent cellular cytotoxicity, which can be altered after irradiation because of the changes in monocyte and macrophage population and their FcgR expression (25). Hence, the response to CTLA-4 in the context of irradiation can be variable depending on the timing of administration.
Other immunosuppressive cells, such as myeloid-derived suppressor cells (MDSCs), were relatively low in the tumor microenvironment during the period of tumor stability and relapse after irradiation. The inflammatory monocytes were rapidly upregulated but then rapidly declined after Treg upregulation. PMN-MDSC slowly increased during tumor progression, but this was a late event rather than a driving force in the failure of irradiation. Some groups reported that local and systemic MDSC numbers decrease drastically 7–14 d after a single high dose of ionizing radiation in murine models (26, 27), whereas others observed that MDSCs were rapidly recruited into the tumor stroma within 3 d after LRT (28, 29). This discrepancy likely reflects the differential kinetics of immune responses in inflammatory monocytes and PMN-MDSC as well as potential differences in the radiation doses and types of animal models (11).
The proportion of M2 macrophages temporarily decreased after irradiation. This polarization of M2 toward M1-like macrophages is consistent with previous report in a pancreatic tumor model with low-dose radiation (30). We previously observed a peak of GITRL+ macrophage on the first day after irradiation (data not shown), which is a marker of M1 macrophage (31). This peak typically resolved as Tregs were upregulated in the tumor. In these series of experiments, we observed that the LRT + DT group had persistent mRNA expression of GITRL (Tnfsf18) in the tumor on day 10 after LRT + DT, suggesting that GITRL+ APCs could play a major role in maintaining the antitumoral immune response generated by nonablative oligofractionated irradiation in the absence of Tregs. Recent work supports this concept by demonstrating that GITRL+ APCs are critical to activate effector CD4+ T cells in the postpriming phase after viral infection (32). Overall, the persistent upregulation of both GITR and GITRL in the LRT + DT group (Fig. 4J, 4Q) suggests that the GITR–GITRL pathway is an important mechanism of immune activation after LRT that is disrupted by Tregs and leads to the reconstitution of M2 macrophages in the tumor microenvironment. The reconstitution of M2 macrophages can be driven by the activation of immunosuppressive pathways, including IL-4, IL-10, and TGF-β, after irradiation (33). The importance of these cytokines, in particular TGF-β, in the upregulation of Tregs will require further investigations in the context of irradiation (34). In our model, the mRNA expression of TGFb1 was similar on day 12 after LRT across the different groups (Fig. 4M), but Tregs had repopulated the tumor by then, and therefore, earlier changes cannot be excluded.
Nonspecific Treg targeting agents, such as CD25-depleting Abs, have already been tested as a promising therapeutic approach in various animal models and entered clinical trials (17, 35). Several studies by our group and others have shown that depletion of CD25+ Tregs before the inoculation of tumor cells leads to significant antitumor effect; however, CD25+ Treg depletion after the inoculation of tumor cells does not show tumor regression in mesothelioma (17, 35, 36) and other types of tumor (37–39). This therapeutic failure is partly due to a concomitant elimination of CD25+ effector T cells and to the persistence of CD25− Tregs (19, 20, 37, 38). Hence, these studies are limited by the lack of specific depletion of Tregs. Although the tumor microenvironment of mesothelioma contains large proportion of M2 macrophages, we undertook the current study in DEREG mice to have a clear understanding of the role of Tregs in the context of mesothelioma. We also previously demonstrated that Treg depletion in combination with chemotherapy had a synergistic effect on the tumor (17). In these series of experiments, we demonstrated that selective Foxp3+ Treg depletion leads to the reduction of growth in established tumor, similarly to some previous studies in other types of tumor (19, 20). However, the impact of selective Foxp3+ Treg depletion alone was limited once the tumor is established and did not cure mesothelioma-bearing mice. Hence, nonablative irradiation was required to achieve the full benefit of Treg depletion once the tumor was established.
The depletion of GFP+Foxp3+ Tregs with DT in DEREG mice was transient, as Foxp3+ cells were almost completely depleted 1 d after DT injection (Fig. 2B) but returned to near normal level in the spleen 10 d after the treatment with DT injection (Fig. 7A). These findings are consistent with previous studies showing that the effect of Treg depletion by DT injection to the DEREG mice is transient, and recovery of Tregs is seen ∼7–14 d after DT injection (19, 20). Although Treg depletion is transitory, the impact is extremely significant after nonablative oligofractionated irradiation, suggesting that long-term Treg depletion is not necessary to obtain a profound benefit. This observation is of importance to reduce the duration of immunotherapy and their side effects in the clinical setting.
We demonstrated that tumor-bearing mice could be cured with LRT and Foxp3+ Treg depletion, whereas LRT alone was never curative in this model. Cured mice completely rejected the tumor after rechallenge, confirming the in situ vaccination generated by the irradiation. This benefit was associated with increased central and effector memory T cell subsets. We previously demonstrated that the same irradiation protocol performed 7 d before radical surgery led to a long-term antitumor immune protection that was primarily driven by CD4+ T cells (10). These findings suggest that nonablative oligofractionated irradiation followed by surgery or adjuvant immunotherapy to deplete Tregs can provide a strong and long-lasting antitumoral immune response.
Recent study in patients treated with radiotherapy and CTLA-4 blockade in metastatic non–small cell lung cancer demonstrated that the release of IFN-β and new TCR clones were associated with an objective abscopal response (12). The importance of new TCR clones in prognosis after oligofractionated irradiation was also demonstrated in preclinical models (40–42). Furthermore, recent clinical investigations demonstrated that the diversity and clonality of the TCR repertoire provide information that can help to predict response to checkpoint inhibitors (43). Patients with a high clonality at baseline can benefit from anti–PD-1 therapy, but not from anti–CTLA-4 therapy, whereas patients with high diversity at baseline can benefit from anti–CTLA-4 therapy, but not from anti–PD-1 therapy (43). Hence, anti–CTLA-4 therapy broadens the TCR repertoire by allowing the development and expansion of a limited number of clones, whereas PD-1/PD-L1 inhibitors drive proliferation of a restricted number of pre-existing clones after irradiation (40, 44, 45). We demonstrated in our model that PD-1 was upregulated on CD8+ tumor-infiltrating T cells in both the irradiated and nonirradiated tumor after LRT + DT, suggesting that immune checkpoint inhibitors could still be beneficial to optimize the effect of irradiation in the context of Treg depletion.
There has been a rapid expansion in the number of clinical trials combining irradiation with various immunotherapies, and the results of ongoing clinical trials are eagerly awaited to determine safety, efficacy, the optimal dose, technique and sequencing of irradiation, type of immunotherapy, and biomarkers (46). We are currently performing a phase 1 clinical trial to characterize the impact of nonablative oligofractionated irradiation prior to surgery on the tumor microenvironment in MPM (NCT04028570). Based on our preclinical models and data from the literature, we are using three fractions of irradiation to boost the gross tumor volume to at least 21 Gy of radiation to potentiate the abscopal effect before surgery (21). This protocol will include specific tissue collections to determine the optimal dose of radiation to generate an immune activation and define potential targets for immunotherapy.
In conclusion, we demonstrated that the accumulation of Foxp3+ Tregs into the tumor microenvironment after LRT is one of the main reasons for irradiation failure after nonablative oligofractionated irradiation in murine mesothelioma. Transient Foxp3+ Treg depletion after LRT showed efficient local and systemic antitumor responses with 15 Gy in three fractions. These findings will help to refine the strategy of immuno-radiotherapy to generate an optimal antitumor immune response in mesothelioma patients.
This work was supported by the following grants and fellowships: the Mesothelioma Applied Research Foundation (to M.d.P.), the Princess Margaret Cancer Foundation (to M.d.P.), the Canadian Mesothelioma Foundation (to M.d.P.), and the Uehara Memorial Foundation Overseas Research Fellowship (to M.K.).
The online version of this article contains supplemental material.
Abbreviations used in this article:
monocytic myeloid-derived suppressor cell
myeloid-derived suppressor cell
malignant pleural mesothelioma
polymorphonuclear myeloid-derived suppressor cell
surgery for mesothelioma after radiation therapy
regulatory T cell
M.d.P. received personal fees from Bayer and Astra-Zeneca unrelated to the submitted work. The other authors have no financial conflicts of interest.