CCR2-Dependent Recruitment of Tregs and Monocytes Following Radiotherapy Is Associated with TNFα-Mediated Resistance

Radiotherapy (RT) is a widely used anticancer treatment and serves as a mainstay for the management of head and neck squamous cell carcinoma (HNSCC; ref. 1). The effects of RT on the priming and effector phases of antitumor immunity make it an appealing strategy to generate immunity against the patient's own tumor (2) and an interesting combinational tool to improve several anticancer immunotherapies (3, 4). Although radiation is known to enhance the infiltration of activated T cells in the tumor microenvironment (TME; refs. 5, 6), local immunosuppression in the TME is most frequently maintained or even amplified by RT. Regulatory T cells (Tregs) and tumor-associated macrophages (TAMs) represent the main protagonists shaping the immunosuppressive TME and promoting tumor growth, and experimental evidence in murine models has demonstrated that the absence of either Tregs or TAMs delays tumor progression (7, 8). Macrophage and Treg infiltration into tumors is associated with poor prognosis, therapeutic failure, and tumor relapse in multiple solid cancers (9).

The chemokine–chemokine receptor network plays key roles in shaping the TME into an immunosuppressive site by inducing the differential recruitment of regulatory and effector immune cells (10). For instance, the CCL2/CCR2 axis represents a major marker of tumor development and has predictive value following cancer therapy in both human and mouse models (11). Studies have demonstrated that RT-induced CCL2 production by the tumor contributes to therapeutic failure (12–14), but the mechanisms involved remain elusive. The CCR2-dependent recruitment of Ly6C^high monocytes can substantially contribute to the TAM compartment (15, 16). We have shown that CCR2 mediates the recruitment of a subset of CCR2⁺ Tregs in different tumor models, as well as in human oral squamous cell carcinoma (OSCC; ref. 17). We, thus, hypothesized that the CCL2/CCR2 axis might be implicated in the corecruitment of TAMs and Tregs following RT and the constitution of TME immunosuppression. The action of Tregs on innate monocyte-derived cells in vivo has been poorly investigated thus far. Here, we showed that performing local RT on oral tumors in mice induced a CCR2-dependent accumulation of Tregs and TNFα-producing monocytes and macrophages in the tumor bed. Tregs interacted with monocytes/macrophages in irradiated tumors and, finally, TNFα inhibition reduced Treg activity and was associated with increased RT efficacy. These observations define an important cross-talk between components of the innate and adaptive immune system contributing to tumor resistance to RT.

Firefly luciferase-expressing TC1/Luc cells, generated by the HPV16 E6/E7 and c-H-ras retroviral transduction of lung epithelial cells of C57BL/6 origin, were kindly provided by T.C. Wu (Johns Hopkins Medical Institutions) in 2009. Expression of E6 and E7 genes was confirmed by RT-PCR, as well as major histocompatibility complex (MHC) haplotype H2b by flow cytometry (18). The tdTomato TC1 cell line was generated by infection of TC1 cells with a tdTomato lentivirus (kindly provided by Dr. N. Bercovici and M. Lambert, Institut Cochin, Paris, France). TC1 cells were cultured in RPMI-1640 with 10% fetal bovine serum (FBS), sodium pyruvate, nonessential amino acids, and penicillin/streptomycin (P/S) antibiotics (Gibco, Invitrogen) and incubated at 37°C in 5% CO₂. The RAW 264.7 monocyte/macrophage cell line was obtained from ATCC and cultured in Dulbecco's Modified Eagle's Medium with 10% FBS and P/S (Invitrogen). Master stocks of the cells were prepared upon receipt and then maintained under limited passage after resuscitation (typically under 15).

C57BL/6 mice were purchased from Janvier CERT and housed at the Gustave Roussy animal facility (Plateforme d'Evaluation Preclinique, PFEP) and the Centre d'exploration fonctionelle, Pitié-Salpêtrière. C57BL/6-Foxp3tm1Mal mice (called Foxp3-EGFP mice herein; ref. 19), Ccr2^−/− mice (The Jackson Laboratory), and Csf1r-Gal4VP16/UAS-ECFP (MacBlue; David Hume, Roslin Institute, UK; ref. 20) mice were intercrossed at least 10 generations to generate Foxp3-EGFP × MacBlue, Foxp3-EGFP × MacBlue x Ccr2^−/− mice. These last, the Foxp3^{tm3(DTR/GFP)Ayr} (Foxp3-DTR; The Jackson Laboratory; ref. 21), and the MMTV-PyMT-P2A-mCherry-P2A-OVA (PyMT-ChOVA; The Jackson Laboratory; ref. 22) strains were bred at the Centre d'Exploration Fonctionelle, Pitié-Salpêtrière. All animal procedures were performed in accordance with protocols approved by the Ethical Committee CEEA 26 and CEEA 05 and validated by the “Service Protection et Santé Animales, Environnement.” Mice were used between 8 and 16 weeks of age and around 25 weeks of age for the PyMT-ChOVA.

To establish a head and neck tumor implantation model, syngeneic tumor grafts were initiated by injection of 50 μL of PBS suspension containing 5 × 10⁵ TC1/Luc or tdTomato TC1 cells at submucosal sites on the right inner lips of C57BL/6, Ccr2^−/−, Foxp3-EGFP × MacBlue, Foxp3-EGFP × MacBlue × Ccr2^−/− mice, or Foxp3-DTR mice as previously described (18). In some experiments, mice were used without prior injection of tumor cells (tumor-free mice), as indicated in the figures. Throughout the study, the health, weight, and behavior of the mice were assessed every 2 to 3 days. Mice were humanely euthanized upon the presentation of defined criteria (tumor size and bioluminescent signal, loss of >20% of the initial weight), and a survival time was recorded to perform a survival analysis for the treatment groups.

RT-treated mice received single-beam local irradiation to the head and neck region 7 days after tumor inoculation using a 200 kV Varian X-ray irradiator. Selective irradiation of the tumor grafts was performed by the interposition of a 4-cm-thick lead shield on a schedule delivering 7.5 Gy in a single fraction at a dose rate of 1.08 Gy/min. Nonirradiated mice were used as controls.

Neutralizing anti-CD25 (InVivoMAb anti-mouse CD25, clone PC-61.5.3) and anti-TNFα (InVivoMAb anti-mouse TNFα, clone XT3.11), as well as the corresponding isotype controls (InVivoMAb rat IgG1 isotype control, anti-horseradish peroxidase, clone HRPN) were purchased from Bio X cell. Anti-CD25 or isotype control was administered at 20 mg/kg in PBS by intraperitoneal (i.p.) injection on days 4 and 6 after RT. Anti-TNFα or isotype control was administered at 25 mg/kg in PBS by i.p. injection on days 2 and 4 after RT. To deplete Tregs in Foxp3-DTR mice, diphtheria toxin (DTx, Merck) was solubilized in PBS and injected i.p. at 1 μg/mouse on days 2 and 3 after RT.

To monitor tumor growth, bioluminescence imaging was performed using the Xenogen In Vivo Imaging System 50 (IVIS; Caliper Life Sciences) as previously described (18). Signal quantification was performed using the Living Image V 4.3.1 software (Caliper Life Sciences).

For multiphoton imaging of Foxp3-EGFP x MacBlue mice, HNSCC tumors were carefully collected and immobilized in an imaging chamber perfused with oxygenated (95% O₂ plus 5% CO₂) RPMI medium containing 10% fetal calf serum. The local temperature was monitored and maintained at 37°C. The two-photon laser scanning microscopy setup used was previously described (20), and the excitation wavelength used was 870 nm. The system included a set of external non-descanned detectors in reflection with a combination of LP-600 nm followed by LP-462 nm and LP-500 nm dichroic mirrors to split the light and collect the second-harmonic generation signal with a 417/60 nm emission filter, ECFP with a 480/40 nm emission filter, and EGFP with a 525/50 nm emission filter.

Cell motility was measured over 30 minutes, every 30 seconds by 9 consecutive 5 μm z spacing stacks (total of 40 μm thickness) using a plan apochromat 20 × (NA = 1) water immersion objective. Fluorescent cells were monitored over time with three-dimensional automatic tracking and manual correction with Imaris software (Bitplane). The acquisition and analysis protocols for all experimental conditions to be compared were identical. Velocity and straightness were determined using Imaris. The track straightness corresponded to the ratio of the distance between the initial and final positions of each cell to the total distance covered by the same cell.

Tumors were harvested, fixed in 4% paraformaldehyde for 4 hours, incubated in 30% sucrose-PBS overnight at 4°C, embedded in Tissue–Tek OCT compound (Sakura Finetek), and frozen at −80°C. Sectioning was completed on an HM550 cryostat (Thermo Fisher Scientific) at −20°C; 5-μm-thick sections were collected on Superfrost Plus Slides (Thermo Fisher Scientific) and stored at −20°C until use. The slides were counterstained and mounted with Vectashield Mounting Medium with DAPI (Vector Laboratories). Images were acquired with a Zeiss Axio Z1 fluorescence microscope (Carl Zeiss).

Phenotypic characterization was performed using the LSRFortessa instrument (Becton Dickinson) for murine cells. For analysis, FlowJo software (Tree Star Inc.) was used. At indicated time points, TC1/Luc and spontaneous MMTV-PyMT oral tumors were digested in RPMI medium (Gibco, Invitrogen) supplemented with collagenase IV (1 mg/mL; Sigma) with occasional pipetting for 30 minutes at 37°C, entirely resuspended in PBS supplemented with 0.5% BSA and EDTA (2 mmol/L; FACS buffer) and filtered using a 70-μm cell strainer (BD Biosciences). Submandibular blood collection was performed in isoflurane-anesthetized mice using a lithium-heparin minivette POCT (Sarstedt), and the cells were lysed in RBC lysis buffer containing NH₄Cl (0.15 mol/L), KHCO₃ (0.01 mmol/L), and EDTA (0.1 mmol/L). Surface staining was performed by incubating 50 μL of the cell suspension (one tenth of the total tumor tissue or 50 μL of whole blood) with 1 μg/mL purified anti-CD16/32 (2.4G2; BD Biosciences) for 10 minutes at 4°C followed by incubation with a titration-based appropriate dilution of specific surface antibodies (listed below) for an additional 20 minutes. Dead cells were excluded using the Molecular Probes LIVE/DEAD Fixable Stain kit (Life Technologies) according to the manufacturer's instructions. Forward and side scatter parameters were used for doublet exclusion.

After incubation, the cell suspensions were washed once in FACS buffer. The CCL2-binding assay was performed as previously described (17). Briefly, after surface staining, the cells were incubated in the dark for 45 minutes at 37°C in RPMI medium containing murine CCL2AF647(25 nmol/L; Almac Sciences) and then washed with FACS buffer. For INFγ, TGFβ, and TNFα staining, cells were preincubated prior to staining for 2 hours with the Cell Activation Cocktail containing PMA, ionomycin, and brefeldin A according to the manufacturer's instructions (BioLegend). For intracellular staining after surface staining, the cells were fixed in 4% paraformaldehyde for 20 minutes, washed twice in Perm/Wash solution (BD Biosciences), incubated for 10 minutes with purified anti-CD16/32 (1 μg/mL) in Perm/Wash at room temperature, and incubated for 30 minutes in Perm/Wash in the presence of anti-IFNγ, anti-TGFβ, or anti-TNFα. The Foxp3/Transcription Factor Staining Buffer Set (BD Biosciences), anti-Foxp3, and anti–CTLA-4 were used according to the manufacturers' instructions. Samples were washed in FACS buffer and resuspended in 200 μL in FACS buffer before acquisition.

Calculation of the absolute numbers of different cell populations was performed by adding a fixed number (10,000) of nonfluorescent 10-μm Polybead Carboxylate Microspheres (Polysciences) to each vial according to the following formula: number of cells = number of acquired cells × 10,000/number of acquired beads. The number of cells obtained for each sample was extrapolated to the tumor weight or blood volume. The panel of antibodies used comprised the following: anti-CD11b (clone M1/70), anti-Ly6C (clone AL-21), anti-Ly6G (clone 1A8), anti-NK1.1 (clone PK136), anti-I-Ab (clone AF6-120.1), anti-CD11c (clone HL3), anti-Siglec-F (clone E50-2440), anti-CD103 (clone M290), anti-CD45 (clone 30-F11), anti-CD4 (clone RM4-5), anti-CD3 (clone 145-2C11), anti-CD8 (clone 53-6.7), anti-CD25 (clone PC61), anti-CD64 (clone X54-5/7.1), anti-TNFα (clone MP6-XT22), anti-IFNγ (clone XMG1.2), anti-CTLA4 (clone UC10-4F10-11), anti-TGFβ (clone TW7-16B4), anti-Foxp3 (clone MF23; Pharmingen, BD Biosciences), anti-CD206 (clone C068C2; BioLegend), and anti-CCR2 (clone 475301; R&D Systems).

CCL2 in cell supernatants (described in the following sections, 10-fold diluted in the calibration diluent provided in the kit) and mouse sera (2-fold diluted) were evaluated by ELISA using the Mouse Quantikine ELISA Kit from R&D Systems following the manufacturer's instructions.

RNA was extracted from cells and tumor samples as indicated in the figures using the Qiagen RNeasy kit and quantified using NanoDrop spectrophotometer (Thermo Fisher Scientific). The samples were treated with DNAse-I (Qiagen), and RNA was reverse-transcribed using the Bio-Rad's I-Script kit. qRT-PCR analysis of CCL2 was performed using SYBR Green Mix (Applied Biosystems) and the GeneAmp 7900HT Sequence Detection System (Applied Biosystems) as previously described (23) using the following primers for CCL2: F = TCCCTGTCATGCTTCTGG, R = CTGCTGGTGATCCTCTTGTA, and for 18s rRNA as normalizing gene: F = GTAACCCGTTGAACCCCATT; R = CCATCCAATCGGTAGTAGCG.

The genomic TC1/Luc cell line was edited using CRISPR-CAS9 technology to disrupt the Ccl2 gene. The Ccl2 mouse gene knockout via CRISPR kit (product number KN302782, Origene) was used to knockout a region of the Ccl2 gene and knock-in a functional cassette containing GFP and the puromycin-resistance gene. Briefly, TC1/Luc cells were cotransfected with donor template DNA and pCas-guide plasmids provided in the KN302782 kit using Lipofectamine 2000 (Thermo Fisher Scientific). Forty-eight hours later, cells were sorted for GFP expression, cultured for 7 weeks, and then selected for puromycin resistance. Individual cell colonies were isolated by limiting dilution. Supernatants of clones were screened for CCL2 by ELISA (described above), and CCL2-negative clones were selected.

TC1/Luc and RAW264.7 cells were mock-treated or irradiated at 7.5 Gy and then respectively seeded into the lower and upper chambers of 6.5 mm transwells with a 0.4-μm Pore Polycarbonate Membrane Insert (Corning). The number of cells plated was adjusted considering the proliferation rate and the response to RT to yield an approximately equivalent number of cells regardless of the radiation dose delivered and the growth time. After 48 hours of incubation, supernatants were collected and analyzed by ELISA (described above), and the mRNAs were separately extracted from TC1/Luc and Raw264.7 cells, and analyzed by RT-qPCR (as indicated above).

Cytokine and chemokine concentrations in tumor tissues were profiled at Eve Technologies Corporation using their Discovery Assay ref. MD31. Protein extracts from tumor tissues were prepared by homogenization in RIPA buffer (Sigma-Aldrich) containing the cOmplete protease inhibitor cocktail (Roche) using BioMasher disposable homogenizers (Nippi). Protein extracts were diluted to 4 μg/μL and analyzed at Eve Technologies using the multiplex immunoassay Mouse Cytokine Array/Chemokine Array 32-Plex (Milliplex, Merck Millipore) with the BioPlex 200 instrument (Bio-Rad). Cytokine and chemokine concentrations were calculated based on the standard curves generated using the standards included in the assay.

Statistical analyses were performed using Prism Version 7 (GraphPad). Survival data were analyzed using the Kaplan–Meier and log-rank tests for survival distribution. The bioluminescent signal (serving as a measurement of tumor size), immune infiltrate, and tumor perfusion levels were analyzed using appropriate tests. Accordingly, multigroup analyses of variances were performed using one-way or two-way ANOVA followed by Sidak or Tukey multiple comparisons tests, respectively, or the Kruskal–Wallis test followed by Dunn multiple comparisons test. For simple comparison analysis, an unpaired Student t test was used to compare parametric distributions, and the Mann–Whitney test was used to compare nonparametric distributions. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns = not significant.

Orthotopic inoculation of the TC1/Luc cell line was performed at a submucosal site of the inner lip in syngeneic wild-type (WT, Ccr2^+/+) and Ccr2^−/− mice to mimic an HPV-related oral cancer as described previously (18). Using the same model and irradiation treatment (Fig. 1A), we showed here that the lack of stromal CCR2 improved the survival rate after localized RT (Fig. 1B; median survival gain after RT for WT = 3 days, for Ccr2^−/− = 6.5 days), which was associated with a reduction in tumor burden (Supplementary Fig. S1A). RT induced a transient CCR2-dependent increase of classic Ly6C^high monocytes in the blood that was not observed after the local irradiation of tumor-free mice, indicating that the increase in blood Ly6C^high monocytes was dependent on irradiation of the tumor (Fig. 1C). We next monitored the impact of RT on Ly6C^high monocyte and macrophage subsets within the tumor according to the gating strategy depicted in Supplementary Fig. S1B. The Ly6C^high monocytosis induced by RT was associated with a massive and transient accumulation of Ly6C^high monocytes into the tumor, a higher CD64 expression, and a subsequent accumulation of Ly6C^low/–CD64⁺ macrophages, suggesting that the latter mainly arose from the differentiation of Ly6C^high monocytes between days 3 and 5 after RT (Fig. 1D). CCR2 deficiency resulted in a severe impairment of both monocyte and macrophage accumulation on day 5 after RT (Fig. 1E). New infiltrating macrophages in WT mice displayed a higher proportion of I-Ab⁺ cells, whereas CD206 was expressed at a lower frequency intensity in these mice than in nonirradiated mice (Fig. 1F). This activation profile was severely impaired in Ccr2^−/− mice, with a lower expression of I-Ab and CD206 on the few infiltrating macrophages detected in these mice (Fig. 1F). Histologic analysis of MacBlue transgenic mice inoculated with tdTomato TC1 cells unveiled that the accumulation of ECFP⁺ monocytes/macrophages mainly occurred at the tumor periphery after RT and was not observed in MacBlue x Ccr2^−/− mice (Supplementary Fig. S1C). Overall, our results suggest that local tumor irradiation induces a CCR2-dependent increase of monocytes in the blood that subsequently infiltrate the tumor and progressively differentiate into macrophages.

We previously demonstrated that CCR2 represents an important homing receptor of Tregs toward the tumor and that a large fraction of Tregs are CCR2⁺ in human HNSCC (17). In this orthotopic model of HNSCC, RT, together with an increased CCR2-dependent tumor infiltration of monocytes/macrophages, mediated a significant increase in CD4⁺ conventional T cells (Tconv) and Tregs in RT versus untreated tumors at 5 days after treatment (Fig. 2A). In the TC1/Luc model, the accumulation of Tconvs and Tregs was severely reduced in Ccr2^−/− mice, in accordance with our previous study (Fig. 2A; ref. 17). In agreement with the CCR2-mediated recruitment of Tregs to irradiated tumors, the proportion of the CCR2⁺ Treg subset was increased in WT mice following RT, whereas the percentage of CCR2⁺ Tconvs was not significantly changed (Fig. 2B). To investigate the effect of Tregs following RT in this HNSCC tumor model, we selectively depleted Tregs using Foxp3-DTR mice. Diphtheria toxin (DTx) treatment was started on day 2 after irradiation to fully impair RT-induced Treg recruitment (Fig. 2C). Treg depletion in irradiated tumor-bearing mice reduced the tumor burden compared with that in nondepleted control mice (Fig. 2D), confirming that Tregs limit RT efficacy. RT-induced accumulation of macrophages and Tregs was also confirmed in a spontaneous model of HNSCC, using male PyMT-ChOVA mice that develop salivary tumors, suggesting that this observation is not restricted to engrafted models (Fig. 3A and B; ref. 22).

To better understand the mechanisms underlying RT-mediated cell recruitment, we investigated the effect of RT on CCL2 production. Local irradiation of tumor-bearing mice induced a transient increase in the serum CCL2 concentration at 3 days after treatment (Fig. 4A), which directly correlated with the Ly6C^high monocyte release into the blood (Fig. 1C). In the absence of a tumor, the serum CCL2 concentration was not increased after RT (Fig. 4B), whereas CCL2 accumulation was still detectable after irradiation of TC1/Luc tumor-bearing Ccl2^−/− mice, indicating that the circulating CCL2 mainly originated from the tumor cells (Supplementary Fig. S2A). In contrast, no serum CCL2 accumulation was observed in either WT or Ccl2^−/− mice bearing CCL2 knockdown (Ccl2^−/−) TC1/Luc cells (Supplementary Fig. S2B). These results suggest that TC1/Luc tumor cells represent the main source of CCL2 production after RT rather than stromal cells. In the tumor, increased CCL2 transcription was detected by quantitative PCR 3 days after RT and was maintained on day 5 (Supplementary Fig. S2C). In vitro, when TC1/Luc tumor cells were cocultured with the macrophage cell line RAW264.7, CCL2 secretion was increased and peaked when both TC1/Luc and RAW264.7 cells were exposed to ionizing radiation (Fig. 4C), indicating that secreted factors from irradiated monocytes/macrophages further contribute to CCL2 upregulation at both the mRNA (Supplementary Fig. S2D) and secreted protein levels (Fig. 4C). We next performed cytokine profiling on whole-tumor extracts and compared the concentrations of a set of cytokine and growth factors relevant to monocyte-derived cell recruitment and differentiation in nonirradiated control tumors and irradiated tumors at 5 days after RT, when macrophages and Tregs accumulated into irradiated tumors (Fig. 4D). Among the cytokines and growth factors analyzed, CCL2 accumulation was confirmed at 5 days after RT. IL3, GM-CSF, TNFα, and IFNγ significantly increased after RT, whereas M-CSF concentrations were similar to those in nonirradiated tumors. We concluded that local RT induces the secretion of CCL2 by tumor cells favoring the recruitment of Ly6C^high monocytes into the tumor and is associated with an increased production of myeloid-cell–related cytokines and growth factors involved in survival and differentiation.

We speculated that the corecruitment of Tregs and monocytes following RT might favor the cross-talk between these two subsets. TNFα is known to have a direct effect on Treg function (24, 25) and was found to be upregulated in irradiated tumors (Fig. 4D). In the tumor, Ly6C^high monocytes and TAMs represented important TNFα producers. The production of this cytokine was maintained in the new infiltrating cells that accumulated 5 days after RT (Fig. 5A). Anti-TNFα treatment after irradiation improved the efficacy of RT (Fig. 5B), whereas the effect of anti-TNFα was abrogated in Foxp3-DTR mice (Fig. 5C), suggesting that the antitumor effect of anti-TNFα was mediated by Tregs. Anti-TNFα treatment reduced the Treg frequency among CD4⁺ T cells following RT (Fig. 5D) and significantly reduced TGFβ expression by Tregs (Fig. 5E), whereas CTLA4 expression was not modified (Supplementary Fig. S3A). Conventional CD4⁺ and CD8⁺ T-cell infiltration was not modulated by anti-TNFα (Supplementary Fig. S3B), but RT increased the proportion of IFNγ-producing CD8⁺ T cells only when mice were treated with anti-TNFα (Fig. 5F). Our results suggest that RT induces the CCR2-dependent accumulation of TNFα-producing myeloid cells that activate Tregs and dampen the CD8⁺ T-cell response to ionizing radiation via TGFβ production.

To further investigate the mechanism of Treg/monocyte cross-talk, we performed live imaging using intravital biphoton microscopy. Using MacBlue x Foxp3-EGFP double transgenic mice, we monitored the dynamic interactions of corecruited monocytes/macrophages and Tregs within irradiated tumors 5 days after RT, at the time Tregs accumulated within the tumor. Clusters of monocytes/macrophages were present in close proximity to Tregs (Fig. 6A), and Tregs displayed multiple interactions with monocytes/macrophages (Fig. 6A; Supplementary Video S1). Treatment with an anti-CD25 antibody is known to neutralize Treg activity without depleting the cell population (Supplementary Fig. S4A; ref. 26). Anti-CD25–neutralized Tregs displayed an increased mean velocity compared with that of IgG-treated controls (Fig. 6B). Although the rate of Treg interactions with monocytes/macrophages was not affected (Fig. 6C), the duration of these interactions was significantly reduced in anti-CD25–treated mice (Fig. 6D), likely explaining the higher speediness of Tregs (Fig. 6B). The very low number of Tregs in nonirradiated mice prevented us from analyzing their potential interactions with monocytes/macrophages. Anti-CD25 treatment was not sufficient to improve the RT effect (Supplementary Fig. S4B–S4C), but the altered pattern of interaction between Tregs and monocytes/macrophages upon CD25 inhibition was associated with an altered activation profile, as depicted by the reduced expression of MHC class II molecules (I-Ab) in both tumor Ly6C^high monocytes and TAMs (Fig. 6E and F), suggesting that Treg–monocyte/macrophage cross-talk occurs via direct cell-to-cell interactions.

It is well recognized that RT severely affects the TME and tumor immune response (27). Radiation promotes the recruitment of inflammatory cells into the TME but can also attract immunosuppressive cells (28). Our data, obtained using an orthotopic model of HPV-associated oral cancer, indicated that an influx of Ly6C^high monocytes, driven by the CCL2/CCR2 axis, is observable 3 days after a single exposure to 7.5 Gy of ionizing radiation. Our observation on Ly6C^high monocyte mobilization is in accordance with others made using subcutaneous models of pancreatic (13), colon, and lung adenocarcinoma (14) and with intramuscular models of colon adenocarcinoma, glioblastoma, and lung carcinoma (12) with different RT regimens (15–20 Gy), suggesting that the CCL2/CCR2-mediated recruitment of Ly6C^high monocytes can be a general mechanism that follows tumor exposure to ionizing radiation, at least at high doses (>7.5 Gy). This suboptimal RT regimen (7.5 Gy in a single dose) was chosen in our study in order to monitor the added value of the combined immunotherapy targeting CCR2 axis. Nevertheless, it remains to be clearly elucidated whether CCR2-mediated recruitment occurs at lower RT doses, and especially after multiple RT exposures, to more closely recapitulate fractionated RT regimes in the clinical setting.

The mechanism was shown to involve the induction of increased CCL2 expression by irradiated tumor cells based on the following: (i) CCL2 was still detectable and increased after RT when TC1/Luc cells were engrafted into Ccl2^−/− mice; (ii) no upregulation of CCL2 was detectable in the blood of mice engrafted with a Ccl2^−/− cell line (in line with Kalbasi and colleagues; ref. 13) or in the blood of tumor-free mice after RT; and (iii) CCL2 was upregulated in tumor cells in vitro after RT. CCL2 upregulation in irradiated tumor cells is likely enhanced by the release of soluble factors from TAMs, as suggested by the tumor–macrophage coculture experiments. Such mediators could include IL6 (29) and type I interferons (14). Further experiments would be necessary to confirm whether primary tumor monocytes, as well as other stromal cells, might enhance CCL2 upregulation after irradiation.

Our work clearly demonstrated that not only Ly6C^high monocyte, but also Treg, recruitment to the tumor after RT involves the CCL2/CCR2 pathway. The intratumor Tregs accumulated 5 days after RT, whereas Treg accumulation in irradiated Ccr2^−/− mice was severely reduced. We cannot exclude that the increase in Tregs was partly due to the local proliferation of intratumor Tregs after radiation, as this observation has been described using different tumor models (30). Nevertheless, the lack of increase in Tregs in irradiated tumors from Ccr2^−/− mice and the increased proportion of CCR2-expressing Tregs in WT mice after RT are supportive of a significant role of CCR2-mediated Treg recruitment in the tumor bed.

The corecruited monocytes/macrophages and Tregs displayed dynamic interactions within the irradiated tumor tissue, suggesting a direct cell-to-cell mediated cross-talk between these two subsets. Anti-CD25 treatment reduced the contact duration between these subsets, showing that these interactions are not random but rather guided by a specific activation state of the different subsets. Romano and colleagues showed that expanded Tregs polarize in vitro monocytes toward alternatively activated macrophages (24). In the tumor context, we observed that functional Tregs induced monocyte/macrophage activation by upregulating the expression of MHC class II molecules, suggesting that they might favor direct T-cell receptor engagement. In this system, anti-CD25 treatment did not improve the RT efficacy, showing that reducing the duration of the interaction between Tregs and monocytes/macrophages is not sufficient to abrogate their immunosuppressive activity. Whether local CCL2 production favors physical interactions between the two subsets deserves further investigation. Among local cytokines produced following RT, TNFα was previously demonstrated to exert a radioprotective effect, which was proposed to occur via VEGF production by TAMs (31). We further demonstrated that TNFα neutralization enhances tumor response to RT by reducing Treg frequency and TGFβ production. Considering the role of TGFβ in Treg-mediated immunosuppressive activity (32), the activation of this axis is relevant to the radioprotective actions of Tregs. Among these protective actions, we observed that anti-TNFα increased the frequency of IFNγ-producing CD8⁺ T cells after RT. Although CTLA-4 expression was increased in Tregs following RT, its expression was not modified by anti-TNFα, suggesting that TNFα might regulate specific pathways of Treg activation. Several pieces of data have indicated that TNFα induces more specific signaling in Tregs via TNFR2 to maintain their suppressive activity (25, 33, 34). Deciphering the pathway of Treg activation following RT is crucial considering the potential of combined immunotherapies targeting Tregs with RT. Considering the cytotoxic effect of TNFα (35), one could assume that a shortened anti-TNFα treatment might efficiently block Treg accumulation while avoiding inhibition of CD8⁺ T-cell activity. In contrast to most cancers, paradoxical observations have been gathered on the role of Tregs in patients with HNSCC, suggesting that tumor Treg density correlates with increased survival (36). This discrepancy is poorly understood, and some observations suggest that Tregs suppress the induction of a chronic inflammatory state in the tissue (37). However, these hypotheses are still being debated, as other research showed no such correlation (38). Our data clearly indicated that Tregs play a detrimental role in the response to RT, as their depletion increases RT efficacy. Thus, targeting Tregs could be an interesting therapeutic approach to enhance the RT index, as previously suggested [see Liu and colleagues (ref. 39) for a review].

The results of our study offer interesting perspectives. Our data provide evidence for the poorly studied cross-talk between adaptive and innate immune system components and indicate that impacting the monocyte/Treg interaction after RT could increase the antitumor effect of RT. The extent of CCL2 and TNFα production was heterogeneous between tumors following RT, possibly reflecting the differential response to radiation. It would, thus, be interesting to evaluate the prognostic value of these cytokines. As modulators of CCL2 and CCR2 are already being tested in clinical settings, translating the results from this study into the clinic is conceivable. An alternative (and potentially synergistic) approach could be to use anti-TNFα or anti-TGFβ therapeutics to overcome the immunosuppressive effects of Tregs and, thus, potentiate the effects of RT. Our data underline the relevance of using combined strategies to improve the biological efficacy of ionizing radiation-based therapies and limit the toxicities related to dose escalation. Indeed, deintensifying treatments to avoid late toxicity is a foremost goal and an active area of research. The use of RT in combination with CCL2/CCR2 and anti-TNFα modulators represents a promising approach to achieve such goals.

E. Deutsch reports receiving commercial research grants from Boehringer Ingelheim, AstraZeneca/Medimmune, Roche, and Lilly and is a consultant/advisory board member for Merck KGA and Medimmune. No potential conflicts of interest were disclosed by the other authors.

Conception and design: M. Mondini, P.-L. Loyher, E. Deutsch, A. Boissonnas

Development of methodology: M. Mondini, P.-L. Loyher, E. Deutsch, A. Boissonnas

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Mondini, P.-L. Loyher, P. Hamon, M. Gerbé de Thoré, M. Laviron, K. Berthelot, B.L. Salomon, A. Boissonnas

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Mondini, P.-L. Loyher, P. Hamon, M. Gerbé de Thoré, M. Laviron, K. Berthelot, C. Clémenson, E. Deutsch, A. Boissonnas

Writing, review, and/or revision of the manuscript: M. Mondini, P.-L. Loyher, P. Hamon, M. Gerbé de Thoré, M. Laviron, K. Berthelot, C. Clémenson, B.L. Salomon, C. Combadière, E. Deutsch, A. Boissonnas

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Mondini, E. Deutsch, A. Boissonnas

Study supervision: M. Mondini, E. Deutsch, A. Boissonnas

The authors wish to thank the “NAC” animal facility for their assistance with mouse breeding and the Plateforme d'Évaluation Preclinique (PFEP) for providing animal care. P.-L. Loyher was funded by the “Fondation ARC pour la Recherche sur le Cancer.” P. Hamon was funded by the "Ligue contre le cancer." K. Berthelot was funded by “Fondation pour la Recherche Médicale” and “Fondation ARC pour la Recherche sur le Cancer.” This work was supported by funding from the European Community's Seventh Framework Programme (FP7/2007–2013) n°304810—RAIDs (to A. Boissonnas, E. Deutsch, and M. Mondini), Inserm, "Fondation ARC pour la recherche sur le cancer" (Projets Fondation ARC to M. Mondini and A. Boissonnas), and the "Ligue contre le cancer" (to A. Boissonnas).

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.