Dendritic Cell Maturation Defines Immunological Responsiveness of Tumors to Radiation Therapy Free

Radiation therapy is used to treat over half of all cancer patients at some point during the course of their treatment (1, 2). However, the treatment response varies significantly across cancer pathologies, and mechanisms describing why particular cancers respond poorly to radiation are lacking. Traditionally, the efficacy of radiation has been attributed to direct killing of cancer cells following radiation-induced DNA damage (3). Recently, this paradigm has shifted, as studies have demonstrated that radiation can trigger immunogenic cancer cell death capable of igniting tumor-specific immunity (4–6). Treatment with radiation leads to the release of endogenous adjuvants and tumor-associated Ags that can be recognized by the immune system to direct anti-tumor immune responses (7–9). Conversely, it has also been reported that radiation therapy can promote upregulation of molecules that foster immunosuppression following treatment (10–14). Thus, the cumulative integration of these signals within individual tumors likely plays a significant role in determining whether a successful anti-tumor immune response is generated following radiation. A better understanding for how individual tumor microenvironments shape the immune response following radiation is needed to improve patient outcomes following treatment.

Dendritic cells (DCs) are key sentinels of the immune system, capable of processing and presenting Ags, sensing innate danger signals, and integrating microenvironmental cues to regulate whether an adaptive immune response is mounted toward foreign invaders. In particular, conventional type 1 DCs (cDC1s) have the specialized ability to uptake exogenous cell-associated Ags and potently cross-prime Ag-specific CD8⁺ T cell responses (15–18). Cross-presenting cDC1s are defined by their expression of the transcription factors BATF3, ZBTB46, ID2, and IRF8 (19). cDC1 can be further divided into those capable of migrating from tissues (CD103⁺ cDC1) and those resident to lymphoid organs (CD8α⁺ cDC1) (20, 21). CD103⁺ cDC1s are present in many murine tumors and are thought to be the predominant cell type capable of trafficking intact tumor-associated Ags to the draining lymph node (dLN) to initiate cross-priming of tumor-reactive CD8⁺ T cells (22, 23).

In preclinical models, cDC1s are required for the rejection of immunogenic tumors, and they are known to play an important role in promoting anti-tumor immune responses following treatment with many immunotherapies (15, 24–26). Moreover, it has been reported that increased cDC1 signatures in patient tumors correlate with improved outcomes in a range of cancers (23, 27, 28). Activation of intratumoral cDC1 is proposed to support the development of anti-tumor immunity through two key mechanisms: 1) cDC1 migration to the dLN to deliver tumor-associated Ag and initiate priming of tumor-reactive CD8⁺ T cells and 2) cDC1 function within the tumor to recruit and reprime tumor-reactive CD8⁺ T cells locally. The role of cDC1s activation and migration in radiation-mediated tumor regression remains to be determined. In certain tumor models, the efficacy of radiation has been shown to depend on the presence of cDC1s (29, 30). However, these studies used mice that lack cDC1s (Batf3^−/−) throughout the course of tumor development, as opposed to only during therapy, making it difficult to draw conclusions regarding the mechanism. Thus, the question remains whether cDC1 activation and migration is required to successfully promote anti-tumor immune responses following radiation therapy and whether this differs across cancers.

In this study, we investigated mechanisms that regulate why particular cancer types are either highly or poorly responsive to radiation. Using tumor models with equivalent radiosensitivity in vitro but differing responsiveness to radiation in vivo, we demonstrate that poorly radioimmunogenic tumors fail to activate intratumoral cDC1 following treatment. Polyinosinic-polycytidylic acid (Poly I:C) has been shown to successfully combine with radiation therapy to improve tumor control (31). We similarly show that combining radiation with the exogenous adjuvant poly I:C successfully drives cDC1 maturation, resulting in tumor cures. We determine the combined efficacy of radiation and poly I:C is dependent on cDC1s, which promote the development of tumor-specific effector CD8⁺ T cells. Finally, we establish that trafficking of CD8⁺ T cells from lymph nodes (LNs) to the tumor is necessary for treatment efficacy. Taken together, these data demonstrate that intratumoral cDC1 activation and migration following radiation is one potential mechanistic factor that limits the response to radiation therapy across different cancer pathologies.

Experiments used 6–8-wk-old C57BL/6 (no. 000664), B6.SJL (no. 002014), and Zbtb46-DTR (no. 019506) mice that were obtained from The Jackson Laboratory. 2C TCR transgenic mice were kindly provided by Dr. T. Gajewski at the University of Chicago. Survival experiments were performed with five to eight mice per experimental group, and mechanistic experiments were performed with four to six mice per group. Animal protocols were approved by the Earle A. Chiles Research Institute Institutional Animal Care and Use Committee (Animal Welfare Assurance no. A3913-01). The Panc02-SIY pancreatic adenocarcinoma line expressing the model Ag SIYRYYGL (SIY) was kindly provided by Dr. R. Weichselbaum at the University of Chicago. MC38 colorectal carcinoma line was obtained from Dr. K. Young at Earle A. Chiles Research Institute. Moc1 and Moc2 oral squamous cell carcinoma lines were kindly provided by Dr. R. Uppaluri at the Dana Faber Cancer Institute. Panc02-SIY, Moc1, and Moc2 cell lines were grown in complete RPMI 1640 containing 10% heat-inactivated FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin. MC38 cell lines were grown in DMEM containing 10% heat-inactivated FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin. Pathogen and mycoplasma contamination testing were performed on all cell lines within the past 6 mo using the IMPACT II Mouse PCR Profiling from IDEXX BioAnalytics.

Tumor cells lines were treated with the indicated dose of radiation using a cesium irradiator. After treatment, 5 × 10² cells were seeded in a six-well plate and allowed to grow for 5 d. On day 5, media was removed, plates were washed with PBS, and cells were fixed with methanol. The number of tumor cell colonies was counted for each well and normalized by dividing the number of colonies in the untreated well to get the percentage of surviving cells for each dose of radiation.

Tumors were implanted s.c. into the right flank as follows: 2 × 10⁵ MC38, 5 × 10⁶ Panc02-SIY, 1 × 10⁶ Moc1, and 1 × 10⁵ Moc2. When tumors were ∼5 mm in average diameter, mice were randomized to receive treatment with computed tomography (CT)–guided radiation using the Small Animal Radiation Research Platform from Xstrahl. Dosimetry was performed using Murislice software from Xstrahl. The Small Animal Radiation Research Platform delivered a single dose of 12 Gy to an isocenter within the tumor using a 10 × 10 mm collimator and a 45° beam angle to minimize dose delivery to normal tissues. For poly I:C treatments, vaccine grade reagent from InvivoGen (no. vac-pic) was administered intratumorally at 50 μg/tumor in a total volume of 10 μl. Control mice received 10 μl of vehicle. The first dose of poly I:C was administered concurrently with radiation, and the second dose was given 5 d later. For CD8 depletion, 200 μg of anti-CD8β Abs from Bio X Cell (clone 53-5.8) were given i.p. 1 d prior to radiation and again 7 d later. To block T cell egress during treatment, FTY720 from Cayman Chemical (no. 10006292) was administered at 1 mg/kg/d i.p., starting 1 d prior to radiation for a total of seven consecutive days. For Flt3L experiments, the compound was provided by Bristol-Myers Squibb and administered i.p. at 30 μg/mouse/d for nine consecutive days. In all survival experiments, tumor length and width were measured two to three times per week using calipers. Mice were euthanized when tumor size exceeded 12 mm in any dimension or when the body condition score declined one level.

Following dissection, tumors were weighed and minced into small fragments and then transferred into C tubes from Miltenyi Biotec containing enzyme digest mixture (mix) with 250 U/ml collagenase IV (no. LS004188; Worthington Biochemical), 30 U/ml DNase I (no. 4536282001; MilliporeSigma), 5 mM CaCl₂, 5% heat-inactivated FBS, and HBSS. Tissue was dissociated using a gentleMACS tissue dissociator from Miltenyi Biotec. This was followed by incubation at 37°C for 30 min with agitation. For the dLNs, capsules were cut open and incubated with enzymatic mix described above at 37°C for 15 min with agitation. Enzyme mix containing dLNs was then vigorously pipette mixed and incubated at 37°C for an additional 15 min. Enzymatic reactions for both the tumor and dLN were quenched using ice-cold RPMI containing 10% FBS and 2 mM EDTA. Single cell suspensions were then filtered through 100-μm (tumor) or 40-μm (dLN) nylon cell strainers to remove macroscopic debris. Cells were washed and counted as described above.

For staining, 2 × 10⁶ cells were stained with Zombie Aqua Viability Dye from BioLegend (no. 423102) in PBS for 10 min on ice, and then Fc receptors were blocked with anti-CD16/CD32 Abs from BD Biosciences (2.4G2) for an additional 10 min. After centrifugation, the supernatant was removed, and cells were stained with a surface Ab mix containing FACS buffer (PBS, 2 mM EDTA, 2% FBS) and Brilliant Stain Buffer Plus from BD Biosciences (no. 566385) for 20 min on ice. The following Abs were purchased from BioLegend: F4/80-PerCP/Cy5.5 (BM8), CD11c-PE/Cy7 (N418), CCR7-PE (4B12), CD90.2-A700 (30-H12), CD19-A700 (6D5), MHC class II (MHC-II)–BV421 (M5/114.14.2), CD11b-BV605 (M1/70), CD8α-BV650 (53-6.7), Ly-6C-BV711 (HK1.4), and IL-12 PE (C15.6). CD40-FITC (HM40-3), CD103–allophycocyanin (2E9), CD24–allophycocyanin e780 (M1/69), and Granzyme B eFluor450 (NGZB) were obtained from Thermo Fisher Scientific. CD80-PE CF594 (16-10A1), CD45-BV786 (30-F11), and Ki-67 FITC (B56) were purchased from BD Biosciences. PE-conjugated Kb-SIY pentamers (no. F1803-2B) were purchased from ProImmune. After surface staining, cells were washed in FACS buffer and fixed for 20 min on ice with Fixation/Permeabilization Buffer from BD Biosciences (no. 554722). For intracellular and intranuclear cytokine analysis, single cell suspensions from tumors were incubated in complete RPMI ± 50 μg/ml poly I:C and 10 μg/ml GolgiPlug from BD Biosciences (no. 555029) at 37°C for 6 h. Cells were then stained as described above, except fixation and permeabilization was performed using the Foxp3/Transcription Factor Staining Buffer Set from Thermo Fisher Scientific (no. 00-5523-00), and then cells were incubated with intracellular Abs for 30 min on ice. All samples were resuspended in FACS buffer and acquired on a BD Fortessa flow cytometer. Data were analyzed using FlowJo software from Tree Star v10.5. cDC1 were gated as leukocytes/single cells/Live/CD45⁺/CD90.2⁻CD19⁻/Ly-6C⁻/MHC-II⁺/CD24⁺F4-80⁻/CD11b⁻/CD103⁺. CD8⁺ T cells were gated as single cells/Live/CD45⁺/CD90.2⁺CD19⁻/CD8⁺CD4⁻.

Bone marrow chimeras were generated using B6.SJL (CD45.1⁺) recipient mice that were irradiated with 1000 rads. Bone marrow cells were isolated from wild-type (WT) C57BL/6 (CD45.2⁺) or Zbtb46-DTR (CD45.2⁺) donor mice femurs and tibias using a 27G needle. Cells were filtered through a 70-μm cell strainer to generate a single cell suspension and resuspended in PBS. Recipient mice received 3–5 × 10⁶ donor bone marrow cells by retro-orbital injection. Tumors were implanted 8 wk following bone marrow reconstitution. Diphtheria toxin from MilliporeSigma (no. D0564) was administered 3 d prior to radiation at 20 ng/g i.p. for initial DC depletion. This was followed by an additional three doses of 5 ng/g of diphtheria toxin that were given every 3 d to maintain depletion.

Tumors were harvested on ice, weighed, and homogenized in PBS containing 4.5 μl Halt Protease Inhibitor Cocktail from Thermo Fisher Scientific (no. 78440) per mg tissue. The cell debris was removed by centrifugation at 14,000 × g for 15 min at 4°C, and supernatants were stored in aliquots at −80°C until analyzed. Cytokines and chemokines were detected using 25 μl of supernatant and the Cytokine & Chemokine 26-Plex Mouse ProcartaPlex Panel 1 kit from Life Technologies (no. EPX260-26088-901). Data were acquired on a Luminex 100 array reader, and cytokine/chemokine concentrations for each tumor sample was calculated using standard curves for each analyte.

Data were analyzed and graphed using Prism from GraphPad Software (v7.0). Individual data sets were compared using Student t test, and analysis across multiple groups was performed using ANOVA with individual groups assessed using Tukey comparison. Kaplan–Meier survival curves were compared using a log-rank test.

First, we set out to identify murine tumor models with equivalent radiosensitivity in vitro but differing responsiveness to the same dose of radiation in vivo. We compared the radiosensitivity of the murine colon tumor cell line MC38 and the pancreatic tumor cell line Panc02-SIY. In vitro, both tumor cell lines had comparable sensitivity to a range of radiation doses (Fig. 1A), These cell lines were then used to establish syngeneic flank tumors in mice and further evaluate their response to radiation in vivo. When tumors reached an average diameter of 5 mm, they were treated with CT-guided radiation to prevent indirect targeting of the tumor dLN (Fig. 1Bi, ii). Both tumor types showed delayed tumor growth kinetics in response to radiation as compared with untreated controls. Despite displaying equivalent radiosensitivity to Panc02-SIY in vitro, MC38 tumors exhibited considerable tumor regression and, in some instances, tumor cures (Fig. 1Ci). We also tested the head and neck tumor cell lines Moc1 and Moc2, which had comparable radiosensitivity in vitro but differing responsiveness in vivo (Supplemental Fig. 1A, 1B). Taken together, these data indicate that tumor cell–intrinsic radiosensitivity is not the limiting factor controlling the response to radiation in vivo in these tumor models. To determine if the improved tumor control in MC38 tumors following radiation was dependent on the adaptive immune response, we depleted CD8⁺ T cells prior to treatment and found that CD8⁺ T cell depletion significantly abrogated the enhanced survival benefit of radiation in MC38 tumors but had no impact on Panc02-SIY (Fig. 1Cii, Supplemental Fig. 1B). We observed similar results in Moc1 tumors, which required CD8⁺ T cells for their enhanced response to radiation, whereas Moc2 tumors did not require CD8⁺ T cells (Supplemental Fig. 1B). Given that MC38 and Moc1 tumors exhibited a CD8⁺ T cell–dependent survival advantage in response to radiotherapy, we will refer to them as radioimmunogenic tumors from this point forward, whereas Panc02-SIY and Moc2 will be referred to as poorly radioimmunogenic tumors.

In radioimmunogenic MC38 tumors, improved tumor control following radiation therapy required CD8⁺ T cells, suggesting a potential failure to generate an effective anti-tumor CD8⁺ T cell response in poorly radioimmunogenic Panc02-SIY tumors. Because cDC1 are known to play an important role in cross-priming CD8⁺ T cell responses, this led us to evaluate whether cDC1 were being activated equivalently in both tumor models following radiation (15). We used flow cytometry to assess changes in both the quantity and maturation state of DC subsets within the tumor after treatment with a range of radiation doses (Fig. 2A, Supplemental Fig. 2A). There was a significant reduction in total DCs, particularly within the CD103⁺ cDC1 compartment following radiation in both tumor models (Fig. 2Bi, ii). Interestingly, the remaining intratumoral cDC1s in MC38 tumors expressed higher levels of markers associated with DC maturation, including CCR7, which is important for migration to the dLN (Fig. 2Ci) and the costimulatory molecule CD80 (Fig. 2Ci, ii) (32). Moreover, expression of these activation markers increased in a dose-dependent manner with higher doses of radiation (Fig. 2Ci, ii). Similarly, there was a trend toward increased intratumoral cDC1 activation following 12 Gy of radiation in the radioimmunogenic Moc1 tumors but not in the poorly radioimmunogenic Moc2 tumors (Supplemental Fig. 1Ci–iii). To determine whether increased accumulations of intratumoral cDC1s could improve the efficacy of radiation in poorly radioimmunogenic Panc02-SIY tumors, we administered the cytokine Fms-like tyrosine kinase 3 ligand in combination with radiation (Supplemental Fig. 3A) (33). Treatment with Fms-like tyrosine kinase 3 ligand significantly increased the accumulation of intratumoral cDC1s, but DC maturation was still impaired (Supplemental Fig. 3Bi, ii), and treatment had no impact on animal survival following radiation (Supplemental Fig. 3Ci, ii). Thus, although radiation is clearly capable of generating signals to promote cDC1 maturation in particular tumor types, these signals are either lacking or actively suppressed in poorly radioimmunogenic tumors, leading to impaired tumor control after radiation. Importantly, these results provide one potential explanation for why equivalent doses of radiation are capable of inducing varying degrees of tumor regression across different tumor types.

Our results thus far had suggested that radiation alone is unable to drive cDC1 activation in poorly radioimmunogenic tumors, and this failure may limit the extent of tumor control following radiation. We hypothesized that externally driving DC maturation by administration of adjuvants directly to the tumor would restore T cell–mediated tumor control. To identify an optimal adjuvant, we examined TLR expression on DCs and found TLR3 expression to be highly enriched on cross-presenting cDC1s (Supplemental Fig. 2B–D). Importantly, signaling through this innate receptor has been shown to induce cDC1 maturation (34, 35). Previous work has demonstrated improved tumor control in murine models when radiation is combined with poly I:C, suggesting that this agent may restore cDC1 function in tumors (31, 36, 37). We administered intratumoral poly I:C concurrently with radiation and then again 5 d later and assessed tumors for cytokine responses and DC maturation (Fig. 3A). Analysis of cytokines in tumors revealed increased levels of type I IFN (IFN-α), proinflammatory cytokines (TNF-α, IL-6, and IL-1β), and chemokines known to recruit T cells (CCL5 and CXCL10) in both single agent poly I:C or the combination of radiation and poly I:C–treated tumors (Fig. 3B). Thus, treatment with poly I:C transforms the milieu within the tumor into an environment that is more favorable for the development of anti-tumor immunity in the context of radiation therapy.

Earlier data indicated that radiation effectively induced cDC1 maturation only in radioimmunogenic tumors (MC38 and Moc1), and this process did not occur in poorly radioimmunogenic tumors (Panc02-SIY and Moc2). To address whether poly I:C was able to induce cDC1 maturation in poorly radioimmunogenic Panc02-SIY tumors, we used flow cytometry to monitor changes in the quantity and activation state of cDC1s within the tumor. Our analysis revealed that all treatment groups had fewer intratumoral cDC1s as compared with untreated controls 1 d following treatment (Fig. 3Ci). However, of the cDC1s that remained in the tumor, we noted an increased expression of markers associated with DC maturation and migration (CCR7, CD40, and CD80) when poly I:C was given alone or in combination with radiation (Fig. 3Cii–iv). Treatment with poly I:C significantly increased the production of IL-12 specifically in intratumoral cDC1s (Fig. 3D), a cytokine associated with enhanced DC priming (35). Interestingly, although single agent poly I:C induced changes in cDC1 maturation and generated a favorable cytokine environment within tumors, it failed to impact tumor growth, whereas the combination of radiation and poly I:C resulted in tumor regression (Fig. 3Ei). Unlike earlier studies, our dosing regimen also resulted in durable tumor cures (Fig. 3Eii) (31). These data demonstrate that in tumor models in which cDC1 maturation is impaired either because of active suppression or a failure for radiotherapy to release sufficient signals, we can overcome this deficit by administering exogenous adjuvants to promote cDC1 maturation following radiation therapy, and this leads to durable tumor cures. Importantly, these results suggest that adjuvant signal in the form of poly I:C alone is insufficient to induce tumor cures.

Because macrophages in the tumor express some TLR3 (Supplemental Fig. 2C) and tumor-associated macrophages can impact tumor control following radiation therapy (38), we evaluated the importance of tumor macrophages to the treatment response. We found that macrophage depletion using anti-CSF1 did not significantly impact tumor control by the combination of radiation therapy and poly I:C (Supplemental Fig. 4Ai, ii), suggesting that cDC1s may be the critical target for TLR3 ligands. Although cDC1s were successfully activated by poly I:C when combined with radiation, the question remained whether these cells were required for treatment efficacy. One widely used approach to deplete cDC1 in murine models are Batf3^−/− mice; however, these mice lack DC through all stages of tumor development, which changes the baseline tumor immune environment prior to treatment initiation (15). To isolate the effect of treatment on DC populations, we required an approach to selectively deplete conventional DCs (cDCs) at the time of treatment. Zbtb46-DTR mice express the diphtheria toxin receptor selectively in cDCs and permits their depletion at any time point by the administration of diphtheria toxin (39). To deplete cDCs, we established Panc02-SIY tumors in Zbtb46-DTR or WT C57BL/6J bone marrow chimeras and treated them with diphtheria toxin 3 d prior to treatment with radiation and poly I:C (Fig. 4Ai). Treatment with diphtheria toxin resulted in a loss of cross-presenting DCs in both the tumor (Supplemental Fig. 4B) and in the tumor dLN (Supplemental Fig. 4Ci, ii) of Zbtb46-DTR bone marrow chimeras but not in WT control bone marrow chimeras. Depletion of cDCs immediately prior to radiation significantly impaired tumor control and abrogated the enhanced survival benefit of radiation and poly I:C when compared with control WT bone marrow chimeras treated with diphtheria toxin (Fig. 4Aii, iii). Notably, in bone marrow chimeras given the combination of poly I:C and radiation therapy without DC depletion, the overall efficacy of treatment was consistently reduced compared with that observed in WT mice (Fig. 3), suggesting some general loss of immune function through the development of bone marrow chimeras. Although cDC1s were clearly important for the efficacy of combination therapy, the mechanism by which they promoted tumor regression remained unclear. To determine whether DC migration was important for therapy, we first quantified the total number of migratory CD103⁺ cDC1s in the tumor dLN following treatment. The data revealed more migratory CD103⁺ cDC1s with an activated phenotype (CD80) in the dLNs of combination-treated animals as compared with untreated or single agent controls (Fig. 4Bi, ii), suggesting increased migration following treatment. These data demonstrate that cDC1s play an important role in the anti-tumor efficacy of radiation and poly I:C.

Our data thus far suggested that CD103⁺ cDC1 migration to the dLN is increased following combination therapy. Although Ag recognition serves as signal 1 for T cell priming, DCs are known to provide additional signals in the form of costimulation (signal 2) and cytokines (signal 3) that further promote the expansion and quality of Ag-specific T cells (40). This led us to first evaluate whether CD8⁺ T cells were required for the combined efficacy of radiation and poly I:C by depleting CD8⁺ T cells (Fig. 5Ai, Supplemental Fig. 4D). Depletion of CD8⁺ T cells completely abolished the efficacy of treatment, indicating that these cells were indeed important for treatment (Fig. 5Aii, iii). Next, we used flow cytometry to assess the phenotype of CD8⁺ T cells in the tumor 7 d after treatment. Although radiation alone increased the number of CD8⁺ T cells in tumors compared with all other treatment groups, this was not the case in the combination of treated animals (Fig. 5Bi). Instead, the combination of radiation and poly I:C significantly expanded the proportion of proliferating (Ki67⁺) CD8⁺ T cells in the tumor with enhanced cytotoxic potential as identified by the protease granzyme B (Fig. 5Bii). This pattern was also observed in Ag-specific 2C CD8⁺ T cells, which recognize the SIY peptide expressed by Panc02-SIY tumor cells (Fig. 5Biii). Moreover, we observed a similar increase in CD8⁺ Ki67⁺ Granzyme B⁺ cells following radiation alone in radioimmunogenic MC38 tumors (Supplemental Fig. 4E). These data suggest that following radiation and poly I:C, cDC1s prime CD8⁺ T cells that have improved cytolytic potential as compared with controls. The question then remained whether these T cells were being activated within the tumor or were instead being primed by cDC1 within the dLN. To address this question, we used S1PR agonist FTY720 to sequester T cells in the LN, thereby preventing their migration to the tumor following priming in the dLN (Fig. 5Ci, Supplemental Fig. 4D) (41). When T cell egress from the LNs was impaired with FTY720, the combined efficacy of radiation and poly I:C was completely abrogated (Fig. 5Cii, iii). Taken together, these results demonstrate that tumor regression following treatment radiation and poly I:C is dependent on cDC1s, which play an important role in generating tumor-reactive effector CD8⁺ T cells within the tumor dLN, and these T cells must be free to migrate through the circulation to the treatment site to result in a tumor cure.

The treatment response to radiation is highly variable across different cancer pathologies. Although radiation is capable of directly killing tumor cells, this is not the sole mechanism responsible for tumor shrinkage following treatment (7). Our studies confirm that tumor cell–intrinsic radiosensitivity in vitro is a poor predictor for the overall response to radiation in vivo and instead implicates other mechanisms. Given that radiation has been shown to elicit tumor-specific adaptive immune responses, we investigated immune-related mechanisms that might explain this variable response across cancer pathologies (8, 42). Our findings demonstrate that when a range of tumor types were treated with equivalent doses of radiation in vivo, improved treatment responses were dependent on the presence of CD8⁺ T cells only in radioimmunogenic tumors (MC38 and Moc1) and independent of tumor cell–intrinsic radiosensitivity. These data highlight the importance of generating a productive tumor-specific adaptive immune response following radiation and provide useful insight into the potential immune-related mechanisms that explain the differential response to radiation across different cancers.

cDC1s are a critical cross-presenting cell type capable of linking the innate and adaptive immune system (15). We discovered that intratumoral cDC1 activation following radiation is not uniform across different tumor types. Instead, radiation induces cDC1 maturation only in particular tumor types (MC38 and Moc1) that correspond with the tumor types reliant on CD8⁺ T cells for an improved response to radiation. These data suggest that cDC1 maturation fails to occur in poorly radioimmunogenic tumors, either because of active suppression or the absence of adequate signals following radiation therapy. Ultimately, this failure results in the impaired generation of tumor-specific CD8⁺ T cell responses and limits the extent of tumor control following radiation. Although we did see a modest increase in the DC-suppressive cytokine IL-10 following radiation in the poorly radioimmunogenic Panc02-SIY tumors (43), each tumor type may have its own unique pathways or cell types potentially responsible for DC suppression following radiation. These could include other cytokines or metabolites such as PGE₂ or IDO that are increased following radiation and function to suppress intratumoral cDC1 activation (28). Additional studies are needed to identify the specific factors and signaling pathways within various tumors that prevent cDC1 maturation after treatment to improve responses to radiation.

Previous studies have demonstrated that bone marrow–derived DCs injected into irradiated tumors can take up Ags and cross-present in the dLNs but have a limited ability to recruit activated T cells back to the irradiated site (44). Similarly, Jahns et al. (45) demonstrated that radiation of monocyte-derived DCs in vitro did not directly cause DC maturation but also did not prevent their maturation following exposure to appropriate stimuli. One approach to overcome the failure of radiation to induce intratumoral cDC1 activation is to provide exogenous adjuvants that drive DC maturation. In this study, we used the adjuvant poly I:C to target the innate receptor TLR3, which is highly expressed by cDC1s (35). Yoshida et al. (31) previously demonstrated that poly I:C in combination with radiation improved tumor control, resulting in DC activation in the tumor dLN. We similarly demonstrate that concurrent administration of poly I:C and radiation with a second dose of poly I:C given 5 d later successfully drives intratumoral cDC1 maturation in poorly radioimmunogenic Panc02-SIY tumors. Importantly, this treatment combination leads to durable tumor cures that are dependent on cDCs. The prior reports have suggested that giving poly I:C 1 d prior to radiation can temporarily delay tumor growth, but treatment ultimately fails to cure tumors (31). Timing adjuvant delivery with radiation-mediated tumor cell death is likely critical in coordinating the release of tumor-associated Ags with the adjuvant signals that function to promote DC maturation.

Our data suggest that although radiation alone is capable of generating signals that promote cDC1 maturation in radioimmunogenic tumors, these signals are either absent or suppressed in poorly radioimmunogenic tumors. We have previously demonstrated that macrophages suppress T cell control of tumors following radiation therapy (11, 46), and others have shown they can secrete factors such as IL-10 that suppress DC maturation in tumors (43). In addition, other cell populations present in the tumor environment can alter patterns of DC maturation following radiation therapy (47), suggesting that the immune milieu may regulate the ability of DCs to mature. In poorly radioimmunogenic tumors, a bolus of innate adjuvant was sufficient to provide the missing signal or overcome suppressive mechanisms. In our studies in poorly radioimmunogenic tumors, we provided this signal in the form of poly I:C, which was selected based on the enriched expression of its rTLR3 in cDC1s, but other innate adjuvants that activate DC maturation have also shown synergy with radiation therapy (48–50). Although we see no evidence of other cells contributing to cDC1 maturation following TLR3 ligation, this possibility has not been excluded. Although TLR3 is expressed by cDC1 and necessary for their activation by poly I:C, cDC1 maturation to full Ag presenting and processing capacity following TLR3 ligation is dependent on their production and response to type I IFN (34, 35, 51). Thus, TLR3 ligation likely causes additional positive proinflammatory effects in the tumor environment secondary to TLR3 ligation in DCs. Together, these data indicate that the presence of immunological adjuvant in the tumor and the capability of DCs to respond to these released adjuvants are critical determinants for the success of radiation therapy.

A long-standing question within the field of radiation therapy is whether treatment can lead to the development of new tumor-reactive CD8⁺ T cell responses and essentially function as an endogenous cancer vaccine. In this study, we provide evidence that radiation fails to drive intratumoral cDC1 maturation in poorly radioimmunogenic tumors, one of the first steps in developing a productive anti-tumor CD8⁺ T cell response. However, by combining radiation with poly I:C, we overcome this barrier and demonstrate that when T cells have been sequestered in the LNs during treatment, tumors fail to cure. DC maturation through signals such as TLR3 ligation results in a decreased phagocytosis and a shift to a migratory and Ag presentation phenotype via expression of markers such as CCR7 and CD80, respectively (52, 53). Our data suggests that in poorly radioimmunogenic tumors, DCs are actively phagocytosing material from irradiated cancer cells but fail to receive the signals that allow them to mature. In radioimmunogenic tumors or in poorly radioimmunogenic tumors given adjuvants, these cells complete their cycle and travel to the dLN to prime T cells (52, 53). These data suggest that in these circumstances, combination therapy is generating new CD8⁺ T cell responses within the dLN and indicate that under optimal conditions, radiation therapy can function as an endogenous cancer vaccine. Importantly, this work also demonstrates the importance of selecting diverse tumor models to evaluate treatments. The nonresponsive tumors may provide the greatest source of information to understand how treatments succeed and critically guide novel interventions to help patient populations who currently do not respond to treatment.

In patients, CD8⁺ T cell infiltration within tumors tends to correlate with improved outcomes across a range of malignancies (54–56). Even in the absence of radiation, recent studies have demonstrated that the presence of DCs within tumors is highly impactful to the success of other therapies (25, 57). We propose that patients with a poor immune environment are similar to our poorly responsive murine models, whereby radiation therapy fails to drive DC maturation either because of the absence of adjuvant signals or by active suppression within the tumor microenvironment. In these patients, radiation would be unable to generate high-quality, tumor-reactive T cell responses despite the release of tumor Ags that have the potential to be recognized by the immune system. Thus, these unresponsive patients may benefit from the addition of adjuvants that enable radiation therapy to fully function as an endogenous cancer vaccine by driving cDC1 maturation and effective cross-presentation of tumor Ags to CD8⁺ T cells. We believe that by combining radiation therapy with adjuvants that target these deficiencies, we can restart the cycle of immunity and convert otherwise dismal radiation responses into more favorable outcomes.

The authors have no financial conflicts of interest.