Impact of proton therapy on antitumor immune response

Radiotherapy delivered using photons induces an immune response that leads to modulation of the tumor microenvironment. Clinical studies are ongoing to evaluate immune checkpoint inhibitors in association with photon radiotherapy. At present, there is no publication on the radio-induced immune response after proton therapy. Balb/c mice bearing subcutaneous CT26 colon tumors were irradiated by a single fraction of 16.4 Gy using a proton beam extracted from a TR24 cyclotron. RNA sequencing analysis was assessed at 3 days post-treatment. Proton therapy immune response was monitored by flow cytometry using several panels (lymphoid, myeloid cells, lymphoid cytokines) at 7 and 14 days post-irradiation. RNA-Seq functional profiling identified a large number of GO categories linked to “immune response” and “interferon signaling”. Immunomonitoring evaluation showed induced tumor infiltration by immune cells. This is the first study showing the effect of proton therapy on immune response. These interesting results provide a sound basis to assess the efficacy of a combination of proton therapy and immune checkpoint inhibitors.

www.nature.com/scientificreports/ and MHC-class 1, has been analyzed following proton radiation. Proton radiation of several types of cancer cells upregulated the expression of these markers of immunogenic modulation, with degrees of upregulation similar to those observed after equivalent exposure to photon radiation 12,13 . Recently, Spina et al., evaluated the effects of carbon-ion therapy on immune modulation. They highlighted an interesting induction of pro-inflammatory cytokines 14 . Yoshimoto et al. evaluated the immunogenic alterations induced by carbon ion irradiation in vitro. They described that carbon ion radiation increased the secretion of high mobility group box 1 (HMGB1) in human cancer cell lines. HMGB1 is an ICD marker, playing an important role in antigen-presenting cell activation and induction of immune response 15 . The levels of HMGB1 induced by carbon ion exposure were similar with equivalent doses of photon irradiation. Although the number of patients treated with protons is clearly higher than with carbon ion, preclinical studies with particles have focused on the impact of carbon ion irradiation on immune response. Preclinical in vivo studies using carbon ion radiation have convincingly demonstrated that this type of particle induces antitumor immune response in immunocompetent animals 16 . Proton therapy (PT) plays an important role in clinical radiotherapy with growing facilities and indications 17 . It has particularly interesting ballistic advantages defined by the Bragg peak, beyond which the dose delivered is almost nil, enabling the total avoidance of surrounding organs at risk. Protons have an increased Relative Biological Effectiveness (RBE) compared to photons, and they cause a different type of damage to DNA than do photons 18 . Moreover, densely ionizing radiation may have other biological advantages induced by different cell death pathways and release of pro-inflammatory cytokines 19 .
Recently, Durante and Formenti proposed that, combined with immunotherapy, particle radiation could be more effective than photon radiation, as protons and heavy ions displayed physical advantages and led to reduced damage to blood lymphocytes that are required for an effective anti-tumor immune response 20 . However, the effect of proton on the intratumor immune response remains currently unexplored.
In this study, we aimed to evaluate the radio-induced immune response with a 16.4 Gy single fraction of proton. We evaluated this response by transcriptomic analysis in order to describe the immune molecular pathways modulated by proton therapy, and by analysis of the tumor microenvironment, by immunomonitoring of intratumoral infiltrated immune cells.

Results
Effect of 16.4 Gy with proton therapy on CT26 tumor growth. A single dose of 16.4 Gy proton therapy was delivered to the CT26 tumors of immunocompetent BALB/c mice. The dose of 16.4 Gy delivered with proton therapy induced enhanced CT26 tumor control compared to the non-irradiated (NI) control tumors (Fig. 1A,B). We highlighted a significant tumor growth delay for the proton therapy group, compared to NI control (P < 0.0001), with median survivals of 75 versus 24 days, respectively. For 6 mice out of 10 treated by 16.4 Gy proton therapy, the maximum tumor volume limit of 1500 mm 3 was reached between 54 and 84 days, while the 4 remaining mice still presented complete response after 100 days.
Proton therapy activated transcriptomic antitumor immune response pathways. Gene transcript analysis was compared between 16.4 Gy and non-irradiated CT26 tumors, 3 days post-irradiation to examine immune response after proton therapy. RNA-Seq profiling identified 68 genes differentially expressed with s-value < 0.005 and a Fold Change ≥ 2 (Table S1).
Enrichment analysis was then performed on the gene set (Table S2), then we focused on pathways linked to "immune response" and "interferon signaling". The enriched GO categories listed in Table 1 represent all the pathways identified that included "immune response" and "interferon signaling".
Next, we extracted all genes involved in the "immune response" pathway obtained with the enrichment (p value = 9.6e−09). This pathway was covered by 25 differentially expressed genes, which are presented in a heatmap below (Fig. 2). All the genes observed presented up-regulation after 16.4 Gy proton therapy, compared to NI www.nature.com/scientificreports/ controls. Among these genes, we identified several genes of Ifit (Interferon-induced protein with tetratricopeptid repeats) and Ifi (Interferon inducible protein) families, which are involved in Interferon alpha and beta signaling pathways. Many induced genes are involved in the type I interferon pathway. Therefore, we looked up for the expression of genes of interest relative to Interferon Type 1 pathway in our RNASeq profiling, Ifnar2, Cxcl10 and Trex1 tended to be over-expressed after 16.4 Gy of proton therapy, but with s-values > 0.005 (data not shown). As Cxcl10 and Trex1 presented Fold Change > 2, we decided to quantify their www.nature.com/scientificreports/ variation of expression using RTqPCR. Both Cxcl10 and Trex1 displayed a significant increase in their relative expression (p < 0.05), validating the results obtained by RNAseq analysis on those genes (Fig. 3).

Proton therapy induced intra-tumor immune cell infiltration. Proton therapy induced infiltration
of two types of immune cells: antitumor and pro-tumor immune cells (Fig. 4). Seven days after PT, significant infiltration of CD8 + T cells, CD4 + T cells and type 1 tumor associated-macrophage (TAM1) was observed. For these cells, infiltration tended to be maintained 14 days after PT, albeit without reaching statistical significance. We did not observe radio-induced infiltration of NK cells in these conditions. CD8 + T cells presented a cytotoxic activity phenotype, with co-expression of Granzyme B in more than 80% of them. Concerning immunosuppressive cells, we observed significant tumor infiltration of Treg, but no significant variation in myeloid derived suppressor cells (Mo-MDCS) and type 2 tumor associated-macrophage (TAM2) infiltrations, 7 and 14 days after PT (Fig. 4). All significant infiltrations were transient, with loss of significance of these infiltrates compared to controls at D14.

Discussion
Proton therapy is an attractive therapeutic approach. It has both physical advantages, limiting the dose to surrounding organs at risk, and biological advantages, due to high linear energy transfer with different types of DNA damage induced by the densely ionizing radiation 22 . Several teams have described the potential effects of protons on the antitumor immune response, illustrating the importance of this issue, but also the difficulty of implementing experiments to demonstrate it 16,20,23,24 . Devices amenable to carrying out such studies in vivo are rare, and offer limited access for research studies. For example, in France, only few machines are available, and only a few slots per month are opened for research. A 16.4 Gy unique dose of proton therapy significantly enhanced the time to reach the tumor volume limit, with 4 mice out of 10 with complete response 100 days post-irradiation. This model was never used with proton therapy. Nevertheless, the data obtained by Grapin et al. on the same colon carcinoma tumor model, after 16.4 Gy X-rays, presented increased growth delays without any complete response compared to our results 10 . In addition, this single dose of photon irradiation was compared to two fractionation schedules, with the same Biologically Effective Dose (BED), 3 × 8 Gy and 18 × 2 Gy ( Figure S1). Whatever the fractionation with photon irradiation, high single dose, hypofractionation or "classical" fractionation, our single dose of proton therapy increased more tumor growth delay.
This present study highlights that proton therapy enables the activation of several pathways implicated in the immune response. Activation of innate and specific immune responses was previously described to be induced by type I interferon (IFN-I), with activation of antigen recognition and cytolytic activity 25,26 . One of the critical signals for the spontaneous activation of antitumor T cells against immunogenic tumors, as well as RT-induced T-cell priming, is double strand DNA (dsDNA) quantity 27 . Cytosolic DNA derived dsDNA is sensed by cyclic GMP-AMP synthase (cGAS) to generate cGAMP required for the activation of the adaptor STING (stimulator of interferon genes), resulting in the induction of several interferon-stimulated genes such as Cxcl10 8,9,26,[28][29][30] . Cxcl10 served as a chemotactic gradient to recruit T cells, expressing CXCR3, inside tumor 31 .
After conventional RT, several papers demonstrated that photon RT induced cytosolic DNA accumulation, which activates the cGAS/STING pathway resulting in activation of IFN-I 9,32,33 . In this same publication, authors also reported that doses at 8 and 10 Gy per fraction led to Trex1 expression, which degrades cytosolic DNA, thus avoiding IFN-I activation 2 .
Harding et al. showed that radio-induced activation of type I IFN signalling was preceded by micronuclei formation 34 . The formation of micronuclei is one of the mechanisms responsible for the presence of cytosolic DNA. DNA double-strand breaks are the most lethal lesions induced by ionizing radiation and can trigger a series of cellular DNA damage responses (DDRs). Recently, a review described the current evidence linking the DNA damage responses to activation of the immune response through micronuclei formation 35 . It has been demonstrated that proton therapy induced more DNA double-strand breaks and micronuclei than photon irradiation at equivalent doses (2-12 Gy) 36 . The IFN-I pathway was induced with a single dose of 16.4 Gy proton therapy in our study (ie. up-regulation of Ifi and Ifit genes in RNASeq). These results suggest that the proton-induced mechanisms could be comparable to those induced by conventional photon radiotherapy.
With a high dose per fraction of photon, Vanpouille-Box et al. described an increase in Trex1 expression as well as a decrease in the amount of cytosolic DNA and a decrease in the activation of cGAS/STING in TSA cells, highlighted by a Cxcl10 expression decrease 9 (supplementary data). With our gene transcript analysis, we also observed an increasing expression of Trex1 in our group irradiated with 16.4 Gy proton therapy. However, we observed Cxcl10 gene expression induction. Other authors have previously described an induction of Cxcl10 expression after hypofractionated photon therapy administrated by 2 × 12 Gy using B16 cancer model 37 . It would be interesting to also evaluate in vitro the effect of protons on the amount of cytosolic DNA.
As described in the review of Zhang et al. 8 , two DNA sensing pathways are known to induce the production of type I IFNs 38 : the first one is the cytosolic DNA which activates cGAS/STING pathway, as described above, and the second one is the endosomal DNA which is detected by TLR9. Then TLR9 detects endosomal DNA and activates IRF7 29 . In our transcriptomic analysis we highlighted a significant up regulation of Tlr9 and Irf7 (Fig. 2). Thus, if the induction of TREX1 expression led to inhibition of the cytosolic DNA pathway, the endosomal pathway could be an alternative for high dose PT to induce the type I interferon pathway.
Moreover, a single fraction of 16.4 Gy proton therapy induced a significant antitumor response in our cancer model, with tumor infiltration of different types of immune cells, such as CD8 + T cells, which express granzyme B + , translating cytotoxic activity. This is related to the induction of expression of certain chemokines observed in RNA-Seq (such as Cxcl11), which contribute to the recruitment of effector T cells in the tumor 30 . CD8 + T cell activation is one of the mechanisms involved in the induction of the abscopal effect, which is a rare systemic effect first described more than 60 years ago after conventional radiotherapy 39,40 . Brenneman et al., recently described the first case report of an abscopal effect after proton therapy in a patient with sarcoma 41 . The incidence of this abscopal effect is increased when photon RT is combined with immunotherapy 42 . The association of proton therapy with immunotherapy could therefore yield a beneficial effect both locally and outside of the radiation field in metastatic patients. To compare the percentages of antitumor and protumor immune cells obtained after 16.4 Gy proton and photon X irradiations (as previously published 10 ) on our colon carcinoma tumor model, the different panels evaluated were gathered in Supplementary Fig. 2. In antitumor immune cells, no difference was induced, except for NK cells which were enhanced after RT, 7 days post-irradiation, but this difference faded after 14 days. For protumor immune cells, no differences were observed for Mo-MDSC and TAM2 cells. Treg cells maintained their percentages of infiltration cells after PT between 7 and 14 days, in contrary to cells treated with RT that diminished 14 days post-irradiation, but the percentages were very low. As infiltration and immune response of PT remain similar to those presented after RT, the difference of PT efficacy may result from DNA www.nature.com/scientificreports/ damage and not from immune response. A comprehensive study of cytotoxic activity of lymphocytes T could confirm these findings.
Combinations of immunotherapy with conventional radiotherapy are increasingly being evaluated in preclinical and clinical situations 43,44 . Some studies have attempted to optimize the combination conditions to achieve a radio-induced immune response, leading to an in situ vaccine, which may be amplified with immunotherapy by inhibiting tumor microenvironment immunosuppression 45 .
Yet, no published preclinical study has described the effectiveness of combining proton therapy with immunotherapy. We searched for the query string "proton" and "immunotherapy" in the ClinicalTrials.gov database in December 2020, and identified only 4 verified trials worldwide evaluating the safety and/or efficacy of a combination of PT and immunotherapy (NCT03765190; NCT03818776; NCT03267836 and NCT03764787) that are currently ongoing or not yet recruiting. They are all early-stage studies (phase I or II) with small sample sizes (maximum 30 patients). All these trials are investigating proton therapy with anti-PD-1 or anti-PD-L1 safety and efficacy for metastatic cancers (neoplasm or head and neck), non-small cell lung cancer, head and neck cancer or meningioma.
In our experiments, we showed that there was tumor infiltration by Tregs, which have an immunosuppressive effect. These interesting findings could encourage the evaluation of an association of proton therapy with for example an anti-CTLA4 or an anti-CCR4, which, by targeting Tregs, induce reactivation of CD8 + T cells against tumor cells 46 . We previously demonstrated with conventional photon treatment that inducing expression of immunotherapy targets by RT could have an impact on the efficacy of the combination of RT with a specific immunotherapy (anti-PD-L1) 10 .
There is a keen interest within the scientific community in developing and evaluating the combination of proton therapy with immunotherapy. Several research teams have expressed an interest, on the assumption that proton therapy could activate the immune response and increase immunotherapy efficacy 20,[47][48][49] . However, to date, no study has demonstrated the concept biologically.
For the first time, we demonstrate here that proton therapy can activate the immune response and can "heat up" the tumor by infiltration of antitumor immune cells. As there is considerable heterogeneity in immune response between different tumor models in the literature, our results obtained on one cell line in a mouse model need to be confirmed using other syngeneic models. Our results pave the way for future studies that could evaluate the effect of the proton therapy dose delivered, and its possible fractionation scheme, the effects of variations in RBE, and the best combination of proton therapy and immunotherapy on different models. It therefore seems essential to evaluate the effect of proton therapy on the expression of specific targets, in order to guide clinicians in the choice of immunotherapy to be combined with proton therapy in future clinical trials.
A single fraction of 16.4 Gy proton therapy led to an interesting induction of immune response biological pathways and immunostimulatory antitumoral effects in an ectopic (subcutaneous) mouse model with a transplanted CT26 colon carcinoma cell line. This study paves the way for future investigations that may explain the improved efficiency of PT via activation of other pathways of immunity response. This study reveals the possible potential of combining proton therapy with immunotherapy in order to enhance tumor control and survival.

Methods
Cell culture and animals. The murine colon carcinoma cell line CT26 was purchased from American Type Culture Collection and cultured in RPMI 1640 (Dutscher, France) supplemented with 10% fetal bovine serum (PAN Biotech GmbH, Aidenbach, Germany) at 37 °C in a humidified atmosphere containing 5% CO 2 and 95% humidity.
Tumor grafting was performed as previously described 10 . Briefly, CT26 cells (5 × 10 5 ) were suspended in 100 µl of NaCl and injected subcutaneously in the right flank of immunocompetent 8-week BALB/c female mice (Charles River Laboratories, Saint-Germain-des-Monts, France). Tumor size was measured until tumor volume (TV) reached the limit point of 1500 mm 3 . TV was calculated according to the equation TV = (L × W 2 )/2, where L and W are the length and width of the tumor, respectively. Ten days after injection, mice were randomized to treatment and control groups to obtain an equivalent average tumor volume in each group of 130 ± 20 mm 3 (n = 5-6 mice for RNA-Seq profiling analysis and n = 4-5 mice for immunomonitoring). Mice were euthanized as soon as the limit point was reached (tumor size ≥ 1500 mm 3 ) or 100 days after treatment for the growth delay study; for RNA-Seq experiments, mice were sacrificed 3 days after irradiation; for immunomonitoring experiments mice were sacrificed 7 and 14 days after irradiation (Fig. 5). As requested by the ethics committee and French regulations, the mice were sacrificed by cervical dislocation after general gaze anaesthesia (Isoflurane 2.5%). www.nature.com/scientificreports/ All mouse procedures were performed with approved protocols in accordance with the legislation on the use of laboratory animals (directive 2010/63/EU), and with ethical rules for the care and use of animals for research from the small animal ethics committee (C2ea Grand Campus n°105 and C2ea Cremeas n°35, C2ea Icomech n°38) and the French Ministries of Research and Agriculture (APAFIS#13961-2018022215413276 v2, APAFIS#22350-201910091738155 v2 and APAFIS#8235-201612161350414_v1). All procedures used follow ARRIVE guidelines to respect 3R recommendations. Thus, animals are randomized before treatment in order to have comparable tumor volume averages. The numbers of mice in each group allow statistical analysis. The control groups of untreated mice allow the analysis of the effect of the treatment.
Treatments. Ten days after injection of cancer cells, tumors were irradiated by a single dose of 16.4 Gy under anaesthesia (isoflurane 2.5% mixed with oxygen compact unit, Minerve, France) with a proton beam extracted from CYRCé platform (CYclotron pour la ReCherche et l'Enseignement) in the Institut Pluridisciplinaire Hubert Curien (Strasbourg, France) with an average energy beam of 25 MeV 50 . Using an in-house immobilization bed, tumors were irradiated directly in contact with the collimator with a dose rate of 3.1 Gy/min. Flow cytometry. The modulation of the immune response by 16.4 Gy proton therapy was evaluated by flow cytometry at 7 and 14 days after treatment, as previously described 10 . Briefly, after dissection, tumors were dissociated using a mouse tumor dissociation kit (Miltenyi Biotech). To analyze myeloid cell infiltration, tumor cell suspension (10 6 cells) was stained in Flow Cytometry Staining Buffer (FSB, eBioscience) with specific antibodies according to manufacturer's recommendation (antibody details are presented in Table S2) for 15 min at room temperature in the dark, washed twice in FSB and analyzed by flow cytometry. To analyze lymphoid cell infiltration, tumor cell suspension was performed according to manufacturer's recommendation (Miltenyi Biotech). To study CD8 + cytotoxicity activity using Granzyme labelling, the tumor cell suspension was cultured on 96-well plates with complete RPMI medium (Dutscher) overnight at 37 °C. During the last 4 h of culture PMA (phorbol 12-myristate 13-acetate; 20 ng/ml; Sigma-Aldrich), ionomycin (1 μg/ml; Sigma-Aldrich), and brefeldin A (2 μl/ ml; eBioscience) were added. For lymphoid and myeloid cell infiltration assays, Viability Dye eFluor 780 was used to identify live cells. Flow cytometry acquisition was performed on a Cytoflex 13C cytometer (Beckman Coulter). CytExpert (Beckman Coulter) was used for analysis. For lymphoid and myeloid cell identification and for lymphoid cell functionality, the gating strategies are presented in Supplementary Figure S3. RNA extraction, RT-qPCR quantification, RNA sequencing (RNA-Seq) and gene set enrichment analysis. Tumor tissue was dissociated with Minilys tissue homogenizer (Bertin, Ozyme). Then, total RNA was extracted from tumor using Trizol method (Invitrogen).
For RT-qPCR, 1000 ng of RNA was reverse transcribed into cDNA using M-MLV Reverse Transcriptase (10,338,842, Invitrogen) and random primers (10,646,313, Invitrogen). cDNAs were quantified by real-time PCR using a Power SYBR Green Real-time PCR Kit (10,219,284, Fisher Scientific) on a QuantStudio 5 Real Time PCR System (Thermo Fisher Scientific). Relative mRNA levels were determined using the ΔCt method. Values were expressed relative to β-actin. The sequences of the oligonucleotides used are described in Supplementary  Table S4.
Single-end transcriptome reads were pseudo-aligned to the UCSC mm 10 reference genome and quantification of gene expressions was performed with the Kallisto algorithm (v 0.44.0) 51 . The program was run with default options.
RNA-Seq profiling analysis was assessed with R software (R version 4.0.3). Differential analysis was performed with DESeq2 R package (version 1.30.0) 52 using log fold change shrinkage 53 . A gene was considered significantly differentially expressed when the corresponding s-value 54 was < 0.005 and a log2 fold change ≥ 1. Gene set enrichment analyses were performed using gProfiler2 (v0.2.0) 55 .

Statistical analysis.
The results are expressed as mean ± SEM (standard error of the mean) for immunomonitoring experiments and as mean ± SD (standard deviation) for quantification of genes mRNA relative expression. Figures were designed using GraphPad Prism V8. Software (GraphPad Software, USA). Comparisons between groups were carried out using a non-parametric Mann-Whitney test for immunomonitoring experiments and quantification of genes mRNA relative expression and a Log-rank test for Kaplan Meier curves. Statistical analyses were achieved using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). A p-value < 0.05 was considered statistically significant.