Proton radiation boosts the efficacy of mesothelin-targeting chimeric antigen receptor T cell therapy in pancreatic cancer

Significance

Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive cancer with limited treatment options, demanding innovative therapeutic strategies. By combining proton radiation therapy (RT) with Chimeric antigen receptor T cell (CAR T) therapy targeting mesothelin (MSLN), this research addresses barriers hindering the success of CAR T therapy in solid tumors. Here, we show that proton RT increases MSLN expression, modulates the immunosuppressive microenvironment, and improves the antitumor response of CAR T therapy, underscoring the transformative potential of this strategy. This combination of proton RT and MSLN-targeting CAR T cells provides a blueprint for overcoming challenges in solid tumor immunotherapy, marking an important step toward improved patient outcomes.

Abstract

Pancreatic ductal adenocarcinoma (PDAC) represents a challenge in oncology, with limited treatment options for advanced-stage patients. Chimeric antigen receptor T cell (CAR T) therapy targeting mesothelin (MSLN) shows promise, but challenges such as the hostile immunosuppressive tumor microenvironment (TME) hinder its efficacy. This study explores the synergistic potential of combining proton radiation therapy (RT) with MSLN-targeting CAR T therapy in a syngeneic PDAC model. Proton RT significantly increased MSLN expression in tumor cells and caused a significant increase in CAR T cell infiltration into tumors. The combination therapy reshaped the immunosuppressive TME, promoting antitumorigenic M1 polarized macrophages and reducing myeloid-derived suppressor cells (MDSC). In a flank PDAC model, the combination therapy demonstrated superior attenuation of tumor growth and improved survival compared to individual treatments alone. In an orthotopic PDAC model treated with image-guided proton RT, tumor growth was significantly reduced in the combination group compared to the RT treatment alone. Further, the combination therapy induced an abscopal effect in a dual-flank tumor model, with increased serum interferon-γ levels and enhanced proliferation of extratumoral CAR T cells. In conclusion, combining proton RT with MSLN-targeting CAR T therapy proves effective in modulating the TME, enhancing CAR T cell trafficking, and exerting systemic antitumor effects. Thus, this combinatorial approach could present a promising strategy for improving outcomes in unresectable PDAC.

Pancreatic ductal adenocarcinoma (PDAC) presents a formidable challenge in oncology, with its relentless progression poised to make it the second leading cause of cancer-related deaths in the United States by 2030 (1). While the conventional approach to early-stage PDAC involves surgical resection paired with adjuvant chemotherapies (2), a significant number of patients are diagnosed at an advanced stage, rendering surgical intervention impractical due to barriers posed by tumor size, location, or metastasis. Given these clinical predicaments, creating novel therapeutic approaches that improve outcomes is imperative (3).

Chimeric antigen receptor T (CAR T) cell therapy has shown impressive responses in hematological malignancies, offering potential in solid malignancies (4, 5). Recently, mesothelin (MSLN), a surface-bound glycosylphosphatidylinositol-anchored protein, has emerged as a compelling target for CAR T therapy for several solid malignancies, including PDAC, paving the way for innovative therapeutic strategies (6–11). However, trials employing anti-MSLN-CAR T cells have shown limited clinical efficacy (12), leading to ongoing research aimed at addressing and overcoming the barriers that impede their success. A key obstacle is the heterogeneous expression of MSLN within PDAC, creating a dynamic landscape that complicates the comprehensive elimination of cancer cells (13). Additionally, the shared expression of MSLN on the pleura, peritoneum, and pericardium introduces the risk of on-target, off-tumor toxicities (14). The resilient, highly desmoplastic stroma surrounding PDAC presents a dual challenge, as a physical barrier impeding CAR T cell infiltration into the tumor site and fostering an immunosuppressive microenvironment (15, 16). Addressing these intricate hurdles is crucial for advancing the application of CAR T therapy in PDAC.

Radiation therapy (RT) is an established treatment modality for unresectable PDAC, alleviating symptoms and improving local control (17–19). Beyond its classical role in eradicating cancer cells, a paradigm shift acknowledges that RT profoundly influences the tumor microenvironment (TME) and the immune system (20). Compelling evidence from preclinical models and clinical studies substantiates its multifaceted stimulatory impact in PDAC and other solid adenocarcinomas, including heightened immunogenic cell death (21, 22), elevated interferon (IFN) signaling (23–25), and synergistic interactions with immune-modulatory therapies (26–28). However, while therapeutically potent, the large exit radiation dose of photon RT raises concerns about collateral damage to normal organs and immune cells within the irradiated region, potentially hindering the optimal functioning of immunotherapies (29–31). Furthermore, the low-dose bath of photon RT may induce the upregulation of the target antigens, including MSLN, in normal organs, amplifying the risk of unintended toxicities.

Addressing these challenges, proton RT emerges as a promising alternative to photon-based therapy. The focused and precise nature of proton RT, characterized by no exit RT dose beyond the Bragg peak (32), allows targeted delivery to the tumor site with reduced exposure to surrounding organs and immune cells (29, 31, 33, 34). This characteristic is particularly significant for optimizing the therapeutic balance between eradicating cancer cells and preserving the integrity of the immune system. In the pursuit of refining the therapeutic landscape for unresectable PDAC, the consideration of proton RT offers a potential avenue to mitigate adverse effects associated with photon-based RT while enhancing the specificity and efficacy of emerging immunotherapeutic approaches, such as CAR T cell therapy.

Here, we demonstrate that proton RT increases the expression of MSLN on tumor cells, modifies the immunosuppressive microenvironment, and improves the local and systemic antitumor response to MSLN-targeting CAR T therapy. This innovative approach of combining proton RT and CAR T therapy could pave the way toward addressing the challenges in solid tumor immunotherapy, thereby improving patient outcomes.

Results

Proton RT Increases MSLN-Targeting CAR T Trafficking and Reshapes the Immunosuppressive PDAC TME.

RT has been shown to affect the expression of cell-surface proteins, therefore potentiating CAR T-mediated killing in solid tumors (35), prompting us to compare MSLN expression on cancer cells before and after proton RT. A mouse PDAC model was established by inoculating PDA7940b cells derived from KPC mice (LSL-Kras^G12D/+; LSL-Trp53^R172H/+; Pdx1-Cre) (36), subcutaneously (S.C.) into the right flank of C57BL/6J mice (Fig. 1A). After 10 d, when the tumors reached the size of ~50 mm³, mice were randomized into a proton RT treatment group in which visible tumors received a single fraction of 12 Gy and a control group that did not receive RT (NR group). Flow cytometry analysis revealed an average increase of approximately 45% and 36% in MSLN Mean Fluorescent Intensity (MFI) at 3- and 8-d following 12 Gy proton RT, respectively, as compared to the NR group (Fig. 1 B–D and SI Appendix, Fig. S1 A and B), suggesting that RT increases MSLN expression in tumor cells. Irradiation with doses of 8 Gy and 16 Gy also increased MSLN expression. There was a borderline significant dose–response in MSLN expression between these two doses (P = 0.06), whereas no statistically significant difference was observed between the 12 Gy and 16 Gy doses (SI Appendix, Fig. S1C). To confirm flow cytometry data and to investigate whether RT leads to higher frequencies of MSLN-expressing tumor cells, we performed RNA in situ hybridization (RNA-ISH) to detect MSLN transcripts from formalin-fixed paraffin-embedded tumor tissues. Significantly higher MSLN-positive areas within analyzed sections were observed in RT-treated tumor tissues when compared to untreated tumors (Fig. 1E), supporting the hypothesis that 12 Gy RT does not only lead to higher MSLN expression but also to higher frequencies of MSLN-expressing tumor cells. Next, we tested whether RT increases CAR T cell trafficking into tumors. We injected PDA7940b cells into the right flank of C57BL/6J mice (Fig. 1F). After 10 d, the mice were divided into four treatment groups: a group that received no treatment (NR), an MSLN-targeting CAR T treatment group that received one dose of CAR T treatment intravenously (I.V.) (CAR T), a proton RT group that received a single fraction of 12 Gy (RT), and a combination group that received a single fraction of 12 Gy proton RT to the tumor and a dose of CAR T (RT and CAR T group). All four groups received a single intraperitoneal injection (I.P.) of cyclophosphamide as a standard lymphodepletion strategy. On day eight after RT (day six after CAR T injection), tumors were harvested for flow cytometry of single-cell suspension and RNA-ISH of bulk tumor tissue. In line with the increase of MSLN protein intensity after proton RT, flow cytometry performed from tumors 6 d after CAR T I.V. injections showed a 9.5-fold increase in CD3⁺CD45.1⁺ cells (infused CAR T cells) in the TME compared to the number of CAR T in the unirradiated group (Fig. 1G). In RNA-ISH, CAR transcripts (vMSGV RNA) were detected, confirming penetrance and trafficking of the CAR T in the irradiated TME (Fig. 1H). A higher percentage of CAR T cells expressing T cell immunoglobulin and mucin domain 3 (TIM-3) and Lymphocyte Activation Gene 3 (LAG-3) were observed in the combination treatment group, whereas the percentage of intratumoral CAR T expressing Programmed Cell Death Protein 1 (PD-1) remained similar between cohorts (Fig. 1 I–K and SI Appendix, Fig. S2).

Fig. 1.

Proton RT increases CAR T trafficking into PDAC tumors through increased MSLN expression. (A) Schematic of in vivo mouse experiments to evaluate the expression of MSLN. 5 × 10⁵ PDA7940b cells were injected S.C. into the right flank of C57BL/6 mice. Tumors were grown until sizes reached ~50 mm³ before treatment with proton RT. Mice were killed, and tumors were harvested for analysis 3- and 8-d post-proton RT. (B) A representative fluorescent microscopy image shows blood vessels (CD31; red) and MSLN (green) of irradiated and unirradiated (NR) PDAC tumors (10× magnification). Scale bar, 100 µm. (C–E) MFI of MSLN levels, as assessed by flow cytometry of tumor single-cell suspension, at (C) day 3 and (D) day 8 post 12 Gy proton RT, as well as (E) the percent of MSLN-positive transcripts detected by RNA-ISH in formalin-fixed paraffin-embedded tumor tissues. Average ± SD and individual values are shown. The Mann–Whitney two-tailed test was used for statistical analysis. (F) Schematic of mouse experiments to evaluate CAR T cell trafficking into tumors. 5 × 10⁵ PDA7940b cells were injected S.C. into the right flank of C57BL/6 mice. Tumors were grown until sizes reached ~50 mm³ before treatment with proton RT, followed by single 120 mg/kg cyclophosphamide administration and tail vein injection of 5 × 10⁶ MSLN-CAR T cells at day two post-RT. Eight days post-RT, mice were killed, and tumors were harvested for analysis. (G) CAR T cell count in tumors assessed by flow cytometry. (H) Representative image from RNA-ISH against CD3 (red), CAR T (pink), and MSLN (green). Scale bar, 50 µm. (I–K) Rates of TIM-3 (I), LAG-3 (J), and PD-1 (K) positive CAR T cells 8 d after 12 Gy proton RT (n = 5 per group). The Mann–Whitney U test was used for statistical analysis. (L and M) Immune cell populations in the TME. (L) M1 macrophages (CD45⁺F4/80⁺CD86⁺MHCII⁺) at 3 d and (M) MDSC (monocytic) (CD45⁺CD11B⁺Ly-6G⁻Ly-6C⁺) at 8 d following proton RT. Average ± SD are shown. Kruskal–Wallis one-way ANOVA for multiple comparisons was used for statistical analysis. For all panels in this figure, * indicates P ≤ 0.05, ** indicates P ≤ 0.01, and ns (nonsignificant) indicates P > 0.05.

To assess changes in immune cell composition within the irradiated TME, tumors were collected at two time points following proton RT. Flow cytometry of single-cell suspension and pathological assessment of bulk tumor tissue was carried out 3 d after 12 Gy proton RT (1 d after CAR T I.V. injection) and 8 d after proton RT (6 d after CAR T I.V. injection). Three days after proton RT, we observed an increase in antitumorigenic (M1) tumor-associated macrophages (TAM) in the TME (CD45⁺F4/80⁺CD86⁺MHCII⁺) in the combined RT and CAR T treatment group compared to the CAR T alone group (Fig. 1L and SI Appendix, Fig. S3). Whereas 8 d after proton RT, a significant decrease in monocytic myeloid-derived suppressor cells (MDSCs) (CD45⁺CD11B⁺Ly-6C⁺Ly-6G⁻) in the combination group compared to the NR group was noted (Fig. 1M and SI Appendix, Fig. S3). Further, at 8 d, the histopathological evaluation revealed karyomegalic neoplastic cells, characteristic of RT-induced morphological changes (37) (SI Appendix, Fig. S1D). No significant differences in the other immune cell populations in the irradiated TME were observed at either time point (SI Appendix, Fig. S4). Our results show that proton RT can modulate the PDAC TME, creating a more favorable environment for CAR T cell function.

Combining Proton RT and CAR T Therapy Attenuates Tumor Growth and Improves Survival in Flank and Orthotopic PDAC Mouse Models.

To test the hypothesis that proton RT mediates improved CAR T cell antitumor efficacy and survival, we implanted PDA7940b cancer cells into the flanks of mice as previously described and divided them into four groups: No treatment, two weekly I.V. CAR T treatments, a single dose of 12 Gy proton RT, and a combination group treated with 12 Gy proton RT and two doses of CAR T (Fig. 2A). We found no difference in tumor growth rate or overall survival between CAR T therapy and NR groups (Fig. 2 B and C). Local proton RT effectively reduced tumor growth compared to the NR and CAR T groups. However, the combination of RT and CAR T treatment elicited the most significant delay in tumor growth and improved survival (Fig. 2 B and C). Furthermore, administering a second RT treatment before the second CAR T therapy dose did not provide any additional benefit (SI Appendix, Fig. S5 A and B).

Fig. 2.

Combining proton RT and CAR T therapy improves survival and attenuates tumor growth in flank and orthotopic PDAC mouse models. (A) Schematic of in vivo mouse experiments to evaluate the tumor growth and survival after the combination of proton RT with MSLN-targeting CAR T. A total of 5 × 10⁵ PDA7940b cells were injected S.C. into the right flank of C57BL/6 mice. Tumors were grown until sizes reached ~50 mm³ before treatment with proton RT, followed by single 120 mg/kg cyclophosphamide administration and tail vein injection of 5 × 10⁶ MSLN-CAR T cells at 2- and 9-d post-RT. Mice were monitored, and tumors were measured with calipers three times per week. Mice were killed once tumors reached a volume of 1,000 mm³. (B) Tumor growth curves following treatment with 12 Gy proton RT, CAR T cells, RT and CAR T, or NR as control. Values represent the mean ± SEM. Two-way ANOVA was used for statistics. Black arrows indicate the CAR T cell injections. (C) Kaplan–Meier survival analysis of mice. The log-rank Mantel–Cox test was used for statistical analysis. (D) Schematic of mouse orthotopic tumor and implanted radiopaque marker at the injection site (Image was created with BioRender.com). (E) Images of the pancreas collected 3 wk after injection of 5 × 10⁵ PDA7940b cells orthotopically in the tail of the pancreas of RT (n = 5 biologically independent samples; one mouse was excluded from the final measurements due to tumor growth that caused its death after 11 d) and RT+CAR T (n = 6 biologically independent samples). The red and blue dotted lines indicate the tumor and normal areas of the pancreas, respectively. (F and G) Box and whisker plots display the percentage of tumor weight normalized to body weight (F) and intratumor necrosis score percent (G). The Mann–Whitney two-tailed test was used for statistical analysis. (H) Representative images of the pancreas from an orthotopic pancreatic tumor model, stained for H&E. The red and blue dotted lines indicate the tumor and normal areas of the pancreas, respectively. (I) Box and whisker plot of the percentage of tumor area normalized to total pancreas area. (Scale bar, 1 mm.) The Mann–Whitney two-tailed test was used for statistical analysis. For all panels in this figure, * indicates P ≤ 0.05 and *** indicates P ≤ 0.001.

To better recapitulate the complex interactions between tumor cells, stroma, and the surrounding tissues and provide a more relevant representation of disease progression in patients, we used an orthotopic PDAC model (38) (Fig. 2D). A radiopaque metal marker was affixed to the site of tumor cell injection orthotopically to the tail of the pancreas. Cone-beam CT was performed to mimic clinical image-guided proton RT for precise alignment with the proton beam during RT (SI Appendix, Fig. S5 C and D and Movie S1). Given the lack of response to CAR T therapy in the flank tumor models (Fig. 2 B and C), we randomized mice into groups receiving 12 Gy proton RT alone or 12 Gy proton RT combined with two doses of CAR T, 2- and 9-d following RT. Throughout the follow-up period, mice receiving combination treatment demonstrated no significant weight reduction, indicating minimal therapy-related toxicity (SI Appendix, Fig. S5E). Tumors in the proton RT group were visually larger than tumors in the CAR T and proton RT cohort (Fig. 2E). Additionally, 2 wk after RT, the terminal tumor growth (determined by tumor weight to mouse weight ratio) was significantly lower in the combination treatment group compared to the RT-only group (Fig. 2F), and displayed higher intratumoral necrosis (Fig. 2G). Moreover, the tumor area to normal pancreas area ratio was higher in the group that received RT alone, indicating that most of the normal pancreas had been replaced by the neoplastic process (Fig. 2 H and I).

Local Proton RT Increases the Proliferation of Extratumoral MSLN-Targeting CAR T, Elevates Serum Cytokines, and Potentiates an Abscopal Response.

Serum cytokines were analyzed from treated mice 3 and 8 d after proton RT (1 d and 6 d after CAR T injections, Fig. 3). Three days after RT, we observed increased serum levels of IFN-γ, IL-5, and IL-17 in the proton RT alone and combination treatment groups compared to the NR and CAR T groups (Fig. 3 A, B, and D). An increase in the serum levels of interleukin IL-22 was noted in the CAR T-only group (Fig. 3 C and E). Eight days after RT, serum IFN-γ remained higher in the RT and CAR T group (Fig. 3F). No statistical differences were observed in serum levels of IL-22 and IL-17 between the treatment groups at day eight (Fig. 3 G–J). Next, we harvested spleens on day eight after RT (day six after CAR T injections) to analyze changes in extratumoral CAR T cells and analyzed CAR T cell counts in peripheral blood. Flow cytometry of single-cell suspensions revealed a significant increase in CAR T in the spleens of mice treated with proton RT compared to MSLN-targeting CAR T alone (Fig. 4A). Dose calculations indicated that the spleen received less than 2 to 3% of the dose delivered to the irradiated field (SI Appendix, Fig. S6A). Furthermore, no visible DNA double-strand breaks, markers of radiation exposure, were detected (SI Appendix, Fig. S6B). These findings suggest that the increase in CAR T cells in the spleen is not attributable to direct radiation exposure. The CAR T cell population in the combination group in the spleens did not display any differences in the expression of PD-1, TIM-3, and LAG-3 compared to the CAR T treatment group only (SI Appendix, Fig. S7 A–C). In addition, the combination group exhibited a higher proliferation rate of endogenous T cells in the spleen (CD3⁺CD45.1⁻) (Fig. 4B) with no significant differences in other immune cell populations (SI Appendix, Fig. S7 D–J). No changes in the number of CAR T were observed in the peripheral blood of mice (Fig. 4C).

Fig. 3.

Proton RT induces serum cytokine levels. (A) IFN-γ, (B) IL-5, (C) IL-22, (D) IL-17A, and (E) heat map summarizing levels of indicated serum cytokines at 3 d after proton RT. (F) IFN-γ, (G) IL-5, (H) IL-22 (I) IL-17A, and (J) heat map summarizing levels of indicated serum cytokines at 8 d after proton RT. (n = 4 to 5 per group and time point). Average and individual values are shown. Kruskal–Wallis’s one-way ANOVA for multiple comparisons was used for statistical analysis. For all panels in this figure, * indicates P ≤ 0.05, ** indicates P ≤ 0.01, and ns (nonsignificant) indicates P > 0.05.

Fig. 4.

Combining proton RT with CAR T potentiates a systemic response. (A) CAR T counts (CD3⁺CD45.1⁺), (B) Endogenous T cell counts (CD45⁺CD3⁺CD45.1⁻) in the spleen at 8 d post-proton RT, and (C) CAR T cell count in peripheral blood 8 d after proton RT (n = 5 per group). Kruskal–Wallis’s one-way ANOVA was used for statistical analysis. (D) Schematic of in vivo mouse experiments to evaluate the tumor growth after combining proton RT with MSLN-targeting CAR T in a dual-flank tumor model. 5 × 10⁵ PDA7940b cells were injected into the right flank (“Primary tumor”) of C57BL/6J mice. After 2 d, an equal number of cancer cells were injected into the opposite flank (“Secondary tumor”). Tumors were grown until the Primary tumor sizes reached ~50 mm³ before it was treated with 12 Gy proton RT, followed by single 120 mg/kg cyclophosphamide administration and tail vein injection of 5 × 10⁶ MSLN-CAR T cells at 2- and 9-d post-RT. The Secondary tumor was not irradiated. Mice were monitored, and the “Primary” and “Secondary” tumors were measured with calipers 3 d per week. Mice were killed once tumors reached a volume of 1,000 mm³. (E and F) Tumor growth curves of Primary (E) and Secondary ("abscopal”) (F) tumors following treatment with RT, CAR T cells, RT+CAR T, or NR as control. Black arrows indicate the CAR T cell injections. Values represent the mean ± SEM. Two-way ANOVA was used for statistics. For all panels in this figure, * indicates P ≤ 0.05, ** indicates P ≤ 0.01, **** indicates P ≤ 0.0001, and ns (nonsignificant) indicates P > 0.05.

Next, we employed a dual-flank tumor mouse model to investigate the potential abscopal effects of CAR T and proton RT treatments. In our model, 5 × 10⁵ PDA7940b cells were injected into the right flank (Primary tumor) of C57BL/6J mice. After 2 d, an equal number of cancer cells were injected into the opposite flank (Secondary tumor) (Fig. 4D). Ten days later, mice were randomly assigned to four groups: 1. NR 2. A single dose of 12 Gy proton RT to the Primary tumor 3. Two CAR T treatments administered 1 wk apart. 4. A combination of 12 Gy proton RT to the Primary tumor followed by two CAR T injections. As observed in our previous experiments, MSLN-targeting CAR T therapy alone did not induce a significant regression in tumor growth in either the Primary or the Secondary tumors (Fig. 4 E and F). RT caused an attenuation in the tumor growth of the irradiated flank tumor. However, adding CAR T therapy to proton RT caused a significant, additional delay in tumor progression in both the Primary and Secondary tumors (Fig. 4 E and F and SI Appendix, Fig. S8 A–F). To enhance the validity and generalizability of our results, we conducted a similar experiment with a dual flank experimental design, but this time using 12 Gy photon RT (XRT) instead of protons. This test confirmed that combining XRT and MSLN-targeting CAR T cell treatment significantly reduced tumor growth in both the Primary and Secondary tumors compared to the individual treatments (SI Appendix, Fig. S9 A–C).

Discussion

MSLN is a glycosylphosphatidylinositol-anchored cell-surface protein that is present in high levels in a plethora of cancers, including PDAC, but only on the mesothelial surface of the body, making it an attractive target for CAR T therapy (39). However, early-phase clinical trials using MSLN-targeting CAR T therapy did not yield sustained clinical responses, possibly due to the inability of the CAR T cells to penetrate the immunosuppressive TME (40). Our results suggest a multifaceted effect of proton RT on MSLN expression on tumor cells, CAR T trafficking, TME modulation, and systemic immune response. Furthermore, we demonstrate that combining proton RT and MSLN-targeting CAR T therapy attenuates PDAC growth, improves survival, and potentiates systemic CAR T therapy. Our results also show that combining treatments results in delayed tumor growth in both the irradiated and distal (abscopal) tumors, thus offering a treatment strategy for unresectable PDAC.

Target antigen heterogenicity and loss of target antigen expression remain significant limitations in CAR T therapy (41). This is also a major challenge in targeting immunotherapeutics toward PDAC due to the low to moderate intensity of MSLN expression in most tumors, despite 75 to 100% of PDAC expressing MSLN (13, 40, 42, 43). Our study demonstrates a significant increase in MSLN protein expression intensity in a mouse model of PDAC following a single focal dose of 8, 12, or 16 Gy proton RT. Although a noticeable trend suggests a dose–response relationship, this did not reach statistical significance. The elevated MSLN expression was accompanied by almost a 10-fold enhanced trafficking of CAR T into the TME compared to the CAR T treatment alone. Similarly, Hassan et al. showed that irradiating A431-K5, a cervical epidermoid carcinoma cell transfected with a plasmid encoding MSLN, with 10 Gy, increased the MFI of MSLN in the cancer cells. Combining RT with SS1P, a recombinant anti-MSLN immunotoxin that consists of an anti-MSLN variable antibody fragment linked to a truncated portion of Pseudomonas exotoxin A, attenuated tumor growth in a murine model (44). Notably, a recent study investigated the effects of low-dose photon RT (noncytotoxic) in combination with MSLN-targeting CAR T in immunodeficient NOD/SCID gamma mice harboring non–small cell lung cancer and mesothelioma models. They found an increased accumulation of CAR T cells in the tumor, attributed to a dose-dependent increase in the expression of several chemokines and chemokine receptors in the infiltrating T cells. Nonetheless, exposure to 4 Gy photon RT did not enhance MSLN expression in cancer cells when tested in vitro (45). Considering these findings collectively, it is possible that proton RT enhances the sensitivity of PDAC tumors to MSLN-targeting CAR T therapy by elevating the presence of the target protein on the cancer cell surface. However, this enhancement is contingent on dosage and necessitates doses surpassing 4 Gy.

PDAC tumors are characterized by highly immunosuppressive TME characterized by tumor cells, protumorigenic suppressive immune cells, and cytokines leading to CAR T exhaustion and dysfunction (46). RT can potentially enhance the efficacy of CAR T cell therapy in solid tumors, mainly by influencing the TME. RT can induce immunogenic cell death and modulate the composition and function of immune cells by promoting the release of tumor-associated antigens (TAA), danger signals, and cytokines, creating a more immunostimulatory milieu (21, 47, 48). Our study indicates that proton RT, combined with MSLN-targeting CAR T therapy, increases MSLN expression and “classically activated” M1 TAMs in the TME. Importantly, this increase is not observed in the CAR T or RT alone treatment groups, suggesting a synergistic effect between the two treatments, which leads to a rise in the M1 macrophage population in the TME. M1 macrophages are characterized by the secretion of proinflammatory cytokines, such as IL-1β, IL-6, tumor necrosis factor-alpha, and reactive oxygen and nitrogen species, therefore promoting phagocytosis, antigen presentation, and cancer cell killing (49, 50). One possible mechanism for the observed enrichment in M1 macrophages in the combination treatment group is the increase in the serum levels of IFN-γ, which has been shown to mediate macrophage polarization toward a classically activated phenotype (51). Furthermore, the TME exhibited fewer MDSCs in the combination therapy. Importantly, MDSCs can significantly hamper CAR T cell function as they possess potent immunosuppressive capabilities that directly target effector T cells. MDSC-mediated CAR T cell suppression is profound, and low levels of MDSCs in B cell lymphoma or leukemia patients receiving CAR T targeting CD19 have been associated with improved response to therapy (52).

In addition to the aforementioned stimulatory effects on the TME, RT may also act as an “in situ cancer vaccine,” inducing a systemic immune-mediated cancer cell killing termed “abscopal effect” (53). In our study, we observed proliferation in the spleen of both endogenous and CAR T cells after proton RT. This was accompanied by a rise in the serum levels of IFN-γ and changes in other cytokines related to the immune system, such as IL-17A, IL-5, and IL-22. Previous studies have revealed that local RT can propagate a cascade of innate and adaptive immune responses by releasing damage-associated molecular patterns (DAMPs). This, in turn, induces the stimulator of interferon genes signaling pathway and promotes cross-priming of TAA and activation of effector T cells (54). Recent studies have found that IFN-γ can further stimulate this immune response by promoting the presentation of TAAs, activating cytotoxic T cells, and inducing cancer cell apoptosis and death (55). Using a mouse colon adenocarcinoma model, Gerber et al. demonstrated that CD8⁺ cells are the primary source of IFN-γ following RT and that RT had no impact on tumor burden in mice lacking IFN-γ (23). IFN-γ has also been proposed to mediate the abscopal effects of RT in mice inoculated with B16-OVA cells and treated with the gram-positive antibiotic vancomycin (56). Importantly, IFN-γ has been implicated in CAR T function in solid tumors. The IFN-γ-receptor signaling pathway plays a crucial role in cell adhesion by regulating CAR T cells’ binding duration and avidity (57). The activation of CAR T cells prompts the production of IFN-γ, which triggers the upregulation of Intercellular Adhesion Molecule 1 (ICAM-1) in glioblastoma cells. High-frequency interaction between ICAM-1 and Lymphocyte function-associated antigen 1 stabilizes the immunologic synapse, which induces effective cytotoxicity of CAR T cells against its targets (57). Therefore, the increase in IFN-γ following proton RT likely amplifies the synergy of combining MSLN-targeting CAR T therapy and RT. Finally, RT can also induce damage to the blood vessels, particularly the endothelial cells lining the microvasculature. This damage can lead to changes in the structure and permeability of the vessels, enabling leakage of cytokines, DAMPs, and immune cells into the TME, which may affect the entrance and activation of CAR T cells (58, 59).

Importantly, the addition of the second proton RT fraction had only a modest benefit in terms of growth delay compared to the RT alone. One possible explanation is that treating RT after CAR T may result in the depletion of the CAR T within the tumor, thus limiting the effectiveness of the treatment. In a recent study, Saifi et al. explored the optimal approach to incorporate RT with CAR T therapy in a large cohort of patients with relapsed/refractory B cell non-Hodgkin lymphoma. They found improved local control for sites that received RT before the CAR T infusion as a bridging strategy compared to patients who received RT as a salvage intervention post-CAR T treatment (60). Taken together, it can be inferred that the optimal timing for adding CAR T therapy should be after the completion of the RT treatment course, which usually confers to after 3 to 5 fractions of stereotactic RT in PDAC patients. This scheduling will allow for the upregulation of MSLN by RT in the tumor and reprogram the TME to be more receptive to the infiltration of the CAR T while not subjecting the CAR T to lymphodepletion by RT. Further studies are warranted to compare the optimal timing for RT and CAR T combination.

This study has limitations that should be considered when interpreting the results. First, the experiments were conducted using only one mouse PDAC cell line. This limits the generalizability of our findings, as different cell lines may exhibit varied responses to the combination of MSLN-targeting CAR T cells and RT. Second, our study was limited to the use of female mice. Biological differences between male and female mice could potentially influence the immune response and treatment efficacy.

In conclusion, we show that proton RT enhances the expression of MSLN on PDAC tumor cells, alters the immunosuppressive microenvironment, and boosts both local and systemic antitumor responses to MSLN-targeting CAR T therapy. These results suggest a combination treatment approach to be tested in clinical trial settings.

Materials and Methods

PDAC Tumor Cells.

The murine PDA cell line, PDA7940b, was a kind gift from Gregory Beatty (Division of Hematology-Oncology, Department of Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA) and was established from a KPC (LSL-Kras^G12D/+; LSL-Trp53^R172H/+; Pdx1-Cre) mouse PDAC tumor model (36), and endogenously expresses mouse MSLN (10). Tumor cells were tested regularly for Mycoplasma contamination by the Department of Genetics at the University of Pennsylvania (MycoAlert Mycoplasma Detection Kit, Lonza) (11, 61). PDA7940b cells were maintained in culture with R10 media: Roswell Park Memorial Institute (RPMI) 1640 (Gibco) supplemented with 10% heat-inactivated fetal bovine serum (FBS, Seradigm), 2% 1 M 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer solution (Gibco), 1% 100× glutaMAX (Gibco), and 1% 10,000 U/mL penicillin + 10,000 μg/mL streptomycin (Gibco) (11, 61).

MSLN-Targeting CAR T Cell Production and Injections.

Prior publications described the production of mouse MSLN-targeting CAR T cells using retroviral vectors (9). Briefly, the spleen from C57BL/6 mice was harvested. Mouse T cells were then purified with a mouse T cell isolation kit (STEMCELL) and activated with anti-mouse CD3/CD28 antibody-coated beads (Dynabeads, Gibco). A bead: T cell ratio of 2:1 was used (11, 61). Two days after bead stimulation, T cells were retrovirally transduced to express a mouse MSLN-specific CAR construct (MSGV-mMesoBBz) on recombinant human fibronectin-coated plates (Retronectin, TaKaRa) (11, 61). Recombinant mouse IL-2 (50 U/mL) was supplemented daily with fresh mouse T cell media [RPMI 1640 (Gibco) supplemented with 10% heat-inactivated FBS (Seradigm), 1% 100× glutaMAX (Gibco), 1% 10,000 U/mL penicillin + 10,000 μg/mL streptomycin (Gibco), 1× 100 mM sodium pyruvate (Gibco), and 50 μM β-mercaptoethanol (SIGMA)] containing 50 U/mL IL-2 every day (11, 61). On day five of stimulation, mouse CAR T cells were harvested, debeaded, analyzed for CAR expression by anti-Fab (2)’ staining and flow cytometry, and used for in vivo experiments the same day (11, 61). MSLN-targeting CAR T was injected I.V. under direct visualization 2 and 9 d after proton RT at a dose of 5 × 10⁶ cells per mouse. One day before the first MSLN-targeting CAR T treatment, mice were treated with I.P. cyclophosphamide (Sigma-Aldrich) at a 120 mg/kg dose for lymphodepletion (11, 61).

Tumor Growth and Survival Experiments.

All animal experiments were conducted in accordance with the University of Pennsylvania’s regulations (animal protocol number 805191) and approved by the University Laboratory Animal Resources and Institutional Animal Care and Use Committee (IACUC). The mice were kept in a controlled environment with 12:12 light–dark cycles, temperatures ranging from approximately 18 to 23 °C, and humidity maintained at 40 to 60%. In all in vivo experimental procedures, 9 to 11-wk-old female C57BL/6 mice were used. For the tumor growth studies, 5 × 10⁵ PDA7940b cells were injected S.C. into the right flank of mice. For the orthotopic pancreatic tumor model, 5 × 10⁴ PDA7940b cells were injected into the tail of the pancreas. In the dual-flank tumor model, 5 × 10⁵ PDA7940b cells were injected into the right flank of mice, and 2 d later, the same number of cells were injected into the contralateral flank.

Proton and Photon RT of Mouse Flank Tumors.

Ten days after tumor flank injections, mice were subjected to proton RT and photon RT (XRT) as described (62). Briefly, mice were irradiated while immobilized with 2.5% isoflurane anesthesia with medical air as the carrier gas (VetEquip). After 5 min in an induction chamber, a fully anesthetized mouse was placed on a customized platform. The head of the mouse was then put in a face mask that allowed the gas to be scavenged (Xerotec Inc.). For all irradiations, the tumor was aligned using a laser positioning system. The system provided vertical and horizontal lasers aligned with the proton and photon beam centers (63). Custom three-dimensional printed jigs with anesthesia gas ports were used to allow irradiation of mice on the Small Animal Radiation Research Platform (SARRP, Xstrahl Life Sciences) platform with cone-beam CT capability (63). In a dedicated research room, IBA Proteus Plus (Louvain-La-Neuve, Belgium) delivered a proton RT beam at a fixed angle beam line at a dose rate of ~1 Gy/s. For photon RT, mice underwent image-guided RT using the SARRP and received a single dose of 12 Gy at a dose rate of approximately 1.65 Gy per minute.

Implantation of the Radiopaque Marker for Image-Guided Proton RT.

Animals were anesthetized using inhaled isoflurane at a concentration of 2 to 2.5%. A 1 to 2 cm midline abdominal incision was made, and the pancreas was exposed in situ. After injecting PDA7940b cells into the tail of the pancreas, one side of the radiopaque marker was coated in Medbond Tissue Glue and then rapidly applied to the tumor injection site. The peritoneum was sutured using 5-0 Poly(glycolic acid)-poly(caprolactone) (PGA-PCL) absorbable sutures (McKesson), and the skin was sutured using 5-0 nylon nonabsorbable sutures (Ethicon). All surgical procedures were performed in a sterile manner. The mice were given 10 mg/kg of meloxicam S.C. every 24 h for 72 h and 1 mg/kg of sustained-release buprenorphine at surgery. From the time of surgery until killing of the animals, animal body weight and survival were monitored.

Cone-Beam CT Imaging and Proton RT of the Orthotopically Implanted Tumors.

All mice were imaged and irradiated for these studies while immobilized with 2.5% isoflurane anesthesia and medical air as the carrier gas. After spending five min in an induction chamber, the fully anesthetized mouse was positioned on a customized platform in the ventral recumbent position. The mouse’s head was set in a face mask that allowed the gas to be scavenged, and the platform was placed off the stage of the SARRP. Using the SARRP Control Interface, a cone-beam CT image was initiated with the X-ray tube operating at 65 kV, 0.5 mA with aluminum filtration. To image the spleen and calculate the prescribed dose, we administered 200 μL of a blood pool contrast agent through tail vein injection (Fenestra HDVC, MediLumine) 3 min before acquiring the CBCT image. The images were reconstructed with Xstrahl’s MuriSlice Software. Dosimetry was performed by measuring EBT2 Gafchromic film exposure at depth using solid-water phantom material with a Microtek Artixcam M1 camera. For image-guided proton RT, the surgically implanted radiopaque marker was visualized, and an isocenter was selected for targeted RT. Once the isocenter was determined, the robotic stage moved the animal to the proper location. The 12 Gy single proton RT dose was delivered using a round 10 × 10 mm collimated beam.

Tumor Growth Measurements and Mouse Survival.

Tumor size was measured with calipers three times per week, and the volumes were calculated as follows: volume = (length in millimeters × width² in millimeters)/2 (11, 61). For survival analysis, mice were killed when endpoints were reached according to the animal protocol. Tumor growth delay was assessed by measuring and plotting the time for each tumor to reach 3, 5, and 7 times its initial volume from the first RT for the Primary tumors and first CAR T treatment for Secondary tumors. For the orthotopic PDAC tumor mouse model, mice were killed 2 wk after image-guided proton RT; tumors were harvested along with the entire remaining pancreas, weighed, and photographed.

Single-Cell Suspension from Tumors and Spleens.

Processing of single-cell suspensions from tumors and spleen was previously described (11, 61). Briefly, tumor tissue was minced into 3 to 5 mm pieces following tumor harvest using scalpels and razor blades (11, 61). Minced tissue was then incubated in DMEM (1×, Gibco) supplemented with 1× Collagenase/Hyaluronidase (STEMCELL) and DNase I Solution (1 mg/mL, STEMCELL) at 37 °C shaking at 200 rpm for 20 min (11). Prior to use for staining and flow cytometer analysis, red blood cells were lysed using ACK Lysis Buffer (Life Technologies) (11, 61). Spleens were minced by utilizing the flat end of a syringe plunger, and red blood cells were lysed using ACK Lysis Buffer (Life Technologies) before staining and flow analysis (11, 61).

CAR T Cell Staining of Peripheral Blood.

Peripheral blood of C57BL/6 mice was obtained by cardiac puncture and stained, and the numbers of cells were quantified using TruCount tubes (BD Biosciences) according to the manufacturer’s instructions (11, 61).

Cytokine Analysis.

The Th1/Th2/Th9/Th17/Th22/Treg Cytokine 17-Plex Mouse ProCAR TaPlex™ Panel (Invitrogen) was used to detect mouse cytokines from the serum of C57BL/6 mice, which was submitted to the Human Immunology Core (HIC) at the University of Pennsylvania for Luminex assay (11, 61).

Pathological Assessment, Immunohistochemistry, and RNA-ISH.

Tumors of C57BL/6 mice were prepared for standard pathological analysis by the Pathology Core laboratory within the Children’s Hospital of Philadelphia (11, 61). Formalin-fixed tissues were routinely processed for paraffin embedding, sectioning, and hematoxylin and eosin (H&E) staining (11, 61). A board-certified veterinary pathologist analyzed H&E stained tissue sections and assessed for the following parameters in a semiquantitative manner (11, 61): tumor necrosis, intra- and peritumoral inflammatory infiltrates, presence of intratumoral stroma, edema, and hemorrhage, and presence of karyomegalic neoplastic cells. RNA-ISH to visualize RNA molecules in individual cells of formalin-fixed, paraffin-embedded tumor tissues was performed according to the manufacturer’s instructions (RNAscope, ACDBio) (11, 61). The following RNAscope probes were used: Mm-Msln (CAT: 443241, ACDBio), Mm-Cd3e-C2 (CAT:314721-C2), vMSGV-C3 (CAT:1140971-C3, ACDBio), and 4-plex Negative Control Probe (CAT: 321831, ACDBio) (11, 61). The RNAscope slides were scanned by the Comparative Pathology Core at the University of Pennsylvania School of Veterinary Medicine using the Aperio Versa 200 whole slide scanner (Leica Biosystems) (11, 61).

Immunofluorescence Staining for MSLN.

Eight days after RT, tumor tissues were cut into 10-μm thick sections, coded, and stored at −80 °C. Slides were thawed at room temperature and subsequently fixed with 2% paraformaldehyde (PFA) for 20 min (11, 61). After three washes with Tris-buffered saline (TBS), tissues were blocked with 8% bovine serum albumin (BSA) and 1% donkey serum in TBS-T (0.025% Triton X-100) at room temperature for 1 h (61). The primary antibodies rabbit anti-mouse CD31 (CAT: NB100-2284; Novus) and rat anti–mouse MSLN (CAT: LS–C179484; LSBio), were incubated overnight at 4 °C at a dilution of 1:50. After three washes with TBS-T, the secondary antibodies donkey anti-rabbit (CAT: A-21207; Invitrogen) goat anti-rat (CAT: A-11006; Invitrogen) were added at 1:200 and incubated for 1 h at room temperature in a humidified chamber. After three 5 min rinses with TBS, tissues were stained with 1 μg mL–1 Hoechst (CAT: H3570; Invitrogen) for 30 min at room temperature, washed with TBS, and coverslips were mounted with antifade mounting medium (11, 61).

Immunofluorescent Staining for Double-Stranded DNA Breaks.

Tissues from PDAC tumors and spleens embedded in optimal cutting temperature compound were processed and stained for DNA double-strand breaks as described previously (64). Briefly, the tissue sections were fixed with 2% PFA in Phosphate-buffered saline (PBS) for 20 min and then washed twice in PBS for 15 min each. Subsequently, the sections were treated with 70% ethanol at −20 °C, followed by three washes in PBS for 10 min each. Next, the sections were blocked in 8% BSA in PBS containing 0.5% Tween-20 and 0.1% Triton X-100 (PBS-TT) for 1 h. After a 5-min wash with PBS-TT, the excess buffer was removed, and gamma-histone H2AX (1:50; CAT: 16-202A MilliporeSigma) diluted in 1% BSA in PBS-TT was added to each section. The slides were then incubated for 2 h at room temperature in a humidified staining trough. A negative control without primary antibody was included in all cases. Following this, the sections were washed three times with PBS-TT for 5 min each, then counterstained with 100 microliters of 5 μg/mL of Hoechst 33342 (Molecular Probes) for 30 min, washed in PBS-TT, and finally mounted in Vectashield medium (CAT: H-1000, Vector Laboratories).

Flow Cytometry and Antibodies.

All markers were stained in 1× Dulbecco’s Phosphate-buffered saline (DPBS) (Gibco) containing 5% heat-inactivated FBS (Seradigm) (11, 61). Monoclonal rat anti–mouse MSLN antibody (CAT: LS–C179484; LSBio), PE mouse anti-rat IgG2a antibody (CAT: 12-4817-82; Invitrogen), and Fluorescein isothiocyanate (FITC) anti-mouse CD326 (Ep-CAM; CAT: 118207; BioLegend) were used to stain for mouse MSLN expression of the murine tumor cells PDA7940b (11, 61). Isotype antibody (CAT: LS-C292311; LSBio) was employed as a control. Mouse CAR T, endogenous T cells, B cells, NK cells, macrophages, DCs, and MDSCs were detected in peripheral blood, spleen, or tumors of syngeneic mice by using the following antibodies (11): Alexa Fluor 700 anti-mouse CD45 (CAT: 103128, BioLegend), Brilliant Violet 650 anti-mouse CD3 (CAT: 100229, BioLegend), Brilliant Violet 510 anti-mouse CD45.1 (CAT: 110741, BioLegend), Brilliant Violet 785 anti-mouse CD19 (CAT: 115543, BioLegend), PE/Dazzle 594 anti-mouse NK1.1 (CAT: 108748, BioLegend), PerCP/Cyanine 5.5 anti-mouse F4/80 (CAT: 123128, BioLegend), Alexa Fluor 647 anti-mouse CD86 (CAT: 105020, BioLegend), Brilliant Violet 510 anti-mouse I-A/I-E (CAT: 107636; BioLegend), PE anti-mouse CD206 (MMR; CAT: 141706; BioLegend), PE/Cyanine7 anti-mouse CD11c (CAT: 117318; BioLegend), Brilliant Violet 711 anti-mouse/human CD11b (CAT: 101242, BioLegend), Brilliant Violet 421 anti-mouse Ly-6C (CAT: 128032, BioLegend), and Alexa Fluor^® 488 anti-mouse Ly-6G (CAT: 127626; BioLegend). To detect the activation markers PD-1, TIM-3, and LAG-3 on T cells, the following antibodies were used: Brilliant Violet 421™ anti-mouse CD279 (PD-1; CAT: 135218; BioLegend), PerCP/Cyanine5.5 anti-mouse CD366 (TIM-3; CAT: 134012; BioLegend), and PE/Dazzle™ 594 anti-mouse CD223 (LAG-3; CAT: 125224; BioLegend). TruStain FcX anti-mouse CD16/32 (CAT: 101320, BioLegend) was used before staining single-cell suspensions of mouse tumors and spleen (11, 61). Cells were stained for viability using Invitrogen™ LIVE/DEAD™ Fixable Near-IR Dead Cell Stain Kit (CAT: L10119; Invitrogen) according to the manufacturer’s instructions (11, 61). Where indicated, CountBright Absolute Counting Beads (CAT: C36950, Invitrogen) were used to get absolute cell counts as per the manufacturer’s instructions. All data were collected by an LSRFortessa cytometer (BD Biosciences) equipped with FACSDiva software (BD Biosciences) (11, 61).

Statistical Analysis.

Statistical analysis was performed using Prism version 9 (GraphPad Software) (11, 61). The legend of each figure denotes the statistical test used. Survival curves were drawn using the Kaplan–Meier method, and the differences of the two curves were compared with the log-rank Mantel–Cox test (11, 61). The Mann–Whitney U test was employed to compare two groups, and the Kruskal–Wallis one-way ANOVA for multiple comparisons was used to compare three or more groups (11, 61). For all figures, ns (nonsignificant) indicates P > 0.05, * indicates P ≤ 0.05, ** indicates P ≤ 0.01, *** indicates P ≤ 0.001, and **** indicates P ≤ 0.0001. The graphs were created using Prism version 9 (GraphPad Software) and PowerPoint (Microsoft).

Supplementary Material

Appendix 01 (PDF)

Movie S1.

Cone-beam CT of implanted marker for IGRT. Reconstructed cone-beam CT image of a marker implanted in a PDAC tumor in the pancreas tail for image-guided proton RT.

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.