SBRT embedded in low-dose RT plus αPD-1 (immuno-EclipseRT, iERT) elicits CD8+ T cell immunity against bulky tumors via an IFN-I/NK/DC axis

Ren Luo; Min Yu; Kai Kang; Yufeng Zhang; Shanghai Liu; Zhuoran Yao; Zichong Peng; Xuanwei Zhang; Shuangsi Liao; Zhou Su; Linglu Yi; Hui Wang; Wenhao Zhao; Yue Zheng; Guo Lin; Ruizhan Tong; Feifei Na; Youling Gong; Meijuan Huang; Bingwen Zou; Long Bai; Renming Zhong; Di Yan; Gabriele Niedermann; Jianxin Xue; You Lu

doi:10.1038/s41467-026-73683-z

. 2026 Jul 8;17:7568. doi: 10.1038/s41467-026-73683-z

SBRT embedded in low-dose RT plus αPD-1 (immuno-EclipseRT, iERT) elicits CD8⁺ T cell immunity against bulky tumors via an IFN-I/NK/DC axis

Ren Luo ^1,^2,^3,^#, Min Yu ^1,^#, Kai Kang ^1,^3,^#, Yufeng Zhang ^1,^2,^#, Shanghai Liu ^1,², Zhuoran Yao ^1,³, Zichong Peng ^1,³, Xuanwei Zhang ^1,², Shuangsi Liao ¹, Zhou Su ^1,⁴, Linglu Yi ^3,⁵, Hui Wang ¹, Wenhao Zhao ¹, Yue Zheng ¹, Guo Lin ¹, Ruizhan Tong ^1,^3,⁵, Feifei Na ^1,², Youling Gong ^1,², Meijuan Huang ¹, Bingwen Zou ^1,², Long Bai ⁶, Renming Zhong ⁶, Di Yan ⁶, Gabriele Niedermann ⁷, Jianxin Xue ^1,^3,^✉, You Lu ^1,^2,^3,^5,^✉

PMCID: PMC13415869 PMID: 42420294

Abstract

Bulky tumors remain challenging to treat, and immune checkpoint inhibitors (ICIs), alone or combined with conventional radiotherapy (RT), yield limited efficacy. We present EclipseRT (ERT), an RT technique that delivers low-dose RT (LDRT) to the gross tumor volume (GTV) and stereotactic body RT (SBRT) to selected subvolume(s) within the GTV. Combined with ICIs (iERT), this approach achieves marked control of bulky tumors through the coordinated activity of NK and CD8⁺ T cells. Single-cell RNA sequencing and validation experiments show that the SBRT component robustly induces type I interferon (IFN-I), which activates NK cells to secrete XCL1, thereby recruiting cross-presenting XCR1⁺ dendritic cells (DCs). SBRT also promotes the release of extracellular vesicles carrying neoantigens, enhancing DC cross-presentation and CD8⁺ T-cell responses. The LDRT component further promotes NK and CD8⁺ T-cell recruitment. iERT also induces precursor exhausted CD8⁺ T cells in tumors and tumor-draining lymph nodes. Collectively, iERT activates the IFN-I/NK/DC/CD8⁺ T-cell axis, driving potent antitumor immunity against bulky tumors.

Subject terms: Tumour immunology, Cancer immunotherapy, Radiotherapy

Bulky tumors are often refractory to immune checkpoint blockade, and effective radioimmunotherapy strategies remain limited. The authors show that immuno-EclipseRT, which consists of low-dose radiotherapy (LDRT) to the whole tumor, partial-tumor stereotactic body radiation therapy (SBRT) and PD-1 blockade, controls bulky tumors via an IFN-I driven XCL1⁺ NK/XCR1⁺ cDC/CD8⁺ T cell axis.

Introduction

Immune checkpoint inhibitors (ICIs) have transformed cancer treatment, yet only about 20% of patients respond (1). Therefore, potential combination partners are under intense investigation. Radiotherapy (RT) can exert antitumor immunity both locally within the irradiated field and systemically beyond it, known as the abscopal effect^1,2. However, attempts to integrate conventional RT with ICIs have largely failed^3–6. This lack of effectiveness may be due to suboptimal study design, inappropriate selection of patient populations, or the administration of standard RT schedules and target volumes^1,2. Adaptation of RT to immunotherapy, named adaptive immuno-radiotherapy (AIRT), represents an emerging RT approach with potential clinical relevance^1,2. As a proof-of-concept, we focus on designing a new treatment strategy for bulky solid tumors (≥7 cm), which are associated with poor prognosis under conventional therapies (e.g., RT and chemotherapy) and show limited response to ICIs^7,8, making them an ideal model for studying AIRT. This poor response may be due to the unfavorable tumor microenvironment (TME) of bulky tumors, characterized by low infiltration of natural killer (NK) cells, T cells, and antigen-presenting cells (APC), together with increased numbers of immunosuppressive cells^7,8. In addition, tumor-infiltrating T cells are often severely exhausted and dysfunctional due to chronic tumor antigen stimulation⁸. In patients, the time to response to ICIs is relatively long (median 8–12 weeks); therefore, cytoreductive therapies should be considered in combination with ICIs to bridge this period and foster an immune response⁹. However, in advanced bulky tumors, chemotherapy is less effective even when combined with ICIs^10,11, as these tumors often develop following the failure of multiple chemotherapy lines¹². Increased tumor hypoxia also renders bulky tumors intrinsically resistant to conventionally fractionated radiation therapy (CFRT, 1.8–2 Gy per fraction)¹³. Whole-tumor high-dose or curative RT for large tumors induces significant radiation-induced toxicity (≥ grade 3), and combined high-dose radiochemotherapy would be less tolerable for such patients^14–16.

Stereotactic body radiation therapy (SBRT), delivering 1–5 large fractions (e.g., 5–30 Gy per fraction), can effectively kill CFRT-resistant tumor cells and damage the vasculature, causing secondary tumor cell death^17,18. In addition, it can induce immunogenic cell death (ICD), thereby stimulating tumor-specific CD8⁺ T cell-dependent antitumor immunity^19–21. SBRT can also induce the T cell–dependent abscopal effect^21,22. For early-stage (I-II) non-small cell lung cancer (NSCLC), SBRT is the standard of care for patients who have peripheral tumors less than 5 cm in diameter, without lymph node invasion, who are either ineligible for surgery or have declined it. A phase II trial demonstrated that adding nivolumab (an αPD-1 antibody) to SBRT significantly improved 4-year event-free survival in individuals with early-stage NSCLC²³. For oligometastatic patients, SBRT combined with ICIs has also been applied with curative intent²⁴. Despite its effectiveness, SBRT is generally not recommended for tumors larger than 5 cm due to the risk of increased toxicity^25,26.

Low-dose radiation therapy (LDRT) with a fraction dose ≤ 2 Gy and a total dose of no more than 6 Gy can inflame the TME by polarizing macrophages to the M1 phenotype and increasing tumor infiltration by NK cells, NK group 2D (NKG2D)⁺ CD4⁺ T cells, and CD8⁺ T cells^27–30. Using bilateral murine tumor models, we and another group have shown that applying SBRT to primary tumors and delivering LDRT to secondary (abscopal) tumor lesions, combined with ΙCI(s), can markedly enhance the abscopal effect^28,29. This effect depends on CD8⁺²⁸ and/or CD4⁺ T cells²⁹ and is associated with the induction of the T lymphocyte chemoattractant CXCL10 in LDRT-irradiated abscopal tumors²⁸. Our recent phase I clinical trial (NCT03812549) demonstrated that this strategy was safe and effective in treatment-naïve metastatic PD-L1⁺ NSCLC patients³¹. However, LDRT is unlikely to be tumoricidal to solid tumors.

Hence, there remains an urgent need for innovative RT techniques that can enhance the therapeutic index in patients with bulky tumors. Spatially Fractionated Radiation Therapy (SFRT) has been proposed as a treatment option for bulky tumors^32–34. Recent preclinical studies have demonstrated that spatially heterogeneous RT, delivered via brachytherapy³⁵ or by treating one half of the tumor with SBRT (16 Gy × 1) and the other half with LDRT (2 Gy × 1)³⁶, can potentiate ICI efficacy. These important studies proposed the concept that heterogeneous intratumor RT can promote spatially diversified antitumor immune responses. Our previous work showed that LDRT could enhance ICI efficacy in mice and patients^28,31. To overcome the challenge of bulky tumors, we combined fractionated SBRT (10 Gy × 3) and LDRT (2 Gy × 3) to develop an RT technique, termed “EclipseRT” (ERT). Its dose distribution map resembles an ‘eclipse,’ as it concurrently delivers LDRT to the entire gross tumor volume (GTV) and SBRT to subvolume(s) within the same GTV. When combined with ICI, we termed this approach immuno-ERT (iERT). This strategy demonstrated promising clinical benefits in our unpublished retrospective cohort of 39 patients with thoracic and abdominal bulky tumors³⁷. These encouraging results prompted us to further investigate the underlying mechanisms and initiate prospective clinical trials (NCT05615142 and NCT06349837).

In this study, in addition to evaluating iERT efficacy in mouse models of bulky tumors, we dissect the underlying immunological mechanisms in depth. Fractionated iERT, but not single-fraction iERT, achieves significant control of bulky tumors in mice, and this effect depends on NK and CD8⁺ T cells. iERT induces type I interferon (IFN-I) signaling, which in turn promotes a distinct XCL1⁺ NK-cell subset that acts as an upstream driver of the XCL1⁺ NK/XCR1⁺ dendritic cell (DC)/CD8⁺ T-cell axis against bulky tumors. Furthermore, SBRT-induced extracellular vesicles (EVs) carry tumor neoantigens to DCs, enhancing their capacity to prime CD8⁺ T-cell responses. Our findings reveal a critical role for NK cells in initiating effective RT–ICI responses and provide a rationale for future clinical trials evaluating iERT in patients with bulky tumors.

Results

iERT achieves marked control of bulky tumors in mice

A high-precision small-animal irradiation machine with a treatment planning system (see Methods) was used to apply the ERT technique to bulky CT26 colon carcinoma, Lewis lung carcinoma (LLC1), and B16-OVA melanoma tumors (400–900 mm³, Fig. 1a). In ERT, LDRT (1 or 3 fractions × 2 Gy) was delivered to the whole tumor, while SBRT (1 or 3 fractions × 10 Gy) was delivered simultaneously to the tumor center (Fig. 1a). Three sectional views illustrating the ERT dose distribution are shown in Supplementary Fig. 1a, and the corresponding dose-volume histogram (DVH) parameters are summarized in Supplementary Table 1. The SBRT-irradiated area was marked with green tissue marking ink (Fig. 1b–d). The green marking ink also enabled precise discrimination and sampling of the SBRT and LDRT zones for subsequent experiments, including single-cell RNA sequencing (scRNA-seq), T-cell receptor sequencing (TCR-seq), and flow cytometry (also see “iERT for mice” Methods). To confirm the accuracy of ERT, we performed γH2AX immunofluorescence (IF) staining and observed that γH2AX foci were more abundant in SBRT zones than in LDRT zones and untreated tumors (Fig. 1e–g). 53BP1 IF staining and comet assay also corroborated the γH2AX results (Supplementary Fig. 1b). Western blotting and enzyme-linked immunosorbent assay (ELISA) for phospho-H2AX (Ser139) further validated the accuracy of our sampling method in separating the SBRT and LDRT zones (Fig. 1h, i). SBRT (10 Gy/fraction) and LDRT (2 Gy/fraction) were chosen based on our and others’ preclinical studies demonstrating their respective strengths in eliciting immunogenic and immunomodulatory antitumor effects^{28–30,38–40}.

Fig. 1 — a Dose distribution in a transverse section of LDRT, partial SBRT and ERT, along with the treatment scheme for mice. b, c Mouse positioning (b) and iERT dose distribution (c) of a bulky tumor. The SBRT (red dashed line) and LDRT (blue dashed line) zones were sampled along the sagittal plane, and the intermediate tissue at their interface (at least 2 mm in thickness, c, left, turquoise dashed line) was carefully removed. The intermediate tissue at the SBRT and LDRT interface along the transverse section (turquoise dashed line, c right and d left panels) was carefully removed. d Green marking ink on tumor (d left) and H&E staining (d, middle and right) confirming the SBRT-irradiated region. e, f γH2AX (green) immunofluorescence demonstrating DNA damage in iERT-treated tumors with distinct LDRT and SBRT zones, compared with untreated controls (f). Nuclei: DAPI (blue). The orange dashed line approximately delineates the SBRT zone. g Quantification of γH2AX foci in tumors 15 min post-ERT versus untreated tumors (n = 6 biologically independent samples per group, representative images shown in e and f). h, i CT26 tumors treated with ERT were dissected ex vivo into LDRT and SBRT zones. h Western blotting analysis of phospho-H2AX (Ser139) and GAPDH, with quantitative densitometry shown on the right (representative image from three independent experiments in which 6 different untreated and ERT-treated tumors were analyzed). i ELISA-based measurement of γH2AX levels presented as fold change relative to untreated tissue (n = 5 biologically independent samples per group). j–l Tumor growth curves for bulky CT26 (j, n = 11 mice per group), LLC1 (k, n = 5 mice per group) or B16-OVA (l, for untreated, n = 7 mice; for others, n = 6 mice per group) tumors in mice treated with partial SBRT/αPD-1, LDRT/αPD-1, or iERT. m–o Survival of mice bearing CT26 (m, n = 11 mice per group), LLC1 (n, n = 5 mice per group) or B16-OVA (o for untreated, n = 7 mice; for others, n = 6 mice per group) tumors. Data are presented as mean ± SEM. P- values (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001) were determined by one-way ANOVA (g–i), a two-tailed t test (j–l) and a log-rank test (m–o). Source data are provided as a Source Data file. a left Created in BioRender. https://biorender.com/7kwswlo.

In the bulky CT26 tumor model, αPD-1 plus one fraction of ERT (whole-tumor 2 Gy and partial-tumor 10 Gy) only slightly improved tumor control compared with untreated mice and extended survival compared with untreated mice and mice treated with 2 Gy + αPD-1 (Supplementary Fig. 1c). Escalating RT fractions to three, LDRT/αPD-1 and partial SBRT/αPD-1 showed moderate antitumor effects, delaying tumor growth and extending mouse survival in the bulky CT26, LLC1, and B16-OVA tumor models (Fig. 1j–o and Supplementary Fig. 1d). Notably, iERT controlled bulky tumors more effectively (Fig. 1j–l) and prolonged median survival compared with LDRT/αPD-1 or SBRT/αPD-1 therapy in these models (Fig. 1m–o). In the bulky CT26 model, iERT achieved a 27.3% (3 out of 11) cure rate, which was not observed with LDRT/αPD-1 or SBRT/αPD-1 (Fig. 1j and m and Supplementary Fig. 1d). No weight loss or significant blood toxicities were detected (Supplementary Fig. 1e–g). To compare toxicity profiles, we treated CT26 tumor-bearing mice with CFRT/αPD-1 (2 Gy × 25 fractions; α/β = 10 Gy, biologically effective dose = 60 Gy, equivalent to 10 Gy × 3) or iERT. iERT had minimal effects on leukocytes, neutrophils, and lymphocytes, whereas CFRT/αPD-1 significantly reduced all these populations, particularly lymphocytes, which are critical for ICI efficacy¹ (see Supplementary Fig. 1g). In addition, CFRT/αPD-1-treated mice developed visible skin damage approximately 2–3 weeks post CFRT, whereas no such toxicity was observed in the iERT-treated group (see Supplementary Fig. 1h). These findings suggest that iERT may reduce hematologic and skin toxicity compared with CFRT/αPD-1.

Of note, ERT alone was more effective than either partial SBRT alone or LDRT alone in treating bulky CT26 tumors. In this αPD-1 non-responsive bulky tumor model (Supplementary Fig. 1i), adding ERT to αPD-1 (iERT) improved efficacy over αPD-1 or ERT alone (Supplementary Fig. 1i). These results indicate that ERT is effective on its own and even more potent when combined with αPD-1 (iERT).

Overall, three-fraction iERT was more effective than single-fraction iERT. iERT outperformed LDRT or partial SBRT combined with αPD-1 in controlling bulky mouse tumors and showed lower blood and skin toxicity than CFRT combined with αPD-1.

iERT elicits early NK cell and subsequent CD8⁺ T cell responses

We performed flow cytometry, multiplex immunohistochemistry (mIHC), and scRNA-seq to evaluate immune cell infiltration in the TME at baseline (d0) and on days 4, 7, 10, and 16 post-RT (Fig. 2a). Green marking ink was applied to the SBRT-irradiated zone before tumor excision, remaining stable during tissue processing and sectioning to enable accurate separation of SBRT and surrounding LDRT regions for downstream analyses (Fig. 1b–i and Supplementary Fig. 1b; see “iERT for mice” “Methods”). Flow cytometry of single-cell suspensions from these dissected regions revealed an early (d4) increase in NK cells, but not CD8⁺ T cells, the two main effector populations for tumor cell elimination (Fig. 2b). CD8⁺ T cells significantly increased only at d10. To capture temporal dynamics from baseline (d0) to a later time point (d16), serial fine-needle biopsies were collected at d0, d4, d10, and d16 post-iERT, following a protocol adapted from Sitnikova et al.⁴¹. Flow cytometry analysis of these biopsies showed distinct kinetic patterns (Supplementary Fig. 2a). NK cells showed a marked increase at d4, suggesting early activation, whereas CD8⁺ T cells peaked at d10, indicating a delayed adaptive response. These results closely paralleled those observed in dissected bulky tumors at d4 and d10 (Fig. 2b).

To further investigate the most suitable time point to study CD8⁺ T cells induced by iERT, mIHC was applied (Fig. 2c). As shown in the left two panels of Fig. 2d, e, LDRT/αPD-1 resulted in the highest increase in CD8⁺ T cells in SBRT zones (tumor center) of bulky CT26 tumors on d7. Three days later (d10), iERT therapy induced a significantly greater increase in CD8⁺ T cells in both the SBRT and LDRT zones of bulky tumors (middle two panels of Fig. 2d, e). On d16, the number of CD8⁺ T cells remained the highest in the iERT group in the SBRT zones but not in the LDRT zones (right two panels of Fig. 2d, e).

Based on these results, the d4 and d10 time points were selected for scRNA-seq analysis of CT26 tumors from each treatment group. After quality control, 138,576 cells were categorized into different cell types based on their typical marker genes (Fig. 2f and Supplementary Fig. 2c). Consistent with the flow cytometry data (Fig. 2b), on d4, iERT (indicated by green, Fig. 2g) induced more NK cells but not CD8⁺ T cells than other treatment groups. However, on d10, iERT induced a more significant increase in CD8⁺ T cells (Fig. 2g) in tumors compared to d4 and other treatment groups. These findings showed that iERT induces early NK cell infiltration followed by subsequent CD8⁺ T cell infiltration in bulky tumors, suggesting a pattern similar to that seen in viral infections, where NK cells act as initial responders (Fig. 2h, i)⁴².

No differences were observed among innate immune cells, including macrophages (M1, M2, CXCL9⁺) or neutrophil subsets with potential antitumor activity (CXCL9⁺⁴³ and MHCII⁺⁴⁴) among the groups at d4 by flow cytometry (Supplementary Fig. 2d, e). This suggests their limited involvement in iERT efficacy early after treatment.

To assess T cell clonality over time following iERT, TCR-seq was performed on tumors collected at d0, d4, d10, and d16. Normalized clonotype numbers increased from baseline, peaked at d7, and declined by d10–d16, indicating early clonal expansion followed by contraction (Supplementary Fig. 2f, left). The normalized Shannon diversity index, initially low, rose sharply at d7 and remained elevated through d16, consistent with repertoire remodeling (Supplementary Fig. 2f, middle). The Morisita similarity index steadily increased from d0 to d16, suggesting the emergence of shared, immunodominant clones across tumors and a strong antigen-driven T cell response (Supplementary Fig. 2f, right). These kinetics indicate that the interval between early innate activation at d4 and peak adaptive T cell responses around d10 may represent an optimal therapeutic window for combining iERT with synergistic treatments.

In conclusion, iERT triggers early NK cell recruitment followed by a sustained adaptive response characterized by robust CD8⁺ T cell infiltration.

iERT induced tumor infiltration by chemokine/cytokine-secreting NK cells

Analysis of our scRNA-seq dataset identified five distinct NK cell subsets based on typical marker genes (Fig. 3a and Supplementary Fig. 3a). Among these, Xcl1⁺ and Ccl5⁺ NK cells expressed the highest level of Ifng (Supplementary Fig. 3b). At day 4 after iERT initiation, both the SBRT and LDRT zones showed the most pronounced increases in the frequencies of Xcl1⁺ and Ccl5⁺ NK cells among the five subpopulations (Fig. 3b).

Flow cytometric analysis at d4 showed that LDRT/αPD-1 and iERT increased infiltration of bulk NK cells as well as IFNγ⁺ and CCL5⁺ NK cells into both the SBRT and LDRT zones of bulky tumors (Fig. 3c, d). In contrast, SBRT/αPD-1 increased these NK cell populations only in the SBRT zone with no increase in the LDRT zone. These findings suggest that LDRT is required for NK cell infiltration – both bulk and IFNγ⁺/ CCL5⁺ subsets – throughout the entire bulky tumor.

As commercial flow cytometry antibodies for XCL1 are not available, XCL1⁺ NK cells at d4 were detected using mIHC (Fig. 3e). In SBRT/αPD-1- or iERT-treated tumors, the SBRT zones showed increased XCL1⁺ NK cells (Fig. 3f, left). In addition, the tumor periphery (LDRT zones) displayed elevated XCL1⁺ NK cells across all three RT modalities when combined with αPD-1 (Fig. 3f, right). These findings suggest that SBRT is required to drive infiltration of XCL1⁺ NK cells into tumor centers (SBRT zones), which are typically hypoxic in bulky tumors¹³.

At d10, bulk NK and CCL5⁺ NK cell populations were higher in iERT-treated tumors, whereas IFNγ⁺ NK cell numbers did not show a similar increase (Supplementary Fig. 3d, e).

To assess the functional importance of NK cells for iERT efficacy, NK cell depletion was performed using αGM1 in Balb/c mice with CT26 tumors and αNK1.1 in C57BL/6 mice with LLC1 tumors. Depletion effectively eliminated NK cells without significantly altering other immune populations, including DCs, CD4⁺ and CD8⁺ T cells, or myeloid cells (Supplementary Fig. 3f, g). NK cell depletion resulted in poorer tumor control in both CT26 (Fig. 3g and Supplementary Fig. 3i) and LLC1 (Fig. 3h and Supplementary Fig. 3j) bulky tumor models.

Together, these results demonstrate that iERT promotes the infiltration of NK cells secreting XCL1, CCL5 and IFNγ into bulky tumors and that NK cells are essential for iERT efficacy.

iERT induces IFN-I which drives NK cell activation

Immune Response Enrichment Analysis (IREA)⁴⁵ revealed a highly active IFN-I response in the Xcl1⁺ and Ccl5⁺ NK cell subsets within the TME, whereas Gene Set Variation Analysis (GSVA) indicated high IFN-I scores only in Xcl1⁺ NK cells (Fig. 4a and Supplementary Fig. 4a). GSVA further revealed that Xcl1⁺ NK cells scored high for NK activation, degranulation, cytokine production, and NK-mediated immune response to tumor cells (Supplementary Fig. 4a). In vitro, co-culture of splenic naïve NK cells with irradiated tumor cells (10 Gy × 3) increased NK cell IFNγ production, indicating activation (Fig. 4b). RT is known to induce IFN-I, and indeed, the radiation – particularly 10 Gy × 3 – significantly increased Ifnb1 and its downstream response marker Cxcl10 in CT26 tumor cells (Fig. 4c). Bulk RNA-seq confirmed an IFN-I response after radiation both in vitro (Supplementary Fig. 4b) and ex vivo (Supplementary Fig. 4c). To determine whether IFN-I is a key ICD molecule activating NK cells, antibodies against interferon alpha and beta receptor subunit 1 (IFNAR1), high mobility group box 1 (HMGB1), or calreticulin were added 30 min before co-incubation of irradiated tumor cells and naïve NK cells. The reduction in CT26 cell death, the proportion of IFNγ-positive NK cells and the mean fluorescence intensity (MFI) of IFNγ was observed only upon IFNAR1 blockade (Fig. 4d). RNA-seq showed that IFNβ1 stimulation of NK cells increased Xcl1, Ifng, and, to a lesser extent, Ccl5 expression (Fig. 4e) and upregulated pathways for cytokine production and antiviral responses (Supplementary Fig. 4d).

To broaden cytokine/chemokine profiling, we performed a multiplex assay (Supplementary Fig. 4e), which revealed iERT-induced upregulation of a wide spectrum of cytokines and chemokines (Supplementary Fig. 4f), including Th1 cytokines (IL-2, TNF, IFNγ, IL-12p70), myeloid activation markers (IL-1β, IL-18), and chemokines involved in myeloid and lymphoid recruitment (CCL2, CCL4, CXCL10). CXCL9, absent from the assay panel, was assessed by RNA-seq (data from Supplementary Fig. 4c) and found to have the highest expression in the iERT group (Supplementary Fig. 4g).

In vivo, IFNAR1 blockade reduced the efficacy of iERT against bulky CT26 tumors (Fig. 4f) and, as revealed by mIHC, decreased XCL1⁺ NK cells in both SBRT and LDRT zones (Fig. 4g). Flow cytometry showed that NK depletion or IFNAR1 blockade reduced bulk, IFNγ⁺ and CCL5⁺ NK cells in the SBRT zone d4 post RT initiation (Fig. 4h, i). CCL5⁺ NK cells also decreased in the LDRT zone (Fig. 4h, i). At d10, both NK depletion and IFNAR1 blockade reduced bulk NK, IFNγ⁺ and CCL5⁺ NK cells (Fig. 4h, i). Notably, IFNAR1 blockade did not significantly impair SBRT/αPD-1 or LDRT/αPD-1 efficacy (Supplementary Fig. 4h).

Collectively, in vitro and ex vivo experiments modeling SBRT, the high-dose component of ERT, demonstrated that SBRT-induced IFN-I is a key driver of NK cell activation and cytokine/chemokine production. In vivo, IFNAR1 blockade reduced infiltration of bulk NK cells as well as XCL1⁺, IFNγ⁺, CCL5⁺ NK cell subsets, and diminished the therapeutic efficacy of iERT.

iERT recruits XCR1⁺ cDC1s through NK-derived XCL1

XCL1 is a key chemokine for recruiting XCR1⁺ cDC1s to the TME⁴⁶, a process essential for subsequent CD8⁺ T cell-mediated antitumor immunity. In our scRNA-seq data, cDCs were classified into four subsets based on typical marker genes (Fig. 5a and Supplementary Fig. 5a). Among these, Xcr1⁺ cDC1s had the highest GSVA scores for antigen processing and presentation pathways (Fig. 5b).

At d4 post-RT initiation, cross-presenting XCR1⁺ cDC1s were most abundant in the SBRT zones, but not LDRT zones, of iERT-treated tumors (Fig. 5c). By d10, an increase in XCR1⁺ cDC1s was also observed in LDRT/αPD-1- and iERT-treated tumors. No significant changes were detected in CCR7⁺ cDC1 or cDC2 among groups on d4 (Supplementary Fig. 5b, c). These findings suggest that XCR1⁺ cDC1 may be the key DC subset contributing to iERT efficacy. Depletion of NK cells reduced the iERT-induced increase in XCR1⁺ cDC1s (Fig. 5c), indicating that NK cells facilitate their infiltration into the TME. cDC2 levels rose at d10 (Supplementary Fig. 5c), coinciding with peak CD8⁺ T cell infiltration and thus potentially contributing to iERT efficacy.

NK cells were the main source of XCL1, while cDCs were the primary DC population expressing XCR1 (Fig. 5d). Blocking the XCL1-XCR1 interaction with an XCL1-neutralizing antibody diminished both iERT efficacy (Fig. 5e) and the density of XCR1⁺ cDC1s in tumors of iERT-treated mice (Fig. 5f). These results indicate that NK cell-derived XCL1, induced by RT-triggered IFN-I, recruits highly cross-presenting XCR1⁺ cDC1s to the TME of bulky tumors during iERT. In contrast, XCL1 blockade did not significantly reduce the efficacy of SBRT/αPD-1 or LDRT/αPD-1 (Supplementary Fig. 5e).

To investigate how DCs present tumor antigens to prime CD8⁺ T cells, we examined tumor-derived EVs. Bone marrow-derived dendritic cells (BMDCs, green) captured more EVs (orange) from SBRT-irradiated CT26 or B16-OVA cells than from non-irradiated cells (Fig. 5g, h). EVs isolated from the supernatant of irradiated CT26 or B16-OVA cells promoted BMDC maturation and CD8⁺ T cell activation (Fig. 5i, j), effects reduced by GW4869, an inhibitor of exosome biogenesis/release⁴⁷ (Fig. 5i, j). CD81⁺ EVs, but not CD9⁺ EVs, increased following SBRT (Supplementary Fig. 5f), suggesting that CD81⁺ EVs may play a more prominent role in RT-induced EV production. In our coculture system, OT-I cells can only be primed via BMDCs presenting OVA-derived peptide SIINFEKL bound to H-2K^b of MHC class I, but not MHC II, indicating that EVs contained neoantigens required cross-presentation by the BMDCs⁴⁸. In addition, SIINFEKL peptide (red, Supplementary Fig. 5g) was detected in both CD81⁺ and CD9⁺ EVs, with a marked increase in CD81⁺ EVs following SBRT (Fig. 5k), reinforcing their role in antigen presentation and CD8⁺ T cell priming. SBRT, but not LDRT, significantly increased EV release (Supplementary Fig. 5h), and together with Fig. 5k, these data indicate that SBRT enhances both EV release and EV cargo content. In vivo, GW4869 administration reduced iERT efficacy, highlighting the importance of EVs in mediating iERT-induced antitumor immunity (Fig. 5l).

Together, these findings demonstrate that iERT recruits XCR1⁺ cDC1s via NK-derived XCL1, with XCL1 neutralization impairing therapeutic efficacy. In vitro experiments indicate that SBRT, the high-dose component of ERT, promotes DC-mediated priming of CD8⁺ T cells through EV-mediated antigen cross-presentation, a mechanism supported in vivo by reduced iERT efficacy following blockade of EV biogenesis.

iERT induces CD8⁺ T cell responses via the IFN-I/NK/XCL1 axis

To explore the heterogeneity of CD8⁺ T cells, scRNA-seq analysis identified six distinct transcriptional states: stem-like precursors of exhausted T cells (Tpex; Tcf7 and Pdcd1), effector-like T cells (Tef; Pdcd1, intermediate Harvcr2), exhausted T cells (Tex; Pdcd1, high Harvcr2), naïve/memory T cells (Lef1, Tcf7, Sell), IFN-stimulated gene (ISG)⁺ T cells, and proliferating T cells (Fig. 6a and Supplementary Fig. 6a, b). Tpex, Tef, and Tex expressed high levels of the exhaustion-associated transcriptional factor Tox (Supplementary Fig. 6b), with Tef and Tex also upregulating effector-related genes, including Ifng, Gzmb, and Prf1. In Fig. 6b, UMAP plots showing the score distributions of stemness, naiveness, cytotoxicity, and exhaustion signatures across CD8⁺ T-cell subsets. Stemness was mainly enriched in the Tpex cluster, whereas cytotoxicity was predominantly enriched in the Tef cluster. T cell exhaustion begins with self-renewing TCF-1⁺ PD-1⁺ Tpex cells, which differentiate into Tef cells and Tex cells⁴⁹. PD-1/PD-L1 blockade can reinvigorate exhausted T cell responses, largely by expanding Tpex cells^50–52. Tef and Tex cells act as the main executors in tumor eradication, although Tex cells are terminally highly dysfunctional^50,52.

a UMAP plot (n = 10,500 cells) showing six CD8⁺ T cell subsets (left) and bubble plot (right) with marker gene expression across subsets. b UMAP plots showing score distributions of each signature; dots are color-coded by estimated kernel density. c Density of AH1⁺ CD8⁺ T cells and subpopulations in SBRT or LDRT zones (d4) across treatments (n = 5 biologically independent samples per group). d Gating strategy for AH1⁺ CD8⁺ T cells and subpopulations corresponding to Fig. 6c, h, i; Fig. 7c. Density of AH1⁺ CD8⁺ T cells and subpopulations in tumor (d10) across treatments (lower panels) (untreated, n = 5; LDRT + αPD-1, n = 6; SBRT + αPD-1, n = 8; iERT, n = 7; biologically independent samples per group). e Representative mIHC images of CT26 tumors (d10 post-iERT initiation) showing TCF-1 (red), PD-1 (orange), TIM-3 (purple) on CD8⁺ (green) T cells and CD11c⁺ (white) cells. f Violin plots of the distance between CD11c⁺ cells and Tpex or more differentiated CD8⁺ T cells (Tdf) in SBRT or LDRT zones of CT26 tumors (d10). Width represents kernel density, dashed lines indicate medians, and dotted lines the 25th/75th percentiles (Tpex to DC (SBRT zone). Untreated, n = 1; LDRT + αPD-1, n = 1; SBRT + αPD-1, n = 0; iERT, n = 279. Tpex to DC (LDRT zone). Untreated, n = 9; LDRT + αPD-1, n = 18; SBRT + αPD-1, n = 31; iERT, n = 220. Tdf to DC (SBRT zone). Untreated, n = 10; LDRT + αPD-1, n = 395; SBRT + αPD-1, n = 11; iERT, n = 4678. Tdf to DC (LDRT zone). Untreated, n = 20; LDRT + αPD-1, n = 1038; SBRT + αPD-1, n = 48; iERT, n = 7753 cells). g Treatment schemes. Tumor growth and survival of mice bearing bulky CT26 tumors treated with iERT ± CD8⁺ T cell-depleting antibodies (LDRT + αPD-1, n = 11; SBRT + αPD-1, n = 11; iERT, n = 11; iERT + CD8⁺ T cell-depleting antibodies, n = 6; biologically independent samples per group). h Representative flow cytometry of AH1⁺ CD8⁺ T cells in tumors (d10) across treatments. The red arrow highlights the highest proportion of AH1⁺ CD8⁺ T cells upon iERT treatment. i Density of AH1⁺ CD8⁺ T cells and subpopulations in the TME (d10) across treatments (untreated, n = 5; LDRT + αPD-1, n = 5; SBRT + αPD-1, n = 5; iERT, n = 5; iERT+NK depletion, n = 5; iERT+αIFNAR1, n = 7; iERT + αXCL1, n = 6; biologically independent samples per group). d, i data are pooled from two independent experiments. Data are presented as mean ± SEM. P-values (ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001) were determined by one-way ANOVA (c, d, f, and i), a two-tailed t test (g, tumor growth curves) and a log-rank test (g, survival curves). Source data are provided as a Source Data file. c left Created in BioRender. https://BioRender.com/r6bqkrw. e upper Created in BioRender. https://BioRender.com/nxm7l07. g left Created in BioRender. https://biorender.com/7kwswlo.

AH1, an H-2L^d-restricted peptide, was identified as the immunodominant neoantigen in CT26 tumors recognized by CD8⁺ T cells⁵³. In flow cytometry analysis, TCF1⁺PD1⁺TIM3^-, TCF-1^-TIM-3⁺PD-1⁺CD101^-, TCF-1^-TIM-3⁺PD-1⁺CD101⁺ CD8⁺ T cells were classified as Tpex, Tef and Tex, respectively, based on previous studies of their markers identified by flow cytometry^50,52. Using the AH1 MHC-I tetramer, we revealed that tumor-specific CD8⁺ T cells were unchanged early after RT (d4, Fig. 6c). Later after RT (d10), the density of AH1⁺ tumor-specific CD8⁺ T cells and their stem-like phenotype (Tpex) were significantly increased in the SBRT and LDRT zones (Fig. 6d). AH1⁺ CD8⁺ Tef and Tex cells were most elevated in the LDRT zone of the iERT group compared to untreated mice and the SBRT/αPD-1 group (Fig. 6d).

Tpex cells, identified as TCF-1⁺ (red) TIM-3^-PD-1⁺ (orange), and the more differentiated CD8⁺ T cells (Tdf, including Tef and Tex), identified by TCF-1^-TIM-3⁺ (purple) PD-1⁺, can be visualized using mIHC (Fig. 6e). Recently, it was shown that DCs can provide appropriate signals for TCF-1⁺ CD8⁺ T cell maintenance and differentiation into TIM-3⁺ CD8⁺ T cells in the TME⁵⁴. Tpex cells were rarely detected in the tumor centers (SBRT zones) of untreated, SBRT/αPD-1, and LDRT/αPD-1 mice, explaining the absence of data in Fig. 6f (Tpex to DC, SBRT zone). In the tumor peripheries (LDRT zones), Tpex cells were closer to DCs (CD11c⁺) in SBRT-containing therapies (iERT and SBRT/αPD-1) compared to LDRT/αPD-1 (Fig. 6f, Tpex to DC, LDRT zone). Moreover, in SBRT-containing therapies, Tdf cells exhibited a shorter distance to DCs (CD11c⁺) in both tumor centers and peripheries compared to LDRT/αPD-1 therapy (Fig. 6f, Tdf to DC). These data indicate that both SBRT and LDRT components of iERT are crucial for optimal DC-T cell interactions.

Single-cell TCR-seq analysis revealed that CD8⁺ T cells isolated from tumors treated with iERT exhibited a higher Shannon diversity index at d4 and a lower index at d10 compared to the untreated group (Supplementary Fig. 6c). This suggests that iERT treatment enhances the diversity of CD8⁺ T cell TCR clones in the TME at an early stage, while promoting the enrichment of dominant clones at later stages. CD8⁺ T cells isolated from tumors treated with iERT at d10 contained a higher proportion of hyperexpanded TCR clonotypes (Supplementary Fig. 6d).

iERT efficacy was strongly dependent on CD8⁺ T cells, as a significant decrease in tumor control was observed when CD8⁺ T cells were depleted using antibodies (Fig. 6g). AH1-specific CD8⁺ T cells were significantly decreased upon NK depletion and the blocking of IFNAR1 or XCL1 (Fig. 6h, i). These data underscore the importance of the NK cell-centric axis, specifically the upstream activator IFN-I and the downstream effector XCL1, for inducing optimal tumor-specific CD8⁺ T cell responses during iERT treatment. To further validate the XCL1⁺ NK/XCR1⁺ DC/CD8⁺ T axis, we analyzed predefined immune interaction axes with established relevance to NK–DC–T cell crosstalk, including XCL–XCR, CCL–CCR, and CXCL–CXCR ligand–receptor pairs, as well as classical MHC I–CD8 interactions, using the CellChat algorithm applied to our scRNA-seq data; only statistically significant interactions are presented. This analysis demonstrates interactions between NK, DC, and CD8⁺ T cells, with particularly strong crosstalk observed within the XCL1⁺ NK/XCR1⁺ DC/Tef axis in the iERT group (Supplementary Fig. 6e, f). These CellChat results indicate a treatment-specific enhancement of the XCL1⁺ NK/XCR1⁺ DC/Tef axis.

The tumor-draining lymph nodes (TDLNs) also contained Tpex cells transcriptionally similar to those in the TME (Fig. 7a, b) and iERT induced more AH1⁺ Tpex cells in the TDLNs (Fig. 7c). In TDLNs draining bulky B16-OVA tumors, the MFI of SIINFEKL/H-2K^b MHC-I complexes on DCs was highest in the iERT treatment group (Fig. 7d), suggesting improved processing of the surrogate tumor antigen OVA by DCs through the antigen cross-presentation pathway. Blocking the egress of T cells using FTY720 reduced iERT efficacy (Fig. 7e). These results highlight the importance of TDLNs in iERT.

Collectively, these data indicate that iERT induces Tpex cells in both the TME and TDLNs, as well as Tef and Tex cells in the TME. Tumor-specific CD8⁺ T cells, induced by the IFN-I/NK/XCL1 axis and originating from TDLNs, are critical for therapeutic efficacy. These data provide further insight into iERT-induced antitumor CD8⁺ T cell responses.

Discussion

Radioimmunotherapy trials have so far achieved only limited success^1,2. Developing an effective AIRT approach to enhance ICI efficacy is therefore urgently needed². As a proof-of-concept for AIRT, we developed an innovative technique, ERT, which concurrently delivers LDRT to the GTV and SBRT to subvolume(s) of the same tumor. This approach leverages the immunogenic effects of SBRT and the immunomodulatory effects of LDRT. In combination with αPD-1, we termed this approach iERT. We evaluated this approach in bulky tumors, which pose major challenges for traditional treatments such as conventional RT, surgery, and chemotherapy. Bulky tumors also often respond poorly to ICIs due to their immunosuppressive TME, characterized by a paucity of NK cells, DCs, and CD8⁺ T cells^7,55,56. iERT enables the effective treatment of large bulky tumors, and our study provides important mechanistic insights into how iERT exerts its therapeutic effects, thereby also advancing the general understanding of the biological mechanisms underlying combined radioimmunotherapy (see also graphical abstract, Fig. 7f): (1) iERT triggers a antitumor response characterized by an early innate immune response (NK cells), which subsequently transitions into an adaptive immune response (CD8⁺ T cells). (2) To our knowledge, this is the first study to demonstrate the essential role of XCL1⁺ NK cells in the context of radioimmunotherapy, as revealed by scRNA-seq and validated through a series of functional experiments. In this setting, NK cells are activated by SBRT-induced IFN-I rather than by translocated calreticulin, a known inducer of NK cell activation during ICD⁵⁷. (3) Unlike its well-established direct role in DC-mediated CD8⁺ T cell priming, we found that IFN-I can also indirectly enhance cross-priming by activating XCL1⁺ NK cells, which in turn recruit XCR1⁺ DCs to the tumor.

In three mouse bulky tumor models, iERT achieved superior tumor control compared to either LDRT/αPD-1 or partial SBRT/αPD-1, leading to complete cure in 48.9% (23 of 47 mice, data pooled from Figs. 1–7 and Supplementary Fig. 1) of mice bearing bulky CT26 tumors. The efficacy of iERT was dependent on NK and CD8⁺ T cells. iERT induces an early NK cell response followed by a subsequent CD8⁺ T cell response in murine bulky tumors. To identify dynamic immune cell infiltration-based predictors of tumor cure, we performed fine-needle biopsies; however, none of the measured parameters predicted complete tumor regression (see Supplementary Data 1). This likely reflects both the limited representativeness of fine-needle biopsies and the variability of tumor-immune dynamics. In bulky tumors, substantial biological heterogeneity^35,36,58 - including differences in baseline TME composition, such as the balance between effector CD8⁺ T cells and NK cells versus immunosuppressive myeloid cells and regulatory T cells - may influence immune infiltration dynamics and therapeutic responses⁸. Moreover, spatial genomic heterogeneity and stromal factors (e.g., fibrosis) can further complicate the interpretation of immune correlates⁸. The low sample yield from fine-needle biopsies also constrained the analysis to a narrow flow cytometry panel and precluded further phenotypic and functional profiling. Future studies integrating multiplex spatial imaging and/or spatial transcriptomics may help disentangle these biological variables from technical noise and better define immune correlates of tumor cure.

Our in-depth analysis revealed a previously unrecognized mechanism by which RT-triggered IFN-I drives CD8⁺ T cell-mediated antitumor immunity. The SBRT component of iERT induces IFN-I, which promotes tumor infiltration by XCL1⁺ NK cells - a distinct subset identified by scRNA-seq - critical for RT-mediated TME remodeling. This subpopulation secretes XCL1, recruiting antigen-presenting XCR1⁺ cDC1s essential for initiating CD8⁺ T cell responses and playing a pivotal role in the efficacy of iERT. SBRT also induced tumor cell-derived EVs carrying tumor antigens such as the OVA-derived model epitope SIINFEKL, while LDRT facilitated NK and CD8⁺ T cell infiltration into the TME. Given their heterogeneity and long-term immune co-evolution, bulky tumors often exhibit an immune-cold phenotype. Our findings indicate that iERT activates the IFN-I/NK/CD8⁺ T cell axis, enhancing immune infiltration, potentially converting immune-cold bulky tumors into ICI-responsive phenotypes, underscoring its translational potential.

SBRT can deliver potentially ablative doses and exerts its immunogenic effects by inducing ICD, which is primarily dependent on the release of damage-associated molecular patterns (DAMPs). DAMPs can facilitate DC recruitment to the tumor and DC-mediated transport of tumor-associated antigens to TDLNs, where DCs initiate T cell priming^20,59–62. In ERT for bulky tumors, partial rather than whole-tumor SBRT is applied to avoid intolerable toxicities^14–16. A fraction dose of 10 Gy was selected for ERT based on previous studies from our group and others^38–40,63, which demonstrated that 8–12 Gy per fraction was optimal for inducing the release of IFN-I, a key DAMP generated by RT-induced DNA damage and crucial to antitumor immunity³⁸. Indeed, here, in vitro SBRT induced a significant increase in Ifnb1, and ex vivo a significantly higher IFN-I response was observed in partial SBRT/αPD-1- and iERT-treated tumors. Previous studies focused on the role of IFN-I in facilitating cross-priming of tumor-specific CD8⁺ T cells via a direct effect on DCs^60–62. However, our study reveals a previously unrecognized mechanism whereby SBRT-induced IFN-I, but not calreticulin or HMGB1, activates NK cells, although a recent study suggests a role for calreticulin in NK cell activation⁵⁷. These activated NK cells secrete XCL1 to recruit XCR1⁺ cDC1 cells into the TME, thereby fostering the initiation of adaptive CD8⁺ T cell-mediated antitumor immunity by cross-presenting DCs. This finding expands the current understanding of how IFN-I, as a DAMP produced during RT-induced ICD, can promote tumor antigen cross-priming/presentation by DCs. Notably, we also demonstrate that SBRT can induce tumor cells to secrete EVs containing neoantigens (e.g., SIINFEKL), which can be cross-presented by DCs.

The rapid activation of NK cells is particularly crucial for patients with bulky tumors, as it enables early tumor control, allowing ICIs to exert their effects subsequently. This mirrors the body’s antiviral response strategy, in which innate immunity, primarily through NK cell activation, is triggered first, paving the way for the subsequent development of adaptive immunity⁴². This is especially important given that the clinical response to ICIs typically occurs relatively late, with a median response time of 8–12 weeks⁹.

However, partial SBRT-induced T cell priming was not sufficient to control bulky mouse tumors (Fig. 1j–o). This may be due to the persistently poor infiltration of NK cells (Fig. 3c–f), DCs (Fig. 5c), and CD8⁺ T cells (Fig. 6c, d, h, and i) into the entire large tumors, even after partial SBRT/αPD-1 treatment. Adding LDRT to partial SBRT (ERT) together with αPD-1 significantly improved bulky tumor control in mice (Fig. 1j–o). Two previous preclinical studies showed that applying SBRT to primary tumors and delivering LDRT to abscopal tumors can enhance the RT-induced abscopal effect, possibly by upregulating T cell-attracting chemokines and promoting effector CD8⁺ T cell recruitment to abscopal tumors^28,29. In the current study, using single-tumor models, LDRT/αPD-1 and iERT treatments significantly increased CXCL10 secretion (Supplementary Fig. 4f), which can attract NK cells (Fig. 3c–f) and CD8⁺ T cells (Fig. 6d). Another study suggested that LDRT delivered to abscopal tumors may remodel the stroma by downregulating Transforming Growth Factor-β (TGFβ) (18). However, in our study, only iERT treatment markedly decreased TGFβ signaling, which was not observed with LDRT/αPD-1 (Supplementary Fig. 4c). Therefore, iERT appears very promising for TME remodeling.

Unlike iERT, whose efficacy is highly dependent on IFNAR1 and XCL1 signaling, blocking either pathway did not significantly reduce the efficacy of LDRT/αPD-1 (Supplementary Fig. 4h) or SBRT/αPD-1 (Supplementary Fig. 5e). Although LDRT induced Ifnb1 expression (Fig. 4c), levels were lower than with SBRT, possibly explaining the limited role of IFN-I and XCL1 in LDRT/αPD-1. In partial SBRT/αPD-1, the irradiated volume was relatively small compared to iERT (e.g., 65 mm³ for SBRT vs. 642 mm³ for LDRT; Fig. 1a). Therefore, while SBRT induced substantial Ifnb1 expression (Fig. 4c) and NK cell infiltration in the SBRT zone (Fig. 3d), these effects were spatially confined, limiting their impact on overall tumor control. These results highlight that both SBRT and LDRT are essential for full iERT efficacy.

The efficacy of PD-1/PD-L1 inhibitors is often hindered not only by the absence of CD8⁺ T cells in the TME but also by their dysfunction. PD-1/PD-L1 inhibitors mainly act on Tpex cells, inducing their proliferation and differentiation into Tef and Tex cells^8,50,52. Here, in untreated, LDRT/αPD-1, or partial SBRT/αPD-1-treated tumors, tumor-specific Tpex and Tef cells were rarely found. In contrast, iERT-treated tumors were markedly infiltrated by both T cell subsets (Fig. 6d). iERT also induced the highest levels of AH1⁺ Tpex cells in the TDLNs and promoted dominant TCR clone enrichment in the TME. These results again emphasize the importance of applying both partial SBRT and LDRT to large tumors to achieve optimal CD8⁺ T cell responses.

Tpex cells require costimulatory signals (e.g., CD28–CD80/86 axis) from DCs in the TME to differentiate into the TIM3⁺ subset and obtain effector functions⁵⁴. CD8⁺ T cells and DCs were located close to each other following SBRT-containing therapy (iERT or SBRT/αPD-1), indicating that the SBRT component of ERT may be key to enhancing crosstalk between these cells (Fig. 6f). This finding highlights the necessity of combining SBRT with LDRT, as LDRT alone is insufficient to elicit optimal immune responses. Tertiary lymphoid structures, intratumoral niches, and APC niches are crucial for T cell and APC interactions and drive antitumor immunity⁶⁴. Although further studies are needed to determine whether pre-existing microstructures or iERT-induced remodeling of the TME are more important for iERT efficacy, our study provides mechanistic insight by identifying RT as a key driver of immune cell interactions, thereby remodeling the TME.

Two recent preclinical studies investigated how RT dose heterogeneity modulates antitumor immunity and advanced the concept of heterogeneous intratumor irradiation to enhance RT–ICI synergy. One study showed that dose variation in single-fraction brachytherapy activates different signaling pathways, reprogramming myeloid cells and enriching effector T cells in the TME in mice, thereby enhancing antitumor effects in combination with ICIs³⁵. The other applied external beam radiation therapy (EBRT) to deliver SBRT (16 Gy one fraction) to half of the tumor and LDRT (2 Gy one fraction) to the other half. This approach, combined with αPD-1, not only enhanced tumor control and promoted cytotoxic CD8⁺ T cells, but also recruited immunosuppressive neutrophils that limited benefit; blocking neutrophil recruitment via a CXCR2 antagonist enhanced efficacy³⁶. These important studies introduced the concept that heterogeneous intratumoral RT can elicit spatially diverse antitumor immune responses^35,36. Together with our work, they provide crucial mechanistic insights into how heterogeneous RT can enhance ICI efficacy. Brachytherapy has limited applicability for many patients, particularly those with thoracic malignancies³⁵. The half-SBRT/half-LDRT technique can be suitable for smaller lesions, yet delivering it safely in bulky thoracic tumors can be challenging due to strict dose constraints³⁶. To address these practical challenges, we developed ERT, an EBRT-based approach that delivers LDRT to the entire GTV and SBRT to selected subvolume(s) in the GTV. This design may offer a more feasible and safe option for large or deeply located tumors. In our unpublished retrospective cohort of 39 patients with bulky tumors (median diameter 6.8 cm), 76.9% had NSCLC, and 92.3% had progressed after more than one prior line of therapy. iERT yielded an objective response rate of 38.5%, with no toxicities above grade III. The median progression-free survival was 5.6 months, and the median overall survival was 23.0 months³⁷. Our study extends prior work^35,36 by demonstrating that fractionated heterogeneous RT, compared with single-fraction heterogeneous RT, more effectively augments immunomodulatory effects in combination with αPD-1 in mice. We also provide an in-depth mechanism of iERT. Using scRNA-seq and validation experiments at different treatment time points, we dissected the specific immunomodulatory effects of the different treatment components. In addition, we discovered a previously unrecognized mechanism for how IFN-I signaling (shown here for hRT-induced IFN-I) drives NK cell activation and initiates the NK/DC/CD8⁺ T cell axis. RT-induced IFN-I activated XCL1-secreting NK cells, which were crucial for recruiting cross-presenting XCR1⁺ cDC1s and priming CD8⁺ T cell responses. This enhanced the responses of Tpex and Tef cells, which are essential for sustained antitumor immunity. Furthermore, we demonstrated that SBRT, but not LDRT, induces EV release, and that SBRT-induced EVs carry tumor neoantigens to DCs, augmenting their ability to prime CD8⁺ T cells.

Combined with pembrolizumab, multisite SBRT to up to four tumor nodules was effective and safe for treating metastatic disease in a phase I trial⁶⁵. Among 97 patients in this trial, 46 received partial SBRT for tumors exceeding 65 cm³, with no significant difference in local failure compared to complete SBRT (P = 0.08)⁶⁶. Only one of 46 patients had grade 3 dose-limiting toxicity. Unlike our study, this trial treated large tumors with partial SBRT without LDRT covering the entire tumor⁶⁶. Our preclinical results show that LDRT in iERT is required to inflame the TME and achieve optimal CD8⁺ T cell responses, suggesting that ERT may be a better partner for ICIs in treating bulky tumors.

SFRT in the form of GRID or Lattice RT can also partially deliver ablative high-dose irradiation (10–25 Gy × 1–5 fractions) to subvolumes of large tumors. While preliminary clinical evidence indicates that SFRT is safe and effective, the underlying mechanisms and immunomodulatory effects, particularly in combination with ICIs, remain poorly understood^32–34. In the era of ICIs, SFRT plus ICIs has been reported in only a few patients³³. In contrast, ERT plus αPD-1 antibodies was evaluated in our unpublished retrospective study involving 39 patients with difficult-to-treat bulky tumors, demonstrating a favorable safety profile. The first patient was treated on May 26, 2020 (as mentioned above)³⁷.

Unlike classic SFRT, which evenly distributes high-dose beams across the entire bulky tumor, ERT enhances immunotherapy by delivering LDRT to the whole tumor and SBRT to subvolumes, such as hypoxic regions that are resistant to CFRT but not SBRT¹³. These regions can be identified through hypoxia PET imaging (e.g., ¹⁸F-labeled fluoroazomycinarabinoside, ¹⁸F-FAZA). Historically, CFRT has aimed to maximize cancer cell destruction while minimizing healthy tissue toxicity, often overlooking RT-induced immunostimulatory effects. AIRT, which modifies conventional RT to better integrate with ICIs, is crucial for designing effective new clinical trials for RT + ICIs^1,2. To achieve this, we applied LDRT, known for its immunomodulatory effects, to the entire tumor to recruit NK and CD8⁺ T cells. Simultaneously, SBRT, with its immunogenic properties, can be directed to specific subvolumes of large tumors, such as hypoxic regions, to activate NK cells and recruit cDCs for the priming of CD8⁺ T cell antitumor responses.

To support clinical translation, we have initiated a Phase I trial (NCT06349837) with standardized target volume delineation and integration of iERT into existing treatment workflows. For LDRT, the GTV-LDRT covers the bulky tumor, with mandatory ¹⁸F-FDG PET/CT to differentiate active tumor from atelectasis or normal tissue. For SBRT, a 1–2 cm spherical subvolume (GTV-SBRT) within the GTV-LDRT is selected, avoiding necrosis, atelectasis, and regions adjacent to the bronchial tree or mediastinum for safety. Optional ¹⁸F-FAZA (a hypoxia tracer) PET may guide selection of hypoxic, metabolically active regions identified by overlapping FAZA and FDG uptake. ERT is given with tislelizumab, with optional chemotherapy or antiangiogenic therapy per clinical judgment. In bulky tumors, marked heterogeneity in the TME (oxygenation, metabolism, vasculature) leads to spatially heterogeneous RT responses. ¹⁸F-FDG PET/CT has been proposed to monitor intratumoral metabolic responses during treatment, enabling adaptive RT guided by metabolic changes to optimize dose and target coverage⁵⁸. Tubin et al. delivered partial SBRT to hypoxic subregions empirically delineated with ¹⁸F-FDG PET and contrast-enhanced CT, targeting the hypovascular (CT contrast-hypoenhanced), hypometabolic junction between the central necrotic core and the peripheral hypermetabolic rim³⁴. In our trial, ¹⁸F-FAZA PET (a hypoxia tracer) may further refine the selection of hypoxic, metabolically active subvolumes by integrating ¹⁸F-FAZA and ¹⁸F-FDG uptake patterns. Such image-defined subvolumes can be prioritized for SBRT boosts within the iERT framework to enhance antigen release and ICD. Future directions include incorporating advanced functional imaging (e.g., 3′-deoxy-3′-[18 F]fluorothymidine PET for cellular proliferation) and molecular imaging (PD-L1 PET) to better characterize intratumoral heterogeneity and optimize SBRT subvolume delineation.

Previous studies have demonstrated that RT-induced antitumor immunity is highly dependent on dose and fractionation^2,18. A preclinical study demonstrates that 8–12 Gy per fraction optimally induces IFN-I release and abscopal effect, whereas > 12–18 Gy activates TREX1, degrading cytosolic DNA, suppressing IFN-I production and abscopal effect³⁸. LDRT (0.5–2 Gy/fraction, total ≤ 6 Gy) has been shown by us and others to promote immune cell infiltration in both preclinical and clinical settings^28–31. In the present preclinical study, a single fraction of ERT combined with αPD-1 achieved only modest tumor control (Supplementary Fig. 1c), whereas three fractions markedly improved tumor control compared with LDRT/αPD-1 or SBRT/αPD-1 (Fig. 1j, m), indicating that 10 Gy × 3 is more effective than 10 Gy × 1. This difference may be attributed to the ability of fractionated ERT to induce a stronger and earlier IFN-I response than single-fraction irradiation³⁸, as early post-treatment IFN-I upregulation is critical for effective ICI therapy^67,68. In addition, fractionated ERT may differentially regulate EV release and cargo composition, favoring the generation of immunostimulatory EVs. Finally, distinct activation of cellular stress-response pathways, including DNA damage signaling and innate immnue sensing, may also facilitate ERT-driven shaping of NK- and CD8⁺ T cell–mediated immune responses. Nonetheless, the optimal RT regimen remains to be determined. To address this in the clinic, our ongoing Phase I trials are testing multiple dose fractionation schemes: NCT05615142 (ECLIPSE-01 trial) is currently evaluating ERT delivered as partial SBRT (10 Gy) combined with LDRT (2 Gy) per fraction, administered over 1 to 3 fractions (i.e., 10 Gy/2 Gy × 1–3); and NCT06349837 (ECLIPSE-02 trial) is testing ERT at 8 Gy/2 Gy × 3, 10 Gy/2 Gy × 3, or 15 Gy/2 Gy × 3 fractions. We also anticipate that incorporating dynamic immune parameters (e.g., IFN-I signature, CXCL9/10 levels, NK/CD8⁺ T cell counts) will further refine iERT dosing strategies⁶⁸.

We acknowledge several limitations. In this study, we focused on analyzing the SBRT- and LDRT-irradiated regions while excluding the intermediate zone that received 2–10 Gy, using ink marking to ensure clear spatial separation for scRNA-seq, TCR-seq and flow cytometry analyses (see “iERT for mice “in Methods). Notably, in iERT, ~ 20% of the tumor volume receives ~ 6 Gy, and recent work³⁵ suggests that such intermediate-dose regions may also contribute to the enhanced efficacy of heterogeneous RT when combined with ICIs. The potential biological relevance of this zone warrants further investigation, as it may represent an important component of the overall therapeutic mechanism. In addition, the in vitro experiments shown in Fig. 4b–e (e.g., RT-induced IFN-I signaling and co-culture of irradiated tumor cells activating NK cells) and Fig. 5g–k (e.g., co-culture of EVs activating BMDCs) employed SBRT (3 × 10 Gy) rather than iERT-treated tumors. These experiments were designed as mechanistic models to study the high-dose SBRT component of iERT and its effects on NK cells and DCs. However, these models do not fully capture the iERT synergy.

In conclusion, we propose iERT, an AIRT strategy that synergistically combines LDRT, SBRT, and ICI therapy to address the challenge of bulky tumors. Mechanistically, iERT triggers an IFN-I response, which subsequently activates the XCL1⁺ NK/XCR1⁺ cDC/tumor-specific CD8⁺ T-cell axis to effectively control bulky tumors in mice. This strategy warrants further investigation in our ongoing phase I clinical trial (NCT06349837).

Methods

Mice and cell lines

Mouse experiments were approved by the Institutional Animal Care and Use Committee of Sichuan University and West China Hospital of Sichuan University (no. 2020415 A and 20231226009). BALB/c and C57BL/6 male mice aged six to eight weeks were purchased from GemPharmatech Co., Ltd. and housed in a specific pathogen-free (SPF) facility. All mice were maintained under controlled conditions with a 12 h light/dark cycle, constant temperature (21 °C–25 °C), and humidity (40–70%), with unrestricted access to standard chow and water. CT26 were obtained from the Cell Center of Peking Union Medical College (1101MOU-PUMC000275). LLC1 (CRL-1642) and B16F10 (CRL-6475) cells were obtained from ATCC. B16F10 cells were transfected with 1 μg pCI-neo-cOVA (Plasmid #25097, Addgene) plasmid and 1.5 μL Lipofetamine 3000 (Invitrogen, # L3000150) overnight in a 6-well plate with Opti-MEM medium (Gibco, # 31985070). B16-OVA cells were selected with 1.5 mg/mL G418 (Gibco, # 10131035) starting 1 day after transfection. OVA expression in B16-OVA cells was examined by OT-I cells. OT-I mice (C001198) expressing a transgenic T-cell receptor specific for the OVA peptide SIINFEKL presented by H-2K^b were purchased from Cyagen Biosciences Inc.

Establishment of tumor models

5 × 10⁵ CT26, LLC1, or B16-OVA tumor cells were suspended in 50% Matrigel (Mogengel-Bio, #082706) and injected s.c. into the left hind limb of BALB/c, C57BL/6 or C57BL/6 mice, respectively. Calipers were used to measure tumor size every three days, and tumor volume was calculated using the formula: 0.5 × length × width². The largest tumor should not exceed 2 cm at the largest diameter according to the ethics committee; otherwise, the mouse should be euthanized. In all experiments, the maximal tumor size was not exceeded. At the end of the experiments, mice were euthanized under deep isoflurane anesthesia followed by cervical dislocation.

iERT for mice

A fraction dose of 10 Gy of SBRT and 2 Gy of LDRT were simultaneously administered on 3 consecutive days with an X-RAD SmART+ irradiator (Precision X-ray). A digital X-ray detector (20 cm × 20 cm) was positioned on the gantry opposite the X-ray source, with a 60 cm source-to-detector distance. Cone-beam CT (CBCT) images were acquired at 60 kV, 0.5 mA with a 2 mm aluminum filter, using a continuous beam and 360° gantry rotation. Target volumes were contoured on the CBCT images, and 3D treatment planning was performed using the SmART-ATP small-animal treatment planning system (TPS) (version 1.1). Irradiation was delivered at 225 kV, 20 mA with a 0.3 mm copper filter.

For partial SBRT, the gross tumor volume (GTV-SBRT) was defined as a 5 mm-diameter sphere at the tumor center; the GTV-LDRT encompassed the entire gross tumor. SBRT was delivered via a 360° arc technique, while LDRT used parallel-opposed beams (225 kVp, 20 mA). Doses of 10 Gy (SBRT) and 2 Gy (LDRT) were prescribed to cover 95% of their respective GTVs. Isocenters were placed at the centers of GTV-SBRT or GTV-LDRT. For ERT, 8 Gy was delivered to GTV-SBRT, followed by 2 Gy to GTV-LDRT, with the isocenter at the GTV-SBRT center. A 5-mm (diameter) circular collimator (C5) was used to deliver 8 Gy SBRT. For LDRT, either a C10 or a 20 mm square collimator (S20) was selected based on tumor size/shape to ensure adequate target coverage. TPS outputs and corresponding DVHs are shown in Fig. 1a, and Supplementary Fig. 1a. The dose rates for C5, C10, and S20 collimators were 3.06, 3.14 and 3.35 Gy/min, respectively.

Dose delivery verification was performed with a self-made phantom constructed from two silicone gel slabs with a density ≈ 1.03 g/cm³, approximating human or animal soft tissue.

Post-RT, the mouse skin was removed, and the SBRT-irradiated area was marked with green marking ink (Fig. 1d, ZLI-9080, ZSGB-BIO). This marking was then used to accurately distinguish the SBRT and LDRT zones for subsequent experiments. Biological validation of dosimetry was performed using γH2AX and 53BP1 IF staining and comet assay (C2041S, Beyotime) (Fig. 1e, f and g, Supplementary Fig. 1b). To ensure clear separation of differentially irradiated regions for subsequent analyses, the SBRT (red dashed line) and LDRT (blue dashed line) zones were sampled along the sagittal plane (Fig. 1c, left). A ≥ 2-mm-wide band of tissue at their interface along the sagittal plane (Fig. 1c, left)—corresponding to the intermediate tissue (~ 2–10 Gy; turquoise dashed line)—was carefully removed to avoid overlap of radiation effects. In addition, tissue at the SBRT–LDRT interface on transverse sections (turquoise dashed line; Fig. 1c, right and Fig. 1d, left) was removed. The weights of the collected SBRT and LDRT zones were generally consistent across animals (Supplementary Fig. 7b, left). Inter-animal variability was observed in viable leukocyte yield (Supplementary Fig. 7b, right).

On the first day of RT (d1), 200 μg of αPD-1 (RMP1-14, BioXCell, Cat# BE0146) was injected i.p., followed by injections every three days. αCD8 antibodies (200 μg, clone 2.43, BioXCell, Cat# BE0061) were injected i.p. on d-2, d1, and once weekly thereafter to deplete CD8⁺ T cells. For neutralization of XCL1, 25 μg of αXCL1 (R&D Systems, Cat# MAB486) were injected i.p. on d1, d4 and d7 post-RT. For neutralization of IFNAR1, 250 μg of αIFNAR1 (MAR1-5A3, Selleck Cat# A2121) were injected i.p. 1 day before RT (d0), d1 and d4 post-RT. For NK-cell depletion in Balb/c mice and C57BL/6 mice, αGM1 (clone poly21460, BioLegend Cat# 146002, 500 μg) and αNK1.1 (clone PK136, BioXCell, Cat# BE0036, 100 μg) i.p., respectively, on d0, d1, d4 and d7. To block T cell egress from lymph nodes, FTY720 (25 μg per mouse, Selleckchem) was i.p. injected d0 and then every other day. For inhibition of EV biogenesis and release, GW4869 (1.25 mg/kg, MCE CAT#HY-19363) was i.p. injected d0 and then every three days. Survival was determined when the tumor had reached 2000 mm³ after treatment started.

Western blotting

Tumor samples were homogenized on ice using a tissue homogenizer (LABGIC, L-HOB-MINI) in RIPA lysis buffer (Beyotime, #P0013B) supplemented with protease and phosphatase inhibitor cocktails (TargetMol, #C0001, #C0002, #C0003). Lysates were incubated on ice for 20 min and centrifuged at 12,000 rpm for 15 min at 4 °C. Protein concentrations in the supernatants were determined using the BCA assay (Epizyme, #ZJ103). Samples were mixed with 5× sample buffer (Epizyme, #LT101S), boiled at 100 °C for 5 min, resolved on 12% PAGE gels (Vazyme, #E304-01), and transferred to PVDF membranes using the Mini-PROTEAN Tetra System (Bio-Rad).

Membranes were blocked for 30 min with protein-free rapid blocking buffer (Epizyme, #PS108P) and incubated overnight at 4 °C with primary antibodies diluted in universal antibody diluent (NCM Biotech, #WB500D): γH2AX (1:1000; CST, #9718) and GAPDH (1:5000; Abcam, #ab8245). After washing with TBST, membranes were incubated for 1 h at room temperature with HRP-conjugated secondary antibodies (Jackson ImmunoResearch, #111-035-003 or #115-035-003; 1:5000). Blots were washed again and developed using Ultrasensitive ECL (Oriscience, #PD202) prior to imaging. The grayscale values of GAPDH and γH2AX were measured by ImageJ software (version 1. 53 k).

ELISA

For ELISA, samples were prepared using the same procedures as for western blotting. The concentration of γH2AX in the supernatant was measured using the PathScan Phospho-Histone H2A.X (Ser139) Sandwich ELISA Kit (Cell Signaling Technology, #50929 C). The assay was performed according to the manufacturer’s instructions.

Flow-cytometric analyses

To prepare single-cell suspensions, tumors were weighed, minced, and then mechanically ground through a 70 μm sieve and filtrated through 40 μm cell strainers (Biofil, CSS013070, CSS013040). The lymph nodes were squeezed through a 70-μm strainer. Red blood cells were lysed using 1X lysis buffer (Leagene, CS0003). After preparing single-cell suspensions, leukocytes were counted by trypan blue exclusion under a microscope based on size/morphology. The frequency of each population within the CD45⁺ compartment was subsequently determined by flow cytometry. Absolute counts per gram tumor were calculated by multiplying the percentage of each population (of CD45⁺ cells) by the total number of viable leukocytes counted. These counts were then normalized to tumor weight and reported as cells per gram of tumor. AH1 tetramer-PE (H-2L^d, gp70, MBL Life Science, Cat# TS-M521-1) and CD8-Alexa Fluor 700 (Bio-Rad Cat# MCA609A700, clone KT15) were used to detect tumor-specific CD8⁺ T cells. CD45-BV510 (Cat# 103137, 30-F11), PD-1-BV785 (Cat# 135225, 29 F.1A12), TIM3-BV605 (Cat# 119721, RMT3-23), CD49b-AF647 (Cat# 108912, DX5), CCL5-PE (Cat# 149103, 2E9/CCL5), IFNγ-FITC (Cat# 505805, XMG1.2), CD11b-PE-Cy7 (Cat# 101216, M1/70), CD19-FITC (Cat# 115505, 6D5), CD4-FITC (Cat# 100509, RM4-5), XCR1-BV650 (Cat# 148220, ZET), MHCII-BV785 (Cat# 107645, M5/114.5.2), CD103-APC (Cat# 121413, 2E7), CD86-BV421 (Cat# 105031, GL-1), CCR7-BV605 (Cat# 120125, 4B12), F4/80-BV711 (Cat# 123147, BM8), CD206-AF700 (Cat# 141733, C068C2), Ly6C-PerCP-Cy5.5 (Cat# 128011, HK1.4), and CXCL9-PE (Cat# 515603, MIG-2F5.5) were purchased from BioLegend; TCF1-Pacific Blue (BV421) (Cat# 9066, C63D9) was purchased from CST; CD101-PE-Cy7 (Cat# 25-1011-82, Moushi101) and Ly6G-BUV737 (Cat# 367-9668-82, 1A8) were purchased from Thermo Fisher Scientific; and CD3-BUV496 (Cat# 612955, 145-2C11) and CD11c-APC-Cy7 (Cat# 561241, HL3) were purchased from BD Biosciences.

A BD LSRFortessa flow cytometer was used to analyze the samples, and FlowJo software (version 10.4.0) was used to analyze the data. NK cells were identified as: CD45⁺, CD3^-, CD49b⁺ or NK1.1⁺. XCR1⁺cDC1: CD45⁺, CD3^-CD19^-CD49b^-, CD11c⁺, CD11b^-, and XCR1⁺; CCR7⁺cDC1: CD45⁺, CD3^-CD19^-CD49b^-, CD11c⁺, CD11b^-, and CCR7⁺; cDC2: CD45⁺, CD3^-CD19^-CD49b^-, CD11c⁺, and CD11b⁺. M1 macrophages: CD45⁺, CD3^-CD19^-CD49b^-, F4/80⁺, CD11b⁺, CD86⁺ and CD206^-. M2 macrophage: CD45⁺, CD3^-CD19^-CD49b^-, F4/80⁺, CD11b⁺, CD86^- and CD206⁺. Neutrophils: CD45⁺, CD3^-CD19^-CD49b^-, F4/80^-, CD11b⁺ and Ly6G⁺. Tpex cells: CD8⁺, PD-1⁺, TCF1⁺, and TIM3^-, Tef cells: CD8⁺, PD-1⁺, TCF1^-, TIM3⁺, and CD101^-; Tex cells: CD8⁺, PD-1⁺, TCF1^-, TIM3^-, and CD101⁺.

NK cell isolation, sorting and coincubation with tumor cells

NK cells from the spleens of BALB/c mice were isolated by using CD49b (DX5) MicroBeads (Miltenyi, Cat# 130-052-501). Splenic NK cells from BALB/c mice were flow-cytometrically sorted (Aria Fusion) using the following gating strategies: live 7AAD^-CD3^-CD49b⁺. The sorted NK cells were then co-cultured in the presence of 10 ng/μL IL-15 (PeproTech, Cat# 210-15-50UG). αIFNAR1 (50 μg/mL, MAR1-5A3, Selleck Cat# A2121), αHMGB1 (15 μg/mL, ab18256, Abcam) and α-calreticulin (3.23 μg/mL, ab92516, Abcam) were incubated with NK cells 30 min before coincubation with tumor cells.

IF of tumor slices

Slices from formalin-fixed paraffin-embedded samples were stained with anti-γH2AX (Ser139) (Cell Signaling Technology Cat# 9718) or anti-53BP1 antibody (ab36823, Abcam). Images were acquired using a PerkinElmer Vectra Polaris multispectral microscope or A1R MP+ multiphoton confocal microscopy. For quantification, the QuPath software (version 0.4.1) was used.

mIHC of tumor slices

Slices from formalin-fixed paraffin-embedded samples were used and stained with Opal 6-Plex Manual Detection Kit (Akoya, NEL861001KT) and the following primary antibodies: rabbit anti-CD8α (Cell Signaling Technology Cat# 98941, 1:100), rabbit anti-CD3 (Abcam Cat# ab16669, 1:150), rabbit anti-TCF1 (Cell Signaling Technology Cat# 2203, 1:100), rabbit anti-PD-1 (Cell Signaling Technology Cat# 84651, 1:100), rabbit anti-TIM3 (Cell Signaling Technology Cat# 83882, 1:200), rabbit anti-CD11c (Cell Signaling Technology Cat# 97585, 1:100), rat anti-CD49b (BioLegend Cat# 108902, 1:100), and goat anti-XCL1 (R and D Systems Cat# AF486, 1:100). mIHC images were acquired using a PerkinElmer Vectra Polaris multispectral microscope. For quantification, the QuPath software (version 0.4.1) was used to create objectives for the tumor center (SBRT zone) or periphery (LDRT zone), where the cells were auto-detected (example, Fig. 2c)⁶⁹. Tpex cells were identified as: CD3⁺, CD8⁺, PD-1⁺, TCF-1⁺, and TIM-3^-. Tdf cells were identified as: CD3⁺, CD8⁺, PD-1⁺, TCF-1^-, and TIM-3⁺. DCs were identified as: CD11c⁺. NK cells in BALB/c mice were identified as: CD49b⁺ and CD3^-. The distance between Tpex (or Tdf) and CD11c⁺ cells was also obtained using QuPath software⁶⁹.

Quantitative reverse transcription polymerase chain reaction (RT-qPCR)

Total RNA from cells was extracted with the TRIzol Reagent (Invitrogen, Cat# 15596018) and reverse transcribed to complementary DNA (cDNA) by using the HiScript III RT SuperMix for qPCR (R323-01, Vazyme). Real-time PCR was performed with ChamQ Universal SYBR qPCR Master Mix (Q711-03, Vazyme) using a Bio-Rad CFX96 Touch Real-Time PCR Detection System. Data were normalized by the level of Gapdh expression in each individual sample. The 2ΔΔCt method was used to calculate relative expression changes. The primers used are: mouse Ifnb1 forward: GCCTTT GCCATCCAAGAGATGC; mouse Ifnb1 reverse: ACACTGTCTGCTGGTGGAGTTC; mouse Cxcl10 forward: GCCGTCATTTTCTGCCTCA; mouse Cxcl10 reverse: CGTCCTTGCGAGAGGGATC; mouse Gapdh forward: CATCACTGCCACCCAGAAGACTG; mouse Gapdh reverse: ATGCCAGTGAGCTTCCCGTTCAG.

Bulk RNA-seq

Total RNA was extracted from in vitro cultured tumor cells harvested on d4 after the first RT or non-irradiated, FACS-isolated NK cells harvested 4 h after treating with IFNb1 (10000 U/ml, MedChemExpress, Cat# HY-P73130) or PBS, or ex vivo tumor samples using TRIzol reagent (Invitrogen, Cat# 15596018). RNA purity and concentration were measured using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific), and RNA integrity was evaluated with an Agilent 2100 Bioanalyzer (Agilent Technologies). Libraries were prepared using the NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biolabs, Cat# E7530) for low RNA input FACS-isolated NK cell samples, the VAHTS Universal V5 RNA-seq Library Prep Kit (Vazyme, Cat# NR605) for in vitro tumor cells, or TruSeq Stranded mRNA LT Sample Prep Kit (Illumina, Cat# 20020594) for ex vivo tumor samples according to the manufacturer’s protocol. Sequencing was performed on an Illumina NovaSeq 6000 platform, generating 150 base pair (bp) paired-end reads. RNA-seq and analysis were performed by OE Biotech Co, Ltd. Raw reads in FASTQ format were processed with fastp (version 0.23.1) to remove low-quality reads, yielding clean reads. These clean reads were aligned to the reference genome (mm10) using HISAT2 (version 2.2.1), and gene-level read counts were quantified with HTSeq-count (version 0.9.1). Differential gene expression (DGE) analysis was performed using edgeR (version 3.36.0). The results of the DGE analysis between two treatment groups were sorted in descending order of log₂(fold change), and gene set enrichment analysis (GSEA) was performed using clusterProfiler (version 4.2.2). Immune-cell-related genes⁷⁰ and hallmark signatures were analyzed with the single-sample GSEA (ssGSEA) method in Gene Set Variation Analysis R package (3.17)⁷¹ (R version 4.2.2). Hallmark signature gene sets were obtained from MSigDB (https://www.gsea-msigdb.org/gsea/msigdb/).

Luminex assays

Tumors were harvested 7 days after the initiation of RT and lysed with RIPA buffer (Beyotime, #P0013B). The levels of 26 mouse cytokines and chemokines were measured in tumor lysis using Luminex assays (Invitrogen, EPX260-26088-901) according to the manufacturer´s protocol. The concentration of cytokines and chemokines was normalized by tumor weight.

Bulk TCR-seq

RNA was extracted from ex vivo tumor samples collected at d0, d4, d7, d10, and d16 post-iERT using the Qiagen kit (Cat. No. 74104) according to the manufacturer’s instructions. RNA quantification was carried out using the Qubit RNA BR Assay Kit (Thermo Fisher Scientific, Catalog No. Q10210) on the Qubit apparatus.

TCR amplification and sequencing were carried out following a previously published protocol with slight modifications^72,73. Briefly, cDNA derived from TCR transcripts was purified, followed by reverse transcription using 5’ Rapid Amplification of cDNA ends. The variable (V) regions of TCR genes were amplified via two rounds of PCR, employing C-region-specific reverse primers coupled with an adapter sequence. The resulting amplicons were then ligated to sequencing adapters for paired-end sequencing (150 bp) on the DNBSEQ-T7 platform to generate raw sequencing reads.

TCR repertoires were generated by using a licensed copy of PyIR, a minimally dependent high-speed wrapper for the immunoglobulin Basic Local Alignment Search Tool (IgBLAST) (v1.20.0), on the paired-end raw sequencing FASTQ files. The default options were used for the alignment and assembly (Leaimmucenter). Global diversity/clonality coefficients have been calculated as follows: Shannon’s diversity was calculated using the formula below, in which pi is the frequency of clonotype i for the sample with N unique clonotypes.

H^{'} = - \sum_{i = 1}^{N} p_{i} \times \ln (p_{i})

Normalized Shannon diversity entropy (NSDE) was:

N S D E = \frac{H^{'}}{\ln (N)}

NSDE index is a score bounded between 0 (monoclonality) and 1 (maximum diversity).

TCR convergence is determined as the aggregate frequency of clones, defined as unique TCR nucleotide sequences, that shared a variable gene and CDR3 amino acid sequence with at least one other clone. The Morisita Index (also known as Morisita’s overlap index) is used to measure the similarity of TCR clonotypes between two samples based on their species composition and abundance. The Morisita Index formula is:

M = \frac{2 \times Σ (Xi \times Yi)}{[(\frac{Σ ({Xi}^{2})}{X^{2}}) + (\frac{Σ ({Yi}^{2})}{Y^{2}})] \times X \times Y}

Where:

Xi is the number of individuals (reads/frequency) of species i in sample X;

Yi is the number of individuals of species i in sample Y;

X is the total number of individuals in sample X;

Y is the total number of individuals in sample Y.

For TCR analysis specifically:

Xi represents the frequency/count of a specific TCR clonotype in sample X;

Yi represents the frequency/count of the same TCR clonotype in sample Y;

X represents the total number of TCR sequences in sample X;

Y represents the total number of TCR sequences in sample Y.

Steps to calculate: 1. Identify shared clonotypes between both samples; 2. For each shared clonotype, multiply its frequencies in both samples (Xi × Yi); 3. Sum all these products. Multiply by 2; 4. Divide by the denominator as shown in the formula. In our study, Python 3.9 to implement this calculation and derive the morisita index.

TCR-seq was performed and analyzed by Leading Biology Inc.

scRNA-seq

In the bulky CT26 model, tumors and TDLNs were resected on days 4 and 10 in four treatment groups, namely untreated, LDRT + αPD-1, SBRT + αPD-1, and iERT. The tumors collected on day 4 were divided into SBRT zones and LDRT zones, while the tumors collected on day 10 were not divided because the tumor volume and structure changed significantly compared with the baseline. Then, samples were enzymatically digested to generate single-cell suspensions by Novogene Co., Ltd. using the Tumor Dissociation Kit (mouse, Miltenyi, Cat# 130-096-730), in accordance with the Cell Preparation for Single Cell Protocols (10x Genomics), which were adjusted to a concentration of 700–1200 cells/μl. Only cell suspensions that passed quality control with a viability exceeding 80% were selected for scRNA-seq. Single-cell library preparation and sequencing were performed by Novogene Co., Ltd. For each sample, a target capture of 10,000 cells was set, and the appropriate loading volume and total cell number were calculated in accordance with the manufacturer’s guidelines for the Chromium Next GEM Single Cell 5’ Reagent Kits v2 (10x Genomics). ScRNA-seq libraries were prepared using the Chromium Next GEM Single Cell 5’ Library & Gel Bead Kit (10x Genomics, Cat# PN-1000165) and Single Cell V(D)J Kit (10x Genomics, Cat# PN-1000071). The sequencing was performed on an Illumina NovaSeq 6000 platform, producing 150 bp paired-end reads. Gene expression and TCR sequences were identified using the multi-pipeline in the Cell Ranger (version 7.0.1) with the mouse reference genome (mm10). Quality control metrics were generated and evaluated.

Gene expression data from each tumor and TDLN sample were processed into Seurat objects using the Seurat package (version 4.1.0). Cells with 500–8000 detected genes and 1000–50,000 unique molecular identifiers were retained. Cells exhibiting a mitochondrial gene proportion exceeding 5% were excluded, along with doublets identified through Scrublet (version 0.2.3). Expression data of samples from tumors and lymph nodes were independently merged and analyzed. The expression matrix was normalized and scaled, and 2000 variable genes were identified using the FindVariableFeatures function for subsequent principal component analysis, retaining the top 50 principal components. Harmony (version 0.1.0) was applied to accurately integrate data from different samples. Subsequent analyses were performed based on the corrected embedding from Harmony principal components, with the RunUMAP function for visualization, and the FindNeighbors and FindClusters functions for unsupervised clustering. Cell types were manually annotated based on canonical marker genes. To achieve finer resolution within cell subsets, the Harmony integration and downstream steps were repeated, enabling more detailed clustering and analysis. Cells with high expression of skeletal muscle marker genes were identified as sampling artifacts, while those with high expression of ribosome-related genes were classified as low-quality populations. Both were excluded from subsequent analyses.

To further analyze the distribution of cell types across samples, odds ratios (OR) were calculated to quantify the relative enrichment or depletion of specific cell types. For each combination of cell type i and sample j, a 2 × 2 contingency table was constructed, summarizing: (1) the number of cells of type i in sample j, (2) the number of cells of type i in all other samples, (3) the number of non-i cells in sample j, and (4) the number of non-i cells in all other samples. Fisher’s exact test was then applied to this contingency table to calculate the OR and the corresponding P-value. P-values were adjusted for multiple comparisons using the Benjamini-Hochberg (BH) method via the p.adjust function in R.

DGE analysis for each cell type was conducted using MAST (version 1.20.0). Genes with adjusted P-values < 0.05 and log₂(fold change) > 0.5, identified as differentially expressed compared to other cells, were used for immune response enrichment analysis (IREA) to evaluate cytokine activation. In addition, genes with adjusted P-values < 0.05 and log₂(fold change) > 0.25 were selected to create heatmaps, highlighting the heterogeneity of gene expression across cell types. Gene set variation analysis (GSVA) was performed using GSVA (version 1.42.0) to assess the pathway enrichment in the cells. Gene sets for NK cells and cDCs were sourced from the MSigDB and Gene Ontology (GO) databases.

The signature scores for CD8⁺ T cells were calculated using AUCell (version 1.16.0). The stemness score was calculated based on the marker genes Tcf7, Cxcr5, Cd28, Gzmk, Ccr7, Il7r, Bcl6, Sell, and Cd27. The naiveness score was calculated based on the marker genes Ccr7, Tcf7, Lef1, Sell, and Il7r. The cytotoxicity score was calculated based on the marker genes Prf1, Ifng, Nkg7, Gzmb, Gzma, Klrk1, Klrb1, Ctsw, and Cst7. The exhaustion score was calculated based on the marker genes Pdcd1, Havcr2, Lag3, Ctla4, Tox, Icos, Tnfrsf4, and Tigit. UMAP plots displaying the distribution density of signature scores were generated using Nebulosa (version 1.4.0) with the weighted kernel density estimation method.

For T cells with paired single-cell V(D)J sequencing data generated on the same libraries, cells with identical TCR genotypes and CDR3 nucleotide sequences were considered to belong to the same TCR clone. Clone sizes were calculated using scRepertoire (version 2.0.4). To assess the diversity of TCR clones in CD8⁺ T cells across samples, the Shannon diversity index $H$ was calculated as follows:

H = - \sum_{i = 1}^{M} p_{i} \ln p_{i}

$p_{i}$ was the frequency of TCR clone $i$ , and $M$ was the total number of TCR clone types. A higher index indicates a more even distribution of TCR clones within the sample, whereas a lower index suggests a dominance of specific TCR clones within the sample.

The cell–cell interaction analyses were performed and visualized using the R package CellChat (version 2.1.2). The computeCommunProb function was used to calculate communication probabilities between cell types within tumor samples from each treatment group. To reduce dropout effects, particularly the zero expression of ligand or receptor subunits, smoothed projected data were used for the analysis by adjusting the corresponding parameter. Furthermore, the relative proportion of each cell type was incorporated into the analysis to minimize potential biases related to differences in population size.

EV isolation uptake imaging cross-priming assay and visualizing

EVs were isolated from the supernatant from untreated or irradiated CT26 and B16-OVA cells. Fetal bovine serum (FBS) used for cell culture was ultracentrifuged at 110,000 × g for 3 h at 4 °C to deplete FBS exosomes. Cell culture supernatant was harvested and filtered through the 0.22 μm filter (Millipore, Cat# SLGPR33RB), and cell-derived EVs were extracted by ultracentrifugation. Briefly, cell supernatant was centrifuged at 300×g for 10 min to remove cells; 2000 × g for 20 min, and 8000×g for 30 min to remove cell debris. Finally, the medium was centrifuged at 110, 000 × g for 70 min at 4 °C in an SW32-Ti rotor and Optima XPN-100 ultracentrifuge (Beckman). The supernatant was discarded, and the EVs were washed with PBS and ultracentrifuged again at 110,000 × g for 70 min at 4 °C to remove impurities further. Finally, the EVs were suspended with PBS and stored at − 80 °C. ZetaView (Particle Metrix) was used to determine EV concentration.

BMDCs from naive C57BL/6 mice and Balb/c mice were cultured in complete RPMI 1640 (Gibco, Cat# 11875093) and DMEM (Gibco, Cat# 11965092) medium supplemented with 10% FBS (Gibco, Cat# A5256701) and 1% penicillin‒streptomycin (SPERIKON, Cat# SP00303) for 7 days in the presence of 20 ng/mL murine granulocyte-macrophage colony stimulating factor (GM-CSF, Genescript, Z03300) and 10 ng/mL murine IL-4 (Genescript, Z02996) at 37 C, 5% CO₂. Then, BMDCs were seeded in CellCarrier-96 Ultra plate (Revvity, Cat# 6055300) at a density of 1 × 10⁴/well before stained with CFDA SE probe (Beyotime, C1031). EVs were isolated from B16-OVA cells and CT26 cells untreated or treated with 10 Gy × 3 irradiation after 48 hours. And EVs derived from 1 × 10⁵ tumor cells were cocultured with BMDCs in a CellCarrier-96 Ultra plate before stained with DiI dye (Beyotime, C1036). EVs uptake imaging was performed then by the Opera Phenix Plus imaging system (PerkinElmer).

B16-OVA cells and CT26 cells were treated in three ways: treated by radiation with 2 Gy × 3, 10 Gy × 3 and 10 Gy × 3 plus EV inhibitor GW4869 (MCE, HY-19363). Untreated cells served as the negative control. EVs were isolated 48 h after treatment. EVs derived from 1 × 10⁷ B16-OVA cells and CT26 cells were incubated with 5 × 10⁵ C57BL/6 BMDCs and Balb/c BMDCs, respectively. These C57BL/6 BMDCs and Balb/c BMDCs were co-cultured with OT-I T cells and TDLN-derived T cells, respectively, at a ratio of 1:3, which were isolated from TDLNs of Balb/c mice bearing CT26 treated with 10 Gy × 3 + αPD-1 on day 9. After 72 h of DC-T cell co-culture, the DC and T cells were analyzed by flow cytometry.

EVs were captured with an Exosome detection kit (NanoView Biosciences, Cat# EV-TETRA-M2), which was coated with exosomal antibodies against CD81 and CD9. IgG isotype was used as a negative control. After capturing EVs, the Exoview Tetraspanin Chips were incubated with anti-CD81 and anti-CD9 to characterize EVs. OVA257-264 (SIINFEKL) peptide bound to H-2Kb monoclonal antibody-APC (Thermo Fisher Scientific Cat# 13-5743-82) to detect EVs containing SIINFEKL. After denaturation of the probe at 75 °C for 10 min, the probe or antibody solution was incubated with the chip overnight. Finally, the chip was imaged using the ExoView R200 Reader.

Patient case

All procedures involving human participants were conducted in accordance with the Declaration of Helsinki. The patient case presented in Fig.7f was approved by the Ethics Committee of West China Hospital, Sichuan University (approval no. 2023-403). Written informed consent was obtained from the patient prior to treatment, including consent for publication of anonymized data and figures.

Statistical analysis

Previous experience served as the basis for calculations of expected averages and deviations to calculate the sample size of mouse experiments to achieve 80% power. This typically results in a minimum sample size of 5 mice, depending on the experiment. Based on pre-treatment tumor volume, mice were blindly randomly assigned to the treatment groups. The number of mice in each group and the statistics are included in the figures and legends, respectively. Outliers were not removed.

The unpaired two-tailed Student’s t test was used for two-group comparisons. For multiple comparisons, one-way or two-way ANOVA followed by the “Two-stage step-up method of Benjamini, Krieger and Yekutieli” was used. Survival data were compared using the log-rank test. P < 0.05 was considered significant. All analyses were performed using GraphPad Prism version 7.0.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary Information^{(16.2MB, pdf)}

41467_2026_73683_MOESM2_ESM.pdf^{(70.3KB, pdf)}

Description of Additional Supplementary Files

Supplementary Data 1^{(27.3KB, xlsx)}

Reporting Summary^{(124.6KB, pdf)}

Transparent Peer Review file^{(9.4MB, pdf)}

Source data

Source Data^{(15.4MB, xlsx)}

Acknowledgements

We thank Li Li, Fei Chen, Chunjun Bao, and Yang Deng from the Institute of Clinical Pathology of West China Hospital, Sichuan University, Sichuan, China, for their generous help with histological staining. We thank Zhenru Wu from the Institute of Clinical Pathology, Key Laboratory of Transplant Engineering and Immunology, NHC, West China Hospital, Sichuan University, Sichuan, China, for her help with the Luminex assays. We thank Qing Yang from the Preclinical Imaging Platform, West China Hospital, Sichuan University, for the help in small animal irradiation.

Author contributions

Y.L. conceptualized and supervised the project. R.L., M.Y., J.X.X., and Y.L. designed the study. R.L., M.Y., K.K., Y.F.Z. and Z.S. executed, interpreted, and analyzed the data. S.H.L., Z.R.Y., Z.C.P., X.W.Z., S.S.L., L.L.Y., H.W., Y.F.Z., W.H.Z., Y.Z., G.L., R.Z.T., F.F.N., Y.L.G., M.J.H., B.W.Z., and G.N. were involved in the collection and assembly of data. R.L., M.Y., K.K., J.X.X., and Y.L. wrote the original draft. L.B., R.M.Z. and D.Y. contributed to quality assurance and control (QA & QC) of the mouse radiotherapy. All authors revised the manuscript and approved the final version submitted for publication.

Peer review

Peer review information

Nature Communications thanks Michele Mondini, who co-reviewed with Paul Bergeron, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work is supported by the Original Exploration Program of the National Natural Science Foundation of China, No.82350128 (YL); 1. 3. 5 project for disciplines of excellence, West China Hospital, Sichuan University, No. 25HXJS004 (Y.L.); West China Hospital, Sichuan University the 1·3·5 Project for Disciplines of Excellence, No. ZYJC21003 (Y.L.) and ZYYC23010 (J.X.X.); National Natural Science Foundation of China, No.82303694 (R.L.); Sichuan Provincial Research Foundation, No. 2026NSFSC0652 (R.L.); Sichuan University “From 0 to 1” Innovation Research Project, No. 2023SCUH0045 (R.L.).

Data availability

Raw sequencing data have been deposited in the Genome Sequence Archive at the National Genomics Data Center under accession numbers CRA023520 and CRA023557. Processed sequencing data are available at Zenodo (https://zenodo.org/records/17662223). All data are included in the Supplementary Information or available from the authors, as are unique reagents used in this Article. The raw numbers for charts and graphs are available in the Source Data file whenever possible. Source data are provided in this paper.

Code availability

Scripts for scRNA-seq analysis are available at GitHub (https://github.com/k2med/iERT_scRNAseq). All software used in this study is publicly available and described in the Methods section. No customized code was generated for this study.

Competing interests

Y. Lu has an invited speaker/project lead/principal investigator role with Roche, AstraZeneca, BeiGene, and Hengrui Pharmaceuticals; and an invited speaker role with Pfizer, Merck Sharp & Dohme, and Bristol-Myers Squibb. The remaining authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors contributed equally: Ren Luo, Min Yu, Kai Kang, Yufeng Zhang.

Contributor Information

Jianxin Xue, Email: radjianxin@163.com.

You Lu, Email: radyoulu@hotmail.com.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-73683-z.

References

1.Lynch, C., Pitroda, S. P. & Weichselbaum, R. R. Radiotherapy, immunity, and immune checkpoint inhibitors. Lancet Oncol.25, e352–e362 (2024). [DOI] [PubMed] [Google Scholar]
2.Galluzzi, L., Aryankalayil, M. J., Coleman, C. N. & Formenti, S. C. Emerging evidence for adapting radiotherapy to immunotherapy. Nat. Rev. Clin. Oncol.20, 543–557 (2023). [DOI] [PubMed] [Google Scholar]
3.Lee, N. Y. et al. Avelumab plus standard-of-care chemoradiotherapy versus chemoradiotherapy alone in patients with locally advanced squamous cell carcinoma of the head and neck: a randomised, double-blind, placebo-controlled, multicentre, phase 3 trial. Lancet Oncol.22, 450–462 (2021). [DOI] [PubMed] [Google Scholar]
4.Theelen, W. S. M. E. et al. Pembrolizumab with or without radiotherapy for metastatic non-small-cell lung cancer: a pooled analysis of two randomised trials. Lancet Respir. Med.9, 467–475 (2021). [DOI] [PubMed] [Google Scholar]
5.Bradley, J. D. et al. LBA1 Durvalumab in combination with chemoradiotherapy for patients with unresectable stage III NSCLC: Final results from PACIFIC-2. ESMO Open9, 102986 (2024).
6.Lim, M. et al. Phase III trial of chemoradiotherapy with temozolomide plus nivolumab or placebo for newly diagnosed glioblastoma with methylated MGMT promoter. Neuro Oncol.24, 1935–1949 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Huang, A. C. et al. T-cell invigoration to tumour burden ratio associated with anti-PD-1 response. Nature545, 60–65 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Binnewies, M. et al. Understanding the tumor immune microenvironment (TIME) for effective therapy. Nat. Med.24, 541–550 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Socinski, M. A. et al. Association of immune-related adverse events with efficacy of atezolizumab in patients with non-small cell lung cancer: pooled analyses of the phase 3 IMpower130, IMpower132, and IMpower150 randomized clinical trials. JAMA Oncol.9, 527–535 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Beck, M. et al. Differences and similarities in the pattern of early metabolic and morphologic response after induction chemo-immunotherapy versus induction chemotherapy alone in locally advanced squamous cell head and neck cancer. Cancers14, 10.3390/cancers14194811 (2022). [DOI] [PMC free article] [PubMed]
11.Grambow-Velilla, J. et al. Total metabolic tumor volume on 18F-FDG PET/CT is a useful prognostic biomarker for patients with extensive small-cell lung cancer undergoing first-line chemo-immunotherapy. Cancers15, 2223 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Kim, S. I., Cassella, C. R. & Byrne, K. T. Tumor burden and immunotherapy: impact on immune infiltration and therapeutic outcomes. Front. Immunol.11, 629722 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Rockwell, S., Dobrucki, I. T., Kim, E. Y., Marrison, S. T. & Vu, V. T. Hypoxia and radiation therapy: past history, ongoing research, and future promise. Curr. Mol. Med.9, 442–458 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Timmerman, R. et al. Excessive toxicity when treating central tumors in a phase II study of stereotactic body radiation therapy for medically inoperable early-stage lung cancer. J. Clin. Oncol.24, 4833–4839 (2006). [DOI] [PubMed] [Google Scholar]
15.Przybysz, D. & Bradley, J. D. Challenging situations for lung SBRT. J. Thorac. Oncol.12, 916–918 (2017). [DOI] [PubMed] [Google Scholar]
16.Dawson, L. A., Hashem, S. & Bujold, A. Stereotactic body radiation therapy for hepatocellular carcinoma. in American Society of Clinical Oncology Educational Book (2012). [DOI] [PubMed]
17.Song, C. W. et al. Biological principles of stereotactic body radiation therapy (SBRT) and stereotactic radiation surgery (SRS): indirect cell death. Int. J. Radiat. Oncol. Biol. Phys.110, 21–34 (2021). [DOI] [PubMed] [Google Scholar]
18.Citrin, D. E. Recent developments in radiotherapy. N. Engl. J. Med.377, 1065–1075 (2017). [DOI] [PubMed] [Google Scholar]
19.Casares, N. et al. Caspase-dependent immunogenicity of doxorubicin-induced tumor cell death. J. Exp. Med.202, 1691–1701 (2005). [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Golden, E. B. et al. Radiation fosters dose-dependent and chemotherapy-induced immunogenic cell death. Oncoimmunology3, e28518 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Ngwa, W. et al. Using immunotherapy to boost the abscopal effect. Nat. Rev. Cancer18, 313–322 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Buchwald, Z. S. et al. Tumor-draining lymph node is important for a robust abscopal effect stimulated by radiotherapy. J. Immunother. Cancer8, e000867 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Chang, J. Y. et al. Stereotactic ablative radiotherapy with or without immunotherapy for early-stage or isolated lung parenchymal recurrent node-negative non-small-cell lung cancer: an open-label, randomised, phase 2 trial. Lancet402, 871–881 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Pitroda, S. P., Chmura, S. J. & Weichselbaum, R. R. Integration of radiotherapy and immunotherapy for treatment of oligometastases. Lancet Oncol.20, e434–e442 (2019). [DOI] [PubMed] [Google Scholar]
25.Zhao, J. et al. Simple factors associated with radiation-induced lung toxicity after stereotactic body radiation therapy of the thorax: a pooled analysis of 88 studies. Int. J. Radiat. Oncol. Biol. Phys.95, 1357–1366 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Videtic, G. M. M. et al. Stereotactic body radiation therapy for early-stage non-small cell lung cancer: executive summary of an ASTRO evidence-based guideline. Pr. Radiat. Oncol.7, 295–301 (2017). [DOI] [PubMed] [Google Scholar]
27.Klug, F. et al. Low-dose irradiation programs macrophage differentiation to an iNOS + /M1 phenotype that orchestrates effective T cell immunotherapy. Cancer Cell24, 589–602 (2013). [DOI] [PubMed] [Google Scholar]
28.Yin, L. et al. Effect of low-dose radiation therapy on abscopal responses to hypofractionated radiation therapy and anti-PD1 in mice and patients with non-small cell lung cancer. Int. J. Radiat. Oncol. Biol. Phys.108, 212–224 (2020). [DOI] [PubMed] [Google Scholar]
29.Barsoumian, H. B. et al. Low-dose radiation treatment enhances systemic antitumor immune responses by overcoming the inhibitory stroma. J. Immunother. Cancer8, e000537 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Herrera, F. G. et al. Low-dose radiotherapy reverses tumor immune desertification and resistance to immunotherapy. Cancer Discov.12, 108–133 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Zhou, X. et al. Safety and tolerability of low-dose radiation and stereotactic body radiotherapy + sintilimab for treatment-naïve stage IV PD-L1+ non-small cell lung cancer patients. Clin. Cancer Res.29, 4098–4108 (2023). [DOI] [PubMed] [Google Scholar]
32.Griffin, R. J. et al. Understanding high-dose, ultra-high dose rate, and spatially fractionated radiation therapy. Int. J. Radiat. Oncol. Biol. Phys.107, 766–778 (2020). [DOI] [PubMed] [Google Scholar]
33.Iori, F. et al. Lattice radiation therapy in clinical practice: A systematic review. Clin. Transl. Radiat. Oncol.39, 100569 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Tubin, S., Popper, H. H. & Brcic, L. Novel stereotactic body radiation therapy (SBRT)-based partial tumor irradiation targeting hypoxic segment of bulky tumors (SBRT-PATHY): improvement of the radiotherapy outcome by exploiting the bystander and abscopal effects. Radiat. Oncol.14, 21 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Jagodinsky, J. C. et al. Intratumoral radiation dose heterogeneity augments antitumor immunity in mice and primes responses to checkpoint blockade. Sci. Transl. Med.16, eadk0642 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Bergeron, P. et al. Non-homogenous intratumor ionizing radiation doses synergize with PD1 and CXCR2 blockade. Nat. Commun.15, 8845 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Luo, R. et al. The eclipse-radiotherapy combined with ICI orchestrates CD8 + T cell- and NK cell-dependent anti-tumor efficacy against bulky tumors. Int. J. Radiat. Oncol.*Biol.*Phys.120, S95 (2024). [Google Scholar]
38.Vanpouille-Box, C. et al. DNA exonuclease Trex1 regulates radiotherapy-induced tumour immunogenicity. Nat. Commun.8, 15618 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Luo, R. et al. Cisplatin facilitates radiation-induced abscopal effects in conjunction with PD-1 checkpoint blockade through CXCR3/CXCL10-mediated T-cell recruitment. Clin. Cancer Res.25, 7243–7255 (2019). [DOI] [PubMed] [Google Scholar]
40.Luo, R. et al. Necroptosis-dependent immunogenicity of cisplatin: implications for enhancing the radiation-induced abscopal effect. Clin. Cancer Res.29, 667–683 (2023). [DOI] [PubMed] [Google Scholar]
41.Sitnikova, S. I. et al. Novel non-terminal tumor sampling procedure using fine needle aspiration supports immuno-oncology biomarker discovery in preclinical mouse models. J. Immunother. Cancer9, 10.1136/jitc-2021-002894 (2021). [DOI] [PMC free article] [PubMed]
42.Björkström, N. K., Strunz, B. & Ljunggren, H. G. Natural killer cells in antiviral immunity. Nat. Rev. Immunol.22, 112–123 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Onyshchenko, K. et al. Expansion of circulating stem-like CD8(+) T cells by adding CD122-directed IL-2 complexes to radiation and anti-PD1 therapies in mice. Nat. Commun.14, 2087 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Wu, Y. et al. Neutrophil profiling illuminates anti-tumor antigen-presenting potency. Cell187, 1422–1439 (2024). [DOI] [PubMed] [Google Scholar]
45.Cui, A. et al. Dictionary of immune responses to cytokines at single-cell resolution. Nature625, 377–384 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Böttcher, J. P. et al. NK Cells stimulate recruitment of cDC1 into the tumor microenvironment promoting cancer immune control. Cell172, 1022–1037 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Luberto, C. et al. Inhibition of tumor necrosis factor-induced cell death in MCF7 by a novel inhibitor of neutral sphingomyelinase. J. Biol. Chem.277, 41128–41139 (2002). [DOI] [PubMed] [Google Scholar]
48.Porgador, A., Yewdell, J. W., Deng, Y., Bennink, J. R. & Germain, R. N. Localization, quantitation, and in situ detection of specific peptide-MHC class I complexes using a monoclonal antibody. Immunity6, 715–726 (1997). [DOI] [PubMed] [Google Scholar]
49.Hashimoto, M. et al. CD8 T Cell exhaustion in chronic infection and cancer: opportunities for interventions. Annu. Rev. Med.69, 301–318 (2018). [DOI] [PubMed] [Google Scholar]
50.Im, S. J. et al. Defining CD8 + T cells that provide the proliferative burst after PD-1 therapy. Nature537, 417–421 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Siddiqui, I. et al. Intratumoral Tcf1+PD-1 + CD8 + T cells with stem-like properties promote tumor control in response to vaccination and checkpoint blockade immunotherapy. Immunity50, 195–211 (2019). [DOI] [PubMed] [Google Scholar]
52.Miller, B. C. et al. Subsets of exhausted CD8(+) T cells differentially mediate tumor control and respond to checkpoint blockade. Nat. Immunol.20, 326–336 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
53.Huang, A. Y. et al. The immunodominant major histocompatibility complex class I-restricted antigen of a murine colon tumor derives from an endogenous retroviral gene product. Proc. Natl. Acad. Sci. USA93, 9730–9735 (1996). [DOI] [PMC free article] [PubMed] [Google Scholar]
54.Prokhnevska, N. et al. CD8(+) T cell activation in cancer comprises an initial activation phase in lymph nodes followed by effector differentiation within the tumor. Immunity56, 107–124 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Dall’Olio, F. G. et al. Tumour burden and efficacy of immune-checkpoint inhibitors. Nat. Rev. Clin. Oncol.19, 75–90 (2022). [DOI] [PubMed] [Google Scholar]
56.Braun, D. A. et al. Progressive immune dysfunction with advancing disease stage in renal cell carcinoma. Cancer Cell39, 632–648 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Sen Santara, S. et al. The NK cell receptor NKp46 recognizes ecto-calreticulin on ER-stressed cells. Nature616, 348–356 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Yan, D. et al. Inter/intra-tumoral dose response variations assessed using FDG-PET/CT feedback images: Impact on tumor control and treatment dose prescription. Radiother. Oncol.154, 235–242 (2021). [DOI] [PubMed] [Google Scholar]
59.Zitvogel, L., Apetoh, L., Ghiringhelli, F. & Kroemer, G. Immunological aspects of cancer chemotherapy. Nat. Rev. Immunol.8, 59–73 (2008). [DOI] [PubMed] [Google Scholar]
60.Diamond, M. S. et al. Type I interferon is selectively required by dendritic cells for immune rejection of tumors. J. Exp. Med.208, 1989–2003 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
61.Deng, L. et al. STING-Dependent cytosolic DNA sensing promotes radiation-induced type I interferon-dependent antitumor immunity in immunogenic tumors. Immunity41, 843–852 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Wennerberg, E. et al. CD73 Blockade promotes dendritic cell infiltration of irradiated tumors and tumor rejection. Cancer Immunol. Res.8, 465–478 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
63.Mathieu, M. et al. Activation of sting in response to partial-tumor radiation exposure. Int. J. Radiat. Oncol. Biol. Phys.117, 955–965 (2023). [DOI] [PMC free article] [PubMed]
64.Tooley, K. A., Escobar, G. & Anderson, A. C. Spatial determinants of CD8(+) T cell differentiation in cancer. Trends Cancer8, 642–654 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Luke, J. J. et al. Safety and clinical activity of pembrolizumab and multisite stereotactic body radiotherapy in patients with advanced solid tumors. J. Clin. Oncol.36, 1611–1618 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
66.Korpics, M. C. et al. Partial tumor irradiation plus pembrolizumab in treating large advanced solid tumor metastases. J. Clin. Invest.133, e162260 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Zemek, R. M. et al. Temporally restricted activation of IFNβ signaling underlies response to immune checkpoint therapy in mice. Nat. Commun.13, 4895 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Zemek, R. M., Anagnostou, V., Pires da Silva, I., Long, G. V. & Lesterhuis, W. J. Exploiting temporal aspects of cancer immunotherapy. Nat. Rev. Cancer24, 480–497 (2024). [DOI] [PubMed] [Google Scholar]
69.Bankhead, P. et al. QuPath: Open source software for digital pathology image analysis. Sci. Rep.7, 16878 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
70.Charoentong, P. et al. Pan-cancer immunogenomic analyses reveal genotype-immunophenotype relationships and predictors of response to checkpoint blockade. Cell Rep.18, 248–262 (2017). [DOI] [PubMed] [Google Scholar]
71.Hänzelmann, S., Castelo, R. & Guinney, J. GSVA: gene set variation analysis for microarray and RNA-Seq data. BMC Bioinform.14, 7 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
72.Schmidt, J. et al. Neoantigen-specific CD8 T cells with high structural avidity preferentially reside in and eliminate tumors. Nat. Commun.14, 3188 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
73.Zvyagin, I. V. et al. Tracking T-cell immune reconstitution after TCRαβ/CD19-depleted hematopoietic cells transplantation in children. Leukemia31, 1145–1153 (2017). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information^{(16.2MB, pdf)}

41467_2026_73683_MOESM2_ESM.pdf^{(70.3KB, pdf)}

Description of Additional Supplementary Files

Supplementary Data 1^{(27.3KB, xlsx)}

Reporting Summary^{(124.6KB, pdf)}

Transparent Peer Review file^{(9.4MB, pdf)}

Source Data^{(15.4MB, xlsx)}

Data Availability Statement

[CR1] 1.Lynch, C., Pitroda, S. P. & Weichselbaum, R. R. Radiotherapy, immunity, and immune checkpoint inhibitors. Lancet Oncol.25, e352–e362 (2024). [DOI] [PubMed] [Google Scholar]

[CR2] 2.Galluzzi, L., Aryankalayil, M. J., Coleman, C. N. & Formenti, S. C. Emerging evidence for adapting radiotherapy to immunotherapy. Nat. Rev. Clin. Oncol.20, 543–557 (2023). [DOI] [PubMed] [Google Scholar]

[CR3] 3.Lee, N. Y. et al. Avelumab plus standard-of-care chemoradiotherapy versus chemoradiotherapy alone in patients with locally advanced squamous cell carcinoma of the head and neck: a randomised, double-blind, placebo-controlled, multicentre, phase 3 trial. Lancet Oncol.22, 450–462 (2021). [DOI] [PubMed] [Google Scholar]

[CR4] 4.Theelen, W. S. M. E. et al. Pembrolizumab with or without radiotherapy for metastatic non-small-cell lung cancer: a pooled analysis of two randomised trials. Lancet Respir. Med.9, 467–475 (2021). [DOI] [PubMed] [Google Scholar]

[CR5] 5.Bradley, J. D. et al. LBA1 Durvalumab in combination with chemoradiotherapy for patients with unresectable stage III NSCLC: Final results from PACIFIC-2. ESMO Open9, 102986 (2024).

[CR6] 6.Lim, M. et al. Phase III trial of chemoradiotherapy with temozolomide plus nivolumab or placebo for newly diagnosed glioblastoma with methylated MGMT promoter. Neuro Oncol.24, 1935–1949 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR7] 7.Huang, A. C. et al. T-cell invigoration to tumour burden ratio associated with anti-PD-1 response. Nature545, 60–65 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR8] 8.Binnewies, M. et al. Understanding the tumor immune microenvironment (TIME) for effective therapy. Nat. Med.24, 541–550 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR9] 9.Socinski, M. A. et al. Association of immune-related adverse events with efficacy of atezolizumab in patients with non-small cell lung cancer: pooled analyses of the phase 3 IMpower130, IMpower132, and IMpower150 randomized clinical trials. JAMA Oncol.9, 527–535 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR10] 10.Beck, M. et al. Differences and similarities in the pattern of early metabolic and morphologic response after induction chemo-immunotherapy versus induction chemotherapy alone in locally advanced squamous cell head and neck cancer. Cancers14, 10.3390/cancers14194811 (2022). [DOI] [PMC free article] [PubMed]

[CR11] 11.Grambow-Velilla, J. et al. Total metabolic tumor volume on 18F-FDG PET/CT is a useful prognostic biomarker for patients with extensive small-cell lung cancer undergoing first-line chemo-immunotherapy. Cancers15, 2223 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR12] 12.Kim, S. I., Cassella, C. R. & Byrne, K. T. Tumor burden and immunotherapy: impact on immune infiltration and therapeutic outcomes. Front. Immunol.11, 629722 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR13] 13.Rockwell, S., Dobrucki, I. T., Kim, E. Y., Marrison, S. T. & Vu, V. T. Hypoxia and radiation therapy: past history, ongoing research, and future promise. Curr. Mol. Med.9, 442–458 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR14] 14.Timmerman, R. et al. Excessive toxicity when treating central tumors in a phase II study of stereotactic body radiation therapy for medically inoperable early-stage lung cancer. J. Clin. Oncol.24, 4833–4839 (2006). [DOI] [PubMed] [Google Scholar]

[CR15] 15.Przybysz, D. & Bradley, J. D. Challenging situations for lung SBRT. J. Thorac. Oncol.12, 916–918 (2017). [DOI] [PubMed] [Google Scholar]

[CR16] 16.Dawson, L. A., Hashem, S. & Bujold, A. Stereotactic body radiation therapy for hepatocellular carcinoma. in American Society of Clinical Oncology Educational Book (2012). [DOI] [PubMed]

[CR17] 17.Song, C. W. et al. Biological principles of stereotactic body radiation therapy (SBRT) and stereotactic radiation surgery (SRS): indirect cell death. Int. J. Radiat. Oncol. Biol. Phys.110, 21–34 (2021). [DOI] [PubMed] [Google Scholar]

[CR18] 18.Citrin, D. E. Recent developments in radiotherapy. N. Engl. J. Med.377, 1065–1075 (2017). [DOI] [PubMed] [Google Scholar]

[CR19] 19.Casares, N. et al. Caspase-dependent immunogenicity of doxorubicin-induced tumor cell death. J. Exp. Med.202, 1691–1701 (2005). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR20] 20.Golden, E. B. et al. Radiation fosters dose-dependent and chemotherapy-induced immunogenic cell death. Oncoimmunology3, e28518 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR21] 21.Ngwa, W. et al. Using immunotherapy to boost the abscopal effect. Nat. Rev. Cancer18, 313–322 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR22] 22.Buchwald, Z. S. et al. Tumor-draining lymph node is important for a robust abscopal effect stimulated by radiotherapy. J. Immunother. Cancer8, e000867 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR23] 23.Chang, J. Y. et al. Stereotactic ablative radiotherapy with or without immunotherapy for early-stage or isolated lung parenchymal recurrent node-negative non-small-cell lung cancer: an open-label, randomised, phase 2 trial. Lancet402, 871–881 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR24] 24.Pitroda, S. P., Chmura, S. J. & Weichselbaum, R. R. Integration of radiotherapy and immunotherapy for treatment of oligometastases. Lancet Oncol.20, e434–e442 (2019). [DOI] [PubMed] [Google Scholar]

[CR25] 25.Zhao, J. et al. Simple factors associated with radiation-induced lung toxicity after stereotactic body radiation therapy of the thorax: a pooled analysis of 88 studies. Int. J. Radiat. Oncol. Biol. Phys.95, 1357–1366 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR26] 26.Videtic, G. M. M. et al. Stereotactic body radiation therapy for early-stage non-small cell lung cancer: executive summary of an ASTRO evidence-based guideline. Pr. Radiat. Oncol.7, 295–301 (2017). [DOI] [PubMed] [Google Scholar]

[CR27] 27.Klug, F. et al. Low-dose irradiation programs macrophage differentiation to an iNOS + /M1 phenotype that orchestrates effective T cell immunotherapy. Cancer Cell24, 589–602 (2013). [DOI] [PubMed] [Google Scholar]

[CR28] 28.Yin, L. et al. Effect of low-dose radiation therapy on abscopal responses to hypofractionated radiation therapy and anti-PD1 in mice and patients with non-small cell lung cancer. Int. J. Radiat. Oncol. Biol. Phys.108, 212–224 (2020). [DOI] [PubMed] [Google Scholar]

[CR29] 29.Barsoumian, H. B. et al. Low-dose radiation treatment enhances systemic antitumor immune responses by overcoming the inhibitory stroma. J. Immunother. Cancer8, e000537 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR30] 30.Herrera, F. G. et al. Low-dose radiotherapy reverses tumor immune desertification and resistance to immunotherapy. Cancer Discov.12, 108–133 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR31] 31.Zhou, X. et al. Safety and tolerability of low-dose radiation and stereotactic body radiotherapy + sintilimab for treatment-naïve stage IV PD-L1+ non-small cell lung cancer patients. Clin. Cancer Res.29, 4098–4108 (2023). [DOI] [PubMed] [Google Scholar]

[CR32] 32.Griffin, R. J. et al. Understanding high-dose, ultra-high dose rate, and spatially fractionated radiation therapy. Int. J. Radiat. Oncol. Biol. Phys.107, 766–778 (2020). [DOI] [PubMed] [Google Scholar]

[CR33] 33.Iori, F. et al. Lattice radiation therapy in clinical practice: A systematic review. Clin. Transl. Radiat. Oncol.39, 100569 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR34] 34.Tubin, S., Popper, H. H. & Brcic, L. Novel stereotactic body radiation therapy (SBRT)-based partial tumor irradiation targeting hypoxic segment of bulky tumors (SBRT-PATHY): improvement of the radiotherapy outcome by exploiting the bystander and abscopal effects. Radiat. Oncol.14, 21 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR35] 35.Jagodinsky, J. C. et al. Intratumoral radiation dose heterogeneity augments antitumor immunity in mice and primes responses to checkpoint blockade. Sci. Transl. Med.16, eadk0642 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR36] 36.Bergeron, P. et al. Non-homogenous intratumor ionizing radiation doses synergize with PD1 and CXCR2 blockade. Nat. Commun.15, 8845 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR37] 37.Luo, R. et al. The eclipse-radiotherapy combined with ICI orchestrates CD8 + T cell- and NK cell-dependent anti-tumor efficacy against bulky tumors. Int. J. Radiat. Oncol.*Biol.*Phys.120, S95 (2024). [Google Scholar]

[CR38] 38.Vanpouille-Box, C. et al. DNA exonuclease Trex1 regulates radiotherapy-induced tumour immunogenicity. Nat. Commun.8, 15618 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR39] 39.Luo, R. et al. Cisplatin facilitates radiation-induced abscopal effects in conjunction with PD-1 checkpoint blockade through CXCR3/CXCL10-mediated T-cell recruitment. Clin. Cancer Res.25, 7243–7255 (2019). [DOI] [PubMed] [Google Scholar]

[CR40] 40.Luo, R. et al. Necroptosis-dependent immunogenicity of cisplatin: implications for enhancing the radiation-induced abscopal effect. Clin. Cancer Res.29, 667–683 (2023). [DOI] [PubMed] [Google Scholar]

[CR41] 41.Sitnikova, S. I. et al. Novel non-terminal tumor sampling procedure using fine needle aspiration supports immuno-oncology biomarker discovery in preclinical mouse models. J. Immunother. Cancer9, 10.1136/jitc-2021-002894 (2021). [DOI] [PMC free article] [PubMed]

[CR42] 42.Björkström, N. K., Strunz, B. & Ljunggren, H. G. Natural killer cells in antiviral immunity. Nat. Rev. Immunol.22, 112–123 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR43] 43.Onyshchenko, K. et al. Expansion of circulating stem-like CD8(+) T cells by adding CD122-directed IL-2 complexes to radiation and anti-PD1 therapies in mice. Nat. Commun.14, 2087 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR44] 44.Wu, Y. et al. Neutrophil profiling illuminates anti-tumor antigen-presenting potency. Cell187, 1422–1439 (2024). [DOI] [PubMed] [Google Scholar]

[CR45] 45.Cui, A. et al. Dictionary of immune responses to cytokines at single-cell resolution. Nature625, 377–384 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR46] 46.Böttcher, J. P. et al. NK Cells stimulate recruitment of cDC1 into the tumor microenvironment promoting cancer immune control. Cell172, 1022–1037 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR47] 47.Luberto, C. et al. Inhibition of tumor necrosis factor-induced cell death in MCF7 by a novel inhibitor of neutral sphingomyelinase. J. Biol. Chem.277, 41128–41139 (2002). [DOI] [PubMed] [Google Scholar]

[CR48] 48.Porgador, A., Yewdell, J. W., Deng, Y., Bennink, J. R. & Germain, R. N. Localization, quantitation, and in situ detection of specific peptide-MHC class I complexes using a monoclonal antibody. Immunity6, 715–726 (1997). [DOI] [PubMed] [Google Scholar]

[CR49] 49.Hashimoto, M. et al. CD8 T Cell exhaustion in chronic infection and cancer: opportunities for interventions. Annu. Rev. Med.69, 301–318 (2018). [DOI] [PubMed] [Google Scholar]

[CR50] 50.Im, S. J. et al. Defining CD8 + T cells that provide the proliferative burst after PD-1 therapy. Nature537, 417–421 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR51] 51.Siddiqui, I. et al. Intratumoral Tcf1+PD-1 + CD8 + T cells with stem-like properties promote tumor control in response to vaccination and checkpoint blockade immunotherapy. Immunity50, 195–211 (2019). [DOI] [PubMed] [Google Scholar]

[CR52] 52.Miller, B. C. et al. Subsets of exhausted CD8(+) T cells differentially mediate tumor control and respond to checkpoint blockade. Nat. Immunol.20, 326–336 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR53] 53.Huang, A. Y. et al. The immunodominant major histocompatibility complex class I-restricted antigen of a murine colon tumor derives from an endogenous retroviral gene product. Proc. Natl. Acad. Sci. USA93, 9730–9735 (1996). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR54] 54.Prokhnevska, N. et al. CD8(+) T cell activation in cancer comprises an initial activation phase in lymph nodes followed by effector differentiation within the tumor. Immunity56, 107–124 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR55] 55.Dall’Olio, F. G. et al. Tumour burden and efficacy of immune-checkpoint inhibitors. Nat. Rev. Clin. Oncol.19, 75–90 (2022). [DOI] [PubMed] [Google Scholar]

[CR56] 56.Braun, D. A. et al. Progressive immune dysfunction with advancing disease stage in renal cell carcinoma. Cancer Cell39, 632–648 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR57] 57.Sen Santara, S. et al. The NK cell receptor NKp46 recognizes ecto-calreticulin on ER-stressed cells. Nature616, 348–356 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR58] 58.Yan, D. et al. Inter/intra-tumoral dose response variations assessed using FDG-PET/CT feedback images: Impact on tumor control and treatment dose prescription. Radiother. Oncol.154, 235–242 (2021). [DOI] [PubMed] [Google Scholar]

[CR59] 59.Zitvogel, L., Apetoh, L., Ghiringhelli, F. & Kroemer, G. Immunological aspects of cancer chemotherapy. Nat. Rev. Immunol.8, 59–73 (2008). [DOI] [PubMed] [Google Scholar]

[CR60] 60.Diamond, M. S. et al. Type I interferon is selectively required by dendritic cells for immune rejection of tumors. J. Exp. Med.208, 1989–2003 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR61] 61.Deng, L. et al. STING-Dependent cytosolic DNA sensing promotes radiation-induced type I interferon-dependent antitumor immunity in immunogenic tumors. Immunity41, 843–852 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR62] 62.Wennerberg, E. et al. CD73 Blockade promotes dendritic cell infiltration of irradiated tumors and tumor rejection. Cancer Immunol. Res.8, 465–478 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR63] 63.Mathieu, M. et al. Activation of sting in response to partial-tumor radiation exposure. Int. J. Radiat. Oncol. Biol. Phys.117, 955–965 (2023). [DOI] [PMC free article] [PubMed]

[CR64] 64.Tooley, K. A., Escobar, G. & Anderson, A. C. Spatial determinants of CD8(+) T cell differentiation in cancer. Trends Cancer8, 642–654 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR65] 65.Luke, J. J. et al. Safety and clinical activity of pembrolizumab and multisite stereotactic body radiotherapy in patients with advanced solid tumors. J. Clin. Oncol.36, 1611–1618 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR66] 66.Korpics, M. C. et al. Partial tumor irradiation plus pembrolizumab in treating large advanced solid tumor metastases. J. Clin. Invest.133, e162260 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR67] 67.Zemek, R. M. et al. Temporally restricted activation of IFNβ signaling underlies response to immune checkpoint therapy in mice. Nat. Commun.13, 4895 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR68] 68.Zemek, R. M., Anagnostou, V., Pires da Silva, I., Long, G. V. & Lesterhuis, W. J. Exploiting temporal aspects of cancer immunotherapy. Nat. Rev. Cancer24, 480–497 (2024). [DOI] [PubMed] [Google Scholar]

[CR69] 69.Bankhead, P. et al. QuPath: Open source software for digital pathology image analysis. Sci. Rep.7, 16878 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR70] 70.Charoentong, P. et al. Pan-cancer immunogenomic analyses reveal genotype-immunophenotype relationships and predictors of response to checkpoint blockade. Cell Rep.18, 248–262 (2017). [DOI] [PubMed] [Google Scholar]

[CR71] 71.Hänzelmann, S., Castelo, R. & Guinney, J. GSVA: gene set variation analysis for microarray and RNA-Seq data. BMC Bioinform.14, 7 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR72] 72.Schmidt, J. et al. Neoantigen-specific CD8 T cells with high structural avidity preferentially reside in and eliminate tumors. Nat. Commun.14, 3188 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR73] 73.Zvyagin, I. V. et al. Tracking T-cell immune reconstitution after TCRαβ/CD19-depleted hematopoietic cells transplantation in children. Leukemia31, 1145–1153 (2017). [DOI] [PubMed] [Google Scholar]

PERMALINK

SBRT embedded in low-dose RT plus αPD-1 (immuno-EclipseRT, iERT) elicits CD8+ T cell immunity against bulky tumors via an IFN-I/NK/DC axis

Ren Luo

Min Yu

Kai Kang

Yufeng Zhang

Shanghai Liu

Zhuoran Yao

Zichong Peng

Xuanwei Zhang

Shuangsi Liao

Zhou Su

Linglu Yi

Hui Wang

Wenhao Zhao

Yue Zheng

Guo Lin

Ruizhan Tong

Feifei Na

Youling Gong

Meijuan Huang

Bingwen Zou

Long Bai

Renming Zhong

Di Yan

Gabriele Niedermann

Jianxin Xue

You Lu

Abstract

Introduction

Results

iERT achieves marked control of bulky tumors in mice

Fig. 1. Combining ERT and αPD-1 induces potent antitumor effects in mice with bulky tumors.

iERT elicits early NK cell and subsequent CD8+ T cell responses

Fig. 2. iERT induces early NK cell and subsequent CD8+ T cell infiltration into bulky mouse tumors.

iERT induced tumor infiltration by chemokine/cytokine-secreting NK cells

Fig. 3. iERT enhances NK cell-mediated antitumor immune responses.

iERT induces IFN-I which drives NK cell activation

Fig. 4. iERT-induced IFN-I triggers NK cell-mediated antitumor responses.

iERT recruits XCR1+ cDC1s through NK-derived XCL1

Fig. 5. iERT enhances XCR1+ cDC1 infiltration and tumor control via XCL1.

iERT induces CD8+ T cell responses via the IFN-I/NK/XCL1 axis

Fig. 6. iERT enhances CD8+ T cell responses through IFN-I/NK/XCL1 axis.

Fig. 7. iERT induces CD8+ T cell responses in tumor-draining lymph nodes.

Discussion

Methods

Mice and cell lines

Establishment of tumor models

iERT for mice

Western blotting

ELISA

Flow-cytometric analyses

NK cell isolation, sorting and coincubation with tumor cells

IF of tumor slices

mIHC of tumor slices

Quantitative reverse transcription polymerase chain reaction (RT-qPCR)

Bulk RNA-seq

Luminex assays

Bulk TCR-seq

scRNA-seq

EV isolation uptake imaging cross-priming assay and visualizing

Patient case

Statistical analysis

Reporting summary

Supplementary information

Source data

Acknowledgements

Author contributions

Peer review

Peer review information

Funding

Data availability

Code availability

Competing interests

Footnotes

Contributor Information

Supplementary information

References

Associated Data

Supplementary Materials

SBRT embedded in low-dose RT plus αPD-1 (immuno-EclipseRT, iERT) elicits CD8⁺ T cell immunity against bulky tumors via an IFN-I/NK/DC axis

iERT elicits early NK cell and subsequent CD8⁺ T cell responses

Fig. 2. iERT induces early NK cell and subsequent CD8⁺ T cell infiltration into bulky mouse tumors.

iERT recruits XCR1⁺ cDC1s through NK-derived XCL1

Fig. 5. iERT enhances XCR1⁺ cDC1 infiltration and tumor control via XCL1.

iERT induces CD8⁺ T cell responses via the IFN-I/NK/XCL1 axis

Fig. 6. iERT enhances CD8⁺ T cell responses through IFN-I/NK/XCL1 axis.

Fig. 7. iERT induces CD8⁺ T cell responses in tumor-draining lymph nodes.