Research ArticleIMMUNOLOGY

Radiation-induced eosinophils improve cytotoxic T lymphocyte recruitment and response to immunotherapy

See allHide authors and affiliations

Science Advances  29 Jan 2021:
Vol. 7, no. 5, eabc7609
DOI: 10.1126/sciadv.abc7609


The efficacy of cancer immunotherapy is dictated by CD8+ T cell infiltration and the nature of the tumor microenvironment (TME). By inflaming the TME to favor CD8+ T cell immunity, radiation is now widely considered as a neoadjuvant for immunomodulation. Here, we observed that local irradiation enhances the infiltration of intratumoral eosinophils, and depletion of eosinophil dampens CD8+ T cell infiltration and diminishes the anti-tumor effectiveness of radiation. Retrospectively, we identified a strong correlation between eosinophilia and survival benefit in radiation-treated cancer patients. Experimentally, we further show that radiation enhances the intratumoral infiltration of adoptive transferred T cells therapy, bolstering eosinophils by intravenous interleukin-5 administration promotes the efficacy of radiation-induced abscopal effect. Together, these results suggest that eosinophil mobilization can be considered as a mechanistically relevant biomarker for predicting the effectiveness of pre-immunotherapy radiation, as well as a new strategy to enhance T cell-mediated immunotherapy against cancers.


Recent clinical advances in immune checkpoint blockade (ICB) therapy and chimeric antigen receptor (CAR)–T cell therapy, used for patients with multiple cancer types and hematologic malignancies, respectively, represent a new era in cancer immunotherapy (1, 2). The antitumor efficacy of ICB therapy, which involves the blocking by antibodies specific for programmed death-1 (PD-1), programmed death-ligand 1 (PD-L1), and cytotoxic T lymphocyte–associated protein 4 (CTLA-4), depends on the existence of antigen-specific CD8+ T cells and their recruitment to the tumor microenvironment (TME) (3). While CAR-T cell immunotherapy has been tremendously successful in treating hematologic malignancies (47), its success against solid tumors has been moderate due to barriers to cytotoxic T cell infiltration into the TME (8). Because intratumoral CD8+ T cell infiltration is essential for the success of both ICB and adoptive cell therapy (ACT), it is extremely important to identify biological mechanisms underlying a favorable TME to optimize the benefits of both immunotherapeutic modalities.

The CXCR3 ligands CXCL9, CXCL10, and CXCL11 are key chemokines enabling the recruitment of CD8+ T cells into the TME (9, 10). A recent study suggests that CD103+ dendritic cells (DCs) represent the main source of CXCL9 and CXCL10 in the TME (10). Moreover, a lack of CD103+ DCs within the TME leads to an absence of CXCL9 and CXCL10, which results in an immune-resistant TME, via the exclusion of CD8+ T cells (11). In addition to CD103+ DCs, activated eosinophils have also been demonstrated to be a source of CXCL9/10 (12, 13), and the adoptive transfer of tumor-specific CD8+ T cells, together with activated eosinophils, leads to enhanced T cell infiltration (13). In addition, the targeted inhibition of either enhancer of zeste homolog 2 (EZH2)–mediated histone H3 lysine 27 trimethylation (H3K27me3) or DNA methyltransferase 1 (DNMT1)–mediated DNA methylation could enhance sensitivity to the PD-1 blockade, owing to the release of CXCL9 and CXCL10 from tumors (14). Together, these studies describe a new avenue for the development of combinational strategies that promote T cell recruitment and infiltration into the TME.

Emerging evidence from multiple solid tumor types support the theory that a combination of radiotherapy (RT) and immunotherapy can deliver a synergistic antitumor effect that is stronger than either monotherapy alone (15, 16). For example, in a retrospective analysis, non–small cell lung cancer (NSCLC) patients who received RT benefited from the subsequent administration of pembrolizumab (17). Similarly, as demonstrated in a prospective clinical trial (PACIFIC), when used after chemoradiotherapy, the anti–PD-L1 antibody durvalumab provides substantial patient benefits (18). Thus, preclinical and clinical data suggest that RT has a potential role in optimizing the TME for ICB therapy (17). It has been proposed and partially demonstrated that the underlying mechanism of this radiation-elicited immune response involves tumor cell death, tumor antigen exposure, and priming of CD8+ T cells in the draining lymph nodes, resulting in the subsequent infiltration of CD8+ T cells into the TME. However, additional mechanisms by which RT creates a TME conducive to CD8+ T cell recruitment remain largely unexplored.

In the current study, we conducted genome-wide expression analysis of irradiated tumors to investigate how radiation improves the TME immune signature. The results revealed that eosinophil-related molecular markers are strongly increased in the TME following radiation exposure. Subsequent experimental studies described herein demonstrate that eosinophil infiltration into the TME (i) increases markedly after irradiation, (ii) correlates with the recruitment of endogenous CD8+ cytotoxic T lymphocytes (CTLs) into the TME, and (iii) is essential for an endogenous CTL-mediated antitumor response. Moreover, a retrospective analysis of three cohorts including human NSCLC and nasopharyngeal carcinoma (NPC) patients established a positive correlation between eosinophil mobilization and favorable response to RT. Of clinical interest, RT also promoted CAR-T cell infiltration and persistence in a patient with B cell non-Hodgkin lymphoma (B-NHL). Together, these findings reveal eosinophil activation as a new mechanism by which radiation promotes CTL recruitment to facilitate immunotherapy against solid tumors.


Radiation promotes eosinophil infiltration into the TME

The systemic effects of radiation on the TME are largely unknown. To assess how radiation remodels the TME, we established a RT model. Specifically, in mice, when subcutaneously inoculated B16-F10 melanoma tumors reached 100 mm3, 20 Gy of focused irradiation was applied to the tumor on the hind leg (fig. S1). Genome-wide RNA-sequencing (RNA-Seq) analyses were then conducted on the tumor tissues 1 week [as described previously (19)] after irradiation. We then investigated the pathways that were typically altered in irradiated lesions [i.e., reactive oxygen species (ROS) formation (20), type I interferon (IFN-I) activation (21), IFNγ (IFN-II) signaling (22), and inflammatory responses (23)]. In these heatmaps, for each pathway, up-regulated genes were substantially enriched in the irradiated group, supporting the rationale of our schema for recapitulating the irradiated immune microenvironment (Fig. 1A). Following previous observations regarding infiltrating granulocytes in the niche receiving irradiation, we subjected these transcriptomic changes to gene set enrichment analysis (GSEA), focused on the signatures of granulocyte chemotaxis, migration, and activation (Fig. 1B and table S1). Cytometric analysis of tumor tissues was performed 7 days after irradiation, and confirmed an overall enrichment of CD45+ cells (24, 25) (fig. S2), including CD8+ T cells and various myeloid cells (Fig. 1C), in irradiated tumors. In particular, we noticed an enriched panel of genes signifying a type 2 inflammatory response; these included genes specifying the lineage identity (Ear2, P2ry13, P2ry14, Gata1, and Siglecf) and genes involved in eosinophil regulation, differentiation (Camk1d), activation (Cd69 and Ccl5), survival (Il4, Il5, Il5ra, Il13, and Il33), chemoattraction (Ccl11, Ccl24, and Chil3), and Ccr3 expression (Fig. 1D). The expression of CD103, which was reported to be expressed on CXCL9/10-producing DCs, exhibited modest up-regulation (Fig. 1D). The increased expression of eosinophil signature genes upon tumor irradiation was validated by cytometric analysis, which demonstrated enhanced eosinophil intratumoral infiltration (Fig. 1C).

Fig. 1 Radiation promotes eosinophil infiltration into the TME.

C57BL/6 mice were subcutaneously injected with 1 × 106 B16-F10 tumor cells. Tumors were locally irradiated at a dose of 20 Gy when they reached approximately 100 mm3 and then resected for analysis 7 days after irradiation. (A) RNA-Seq analysis of irradiated tumors versus nonirradiated tumors across four gene sets that respond to radiation (n = 4 per group). (B) GSEA of gene sets associated with chemotaxis [normalized enrichment score (NES) = 1.59], migration (NES = 1.75), and activation (NES = 1.80) of granulocytes. (C) Flow cytometry analysis of the indicated tumor-infiltrating lymphocyte subsets after irradiation. The upper panel shows percentages, and the lower panel shows subset counts. NK, natural killer. (D) Heatmap depicting specific gene signatures of the type 2 inflammatory response. (E) Schematic illustrating a bilateral B16-F10 tumor–bearing mouse model receiving an intravenous injection of 4 × 105 CFSE-stained eosinophils on the day of irradiation. (F) Representative data of intratumoral CFSE+ cells in irradiated and nonirradiated side. (G) Flow cytometry quantification of the percentage and number of tumor-infiltrating eosinophils (n = 5 per group, paired Student’s t test). (H) Quantitative RT-PCR analysis of genes known to be involved in eosinophil chemoattraction. (I) Migration rates of transferred eosinophils that infiltrated into the nonirradiated or irradiated tumors. (J) Quantitative RT-PCR analysis of genes known to support eosinophil survival and proliferation. Data are obtained from one experiment that was representative of three independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001 (unpaired Student’s t test). n.s., not significant.

We hypothesized that the higher number of eosinophils in irradiated tumors was attributable to increased eosinophil migration. To test this, we subcutaneously inoculated B16-F10 tumor cells on both hind legs of C57BL/6 mice; after allowing tumor formation to occur, we irradiated one tumor with a single dose of 20 Gy. Upon irradiation, we also adoptively transferred carboxyfluorescein diacetate succinimidyl ester (CFSE)–labeled eosinophils into tumor-bearing mice (Fig. 1E). Two days after radiation, mice were sacrificed, and eosinophils were quantified in tumors obtained from both sides. In comparison to the control tumor, the proportion of infiltrating eosinophils was significantly elevated in tumors on the irradiated side (Fig. 1F). On average, CFSE-labeled eosinophils constituted about 41% of the total eosinophils in irradiated tumors (the mean percentage of CFSE+ eosinophils in irradiated tumors; Fig. 1G, left). Similarly, the average number of CFSE+ eosinophils that infiltrated in irradiated tumors was significantly higher than in nonirradiated tumors (the mean count of CFSE+ eosinophils in irradiated tumors versus that in nonirradiated tumors was 76:1; Fig. 1G, right). We hypothesized that attractants from the irradiated tumor might guide circulating eosinophil migration. Quantitative polymerase chain reaction (qPCR) analysis revealed the radiation-induced intratumoral up-regulation of Ccl11 (Eotaxin1) and Ccr3, the chemokine and chemokine receptor pair known to direct eosinophil migration (Fig. 1H). Moreover, these elevations were validated at the protein level using enzyme-linked immunosorbent assay and flow cytometry for CCL11 and CCR3, respectively (fig. S3). Correspondingly, we observed an enhanced mobility of eosinophils in irradiated tumors using in vitro migration assays (Fig. 1I and fig. S4). In addition, irradiation also stimulated the transcription (Il5, ***P = 0.0006; Gmcsf/Csf2, *P = 0.0321; Fig. 1J) and translation [interleukin-5 (IL-5), *P = 0.0113; granulocyte-macrophage colony-stimulating factor (GM-CSF)/CSF2, **P = 0.0044; fig. S5] of cytokines promoting eosinophil survival and proliferation. After analyzing the cell composition of the isolated tissues and assessing the expression of eosinophil attractants (Hmgb1 and Ccl11) in each cell type, we observed that tumor cells were the dominant producers of eosinophil attractants (fig. S6). These compounds were produced in a dose-dependent manner with respect to radiation (fig. S7), supporting the fact that high-dose RT was a superior inducer of eosinophilia. Together, these data suggest that irradiation creates a tumor niche favorable for eosinophil infiltration and expansion.

Eosinophil infiltration is accompanied by enhanced CD8+ T cell infiltration and cytolytic activity

Activated CD8+ CTLs are critical for the antitumor response, as the depletion of CD8+ T cells significantly attenuates the therapeutic efficacy of RT (fig. S8) (19). In our chamber migration assays, we found that migration substantially increased when CD8+ T cells were incubated with more eosinophils, which suggests a chemokine-dependent mechanism for recruiting CD8+ T cells (fig. S9). Next, we evaluated the eosinophil-CTL correlation at 5 days after radiation treatment in vivo (the peak of eosinophil infiltration based on our pilot experiment; fig. S10). Using the two-sided melanoma model, we examined key chemokines known to recruit effector CD8+ CTLs and observed that local irradiation significantly increases the expression of Ccl5, Cxcl9, and Cxcl10 (Fig. 2A). Moreover, when bone marrow–derived eosinophils (BMDEs) were incubated with the supernatant from irradiated or nonirradiated tumors, we confirmed the elevated expression of Ccl5, Cxcl9, and Cxcl10 in the irradiated condition (fig. S11). Consistent with these results, the intratumoral infiltration of CD8+ T cells was significantly elevated in irradiated tumors (fig. S12 for immunofluorescent staining evidence). In addition, irradiation also enhanced the local cytotoxicity of these T cells, as indicated by the expression of Ifng, Gzma, Gzmb, and Perforin1 (Prf1), as per examinations of the entire tumor tissue (Fig. 2B) and isolated CD8+ T cells (fig. S13). This phenotype is not melanoma specific; in both 4T1 breast and MC38 colon cancer models, irradiation initiated enhanced eosinophil and CD8+ T cell infiltration (Fig. 2, C and D, and fig. S14). The eosinophil niche and CTL response genes were also activated upon irradiation (Fig. 2, E and F). These results suggest that eosinophils are an integral cellular component of an immunologically “hot” tumor.

Fig. 2 Eosinophil infiltration is accompanied by enhanced CD8+ T cell infiltration and cytolytic activity.

Tumors were resected from B16-F10 tumor–bearing mice 5 days after irradiation. (A) Quantitative RT-PCR analysis of mRNA transcripts related to T cell chemoattraction. (B) Quantitative RT-PCR measurements of Ifng, Gzma, Gzmb, and Prf1 expression in irradiated and nonirradiated tumor tissues. Data are from one experiment representative of three independent experiments (n = 5 mice in the control group and n = 4 in the radiation group). (C and D) Flow cytometry analysis of CD45+ cells, eosinophils, and CD8+ T cells in irradiated and nonirradiated tumors from 4T1 (C) or MC38 (D) tumor–bearing mice 5 days after irradiation. (E and F) Quantitative RT-PCR analysis of cytokine and chemokine expression in 4T1 (E) or MC38 (F) tumors from locally irradiated mice. 4T1 tumor–bearing mice: n = 7 in the control group and n = 5 in the radiation group; MC38 tumor–bearing mice: n = 6 in the control group and n = 7 in the radiation group. Results are mean ± SEM values. *P < 0.05, **P < 0.01, and ***P < 0.001 (unpaired Student’s t test).

Eosinophilia predicts longer progression-free survival in RT-treated patients

To test the clinical relevance of these observations, we retrospectively evaluated the correlation between eosinophilia and RT efficacy in three independent cohorts. Because it is not feasible to obtain tumor biopsies after RT due to ethical restrictions, we first examined whether eosinophil numbers in the peripheral blood (PB) were reflective of intratumoral infiltration. We compared 35 pairs of samples collected from surgically resected tumor tissues and PB from patients with NSCLC and found that intratumoral eosinophil abundance, as determined using immunohistochemistry, correlates with peripheral eosinophil numbers measured by determining the complete blood count (CBC) (fig. S15).

We retrospectively studied the records of consenting patients, collected over the past 7 years, by the Department of Oncology and Radiotherapy of Xinqiao Hospital. We focused on NSCLC and NPC patients treated with localized RT and analyzed all available case records containing the following information: (i) CBC, (ii) complete radiological scans of the involved organs, and (iii) prognosis records from the care-providing doctor and the radiologist’s image assessment. Overall, cohort 1 (234 NSCLC patients), cohort 2 (123 NSCLC patients), and cohort 3 (249 NPC patients) were included in our analysis (baseline patient information is summarized in table S2). The median intervals from the baseline to peak eosinophil counts (EOS) were 15, 37, and 28 days for cohorts 1, 2, and 3, respectively (fig. S16). For cohort 1, eosinophil counts in PB showed an 87.5% increase after irradiation in 175 of 234 (75%) patients (median: EOSbefore = 0.08 × 109/liter and EOSafter = 0.15 × 109/liter, ***P < 0.0001; Fig. 3A). Patients with higher eosinophil counts after radiation had a longer progression-free survival [PFS; median cutoff (1.67-fold): *P = 0.0291, hazards ratio (HR) = 0.73 (0.54 to 0.97), median PFS = 227 days versus 182 days; Fig. 3B; twofold cutoff: *P = 0.0294, HR = 0.73 (0.54 to 0.98), median PFS = 229 days versus 182 days; Fig. 3C]. For cohort 2, another cohort of patients with NSCLC, a similar eosinophil dynamic (median: EOSbefore = 0.08 × 109/liter and EOSafter = 0.16 × 109/liter, ***P < 0.0001; Fig. 3D) and survival benefit [median cutoff (2.17-fold): *P = 0.0138, HR = 0.53 (0.32 to 0.87), median PFS = 591 days versus 303 days; Fig. 3E; twofold cutoff: *P = 0.0434, HR = 0.60 (0.36 to 0.99), median PFS = 591 days versus 315 days; Fig. 3F] was observed as in cohort 2, confirming the predictive significance of eosinophil elevation in patients with NSCLC. In cohort 3 (patients with NPC), mobilization of eosinophils into PB was observed in 206 of 249 (83%) patients. A twofold increase in eosinophil counts was observed after RT (median: EOSbefore = 0.07 × 109/liter and EOSafter = 0.14 × 109/liter, ***P < 0.0001; Fig. 3G). Moreover, a more notable PFS survival benefit was observed in patients exhibiting a pronounced increase in eosinophil counts after RT [median cutoff (2.20-fold): **P = 0.0053, HR = 0.57 (0.39 to 0.84), median PFS = 1894 days versus 917 days; Fig. 3H; twofold cutoff: **P = 0.0046, HR = 0.57 (0.38 to 0.84), median PFS = 1894 days versus 917 days; Fig. 3I].

Fig. 3 Eosinophilia predicts longer PFS in radiotherapy-treated patients.

Retrospective analysis of three cohorts. (A) PB eosinophil counts were elevated after RT in NSCLC patients of cohort 1 (175 of n = 234). (B) PFS in the median fold (1.67-fold) and (C) twofold arms of cohort 1. CI, confidence interval. (D) PB eosinophil counts were elevated after RT in NSCLC patients of cohort 2 (95 of n = 123). (E) PFS in the median fold (2.17-fold) and (F) twofold arms of cohort 2. (G) Change of eosinophil counts in the PB of NPC patients of cohort 3 (206 of n = 249). (H) PFS in the median fold (2.20-fold) and (I) twofold arms of NPC patients. *P < 0.05; **P < 0.01; ***P < 0.001 [Mann-Whitney U test for (A), (D), and (G), and log-rank Mantel-Cox test for survival analysis].

To examine the predictive significance of other cell types in CBC reports, cell counts of lymphocytes, red blood cells (measured based on hemoglobin, g/dl), monocytes, neutrophils, basophils, and platelets were subjected to the above analyses for eosinophils, if applicable (figs. S17 to S19). Survival analysis was only performed for cell types with significant changes. Collectively, none of the investigated cell types were clearly predictive with respect to the survival of patients following RT (figs. S17 to S19). These findings indicate that the PB eosinophil count might be used as a biomarker to predict RT outcomes, at least for patients with NSCLC and NPC.

Eosinophil trafficking into tumors is critical for radiation-induced antitumor immunity

Systemic analysis of the correlation between activated CD8+ T cells and other immune cells indicated that this correlation was not eosinophil specific (fig. S20). We attempted to determine whether eosinophils played a causative role in determining radiotherapeutic efficacy, which is dependent on the recruitment of CTLs (26). Hence, we developed an eosinophil depletion regimen in locally irradiated B16-F10 tumor–bearing mice by administering an anti–Siglec-F antibody for four consecutive days (Fig. 4A). This method was successful at depleting eosinophils (figs. S21 and S22), with no evidence of anti–Siglec-F antibody–associated phenotype change in CD8+ T cells (fig. S23). We observed that the therapeutic effect of radiation in the eosinophil-depleted group was significantly reduced (Fig. 4B), in a manner similar to CD8+ T cell depletion (fig. S8). Eosinophil depletion was persistent, with levels remaining the same 10 days after the injection (fig. S24). Cytometric analysis revealed that antibody-mediated eosinophil depletion caused a drastic reduction in intratumoral CD8+ T cell levels (percentage: ***P < 0.0001; counts: ***P = 0.0001), especially effector CTLs (percentage: ***P < 0.0001; counts: ***P = 0.0002; Fig. 4, C to E). In addition, a clinically validated gene panel signature that reflected inflammatory IFNγ responses and predicted the prognosis of the ICB efficacy (27) was diminished as a consequence of eosinophil depletion (Fig. 4F). Immunological features reduced by eosinophil depletion included T cell chemotaxis (Cxcl9 and Cxcl10), T cell activation (Cd2, Cd3d, Cd3e, Nkg7, Il2rg, Stat1, and Cxcr6), CTL effector functions (Ifng, Gzmk, and Lag3), and antigen presentation (H2eaps, Ciita, and Tagap). When an irradiation-free TME was examined after eosinophil depletion, the significant decreases were only seen in eosinophils and CD8+ T cells (fig. S25). Together, these results suggest that irradiation-induced intratumoral eosinophil recruitment is necessary for TME reprogramming to favor CTL-mediated antitumor immunity.

Fig. 4 Eosinophil trafficking into tumors is critical for radiation-induced antitumor immunity.

(A) Wild-type C57BL/6 mice were subcutaneously injected with 1 × 106 B16-F10 tumor cells in the right flank on day 0. On day 10, mice received local tumor irradiation at a dose of 20 Gy, and from this day onward, 15 μg of the anti–Siglec-F antibody (Ab) or isotype antibody was intraperitoneally injected daily for four consecutive days. On day 14, tumors were resected for analysis. (B) Monitoring and measurement of tumor sizes were performed after irradiation alone or irradiation combined with the anti–Siglec-F or isotype antibody, as compared with the control group (n = 5 to 7 per group). *P < 0.05, **P < 0.01, and ***P < 0.001 [two-way analysis of variance (ANOVA) with multiple comparisons]. (C and D) Flow cytometry quantification of percentages and counts of CD8+ and effector T cells in irradiated tumors with the anti–Siglec-F antibody or isotype antibody. (E) Correlation analyses of the percentage of tumor-infiltrating eosinophils and CD8+ or effector T cells, as indicated using the Pearson correlation test. (F) Quantitative RT-PCR analysis of genes relevant to the immune response in tumors treated with radiation, combined with the eosinophil neutralizing antibody or isotype antibody. Data presented here are from one representative experiment of three independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001 (unpaired Student’s t test).

Radiation promotes T cell infiltration in adoptive transfer therapy

Intratumoral infiltration and the accumulation of transferred T cells are vital for the efficacy of ACT (28, 29). While CAR-T cell therapy, the most successful form of ACT, has been approved as a “cell drug” to treat B cell lineage lymphoma and leukemia (30), progress has been slow for solid tumor treatments (31, 32). One major obstacle to the success of CAR-T cell therapies for solid tumors is the poor infiltration of transferred CAR-T cells (33). To address this obstacle, we examined whether local irradiation could mobilize eosinophils to facilitate adoptive T cell infiltration.

We applied a proof-of-concept test to two types of ACT therapies: tumor-specific effector mouse CD8+ T cells in an immunocompetent model and antigen-specific redirected human CAR-T cells in a xenograft model. With regard to the former, we grafted B16-F10 melanoma bilaterally on CD90.2+ C57BL/6 mice and applied irradiation to one established tumor. Subsequently, congenially marked (CD90.1+) gp100-specific Pmel transgenic CD8+ T cells were activated (fig. S26) and transferred through the tail vein on day 5 after irradiation. One day later, the intratumoral distribution of Pmel cells was assessed to minimize the influence of Pmel proliferation (Fig. 5A). The increased infiltration of the irradiated tumor was observed for newly transferred CD90.1+ Pmel T cells, indicating a guided migration (Fig. 5, B and C).

Fig. 5 Radiation promotes T cell infiltration in adoptive transfer therapy.

(A) Experimental scheme. Activated Pmel CD8+ T cells (1 × 106) were intravenously injected into B16-F10 tumor–bearing C57BL/6 mice 5 days after irradiation (n = 7 per group). TBI, total body irradiation. (B) Flow cytometry quantification of intratumoral CD8+ T cells and Pmel T cells. (C) Correlation analyses (Pearson correlation test). (D) Experimental scheme. Raji tumor–bearing NPG mice were intravenously injected with 1 × 107 CAR-T cells (n = 5 per group). (E and F) Flow cytometry quantification of tumor-infiltrating eosinophils (E) and transferred CAR-T cells (F) as indicated. (G) Correlation analyses. RT group, blue circles; non-RT group, gray circles. (H) Experimental scheme. Activated Pmel CD8+ T cells (1 × 107) were intravenously injected into B16-F10 tumor–bearing mice on day 10. (I) Tumor sizes of mice after different treatments for all the experimental groups: irradiation (n = 11), Pmel CD8+ T cell transfer (n = 10), irradiation along with Pmel CD8+ T cell transfer and isotype antibody (n = 6), irradiation along with Pmel CD8+ T cell transfer and anti–Siglec-F antibody (n = 11), or without treatment (n = 10). Two-way ANOVA with multiple comparisons. (J) Survival of mice (log-rank Mantel-Cox test). (K) Bilateral B16-F10 tumor–bearing mice were treated with 15 μg of anti–Siglec-F antibody (group 2), 100 ng of rmIL-5 (group 3), or vehicle (group 1); n = 9 for each group. (L) Tumor sizes for the irradiated and (M) nonirradiated side. Data are from one representative experiment of two independent experiments (two-way ANOVA with multiple comparisons). *P < 0.05, **P < 0.01, and ***P < 0.001.

For the second ACT model, we investigated whether radiation also facilitated human CAR-T cell infiltration. For this, we subcutaneously inoculated NPG (NOD-Prkdcscid-Il2rgnull) mice with Raji lymphoma cells bilaterally and irradiated one of the tumors. Anti-CD19 CAR-T cells prepared from PB mononuclear cells (PBMCs) (34) of healthy donors were intravenously injected; five days later, CAR-T cell and eosinophil distribution was analyzed (Fig. 5D). In comparison to the nonirradiated side, radiation promoted the intratumoral infiltration of both eosinophils and CD3+Fab+ CAR-T cells (percentage: *P = 0.0476 and R2 = 0.4058; counts: *P = 0.0124 and R2 = 0.5632; Fig. 5, E to G). These results suggest that local radiation might facilitate the infiltration of adoptively transferred T cells into tumors via eosinophil accumulation.

Next, we attempted to determine whether eosinophilia could translate into therapeutic efficacy and whether the irradiation-induced eosinophilia could facilitate subsequent transferred T cell immunotherapy. By transferring ex vivo activated eosinophils, we observed improved tumor control for tumor-bearing mice receiving irradiation-CTL transfer therapy, suggesting that eosinophilia is sufficient for improving outcomes from radiation–T cell transfer therapy (fig. S27). To demonstrate the necessary role of eosinophils in this paradigm, and to test it in an antigen-specific manner, unilateral tumor-bearing mice were irradiated and Pmel cells were transferred into them (Fig. 5H). In comparison to the eosinophil-competent group, eosinophil-depleted mice showed significantly impaired tumor control (Fig. 5I) and poor survival outcomes (Fig. 5J) from Pmel adoptive transfer therapy.

We next aimed to investigate whether the RT-induced eosinophilia contributes to the systemic antitumor immune response. To this end, a mouse model with bilateral tumor implantation was used. Anti–Siglec-F antibody and recombinant mouse IL-5 (rmIL-5) (35) were used to diminish or elicit the eosinophils (Fig. 5K). Tumor growth after eosinophil manipulation was recorded for both the irradiated and nonirradiated tumors (fig. S28A). In these analyses, an expected universal reduction in tumor volume was observed for the irradiated tumors, regardless of the manipulation to eosinophils (fig. S28B). Furthermore, in addition to the eosinophil transfer, we observed improved control of irradiated tumors in mice treated with rmIL-5, indicating a potential clinic-applicable maneuver to improve the therapeutic efficacy of RT and immunotherapy (Fig. 5L). When focusing on the nonirradiated tumor, we identified a mirrored pattern of tumor growth as in the irradiated side, demonstrating that eosinophilia was also involved in eliciting the systemic immune response against cancers (Fig. 5M).

RT enhances anti-CD19 CAR-T cell infiltration in B-NHL

In our closed anti-CD19 CAR-T cell therapy against B-NHL ( identifier: NCT02652910), because of the rapid progression of the right submandibular lymph nodes during the manufacture of CAR-T (Fig. 6A, upper panel), one patient received palliative RT before CAR-T infusion, according to the researcher’s instruction to prevent possible apnea (Fig. 6A, lower panel). This sporadic case enabled us to preliminarily investigate the clinical relevance of radiation-induced eosinophil and T cell infiltration in clinical practice.

Fig. 6 Radiotherapy enhances anti-CD19 CAR-T cell infiltration in B-NHL.

(A) Upper panels: Photograph of a large lesion on the patient’s neck and CT scan of bilateral lymphoma. Lower panel: Treatment regimen schematic: the patient’s right lesion received five fractionated doses of 3-Gy radiation 7 days before CAR-T cell infusion. (B) Positron emission tomography–CT scans of the patient before and 3 months after treatment. (C) CT images obtained during early treatment and 3 months and 1 year after treatment. (D) Bilateral lesion sizes were monitored regularly and compared to their original sizes. Black and blue lines represent changes in the peak particle diameters of the left nonirradiated and right irradiated tumors, respectively. PPD, product of perpendicular diameter. (E) The number of eosinophils in the patient’s PB was elevated 17 days after RT. (F) H&E staining of human lymphoma tissues in the left nonirradiated and right irradiated lesions before and 2 days after CAR-T cell infusion. Typical eosinophils are circled in yellow; yellow boxes indicate enlarged eosinophils. The black scale bar represents 50 μm. (G) CAR-T cell copy numbers in lymph node samples before and 2 days after CAR-T cell infusion. gDNA, genomic DNA. Photo credit: Qingzhu Jia (Chongqing Key Laboratory of Immunotherapy; Department of Oncology, Xinqiao Hospital, Third Military Medical University).

This patient, who received a diagnosis of stage IV lymphoma, was unresponsive to previous chemotherapies and was enrolled on 28 March 2016. After five fractions of 3-Gy irradiation, the right-side lesion was controlled, and anti-CD19 CAR-T therapy was performed 3 days later. This treatment was effective, because the patient reached complete response 3 months after CAR-T cell transfer (Fig. 6B). One year later, the nonirradiated tumor on the left relapsed, while the irradiated tumor on the right remained undetectable via computed tomography (CT) scans (Fig. 6, C and D). Her CBC record showed that the number of eosinophils gradually increased from day 5 after RT (Fig. 6E). We analyzed biopsies sampled from her left (nonirradiated) and right (irradiated) lesions; hematoxylin and eosin (H&E) staining enabled the visualization of the right irradiated lesion with more infiltrated eosinophils (Fig. 6F). The tumor-bearing mouse model also demonstrated the effectiveness of the 3 Gy for 5 fractions in inducing intratumoral eosinophilia (fig. S29). Although the right lesion was 4.3-fold larger than the left lesion, which normally presents an obstacle for T cell infiltration, anti-CD19 CAR-T cells rapidly penetrated the irradiated lesion 2 days after CAR-T transfer (Fig. 6G). This clinical evidence suggests that for solid tumors, localized RT might be a feasible immunomodulatory approach for enhancing the efficacy of CAR-T cell immunotherapy.


In this study, we profiled an inflamed TME reprogrammed by radiation, within which infiltrated eosinophils were indispensable for the antitumor immune response. Antibody-mediated eosinophil depletion greatly reduced the recruitment of CD8+ T cells to irradiated tumor tissues and ultimately weakened the radiation-mediated control of tumors. Thus, as supported by clinical cases, the optimal recruitment of activated CD8+ T cells was dependent on eosinophils. This conclusion is highly relevant to current clinical treatments: Eosinophil abundance represents a favorable prognostic marker for RT that can be readily applied to enhance the immunotherapeutic efficacy and expand CAR-T cell therapy to the treatment of solid tumors.

Tumors evolve to evade the immune response and reprogram the surrounding environment to promote tumor progression. This cooperative state is maintained via suppressive cytokine signaling [IL-1β, IL-6, IL-10, tumor necrosis factor (TNF), and transforming growth factor β (TGFβ)] and suppressive immune cell populations [myeloid-derived suppressor cells (MDSCs), regulatory T cells (Tregs), and tumor associated macrophages (TAMs)] (36, 37). Radiation is thought to impair this immunosuppressive cross-talk by activating cell survival pathways and stimulating the immune system. The intracellular actions of ROS affect TNF signaling pathways and trigger inflammatory cytokine signaling [IL-2–STAT5 (signal transducer and activator of transcription 5) signaling and IFNγ signaling]; then, increased antigen exposure and presentation elicit an immune response and induce cascade effects, including the recruitment of numerous types of circulating innate or adaptive immune cells (38, 39). Enhanced T cell infiltration into the TME after irradiation has been reported in several mouse models (40). We also observed elevated numbers of tumor-infiltrating CD8+ CTLs in irradiated B16-F10 tumors; the depletion of CD8+ T cells significantly weakened the therapeutic effect of radiation. Lee et al. (19) also demonstrated the fact that T cells are required for radiation-elicited tumor shrinkage. In comparison to immunocompetent mice, nude mice responded poorly to local tumor radiation. Even in mice with robust immune systems, radiation would be less effective when CD8+ T cells were cleared via antibody depletion (19). While it has been reported that T cell recruitment depended on IFNγ–ICAM-1 (intercellular adhesion molecule–1) feedback (41, 42), autophagy of tumor cells (43), or CXCL9/10-producing M1 polarized macrophages (44, 45), the precise molecular mechanisms that regulate radiation-induced T cell recruitment remain obscure.

Here, using GSEA data, we found that among granulocyte populations, eosinophils are activated by irradiation, while neutrophil percentages do not change, and basophils are rarely detected. In response to eotaxins (CCL11, CCL24, and CCL26), eosinophils leave the bloodstream and migrate into inflamed tissues. It has been reported that DAMPs (danger-associated molecular patterns), including the HMGB1 protein, also trigger eosinophil recruitment (46, 47). We found that locally irradiated tumor tissue attracted eosinophils into the tumor niche. The maintenance of eosinophils in the TME depends on IL-5 and GM-CSF; IL-5 is the most important eosinophil growth, differentiation, and activating factor, and GM-CSF is necessary to support eosinophil maturation (48, 49). Consistent with these previous findings, both IL-5 and GM-CSF were specifically up-regulated in two independent irradiated tumor models. Our study demonstrates that irradiated TME is conducive to eosinophil survival and proliferation. Future studies need to confirm the contribution of eosinophil attractants by neutralization, and would be necessary to elucidate the mechanisms by which radiation induces the expression of molecules mediating eosinophil recruitment, survival, and maturation.

Previous studies have reported that eosinophils have an antitumor role in both preclinical mouse models and human cancers (50, 51). Eosinophils act as effector cells in tumor tissues by releasing cytotoxic molecules such as ROS, TNF-α, IL-18, IL-33, IFNγ, granzyme, and eosinophil cationic proteins (52, 53). In a screening analysis, the establishment of eosinophil-CD8+ T cell correlation was accompanied by enhanced infiltration of other immune cell types (fig. S20), which suggests that the correlation of eosinophils does not imply its causality. Our eosinophil-depleting experiment provides direct evidence for the fact that eosinophils enhance the immune system by recruiting activated CD8+ T cells to irradiated tumors. An independent study, which demonstrated that eosinophils have the capacity to secrete cytokines responsible for the recruitment of CD8+ T cells, provides further support for this mechanism (13). Noteworthily, despite the peripheral eosinophil counts being significantly elevated after RT, we failed to recapitulate this phenomenon in our mouse model, up to 7 days after radiation exposure. However, the eosinophil counts between the baseline and 7 days for patients with NSCLC/NPC were also nondiscriminatory, suggesting that a wider timeframe was necessary to achieve eosinophilia in the peripheral compartment. Thus, both preclinical and clinical data provide strong evidence for the antitumor activity of eosinophils.

With regard to clinical practice, three aspects of our study merit further emphasis. First, eosinophil counts might serve as a positive prognostic marker for cancer patients who had received RT. In our study, higher peripheral eosinophil counts in NSCLC and NPC patients who received RT were directly correlated with a longer PFS. Second, as shown for a CD19–CAR-T cell–infused lymphoma patient, adjuvant RT can enhance the efficacy of ACT. Last, in addition to RT, eosinophilia is often observed after immunotherapy with IL-2, IL-4, GM-CSF, and anti–CTLA-4 therapy (54, 55); thus, high levels of circulating eosinophils are associated with improved overall survival in melanoma patients treated with ipilimumab and pembrolizumab (5658). Thus, the clinical benefit of combinatory RT, anti–CTLA-4 therapy, and anti–PD-L1 therapy might stem from the capacity of eosinophils to recruit CD8+ T cells to the TME. Together, our findings provide strong evidence for the fact that the role of eosinophils in combinatory radio-immunotherapy should be further explored to deliver improved treatment strategies for patients with cancer.


Experimental design

This study was designed to optimize clinical immuno-radiotherapy by comprehensively exploring the effects of radiation on the TME. To this end, we established tumor-bearing mouse models and locally irradiated tumors with a single dose of 20 Gy. RNA-Seq analysis and flow cytometry detection were used to compare irradiated and nonirradiated tumor tissues. In addition to wild-type C57BL/6 and BALB/c mice, Pmel transgenic mice were also used to study the antigen-specific immune response. A panel of cytokines and chemokines that are crucial for the survival, migration, activation, and functioning of eosinophils or CD8+ T cells was analyzed by quantitative real-time PCR (RT-PCR). Eosinophil migration was assessed using a transwell system. For functional assays, we used an anti-CD8 neutralizing antibody or anti–Siglec-F neutralizing antibody to specifically deplete CD8+ T cells or eosinophils, respectively. To prove that RT-induced T cell infiltration is mediated through eosinophils, an eosinophil depletion experiment was performed in an RT-adoptive transfer combinational therapy model. Immunofluorescence staining was performed to directly depict CD8+ T cell and eosinophil infiltration into irradiated tumor tissues. Human specimens were obtained from surgically resected tissue samples from NSCLC or eosinophilic lymphoid granuloma (ELG) patients. Eosinophils in human tumors were observed by immunohistochemical staining. Retrospective analysis of the clinical records of three cohorts of NSCLC and NPC patients was conducted to determine the relationship between eosinophils and PFS in patients who received RT. The study protocol was approved by the Ethics Committee of the Third Military Medical University (TMMU; Chongqing, China).


Wild-type C57BL/6, BALB/c (male, 6 to 8 weeks old), and Pmel transgenic mice (C57BL/6 background) carrying a transgenic T cell receptor (TCR) specific for the B16-F10 melanoma antigen gp100 were purchased from the Center of Experimental Animals of TMMU. NPG mice were generated by VitalStar Biotechnology Co. Ltd. (Beijing, China). All mice were housed in standard cages under specific pathogen–free conditions at a temperature of 22° to 26°C and humidity of 50 to 60%. They were maintained under a 12-hour dark/light cycle and supplied with sterile food and water ad libitum. Protocols were approved by the Institutional Animal Care and Use Committee of TMMU.

Cell lines

The melanoma cell line B16-F10 and breast cancer cell line 4T1 were obtained from the American Type Culture Collection. The colon adenocarcinoma cell line MC38 was a gift provided by L. Deng (Shanghai Jiao Tong University School of Medicine, China). The B16-F10 melanoma, 4T1 breast cancer, and MC38 colorectal tumor cell lines were cultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), penicillin (100 IU/ml), streptomycin (100 μg/ml), 2 mM l-glutamine, and 1 mM sodium pyruvate in a humidified incubator at 37°C and 5% CO2. The Burkitt’s lymphoma cell line Raji, which was provided by HRAIN Biotechnology Co. Ltd. (R&D Department, Shanghai, China), was cultured in RPMI 1640 medium supplemented with 10% FBS, penicillin (100 IU/ml), streptomycin (100 μg/ml), 2 mM l-glutamine, and 1 mM sodium pyruvate.

Syngeneic tumor establishment and local irradiation

C57BL/6 mice were subcutaneously injected with 1 × 106 B16-F10 tumor cells or 5 × 105 MC38 tumor cells; BALB/c mice were subcutaneously injected with 4 × 105 4T1 tumor cells; and NPG mice were subcutaneously injected with 1 × 106 Raji cells. Tumor cells were suspended in 100 μl of phosphate-buffered saline (PBS) buffer and injected into the flank. B16-F10, MC38, 4T1, and Raji tumors were allowed to grow in mice for approximately 10, 7, 7, and 25 days, respectively. Tumor sizes were calculated as follows: volume = 0.5 × length × width2. When tumor sizes became approximately 100 mm3, mice were anesthetized with pentobarbital (80 mg/kg) and locally irradiated using the Varian Trilogy Stereotactic System with a single dose of 20 Gy.

Human samples

Human tumor samples were obtained from surgical specimens of NSCLC patients. The lymphoma patient received 15 Gy of fractionated RT in five fractions before CAR-T cell infusion. All subjects provided informed consent, and the study protocol was approved by the Ethics Committee of the TMMU (identification no. 2016003; Chongqing, China).

Flow cytometry

To obtain single-cell suspensions, tumor tissue samples were digested with collagenase IV (1 mg/ml; Sigma-Aldrich) and deoxyribonuclease (DNase) I (0.2 mg/ml; Sigma-Aldrich). The following fluorochrome-conjugated antibodies were used: anti-CCR3 (J073E5), anti-CD45 (30-F11), anti-CD11b (M1/70), anti-CD11c (N418), anti-CD103 (2E7), anti-CD64 (X54-5/7.1), anti-CD3 (17A2), anti-CD8 (53-6.7), anti-CD4 (GK1.5), anti-CD44 (IM7), anti-CD62L (MEL-14), anti-CD69 (H1.2F3), anti-CD90.1 (OX-7), anti-CD90.2 (30-H12), anti-F4/80 (BM8), anti–Gr-1 (RB6-8C5), anti-GZMB (GB11), anti–Ly-6C (HK1.4), anti-MHCII (M5/114.15.2), anti-NK1.1 (PK136), and anti-TCR Vβ13 (MIR12-4) (all from BioLegend); anti–Siglec-F (E50-2440) (BD Biosciences); and anti-IFNγ (XMG1.2), anti-perforin (eBioOMAK-D), and Fixable Viability Dye eFluor 780 (65-0865) (from eBioscience). Data were collected using a Gallios flow cytometer (Beckman Coulter) and analyzed using FlowJo software.

Depletion of CD8+ T cells and eosinophils

To achieve CD8+ T cell depletion, 250 μg of anti-CD8 antibody (clone 2.43; Bio-X-Cell) was intraperitoneally injected into each mouse every 3 days for a total of four injections. InVivoPlus rat isotype antibody (BP0090; Bio-X-Cell) was also injected into mice and used as a control. To achieve eosinophil depletion, 15 μg of anti–Siglec-F antibody (238047; R&D Systems) was intraperitoneally injected into each mouse for four consecutive days. Rat isotype antibody (MAB006; R&D Systems) was also injected into mice and used as a control. To validate the depletion strategy, 100 μg of anti-CCR3 antibody (6S2-19-4; Bio-X-cell) or isotype antibody (BE0090; Bio-X-cell) was injected intraperitoneally.

Quantitative RT-PCR

Total RNA was extracted from tumor tissues using the TRIzol reagent (Invitrogen), followed by complementary DNA (cDNA) synthesis using a PrimeScript RT reagent kit (Takara) according to the manufacturer’s protocol. Gene expression was assessed using a CFX384 Real-Time PCR detection system with Premix Ex Taq II (Takara) and gene-specific primers (table S3). Cycling conditions included a single denaturing step at 95°C for 30 s, followed by 40 cycles at 95°C for 5 s and 60°C for 34 s. The specificity of qPCR analysis was confirmed with dissociation curves.

CD8+ T cell purification and adoptive transfer

To generate antigen-specific CD8+ T cells for adoptive transfer, Pmel mouse spleens were collected and dissociated into single-cell suspensions. Red blood cells were removed using a red cell lysis solution (BL503A; Biosharp). Single-cell suspensions were pulsed with 1 μM hgp10025–23 (KVPRNQDWL) of immunograde quality, which was purchased from SBS (Beijing SBS Genetech Co. Ltd., Beijing, China) and cultured with recombinant human IL-2 (rhIL-2) (4 ng/ml) in RPMI 1640 complete medium for 3 days. Expanded cells were isolated using an EasySep mouse CD8+ T cell isolation kit (STEMCELL) to achieve 91.2% purity. Then, 1 × 106 hgp100-specific CD8+ T cells were intravenously transferred into B16-F10 tumor–bearing mice.

Purification of eosinophils from the bone marrow

Bone marrows from the femur and tibia of 6- to 8-week-old mice were flushed with PBS and processed into a single-cell suspension. After red blood cells were lysed with red blood cell lysis buffer, cells were resuspended in bone marrow medium [RPMI 1640 containing 20% FBS, 55 μM β-mercaptoethanol, 10 mM nonessential amino acids, 1 mM sodium pyruvate, penicillin (100 IU/ml), streptomycin (100 μg/ml), 2 mM glutamine, and 25 mM Hepes buffer] at a concentration of 1 × 106 cells/ml. Cells were stimulated with stem cell factor (SCF; 100 ng/ml) and fms-like tyrosine kinase 3 ligand (FLT3L; 100 ng/ml) (both from PeproTech, Rocky Hill, NJ) for 4 days. On day 4, the medium containing SCF and FLT3L was replaced with a medium containing only rmIL-5 (10 ng/ml; PeproTech, Rocky Hill, NJ). On day 8, cells were transferred to new flasks and maintained in fresh medium supplemented with rmIL-5. Half of the medium was replaced with fresh medium containing rmIL-5 every other day, and the cell concentration was adjusted each time to 1 × 106 cells/ml. On day 14, BMDEs were stimulated with mIFNγ (15 ng/ml) and mTNF (15 ng/ml) for 16 hours to activate cells. BMDE purity and viability were analyzed by flow cytometry.

Transwell migration assay

Migration was assessed in 24-well transwell plates with a 5-μm pore size (Corning Life Sciences) using RPMI 1640 medium supplemented with 0.2% FBS, 2 nM glutamine, penicillin (100 IU/ml), streptomycin (100 μg/ml), and 0.05 mM β-mercaptoethanol as the migration medium. For tracking eosinophils, activated BMDEs were tagged with 5 μM CFSE (BioLegend). The upper chamber was loaded with 1 × 105 activated BMDEs with 100 μl of migration medium. A total of 1 × 105 tumor cells (nonirradiated or irradiated tumor cells from B16-F10 tumor–bearing mice) were added to a volume of 600 μl to the lower chamber. In dose-response studies, tumor cell density in the lower chamber was adjusted to 2 × 106 cells. Plates were incubated at 37°C in the presence of 5% CO2 for 21 hours. Then, the contents of the lower chamber were carefully collected and quantified using flow cytometry.

For tracking CD8+ T cells, the upper chamber was loaded with 1 × 105 activated CD8+ T cells (activated with anti-CD3 and anti-CD28 antibody; both from BioLegend) via 100 μl of migration medium. A total of 1 × 105 eosinophils (activated by nonirradiated or irradiated tumor supernatant) were added to a 600-μl volume in the lower chamber. In dose-response studies, the density of eosinophils in the lower chamber was adjusted from 1 × 105 and 5 × 105 to 2.5 × 106 cells. Plates were incubated at 37°C in the presence of 5% CO2 for 5 hours. Then, the contents of the lower chamber were carefully collected and quantified by flow cytometry. Migration rate was calculated as the number of cells in the lower chamber/the number of input cells × 100.

Sorting and adoptive transfer of eosinophils

Mice were intraperitoneally injected with 2.5 ml of thioglycolate medium; after 3 to 4 days, a DMEM-resuspended peritoneal lavage solution was collected for sorting using a cell sorter (BD FACSAria II). For in vivo eosinophil migration experiment, we intravenously injected 4 × 105 eosinophils in 200 μl of PBS with a purity of >90% on the day of irradiation. For adoptive transfer, 1 × 106 in vitro activated eosinophils were transferred.

Histological analysis

Patient specimens were fixed in 10% formalin overnight and embedded in paraffin. Four-micrometer-thick sections of patient specimens were stained with H&E, observed under a light microscope (Olympus BX3-CBH; Japan), and scanned for digital imaging (Olympus DP50; Japan). Histopathological examinations were performed by a pathologist.

Immunofluorescent staining of frozen tumors

Frozen tumor tissue samples were cut into 4-μm-thick sections, placed on glass slides, and fixed with paraformaldehyde (AR1069; Boster) for 10 min. Sections were washed with PBS for three times, incubated with bovine serum albumin (BSA) for 30 min at 37°C, and then incubated with a 1:500 dilution of rabbit anti-CD8α antibody (ab217344; Abcam) or isotype antibody (ab172730; Abcam) overnight at 4°C. Tissue sections were rewarmed at 37°C for 30 min and incubated with a secondary Alexa Fluor 647–conjugated goat anti-rabbit antibody (Beyotime) at a 1:100 dilution at 37°C for 30 min. The anti–Siglec-F antibody (E50-2440; BD Biosciences) or isotype control (553930; BD Biosciences) was added at a 1:100 dilution to these sections for 30 min at 4°C in the dark. Then, sections were washed with PBS three times and stained with 4′,6-diamidino-2-phenylindole (DAPI) for 4 min at room temperature. After mounting, slides were observed using a confocal microscope (TCS SP5X; Leica, Wetzlar, Germany).

Immunohistochemical staining of paraffin-embedded tumors

Surgically resected tumor samples obtained from lung cancer patients were cut into 4-μm-thick sections, placed on glass slides, incubated in a 60°C incubator for more than 2 hours, deparaffinized in xylene, and rehydrated in decreasing concentrations of ethanol. Then, tissue sections were incubated in a retrieval solution for antigen retrieval at 95°C for 30 min and incubated in 3% H2O2 for 15 min and in BSA (AR004; Boster) for 30 min at room temperature, before being incubated overnight with an anti–Siglec-8 primary antibody (ab38578; Abcam) or rabbit isotype antibody (ab171870; Abcam). Tissue sections were rewarmed at 37°C for 30 min and incubated with a secondary antibody (KIT-5020; MXB Biotechnologies) at 37°C for 30 min. Chromogenic detection was performed using DAB (3,3′-diaminobenzidine); tissue sections were counterstained with hematoxylin and mounted with neutral balsam. Slides were observed using a light microscope (Olympus BX3-CBH; Japan) and scanned for digital imaging (Olympus DP50; Japan).


Data were analyzed using GraphPad Software (Prism version 5.0), SPSS (IBM), and the RStudio. Except at instances where it was indicated otherwise, experiments were repeated two to three times and comparisons between two groups were assessed by a two-tailed unpaired Student’s t test. Tumor sizes were assessed by two-way analysis of variance (ANOVA) with multiple comparisons. Survival data were analyzed using the Kaplan-Meier method and log-rank (Mantel-Cox) test. P values <0.05 were considered statistically significant.


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We thank L. Deng (Shanghai Jiao Tong University School of Medicine, China) for providing the MC38 cell line. Funding: We acknowledge support from the National Youth Science Foundation of China (no. 31700776 to Q.J.), the National Natural Science Foundation of China (no. 81620108023 to B.Z.), the National Science Fund for Distinguished Young Scholars (no. 81925030 to B.Z.), and the Foundation for Innovative Research Groups of the National Natural Science Foundation of China (no. 81821003 to B.Z.). Author contributions: J.-N.C. conducted most experiments. Q.J. analyzed and interpreted the genomic data. J.-N.C., Y.Y.W., Q.J., Q.-J.L., and B.Z. designed all experiments. W.L., C.S., and X.Z. helped with the collection of patient records. C.S., Z.J., and Q.J. helped with statistics and bioinformatics to analyze data. Z.G. provided advice for immunohistochemical staining. X.X., X.D., S.X., and P.Z. provided technical assistance with most experiments. J.-N.C. prepared the manuscript. P.B.A., Q.J., Y.Y.W., Q.-J.L., and B.Z. directed the research. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data associated with this study are presented in the paper or Supplementary Materials, and the raw RNA-Seq data have been submitted to the National Center for Biotechnology Information (accession number: PRJNA493738).

Stay Connected to Science Advances

Navigate This Article