High-dose irradiation promotes macrophage-mediated engulfment of triple-negative breast cancer through M1 polarization via the IKZF1-CCL5 axis.

Introduction
Breast cancer is the most common cancer and the leading cause of cancer-related death among women worldwide [1, 2]. Breast cancer can be categorized into four subtypes based on the expression of different receptors: estrogen receptor (ER)-positive, progesterone receptor (PR)-positive, human epidermal growth factor receptor 2 (HER2)-positive, and triple-negative breast cancer (TNBC) [3]. Among these, TNBC is the most aggressive subtype, characterized by poor prognosis due to the absence of key receptors [4–6]. Typical treatment options for early-stage TNBC include surgery, radiotherapy, and chemotherapy [7]. Radiotherapy is commonly used after surgery to eliminate any residual cancer cells in the breast or surrounding areas [8]. Beyond its direct cytotoxic effects, increasing evidence indicates that radiotherapy modulates the innate tumor immune microenvironment, including macrophage activation and polarization [9, 10].
The tumor microenvironment (TME) is a complex system in which cancer cells interact with neighboring cells. Tumors recruit various cell types to support their survival by secreting cytokines or chemokines, thereby creating a diverse TME [11]. Among these surrounding cells, immune cells play dual roles in regulating tumor progression [12]. Typically, tumors interact with immunosuppressive cells to evade the immune response [12, 13]. Previous studies have reported that tumor-associated macrophages (TAMs) are the most prevalent immune cell type in the TME of triple-negative breast cancer (TNBC) [14, 15]. Moreover, macrophages can polarize into two main subtypes, M1 and M2, in response to different stimuli such as cytokines or irradiation [15–17]. These two macrophage subtypes play opposing roles in the TME: M1 macrophages secrete pro-inflammatory cytokines against tumor growth, whereas M2 macrophages secrete anti-inflammatory cytokines and promote tumor survival [18–21]. However, TAMs in the TME are typically polarized toward the M2 phenotype [22]. Reprogramming TAMs toward an M1 phenotype restores their phagocytic function and enhances antigen presentation through increased expression of specific surface molecules, ultimately promoting adaptive immune activation and antitumor immunity [23–27]. Given these phenotypic alterations in TAMs, it is important to investigate their potential therapeutic strategies further.
While radiotherapy remains a key component of local management, it also plays a critical palliative role in advanced disease, offering effective relief from pain, bleeding, and mass-effect symptoms. These clinical needs highlight the importance of optimizing radiation strategies for patients with TNBC [28]. Radiotherapy is a common treatment for TNBC, not only for its cytotoxic effects but also for its ability to remodel TME and potentially enhance therapeutic efficacy [29]. Although macrophage polarization following ionizing radiation (IR) has been extensively studied [30, 31], the results remain inconsistent. Several reports indicate that high-dose IR promotes M1 polarization and suppresses tumor growth [32, 33], whereas low-dose IR (e.g., 0.5 Gy) tends to increase M2-like populations [34]. Moreover, macrophage polarization can vary depending on the timing and doses post-IR [35, 36], and the polarization outcome appears highly dependent on tumor context. For instance, M2 polarization has been reported to be enhanced in glioblastoma and cervical cancer following radiation doses of 10 Gy and 8 Gy, respectively, instead of M1 polarization [37, 38]. , contradicting findings from other tumor models. Despite these complexities, few studies have systematically explored how IR influences TAM polarization in TNBC, a subtype known for limited treatment options and poor prognosis. This knowledge gap underscores the importance of tumor-specific investigations. Therefore, in this study, we aimed to investigate how low- and high-dose IR affect TAM polarization and macrophage-mediated tumor inhibition in TNBC, and to elucidate the underlying molecular mechanisms. Clarifying this dose-dependent immune modulation may provide valuable insights into improving radiotherapy efficacy in TNBC by targeting macrophage reprogramming.

Materials and methods

Cell culture and IKZF1 overexpression (OE) model establishment
RAW 264.7 and THP-1 cells were kindly provided by Prof. Shu-Lin Fu (National Yang Ming Chiao Tung University, Taiwan) and Prof. Ming-Huang Chen (Taipei Veterans General Hospital, Taiwan), respectively. The murine TNBC cell line 4T1 was purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA), and the 4T1-3R cell line expressing monomeric red fluorescent protein (mRFP), luciferase 2 (luc2), and herpes simplex virus type I-thymidine kinase (HSV1-tk), was gifted by Prof. Yi-Jang Lee (NYCU, Taiwan). RAW 264.7, 4T1, and 4T1-3R cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Gibco, Billings, MT, USA), whereas THP-1 cells were maintained in Roswell Park Memorial Institute 1640 (RPMI-1640; Gibco). All media were supplemented with 10% FBS (CORNING, Manassas, VA, USA) and 1% penicillin–streptomycin (CORNING). All cells were maintained in a 37 °C incubator with 5% CO2. Cells were subcultured when confluence reached 80–90% to ensure that all cells were in an exponential growth phase during the experiments. IKZF1-overexpressing cells were generated by lentiviral transduction using a mouse Ikzf1 (NM_009578) tagged ORF clone (OriGene Technologies, Rockville, MD, USA). Cells were infected for 4 days and subsequently selected with puromycin (1 µg/mL; InvivoGen, Waltham, MA, USA) for 2 weeks to establish stable cell lines.

Animal model
Six female immunocompetent BALB/c mice (6 weeks old) were obtained from the National Center for Biomodels (Nangang, Taipei, Taiwan) and housed under specific pathogen-free (SPF) conditions with controlled temperature (22 ± 1 °C), humidity (55 ± 10%), and a 12-hour light/dark cycle, with ad libitum access to food and water. To establish an orthotopic TNBC model, 5 × 10⁵ 4T1-3R cells in 100 µL PBS were injected into the fourth mammary fat pad of each mouse. Fourteen days post-inoculation, mice were randomly divided into control (0 Gy) and IR (8 Gy) groups. Tumor growth was measured every 2–3 days using calipers, and volume was calculated as (length × width²)/2. Tumor progression and metastasis were monitored weekly using an in vivo imaging system (IVIS Spectrum; PerkinElmer). All tumors remained below the 1500 mm³ limit in accordance with IACUC guidelines, as confirmed by the growth curves shown in Fig. 3. At the study endpoint, mice were euthanized for tissue collection and histological analysis. All procedures were approved by the IACUC of National Yang Ming Chiao Tung University (Approval No. 1131112).

X-ray irradiation
X-ray irradiation was conducted using a small-animal irradiator (dose rate, 1.34 Gy/min; X-Rad225xL, Precision, Madison, CT, USA). The cells and mice were placed on a platform at a source-to-surface distance (SSD) of 50 cm and irradiated with designated doses (0.5 and 8 Gy). Additionally, the mice were partially shielded with custom lead coverings to protect non-targeted areas from IR exposure. For cell-based experiments, irradiated cells were collected at 24 h post-IR for subsequent analyses. For in vivo experiments, tumor progression and metastasis were evaluated as described above.

Cytokine treatment
RAW 264.7 cells were exposed to 1 µg/mL lipopolysaccharide (LPS) (Cayman Chemical, Ann Arbor, Michigan, USA), or 20 ng/mL interleukin 4 (IL-4) and 20 ng/mL interleukin 13 (IL-13) (PeproTech, Beit Tamar, Rehovot, Israel) for 24 h. The cells were collected for further analysis. In addition, 400 ng/mL C-C motif ligand 5 (CCL5) recombinant proteins (PeproTech) and vascular endothelial growth factor (VEGF) recombinant proteins (PeproTech) were added 8 h after 8 Gy IR, and the cells and medium were collected for further experiments. The recombinant proteins are listed in Supplementary Table S1.

Indirect co-culture
The indirect co-culture method is also known as conditioned medium (CM) treatment. The medium from the unirradiated or irradiated (0.5 Gy or 8 Gy) macrophages was collected. Subsequently, 4T1 cells were treated with the CM derived from the unirradiated or irradiated (0.5 or 8 Gy) RAW 264.7 cells mixed with serum-containing DMEM (1:1).

Assay for migration by using Boyden chamber
A Boyden chamber (48-well, Neuro Probe, Inc., Gaithersburg, MD, USA) was assembled with a fibronectin-coated track-etched membrane (GVS Filter Technology, Sanford, ME, USA) that was dried for 10–15 min. The lower chamber was loaded with 32 µL of complete DMEM, and the membrane was placed over it, followed by the silicone gasket and upper chamber secured with thumbnuts. 4T1 cells were resuspended with 50% CM from unirradiated or irradiated (0.5 or 8 Gy) RAW 264.7 cells and 50% serum-free DMEM and then seeded in the upper chamber. The chamber was covered with a plastic wrap and incubated. After 6 h, cells on the membrane were fixed with 100% methanol for 10 min and stained with Giemsa solution for 2 h. Cell migration was evaluated by counting the purple-stained nuclei under a bright-field microscope (40X).

Cytokine array
The supernatants from unirradiated or irradiated (0.5 or 8 Gy) RAW 264.7 cells were collected and concentrated using Amicon® Ultra Centrifugal Filters (Cat. No. UFC801096, Merck Millipore, Burlington, MA, USA) through high-speed centrifugation (4,000 × g, 15 min). To identify the specific cytokines regulated by IR, we performed a cytokine array with the collected supernatants using the Mouse Cytokine Antibody Array C3 Kit (Cat# AAM-CYT-3-8, RayBiotech Life, Inc. Peach Tree Corners, GA, USA), following the manufacturer’s protocol. Sixty-two targets were pre-coated onto the membranes provided by the kit. Images were captured using the Luminescence/Fluorescence Image System and analyzed using ImageJ software (version 1.8). The luminescence signal for each group was calibrated using Positive Control Spots.

Enzyme-Linked Immunosorbent Assay (ELISA)
The supernatants from unirradiated or irradiated (0.5 or 8 Gy) RAW 264.7 cells were concentrated using Amicon® Ultra centrifugal filters (Cat. No. UFC801096, Merck Millipore) (4,000 × g, 15 min). Serum was collected from clotted blood samples after centrifugation (2,000 × g, 10 min), and all samples were stored at − 80 °C. CCL5 levels were quantified using the Mouse CCL5/RANTES DuoSet ELISA kit (Cat. No. DY478, R&D Systems) and the Human CCL5 (RANTES) ELISA MAX™ Deluxe Set (Cat. No. 440804, BioLegend), according to the manufacturers’ instructions. Cell culture supernatants were analyzed after 1:10 dilution, whereas serum samples were measured undiluted, with concentrations calculated based on standard curves. Absorbance was measured at 450 nm and 570 nm (reference) using a microplate reader (Infinite® 200 PRO, TECAN). Final values were calculated by subtracting 570 nm readings from 450 nm and interpolating from a standard curve.

In Vivo Imaging System (IVIS)
To monitor tumor growth and metastasis non-invasively, the mice were imaged using an IVIS imaging system (PerkinElmer). Mice were anesthetized with 2–3% isoflurane in oxygen and intraperitoneally injected with Pierce™ D-luciferin (150 mg/kg body weight; Cat. No. 88292; Thermo Fisher Scientific, Waltham, MA, USA) 10 min prior to imaging. Bioluminescence signals were captured using standardized acquisition parameters, and signal intensity was quantified using Living Image software 4.7.4 (PerkinElmer). Regions of interest (ROIs) were manually drawn around the primary tumor site or metastatic regions, and the total photon flux (photons/s) was calculated for comparison between groups.

Clinical specimens
Human breast cancer tissue microarrays have been established from Kaohsiung Veterans General Hospital archives, and clinical tissues were collected with informed consent and with approval from Institutional Reviewed Board approval (VEGHKS13-CT11-18). All tissue sections with tumor parts were fixed by 10% formalin, dehydrated and paraffin-embedded, and further histologically examined for the presence of tumor with hematoxylin and eosin (H&E) and Immunohistochemistry (IHC) stain.

Ingenuity canonical Pathway Analysis (IPA)
The whole gene level from the RNA-sequencing dataset was normalized to the 0 Gy group to obtain the fold change caused by the treatments. Fold changes (cutoff = 1.5) were analyzed using IPA to identify the potentially related pathways. Z-scores were used to evaluate the extent to which specific pathways were activated or inhibited. Z-scores are presented as a color spectrum. The red square represents a positive correlation with increased doses, while the blue square represents a negative correlation.

Phagocytosis assay in vitro
Phagocytic activity was assessed using a Phagocytosis Assay Kit (IgG FITC; Cat# 500290, Cayman Chemical). Twenty-four hours after IR exposure (0.5 or 8 Gy), RAW 264.7 cells were incubated with FITC-labeled beads for 4 h. Fluorescent images were acquired using a CKX41 fluorescence microscope (Olympus), and signal intensity was quantified using ImageJ (v1.8). Cell pellets were washed with 1X trypan blue in assay buffer to remove non-phagocytosed beads, then fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100. Cells were stained with anti-RAB7 monoclonal antibody (A12308, Abclonal) followed by Alexa Fluor 594–conjugated secondary antibody (#A-11012, Thermo Fisher). After resuspension in 1% FACS buffer, samples were analyzed using CytoFLEX flow cytometry (Beckman Coulter). Mean fluorescence intensity (MFI) was normalized to the untreated control (no beads).

Direct co-culture
4T1 cells were pre-labeled with carboxyfluorescein succinimidyl ester (CFSE) (ab113853, Abcam), and co-cultured with RAW 264.7 cells that had been irradiated with 0.5 or 8 Gy, 24 h post-irradiation. After 4 h of co-culture, cells were stained with APC-conjugated anti-mouse F4/80 antibody (clone BM8, BioLegend) for 30 min at room temperature, washed, and resuspended in FACS buffer. Flow cytometry (CytoFLEX, Beckman Coulter) was used to detect FITC+APC+ double-positive populations, representing macrophage–tumor cell interactions. Percentages were normalized to the unirradiated group.
For confocal imaging, RAW 264.7 cells were pre-labeled with IVISense™ 680 (VivoTrack™, Revvity) for 24 h, irradiated (0.5 or 8 Gy), and co-cultured with CFSE-labeled 4T1 cells for 4 h. Live-cell imaging was performed using a confocal microscope (LSM880) at 488 and 633 nm.

Statistical analysis
Data are presented as the mean ± SD or S.E.M. Statistical analyses were performed using GraphPad Prism (version 7.04, Boston, MA, USA). Statistical significance between the two groups was determined using a two-sided Student’s t-test. One-way analysis of variance (ANOVA) was performed for comparisons among three or more groups. Statistical significance was evaluated using Student’s t-test at *, p < 0.05; **, p < 0.01; ***, p < 0.001; NS, not significant to represent differences between groups.

Results

High-dose IR drives M1 macrophage polarization
We treated RAW 264.7 cells with LPS or IL-4/IL-13 to validate the selected M1 and M2 markers used in subsequent studies. RT-qPCR confirmed significant upregulation of Nos2 and Tnf after LPS treatment (Supplementary Figure S1A), and protein analysis, including western blotting and flow cytometry, showed increased inducible nitric oxide synthase (iNOS) and Cluster of Differentiation 80 (CD80) (Supplementary Figure S1B-D). Similarly, Arg1 and Il10 levels were enhanced after IL-4/IL-13 treatment (Supplementary Figure S2A) along with increased arginase 1 (Arg-1) and interleukin 10 (IL-10) protein levels (Supplementary Figure S2B-D). These results validate the appropriateness of our chosen markers for distinguishing between M1 and M2 polarizations.
We investigated macrophage polarization after exposing RAW 264.7 cells to low (0.5 Gy) and high (8 Gy)-dose X-ray irradiation. RNA was extracted from unirradiated and irradiated cells and transcriptomic profiling was performed using RNA sequencing. In the heatmap generated from the RNA sequencing dataset, we observed an upregulation of some M1 markers along with a downregulation of M2 markers following 8 Gy IR (Fig. 1A), which was less obvious in the 0.5 Gy group. To validate these findings, we performed RT-qPCR to assess the gene expression of Nos2 and Cd80 (M1) and Vegfa and Arg1 (M2) in cells after exposure to low-(0.5 Gy) and high-dose (8 Gy) IR. The RT-qPCR results were consistent with the RNA sequencing data (Fig. 1B). Additionally, we examined related protein levels using western blotting and flow cytometry. The expression of M1 markers, iNOS and CD80, increased significantly, while that of M2 markers, CD206 and Arg-1, decreased after 8 Gy IR (Fig. 1C and D). CD80 expression in the 8 Gy group was higher than that in the control group, as a rightward shift was observed in the flow cytometry results. However, there was no significant difference in the Arg-1 signal between the groups. (Fig. 1E and F). To evaluate the translational relevance of our findings, we additionally performed key experiments in a human macrophage system. High-dose IR similarly induced M1-like polarization in THP-1 cells, as confirmed by qPCR and western blotting (Fig. 1G and H), indicating that IR-induced macrophage reprogramming is conserved in human systems. These findings suggest that high-dose (8 Gy) IR can promote M1 polarization in macrophages.

IR-modulated M1 macrophages suppress the proliferation and migration of TNBC cells
After discovering that macrophages could be polarized to the M1 phenotype after 8 Gy IR treatment, we investigated how these IR-modulated M1 macrophages affected the behavior of TNBC cells using 4T1 cells. We conducted indirect co-culture by adding CM collected from macrophages to 4T1 cells (Fig. 2A). We observed that 8 Gy CM treatment resulted in the lowest cell count and the highest proportion of cells arrested in the G1 phase among all groups (Fig. 2B and C, and Supplementary Figure S3). This suggested that 8 Gy IR can modify the cytokine profile of RAW 264.7, which suppresses the growth and enhances G1 arrest of 4T1 cells. Similarly, Giemsa-staining images showed that the 8 Gy CM treatment group had the fewest migrated cells compared to the other groups (Fig. 2D, top, and Fig. 2E), indicating that high-dose IR significantly inhibited the migration ability of 4T1 cells. Accordingly, we hypothesized that epithelial-mesenchymal transition (EMT) might be inhibited in our model. To test our hypothesis, we examined the expression of EMT-related molecules and found that the 8 Gy CM group had the lowest expression of the mesenchymal marker MMP-9 and the highest expression of the epithelial marker E-cadherin (Supplementary Figure S4). Moreover, CM from 8 Gy–irradiated THP-1 cells exerted a similar inhibitory effect on the migratory capacity of MDA-MB-231 cells (Fig. 2D, bottom, and Fig. 2F), consistent with the inhibitory effects observed in murine models. Taken together, these findings suggest that high-dose IR-modulated M1 macrophages may exert antitumor effects by inhibiting TNBC cell proliferation, survival, migration, and EMT.

High-dose IR reprograms cytokine profiles in macrophages to regulate polarization and tumor suppression
Based on our observations in the indirect co-culture system, we hypothesized that high-dose IR alters macrophage cytokine secretion, thereby influencing polarization and inhibiting tumor progression. To investigate this, we utilized cytokine arrays to identify the key regulatory factors (Fig. 3A). By analyzing the secretion levels of all candidates, we found that CCL5 was the most abundantly secreted cytokine in untreated macrophages (Fig. 3B). In addition, we observed a significant dose-dependent decrease in CCL5 and VEGF expression (Fig. 3C). These findings were validated by western blot analysis of CM collected from IR-treated macrophages. The 8 Gy group exhibited the lowest levels of CCL5 and VEGF compared to the other groups (Fig. 3D and E), consistent with the RNA sequencing results (Fig. 3F), especially in the 8 Gy group. We further analyzed the relationship between CCL5 and VEGF using the online tool TISIDB (http://cis.hku.hk/TISIDB/), which calculates the associations between tumors and the immune system. Our analysis revealed that CCL5 was positively correlated with macrophage abundance in breast cancer, whereas VEGF showed a negative correlation (Supplementary Figure S5A and S5B). To verify these findings, we measured CCL5 concentration in the CM from irradiated murine macrophages using ELISA and observed a dose-dependent decrease in CCL5 levels (Fig. 3G). Consistently, in the human system, ELISA analysis showed that CM from 8 Gy–irradiated THP-1 cells exhibited significantly reduced CCL5 secretion (Supplementary Figure S6), supporting the translational relevance of this effect.

High-dose IR inhibits TNBC growth and metastasis through downregulating the CCL5 level
To mimic the in vivo environment, we established an orthotopic TNBC model by injecting 4T1-3R cells into BALB/c mice. After administering 8 Gy of IR to the tumor site, luminescence signals were monitored weekly using an IVIS imaging system until the experimental endpoint (Fig. 4A). Compared with the control group, tumors treated with 8 Gy IR exhibited significantly lower signals within two weeks post IR exposure. (Figure 4B and C). Additionally, primary tumor sizes were significantly reduced in the 8 Gy IR group compared to the control group (Fig. 4D and E), while the body weights of mice remained stable (Supplementary Figure S7), indicating good treatment tolerability. These findings, consistent with IVIS imaging data, demonstrate that 8 Gy IR effectively suppressed TNBC progression. Furthermore, metastatic signals in the lung and liver were observed in the control group, whereas no detectable metastatic signals were observed in the 8 Gy IR group on day 21.Partial images are shown to highlight representative metastatic regions for clarity (Fig. 4F). At the endpoint, tumors and organs (e.g., liver and lung) were collected for pathological analysis. Multiple metastatic nodules were detected in the control group but were absent in the 8 Gy IR group (Fig. 4G). H&E staining further confirmed these findings, revealing abundant metastatic lesions in liver and lung tissues of the control group, whereas such nodules were not observed in those of mice treated with 8 Gy IR (Supplementary Figure S8). We further observe an upregulation of the M1 marker (CD80) and a downregulation of the M2 marker (CD206) in 8 Gy group compared to control group (Supplementary Figure S9). We also confirmed that CCL5 expression was markedly decreased in both the serum and tumor tissues of mice receiving 8 Gy irradiation (Fig. 4H and I). Furthermore, clinical specimens collected after radiotherapy exhibited lower CCL5 expression compared to those from untreated patients (Fig. 4J). All findings are consistent with our in vitro experiments.

High-dose IR activates phagocytic ability of macrophages against TNBC cells
In addition to cytokine release and dynamic changes in the M1/M2 ratio within the tumor microenvironment (TME), macrophages also rely on their phagocytic function to eradicate cancer cells. This aligns with our transcriptomics data and canonical pathways predicted by IPA, which indicated that 8 Gy IR may activate the phagosome formation pathway in RAW 264.7 cells (Fig. 5A). To further evaluate phagocytic ability, we conducted a phagocytosis assay using flow cytometry and fluorescent imaging. Macrophages were treated with FITC-labeled latex beads that emitted green fluorescence. After a 4-hour incubation, the cells were stained with a red fluorescence-conjugated antibody against the late phagosome marker RAB7 to assess phagosome formation. As shown by our flow cytometry data, both green and red fluorescence signals increased in a IR dose-dependent manner, indicating that macrophages engulfed more FITC-labeled beads and formed more phagosomes after IR exposure (Fig. 5B and C). Additionally, fluorescence images showed a marked increase in green fluorescence in the 8 Gy IR-treated group (Fig. 5D). These results suggest that 8 Gy IR enhances the phagocytic ability of macrophages.

Furthermore, we aimed to determine whether high-dose radiation can activate macrophages to engulf TNBC cells. After labeling the macrophages and 4T1 cells with cell labeling dye, we directly co-cultured RAW 264.7 and 4T1 cells and detected fluorescent signals using flow cytometry and confocal microscopy. We observed more fused signals in the 8 Gy group, as indicated by the yellow arrows in the fluorescence images (Fig. 5E) and purple squares in the flow cytometry results (Fig. 5F and G). These results demonstrate that macrophages exhibited increased phagocytic ability against TNBC cells post-IR, particularly after 8 Gy IR exposure.

CCL5 supplementation reverses IR-induced M1 polarization, impairs phagocytosis, and restores cancer hallmarks of TNBC cells
Based on our findings, CCL5 may be a crucial chemokine in CM collected from irradiated macrophages and could inhibit TNBC progression. Therefore, we applied recombinant CCL5 protein (rmCCL5) to macrophages and examined whether it complemented the expression and secretion of CCL5 and reversed the anti-tumor ability of macrophages. Through the concentration test of rmCCL5, we found that 400 ng/mL of rmCCL5 not only induced the highest CCL5 expression (Supplementary Figure S10A and S10B) but also caused the highest CCL5 secretion (Supplementary Figure S11A and S11B). In addition, rmCCL5 was not cytotoxic to macrophages (Supplementary Figure S12); therefore, we used 400 ng/mL of rmCCL5 in subsequent experiments. After rmCCL5 treatment, CCL5 secretion in macrophage-conditioned medium was significantly increased (Fig. 6A and B). As CCL5 has been reported as an indicator of M2 polarization [39], analysis of the GSE111315 dataset showed an increased CCL5/CD80 ratio following overexpression of the M2 inducer CD5 molecule-like (CD5L) (Supplementary Figure S13). Consistently, in our model, the Ccl5/Cd80 ratio was reduced after 8 Gy IR and restored by rmCCL5 treatment (Fig. 6C). In parallel, given that cytokine array analysis identified VEGF as another IR-suppressed cytokine alongside CCL5, we also validated the comparable effects of CCL5 and VEGF recombinant protein on macrophage M2 polarization. Comparative analyses demonstrated that CCL5 exerted a stronger effect on M2 polarization than VEGF under equivalent experimental conditions (Supplementary Figure S14).

We further observed that rmCCL5 treatment reduced the phagocytic ability of macrophages, as indicated by the decrease in green signals using the phagocytic assay (Fig. 6D). In addition, flow cytometry revealed that rmCCL5 treatment caused a reduction in the fusion signals of macrophages and 4T1 cells, as shown by the purple square in the flow cytometry results (Fig. 6E and F). These results suggested that CCL5 plays a key role in inhibiting the phagocytic activity of TNBC cells. We repeated the previous assays to evaluate the proliferation, cell cycle, and migration abilities of 4T1 cells after rmCCL5 treatment. Consistent with our previous results, 8 Gy-CM treatment resulted in the slowest cell growth (Fig. 6G), accompanied by significant G1-phase arrest (Fig. 6H and Supplementary Figure S15). Additionally, these effects were reversed by the addition of rmCCL5 to CM (Fig. 6G and H and Supplementary Figure S15). Furthermore, the migration ability of 4T1 cells decreased in the 8 Gy-CM group, but was restored after treatment with CM containing rmCCL5, as shown in the Giemsa staining images (Fig. 6I and J). These findings demonstrated that CCL5 counteracts IR-modulated M1 polarization and its associated phagocytic ability, thereby promoting the proliferation and migration of TNBC cells.

IR-modulated M1 macrophages suppresses TNBC progression via IKZF1-CCL5 axis by promoting IKAROS Family Zinc Finger 1 (IKZF1) ubiquitination
We screened IR-modulated transcription factors using IPA to explore the potential regulatory role of CCL5 in macrophage polarization. IKZF1 emerged as one of the significantly downregulated transcription factors detected after 8 Gy IR exposure (Fig. 7A). Previous studies have stated that IKZF1 could regulate the activity of the CCL5 [40]. Consistently, TISIDB analysis showed that IKZF1 has a strong correlation with CCL5 expression in breast cancer (Fig. 7B and Supplementary Figure S16A). In the GSE93602 dataset, the IKZF1 knockout model suppressed Ccl5 expression compared to that in the wild-type group (Supplementary Figure S16B).

We continuously investigated the expression of IKZF1 under different IR conditions and found that IKZF1 was downregulated under 5 Gy conditions in the GSE145577 dataset (Supplementary Figure S16C). In our study, IKZF1 expression decreased as IR dose increased (Fig. 7C and D). Because IKZF1 is a transcription factor, we performed an immunofluorescence assay to examine its localization and expression. The red fluorescence signal co-localized with Hoechst-stained nuclei, indicating nuclear localization. Moreover, IKZF1 expression decreased in an IR-dependent manner (Fig. 7E and F). According to previous studies, IKZF1 can be stabilized by USP7 by regulating its ubiquitination [41]. Therefore, we speculated that IKZF1 might be degraded in macrophages after exposure to 8 Gy IR. To test this hypothesis, cells were treated with the proteasome inhibitor MG-132, which demonstrated that 8 Gy IR reduces IKZF1 protein stability (Fig. 7G). In addition, to determine whether IR-induced IKZF1 suppression is specific to macrophages or also occurs in TNBC cells, we examined IKZF1 levels across both cell types. Interestingly, we found that IKZF1 expression is intrinsically lower in 4T1 cells compared with RAW 264.7 macrophages (Supplementary Figures S17A and S17B).
To further support the causal involvement of the IKZF1–CCL5 axis, we genetically upregulated IKZF1 in RAW 264.7 cells. IKZF1 overexpression, confirmed by fluorescence microscopy and flow cytometry (Supplementary Figure S18), attenuated IR-induced M1 polarization and partially reversed macrophage-mediated tumor suppression, consistent with the effects of recombinant CCL5 supplementation (Supplementary Figures S19 and S20). Collectively, these data indicate that high-dose IR can promote IKZF1 degradation, suppress the IKZF1–CCL5 axis, and drive macrophage polarization toward an M1 phenotype, providing direct genetic evidence for a causal role of this axis in regulating macrophage polarization and downstream tumor cell behavior.

Discussion
Some studies have shown that macrophage polarization in the TME varies with the radiation dose [36, 42]. Therefore, it is worth investigating how different radiation doses affect macrophage polarization in our co-culture model of TNBC and macrophages. In our study, we only applied high (8 Gy) and low (0.5 Gy) doses of IR to RAW 264.7 cells, and high-dose IR promoted polarization toward the M1 subtype. However, 8 Gy of IR may cause normal tissue damage and other side effects. To improve the situation, we designed experiments with a fractionated strategy to observe whether these approaches could lead to similar results in macrophage polarization. We compared a regimen of 2 Gy for four consecutive days with a single dose of 8 Gy. Surprisingly, we observed that the macrophages were still polarized to the M1 type under the fractionated regimen (2 Gy×4) (Supplementary Figure S21A and S21B). The results suggest that the fractionated schedule of 2 Gy could achieve similar effects as a single dose of 8 Gy (Fig. 1A-F).
In this study, we used RAW 264.7 cells as an in vitro macrophage model. These cells, originally derived from a tumor induced by Abelson murine leukemia virus, are considered macrophage-like and are widely used in immunological studies. However, several reports and our own observations have noted that RAW 264.7 cells readily polarize toward the M1 phenotype but exhibit limited responsiveness to M2-polarizing stimuli (Figure S1 and S2). Given the limitations, the use of primary macrophage cultures, such as bone marrow-derived macrophages (BMDMs), is recommended for more physiologically relevant modeling. As part of our preliminary efforts, we previously isolated bone marrow cells from mice and successfully differentiated them into BMDMs. In addition to assessing cell morphology (Supplementary Figure S22A), we confirmed their polarization capacity by western blot analysis. These results showed that BMDMs can effectively polarize to the M2 phenotype in response to IL-4/IL-13 treatment and significantly shift toward the M1 phenotype following 8 Gy IR (Figure S22B). These findings not only support the functional relevance of our in vitro observations but also underscore the necessity of validating key findings in primary macrophage systems prior to in vivo applications.
In our study, we employed both direct and indirect co-culture systems to investigate contact-dependent and contact-independent regulation mediated by soluble factors. While direct co-culture revealed enhanced macrophage-mediated phagocytosis, indirect co-culture experiments demonstrated suppression of tumor cell proliferation and migration, as well as induction of cell cycle arrest. These findings suggest that irradiated macrophages can exert substantial paracrine effects on tumor cells independent of direct physical contact. Given that immune cells within the TME are not always in direct contact with tumor cells, the indirect co-culture model provides a physiologically relevant framework to study cytokine-mediated macrophage–tumor communication.
Beyond macrophages, high-dose irradiation modulates multiple immune components within the TME. Radiotherapy can enhance dendritic cell activation and type I interferon–linked antigen presentation to promote CD8⁺ T-cell priming and infiltration, while also inducing adaptive immunosuppressive responses such as PD-L1 upregulation [10, 43]. In addition, irradiation has been reported to promote myeloid-derived suppressor cell (MDSC) accumulation and alter natural killer (NK) cell recruitment and function through changes in chemokine signaling, depending on radiation dose and experimental context [44]. Notably, advanced TNBC is characterized by a highly immunosuppressive TME enriched with MDSCs, regulatory T cells, granulocytes, and cancer-associated fibroblasts, which collectively limit effective anti-tumor immunity and contribute to the modest clinical efficacy of radiotherapy [45, 46]. Accordingly, macrophage-mediated effects observed in this study are likely shaped by coordinated changes across multiple immunosuppressive compartments within the TME [46, 47].
In this study, we observed a dose-dependent reduction in CCL5 and VEGF expression following IR, suggesting that high-dose IR may suppress pro-tumorigenic cytokines secreted by macrophages. However, the immunoblotting results show weakly significant differences among the groups (Fig. 2C and D), likely due to the limited sensitivity of western blotting in detecting secreted proteins. In contrast, ELISA analysis (Fig. 3G) revealed a significant reduction in CCL5 secretion following 8 Gy IR, underscoring the importance of using sensitive and quantitative methods to assess cytokine changes in the tumor microenvironment.
Along with previous reports, both CCL5 and VEGF have been identified as M2-associated cytokines in TNBC [39, 48]. Notably, both EGFR⁺/VEGF⁻ and EGFR⁻/VEGF⁺ tumor subtypes are associated with high levels of macrophage infiltration across different breast cancer types [49]. In our study, VEGF showed a negative correlation with macrophage abundance in breast cancer based on platform prediction (Figure S5). Therefore, we selected CCL5 as the target for subsequent analyses. Additionally, a study demonstrated that inhibiting the CCL5-CCR5 pathway can promote M1 polarization and suppress liver cancer progression [50]. In our study, we found that CCL5 impaired macrophage phagocytosis, thereby promoting tumor progression (Fig. 6D-J). Moreover, CCL5 has been reported to correlate with tumor mutation burden (TMB), microsatellite instability (MSI), and immunotherapy response across multiple cancer types [51]. Although the biological function and prognostic value of CCL5 require further in vivo studies, the potential synergy between CCL5 antagonists and radiotherapy merits further investigation, particularly in the context of modulating immune infiltration and enhancing therapeutic efficacy.
IKZF1 is a member of the zinc finger protein family and is essential for DNA repair. IKZF1 participates in RPA2 foci formation when DNA double-strand break (DSB) occurs [41]. IKZF1 then undergoes DNA end resection and initiates HR, whereas NHEJ is not significantly altered. Interestingly, in our study, IKZF1 decreased with the IR dose, which we hypothesize exceeds the threshold that leads to severe and irreparable DNA DSB. As a result, IKZF1 may no longer be required and is subsequently targeted for degradation. Detailed radiation-induced DNA repair and the performance of IKZF1 need to be arranged in more dose- and time-effect experiments to understand the actual situation. Based on our results, 8 Gy IR reduced IKZF1 expression and downregulated downstream CCL5 expression in macrophages.
In conclusion, we found that 8 Gy IR reprograms macrophages to create an antitumor environment by downregulating the IKZF1-CCL5 axis by promoting IKZF1 ubiquitination, ultimately suppressing the progression of TNBC cells. Moreover, 8 Gy IR resulted in a stronger effect that promoted macrophage phagocytosis against TNBC cells than 0.5 Gy IR (Fig. 8). Our findings highlight that a single 8 Gy IR is a powerful strategy for reshaping macrophage phenotypes and amplifying their ability to target and eliminate tumor cells. This breakthrough suggests that combining radiotherapy with CCL5 antagonists may further improve the therapeutic outcomes in TNBC.

Supplementary Information