Targeting senescence-like tumor-associated macrophages sensitizes chemotherapy in triple-negative breast cancer.

Introduction
Triple-negative breast cancer (TNBC) is defined by the absence of estrogen receptor (ER), progesterone receptor (PR), and HER2 expression, accounting for about 10% of all breast cancers. Because there are no approved targeted therapies for TNBC, chemotherapy remains the standard treatment, with regimens such as taxanes and anthracyclines given in the neoadjuvant or adjuvant setting to maximize pathologic complete response, while platinum agents are added in BRCA-mutant or high-risk tumors [1, 2]. Although initial responses can be transient, TNBC frequently exhibits poor durability of response due to intrinsic and acquired chemoresistance and a tendency for metastasis. TNBC is highly heterogeneous and biologically aggressive, thus limiting the overall benefit of systemic therapy, with or without immunotherapy [3]. The lack of actionable targets, complex tumor biology, and pervasive chemoresistance present a major obstacle in TNBC management. Therefore, deciphering the resistance mechanisms in TNBC is essential to identify novel targets and develop strategies to improve chemotherapy efficiency.
The tumor microenvironment (TME) and its dynamic cross-talk with cancer cells reprogram signaling networks, metabolism, apoptosis, and drug response, creating a multifactorial barrier to chemotherapy [4]. Cellular components within the TME, including cancer-associated fibroblasts (CAFs), tumor-associated endothelial cells (TECs), tumor-associated macrophages (TAMs), diverse immune cells, and cancer stem cells (CSCs), collectively shape chemoresistance in cancer [5–10]. For instance, CD10+GPR77+ CAFs sustain cancer stemness to counteract chemotherapy through remodeling the extracellular matrix in TNBC [11]. TECs can secrete SDF-1, which disrupts the anti-tumor crosstalk between CXCL10-expressing macrophages and CXCR3 + CD8+ T cells, thereby promoting tumor progression in ovarian and breast cancers [9]. Moreover, TAMs often adopt pro-tumorigenic phenotypes that promote survival signaling and immunosuppression [12, 13]. Collectively, TNBC chemoresistance arises from both tumor-intrinsic changes and persistent stromal–tumor interactions, highlighting the need to target TME components alongside malignant cells.
Cellular senescence is a stress-induced, durable arrest of the cell cycle accompanied by widespread phenotypic changes, most notably the senescence-associated secretory phenotype (SASP) [14, 15]. In cancer, senescence can act as a tumor-suppressive barrier by halting the proliferation of potential malignant cells [16]; however, aging-associated accumulation of senescent cells and their SASP can shape tumor progression [17]. The SASP is a diverse mixture of chemokines, cytokines, growth factors, and matrix-remodeling enzymes that remodel the TME and modulate immune responses, inflammation, angiogenesis, and therapy resistance [18, 19]. Senescence is not limited to cancer cells. Various non-malignant cell types within the TME, such as TAMs, can also enter a senescent state under stress or therapy [20]. Whether stromal cells, and in particular TAMs, undergo chemotherapy-induced cellular senescence remains unclear, in part because TAMs are more migratory and plastic than many other stromal cell types. Senescent non-cancer cells can either suppress or promote tumorigenesis depending on context [21–23], largely through SASP-mediated effects such as immune cell recruitment, extracellular matrix remodeling, and treatment resistance. Despite the ability of many cytotoxic regimens to elicit therapy-induced senescence, the prospect of routinely employing senescence-inducing therapies in TNBC treatment remains uncertain.
In this study, we first screened the cellular components of the TME that may contribute to chemoresistance in TNBC and identified a distinctive contribution from TAMs. Notably, TAMs within tumors treated with chemotherapy acquire a senescence-like phenotype and exhibit SASP features. Among SASP factors, interleukin-6 (IL-6) stands out as a key mediator of chemoresistance in TNBC by inducing drug resistance and promoting cancer-stemness gene expression via the IL-6 receptor (IL-6R)/STAT3 signaling axis. Depletion of senescence-like cells or blockade of the IL-6–IL-6R–STAT3 axis effectively abrogated TNBC drug resistance.

Materials and methods

Cell culture
Murine 4T1 cells were obtained from the Cell Resource Center at the Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China. For routine culture, cells were grown in RPMI-1640, supplemented with 10% fetal bovine serum (Gibco, USA) and 1% penicillin-streptomycin (Thermo Fisher Scientific; catalog no. 15140163). Cells were maintained in a humidified incubator at 37 °C with 5% CO2. All cell lines used were authenticated by short tandem repeat (STR) analysis and routinely verified to be mycoplasma-free using the Mycoplasma Elimination Kit (Sigma, MP0030). For routine maintenance, cells were passaged at 70–90% confluence using 0.25% trypsin-EDTA, counted for viability, and reseeded at appropriate densities. Cryopreservation was performed in 90% FBS with 10% DMSO and stored in liquid nitrogen.

Clinical samples
Human TNBC specimens used in this study were previously reported [24]. The study received ethical approval from The First Hospital of Jilin University Ethics Committee, and written informed consent was obtained from all participants prior to surgery for the use of their tissue samples in research.

Isolation of primary TEC, TAMs, and CAFs
Fresh TNBC surgical specimens were collected under sterile conditions and maintained on ice during processing. Tissue was rinsed in phosphate-buffered saline containing 1% penicillin/streptomycin to reduce debris, and necrotic regions were excised. Viable tissue was minced into approximately 1–2 mm3 fragments. Enzymatic digestion employed a cocktail of collagenase II (1 mg/mL; Gibco, 17101015), collagenase IV (1 mg/mL; Gibco, 17104019), and dispase (2 U/mL) incubated at 37 °C with gentle agitation for 30 min, with intermittent mixing to promote efficient disaggregation. Following digestion, the suspension was gently triturated and passed through a sterile 70-µm cell strainer to yield a single-cell suspension. Endothelial cells were enriched by magnetic-activated cell sorting using the CD31 MicroBead Kit (Miltenyi Biotec Cat# 130-091-935, RRID: AB_3699142), TAMs were enriched with the CD163 MicroBead Kit (Miltenyi Biotec, Cat# 130-124-420), and CAFs were enriched using Anti-Fibroblast MicroBeads (Miltenyi Biotec, Cat# 130-050-601), in accordance with the manufacturers’ protocols. For RT-qPCR-based identification, TAMs should show high expression of macrophage/monocyte markers such as Ptprc (CD45) and Itgam with low/undetectable Epcam to exclude non-macrophage contaminants; TECs should display robust Pecam1, with minimal Ptprc and Epcam; CAFs should express Fap and Acta2 (α-SMA), while being negative for Epcam, Ptprc, and Pecam1.

Cell survival assay
Cell viability was evaluated in 4T1 cells treated with Adriamycin (ADM; S1208, Selleck) or docetaxel (DTX; S1148, Selleck) versus vehicle for 24 h using the CCK-8 (B4007, bioexplorer) assay. Briefly, 1.5 × 104 4T1 cells per well were seeded in the upper chamber of 0.4 μm Transwell inserts, with TECs, TAMs, or CAFs in the lower chamber. Cells were treated with ADM at indicated concentrations, with untreated wells as controls. After 24 h, the upper chambers were transferred to fresh 24-well plates. To each well, 10 µL of CCK-8 solution and 90 µL of culture medium were added, and incubation proceeded for 1.5 h at 37 °C protected from light. Absorbance was measured to assess viability. Absorbance was measured at 450 nm using a BioTek microplate reader. Viability was expressed as a percentage of the control. The following blocking antibodies were used: anti-IL-1α (R&D systems, Cat. AF-400-NA; RRID: AB_354473), anti-IL-6 (R&D systems, Cat. MAB406; RRID: AB_2233899), anti-CXCL1 (R&D systems, Cat. MAB453; RRID: AB_2087696), and anti-TNF-α (R&D systems, Cat. MAB4101; RRID: AB_2240643). Recombinant IL-6 protein was purchased from R&D systems (Cat. 406-ML).

Animal experiments
For in vivo investigations, the orthotopic mammary tumor model was established in female BALB/c mice using the murine 4T1 cell line. Briefly, 5 × 105 4T1 cells were injected into the mammary fat pad to initiate tumor formation. Approximately one week after injection, when tumors reached about 100 mm3, mice were randomized to receive therapy. For monotherapy, ADM was administered at 5 mg/kg twice weekly for three weeks. For combination therapy, ABT263 (10 mg/kg daily) and anti-IL-6 (0.5 µg/g, twice weekly) were added after the ADM course and continued for an additional two weeks. Dasatinib (D; Selleck, S1021) and Quercetin (Q; Selleck, S2391) were administered by oral gavage at dosages of 5 mg/kg and 50 mg/kg for consecutive 14 days. Tumor growth was tracked by caliper measurements of length and width, with volume calculated as (length × width2)/2. Animals were housed five per cage in pathogen-free clear polycarbonate cages, provided with standard chow and water ad libitum, and maintained under controlled environmental conditions with a 12-hour light–dark cycle. Mice were monitored daily for welfare; humane endpoints were predefined, and euthanasia was performed if tumor burden or distress exceeded established thresholds.

Macrophage deletion assay
Macrophage depletion was achieved using clodronate liposomes, with PBS liposomes serving as controls. Both liposome formulations were purchased from MedChemExpress (HY-172202). Each injection consisted of 100 µL per mouse, corresponding to a clodronate dose of 40 mg/kg, delivered intraperitoneally once weekly. Two weeks before tumor cell inoculation, clodronate liposomes or PBS liposomes were given twice a week. Following tumor cell inoculation, the same regimen was continued on a weekly basis to maintain macrophage suppression during tumor progression. The efficacy of macrophage depletion was monitored in a subset of animals by assessing F4/80+ (Cell Signaling Technology, Cat. 30325; RRID: AB_2798990) macrophages in tumor tissues using immunofluorescence (IF) analysis. The detailed protocol for IF analysis was reported previously [25].

Multiplexed immunofluorescence staining
Paraffin-embedded tissue sections from human or mouse TNBC were dewaxed in 100% xylene and rehydrated through a graded ethanol series. Antigen retrieval was performed by heating the slides in sodium citrate buffer (P0081, Beyotime, Shanghai, China) to boiling for 10 min. Endogenous peroxidase activity was quenched with 3% hydrogen peroxide for 10 min, followed by blocking with species-specific serum to reduce nonspecific binding before proceeding with iterative multiplex staining. Multiplex immunofluorescence was carried out using the TSAPLus multiplex immunofluorescence kits (G1236, G1255; Servicebio). The primary antibodies employed were: F4/80 (Cell Signaling Technology, Cat. 30325; RRID: AB_2798990), CD163 (Abcam, Cat. 156769; RRID: AB_3076143), p21 (Cell Signaling Technology, Cat. 2947; RRID: AB_823586), and p16 (Abcam, Cat. 211542; RRID: AB_2891084). After completing all TSA cycles, nuclei were counterstained with DAPI. All fluorophores and DAPI were prepared in accordance with the manufacturers’ guidelines.

Conditioned medium
Conditioned medium was prepared from primary TAMs isolated from both treatment-naïve and ADM-exposed tumors, which were then cultured under standard conditions. Tumors were collected from the orthotopic 4T1 tumor model. After a 12-h serum-starvation period, the culture supernatants were collected and clarified by centrifugation to remove cells and debris. The clarified medium was subsequently filtered through a 0.22 μm membrane to ensure sterility, aliquoted to minimize freeze-thaw cycles, and stored at − 80 °C until use.

ELISA
The SASP-related cytokines and chemokines in TAM-conditioned medium were quantified by ELISA. Mouse IL-1α and CXCL-1 were measured with ELISA kits from R&D Systems (cat. MLA00 and MKC00B), and mouse IL-6 and TNF-α were measured with ELISA kits from Invitrogen (cat. 88-7064-22 and 88-7324-22). All assays were performed strictly according to the manufacturers’ protocols. Samples were run in triplicates alongside appropriate standards and blanks on each plate to generate standard curves. Concentrations were calculated from the standard curves.

SA-β-Gal staining
For SA-β-Gal staining, 10-µm thick cryosections from OCT-embedded tissues were fixed in 2% formaldehyde/0.2% glutaraldehyde for 10 min, washed in PBS, and incubated overnight at 37 °C in a humidified chamber with a staining solution containing 1 mg/mL X-Gal, 5 mM potassium ferrocyanide, 5 mM potassium ferricyanide, 2 mM MgCl₂, and 40 mM citrate-phosphate buffer (pH 6.0). Images were acquired using an Olympus BX53 microscope. Quantitative analysis of blue-stained cells was performed in Fiji/ImageJ.

RNA isolation and real-time quantitative PCR
Total RNA was isolated from 4T1 and primary TAMs using TRIzol reagent (Thermo Fisher) according to the manufacturer’s protocol. RNA integrity and concentration were assessed prior to cDNA synthesis, which was performed from 1 µg of total RNA using the HiScript II First Strand cDNA Synthesis Kit (R211-002, Vazyme) following the kit instructions. Real-time qPCR was then carried out on an ABI 7500 Real-Time PCR System using SYBR Green Master Mix (HY-K0523, MedChemExpress). Reactions were run in triplicate, and relative gene expression was calculated by the 2^−ΔΔCt method with endogenous reference gene (Actb for mouse samples). Primer sequences used are listed in Supplementary Table S1.

Western blotting
Cells or tissues were lysed on ice in RIPA buffer (50 mM Tris-HCl pH 8.0, 1% NP-40, 150 mM NaCl, 5 mM EDTA, 1 mM PMSF) supplemented with protease and phosphatase inhibitors, and lysates were cleared by centrifugation at 12,000 g for 15 min at 4 °C. Protein concentration was measured by the BCA assay, and equal amounts were separated by SDS-PAGE (6–10% gels as appropriate) and transferred to PVDF or nitrocellulose membranes. Membranes were blocked with 5% skimmed milk in TBST for 1–2 h at room temperature, then incubated overnight at 4 °C with primary antibodies followed by HRP-conjugated secondary antibodies (1:5,000) for 1 h at room temperature. The primary antibodies includes: p21 (Cell Signaling Technology, Cat. 2947; RRID: AB_823586), p16 (Abcam, Cat. 211542; RRID: AB_2891084), and β-actin (Abcam, Cat. 8227; RRID: AB_2305186). Immunoreactive bands were visualized using an enhanced chemiluminescence kit (PK10003, Proteintech) and imaged on the GelDoc Go system (Bio-Rad). Densitometric analysis was performed to quantify band intensities relative to β-actin.

Bioinformatic analysis
For determining the changes of TAMs before and after chemotherapy, we leveraged public proteomic data from the PRIDE database (https://www.ebi.ac.uk/pride/archive). Two datasets, PXD022673 (human samples) and PXD022674 (PyMT mice), provide LC-MS/MS-based proteomic profiles of TAMs derived from human and murine samples, collected before and after chemotherapy. Differential protein expression between pre- and post-therapy TAMs was identified using standard proteomics workflows (data processing, normalization, and statistical testing with correction for multiple comparisons). Proteins showing significant changes were subjected to functional interpretation through gene set enrichment analysis (GSEA, RRID: SCR_003199) against the Molecular Signatures Database (MSigDB, RRID: SCR_016863), focusing on macrophage-related pathways and signatures. Enriched pathways with robust significance (FDR-corrected q-values) were highlighted as candidate mechanisms by which chemotherapy reshapes TAM function and thereby influences tumor behavior. The TCGA drug versus survival of a predefined gene signature (CD68, CD163, IL6, IL1B, IL1A, IL10, CXCL1, CCL2, TNF, CD40, and COX2) for senescent TAMs in TNBC was assessed using the GEPIA3 online tool (https://gepia3.bioinfoliu.com/). The gene signature was selected based on a systematic review by Moss et al. [26].

Statistical analysis
All experiments were performed in triplicate or as noted in the figure legends, and continuous data are presented as means ± SD. Analyses were conducted with GraphPad Prism version 9.0. For comparisons between two groups, an unpaired two-tailed t-test was employed. For comparisons among three or more groups, one-way ANOVA followed by Tukey’s post hoc test was used to correct for multiple comparisons. Tumor growth curves were analyzed by two-way ANOVA. A p-value < 0.05 was considered statistically significant, with significance levels denoted as: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Results

TAMs contribute to chemoresistance in TNBC
To delineate how the TME contributes to TNBC chemoresistance, we employed a syngeneic orthotopic model by injecting murine 4T1 cells into the mammary fat pad of Balb/c mice and treating established tumors with ADM (Fig. 1A). Then, we isolated primary TECs, CAFs, and TAMs from treatment-naïve and ADM-exposed tumors (Supplementary Fig. 1), and tested their ability to protect 4T1 cells in co-culture across different ADM concentrations (0, 1, 2, 5, 10 µM). Consistent with similar notions of prior findings [27–29], all three stromal components conferred chemoresistance to 4T1 cells compared with monoculture controls (Fig. 1B-D). Strikingly, TAMs derived from ADM-treated tumors exhibited an additional and stronger protective effect than TAMs from treatment-naïve tumors, suggesting that chemotherapy reprograms TAMs toward a pro-resistance phenotype (Fig. 1D).

To directly assess the contribution of macrophages in vivo, we depleted macrophages with Clodronate liposomes [30]. Clodronate liposomes were administered twice weekly for two weeks before tumor inoculation and then weekly for four additional weeks (Fig. 1E). Clodronate liposome treatment markedly reduced F4/80+ macrophage infiltration in orthotopic tumors, indicating effective macrophage depletion (Supplementary Fig. 2). While Clodronate liposomes alone produced a modest reduction in tumor growth, combining macrophage depletion with ADM markedly suppressed tumor growth and showed additional effect (Fig. 1F). Collectively, these data indicate that TAMs, especially after chemotherapy, are critical mediators of TNBC chemoresistance.

Senescence-like TAMs are induced upon chemotherapy in vivo
To evaluate chemotherapy-induced remodeling of TAMs, we integrated human and mouse datasets with targeted validation. By utilizing a prior LC-MS/MS dataset generated by Liu et al. [31], which profiles CD163+ macrophages from human breast cancer samples collected before and after chemotherapy (Fig. 2A), we performed GSEA of GO biological processes and revealed no significant changes in macrophage activation, phagocytosis, or differentiation signatures post-chemotherapy (Fig. 2B). Instead, post-chemotherapy CD163+ TAMs showed significant enrichment of cellular senescence and SASP pathways (Fig. 2B), indicating the presence of senescence-like TAMs after chemotherapy. To confirm this phenomenon in vivo, we further adopted the dataset [31] and examined TAMs from PyMT mice with or without ADM treatment. Interestingly, we noticed a concordant enrichment of the cellular senescence pathway in ADM-treated TAMs (Fig. 2C).

We further validated senescence in TAMs by multiple orthogonal approaches. First, immunofluorescence analysis demonstrated a higher proportion of p16- or p21-positive F4/80+ macrophages in ADM-exposed tumors versus treatment-naïve controls (Fig. 2D). Then, real-time PCR (Fig. 2E) and Western blot (Fig. 2F) analyses revealed upregulation of canonical senescence markers, Cdkn2a (p16) and Cdkn1a (p21) in TAMs isolated from ADM-treated tumors. Finally, SASP components including IL-1α, IL-6, CXCL1, and TNF-α were markedly elevated in ADM-exposed TAMs (Fig. 2G). We also isolated TAMs from orthotopic 4T1 tumors after docetaxel treatment (DTX, 10 mg/kg, twice weekly) (Supplementary Fig. 3A). Consistently, DTX induced senescence-like phenotypes in TAMs (Supplementary Fig. 3B and 3 C). Together, these human and mouse data support a model in which chemotherapy induces senescence-like TAMs in TNBC.

Removal of senescence-like cells is associated improved chemotherapy efficiency
Given that ADM markedly induces senescence-like in TAMs and drives SASP secretion, we hypothesized that selectively eliminating senescence-like cells would enhance chemotherapy efficacy. To test this, we used ABT263 (Navitoclax), a senolytic Bcl-2 family inhibitor known to target senescent cells [32]. One week after ADM administration, ABT263 treatment was initiated (Fig. 3A) and continued for 2 weeks. ABT263 was effective to remove the senescent cells, as supported by SA-β-Gal staining (Fig. 3B). ABT263 alone produced only a modest reduction in tumor growth, whereas the combination of ADM plus ABT263 yielded a substantially greater inhibition of tumor growth (Fig. 3C). Moreover, we employed a second senolytic strategy using dasatinib plus quercetin (DQ) (Fig. 3D). Dasatinib inhibits Src family kinases and BCR-ABL, while quercetin targets multiple pro-survival pathways; together, DQ disrupts senescent-cell-associated signaling, leading to selective clearance of senescent cells [33, 34]. DQ treatment significantly reduced the number of senescent cells (Fig. 3E). Notably, DQ treatment enhanced chemotherapy efficacy in the orthotopic 4T1 tumor model (Fig. 3F). Collectively, these observations reveal a strong association between senescent cell depletion and diminished tumor growth, while acknowledging that the clearance of senescent tumor cells could also contribute to the anti-tumor outcome.

IL-6, a key factor of SASP, mediates chemoresistance in TNBC
Senescent cells exert their effects through SASP factors [35]. To interrogate the functional relevance of SASP in this context, we collected conditioned media (CM) from macrophages isolated from ADM-exposed tumors, DTX-exposed tumors and from treatment-naïve controls, and used these CMs to treat 4T1 cells in the presence of different concentrations of ADM and DTX. The results showed that ADM-exposed macrophage CM conferred a stronger protective effect against ADM-induced cytotoxicity than CM from control macrophages. Building on this, we neutralized SASP factors in ADM CM using corresponding neutralizing antibodies, including anti-IL-1α, anti-IL-6, anti-CXCL1, and anti-TNF-α, and found that IL-6 blockade largely abrogated the protective effect of ADM CM or DTX CM (Fig. 4A and Supplementary Fig. 4A). Therefore, we further focused on examination of IL-6. Consistently, stimulation of 4T1 cells with recombinant murine IL-6 similarly mitigated ADM-or DTX-induced cell death (Fig. 4B and Supplementary Fig. 4B). In primary TECs, CAFs, and TAMs isolated from treatment-naïve and ADM-exposed tumors, ELISA results showed that TAMs are the predominant source of IL-6 under the conditions tested (Fig. 4C).

Subsequently, we evaluated whether combining IL-6 blockade with ADM could improve therapeutic outcomes (Fig. 4D). IL-6 neutralizing antibodies, when combined with ADM, significantly enhanced anti-tumor efficacy compared with ADM alone (Fig. 4E). Although extensive prior work has linked IL-6 to TNBC chemoresistance [13, 36], we did not attempt to recapitulate those studies here. Instead, we focused on mechanistic readouts within the ADM and DTX context. We further observed that ADM CM or DTX CM induces the expression of drug resistance- and cancer stemness-associated genes (Abcb1, Abcc1, Cd44, Aldh1a1, and Prom1) [37, 38] in 4T1 cells, and that neutralizing IL-6 or blocking IL-6R with tocilizumab (10 µg/ml) attenuates these ADM or DTX CM-induced effects (Fig. 4F and Supplementary Fig. 4C). Given that the IL-6/IL-6R axis signals via STAT3 [39], inhibition of STAT3 with 5 µM Stattic yielded a similar reversal of ADM or DTX CM-driven transcriptional effects (Fig. 4F and Supplementary Fig. 4C). Collectively, these data identify IL-6 as a key SASP-derived mediator of chemoresistance in TNBC, likely acting through the IL-6R/STAT3 axis to drive a program of drug resistance and stemness gene expression.

Senescence-like TAMs in clinical specimens
To enhance the clinical relevance of our findings, we evaluated senescence-like TAMs in treatment-naïve TNBC and post-neoadjuvant chemotherapy samples. Using multiplex immunostaining to detect macrophage marker CD163 in combination with senescence marker p16, we quantified TAMs across all the tissue regions per sample and observed that p16+CD163+ TAMs were more frequently detected in chemotherapy-treated tumors than in treatment-naïve TNBC (Fig. 5A). Moreover, among chemotherapy-treated cases, p16+CD163+ TAMs were significantly enriched in progressive disease compared with those achieving complete or partial remission (Fig. 5B). Given that treatment-naïve TNBC samples are obtained by core needle biopsy, whereas post-treatment specimens are derived from surgical resections, acquisition-related differences may influence the measured abundance of senescence-like TAMs. As a secondary line of evidence, we employed a senescence-like TAM gene signature to assess its association with drug response. In the TCGA-BRCA cohort, progressive disease exhibited higher signature scores than patients achieving complete or partial remission (Fig. 5C). In BRCA patients receiving Carboplatin, higher senescence-like TAM gene signature scores were associated with poorer overall survival (Fig. 5D), although the difference was not statistically significant (p = 0.088). Collectively, these data indicate that chemotherapy can promote senescence-like TAM phenotypes in human TNBC and that senescence-like TAMs may contribute to therapeutic resistance and disease progression.

Discussion
In this study, using integrated approaches spanning in vivo models, clinical specimens, and mechanistic investigation, we demonstrate that chemotherapy induces senescence in TAMs and senescence-like TAM-derived IL-6 activates the IL-6R/STAT3 axis in TNBC cells (Fig. 6). Targeted elimination of senescence-like TAMs or blockade of IL-6 signaling significantly enhances chemotherapy efficacy in vivo. Senescence-like TAMs accumulate in chemotherapy-treated TNBC patients and correlate with poorer therapeutic response. Collectively, these findings reveal a previously unrecognized mechanism whereby chemotherapy, intended to eliminate tumor cells, inadvertently reprograms stromal TAMs into senescence-like, pro-resistance entities that undermine therapeutic success.

Our work positions senescence-like TAMs within the broader landscape of stromal-mediated chemoresistance. While prior studies implicated CAFs or CSCs in TNBC treatment failure, the transformation of TAMs into a senescence-like state emerges as a critical, therapy-induced adaptation [40, 41]. Senescent TAMs exhibit profound functional alterations, such as reduced phagocytosis, metabolic rewiring, and most notably, a robust SASP dominated by IL-6 [26, 42, 43]. Our multiplexed IF data from human TNBC samples underscore the clinical relevance: senescence-like TAM enrichment post-chemotherapy directly correlates with poor therapeutic outcomes, suggesting their utility as biomarkers of resistance. Macrophages surrounding tumors can acquire a senescence-like state and express Arginase-1, an immunosuppressive enzyme that dampens T-cell response [44]. Whether senescence-like TAMs promote chemoresistance through immune-related pathways, either alone or in conjunction with SASP-mediated nonimmune effects on tumor cells, warrants further investigation. Notably, endogenous senescent CAFs are present in the TME, and depletion of these senescent stromal cells slows tumor progression and improves responsiveness to chemotherapy [21]. Moreover, TSPAN8-positive myofibroblastic CAFs exhibit senescence-like phenotypes and promote stem-like properties in neighboring breast cancer cells by secreting SASP-related cytokines IL-6 and IL-8, thereby promoting chemoresistance [27]. Together, our findings and these reports suggest that stromal senescence remodels the TME into an immunosuppressive, chemoresistance, and metastasis-permissive milieu.
SASP-dominated output tends to promote a pro-tumor microenvironment: it drives cancer cell proliferation, epithelial-to-mesenchymal transition, stemness, angiogenesis, and matrix remodeling, and it can dampen effective anti-tumor immune responses. Because SASP factors provide paracrine pro-survival signals, senescence-like TAMs can contribute to chemoresistance by enhancing DNA repair, survival signaling, and stress adaptation in tumor cells, thereby supporting tumor growth. In this study, IL-6 emerged as the non-redundant mediator of chemoresistance. IL-6 engages the IL-6R/STAT3 axis on TNBC cells, activating a transcriptional program that simultaneously enhances drug efflux (ABCB1/ABCC1) and reinforces stemness (CD44/ALDH1A1). While we focused on IL-6, we cannot exclude contributions from other SASP components. For instance, CXCL-1 may recruit immunosuppressive CD11b(+)Gr1(+) myeloid cells and IL-8 could promote the formation of chemotherapy-induced neutrophil extracellular traps to drive chemoresistance [45, 46]. Future studies should delineate whether SASP factors act synergistically or hierarchically.
Our study reveals potential actionable strategies to overcome TAM-driven resistance. ABT263 selectively eliminated senescence-like TAMs, restored chemosensitivity, and improved tumor control. Consistently, depletion of senescent macrophages from the TME exerts anti-tumor effects in lung cancer [20, 22]. These studies highlight senolysis as a viable approach to disrupt therapy-induced stromal adaptation. Moreover, neutralizing IL-6 or its receptor abrogated SASP-mediated protection, confirming the druggability of the IL-6/IL-6R axis. Both strategies synergize with chemotherapy, suggesting their potential in clinical regimens. Importantly, IL-6 inhibitors (e.g., tocilizumab) are FDA-approved for autoimmune disorders, enabling rapid repurposing [47]. However, key questions remain unanswered: Do senescence-like TAMs interact with other stromal cells (e.g., CAFs, endothelial cells) to amplify resistance? How do senescence-like TAMs evade immune clearance? Addressing these could uncover combinatorial opportunities.
While our study establishes senescence-like TAMs as key mediators of TNBC chemoresistance, several aspects warrant further investigation. First, single-cell analyses are needed to define senescence-like TAM subsets and their spatial distribution within TNBC niches. Second, whether and how senescence-like TAMs induce immunosuppression remain unexplored. Third, chemotherapy may induce parallel senescence in cancer cells or CAFs; their crosstalk with senescence-like TAMs could create resistance networks. Regarding senescence-like TAMs, the mechanism of induction remains undefined. It is unclear whether chemotherapy directly drives macrophage senescence via DNA damage/ROS/telomere stress or whether senescence is primarily induced by indirect signals from stressed tumor cells or microenvironmental factors.
In conclusion, we establish a treatment-compromising circuit in TNBC. The induction of TAM senescence enables IL-6 signaling through tumor cell IL-6R/STAT3, resulting in chemoresistance and stemness acquisition. Targeting senescence-like TAMs via senolysis or IL-6 inhibition restores chemosensitivity and improves outcomes in preclinical models. Given the poor prognosis of TNBC and the clinical availability of IL-6 pathway inhibitors, our work suggests stromal senescence as a tractable therapeutic target and paves the way for combination trials aimed at reprogramming the TME to overcome resistance. Moreover, the senescence-like TAM signature and senescence markers in TAMs could serve as prognostic or predictive biomarkers to identify patients at risk of chemoresistance and to guide treatment intensification or modification.

Supplementary Information
Below is the link to the electronic supplementary material.