Patients and samples

Primary invasive breast carcinoma tissues were obtained from 259 patients at Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, between 2006 and 2019. All patients received neoadjuvant chemotherapy after a definitive diagnosis using puncture specimens, and the chemotherapy regimens were as follows: four cycles of AC (doxorubicin 60 mg/m2 plus cyclophosphamide 600 mg/m2) every 3 weeks and paclitaxel (80 mg/m2) weekly for 12 weeks or four cycles of TC (docetaxel 75 mg/m2 plus cyclophosphamide 600 mg/m2) every three weeks. RECIST (Response Evaluation Criteria in Solid Tumors) was used to evaluate therapeutic efficacy. Patients with a complete response (CR) and a partial response (PR) were classified as chemotherapy-sensitive, while those with stable disease (SD) and progressive disease (PD) were classified as chemotherapy-resistant. The scRNA-seq data from 50 breast cancer specimens and 13 normal breast samples were obtained from GEO: GSE161529 and the Pancancer TME Blueprint (https://lambrechtslab.sites.vib.be/en/pan-cancer-blueprint-tumour-microenvironment-0).

Primary cell and tissue culture

Primary normal breast fibroblasts (NBFs) were isolated from reduction mammaplasty samples, and primary cancer-associated fibroblasts (CAFs), tumor-associated macrophages (TAMs) and T lymphocytes were isolated from treatment-naive breast carcinoma samples obtained during breast cancer resection. Briefly, tissues were cut into fragments of approximately 1 mm3 and then digested by enzymatic hydrolysis (DMEM supplemented with 10% FBS, 1.5 mg/mL collagenase type I, 1.5 mg/mL collagenase type III and 1.5 mg/mL hyaluronidase) at 37 °C with gentle agitation for the indicated time (2.5 h for fibroblasts and 1 h for TAMs). The dissociated tissues were resuspended and filtered through a 70-μm cell strainer to obtain a cell suspension, which was centrifuged at 250 x g for 5 min to acquire the stromal fraction. Fibroblasts were further purified using anti-fibroblast microbeads (Miltenyi Biotec, Cat.No.130-050-601). To isolate TAMs and T lymphocytes, the primary cell suspension was centrifuged at 400 × g for 5 min, and then, CD14-positive cells or T cells were further purified by using CD14 microbeads (Miltenyi Biotec, Cat.No.130-050-201) or a Pan T Cell Isolation Kit (Miltenyi Biotec, Cat.No.130-096-535). After verification [3, 40, 42], the isolated primary fibroblasts and TAMs were cultured in DMEM supplemented with 10% FBS, and T lymphocytes were cultured in RPMI-1640 supplemented with 10% FBS. Primary fibroblasts at passages 1–5 were used in our experiments. In some experiments, fibroblasts were cultured with rCCL18 (20 ng/mL, # 300-34, PeproTech), rTGF-β (10 ng/mL, # 100-21, PeproTech) or the indicated conditioned medium (CM) for 7 days. The concentration of CCL18 used to treat cells was set based on the concentration of CCL18 that secreted by TAMs detected by us or others [29, 42, 62]. For gene silencing, fibroblasts were transduced with LV3 lentivirus carrying target gene-specific shRNA constructs (Supplementary Table 1). Lentivirus was provided by Gene Pharma Inc. (Shanghai, China). To inhibit specific signaling pathways, fibroblasts were pretreated with vehicle (DMSO), JSH-23 (6 mM, Selleck, #S7351) or Bay11-7082 (2 mM, Selleck, #S2913). TAMs were pretreated with IgG, TGF-β neutralizing antibody (10 µg/mL, R&D, #MAB1835) or CCL18 neutralizing antibody (10 µg/mL, R&D, #AF394). Tumor CM was supplemented with CCL18 neutralizing antibody, IL6 neutralizing antibody (10 µg/mL, BD, #554543) or IL8 neutralizing antibody (10 µg/mL, BD, #554726) prior to the experiments.

To harvest tumor CM, treatment-naive breast carcinoma samples were cut into fragments of approximately 1 mm3 and then placed on top of sponges in a 12-well cell culture dish. Then, all the tissues were cultured with explant medium (DMEM supplemented with 10% FBS, 10 µg/mL insulin and 5 µg/mL hydrocortisone) at 37 °C and 5% CO2 for 36 h.

All samples were collected from patients who provided written informed consent, and the related protocols were reviewed and approved by the Ethics Committee of Sun Yat-Sen Memorial Hospital.

Western blot

Cells were lysed in RIPA buffer (Millipore) supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific, #78444), and the proteins were collected by centrifugation at 4 °C. A Pierce BCA Protein Assay kit (Cat# 23225, Thermo Fisher Scientific) was used for protein quantification. Equivalent amounts of protein from each sample were resolved by 10% SDS–PAGE and then transferred to PVDF membranes, which were then probed with primary antibodies against α-SMA (1:1000, R&D, #MAB1420), FAP (1:500, R&D, #AF3715), CD10(1:1000, Abcam, #ab255609), GPR77((1:1000, Abcam, #ab96808), caspase-3 (1:1000, CST, #9662), cleaved caspase-3 (1:1000, CST, #9664), PARP (1:1000, CST, #9532), cleaved PARP (1:1000, CST, #5625), phospho-IKKα/β (1:500, CST, #2697), IKKα (1:1000, CST, #2682), IKKβ (1:1000, CST, #2678), SMAD2/3 (1:1000, CST, #8685), phospho-SMAD2 (1:1000, CST, #18338), phospho-SMAD3 (1:1000, CST, #9520), PITPNM3 (1:1000, Novus Biologicals, #NBP1-31070), and GAPDH (1:1000, Proteintech, #HRP-60004), followed by incubation with an HRP-linked secondary antibody (CST). The antigen–antibody reactions were visualized by chemiluminescence-based immunodetection (ECL, Thermo).

Flow cytometry

To detect cell surface markers, cells suspended in PBS containing 1% FBS and 2 mM EDTA were treated with FcR blocking reagent (Miltenyi Biotec, #130-059-901) and then incubated with specific fluorescence-linked antibodies against CD10 (BioLegend, #312204), GPR77 (BioLegend, #342406), CD44 (BD Biosciences, #555478), CD24 (BD Biosciences, #55542), CCR6 (BioLegend, #353409) and CCR8 (BioLegend, #360603) for 30 min at 4 °C. To evaluate cell surface PITPNM3 expression, cells were incubated at 4 °C with anti-PITPNM3 primary antibody (Novus Biologicals, #NBP1-31070) for 60 min and then incubated with a fluorescein-conjugated secondary anti-rabbit IgG antibody (Jackson ImmunoResearch, Bar Harbor, ME) at 4 °C for 45 min. In addition, to detect ALDH1 activity, an ALDEFLUOR kit was used according to the manufacturer’s instructions (Stem Cell Technologies, #01700). To detect apoptosis, tumor cells with indicated treatment were dissociated using 0.25% trypsin-EDTA and then harvested by centrifugation. Cell apoptosis was assessed by using the Annexin V Apoptosis Detection Kit (eBioscience, #88-8005-74). Briefly, cells were incubated with 5 µL FITC-conjugated Annexin V antibody in 100 μL binding buffer at room temperature for 15 min. Then, the cells were rinsed and resuspended in 200 μL buffer supplemented with 5 μL propidium iodide. Specimens were subsequently analyzed on a BD Accuri C6 or Thermo Attune NxT flow cytometer.

Immunofluorescence

For immunostaining of paraffin sections, the sections were deparaffinized, and antigen retrieval was then performed in 0.01 M citrate buffer (pH 6.0). For immunostaining of cultured cells, the cells were fixed with paraformaldehyde and then permeabilized with 0.1% Triton X-100 on ice for 15 min. Nonspecific antigen epitope binding was blocked by incubation with phosphate buffer containing 5% BSA for 1 h. Next, sections or cells were probed overnight at 4 °C with specific primary antibodies, including goat anti-human α-SMA (Abcam, # ab21027, 1:100), rabbit anti-human α-SMA (Abcam, # ab124964, 1:100), mouse anti-human ALDH1 (R&D, # AF5869, 1:100), rabbit anti-human CD10 (Abcam, # ab73409, 1:30), mouse anti-human GPR77 (BioLegend, #342402, 1:30), sheep anti-human FAP (R&D, # AF3715, 1:100), rabbit anti-human CD68 (Abcam, # ab213363, 1:100), goat anti-human CCL18 (R&D, # AF394, 1:100), rabbit anti-human p65 (CST, # 8242, 1:50), rabbit anti-human ac-p65 (Abcam, # ab19870, 1:50), goat anti-human IL6 (R&D, # AF-206, 10 µg/mL) and mouse anti-human IL8 (R&D, # MAB208, 20 µg/mL). Then, antigen–antibody binding was visualized by using Alexa Fluor-conjugated secondary antibodies (Invitrogen) according to the manufacturer’s instructions. DAPI (# D3571, Thermo Fisher Scientific) was used for nuclear counterstaining, and laser scanning confocal microscopy (LSM780, Zeiss) was used for imaging. Staining cells quantification was counted in at least five fields per section and the mean of counts was used for statistical analysis.

Cytokine antibody arrays

The cytokine profile of tissue culture medium was determined by using a Human Cytokine Antibody Array V kit (RayBiotech). Briefly, membrane arrays were incubated with 1 mL tissue culture medium overnight at 4 °C, and the next day, biotin-linked antibodies were used to create an antibody–antigen–antibody sandwich. Next, an HRP-conjugated secondary antibody was used to amplify the signal, which was detected by incubation with a chemiluminescent substrate and X-ray exposure. Quantitative analysis was conducted with Array Vision Evaluation 8.0 (GE Healthcare Life Science).

ELISA

Fresh breast cancer puncture specimens or cells exposed to the indicated treatment were cultured in DMEM supplemented with 10% FBS for 24 h, and the supernatant was collected by centrifugation for subsequent ELISA analysis. IL-6 (eBioscience Cat# 88-7066-86), IL-8 (eBioscience Cat# 88-8086-86) and CCL18 (R&D Systems Cat #DCL180B) ELISA kits were used according to the manufacturers’ instructions.

Immunohistochemistry

Paraffin sections were deparaffinized, and then, antigen retrieval was performed in 0.01 M citrate buffer (pH 6.0). The sections were then incubated overnight at 4 °C with a primary antibody against CCL18 (1:100, R&D, #AF394), and the signal was amplified with an HRP-conjugated secondary antibody and visualized using DAB (Dako). Upright metallurgical microscope (BX53, Olympus) was used for imaging. Staining cells quantification was counted in at least five fields per section and the mean of counts was used for statistical analysis and the calculation and the statistical analysis was performed by two independent researchers.

scRNA-seq transcriptome data processing

Multiple breast samples from GEO series GSE161529 and the Pancancer TME Blueprint (https://lambrechtslab.sites.vib.be/en/pan-cancer-blueprint-tumour-microenvironment-0) were combined using the Seurat analysis package. Conservative quality control cutoffs were set according to the number of genes/cell (>500) and the percentage of mitochondrial unique molecular identifier (UMI) counts (<20%). Unless otherwise stated, the cluster resolution was set to 0.5 for t-distributed stochastic neighbor embedding (t-SNE). EPCAM-negative nonepithelial cells were subjected to further regrouping, and cell clusters in the microenvironment were annotated based on canonical cell type markers for fibroblasts (PDGFRA, PDGFRB, ACTA2, PDPN and FAP), macrophages (CD14 and CD68), endothelial cells (VWF), T cells (CD3D, CD8A and CD4), B cells (CD79A) and pericytes (NG2). The positive threshold was set as log expression level >0.5 for subsequent analysis.

Collagen gel contraction assay

The collagen gel contraction assay was a widely-used and classical functional assay used to examine the activation status of fibroblasts [35]. To evaluate NBF function, NBFs were treated with indicated cytokine or cocultured with primary TAMs in an indirect transwell coculture system. Then, NBFs were harvested by trypsin digestion and a mixture of 1.0 × 104 fibroblasts and 200 µl collagen mix was plated in 24-well culture dishes. The gel was allowed to polymerize for 30 min at 37 °C, after which 500 μL medium was added to the wells, and a sterile pipette tip was used to detach the gels from the well walls. The wells were imaged after 24 h to measure the gel dimensions compared to the well diameter using ImageJ software.

RT–qPCR

Total RNA was extracted from cultured cells by using TRIzol Reagent (Invitrogen). Then, quantitative reverse transcription PCR (qRT–PCR) was performed using a SYBR Premix Ex Taq kit (TaKaRa, Japan) according to the manufacturer’s instructions, and data were collected and analyzed with a LightCycler 480 instrument (Roche). The primer sequences are listed in Supplementary Table 2.

Co-culture experiments

MCF-7, BT474, BT549 and SKBR3 breast cancer cells were obtained from American Type Culture Collection (ATCC) and tested regularly for Mycoplasma infection. A six-well Transwell apparatus with a 0.4 µm pore size (Corning Incorporated) was used for the coculture experiments. A total of 5 × 104 cancer cells were seeded in each upper chamber, and 5 × 104 pretreated fibroblasts were seeded in each lower chamber. The cocultured cells were passaged when they reached 90% confluence.

MTT assay

A 3-(4,5-dimethylthiazol-2-yl)22,5-diphenyltetrazolium bromide (MTT, Sigma) assay was used to determine cell viability. Briefly, 1 × 103 tumor cells/well were plated in 96-well plates and then treated with the indicated chemotherapy agent for the indicated time. Then, MTT solution was added, and the plates were incubated at 37 °C for 4 h. Thereafter, the media were removed, and the formazan crystals were dissolved in DMSO (150 mL/well). Then, the absorbance was measured at 540 nm using an Infinite F500 (Tecan).

Sphere formation assay

Tumor cells (1 × 103 cells/mL) were cultured in DMEM-F12 (GIBCO) supplemented with 20 ng/mL EGF, B27 (1:50, Invitrogen), 4 mg/mL insulin, and 0.4% BSA in 6-well ultralow adhesion plates for two weeks. Then, cell spheres with a diameter >75 mm were counted.

Fluorescence-activated cell sorting

Using a BD Influx flow cytometer, different subsets of CAFs (CD10+GPR77+ and CD10+GPR77+-depleted) were selected by fluorescence-activated cell sorting (FACS). Before cell sorting, primary CAFs were resuspended in PBS containing 1% FBS and incubated with antibodies against CD10 and GPR77 for 30 min at 4 °C. The purity of the sorted populations was verified by flow cytometry.

Luciferase reporter assay

pNF-κB-Luc, pRL-TK, and pTAL-Luc vectors (Promega Madison, WI) were transfected into cells using Lipofectamine 3000 (Invitrogen, Carlsbad, CA). The luciferase activity of transfected cells treated as indicated was determined with the Dual-Luciferase Reporter Assay System (Promega). Firefly luciferase activity was normalized to Renilla luciferase activity for each sample.

TUNEL assay

Paraffin sections were deparaffinized, and then, cell death was detected by using the In Situ Cell Death Detection Kit POD (Cat# 11684817910, Roche) according to the manufacturer’s instructions.

Coinjection animal experiments

Sample size and number of animals were chosen based on our prior experience [3] to achieve statistical significance. No mice were excluded from any analyses and no blind analysis was performed. All procedures were approved by the Institutional Review Boards and Animal Care and Use Committees of Sun Yat-Sen University.

Tumorigenesis

Female NOD/SCID mice were randomly selected to each group (12 mice per group). Model A: serial concentrations of MCF-7 cells alone or mixed with primary NBFs were coinjected into the mammary fat pads of 6-weeks-old NOD/SCID mice at a ratio of 1:3 as previously described [1]. After cell injection, the mice received biweekly intratumoral injections of PBS or rCCL18 (0.1 mg/kg, Cat# 300-34, PeproTech). Tumor formation was assessed weekly for up to 3 months.

Chemoresistance

Female NOD/SCID mice were randomly selected to each group (8 mice per group). The MCF-7 xenograft mouse model was generated as described above (model A) or generated by co-inoculating MCF-7 cells with or without NBFs and TAMs into the mammary fat pads of NOD/SCID mice (model B). In model B, MCF-7 cells were treated with 10 ng/ml TNF-α for 2 weeks as described in our previous study [42] and PBMC were cultured with the CM of treated MCF-7 to induce TAMs before co-inoculating. After cell injection, the mice in model A received PBS or rCCL18 as described above and the mice in model B received IgG or anti-CCL18 neutralizing antibody (2 mg/kg, R&D, #AF394) weekly by intratumoral injecting. When the tumors reached 3 mm in diameter, 10 mg/kg docetaxel was administered by intraperitoneal injection once per week for 6 weeks.

The xenografts were harvested, fixed in 10% formaldehyde solution, and embedded in paraffin for further analysis. Tumor size was assessed weekly by calipers, and tumor volume was determined according to the standard formula: Volume (mm3) = (length × height2)/2.

Statistical analysis

The detailed statistical analysis results are indicated in the figure legends or methods. Unless otherwise described in the figure legends or methods, the statistical analyses were performed using GraphPad Prism 8.0. Pearson correlation was used to assess the association between the infiltration of CCL18+ TAMs and CD10+GPR77+ CAFs. The number of CCL18+ TAMs and CD10+GPR77+ CAFs in 259 cases coincided with normal distribution. All in vitro experiments were repeated at least three independent times, and the specific number of experiments is indicated in the figure legends. The two-tailed Student’s t test was used to identify significant differences between two groups, and the two-tailed one-way ANOVA with Dunnett’s test was used to determine significant differences in experiments with more than two groups. Estimating of variation within each group was performed before comparation. The variance similar between groups was statistically compared. All bar graphs show means and error bars (indicating the standard error of the mean (SEM) or the standard deviation (SD)), as mentioned in each figure legend. P < 0.05 was considered to indicate statistical significance. No blind analysis was performed in this study.