Anti-inflammatory signaling by mammary tumor cells mediates prometastatic macrophage polarization in an innovative intraductal mouse model for triple-negative breast cancer

The BALB/c-derived 4T1 mammary tumor cell line used in this study constitutively expresses the firefly luciferase gene and was a kind gift from Prof. Clare Isacke (Breakthrough Breast Cancer Research Centre, London, UK). This tumor cell line resembles the aggressive phenotype and metastasis seen in human TNBC (estrogen receptor (ER)-negative, progesterone receptor (PR)-negative and human epidermal growth factor receptor 2 (HER2)-negative) [15, 16]. The BALB/c-derived RAW264.7 macrophage cell line was a kind gift from Prof. Rudi Beyaert (Unit of Molecular Signal Transduction in Inflammation, Inflammation Research Center, Ghent University-VIB, Ghent, Belgium). Both cell lines were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 100 U/ml penicillin and 100 μg/ml streptomycin (Thermo Fisher Scientific, Waltham, MA, USA) in culture flasks. Harvesting of cultured 4T1 cells was performed using 0.25% trypsin- ethylenediaminetetraacetic acid (EDTA) (Thermo Fisher Scientific), whereas RAW264.7 macrophages were harvested using a cell scraper. The harvested cells were subsequently washed through centrifugation (805 g for 5 min) and the cell pellets were resuspended in phosphate buffered saline (PBS). Cell numbers were determined through counting using a Bürker chamber.

For preliminary in vitro experiments, 4T1 mammary tumor cells and RAW264.7 macrophages were cultured either alone (5 × 10⁵ cells in mono-culture) or together (5 × 10⁵ of each cell type in co-culture) supplemented with 1 ml of cell culture medium per well in 24 well plates. The cell cultures were incubated (37 °C, 5% CO₂) for 24 h (to examine CHI3L1 and LCN2 secretion) or 96 h (to examine RAW264.7 macrophage polarization) with daily change of the cell culture medium. The harvested cell culture media were spun down (17,000 g) for 10 min to remove cellular debris for further analyses. Cells from 3 wells of 96 h RAW264.7 mono- and 4T1 + RAW264.7 co-cultures were harvested using a cell scraper, pooled and washed through centrifugation (805 g for 5 min) for subsequent flow cytometric analysis.

Harvested and pelleted cells from RAW264.7 mono- and 4T1 + RAW264.7 co-cultures were suspended in 2.5 ml FACS buffer (containing PBS, 1% bovine serum albumin (BSA), 2.5 mM EDTA and 0.01% sodium azide) and 100 μl of the cell suspension was plated in a well of a 96 well plate for counting through flow cytometry (Analis, Cytoflex). Propidium iodide (PI, 2 μl at 50 μg/ml) was also added to the well to evaluate the viability of the cells. Remaining cell suspensions were plated at 100 μl per well in a 96 well plate and the well plate was centrifuged to pellet the cells (805 g for 5 min). To block Fc receptors found on the RAW264.7 macrophages, cell pellets were subsequently resuspended in FcR blocking reagent (1:10 diluted in FACS buffer; Miltenyi Biotec, Leiden, Netherlands) and incubated for 10 min at 2–8 °C. Following centrifugation, cell pellets derived from 4T1 + RAW264.7 co-cultures were stained for 30 min at 2–8 °C with APC-labeled anti-F4/80 (diluted 1:20 in FACS buffer; clone CI:A3–1; Bio-Rad, CA, USA) to distinguish RAW264.7 macrophages from 4T1 tumor cells. This staining was not performed on cells derived from RAW264.7 mono-cultures as no distinction is needed between RAW264.7 macrophages and 4T1 tumor cells. To allow intracellular staining, the pelleted cells were fixed using BD Cytofix/Cytoperm solution (Becton Dickinson, Erembodegem, Belgium) for 20 min at 2–8 °C and permeabilized afterwards by washing twice in 1× BD Perm/Wash Buffer (Becton Dickinson). Cell pellets derived from RAW264.7 mono- and 4T1 + RAW264.7 co-cultures were stained for 30 min at 2–8 °C with PE-labeled anti-IL-12 (diluted 1:20 in 1× BD Perm/Wash Buffer; clone B211220; BioLegend, CA, USA) or anti-TGF-β1 (diluted 1:40 in 1× BD Perm/Wash Buffer; clone TW7-16B4; BioLegend). Istoype-matched and autofluorescence controls were also included for analyses. Following cellular stainings, cell pellets were washed twice with 1× BD Perm/Wash Buffer prior to analysis with a flow cytometer. A maximum of 5% overlap with the isotype control signal was allowed when setting the gates to identify positive F4/80, IL-12 and TGF-β1 signals.

Intraductal inoculations were conducted in the third mammary gland pair of female lactating BALB/c mice under inhalation anesthesia (induction: 2–3% isolflurane supplemented with oxygen (O₂); maintenance: 1–1.5% isoflurane supplemented with O₂) in combination with intraperitoneal (i.p.) administration of buprenorphine (long-acting analgesic, 10 μg/kg Vetergesic, Patheon UK Ltd., Swindon, UK). To attain full lactation, female BALB/c mice of 8–13 weeks (w) were mated with male BALB/c mice of 10 w and pups were weaned 12–14 days following delivery. Mice were inoculated through the mammary teat canal 1 h after weaning using a 32-gauge blunt needle with either 5 × 10⁴ 4T1 cells, 5 × 10⁴ RAW264.7 macrophages or 5 × 10⁴ of both 4T1 cells and RAW264.7 macrophages, suspended in a 100 μl mixture of 1:10 PBS and Matrigel^® (Corning, Bedford, MA, USA). All materials used for inoculation were kept cold to prevent clotting of the Matrigel^®.

Intraductally inoculated mice were screened weekly through in vivo bioluminescence imaging using the IVIS lumina II system (PerkinElmer, Zaventem, Belgium) to monitor growth of 4T1 firefly luciferase-expressing primary tumors. Mice were injected i.p. with 200 μl D-luciferin suspended in PBS (2 mg/100 μl; Gold Biotechnology, St. Louis, MO) and approximately 10 min later images were acquired under inhalation anesthesia with isoflurane. Metastases were detected through ex vivo bioluminescence imaging, which required euthanasia of the intraductally inoculated mice and subsequent harvesting of metastases-bearing organs. Mice were therefore sedated by i.p. injection of a mixture of 100 mg/kg ketamine (Ketamidor, Ecuphar nv/sa, Oostkamp, Belgium) and 10 mg/kg xylazine (Xylazini Hydrochloridum, Val d’Hony-Verdifarm, Beringen, Belgium) and then sacrificed through cervical dislocation. Axillary lymph nodes and lungs were isolated and screened for 4T1 metastases. Primary tumors and spleens were also isolated and weighed upon sacrification. Quantification of the 4T1-derived bioluminescent signals calculated by the living image analysis software 3.2 from in vivo and ex vivo images was performed by dividing the measured total flux with the selected area.

Primary tumors, mammary glands, and metastases-bearing axillary lymph nodes and lungs were isolated upon sacrifice of the mice, fixed in buffered 3.5% formaldehyde for 24 h at room temperature (RT) and embedded in paraffin wax. Sections (5 μm) were deparaffinized, hydrated and stained with hematoxylin and eosin (H&E). Sections were then dehydrated and mounted with a cover glass for evaluation.

For immunohistochemical stainings, sections of 2–3 μm were deparaffinized and antigen retrieval was performed with citrate buffer (10 mM tri-sodium citrate (Santa Cruz Biotechnology, Heidelberg, Germany)), pH 6, (for Ki67, carbonic anhydrase IX (CAIX), CD45, Ly6G, CD163, CD8a, CD31, F4/80) or with Tris-EDTA buffer (10 mM Tris, 1 mM EDTA (Thermo Fisher Scientific)), pH 9, (for cytokeratin 5) both supplemented with 0.05% Tween-20 (Sigma-Aldrich, Bornem, Belgium) at 95 °C under pressure for 30 min using a Decloaking Chamber NxGen (Biocare Medical, CA, USA). The slides were allowed to cool down to RT for 30 min, after which endogenous peroxidase activity was blocked with 3% H₂O₂ in methanol (for Ki67, CAIX, cytokeratin 5, CD45, Ly6G, CD163, CD31 and F4/80) or 0.6% H₂O₂ in methanol (for CD8a) for 10 min at RT. Serum-free protein block (Dako, Heverlee, Belgium) was applied for 10 min at RT to block non-specific binding sites. The slides were subsequently incubated with primary antibody diluted in Antibody Diluent (Dako) for 1 h at RT and secondary antibodies were incubated for 30 min at RT. Primary antibodies and dilutions used were: anti-Ki67 (1:50; clone SP6; Thermo Fisher Scientific); anti-CAIX (1:1000; NB100–417; Novus Biologicals, Littleton, CO, USA); anti-cytokeratin 5 (1:100; clone EP1601Y; Abcam, Cambridge, UK); anti-CD45 (1:1000; clone 30-F11; Thermo Fisher Scientific); anti-Ly6G (1:1000; clone 1A8; BioLegend); anti-CD163 (1:500; clone EPR19518; Abcam); anti-CD31 (1:2000; clone EPR17259; Abcam); anti-F4/80 (1:10; clone CI:A3–1; Bio-Rad); anti-CD8a (1:50; clone 4SM15; Thermo Fisher Scientific). Secondary antibodies used were: Rat-on Mouse HRP-Polymer (Biocare Medical) for CD45, Ly6G, F4/80 and CD8a; Dako EnVision+ Rabbit (Dako) for Ki67, CAIX, cytokeratin 5, CD163 and CD31. A 3,3′-diaminobenzidine (DAB)-containing buffer (generated by adding 1 drop of Dako DAB+ chromogen to 1 ml of Dako DAB+ substrate buffer) was applied for 10 min at RT, which created a brownish signal that was visualized as a positive stain on the slide following counterstaining in hematoxylin for 3–4 min at RT. All rinsing steps in between the incubation steps were performed by applying tris-buffered saline (TBS, Biocare Medical) 3 times for 2 min at RT. All incubation and rinsing steps from peroxidase activity blocking to DAB chromogen staining were performed in a closed microscope slide box (humidified with TBS-wetted tissue paper) on an orbital shaker at 20 rpm. Next to qualitative assessment, differences in immunohistochemical staining between the inoculation groups at each time point were quantitated with ImageJ through color deconvolution (3 color split) and automatic counting.

Primary tumors, mammary glands, spleens, axillary lymph nodes and lungs were isolated and homogenized. The homogenate was subsequently mixed with 300 μl lysis buffer supplemented with protease inhibitors (1% Nonidet P-40, 10 mM Tris-HCl at pH 7.4, 200 mM NaCl, 5 mM EDTA, 10% glycerol, 100 μM phenylmethylsulfonyl (PMSF), 1 mM oxidized L-glutathione (all from Sigma-Aldrich), 0.15 μM aprotinin and 2.1 μM leupeptin (Roche, Mannheim, Germany)). The suspensions were frozen overnight to allow their lysis and centrifuged twice at 17,000 g for 1 h at 4 °C the following day to precipitate the pellet. Protein concentration of the lysates was determined using the Bradford Protein Assay (BioRad, Hercules, CA) followed by spectrophotometrical measurement (Genesys 10S). Lysates were diluted to a concentration of 5 μg/μl with lysis buffer for further analysis. Serum was prepared from cardiac puncture harvested blood that was clotted through incubation at 37 °C for 1 h and then centrifuged at 17,000 g for 1 h at 4 °C.

A selected profile of cytokines (BAFF, G-CSF, IFN-γ, IL-1β, IL-4, IL-6, IL-10, MCP-1, MIP-2 and TNF-α) was quantified in lysates (50 μg protein), sera (diluted 1:4 in assay diluent) and culture media (undiluted) using Luminex Multiplex Assays (ProcartaPlex from Thermo Fisher Scientific) according to the manufacturer’s instructions. TGF-β1, MMP-9, VEGF, CHI3L1 and LCN2 levels were measured in lysates and sera using enzyme-linked immunosorbent assay (ELISA) according to the manufacturer’s instructions (TGF-β1: Mouse uncoated ELISA Kit, Thermo Fisher Scientific; MMP-9, VEGF, CHI3L1 and LCN2: Mouse Quantikine ELISA Kit, Biotechne, Minneapolis, MN, USA). A microplate reader was used to measure the absorbance at 450 nm and 550 nm. Readings at 550 nm were subtracted from those at 450 nm for correction purposes. A standard curve from recombinant mouse protein was constructed according to the manufacturer’s instructions to determine protein levels from the corrected optical densities using Deltasoft.

Primary tumor and mammary gland lysates were equally loaded at a concentration of 20 μg protein per lane on a 12% polyacrylamide gel (GE Healthcare, Buckinghamshire, UK). Proteins were separated and transferred to a 0.45 μm nitrocellulose membrane (Bio-Rad). The membrane was incubated for 1 h at RT with blocking buffer (TBS supplemented with 0.1% Tween-20 and 5% non-fat dry milk) and subsequently with a primary rabbit anti-mouse CD163 monoclonal antibody (clone EPR19518, Abcam, diluted 1:1000 in blocking buffer) overnight at 4 °C. The membrane was then washed 3 times (5 min for each step) at RT with blocking buffer and incubated with a secondary donkey anti-rabbit IgG horseradish peroxidase (HRP) conjugated polyclonal antibody (Thermo Fisher Scientific, diluted 1:100,000 in blocking buffer) for 1 h at RT. The membrane was again washed 3 times (5 min for each step) at RT with blocking buffer and 1 time with destilled H₂O (5 min) at RT before detection of the protein bands. For loading control, the membrane was incubated overnight at 4 °C with a primary rabbit anti-mouse glyceraldehyde 3-phosphate dehydrogenase (GAPDH) monoclonal antibody (clone EPR16891, Abcam, diluted 1:5000 in blocking buffer). Donkey anti-rabbit IgG HRP-conjugated polyclonal antibody (Thermo Fisher Scientific, diluted 1:2000 in blocking buffer) was used as secondary antibody to detect the protein bands. Visualization of the protein bands was performed on a ChemiDoc MP Imaging System (Bio-Rad) using a SuperSignal West Femto Kit (Thermo Fisher Scientific). All incubation steps were performed on a shaking platform. Precision Plus Protein Dual Color Standards (Bio-Rad) were loaded on the blot to identify the molecular weight of the CD163 and GAPDH protein bands. Quantitative analyses were performed with ImageJ.

Statistical analyses were performed using SPSS Statistics 23 (IBM Analytics) and Prism (GraphPad). Data were checked for normality and, if necessary, log₁₀ transformation was performed to normalize the data. Analysis of Variance (ANOVA) tests (with Tukey and Tamhane’s T2 post-hoc tests) and unpaired Student’s t-tests were applied to calculate p-values and determine whether differences between inoculation groups were statistically significant (P < 0.05).

To investigate the influence of macrophages and the macrophage-tumor cell crosstalk on primary tumor growth and ductal breakthrough 4T1 mammary tumor cells were intraductally inoculated with or without RAW264.7 macrophages (4T1 + RAW264.7 and 4T1 inoculation group, respectively) in lactating immunocompetent mice. Compared to 4T1 inoculated mice, the 4T1 + RAW264.7 inoculated mice showed no significant difference in primary tumor growth up to 5 w post-inoculation (p.i.) (Fig. 1a and b). Primary tumor weight at 3 w p.i. (i.e. when primary tumors were palpable, but no distant metastases were detectable) and 5 w p.i. (i.e. when metastases became apparent) in both the 4T1 and 4T1 + RAW264.7 inoculation group corroborated the similar primary tumor growth kinetics in both inoculation groups (Fig. 1c). Histology of the primary tumors at 1, 3 and 5 w p.i. indicated the invasiveness of the 4T1 cells in the lactating mammary ducts with tumor cells displaying characteristics of metastasis-preceding epithelial-to-mesenchymal transition (EMT) at 3 w p.i. and tumor cell invasion of blood vessels at 5 w p.i. (Fig. 1d). Ki67 staining qualitatively and quantitatively confirmed a similar tumor cell proliferation in both inoculation groups at 1, 3 and 5 w p.i. (Fig. 2). Stainings for carbonic anhydrase IX (CAIX), an enzyme that is upregulated in hypoxic tumors [17‐19], allowed to investigate tumor hypoxia and identified a similar progressive increase in CAIX expression from 1 to 5 w p.i. in primary tumors from both inoculation groups (Fig. 2). Disruption of the ductal architecture by the tumor cells was quantitated based on cytokeratin 5 stainings for myoepithelial cells in primary tumors of 4T1 + RAW264.7 and 4T1 inoculated mice. Cytokeratin 5 positivity progressively decreased in the primary tumors and was significantly lower at 3 w p.i. in the 4T1 + RAW264.7 compared to the 4T1 tumors, indicative for enhanced tumor cell breakthrough in the presence of additional macrophages (Fig. 2). In order to evaluate distant metastases of 4T1 primary tumors both with and without RAW264.7 macrophage co-inoculation, the axillary lymph nodes and the lungs - typically the first sites to be affected by metastatic 4T1 cells - were isolated at 5 w p.i. and screened through ex vivo bioluminescence imaging. Both sites had traces of metastases (Fig. 3a). The liver was also isolated and screened but showed no traces of metastases (data not shown). Quantification of the bioluminescent signals identified increased metastasis in axillary lymph nodes and lungs of the 4T1 + RAW264.7 inoculation group (mean signal of 5.022 log₁₀ ± 0.2359 p/s/cm² in axillary lymph nodes and 5.003 log₁₀ ± 0.2957 p/s/cm² in lungs) compared to the 4T1 inoculation group (mean signal of 4.331 log₁₀ ± 0.1972 p/s/cm² in axillary lymph nodes and 4.097 log₁₀ ± 0.2519 p/s/cm² in lungs) (Fig. 3b). Comparative H&E and Ki67 stainings of axillary lymph node and lung sections from metastasis-bearing mice in the 4T1 + RAW264.7 versus 4T1 inoculation group corroborated the ex vivo bioluminescence imaging results with enhanced metastatic tumor burden and tumor cell proliferation in the 4T1 + RAW264.7 inoculation group (Fig. 3c). The lung metastatic growth in both inoculation groups was further characterized immunohistochemically by staining of the immune cell infiltrates (identified by immune cell markers CD45 (pan-immune cell marker), Ly6G (neutrophil marker), CD8a (cytotoxic immune cell marker) and CD163 (anti-inflammatory macrophage marker)) and vascular structures (identified by the endothelial cell marker CD31) (Additional file 1: Figure S1).

The advancement in primary tumor growth and metastasis was reflected by a progressive splenomegaly in both inoculation groups. Whereas the spleen weight did not differ at 3 w p.i. between the 4T1 + RAW264.7 and 4T1 inoculation group, splenomegaly became more pronounced at 5 w p.i. in the 4T1 + RAW264.7 inoculation group (Fig. 3d and e). Indeed, spleen weights and sizes at 5 w p.i. were approximately 1.5 times larger in the 4T1 + RAW264.7 inoculation group compared to the 4T1 inoculation group.

Macrophages are known as important immunomodulators in the breast tumor microenvironment and can strongly affect disease progression. More specifically, macrophages can stimulate either pro- or anti-inflammatory immune responses based on their activation state, i.e. either M1 or M2. Flow cytometric analysis identified a M1 to M2 polarization of RAW264.7 macrophages co-cultured with 4T1 tumor cells based on their significantly decreased intracellular expression of the M1-related (i.e. pro-inflammatory) cytokine interleukin (IL)-12 and significantly increased intracellular expression of the M2-related (i.e. anti-inflammatory) cytokine transforming growth factor (TGF)-β1 compared to RAW264.7 macrophages that were cultured alone (Fig. 4a). In addition, culture media of 4T1 + RAW264.7 co-cultures showed significantly decreased levels of the M1-related cytokine tumor necrosis factor (TNF)-α and significantly increased levels of the M2-related cytokines B-cell activating factor (BAFF), granulocyte colony-stimulating factor (G-CSF) and TGF-β1 compared to RAW264.7 mono-culture media (Fig. 4b). To provide more insights into the immunological impact of the co-inoculated macrophages on the observed tumor progression in vivo, local and systemic M1−/M2-related cytokine profiles were investigated and compared at 3 and 5 w p.i. in the primary tumor lysates and sera from the 4T1 + RAW264.7 and 4T1 inoculation group. At 3 w p.i. 4T1 + RAW264.7 primary tumor lysates showed a significant increase in the M1-related cytokines IL-1β, monocyte chemoattractant protein (MCP)-1, macrophage inflammatory protein (MIP)-2 and TNF-α compared to the 4T1 primary tumors, whereas the M2-related cytokines BAFF, G-CSF, IL-4 and TGF-β1 showed significantly lower levels in the 4T1 + RAW264.7 versus the 4T1 primary tumors (Fig. 5a and b). However, at 5 w p.i. these differences in M1- and M2-related cytokine profiles could no longer be detected between both inoculation groups (Fig. 5c and d). M1- and M2-related cytokines were also measured in lysates of RAW264.7 macrophage-only intraductally inoculated mammary glands. These results showed that the 4T1 tumor cells strongly influence the local and RAW264.7 macrophage-mediated immune responses as, compared to 4T1 + RAW264.7 and 4T1 primary tumors, RAW264.7-only mammary glands contained higher M1-related cytokine levels (interferon (IFN)-γ, IL-1β and TNF-α at 3 and 5 w p.i., and MCP-1 and IL-6 at 5 w p.i. compared to 4T1 + RAW264.7 and 4T1; MCP-1 and MIP-2 at 3 w p.i. compared to 4T1) and lower M2-related cytokine levels (BAFF and IL-4 at 3 and 5 w p.i., IL-10 at 3 w p.i., and TGF-β1 at 5 w p.i. compared to 4T1 + RAW264.7 and 4T1; G-CSF and TGF-β1 at 3 w p.i. compared to 4T1; G-CSF at 5 w p.i. compared to 4T1 + RAW264.7) (Fig. 5).

In contrast to the local primary tumor the serum cytokine levels were far less different between the 4T1 + RAW264.7 and 4T1 inoculation group at 3 and 5 w p.i. More specifically, at 3 w p.i. the serum levels of the M1-related cytokine MCP-1 were higher and the M2-related TGF-β1 serum levels were lower in the 4T1 + RAW264.7 compared to the 4T1-only inoculation group (Additional file 2: Figure S2A and B). These systemic differences were no longer detected at 5 w p.i. (Additional file 2: Figure S2C and D). In contrast to tumor-bearing mice, serum cytokine profiles at 3 and 5 w p.i. in the RAW264.7 inoculation group showed increased levels of M1-related cytokines (MCP-1 and MIP-2 at 3 and 5 w p.i., and TNF-α at 5 w p.i. compared to 4T1 + RAW264.7 and 4T1) and decreased levels of M2-related cytokines (BAFF at 3 and 5 w p.i., and G-CSF at 5 w p.i. compared to 4T1 + RAW264.7 and 4T1; TGF-β1 at 3 and 5 w p.i. compared to 4T1), reflecting the local M1−/M2-related cytokine profile differences (Additional file 2: Figure S2).

In order to further investigate the immunological profile of the primary tumors in both the 4T1 + RAW264.7 and 4T1 inoculation group, immunohistochemistry for CD45, CD163, Ly6G and CD8a was performed on primary tumor sections at pre- (1 and 3 w p.i.) and metastatic (5 w p.i.) time points (Fig. 6). Based on the pan-immune cell marker CD45, the percentage of immune cells increased and these cells invaded both the 4T1 + RAW264.7 and 4T1 primary tumors over time, but also remained partly in the surrounding stroma (Fig. 6). At 1 w p.i., when tumor cells were still in situ and did not yet invade the mammary fat pad, immune cells were scarce. Only few CD163-positive cells could be detected, which mainly comprise tumor-recruited anti-inflammatory/M2 macrophages. The CD163 stainings could not detect the intraductally inoculated RAW264.7 macrophages as these cells are initially M1 polarized and also do not express CD163 based on immunohistochemistry and western blotting for CD163, respectively on paraffin slides and in lysates of RAW264.7 inoculated mammary glands (Additional file 3: Figure S3). At 3 w p.i. the CD163-positive tumor-recruited cells remained outside the tumor area, but strongly increased in the 4T1 inoculation group whereas in the 4T1 + RAW264.7 inoculation group a similar CD163-positive staining was observed at 3 w p.i. compared to 1 w p.i (Fig. 6). At 5 w p.i. the 4T1 + RAW264.7 and 4T1 primary tumors showed similar CD163 stainings, but the CD163-positive cells still minimally invaded the tumor tissue. Western blotting for CD163 corroborated the lower CD163 expression at 3 w p.i. and similar CD163 expression at 5 w p.i in 4T1 + RAW264.7 primary tumors compared to 4T1 primary tumors (Additional file 3: Figure S3B). Additional immunohistochemistry for Ly6G and CD8a in 4T1 + RAW264.7 and 4T1 primary tumors also showed few staining around the tumor-harboring mammary ducts at 1 w p.i. However, at 3 and 5 w p.i. the Ly6G and CD8a positivity increased and, in marked contrast to CD163, were located within the primary tumor area. Compared to the 4T1 primary tumors, 4T1 + RAW264.7 primary tumors qualitatively and quantitatively displayed increased staining for Ly6G at 3 w p.i., whereas similar staining was detected in both inoculation groups for CD8a at that time point, and for Ly6G and CD8a stainings at 5 w p.i (Fig. 6).

Extracellular matrix degradation and angiogenesis were investigated at the protein level by measuring the primary tumor lysate and serum levels of MMP-9 and VEGF in the 4T1 + RAW264.7 and the 4T1 inoculation group at 3 and 5 w p.i. Local MMP-9 and VEGF levels in the primary tumor lysates increased as tumor growth progressed from 3 to 5 w p.i. (Fig. 7a and b). However, at 3 w p.i. the MMP-9 and VEGF levels significantly decreased in the 4T1 + RAW264.7 compared to the 4T1 inoculation group, whereas this difference was no longer seen at 5 w p.i. (Fig. 7a and b). In accordance with the local and systemic cytokine profiles, significantly decreased MMP-9 and VEGF levels were detected in the RAW264.7 inoculated mammary gland lysates compared to the 4T1 + RAW264.7 and 4T1 inoculated counterparts at 3 and 5 w p.i. (Fig. 7a and b). Systemic MMP-9 levels also progressively increased over time and were significantly higher at both 3 and 5 w p.i. in the 4T1 + RAW264.7 inoculation group compared to the 4T1 inoculation group and significantly lower in the RAW264.7 inoculation group at 3 w p.i. compared to the 4T1 + RAW264.7 inoculation group and at 5 w p.i. compared to both the 4T1 + RAW264.7 and 4T1 inoculation group (Fig. 7c). In contrast, VEGF serum levels were close to the detection limit and moreover did not increase between 3 and 5 w p.i. (Fig. 7d). Whereas the systemic VEGF levels in the 4T1 + RAW264.7 and 4T1 inoculation group did not differ at 3 w p.i., at 5 w p.i. the 4T1 + RAW264.7 inoculation group showed significantly higher systemic VEGF levels compared to the 4T1 inoculation group (Fig. 7d). Systemic VEGF levels in the RAW264.7 inoculation group were significantly lower compared to the 4T1 + RAW264.7 and 4T1 inoculation group at both 3 and 5 w p.i. (Fig. 7d).

Immunohistochemical stainings of 4T1 + RAW264.7 and 4T1 primary tumors with the endothelial cell marker CD31 were indicative for an increase in vascular growth over time. Upon quantification, significantly less CD31 staining was observed in 4T1 + RAW264.7 compared to 4T1 tumors upon invasion of tumor cells in the mammary fat pad at 3 w p.i., mirroring the difference in primary tumor VEGF response at that early time point (Fig. 7e).

CHI3L1 and LCN2 have previously been identified as immune-related biomarkers for disease outcome in breast cancer patients as well as for monitoring tumor progression in the intraductal mouse model for TNBC [6, 12‐14]. Measurement of these two parameters in 4T1 + RAW264.7 and 4T1-only tumor-bearing mice could therefore verify the observed differential breast cancer burden between the two inoculation groups. Preliminary in vitro tests showed that co-cultures of 4T1 tumor cells with RAW264.7 macrophages secreted higher amounts of both CHI3L1 and LCN2 than either 4T1 or RAW264.7 mono-cultures (Fig. 8a and b). CHI3L1 and LCN2 levels were subsequently measured in primary tumor lysates and sera from 4T1 + RAW264.7 and 4T1 inoculated mice and showed a progressive increase from 3 to 5 w p.i. (Fig. 8c-f). Local levels did not differ between 4T1 + RAW264.7 and 4T1 primary tumors, which is indicative for the similar primary tumor growth observed in both inoculation groups (Fig. 8c and d). In marked contrast, systemic levels significantly increased in 4T1 + RAW264.7 inoculated mice, mirroring the increased CHI3L1 as well as LCN2 secretion found in 4T1 + RAW264.7 co-cultures (Fig. 8e and f). Local and systemic levels in the RAW264.7 inoculation group were also lower than those in the 4T1 + RAW264.7 and 4T1 counterparts at 5 w p.i., yet not at 3 w p.i. (Fig. 8c-f). Additional measurements of the CHI3L1 and LCN2 levels in spleen lysates from 4T1 + RAW264.7 and 4T1 inoculated mice showed that in the two inoculation groups both biomarkers increased over time with enhanced splenomegaly, but remained low in the RAW264.7-only inoculated mice. At 5 w p.i., when spleen sizes in the 4T1 + RAW264.7 inoculation group were more severely increased compared to their 4T1 counterparts, CHI3L1 and LCN2 levels were also significantly higher in the 4T1 + RAW264.7 spleen lysates than in the 4T1 spleen lysates (Fig. 8g and h). Spleens from RAW264.7-only inoculated mice showed significantly decreased CHI3L1 and LCN2 levels at 5 w p.i. compared to both the 4T1 + RAW264.7 and 4T1 inoculation group (Fig. 8g and h).