The study involving human participants was approved by the Ethics Committee of Fudan University Shanghai Cancer Center. Written informed consent was obtained from all the patients before the enrollment. All animal protocols were approved by the Animal Care Committee of Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences (permit no. IBS14-0806). Animals were kept under stable temperature and humidity conditions. All surgery was performed by cervical dislocation to minimize suffering.

Paraffin-embedded blocks were obtained from the Breast Malignancy Database established by the Department of Breast Surgery, Fudan University Shanghai Cancer Center (Shanghai, China). All of the enrolled patients have fully detailed clinicopathological information and follow-up results. Written informed consent was obtained from the patients before enrollment. We are authorized to use the tissues for research only and have reported the database information previously (14). For immunohistochemical analysis, 89 paraffin-embedded blocks were cut into 5-μm serial sections, and following the confirmation of breast cancer diagnosis by H&E staining, immunohistochemistry was performed following standard procedures. CD163 antibody (clone 10D6; Novocastra/Leica Microsystems, Newcastle, UK) used for immunostaining was titered to find the optimal concentration (1:100). Sections were counterstained with hematoxylin for the identification of the nuclei. Detection was performed using the Dako EnVision System. Images were captured under a microscope with a CCD camera.

Breast cancer cell lines MDA-MB-231, MD-MB-468 and MCF-7 were originally obtained from the American Type Culture Collection (ATCC; Manassas, VA, USA); the human promonocytic cell line U937 originating from the ATCC was kindly provided by Professor Duan Ma (Key Laboratory of Molecular Medicine, Ministry of Education, Shanghai Medical College, Fudan University). All cells were maintained in RPMI-1640 supplemented with 10% fetal bovine serum (FBS) and 100 U/ml penicillin/streptomycin (Gibco/Invitrogen, Carlsbad, CA, USA) at 37°C in a humidified atmosphere of 5% CO₂.

MDA-MB-231, MD-MB-468, MCF-7 and U937 cells were transfected with green or red fluorescence protein (GFP or RFP) expressing lentiviral particles (cat. nos. LVP340 and LVP299; GenTarget, Inc., San Diego, CA, USA) according to the manufacturer's instructions. MCF-7 was transfected with GFP with puromycin resistance (GFP-puro)-tagged CSF1-overexpressing lentiviral particles (cat. no. GCK970139; GeneChem Co., Ltd., Shanghai, China), and MDA-MB-231 was transfected with GFP-puro-tagged shRNA lentiviral particles specifically targeting CSF1 (cat. no. ST3071-A; Shanghai SunBio Medical Biotechnology Co., Ltd., Shanghai, China). Briefly, for adhesive cells the cells were seeded in complete medium at appropriate density (at the time of transduction, cells should be 50–75% confluent) and incubated overnight, the medium was removed and fresh warm complete medium was added with appropriate amount of pre-made lentivirus to obtain the desired MOI, then incubated for another 72 h at 37°C in a humidified atmosphere of 5% CO₂. For suspension cells, the cells were seeded in complete medium at appropriate density and incubated until density reaching 3×10⁶ cells/ml, then cells were diluted into 1×10⁶ cells/ml with fresh warm complete medium and appropriate amount of pre-made lentivirus was added to obtain the desired MOI, incubated for 24 h at 37°C in a humidified atmosphere of 5% CO₂ with a shaking flask, then an equal amount of fresh medium containing relevant antibiotics was added for another 48 h. Positive transduction of cells were visualized by fluorescence microscopy and isolated by fluorescence-activated cell sorting (FACS) followed by puromycin or neomycin selection for 2 weeks.

For the co-culture assay, MDA-MB-231, MDA-MB-468, MCF-7 and U937 cells cultured in RPMI-1640 supplemented with 10% FBS and 100 U/ml penicillin/streptomycin at a final concentration of 5×10⁵/ml using 6-well insert system with 0.4-μm pores (cat. no. 3450; Corning, Inc., Corning, NY, USA). Breast cancer cell suspension (1 ml) was added to each insert well and 1 ml of U937 cell suspension was added to each lower well. After a 72-h incubation, U937 cells in the lower well were harvested for further examinations. Microscopy images were captured for each well at five randomly picked fields under a microscope with a 10X objective lens, and the number of adhesive cells for each field was quantified. The number of adhesive cells was determined by comparing different breast cancer cell lines and blank control, which was used as reference.

For the Transwell migration assay, GFP expression MDA-MB-231, MCF-7 stable cells were cultured in complete RPMI-1640 medium and RFP expression U937 stable cells cultured in serum-free RPMI-1640 using a 24-well insert system with 5-μm pores (cat. no. 3421; Corning, Inc.). A total of 0.5 ml of 5×10⁵/ml breast cancer cell suspension was added to each lower well, and after a 24-h incubation, 0.5 ml of 5×10⁵/ml U937 cell suspension was added to each insert well. The breast cancer conditioned medium in the lower chamber served as a chemoattractant. After a 72-h incubation, migratory cells were visualized by fluorescence microscopy and images were captured under a fluorescence microscope with a CCD camera. Microscopy images were captured for each well at five randomly picked fields with a 10X objective lens, and the number of migratory cells for each field was quantified. The number of migratory cells was determined by comparing different breast cancer cell lines and blank control, which was used as reference.

The invasion assay was performed similarly to migration assay using 24-well insert system with 8-μm pores (cat. no. 3422; Corning, Inc.) except the inside of the insert was precoated with a thin layer of Matrigel™ Basement Membrane Matrix (BD Biosciences) diluted with serum-free medium to a final concentration of 6 mg/ml. A total of 0.5 ml of 5×10⁵/ml breast cancer cell suspension was added to each lower well, and after a 24-h incubation, 0.5 ml of 5×10⁵/ml U937 cell suspension was added to each insert well. The breast cancer conditioned medium in the lower chamber served as a chemoattractant. After a 72-h incubation, invaded cells were visualized by fluorescence microscopy and images were captured under a fluorescence microscope with a CCD camera. Microscopy images were captured for each well at five randomly picked fields with a 10X objective lens, and the number of invaded cells for each field was quantified. The number of invaded cells was determined by comparing different breast cancer cell lines and blank control, which was used as reference.

Cytokine secretion was quantified by Bio-Plex^® cytokine assay. Briefly, conditioned medium samples were obtained after centrifugation to remove cells and their debris and stored at −80°C until cytokine profiling. We used the fluorescent bead-based detection assay system, with an array of beads in liquid suspension, each containing different ratios of two spectrally distinct fluorophores, thereby assigning a unique spectral identity. The beads, which had been conjugated with a monoclonal antibody specific for a target protein, were incubated with the samples to be tested, washed, followed by addition of a biotinylated detection antibody, washed again, and finally incubated with streptavidin-PE. A wide range of standards (1.78–29,107.00 pg/ml) was used to enable quantitation of the individual cytokines using a Bio-Plex array reader with a dual laser detector and real-time digital signal processing. Cytokine levels were measured using a Multiplex kit (Bio-Rad Laboratories, Inc., San Diego, CA, USA) according to the manufacturer's instructions. Standard curves for each cytokine were generated using the reference concentrations provided in the kit. The plate was run on a Luminex 200 Bio-Plex Instrument (Bio-Rad, Hercules, CA, USA). Raw fluorescence data were analyzed with software using the 5-parameter logistic method. Detailed procedures of the Bio-Plex^® Suspension Array System have been described elsewhere (15–17).

Total RNA was isolated from 1×10⁶ cells using TRIzol (Invitrogen) according to the manufacturer's instructions. One microgram of total RNA was reverse transcribed using PrimeScript^® RT-PCR Kit (Takara Bio, Dalian, China) primed with oligo(dT). One microliter of cDNA was subjected to PCR amplification using gene-specific primers. PCR products were resolved on a 2% agarose gel and visualized by ethidium bromide staining. Quantitative PCR was carried out using the SYBR^®-Green PCR Master Mix Kit (Takara Bio). Primer sequences are detailed in Table I. The quantification of gene expression was normalized to the expression of GAPDH.

For immunoblot analysis, cell lysates containing 50 μg of total protein were separated by electrophoresis on 10% sodium dodecyl sulfate-polyacrylamide gels and electroblotted onto polyvinylidene difluoride membranes (EMD Millipore, Billerica, MA, USA). After blocking of the unoccupied sites with 5% skim milk-PBS, the membranes were probed with the desired primary antibodies at 4°C overnight, followed by incubation with the appropriate HRP-conjugated secondary antibodies, according to the manufacturer's instructions. The primary antibodies were anti-β-actin, anti-CSF1, and the proper dilution of them was according to the instructions. The signal was visualized by ECL Western Blotting Detection Reagents (Pierce Biotechnology, Inc.) and photographed with ImageQuant LAS 4000 (GE Healthcare, Piscataway, NJ, USA). β-actin was used as a loading control.

To assess the tumorigenicity in vivo, 8-week-old female non-obese diabetic/severe combined immunodeficient (NOD/SCID) mice were used. Eighteen mice were assigned to six groups equally. A total of 5×10⁵ cells suspended in 100 μl PBS were injected orthotopically into the right lower mammary fat pad according to standard injection procedures. Once tumors were palpable, tumor volume (1/2 × length × width²) was monitored every other day for 8 weeks. Tumor growth curves were plotted.

The quantitative results are presented as the mean ± standard error of at least three independent experiments. Significant differences were determined with Student's t-test using Excel or GraphPad Prism 5 Demo software (GraphPad Software, Inc., San Diego, CA, USA). P<0.05 was considered statistically significant.

TAMs play an important role in the growth and progression of cancers. CD163 has been reported as a specific surface antigen of M2 macrophages (12). Herein, we first detected CD163 expression in breast cancer tissues or non-cancer breast tissues by immunohistochemistry. The results showed that the percentage of CD163-positive cells was significantly higher in hormone-independent breast cancers than that in hormone-dependent counterparts and normal breast tissues, in some cases, TAMs may make up as much as 50% of tumor volume. Moreover, the difference was also significant between hormone-independent DCIS and IDC, whereas CD163 expression was hardly detectable in normal breast tissues (Fig. 1). However, we did not observe a significant difference among tumor size, HER2 expression, lymph node involvement, proliferation index (Ki67) and menstrual status (Table II).

The impact of TAM infiltration on overall survival (OS) was evaluated in breast cancer patients. Both CD163 and CD204 are specific surface antigens of M2 macrophages, as previously reported (18,19). We assessed CD163 and CD204 mRNA expression level in a cohort of 1,117 breast cancer tissue samples (containing tumor stroma) based on publically available gene expression datasets (20). For CD163⁺ macrophage frequency, patients were grouped as ‘high’ or ‘low’ using the lower quartile as a cut-off point; while for CD204⁺ macrophage frequency, patients were grouped as ‘high’ or ‘low’ using the median as a cut-off point. Increased expression of CD163 and CD204 was observed in ER-negative patients, and the patients with high density of CD163⁺ and CD204⁺ macrophages also showed a poor OS, which was an independent poor prognostic factor for OS (Fig. 2).

The human promonocytic U937 cell line was used as a model to simulate the interaction between monocytes and breast cancer cells in vitro. U937 cells normally grow in suspension and have a smooth surface, as previously reported (21). However, after co-cultured with hormone-independent breast cancer cell lines MDA-MB-231 and MDA-MB-468 in a 6-well insert system with 0.4-μm pores, they differentiated into M2 macrophages, as evidenced by a thorny morphology, large cell volume and increase in adherence, and increasing expression of surface antigen CD163 and CD204, all of which are characteristics of M2 macrophages. Whereas, no further obvious changes were observed when co-cultured with hormone-dependent MCF-7 cell line (Fig. 3).

To determine the factors that contributed to monocyte differentiation leaving the vasculature and entering the tumor tissue, we generated stably expressing GFP-puro MDA-MB-231, MCF-7 and RFP with neomycin resistance (RFP-neo) U937 cell lines using lentiviral particles according to the protocol.

We performed the Transwell assay without extracellular matrix barrier to study the capability of breast cancer cells to recruit monocytes using a 24-well insert system with 5-μm pores. After co-cultured for 72 h, migratory cells were visualized by fluorescence microscopy and images were captured under a fluorescence microscope. Microscopy images were captured for each well at five randomly picked fields with a 10X objective lens, and the number of migratory cells for each field was quantified (Fig. 4). We further mimicked the in vivo circumstance applying the Transwell assay with extracellular matrix barrier (BD Biosciences) using a 24-well insert system with 8-μm pores. Our results showed that co-cultured MDA-MB-231 with U937 caused an increase in migration and invasion compared with MCF-7, which indicated that hormone-independent breast cancer cells recruited more monocytes than hormone-dependent breast cancer cells (Fig. 5).

The fluorescent bead-based detection assay is a highly sensitive method for profiling multiple cytokines, which has an advantage over conventional enzyme-linked immunosorbent assay (ELISA), which enables analysis of a number of analytes simultaneously (22). The strong effects of the MDA-MB-231 cells in monocyte recruitment and differentiation led us to inspect their conditioned media (CM) for dissimilarly secreted monocyte chemotaxis and differentiating factors. Among the cytokines tested, there was a significant difference in CSF1 secretion level among different breast cancer cell lines. We found that mean CSF1 level was much higher in MDA-MB-231 cell conditioned medium compared to that of MCF-7 (Fig. 6).

In order to further study the correlation of CSF1 secretion level with monocyte differentiation and homing, and with tumorigenic ability, we overexpressed CSF1 in MCF-7 cells and generated the stable cell line using lentivirus particles following the protocol. The CSF1-overexpressing MCF-7 stable cells were identified by RT-PCR and western blotting, which showed that CSF1 was overexpressed in MCF-7 artificially (Fig. 7A and B). However, CSF1 level was not increased in MCF-7 CM (Fig. 7C).

We co-cultured CSF1-overexpressing MCF-7 stable cells with U937 cells to identify its ability to induce monocyte differentiation using Transwell assay in vitro. We observed that CSF1-overexpressing MCF-7 stable cells did not make U937 cells to differentiate into M2 macrophage (data not shown). This seems to be contradictory superficially, however, as mentioned previously, the active form of CSF1 is an extracellular disulfide-linked homodimer, which is evidenced by our data. No obvious increased CSF1 secretion profile in CSF1-overexpressing MCF-7 stable cell conditioned medium, since we did not envelope the CSF1 plasmid with a proper signal peptide.

Subsequently, we knocked down CSF1 in MDA-MB-231 cells using a series of siRNA sequences (Table III) specifically targeting the CSF1 gene using shRNA lentivirus particles following the protocol. The CSF1-knockdown MDA-MB-231 cells were identified by real-time PCR, which showed that CSF1 was knocked down somewhat in MDA-MB-231 artificially (Fig. 7D). We applied the most satisfactory shRNA lentivirus particles (cat. nos. ST3071-A and pLVT742) to generate a stable knockdown cell line for further study. Our data suggested that CSF1-knockdown stable MDA-MB-231 cells secrete much less CSF1 in its CM and thus failed to induce U937 cell differentiation into M2 macrophages.

In an in vivo tumorigenic ability study, 5×10⁵ cells suspended in 100 μl PBS were injected orthotopically into the mammary fat pad of 8-week-old female NOD/SCID mice. Similarly, only MDA-MB-231 cell lines form a palpable xenograft tumor successfully in 6 weeks, while being sacrificed in 8 weeks was confirmed again (Fig. 7E). Both the results of in vitro and in vivo studies confirmed that CSF1 played its role in promoting tumor progression by inducing the differentiation of monocytes.