Annexin-A1 enhances breast cancer growth and migration by promoting alternative macrophage polarization in the tumour microenvironment

Macrophages are potent immune cells with well-established roles in the response to stress, injury, infection and inflammation. The classically activated macrophages (M1) are induced by lipopolysaccharide (LPS) and express a wide range of pro-inflammatory genes. M2 macrophages are induced by T helper type 2 cytokines such as interleukin-4 (IL4) and express high levels of anti-inflammatory and tissue repair genes. The strong association between macrophages and tumour cells as well as the high incidences of leukocyte infiltration in solid tumours have contributed to the discovery that tumour-associated macrophages (TAMs) are key to tumour progression. Here, we investigated the role of Annexin A1 (ANXA1), a well characterized immunomodulatory protein on macrophage polarization and the interaction between macrophages and breast cancer cells. Our results demonstrate that ANXA1 regulates macrophage polarization and activation. ANXA1 can act dually as an endogenous signalling molecule or as a secreted mediator which acts via its receptor, FPR2, to promote macrophage polarization. Furthermore, ANXA1 deficient mice exhibit reduced tumour growth and enhanced survival in vivo, possibly due to increased M1 macrophages within the tumor microenvironment. These results provide new insights into the molecular mechanisms of macrophage polarization with therapeutic potential to suppress breast cancer growth and metastasis.


Results
4T1 and 67NR murine breast cancer cells polarize macrophages to M2-phenotype. 4T1 and 67NR breast cancer cell lines were derived from a single mammary tumour that develop spontaneously in a BALB/c mouse. 4T1 cells are highly invasive leading to metastases, whereas 67NR cells were derived from primary tumours, but do not metastasize or invade 29,30 .
Breast tumours are highly infiltrated by different types of host leukocytes, and importantly, T cells and monocytes which differentiate into TAMs at the tumour site 30 . Given the ability of TAMs to be recruited to tumours by a range of growth factors and chemokines, which are often produced by the tumour cells themselves, we sought to determine the effect of 4T1 and 67NR-conditioned media (CM) on macrophage polarization. The proportion of M1 and M2 macrophages expressing specific M1 (CD86) or M2 (CD206) cell surface markers was assessed by FACS analysis (Fig. 1A,B).
Next, we examined if 4T1-CM could enhance the phagocytic function of macrophages. RAW 264.7 mouse macrophages were treated with 4T1, 67NR-conditioned culture media, LPS (10 ng mL −1 ) or IL4 (20 ng mL −1 ) for 24 h prior to incubation with E. coli bioparticles (Fig. 1C). Treatment of RAW cells with 4T1-CM enhanced the phagocytic activity by 30% compared to the control. This effect was similar in macrophages treated with IL4 for 24 h, our positive control for alternatively activated, or M2 macrophages. However, treatment of macrophages with 67NR-CM or LPS did not enhance phagocytic ability, suggesting that alternatively activated macrophages, through the activation with 4T1-CM or IL4 exhibit higher phagocytic ability.
The extent of macrophage polarization by 4T1 and 67NR-CM was also confirmed at the gene expression level. As expected, M2 polarization was observed in RAW macrophages following treatment with IL4 (Supplemental data Fig. 1A,B). A time-dependent increase in the expression of M2-signature mRNAs (Arginase-1 and PPARγ) was observed with 4T1-CM treatment, with the highest expression at 48 h post-treatment. (Fig. 1D). In comparison, arginase-1 and PPARγ expression was significantly increased 24 h post-treatment with 67NR-CM, which was shortlived and reduced back to control levels by 48 h (Fig. 1E). M2 polarization status at protein level after 4T1-CM treatment was confirmed by examining the expression of Arg1, PPARγ and iNOS by immunoblotting analysis (Supplemental data Fig. 1C). Consistent with mRNA expression, stimulation with 4T1-CM for 48 h induced high expression of M2 markers compared to control, while 67NR-CM treatment did not change iNOS, Arg-1 or PPARγ expression at this time point. These results demonstrate that treatment of macrophages with breast tumor conditioned media, in particular 4T1, can induce a skewing to the M2 subset with different kinetic profiles.
SCIENTIFIC RepoRts | (2017) 7:17925 | DOI:10.1038/s41598-017-17622-5 M2 polarization is positively correlated with tumour growth in a MMTV-Wnt mouse model. A spontaneous mammary tumour model -MMTV-Wnt has been used previously to examine the genetic basis of breast cancer. It was reported that transgenic expression of Wnt using MMTV LTR enhances ductal hyperplasia in early life and 50% of female transgenic mice exhibit mammary adenocarcinomas by 6 months 31 . We next determined the macrophage immune signature of the primary tumour in MMTV-Wnt mice by analysis of cell numbers and phenotype. Phenotypic analysis of TAM derived from MMTV-Wnt mice was performed on isolated macrophages for distinct populations using surface expression markers by flow cytometry. We found that there were more CD11b-leukocytes in normal mammary glands while more leukocytes in MMTV-WNT mammary tumours were CD11b+ (Fig. 2B). In the CD11b + enriched population, there were significantly more CD11b+ Ly6G-GR1-F480intCD11c-macrophages in tumours compared to normal mammary glands (Fig. 2C), which may be recruited macrophages and more of these CD11c-F480int macrophages exhibit high expression of CD206 and low expression of CD86 ( Fig. 2A,D), suggesting that TAMs in the MMTV-Wnt tumours may be more M2 polarized.
ANXA1 plays a role in macrophage polarization. Annexin-a1 is a multifunctional molecule involved in a range of cellular signal transduction pathways, particularly in inflammation, innate and adaptive immune system, tumour progression and metastasis 32 . Previous studies shown that expression of ANXA1 is increased in certain cancers such as pancreatic and gastrointestinal cancer 33,34 , and decreased in others such as esophageal and prostate cancer 35,36 . To establish whether ANXA1 is required for macrophage polarization, the ability of lipopolysaccharide (LPS), gamma interferon (IFN-γ) to induce M1 polarization and interleukin 4 (IL-4) to induce M2 polarization was examined in murine bone-marrow-derived macrophages (BMDM) from WT BALB/c mice (ANXA1 +/+ ) or mice deficient in ANXA1 (ANXA1 −/− ). IL4 treatment upregulated M2 markers YM1 and Retnia (Fizz1) in BMDM from ANXA1 +/+ , whereas macrophages derived from ANXA1 −/− mice were less sensitive to IL4-induced M2 polarization. Low iNOS levels were observed upon IL4 treatment in both BMDM from ANXA1 +/+ and ANXA1 −/− mice (Fig. 3A), and LPS + IFNγ stimulation enhanced iNOS expression more significantly in ANXA1 −/− compared to ANXA1 +/+ (p < 0.01, Fig. 3B), suggesting that in the absence of ANXA1, a polarization shift to M1-phenotype, rather than M2 phenotype is observed.
To investigate whether the effect of 4T1-CM on macrophage polarization is ANXA1-dependent, BMDM from ANXA1 +/+ and ANXA1 −/− mice were treated with 4T1-CM for 24 h and CD206 expression (M2 marker) was analysed by flow cytometry. CD206 expression were significantly increased 24 h post-treatment with 4T1-CM in ANXA1 +/+ cells, but not in macrophages from ANXA1 −/− mice (Fig. 3C). These data indicate that ANXA1 is required for the macrophage phenotype shift from M1 to M2 by 4T1-CM.
We further determined whether ANXA1 N-terminal peptide product Ac2-26 could induce M2 macrophage polarization. Macrophages were treated with ANXA1 peptide Ac2-26 (1 µM) and mRNA levels of IL-12 and Arg-1 was evaluated by RT-qPCR. At 24 h post-treatment, the overall polarization of macrophages shifted to M2, as they expressed high levels of Arg1 and low levels of IL-12 mRNA (Supplemental Fig. 2A,B). These data implies that ANXA1 itself can induce M2 phenotypic shift.

ANXA1 is required for functional M2 macrophage dependent promotion of tumour growth and invasion. To address if macrophages can influence tumour cell proliferation and migration in vitro, and if this
is dependent on ANXA1, macrophages derived from ANXA1 +/+ and ANXA1 −/− mice were co-cultured in a transwell separation experiment with 4T1 breast cancer cells. BMDM were polarized for 24 h to M1 or M2-phenotype upon treatment with IFN-γ + LPS or IL4, respectively. 4T1 cells co-cultured with ANXA1 +/+ M2-polarized macrophages exhibited enhanced growth and invasion compared to 4T1 cells treated with non-polarized media. However, 4T1 cells co-cultured with ANXA1 −/− M2-polarized macrophages demonstrated significantly lower growth and invasion than ANXA1 +/+ macrophages (Fig. 3D,E). Taken together, these data suggest that macrophage ANXA1 plays an important role in the tumour microenvironment which can promote breast cancer growth and migration, possibly through the induction of macrophage polarization to an alternatively activated immunosuppressive phenotype (M2). CCL5 induces ANXA1 expression and modulates macrophage polarization. The pro-inflammatory cytokine profile of the tumour microenvironment is extremely important to translate signals that communicate with immune cells, leading to inflammation and cancer progression. We therefore, examined the cytokine profile of conditioned media derived from 4T1 cells. Using a cytokine array, we observed that 4T1-CM contained 3-fold higher levels of CCL5 compared with serum-free media (Supplemental Fig. 3). In line with data obtained from the cytokine array, our validated ELISA results demonstrate that CCL5 was highly secreted by 4T1 cells (Fig. 4A).

ANXA1 enhances breast tumour growth and inhibits macrophage activation in vivo. Finally, to
analyse the potential impact of tumor microenvironment derived ANXA1 on the regulation of tumour growth in vivo, 4T1 murine breast cancer cells which harbor a luciferase promoter (4T1-12B) were injected into the 4 th left mammary gland of female BALB/c WT and ANXA1 −/− mice. After injection, tumour growth was measured by manual measuring and bioluminescence imaging every week for 6 weeks. At 5 weeks post injection, tumour growth was significantly higher in the ANXA1 +/+ group compared to ANX1 −/− group (Fig. 6A). Bioluminescence imaging confirmed that the ANXA1 +/+ mice injected with 4T1 murine breast cancer cells exhibited increased luminescence compared with ANXA1 −/− mice (Fig. 6B). Similarly, metastasis in isolated lungs, liver and spleen were observed in ANXA +/+ mice after 6 weeks (Fig. 6C), whereas metastasis was not observed in ANXA1 −/− mice (data not shown). Consistent with this, ANXA1 −/− mice survived longer than ANXA1 +/+ mice post-injection (Fig. 6D). To establish a direct connection between tumour growth and macrophage polarization, macrophages were isolated after 5 weeks post injection and the immune signature of the tumours were determined. Phenotype analysis of the tumour showed a smaller CD11b+ population derived from the tumour from ANXA1 −/− mice (Fig. 6E), and in this CD11b+ population, CD206+ expression levels were similar while CD86+ expression was significantly higher in the ANXA1 −/− mice group (Fig. 6F,G), suggesting that in the absence of ANXA1, the polarization of macrophages in the tumour are more skewed to M1-phenotype. This data strongly indicates that ANXA1 in the tumour microenvironment plays an important role in the development, progression and metastasis of breast cancer.

Discussion
A hallmark of TAM-associated inflammation is the infiltration of leukocytes and stromal cells, which results in the enhanced macrophage recruitment and tumour development 11 . TAM infiltration are correlated with poor patient prognosis and metastasis in breast carcinoma 14 , and these effects may be due the ability of TAM to secrete cytokines, growth factors, chemokines and proteins that can stimulate cancer proliferation and cell invasion. The plasticity of macrophages polarization is rapid and occur at the levels of gene expression, protein, metabolite and microbicidal activity in response to changes in the cytokine environment 1-7,10,11 . The molecular mechanisms underlying the contribution of an inflammatory tumour microenvironment is not totally elucidated. Therefore, the characterisation of the phenotype of TAM is essential to the understanding of tumour-derived signals polarization of innate and adaptive immunity in tumour progression.
ANXA1 plays an important role in tumour development and progression of basal-like breast cancer 37 . We have shown previously that suppression of ANXA1 in highly metastatic breast cancer cells impedes migration and metastasis capabilities in vitro and in vivo 21 . In the current study, we evaluated the effects of ANXA1 signalling on macrophage polarization in the tumour microenvironment. Our data demonstrates that ANXA1 is able to polarize macrophages to an alternatively activated subtype (M2), which binds to the G protein-coupled receptor, FPR2, mediating cellular effects in paracrine manner. Furthermore, our in vivo experiments confirm the ability of tumor microenvironment or host ANXA1 to control tumour growth and metastasis.
CCL5/RANTES is part of the CC chemokine family protein that interacts with G protein-coupled receptors CCR1, CCR3 and CCR5 in many cell types 38,39 . CCL5 has been proposed to be a novel and promising therapeutic target for breast cancer 39,40 . The implications of the chemokine CCL5 promotes progression and metastasis is a subject of debate. A previous study suggested that host-derived CCL5 promotes breast cancer growth and metastasis by restraining the normal differentiation of myeloid-derived suppressor cells subsets in a 4T1 mammary carcinoma model that mimics the triple-negative breast cancer in patients 39 . As CCL5 is able to mediate cross-talk between the tumour cells and the tumour microenvironment 40 , it may be possible that CCL5 modulates macrophage polarization in our study, as we have shown that CCL5 is the major chemokine secreted by 4T1 carcinoma cells and is able to induce macrophage polarization to an alternatively activated subtype. This was prevented by CCL5 blockade which led to the skewing of TAM phenotype towards the classically activated or M1 phenotype. Furthermore, macrophages treated with recombinant CCL5 exhibited increased ANXA1 expression, suggesting a crosstalk between CCL5 and ANXA1. Recently, in a very elegant study, a novel pro-resolving mechanism revealed that ANXA1 plays a new role as a potent monocyte chemoattractant to orchestrate the resolving phase of acute inflammation 41 . Consistent with this, our study reveals a new dimension between ANXA1 and CCL5 to modulate macrophage polarization and tumour cell proliferation. Whether the effects of ANXA1 on macrophage polarization are related to their shared abilities with chemokines (CCL5) or other target molecules remains to be established.
Due to their tissue expression in invasive breast cancer, ANXA1 and FPR2 have become a target for drug discovery. We have shown previously that ANXA1 induces the constitutive activation of NF-κB and subsequent effects on migration and metastasis in breast cancer 21 . In addition, ANXA1 can enhance ERK and RhoA activity in breast cancer cells 42 . Previous studies in different cell types indicate that activation of FPR2 may stimulate a number of signalling pathways, including ERK1/2, PI3K and MAPK signalling 28,43 . Indeed, it has been reported that ANXA1 associated with NF-κB FPR2 activation by distinct ligands can trigger the GPCR-mediated signalling cascade to modulate cytokine signalling and tumour microenvironment, resulting in macrophage polarization and tumorigenesis 8 .
In recent years debate has surround between ANXA1 expression and function in cancer. Recent studies provide evidence linking ANXA1 as an endogenous inhibitor of NF-κB that can be induced by glucocorticoids and modified nonsteroidal anti-inflammatory drugs (NSAIDs) in colon and prostate cancer cells 44 . Our study demonstrates that FPR2 activation by 4T1-CM or ANXA1 peptide Ac2-26 activates ERK1/2-Akt-NF-κB, which in turn may facilitate macrophage polarization into an M2 subtype. Whether the effect of ANXA1 on NF-κB or other signalling molecules are related to cancer tissue specificity remains to be established. It is important to note that ANXA1 is highly expressed in metastatic and triple negative breast cancer cells 21 , where their migration is dependent on TAM phenotype and function. It is tempting to speculate that the analysis of a classically to alternatively activated TAMs spectrum is likely further controlled by temporal and spatial variables within the microenvironment. Therefore, tumours of different origin are heterogeneous in their interaction with the host based on their ability to release factors, such cytokines, into macro and micro-environment to elicit specific host responses, which may be critical for the tumour-promoting activity. Our study provides evidence that ANXA1 modulates TAM function and phenotype through FPR2-ERK signalling, involving CCL5 in the microenvironment, suggesting a potential new axis for targeting cancer therapy.

Methods
Animals. All animal work was approved by the Institutional Animal Care and Use Committee and followed National Advisory Committee for Laboratory Animals Research (NACLAR) approved Institutional Animal Care and Use Committee (IACUC) protocols at the National University of Singapore. All experiments were performed in accordance with relevant guidelines and regulations. BALB/c mice (8-12 weeks) were obtained from Laboratory Animal Centre (Singapore). ANXA1 −/− mice on BALB/c background (backcrossed over 10 generations) were a gift from Prof. Roderick Flower, QMUL, UK. MMTV-Wnt mice were obtained from Prof David Virshup Laboratory (Duke NUS, Singapore) and were maintained by backcrossing to C57BL/6. Mice were kept on a 12-h light/dark cycle with food and water provided ad libitum and maintained under pathogen-free conditions in the animal housing unit.
Cell Culture. Murine cell lines 4T1 and RAW 264.7 were obtained from American Type Culture Collection (ATCC, Manassas, VA, USA). Sublines from the same mammary tumour as 4T1, 67NR was a kind gift from Prof Jeal Paul Thiery, IMCB, A*STAR, Singapore. 4T1-12B cells stably transfected with a luciferase plasmid was a kind gift from Dr Gary Sahagian from Tufts University, USA. RAW 264.7 cells were grown as monolayers in Dulbecco's Modified Eagle's Medium (DMEM, Serana, Australia), while 4T1 and 67NR were cultured in Roswell Park Memorial Institute (RPMI, Serana, Australia) medium supplemented with 10% heat-inactivated fetal bovine serum (FBS, Biowest LLC, Kansas, MO, USA), 1% penicillin-streptomycin (GE Healthcare Life Sciences, HyClone Laboratories, Utah, USA) at 37 °C in a humid atmosphere containing 5% CO2. The cell lines were regularly authenticated through cell morphology monitoring, growth curve analysis and species verification. ANXA1 peptide ac2-26 and FPR2 antagonist WRW4 were obtained from Tocris Bioscences (Bristol, UK).

Isolation of Bone Marrow-derived macrophages (BMDM).
BMDMs from wild-type and ANXA1 −/− mice (8-12 weeks) were obtained by flushing the femurs and tibias of mice with DMEM media, as previously described 45 . Briefly, red blood cells were removed through osmotic lysis and the bone marrow cell suspension was washed twice with PBS, and cultured with BMDM media (DMEM medium containing 10% FBS (v/v), 100 U/mL penicillin, 100 µg mL −1 streptomycin, 2 mM L-glutamine and 20% L-929-conditioned DMEM (v/v) as a source of M-CSF). After 3 days of culture, the cells were supplemented with fresh BMDM media. At day 7, culture media and non-adherent cells were removed and the remaining adherent cells were replenished with fresh BMDM media before experiments.
Generation of tumour-conditioned media. 4T1 and 67NR murine carcinoma cells were cultured in RPMI media containing 10% FBS and 1% penicillin-streptomycin. After 3 days of culture and between 80-90% confluence, the media was removed and replaced with fresh media containing 1% FBS and 1% penicillin-streptomycin. After 24 h, the conditioned-media (CM) was collected and passed through a filtropure syringe filter membrane (0.2 µm) and stored at −80 °C prior experiments.
RNA extraction and RT-qPCR. Total RNA was extracted from the cells by using RNeasy kit (Qiagen, Limburg, Netherlands) according to the manufacturer's protocol. The quantitative and qualitative RNA analyses were performed by using NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, Massachusetts, USA).
Total RNA (1 µg) was used to synthesize cDNA by using a reverse transcription kit (GoTaq ® qPCR master mix, Promega Corporation, Madison, USA) as previously described 22 . The results were normalized to the expression of glyceraldehyde-3phosphate dehydrogenase (GAPDH). The amplification program was as follows: 95 °C for 5 min followed by 40 cycles of 95 °C for 10 s, 60 °C for 10 s and 72 °C for 10 s. The specificity of the assay was confirmed by melting curve analysis at the end of the amplification program. Primers for ANXA1, PPARγ, INOS, Arg1, IL12, Fizz, YM1 are described in supplemental file (Table S1).
Western Blot. Cells were washed twice with ice cold 1 × phosphate buffer saline (PBS). Proteins were extracted from the cells by using RIPA lysis and extraction buffer as previous described 22 . Protein concentration was estimated according to the Bradford's protein assay (BioRad Laboratories, Hercules, California, USA). Equal amounts of protein from each sample were subjected to 10% SDS-PAGE at a constant voltage of 125 V. The proteins were transferred onto nitrocellulose membranes (Bio-Rad, Hercules, CA, USA). Proteins were determined by Western blotting with specific antibodies, and expression signals were obtained by enhanced chemiluminescence. Protein expression was normalized to α-tubulin levels. Specific antibodies against phosphor-Akt-Ser 473, Akt, phosphor-ERK, ERK, phosphor-NFΚB p-65 (Cell signalling Technology), iNOS (Invitrogen), PPARγ (Santa Cruz), α-tubulin (Abcam) and Arg1 (Sigma) were used for immunoblot analysis.
Transwell co-culture assay. In vitro cell migration assay was performed using the Transwell system (24-wells, 8-μm pore size with polycarbonate membrane; Corning Costar, Lowell, MA, USA). Briefly, BMDM and 4T1 cells were harvested and suspended in serum-free media and 1 × 105 cells were added to the upper wells. The 4T1 and BMDM-conditioned media were pre-treated with IFNγ (Peprotech) + LPS (Sigma) or IL4 (R&D systems) and mixed with media (v/v 1:1), prior addition to the lower chamber. After 24 h, the cells attached to the lower surface were counted. The number of cells migrated were acquired in five randomized fields using an Olympus light microscope13 to obtain the invasion index.
In vivo bioluminescence imaging of 4T1 tumour. Sixteen mice were randomized into two groups (WT and ANXA1 deficient mice, 8 mice per group). The mice were subcutaneously injected with stably transfected 4T1-luciferase cells (7500 cells per mouse) into the mammary fat pad and mice were monitored for up to 40 days. The size of the tumours and tissue metastasis were measured by a bioluminescence imaging assay using the Xenogen IVIS Spectrum Iamging System (Caliper Life Sciences) and manual measurement by Vernier caliper. Tumour volume was calculated as (length × width × width/2). Mice were killed by anaesthetic overdose either at the end of the study or earlier if they displayed significant weight loss, signs of distress or palpable tumours ≥1.5cm in diameter.

Statistical analysis.
Results are the means ± SEM of three independent experiments performed in triplicate.
Statistical comparisons between groups were made by using one-way ANOVA and Bonferroni post-tests were performed for intergroup significance. Unpaired two-tailed Student's t-test wsa used for comparing two variables. The differences were considered statistically significant at *p < 0.05.