Apoptotic Cancer Cells Suppress 5-Lipoxygenase in Tumor-Associated Macrophages Free

The enzyme 5-lipoxygenase (5-LO) is key in the production of leukotrienes, a group of proinflammatory lipid mediators derived from arachidonic acid (AA). 5-LO converts AA into leukotrienes in a two-step catalysis. The first step is oxygenation to the intermediate 5(S)-hydroperoxy-6,8,11,14-eicosatetranoic acid, which can be further reduced by glutathione peroxidases to 5(S)-hydroxy-6-trans-8,11,14,-cis-eisosatetranoic acid. In a second step the unstable epoxide leukotriene A₄ (LTA₄) is formed through dehydration of 5(S)-hydroperoxy-6,8,11,14-eicosatetranoic acid. LTA₄ can either be transformed to leukotriene B₄ (LTB₄) by LTA₄-hydrolase or to cysteinyl leukotrienes (LTC₄, LTD₄, or LTE₄) by conjugation with glutathione by LTC₄-synthase (1, 2). The expression of 5-LO is mainly restricted to leukocytes. Notably, 5-LO is abundant in neutrophils and monocytes/macrophages (MΦ), which produce LTB₄ to recruit neutrophils and activate phagocytes at sites of inflammation. Production of cysteinyl leukotrienes by eosinophils, basophils, and mast cells constricts vascular smooth muscles in the respiratory tract during anaphylactic reactions. Aberrant formation of 5-LO metabolites is therefore tightly associated with several diseases such as asthma, arthritis, atherosclerosis, Alzheimer’s disease, and type 2 diabetes (3–6). Moreover, enhanced expression of 5-LO and 5-LO product formation is found in several types of cancer such as breast (7), prostate (8, 9), pancreatic (10, 11), and colon (12), as well as in cancer stem cell–driven acute myeloid leukemia (13). Functionally, 5-LO products increase cellular proliferation and suppress cancer cell apoptosis. MΦ are mononuclear phagocytes with essential roles in innate and adaptive immune responses. Their high plasticity allows them to respond to environmental stimuli such as tissue damage, hypoxia, allergies, and infection by differentiating into distinct functional phenotypes. For simplicity, MΦ are classified by different polarization states. The classically activated M1 phenotype, which is induced by IFN-γ alone or in combination with microbial LPS, induces a proinflammatory response. M1 MΦ are characterized by the production of TNF-α, IL-1β, and IL-6 and shape the host defense against microorganisms and tumor cells (14, 15). In contrast, the alternatively activated M2 phenotype is established by anti-inflammatory stimuli like IL-4 or IL-13. M2 MΦ are characterized by the production of large amounts of IL-10 as well as high expression of the MΦ mannose receptor CD206. In contrast to M1 MΦ, M2 cells contribute to resolution of inflammation. They take up cell debris and mediate tissue remodeling by inducing angiogenesis and repair (16). The M1/M2 nomenclature has been critically viewed due to heterogeneous differentiation and stimulation protocols and the use of variable marker profiles. For accuracy we therefore follow the nomenclature suggested in a recent article (17) and specify the stimulus used for MΦ activation, e.g., M(LPS+IFN-γ), or M(IL-4) to refer to stimulation with LPS+IFN-γ or IL-4, respectively.

A tumor is composed not only of cancer cells, but also extracellular matrix and a variety of stromal cells. Among the immune cell infiltrate, MΦ are the major leukocyte component of the tumor microenvironment (TME) that contributes to all stages of tumor progression (18, 19). Once recruited into the primary or metastatic site, MΦ polarize toward a protumoral phenotype (tumor-associated macrophages, TAMs) under the influence of a number of microenvironmental factors. TAMs possess anti-inflammatory and immunosuppressive properties, and promote angiogenesis and invasiveness (20). Accordingly, TAMs are an interesting therapeutic target for cancer treatment (19). Compared to the role of 5-LO in cancer cells, there is limited knowledge about the function of 5-LO specifically in the TME. We therefore aimed to investigate regulation of 5-LO in MΦs in a tumor context.

LPS, Ca²⁺ ionophore (A23187), CaCl₂, AA, staurosporine (STS), zymosan A, 5,6-dichlorobenzimidazole 1-β-d-ribofuranoside (DRB), and the 5-LO inhibitor BWA4C were purchased from Sigma-Aldrich (St. Louis, MO). IL-4 and IFN-γ were from Peprotech (Hamburg, Germany). LTB₄ was purchased from Biomol (Hamburg, Germany). The Mer tyrosine kinase (MerTK) blocking Ab (ab52968) and corresponding IgG control (ab125938) were ordered from Abcam (Cambridge, U.K.). Carboxyl Latex Beads for phagocytosis assay came from Thermo Fisher Scientific (Waltham, MA). CD36 short blocking peptide SLINKSKSSMF was synthesized according to Kar et al. (21). All reagents were dissolved following the manufacturer’s instructions.

Human MΦ were cultured in RPMI 1640 containing 5% AB-positive human serum (DRK-Blutspendedienst Baden-Würtemberg-Hessen, Frankfurt, Germany), 100 U/ml penicillin, and 100 μg/ml streptomycin. MCF-7 cells were obtained from American Type Culture Collection and cultured in RPMI 1640 containing 10% heat-inactivated FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, 1% sodium pyruvate, and 1% nonessential amino acids (PAA Laboratories, Coelbe, Germany).

Human MΦ were differentiated from PBMCs isolated from buffy coats of anonymous healthy donors obtained from DRK-Blutspendedienst using Ficoll gradient centrifugation (22). After culturing PBMCs for 1 h in RPMI 1640, supplemented with 100 U/ml penicillin and 100 μg/ml streptomycin, nonadherent cells were washed off and cells were cultured in MΦ media (RPMI 1640 containing 5% human serum) for 7 d to allow differentiation from monocytes toward MΦ. The density of MΦ was roughly 5 × 10⁵ cells per well in six-well dishes. M(LPS+IFN-γ) were obtained by stimulation with 100 ng/ml LPS and 100 U/ml IFN-γ for 24 h, whereas M(IL-4) were stimulated with 20 ng/ml IL-4 for 72 h. For a coculture of MΦ with MCF-7 cells, MCF-7 cells were harvested with trypsin-EDTA, washed once with PBS, and suspended in MΦ media. MΦ and cancer cells were cultured in a 1:1 ratio in MΦ media. To obtain pure MΦ at the end of the coculture, remaining MCF-7 cells were detached using trypsin-EDTA, and removed from the culture dish.

MΦ were harvested, washed with PBS, and pelleted at 500 × g at 4°C for 5 min. Cells were blocked with 2% Fc Receptor Binding Inhibitor (eBioscience, Frankfurt, Germany) in PBS for 10 min on ice. Afterwards, the following Ab mix was added in 100 μl PBS: anti–CD80-BV711, anti–CD86-FITC (both from BD Biosciences, Heidelberg, Germany), anti–CD163-APC, anti-CD206-PE-Cy7, and anti-CD326-BV421 (all from BioLegend, San Diego, CA), followed by incubation for 20 min on ice in the dark. Samples were washed and analyzed by flow cytometry using an LSRII Fortessa cell analyzer (BD Biosciences). CCL2, CCL4, CCL5, IL-6, IL-10, IL-8, IL-12 p70, IP10, and TNF-α in cell-culture supernatants were quantified using cytometric bead array Flex Sets (BD Biosciences). Samples were acquired by flow cytometry and processed with FCAP software V1.0.1 (BD Biosciences). CCL17 levels in cell-culture supernatants were quantified using the Legend Max ELISA Kit for human CCL17 (BioLegend).

RNA was isolated using the PeqGold protocol (Peqlab Biotechnologie, Erlangen, Germany). RNA was transcribed into cDNA using Fermentas Reverse Transcriptase Kit (Thermo Fisher Scientific). Real-time quantitative PCR was performed using the MyIQ real-time PCR system and SYBR green (both from Bio-Rad, Munich, Germany). For small cell numbers, RNA isolation and cDNA synthesis were carried out using the MessageBOOSTER cDNA Synthesis from Cell Lysates Kit (BioCat, Heidelberg, Germany) according to the manufacturer’s instructions. Primers for human 5-LO and proto-oncogene c-Myb (c-Myb) were from QuantiTect (Qiagen, Hilden, Germany). The following primers were used from Biomers (Ulm, Germany): human TATA-box binding protein sense 5′-GGGCCGCCGGCTGTTTAACT-3′ and antisense 5′-AGCCCTGAGCGTAAGGTGGCA-3′, human β-2-microglobulin sense 5′-ATCTGCCCTTTCCCGAGATCA-3′ and antisense 5′-CTCACGCGCTTTGTTTTGGT-3′, human 15-lipoxygenase1 sense 5′-GTATCGCAGGTGGGGAATTA-3′ and antisense 5′-GGACACTTGATGGCTGAGGT-3′, firefly luciferase (FL) sense 5′-GGTTCCATCTGCCAGGTATCAGG-3′ and antisense 5′-CGTCTTCGTCCCAGTAAGCTATG-3′, mouse 5-LO sense 5′-ATTGCCATCCAGCTCAACCA-3′ and antisense 5′-ACTGGAACGCACCCAGATTT-3′, and mouse ubiquitin-40S ribosomal protein S27a sense 5′-GACCCTTACGGGGAAAACCAT-3′ and antisense 5′-AGACAAAGTCCGGCCATCTTC-3′.

Cell pellets were incubated in 100 μl SDS-loading dye (50 mM Tris/HCl [pH 6.8], 2% SDS, 10% glycerol, 0.1% bromphenol blue, 10 mM DTT) and heated for 20 min at 60°C. Then 10 μl of the lysate was loaded on 10% SDS polyacrylamide gels. Proteins were transferred to nitrocellulose membranes, incubated with 5-LO Ab (#610694; BD Biosciences), and visualized by IRDye 680- and IRDye 800-coupled secondary Abs using the Li-Cor Odyssey imaging system (LICOR Biosciences, Bad Homburg, Germany). For immunoblotting of c-Myb, nuclear extracts of MΦ were isolated following the nuclear extraction protocol described by Wu et al. (23). Briefly, MΦ were harvested with ice-cold PBS containing 10 mM NaF and 0.5 mM PMSF. Cell pellets were suspended in two package volume of lysis buffer A (HEPES [pH 7.9], 1.5 mM MgCl₂, 10 mM KCl, 300 mM sucrose, 0.5% NP-40), kept on ice for 10 min, and centrifuged at 2600 × g for 30 s at 4°C. The supernatant containing the cytosolic fraction was removed and the pellet was suspended in two-thirds package cell volume of buffer B (HEPES [pH 7.9], 1.5 mM MgCl₂, 420 mM NaCl, 0.2 mM EDTA, 2.5% glycerol). The nuclei pellet was sonicated with 30% for 5 s and centrifuged afterward at 10,400 × g for 5 min at 4°C. Nuclear protein was determined and 200 μg of nuclear extracts were used as indicated above. For detection of c-Myb expression a c-Myb Ab (ab117635; Abcam) was used.

Cells were harvested, washed three times with PBS, and incubated with 2.5 μM Ca²⁺ ionophore (A23187) and 1 mM CaCl₂ in PBS for 15 min on ice. Following the addition of 20 μM AA, samples were incubated for 5 min at 37°C. Cells were pelleted by centrifugation (5 min, 4°C, 10,000 × g) and snap-frozen in liquid nitrogen. LTB₄ and 5(S)-HETE content of the extracted samples were analyzed employing liquid chromatography tandem mass spectroscopy (LC-MS/MS). The LC-MS/MS system comprised a 5500 QTrap mass spectrometer (Sciex, Darmstadt, Germany), equipped with a Turbo-V source operating in negative electrospray ionization mode, an Agilent 1200 binary HPLC pump and degasser (Agilent, Waldbronn, Germany), and an HTC Pal autosampler (Chromtech, Idstein, Germany) fitted with a 25 μl LEAP syringe (Axel Semrau, Sprockhövel, Germany). High-purity nitrogen for the mass spectrometer was produced by a NGM 22-LC/MS nitrogen generator (cmc Instruments, Eschborn, Germany). LTB₄-d4_, 5(S)-HETE-d8, LTB₄, and 5(S)-HETE were obtained from Cayman Chemicals (Ann Arbor, MI). Sample extraction was performed with liquid-liquid extraction. Therefore, cell pellets were resuspended in 200 μl PBS, gently mixed with 20 μl of internal standards (LTB₄-d4, 5(S)-HETE-d8 12 ng/ml), and extracted twice with ethyl acetate. The organic phase was removed at 45°C under a gentle stream of nitrogen. The residues were reconstituted with 50 μl of methanol/water (50:50, v/v), centrifuged for 2 min at 10,000 × g, and then transferred to glass vials (A-Z Analytik-Zubehör, Langen, Germany) prior to injection into the LC-MS/MS system.

Spheroids were generated from MCF-7 cells using the liquid overlay technique as described (24). Next, 5 × 10³ cells per ml were plated onto nonadherent 1% agarose-coated 96-well plates and incubated for 5 d. Primary human monocytes were isolated from human blood PBMCs by using CD14 microbeads (Miltenyi Biotec, Gladbach, Germany) and the AutoMACS Separator (Miltenyi Biotec). Cocultures with 1 × 10⁵ monocytes were induced and maintained for 5 d to allow monocyte infiltration.

For zymosan A–induced peritonitis, C57BL/6 mice were injected i.p. with 100 mg/kg zymosan A. After up to 3 d, mice were euthanized, and a peritoneal lavage was obtained. To gather peritoneal MΦ, the peritoneal lavage was seeded on six-well plates to let MΦ attach to the plate for 1 h. The attached MΦ were then used for RNA isolation. Furthermore, mice expressing the polyoma virus middle T oncoprotein (PyMT) under the mouse mammary tumor virus promoter (25) were bred into a C57BL/6 background to induce mammary carcinoma. In the PyMT model, the expression of the PyMT oncoprotein is restricted to the mammary epithelium, which results in the appearance of mammary tumors starting from 4 wk after birth in C57BL/6 mice and the occurrence of pulmonary metastases starting after 14 wk (26). Then 20 wk after birth, tumor-bearing PyMT mice were sacrificed and perfused with PBS. After perfusion, PyMT tumors and pulmonary metastases were isolated. For animal experiments the guidelines of the Hessian animal care and use committee were followed.

MCF-7 spheroids and PyMT tumors were fixed with 4% paraformaldehyde and paraffin embedded. Tissue samples of human invasive breast cancer were provided by the Cooperative Human Tissue Network and the Cancer Diagnosis Program, which are funded by the National Cancer Institute. Other investigators may have received specimens from the same subjects. Spheroids and tumor sections were stained and analyzed using the Opal staining system according to the manufacturer’s instructions (PerkinElmer, Rodgau, Germany). The following Abs were used for spheroid staining: 5-LO (sc-8885; Santa Cruz, Dallas, TX), CD45 (ab10558; Abcam), cleaved caspase-7 (CASP7; #8438; Cell Signaling, Cambridge, U.K.), and cMyb (ab117635; Abcam). PyMT tumor sections were stained with following Abs: 5-LO (sc-8885; Santa Cruz), adhesion G protein-coupled receptor E1 (F4/80) (14-4801-85; eBioscience), and pan-cytokeratin (PC) (ab27988; Abcam). Human breast tumor sections were stained with the following Abs: 5-LO (sc-8885; Santa Cruz), CD163 (ab182422; Abcam), and PC (ab7753; Abcam). Corresponding secondary HRP-coupled Abs were anti-mouse IgG-HRP, anti-rabbit IgG-HRP (both GE Healthcare, Freiburg, Germany), anti-goat IgG-HRP (Santa Cruz), and anti-rat IgG-HRP (Bioss Antibodies, Woburn, MA). Nuclei were counterstained with DAPI and slides were mounted with Fluoromount-G (SouthernBiotech, AL). Slides were imaged at 4× and 20× using Vectra3 imaging software and images were analyzed using inForm2.0 Software (PerkinElmer).

For luciferase activity assays, human MΦ were transiently cotransfected with 4 μg of 5-LO promoter expression plasmid pN0 (kindly provided by Prof. Dr. D. Steinhilber, Goethe-University, Frankfurt, Germany) or an empty control plasmid pGL3B, together with 0.4 μg Renilla luciferase control vector pRL-TK (Promega, Madison, WI) using ViromerRED transfection reagent (Lipocalyx, Halle, Germany) according to the manufacturer’s instructions. Then 1 d after transfection cocultures with MCF-7, cells were set up and maintained for 3 d. Luciferase activities in cell lysates were measured and FL activity was normalized to Renilla luciferase activity in the lysate. For luciferase reporter RNA expression assays MΦ were transfected with 2 μg of the following plasmids: the empty control plasmid pGL3B; pGL3B containing the 5-LO coding sequence (cds) (pGL3cds), pGL3B containing the 5-LO cds followed by the last four introns J-M (pGL3cdsInJ-M), pGL3B containing a deletion of 1600 bp of the 5-LO cds (pGL3cds-del1600), pGL3B containing a deletion of 1700 bp of the 5-LO cds (pGL3cds-del1700), pGL3B containing a deletion of 1900 bp of the 5-LO cds (pGL3cds-del1900), and pGL3B containing a deletion of 1600 bp of the 5-LO cds in addition to carrying a mutation for the predicted binding site of c-Myb (pGL3cds-del1600 mut c-Myb). Cocultures and RNA isolation procedures were performed as described above. Firefly luciferase expression was normalized to FL expression of samples transfected with the empty control plasmid pGL3B.

Deletion mutants pGL3cds-del1600, -del1700, and -del1900 and site-directed mutagenesis to produce pGL3cds-del1600 mut c-Myb were achieved by PCR using PfuUltra II Fusion HotStart DNA Polymerase (Agilent, Santa Clara, CA). The following primers were used to produce the deletion mutants: sense 5′-ATGGAAGACGCCAAAAACATA-3′ (used for all deletion mutants), pGL3cds-del1600 antisense 5′-TTGCCGTGTTTCCAGTTCTTT-3′, pGL3cds-del1700 antisense 5′-TCGCCGGTGATCCAGCGGTAGCAGGG-3′, and pGL3cds-del1900 antisense 5′-CTTGTCCAGCAGGTGCTTCTCGCT-3′. Site-directed mutagenesis of pGL3cds-del1600 to produce pGL3cds-del1600 mut c-Myb was achieved by using the following primer: sense 5′-GTCCTGAGGGATGGACGCGTACATTCGGCCCGAGATGACCAAATT-3′ and antisense 5′-AATTTGGTCATCTCGGGCCGAATGTACGCGTCCATCCCTCAGGAC-3′.

For migration of human T cells toward control MΦ and TAM coculture supernatants, PBMCs were isolated from buffy coats (DRK Blutspendedienst) using Ficoll gradient centrifugation (22). After culturing PBMCs for 1 h in RPMI 1640 supplemented with 100 U/ml penicillin and 100 μg/ml streptomycin, nonadherent lymphocytes were harvested. MΦ and TAMs were stimulated with Ca²⁺ ionophore (A23187, 2.5 μM) and AA (10 μM) for 30 min. Subsequently, cells were washed with PBS and 2 × 10⁶ PBMCs were added into Transwell inserts (6.5 mm Transwell with 5.0 μm pores; Corning, New York, NY). Lymphocytes were allowed to migrate for 1–4 h. For controls, MΦ were incubated prior to stimulation with A23187 and AA with the 5-LO inhibitor BWA4C (1 μM), which was added freshly after washing off A23187 and AA. In some experimental groups, TAM supernatants were spiked with LTB₄ (10 nM) after washing off A23187 and AA. Migration was determined using flow cytometry with Flow-Count Fluorospheres (Beckman Coulter, Krefeld, Germany) as an internal counting standard. The percentage of migration was determined comparing migrated versus nonmigrated cells. Abs used to discriminate lymphocytes were: anti–CD3-BV605 and anti–CD19-AlexaFluor700 (both BD Biosciences). Samples were acquired on a BD LSR II Fortessa flow cytometer and analyzed with FlowJo Software (Tree Star).

Bone marrow was isolated from tibia and femur of C57BL/6 mice, and 6 × 10⁶ bone marrow cells were incubated in RPMI 1640 containing L929 cell culture supernatant (20%). Cells were incubated for 7 d and medium was replaced every 2 d.

Tumor-bearing PyMT mice were sacrificed and breast tumors as well as lung metastases were isolated. Human breast tumor samples used in this study were provided by the University Cancer Center Frankfurt. Written informed consent was obtained from all patients and the study was approved by the Institutional Review Boards of the University Cancer Center Frankfurt and the Ethical Committee at the University Hospital Frankfurt (project number SGO-01-2014). The guidelines of the World Medical Association’s Declaration of Helsinki were followed. Tumors were dissociated using human and mouse Tumor Dissociation Kits and the GentleMACS (both from Miltenyi Biotec) using standard protocols. For FACS sorting of primary murine TAMs and resident macrophages (RMs), single-cell suspensions were stained with anti–CD326-BV711, anti–CD11b-BV650, anti–CD11c-AlexaFluor700 (all BD Biosciences), anti–CD45-VioBlue (Miltenyi), and anti–F4/80-PE-Cy7 (BioLegend) according to Olesch et al. (27). Cell suspensions were filtered through 30 μm cell strainers, diluted to ideal concentrations for cell sorting. CD11b^low TAMs and CD11b^high RMs were sorted into PBS and RNA was isolated. For FACS sorting of primary human TAMs, cell suspensions were stained with 7-aminoactinomycin D, anti–CD45-BB515, anti–CD19-AlexaFluor700, anti–MHC class II-PE-Cy7, anti–CD64-BV510, anti–CD326-BV421, anti–CD3-PE-CF594 (all from BD Biosciences), anti–CD11c-PerCP-Cy5.5, and anti–CD206-APC (both from BioLegend). TAMs were characterized as CD64, MHC class II, and CD11c positive. Cells were sorted in PBS and RNA was isolated as described above.

Data were analyzed using GraphPad Prism 5.0 (GraphPad Software, San Diego, CA). The p values were calculated using one-sample t test, Student t test, or ANOVA with Bonferroni postcorrection as stated in the figure legends. To check for Gaussian distribution, Kolmogorov–Smirnov or D’Agostino and Pearson omnibus normality tests were performed. Parametric or nonparametric tests were applied accordingly. Asterisks indicate significant differences between experimental groups (*p < 0.05, **p < 0.01, ***p < 0.001).

To analyze regulation of 5-LO in human primary MΦ in a tumor context, we followed a previously established coculture protocol with MCF-7 breast cancer cells that allowed MΦ to adopt a TAM-like phenotype (28) (Fig. 1A). This phenotype was corroborated by analyzing cell surface marker expression and supernatant contents 4 d after coculture, compared with in vitro–generated M(LPS+IFN-γ) and M(IL-4). The morphology of the different phenotypes is shown in Supplemental Fig. 1A. After removing cancer cells, cocultured MΦ were free of MCF-7 cells, as no CD326 expression was detected, expressed the classic inflammatory MΦ markers CD80 and CD86, the immunosuppressive MΦ marker CD206, as well as higher levels of CD163, indicating phagocytic activity (Fig. 1B, 1C). Likewise, cell culture supernatants of cocultured MΦ are neither indicative of classic inflammatory nor anti-inflammatory MΦs (Supplemental Fig. 1B–J). 15-LO1 mRNA, which is typically expressed in M(IL-4), was not detected in our in vitro–generated TAMs (Supplemental Fig. 1K). These findings substantiate previous reports that TAMs exhibit distinct orientations within tumors and are neither classically nor alternatively polarized (29, 30). However, MΦ constitutively express 5-LO to immediately produce leukotrienes when facing inflammatory stimuli (4, 31). In the MΦ/MCF-7 coculture 5-LO mRNA expression time-dependently decreased, starting after 2 d (Fig. 1D), whereas 5-LO protein disappeared after 4 d (Fig. 1E, 1F). Furthermore, we measured LTB₄ and 5(S)-HETE formation in these in vitro–generated TAMs. Compared to naive MΦ, TAMs significantly produced less LTB₄ and 5(S)-HETE after stimulation with Ca²⁺ ionophore and AA (Fig. 1G, 1H). Absolute values of LTB₄ were in the range of 0.1 ng per 10⁶ TAMs and 5(S)-HETE was produced in the range of 40 ng per 10⁶ TAMs. In conclusion, in vitro–generated TAMs lost 5-LO expression and, therefore, the ability to produce leukotrienes.

In our coculture model, MCF-7 cell apoptosis is needed to establish the TAM phenotype (28). To explore the mechanism of 5-LO downregulation, we first used MCF-7 cells that overexpress Bcl2 and thus, are resistant to cell death (28). When coculturing Bcl-2 overexpressing MCF-7 cells with MΦ for 4 d, 5-LO mRNA expression remained stable and comparable to control MΦ, whereas coculturing control MΦ with naive MCF-7 cells consistently reduced 5-LO mRNA (Fig. 2A) and protein (Fig. 2B, 2C). Next, we incubated MΦ with naive MCF-7 cells in Transwell inserts to prevent direct contact between MΦ and tumor cells, but to allow diffusion of soluble factors. Using this setup, 5-LO mRNA expression in MΦ remained high (Fig. 2A). As cancer cell death likely is a prerequisite for 5-LO downregulation in MΦ, we used STS for 3 h to induce cell death of MCF-7 cells. Likewise, coculturing these cells with MΦ for 2 d decreased 5-LO in MΦ. However, adding STS-killed MCF-7 cells into Transwell inserts for 2 d did not lower 5-LO mRNA expression in MΦ compared with controls. Furthermore, incubating MΦ with apoptotic cell conditioned media, generated from STS-killed MCF-7 cells, left expression of 5-LO mRNA unaltered (Fig. 2D). Conclusively, 5-LO downregulation in TAMs is triggered by a direct contact between MΦ and apoptotic tumor cells. This notion was substantiated by studies in mice subjected to zymosan A–induced peritonitis. In this model, inflammation peaked in the peritoneum 1 d after zymosan A injection, whereas 3 d after injection inflammation was largely resolved (32, 33). Resolution of inflammation in the zymosan A–induced peritonitis model requires the uptake of apoptotic neutrophils by peritoneal MΦ starting around day 1. Thus, peritoneal MΦ are expected to have been in contact with apoptotic cells when collecting them at day 3. Indeed, 5-LO mRNA expression decreased in peritoneal MΦ at day 3 compared with day 1, a time-point when apoptotic neutrophils are not abundant (Fig. 2E). Additionally, we confirmed our hypothesis that a 5-LO decrease in TAMs depends on their interaction with apoptotic cells (ACs) in an MCF-7 tumor spheroid model. CD14-positive monocytes were allowed to infiltrate spheroids for 5 d, followed by the detection of 5-LO, active CASP7 to mark apoptosis, and CD45 to identify monocytes/ MΦ by immunohistochemistry. Spheroids were segmented into the periphery and the core (Fig. 2F). The periphery contains mainly viable MCF-7 cells, whereas in the core most of the tumor cells are apoptotic, as indicated by expression of CASP7, or have undergone secondary necrosis. In this system, monocytes/MΦ are CD45 positive and appear equally in the periphery as well as the core region. Within the periphery they express 5-LO, although in the core region, where apoptotic tumor cells and monocytes/MΦ coexist, 5-LO expression was barely visible (Fig. 2F, 2G). Importantly, we found a significant negative correlation between the expression of CASP7 and 5-LO (Fig. 2H). These findings support our notion that the 5-LO decrease in MΦ only occurs upon their contact with apoptotic cancer cells.

Engulfing ACs causes a rearrangement of the cytoskeleton, which initiates downstream signaling cascades within phagocytes (34). We asked whether the 5-LO decrease in TAMs could be a consequence of the phagocytosis process per se. Therefore, we incubated carboxyl group-carrying latex beads, mimicking ACs, with MΦ for 6–96 h. However, phagocytosis of these beads did not alter 5-LO expression (Fig. 3A). Because a direct cell contact of ACs and MΦ was required for 5-LO downregulation, we analyzed critical receptor-ligand interactions required for recognition of ACs by phagocytes. CD36 is a scavenger receptor expressed on MΦ, which recognizes oxidized phospholipids on ACs (35). Blocking CD36 on MΦ in our coculture system did not prohibit the 5-LO decrease (Fig. 3B). Another MΦ receptor contributing to AC recognition and engulfment is MerTK (36). Attenuating the interaction between STS-killed MCF-7 cells and MerTK on MΦ, by adding a blocking Ab, prevented 5-LO downregulation in TAMs compared with an IgG isotype control (Fig. 3C). Thus, downregulation of 5-LO in TAMs is apparently mediated via MerTK.

We next determined the molecular mechanism of 5-LO downregulation in TAMs. First, we hypothesized that mRNA destabilization might contribute to the 5-LO decrease in TAMs. Therefore, we analyzed 5-LO mRNA stability after the MΦ/MCF-7 coculture using DRB to block transcription, which, however, did not destabilize 5-LO mRNA, neither in a 24 h nor a 96 h coculture period compared with control MΦ (Fig. 4A). This suggests transcriptional regulation of 5-LO in TAMs. Following this assumption, we first excluded the 5-LO promoter to be regulated because a 5-LO promoter construct (pN0) exhibited similar activity in control MΦ and TAMs in luciferase activity assays (Fig. 4B). In subsequent studies, utilizing reporter constructs containing cds (pGL3cds) and introns (pGL3cdsInJ-M) of the 5-LO gene, we demonstrated that 5-LO in TAMs was transcriptionally controlled by regulating elements within its cds (Fig. 4C). FL mRNA expression in MΦ of all reporter constructs normalized to pGL3B are shown in Supplemental Fig. 2A. By producing deletions of the cds in the reporter construct containing only the cds, we identified that the first three exons of 5-LO (plasmid pGL3cds-del1600) contained the responsible regulating element. Further truncation of the cds (plasmids pGL3cds-del1700 and pGL3cds-del1900) abrogated downregulation of reporter expression in TAMs compared with control MΦ (Fig. 4C), identifying a 100 bp region in the 5-LO cds to contain the regulating element. Next, we used in silico analysis to inspect the fragment. Using the PROMO 3.0 transcription factor binding site prediction tool (37), we found binding sites for a number of transcription factors such as NF-κB, CEBPβ, p53, and YY1 that could be excluded to be involved in 5-LO downregulation (data not shown). Besides, we also found a potential binding site for the transcription factor c-Myb. To investigate its involvement in 5-LO regulation we mutated this predicted binding site in pGL3cds-del1600 by exchanging 3 bp. The corresponding plasmid del1600 mut c-Myb showed increased expression of the reporter gene in TAMs compared with the nonmutated plasmid (Fig. 4D). Furthermore, c-Myb mRNA as well as protein were increased in cocultures of MΦ with naive MCF-7 cells after 4 d (Fig. 4E, 4F), whereas c-Myb mRNA expression remained low in cocultures containing Bcl2-overexpressing MCF-7 cells or when naive MCF-7 cells were added in Transwell inserts (Fig. 4E). Moreover, upregulation of c-Myb was significantly reduced by using the MerTK blocking Ab during cocultures with STS-killed MCF-7 cells (Fig. 4G). Immunohistochemistry of MCF-7 spheroids infiltrated with CD14 positive monocytes for 5 d revealed that c-Myb expression was exclusively found within the core region, where 5-LO expression in monocytes/ MΦ was low (Fig. 4H). Quantification of c-Myb–positive cells, comparing their abundancy in the periphery versus core regions showed significantly more c-Myb–positive cells within the core (Fig. 4I). This part of the spheroid also contains CASP7-expressing cells (Supplemental Fig. 2B–H). In conclusion, c-Myb expression is induced in TAMs, which is functionally connected to transcriptional repression of 5-LO.

5-LO products are known to recruit immune cells including T cells (3), and a reanalysis of the METABRIC dataset (38) showed that 5-LO expression in human mammary carcinoma strongly correlated with CD3 expression (Fig. 5A). Because 5-LO is not expressed by CD3-positive T cells, these data indicate that 5-LO products may recruit T cells to human tumors. To analyze functional consequences of the 5-LO decrease in our in vitro–generated TAMs, we performed a time-dependent T cell migration assay toward these cells. Therefore, we isolated lymphocytes from human blood donors and added these cells into modified Boyden chambers (Fig. 5B). Migrated cells were harvested and analyzed via multicolor FACS analysis (Supplemental Fig. 3A). Migration of T cells toward TAMs was significantly reduced compared with control MΦ at 4 h, when migration was high compared with earlier time points (Fig. 5C). Interestingly, treatment of MΦ with the 5-LO inhibitor BWA4C reproduced the effect of the MCF-7 cell coculture and supplementation of LTB₄ to TAMs increased T cell migration (Fig. 5D). To summarize, 5-LO downregulation in TAMs prevented T cell migration, an effect that was rescued by the addition of LTB₄.

Our findings indicated low 5-LO expression in TAMs in vitro. To confirm these observations in primary TAMs, we isolated distinct MΦ populations from murine PyMT mammary carcinomas. As controls, bone marrow–derived MΦ (BMDMΦ) from the same animals were generated. Compared to BMDMΦ, 5-LO mRNA expression in TAMs and mammary RMs was significantly reduced (Fig. 6A). The identity of MΦ in all populations was ensured by the sorting procedure (Supplemental Fig. 3B) and by analyzing F4/80 expression (data not shown). Besides analyzing 5-LO expression in TAMs from PyMT tumors at the mRNA level, histological sections of PyMT tumors and pulmonary metastases were prepared and stained for 5-LO, F4/80, and PC (Supplemental Fig. 4A, 4B). PyMT tumor cells expressed varying amounts of 5-LO in primary tumors and in lung metastases. However, in both tissues TAMs did not show any notable expression of 5-LO, which, however, was detected in control BMDMΦ with the same procedure (Fig. 6B). We also analyzed 5-LO expression in TAMs from primary human breast tumors. The gating strategy is shown in Supplemental Fig. 3C. Confirming data from the murine system, TAMs from human tumors also did not express 5-LO mRNA. In this case monocyte-derived MΦ isolated from human buffy coats served as controls (Fig. 6C). Moreover, we evaluated sections of human invasive breast tumors and normal breast tissue by immunohistochemistry (Fig. 6D). The sections were stained for 5-LO, PC, and CD163 (Supplemental Fig. 4C). We observed low 5-LO expression in the TME compared with tumor cells (Fig. 6E). Furthermore, 5-LO expression in MΦ from invasive breast cancer was low compared with 5-LO expression in normal breast tissue (Fig. 6F). In conclusion, neither primary human nor mouse TAMs express 5-LO, corroborating our in vitro data.

Tumors not only consist of malignant cancer cells, but also of tumor-infiltrating immune cells. Often, immune cells represent a significant proportion of the tumor mass. Among them are monocytes, which are recruited into the tumor, differentiate to TAMs, and support tumor growth, invasion, stroma remodeling, as well as angiogenesis (39). In this study we specifically addressed the role of 5-LO in TAMs, which to our knowledge has not been done before. We used a coculture setup of human MCF-7 breast cancer cells and human monocyte-derived MΦ. This setup has been established previously and characterized as a useful in vitro system to study tumor cell death with concomitant phagocytosis of apoptotic debris and subsequent MΦ polarization (28). Using this model, we observed polarization of MΦ toward TAMs, interestingly tied with downregulation of 5-LO. During their coculture with MΦ, MCF-7 cells undergo apoptosis, which recapitulates the situation within tumors, where apoptotic cancer cells are continuously abundant (40, 41). In our two-dimensional culture system, as well as in MCF-7 tumor spheroid cocultures, direct cell contacts of MΦ with apoptotic MCF-7 cells were required to decrease 5-LO in TAMs. This excluded soluble mediators within the cocultures to facilitate 5-LO downregulation. For this reason, we analyzed cell-cell contact-dependent interactions between phagocytes and apoptotic cells to explore the molecular mechanism of 5-LO regulation. MerTK is a receptor tyrosine kinase involved in the ingestion of apoptotic cell material and is mostly expressed in cells of the myeloid lineage, often residing in the TME (42). Following the uptake of apoptotic cells, MerTK suppresses innate immunity by preventing the release of proinflammatory cytokines (IL-12, IL-6), while inducing the production of wound-healing cytokines, e.g., IL-10, which enhances tumor MΦ polarization toward a wound healing/TAM phenotype, which is highly abundant in the TME. We now provide evidence that MerTK also mediates suppression of 5-LO, a proinflammatory enzyme, in TAMs. This feature might be a part of the tumor-supporting impact of MerTK to foster tumor progression. Based on our finding, a 5-LO decrease may be considered as a marker for TAM polarization.

We also determined molecular mechanisms provoking a 5-LO decrease in TAMs. In general, differentiation of leukocytes goes along with 5-LO upregulation. In cells with low 5-LO expression, i.e., progenitor cells, the 5-LO promoter is silenced by DNA methylation. Furthermore, 5-LO expression can be regulated by miRNAs in MΦ (4, 45). However, 5-LO downregulation in TAMs was neither mediated by mRNA destabilization nor transcriptionally regulated via the 5-LO promoter. Instead, we demonstrate transcriptional regulation of 5-LO in TAMs in the 5-LO cds. Regulatory elements located within the distal 5-LO gene were earlier shown to be responsible for 5-LO upregulation during calcitriol- and TGF-β–induced myeloid cell differentiation (46, 47). 5-LO downregulation in TAMs in our system was mediated by the transcriptional regulator c-Myb, potentially interacting with a predicted binding site in exon 3 of the 5-LO cds. However, several efforts of knocking down c-Myb in MΦ, as well as chromatin immunoprecipitation of c-Myb to prove specific involvement of c-Myb in 5-LO downregulation, failed due to various experimental issues. c-Myb is a nuclear proto-oncogene, a homolog of the avian myeloblastosis virus v-myb (48), and is primarily expressed in immature hematopoietic cells. In MΦ, downregulation of c-Myb is essential for their maturation from monocytes to MΦ (49). Along these lines, we failed to detect c-Myb expression in control MΦ, whereas in TAMs c-Myb was detected at the RNA level and in nuclear extracts at the protein level. c-Myb contributes to the development of leukemia (50) and several solid tumors, and previous works predominantly focused on c-Myb in cancer cells (51, 52). The role of c-Myb within cells of the TME has not yet been adequately addressed. Our data suggest that upregulation of c-Myb causes 5-LO repression in TAMs. A correlation between c-Myb and 5-LO expression has been reported earlier in the context of myeloid cell differentiation, suggesting that c-Myb suppresses 5-LO expression in undifferentiated myeloid precursor cells (53). Our findings indicate that activation of TAMs upregulates c-Myb. This process might shift MΦ toward a more undifferentiated state. It will be interesting to explore whether this is a general phenomenon for other MΦ/phagocyte activation protocols that downregulate 5-LO, such as treatment with IL-4 (54).

In TAMs 5-LO is downregulated upon recognition of apoptotic cancer cells via MerTK, provoking its transcriptional repression via c-Myb. Several past studies have focused on the role of 5-LO in cancer cells, but there is little information about the function of 5-LO in the TME. Inhibition of 5-LO in tumor-associated neutrophils in breast cancer abolishes their prometastatic activity and reduces lung metastasis (55). This implies a protumor role of 5-LO within the TME, which goes along with the finding that hematopoietic deletion of 5-LO decreases polyp formation in colon cancer. Notably, the recruitment of mast cells, which produce leukotrienes in an autocrine manner, was reduced in these tumors. Concomitantly, a decreased attraction of myeloid-derived suppressor cells enables antitumor cytotoxic T cells to mount their antitumor function (56). Contrary to the proposed protumor role of 5-LO in the TME, Poczobutt et al. (57) demonstrated an antitumor function of 5-LO in the TME. They showed increased tumor growth as well as liver metastasis in an orthotopic Lewis lung carcinoma syngraft model in 5-LO–deficient mice, proposing decreased T cell recruitment as the underlying mechanism. In line, we show that reduced 5-LO expression in TAMs decreased the recruitment of T cells in an in vitro migration assay, also suggesting an antitumor role of 5-LO within the TME. More recently Poczobutt et al. isolated myeloid cell subsets from lung tumors of immunocompetent mice, which were injected with Lewis lung carcinoma cells. One MΦ population, designated as MacB, was present in high numbers in tumor-bearing lungs. In healthy lungs only a few of these cells were abundant, suggesting this population is recruited from circulating blood monocytes through cancer cell–produced factors. Interestingly, this MΦ population expressed less 5-LO in tumor-bearing lungs compared with the same population within healthy lungs (58). This corresponds well with our data showing that 5-LO is downregulated in TAMs in vitro, but also in TAM populations of human and murine breast tumors. Considering the wealth of information that 5-LO expression is low in TAMs and RMs in mammary tumors, the physiological significance of this observation needs further investigation. Based on the findings that TAMs with low 5-LO expression had a reduced capacity to recruit T cells, we propose an antitumor role of 5-LO in TAMs within the TME. Therefore, attempts of cancer treatment with 5-LO inhibitors should critically consider these findings. Over the few last years, several studies have used 5-LO inhibitors for cancer therapy. For example, inhibition of 5-LO reduced tumor growth, whereas the addition of 5-LO products to cultured tumor cells increased proliferation and activated antiapoptotic pathways (13, 59). However, most of these studies used nonphysiological doses of 5-LO inhibitors and likely induced off-target effects (5). Thus, it seems justified that the abundancy of 5-LO and its products should be investigated more in detail in different types of tumors considering the cellular composition of the TME, to achieve selectivity toward 5-LO in cancer and associated immune cells. Moreover, we underscore the notion that inhibition of MerTK in the TME may increase antitumor immune responses (43).

We thank Praveen Mathoor, Margarete Mijatovic, and Gudrun Beyer for excellent technical assistance.

The authors have no financial conflicts of interest.