Alveolar Macrophages Drive Hepatocellular Carcinoma Lung Metastasis by Generating Leukotriene B₄

Hepatocellular carcinoma (HCC) is the sixth most common neoplasm and the third most frequent cause of cancer death worldwide (1). The median overall survival of HCC patients with an intrahepatic tumor has been prolonged with the advances in diagnostic modalities and multidisciplinary therapies such as surgical resection, transplantation, transcatheter arterial chemoembolization, percutaneous ablation, and chemotherapy, even in the case of recurrence (1). On the other hand, the prognosis for patients with extrahepatic metastases is extremely poor. Lung metastases develop through a hematogenous route and the incidence rate is 25–30% in the patients with malignant tumors at autopsy (2), whereas the rate is 41.6–43.6% in HCC patients (3). The median overall survival of HCC patients with lung metastasis is 5.9 mo, which is markedly shorter than that of HCC patients without lung metastasis (16.2 mo) (3). Hematogenous metastasis is a multistep process in which interactions between metastatic cancer cells and host resident cells are considered essential; however, the precise mechanism of interaction remains unclear.

There are two distinct types of macrophages in lung tissues, alveolar macrophages (AMs) and interstitial macrophages (IMs) (4). In several lung metastasis models, IMs are recruited from the circulation into the pulmonary interstitium, thereby helping tumor cells extravasate, survive, and grow in the lungs (5, 6), and therefore, they are designated as metastasis-associated macrophages. In contrast, AMs are tissue-resident cells lining the inner epithelial surface of the alveoli and playing a crucial role in lung development, surfactant homeostasis, and immune surveillance (7). AMs are mainly derived from fetal liver monocytes and self-renew throughout the life cycle after birth under a steady state (8–10). Until recently, AMs have not been considered to be involved in cancer metastasis (6). Sharma et al. (11) have proven that AMs secrete TGF-β in a premetastatic niche, thereby suppressing the antitumor T cell functions and promoting breast cancer metastasis to lungs. However, the precise roles of AMs in lung metastasis are still unclear.

Chronic inflammation plays a crucial role in the development and progression of cancer by generating various inflammatory mediators (12). Arachidonic acid (AA) is released by cytosolic phospholipase A2 from cellular membranes, predominantly into the cytosol, and is enzymatically metabolized through two major pathways, cyclooxygenases (COXs) and lipoxygenases (LOXs), generating various biologically active eicosanoids with inflammatory activities, such as PGs, leukotrienes (LTs), and hydroxyeicosatetraenoic acids (HETEs) (13). A number of studies have also unraveled involvement of these metabolites in cancer development and metastasis via a myriad of cellular processes (14–16). However, these eicosanoids were analyzed individually, and thus, their comprehensive analysis is still required, particularly in lung metastasis.

In this study we conducted a comprehensive determination of AA-derived lipid mediators in lung metastasis induced by i.v. injection of HCC cells. Moreover, to our knowledge we obtained the first definitive evidence of an essential contribution of AMs and IMs to the progression of lung metastasis. IMs in lung metastatic foci produced an inflammatory chemokine, CCL2, and eventually recruited blood-borne CCR2–expressing AMs into the lungs. AMs expressed 5-LOX, enabling the generation LTB₄ with a potent inducer of tumor cell proliferation, thereby promoting the growth of tumor cells in lungs.

Specific pathogen-free 6–7 wk-old male BALB/c mice were purchased from Charles River, Japan and designated as wild-type (WT) mice. CD45.1 BALB/c congenic mice were obtained from the Jackson Laboratory (Bar Harbor, ME). CCR1-deficient (CCR1^−/−) (17), CCR2^−/− (18), and CCL2^−/− mice (19) were prepared as described previously and were backcrossed to BALB/c mice for more than eight generations. All mice were kept under specific pathogen-free conditions. All the animal experiments in this study complied with the Guidelines for the Care and Use of Laboratory Animals of Kanazawa University and were approved by the local ethical committee on the animal experiments.

The following Abs were used as the primary Abs for immunohistochemical analyses: rat anti-CD3 Ab and rat anti-F4/80 Ab (Serotec), rat anti-Ly6G Ab and rat anti-CD11b Ab (BD Biosciences), rat anti–Ki-67 Ab (Dako), hamster anti-CCL2 biotinylated Ab and Alexa Fluor 488–labeled hamster anti-CD11c Ab (eBioscience), mouse anti-human CD68 Ab (Leica Biosystems), and rabbit anti–5-LOX Ab (Abcam). Isotype-matched control IgGs for individual mouse (Dako), rat (BD Biosciences), rabbit (Sino Biological), and hamster (BioLegend) mAbs were purchased. The following Abs were used as the primary Abs for the flow cytometric analysis: rat anti-CD11b (M1/70) and hamster anti-CD11c (HL3) (BD Biosciences), rat anti-CD45 (30-F11) (eBioscience), rat anti-CD45.1 (A20), rat anti-CD45.2 (104), rat anti-CD115 (AFS98), and rat anti-F4/80 (BM8.1) (Tonbo Biosciences), rat anti-mouse CCR1 (643854) and rat anti-mouse CCR2 (475301) (R&D Systems), and hamster anti-mouse CCR5 (HM-CCR5) (BioLegend). Isotype-matched control IgGs for individual rat and hamster mAbs were purchased from BD Biosciences.

A mouse HCC cell line, BNL 1ME A.7R.1 (BNL), was purchased from the American Type Culture Collection and maintained as described previously (20). Human HCC cell lines Huh7 and PLC/PRF/5 were purchased from Japanese Collection of Research Bioresources Cell Bank and another human HCC cell line, Hep3B, was purchased from the American Type Culture Collection; both were maintained in DMEM (Sigma Chemical) as described previously (21). A mouse mammary carcinoma cell line, 4T1, was obtained from American Type Culture Collection and maintained as described previously (22).

Subconfluent BNL or 4T1 cells were collected and resuspended in PBS at a cell density of 1 × 10⁶ cells per 200 μl or 5 × 10⁴ cells per 200 μl, respectively. The cell viability was always >95% by a trypan blue exclusion test. BNL or 4T1 cell suspensions (200 μl) were injected into the tail vein of untreated BALB/c, AM-depleted WT mice, AM-transferred WT mice, or bone marrow (BM) chimeric mice. In some experiments, mice were administered i.p. with zileuton (Cayman Chemical) dissolved in PBS with 40% DMSO or aspirin (Sigma-Aldrich) dissolved in PBS with 20% DMSO at the indicated doses, once every day for the indicated time intervals starting at 12 d after the tumor cell injection. An LTB₄ receptor antagonist, (5-[2-(2-carboxyethyl)-3) 3-[[(E)-6-(4-methoxyphenyl)-5-hexenyl]oxy]phenoxy] pentanoic acid (ONO-4057) (Ono Pharmaceutical, Japan) (23), was dissolved in sodium hydrogen carbonate at 7.5 mg/ml and administered orally once daily for 9 d at a dose of 30 mg/kg starting 12 d after BNL cell injection. At the indicated time points after tumor cell injection, lungs were removed for the determination of the numbers and sizes of tumor foci on the lung surface, and for the following analyses.

Cancer cell suspensions (1.5 × 10³ per 100 μl) were added to each well of 96-multiwell plates and incubated at 37°C for 24 h. LTB₄, an LTB₄ receptor antagonist, LY293111, and LTC/D/E₄ mixture (cysteinyl LT HPLC mixture I) were obtained from Cayman Chemical. The cells were then treated with the indicated concentrations of LTB₄, 15 μM LY293111, or 1 μM LTC/D/E₄ mixture for additional 72 or 144 h. The cell viability was determined using the cell counting kit 8 (Dojindo), a modified MTT assay. The ratios of cell numbers were determined by comparing the OD value of untreated or vehicle-treated cells.

Liquid chromatography–mass spectrometry/mass spectrometry–based lipidomic analyses were performed as described previously (24, 25) for lungs isolated from untreated WT mice or those that received i.v. BNL cells 21 d earlier. Lipid fractions were obtained by solid-phase extraction with a Sep-Pak C18 cartridge (Waters) with deuterium-labeled internal standards (AA-d8, LTB4-d4, and 15-HETE-d8). Samples were analyzed by a triple quadrupole linear ion trap mass spectrometer (QTRAP5500; SCIEX) equipped with a 1.7 mm, 1.0 × 150 mm Acquity UPLC BEH C18 column (Waters). The mass spectrometry/mass spectrometry analyses were performed in negative ion mode, and the eicosanoids were identified and quantified by using multiple reaction monitoring. Calibration curves and the liquid chromatography retention times for each compound were constructed with synthetic standards.

Frozen lung tissues were homogenized with cold methanol and eluted for 16 h at −20°C. The lipid extraction process was performed with MonoSpin C18-AX (GL Sciences) according to the manufacturer’s instructions. The eluate was dried and reconstituted in enzyme immunoassay buffer. In another experiment, AMs and IMs were isolated from lung tissues using a FACSAria cell sorter, and the cells were cultured at a concentration of 5.5 × 10⁴ cells per 50 μl in FCS-free RPMI 1640 at 37°C for 6 h to determine LTB₄ content in the supernatants. LTB₄ levels were measured using the LTB₄ EIA Kit (Cayman Chemicals) according to the manufacturer’s instructions.

The following BM chimeric mice were prepared as described previously (26). BM cells were collected from the femurs and tibia of CD45.1⁺ WT, CD45.2⁺ WT, CD45.2⁺ CCR1^−/−, and CD45.2⁺ CCR2^−/− mice, by aspiration and flushing, and were suspended in PBS. Recipient CD45.1⁺ WT mice were lethally irradiated to 5.5 to 6.0 Gy using an X-ray irradiator (MBR-1520R-3; Hitachi). Then, the mice were injected intravenously with 4 × 10⁶ BM cells consisting of a 1:1 mixture of CD45.1⁺ WT and CD45.2⁺ WT mice or CD45.1⁺ WT and CD45.2⁺ CCR1^−/− mice, or 5 × 10⁶ BM cells consisting of a mixture of 1:4 of CD45.1⁺ WT and CD45.2⁺ CCR2^−/− mice. Additionally, BM chimeric mice were generated between WT or CCL2^−/− mice by injecting i.v. 2 × 10⁶ BM cells collected from donor mice to the irradiated recipient ones.

Under anesthesia, the trachea was exposed via a midline neck incision and cannulated with a 30 gauge needle. Mice were then intratracheally administered with 100 μl of clodronate liposomes (CLLs) (Cosmo Bio), which were diluted at a ratio of 1:1 in PBS. Control mice received the same volume of PBS.

Bronchoalveolar lavage (BAL) fluid was isolated by cannulation of the trachea and subsequent flushing of the lungs five times with cold 1 ml MACS buffer (PBS supplemented with 2 mM EDTA and 3% FBS) from untreated WT mice or those that received i.v. BNL cells 21 d ago. Isolated cells were resuspended in DMEM containing 10% FBS, 0.1 mmol/l nonessential amino acids, 1 μmol/l sodium pyruvate, 2 mmol/l l-glutamine (3 × 10⁵ per 1 ml) (Life Technologies), and were added to each well of 12-multiwell plates and incubated at 37°C for 2 h. After the nonadherent cells were discarded, adherent cells were used as the AM-enriched fraction. In another series of experiments, erythrocytes were depleted from BAL-derived cell suspensions with ammonium chloride lysing buffer containing 0.826% NH₄Cl, 0.1% KHCO₃, and 0.004% EDTA-4Na. The resultant single-cell preparations were further incubated with FITC-conjugated rat anti-mouse F4/80 mAb followed by anti-FITC microbeads (Miltenyi Biotec), according to the manufacturer’s instructions. F4/80⁺ cells were sorted by using an AutoMACS Pro Separator (Miltenyi Biotec). The resultant cell preparations were measured as more than 95% F4/80⁺ cells with flow cytometric analysis (data not shown), and were used as the highly enriched AM fraction.

The AM-enriched fraction was obtained from WT mice, which were untreated or injected with BNL cells 21 d ago and incubated in each well of 12-multiwell plates. BNL cells were labeled with 2 μM of CFSE (Life Technologies) and were added at a concentration of 2 × 10⁴ cells per 100 μl to the wells, in the presence or the absence of adherent AM-enriched fraction. Then, the cells were incubated in the presence or absence of 15 μM LY293111 for 24 h. The cells were collected and stained with PerCP-conjugated rat anti-mouse CD45 mAb. The proliferation of BNL cells was evaluated by the mean fluorescent intensity of tumor cells (CD45⁻) by flow cytometry.

Highly enriched AMs derived from CD45.2 WT mice were suspended in PBS at a cell density of 5 × 10⁶ cells per ml and transferred into CD45.1 mice through the trachea under anesthesia.

Resected lung tissues were fixed in 10% formaldehyde (Wako), embedded in paraffin, cut at a 3 μm thickness, and stained with H&E solution. In parallel, 3 or 5 μm thick sections were prepared and stained for immunohistochemical analysis as described previously (27). For Ag retrieval, the deparaffinized slides were either treated with 0.1% trypsin solution for 15 min at 37°C, autoclaved in 10 mM citrate buffer (pH 6) for 15 min at 121°C, or autoclaved in Bond Epitope Retrieval Solution 2 (Leica Biosystems) for 10 min at 100°C. Endogenous peroxidase activity was blocked using 0.3% H₂O₂ for 15 min, followed by incubation with Blocking One Histo (Dako) (27). The sections were further incubated with the optimal dilution of the Abs. The resultant immune complexes were detected by the ABC Elite kit (Vector Laboratories) and peroxidase substrate 3-3′-diaminobenzidine kit (Vector Laboratories), according to the manufacturer’s instructions. The samples were examined with a microscopy system (BZ-X700). Images were obtained with the BZ-X700 microscope and were quantified by Keyence Analysis Software. Ki-67⁺ cells in the area of metastatic foci were measured on five randomly chosen visual fields at ×40 magnification.

Cryostat sections 6 μm thick were fixed with cold methanol for 10 min and incubated with Blocking One Histo (Dako). The sections were further incubated with the optimal dilution of the Abs. For a double-color immunofluorescence analysis, Alexa Fluor 488 donkey anti-rat or Alexa Fluor 594 donkey anti-rabbit or hamster Abs were used as secondary Abs (Invitrogen). Immunofluorescence was detected in the setting that excluded the nonspecific signal of the isotype control, using a microscopy system (BZ-X700).

Single-cell suspensions were prepared by cutting lung tissues into small pieces, and treating them with 1 mg/ml collagenase D (Roche) and 40 μg/ml bovine pancreas-derived DNase I (Sigma-Aldrich) for 20 min at 37°C with shaking at 180 rpm. The resulting digestion mixtures were filtered through a 100 μm cell strainer (Corning) to obtain single-cell suspensions. Cell suspensions from lung tissues and BAL fluid were further treated with ammonium chloride lysing buffer to deplete erythrocytes. Isolated single-cell suspensions were stained with various combinations of fluorescent dye–conjugated Abs in MACS buffer. Dead cells were removed from acquired data with a fixable viability dye (eBioscience). Expression of each molecule was determined using FACSCanto II (BD Biosciences) and analyzed with FlowJo software (Tree Star). IMs were defined as CD45⁺F4/80⁺CD11b^highCD11c⁻ in lung tissue and were sorted using a FACSAria cell sorter (BD Biosciences).

Quantitative real-time PCR (qRT-PCR) was conducted on total RNAs extracted from cells or lung tissues using the Power SYBR Green Master Mix (Life Technologies) and the primers listed in Table 1. Expression levels of the target genes were analyzed using the ΔΔ cycle threshold comparative method. The Gapdh and β2m genes were used for macrophage and lung tissues as an internal control, respectively.

BAL fluid was obtained 24 h after the mice were injected i.p. with 1 mg 5-ethynyl-2-deoxyuridine (EdU) (Invitrogen) and EdU incorporation was evaluated by using Click-iT Plus EdU Flow Cytometry Assay Kits (Invitrogen) according to the manufacturer’s instructions.

Lung tissue samples were obtained from nine autopsied HCC patients, who died at University of Fukui Hospital between April 2007 and October 2015; five patients had lung metastases and four did not. The tissues were paraffin embedded and used for histological and immunohistochemical analyses using anti-human CD68 or anti-human 5-LOX Ab. A double-color immunohistochemical analysis was conducted to detect CD68 and 5-LOX using an ABC-AP kit (Vector Laboratories) and Vector Blue Alkaline Phosphatase Substrate (Vector Laboratories) in the presence of levamisole (Vector Laboratories) according to the manufacturer’s instructions. CD68- or 5-LOX–expressing cells in the alveolar spaces were enumerated on five randomly chosen visual fields at ×100 magnification. This study was conducted in accordance with the Declaration of Helsinki, and the study design was approved by the Ethics Committee of the University of Fukui (registration no. 20160145).

Total RNAs were extracted from the AM fraction of WT mice, which were untreated or injected i.v. with BNL cells 21 d previously, using an RNeasy Mini Kit (Qiagen), and their quality was confirmed by using an Agilent 2100 Bioanalyzer (Agilent Technologies). Then, 100 ng of total RNA was mixed with 2 μl of anchored oligonucleotide-dT primer (10 μM, 5′-AAGCAGTGGTATCAACGCAGAGTACAGCAGT₃₀VN-3′) and 0.8 μl of 2'-deoxynucleoside 5'-triphosphate mix (25 mM; Invitrogen), denatured at 72°C for 3 min, and immediately placed on ice. Then 20 μl of the first-strand reaction mix, containing 1 μl SuperScript II reverse transcriptase (200 U/μl; Invitrogen), 0.5 μl RNasin RNase Inhibitor (40 U/μl; Promega), 4 μl Superscript II First-Strand Buffer (5×; Invitrogen), 1 μl DTT (100 mM; Invitrogen), 4 μl betaine (5 M; Sigma), and 1 μl template switching oligonucleotide [20 μM, 5′-GCGGCTGAAGACGGCCTATGTGGG(L)-3′] were added to each sample. Reverse transcription reaction was carried out by incubating at 42°C for 90 min, followed by 10 cycles of (50°C for 2 min and 42°C for 2 min). Finally, the reverse transcriptase was inactivated by incubation at 70°C for 15 min (28). After reverse transcription, PCR was then carried out adding 25 μl KAPA HiFi HotStart ReadyMix (2×; KAPA Biosystems), 1.5 μl biotin primer (10 μM, 5′-/5BiotinTEG/AAGCAGTGGTATCAACGCAGAGT-3′; IDT), 1.5 μl primer (10 μM, 5′-GCGGCTGAAGACGGCCTATGT-3′; IDT), and 2 μl nuclease-free water (Life Technologies) to a final reaction volume of 50 μl. The reaction was incubated at 98°C for 3 min, then cycled 18 times between (98°C for 20 s, 60°C for 20 s, and 72°C for 6 min), with a final extension at 72°C for 10 min. PCR was purified using a 1:1 ratio of AMPure XP beads (Beckman Coulter), with the final elution performed in 50 μl of buffer EB (Qiagen) (28). Purified cDNA was digested with anchoring enzymes (NlaIII), and resulting fragments were bound to streptavidin-coated beads (Dynabeads streptavidin M-280), and nonbiotinylated cDNA fragments were removed by washing. Adapter-1 (5′-ACACTCTTTCCCTACACGACGCTCTTCCGATCTCAGCAGCATG -3′, 5′-/Phos/CTGCTGAGATCGGAAGAGCGTCGTGTAGGGAAAGAG TGT/AmMO/-3′) was ligated to cDNA fragments on the beads and after washing digested with EcoP15I. EcoP15I-digested and released fragments (adapter-1- tags) were ligated to adapters-2 (5′-/Phos/AGATCGGAAGAGCACACGTCTGAACTCCAGTCAC/AmMO/-3′, 5′-GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT-3′). Tags sandwiched between two adapters were amplified by PCR using PhusionHigh polymerase and GEX primers (5′-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTT CCGATCT-3′ and 5′-CAAGCAGAAGACGGCATACGAGATXXXXXXGTGACTGGAGTTC-3′, X; barcodes). The PCR regimen consisted of 98°C for 30 s, 10–15 cycles at 98°C for 10 s, 60°C for 30 s, and 72°C for 30 s. PCR products were run on an 8% non-denaturing polyacrylamide gel. After staining with SYBR green (Takara Bio), the band at 155–160 bp was cut out from the gel, and DNA purified after its elution from the gel pieces. The PCR product from each sample was analyzed on an Agilent Bioanalyzer 2100. Equal concentrations of PCR products from all the samples were mixed and applied to Illumina MiniSeq sequencing (Illumina). Libraries were sequenced paired-end with 26–30 cycles. All datasets have been deposited at National Center for Biotechnology Information, Gene Expression Omnibus under accession number GSE106987 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE106987).

The means ± SD were calculated for all parameters determined. Statistical significance was evaluated with GraphPad Prism software version 6 using one-way ANOVA, followed by Tukey–Kramer post hoc test or Mann–Whitney U test. A p value <0.05 was considered statistically significant.

After BNL cells were injected intravenously, metastatic foci progressively grew in the lung tissues (Supplemental Fig. 1A). Comprehensive determination revealed that almost all AA-derived lipid mediators in the COX-, 5-LOX–, and 12/15-LOX–dependent pathways or nonenzymatic pathways increased in the lungs 21 d after the BNL cell injection (Fig. 1).

Accumulating evidence indicates a crucial contribution of 12/15-LOX and its metabolites, 12-HETE and 15-HETE, to lung metastasis (14, 29). On the contrary, the roles of metabolites from the COX and 5-LOX pathways still remain unclear in lung metastasis. Hence, we treated mice bearing lung tumor metastasis with a 5-LOX inhibitor, zileuton, or a COX inhibitor, aspirin, to examine the pathophysiological roles of these pathways. Zileuton, but not aspirin, reduced the numbers of metastatic tumor foci, particularly those with a diameter larger than 2.0 mm (Fig. 2A–C, Supplemental Fig. 1B). Zileuton further reduced the number of tumor cells expressing Ki-67, a cellular marker for proliferation, in metastatic foci (Fig. 2D, Supplemental Fig. 1C). Zileuton similarly reduced lung metastasis formation arising from an i.v. injection of a mouse breast cancer cell line, 4T1 (Supplemental Fig. 1D, 1E). BNL cells exhibited abundant mRNA expression of LTB₄ receptors, Blt1 and Blt2, but scarce mRNA expression of LTC/D/E₄ receptors, Cyslt1 and Cyslt2 (Fig. 2E, Table I). Consistently, LTB₄ but not an LTC/D/E₄ mixture augmented in vitro proliferation of BNL cells, and their enhanced proliferation was inhibited by a specific LTB₄ receptor antagonist, LY293111 (Fig. 2F). Likewise, Huh7, Hep3B, and PLC/PRF/5 human HCC cell lines also showed abundant mRNA expression of LTB₄ receptors (Fig. 2G), and proliferated in response to LTB₄ (Fig. 2H). Moreover, the administration with an LTB₄ receptor antagonist, ONO-4057, reduced the numbers of lung metastasis foci (Fig. 2I, 2J), similar to the administration of the 5-LOX inhibitor (Fig. 2A, 2B). Thus, LTB₄ can directly support the proliferation of HCC cells, eventually promoting lung metastasis.

Wculek and Malanchi have revealed that neutrophil-derived LTs in the premetastatic niche could support the colonization of lungs by breast cancer cells (16). On the contrary, in the present lung metastasis model, immunohistochemical analysis revealed that F4/80⁺ macrophages were the predominant cell type present in lung metastatic foci, as compared with Ly6G⁺ granulocytes or CD3⁺ T lymphocytes (Fig. 3A). Lung macrophages can be classified into two distinct subsets, IMs and AMs (6), which can be defined as CD45⁺F4/80⁺CD11b^highCD11c⁻ and CD45⁺F4/80⁺CD11b^lowCD11c⁺ populations, respectively (Fig. 3B). As lung metastasis progressed, IMs were accumulated in the lungs (Fig. 3C), consistent with a previous report (5). Likewise, AMs accumulated in the total lung including tumor and nontumor areas (Fig. 3C) as well as in the BAL fluid (Fig. 3D). Hence, we next examined the expression levels of enzymes participating in AA metabolism in these two types of macrophages. IMs expressed higher mRNA levels of several PG-synthesizing enzymes such as Cox2, Ptgis, and Ptgds than those expressed by AMs (Fig. 3E). On the contrary, AMs, particularly those present in lung metastatic foci, exhibited a higher mRNA expression level of the pivotal LTB₄-synthesizing enzyme, 5-lox, than that of IMs (Fig. 3E). Moreover, CD11c⁺ AMs expressed the 5-LOX protein, whereas it was barely detected in CD11b^highCD11c⁻ IMs in the lung with tumor metastasis (Fig. 3F). Consistent with a high expression of the synthetic enzyme, AMs present in lung metastasis foci secreted more LTB₄ than IMs in the same regions (Fig. 3G). Furthermore, the examination of the lungs of HCC patients revealed that the numbers of 5-LOX– and CD68-epxressing cells were higher in the lungs of HCC patients with lung metastasis than in those of HCC patients without lung metastasis (Fig. 3H, Supplemental Fig. 2). Moreover, most CD68⁺ macrophages in the alveolar space expressed 5-LOX (Fig. 3I). Thus, 5-LOX–expressing CD68⁺ AMs may have a similar pathogenic role in human HCC metastasis to lungs as in the mouse lung metastasis model. Intratracheal injection of CLLs selectively depleted AMs but not IMs only until 6 d after the injection (Fig. 4A). Hence, we administered CLLs at different time points (pre-, early, and late phases as depicted in Fig.4B) to examine the roles of AMs at different windows of lung metastasis formation. Depletion of AMs at the late phase but not at the pre- or early phase reduced the numbers of metastatic foci, particularly those with a diameter larger than 2.0 mm (Fig. 4C, 4D), similar to the administration of the 5-LOX inhibitor (Fig. 2A, 2B). Simultaneously, the CLL administration at the late phase significantly decreased the LTB₄ content in the lungs (Fig. 4E). Furthermore, in vitro BNL cell proliferation was significantly enhanced when the cells were cocultured with AMs harvested from metastatic lungs (Fig. 4F) but not with those from untreated lungs (data not shown), and the enhancement was abrogated by treatment with an LTB₄ receptor antagonist, LY293111 (Fig. 4F). Thus, AMs can directly promote the tumor cell growth in metastatic foci of lungs by secreting LTB₄.

Accumulating evidence indicates that AMs are mostly maintained through self-renewal in a steady state (8–10). Thus, enhanced self-renewal might account for the increase of AMs in lungs with metastasis. Contrary to our expectation, an EdU incorporation assay revealed that EdU incorporation was not significantly enhanced in AMs harvested from metastatic lungs compared with those harvested from untreated mice (Supplemental Fig. 3A). Moreover, CD45.2 donor-derived mature AMs were not significantly increased in metastatic lungs when mature AMs were injected intratracheally (Supplemental Fig. 3B, 3C). These observations would indicate that AM accumulation in the metastatic lung did not arise from the local proliferation of mature AMs. Because the total body irradiation-induced decrease in AMs could be supplemented with blood-borne precursors (30), we next examined whether AMs could newly infiltrate from the bloodstream into metastatic lung tissues. Serial analysis of gene expression, together with gene set enrichment analysis, revealed that AMs harvested from metastatic lungs exhibited a gene signature similar to that previously reported for blood-borne AMs (30) (Fig. 5A). Analysis of gene expression of major macrophage-tropic chemokines, including CCL2, CCL3, CCL4, and CCL5, revealed that Ccl2 mRNA expression significantly increased in the lungs later than 14 d after the tumor cell injection (Fig. 5B), when the intralung AM accumulation became obvious (Fig. 3C). On the contrary, Ccl3 and Ccl4 but not Ccl5 mRNA expression was only observed later (Fig. 5B). Moreover, AMs expressed CCR1 and CCR2 but not CCR5 among receptors for CCL2, CCL3, and CCL4 (Fig. 5C). Hence, we performed competitive BM transplantation to elucidate the involvement of CCR1 or CCR2 in the recruitment of AMs from the bloodstream (Fig. 5D). Consistent with the previous report describing the crucial role of CCR2 on the replenishment of AMs with blood-borne monocytes after BM transplantation (8), the chimerism of CCR2^−/− mouse-derived CD45.2⁺ donor cells was slightly reduced spontaneously in the absence of metastasis in the lungs, compared with that of peripheral blood monocytes (Fig. 5E). Chimerism in lung metastasis foci were further depressed in mice receiving CCR2^−/− mouse-derived cells but not those receiving WT and CCR1^−/− mouse-derived cells as donor cells (Fig. 5E, 5F). Thus, the CCL2–CCR2 interactions were mainly responsible for the metastasis-triggered migration of AMs from the bloodstream to the lungs.

Qian et al. (31) previously reported that the CCL2–CCR2 axis induces the accumulation of IMs in lung metastasis. Thus, we presumed that the blockade of CCR2 signal efficiently inhibits the migration of both AMs and IMs, and thereby prevents lung metastasis. When CCR2^−/− mice were intravenously injected with BNL cells, the numbers and sizes of metastatic tumor foci were dramatically reduced in CCR2^−/− lungs (Fig. 6A, 6B), with reduced numbers of both AMs and IMs, compared with WT mouse lungs (Fig. 6C).

To determine the contribution of CCL2 to the tumor metastasis, BM chimeric mice generated between WT and CCL2^−/− mice were intravenously injected with BNL cells. CCL2 deficiency in BM but not non-BM cells reduced the tumor focus sizes together with depressed AM numbers (Fig. 7A, 7B), similar to observations on CCR2^−/− mice (Fig. 6A, 6B), suggesting the contribution of BM cell–derived CCL2 to lung metastasis and AM migration into tumors. Indeed, CCL2 protein expression was detected within tumor foci in the lungs (Fig. 7C), particularly in CD11b^high IMs but not in CD11c⁺ AMs inside tumor foci (Fig. 7D), indicating that IMs were a main producer of CCL2 in metastatic foci.

In the current study, our comprehensive analysis revealed an increase of almost all AA-derived mediators in metastatic lungs and showed that a 5-LOX inhibitor, zileuton, but not a COX inhibitor, aspirin, reduced the numbers of metastatic tumor foci, particularly of those of a larger size. Moreover, LTB₄ but not an LTC/D/E₄ mixture directly augmented the in vitro proliferation of mouse and human HCC cells. Furthermore, LTB₄ was mainly produced by CCR2-expressing AMs, which accumulated into the lungs from the bloodstream under the guidance of CCL2 provided by IMs (Supplemental Fig. 4). Thus, the CCL2–CCR2 axis-mediated interplay of AMs with IMs can crucially contribute to HCC lung metastasis.

Several lines of evidence suggest that PGs have a crucial role in tumor progression (32, 33). Moreover, both genetic and pharmacological approaches have revealed that COX2, an enzyme crucial for PG synthesis, is one of the key metastasis progression factors for seeding lung metastasis (34). Furthermore, it has been shown that IMs expressed COX2 to a larger extent than AMs when mice were orthotopically transplanted with LLC cells (35). Likewise, in our present lung metastasis model, the Cox2 mRNA expression level was higher in IMs than in AMs. However, we observed that a COX2 inhibitor, aspirin, failed to reduce lung metastasis. Thus, the HCC lung metastasis development may not depend on metabolites of the COX2 pathway. Alternatively, a COX2 pathway blockade may lead to the enhancement in another AA metabolism enzyme, 5-LOX, and subsequent LTB₄ generation (36), thereby sustaining lung metastasis formation.

LTB₄, a metabolite of 5-LOX, is a potent lipid inflammatory mediator, and its major activities include the recruitment and activation of leukocytes (37). In addition to its effects on leukocytes, LTB₄ can have profound impacts on other types of cells, including endothelial cells and fibroblasts (37). Moreover, LTB₄ can directly enhance the cancer cell proliferation (15) and invasiveness (38), confer drug resistance (39), and induce epithelial-mesenchymal transition (40, 41), thereby promoting the tumor progression in various cancers. Consistently, an LTB₄ receptor antagonist, LY293111, can induce the S-phase cell cycle arrest and apoptosis, thereby inhibiting cell proliferation (42, 43). We also observed that LY293111 inhibited the LTB₄-mediated augmentation of in vitro proliferation of mouse and human HCC cell lines and that another LTB₄ receptor antagonist, ONO-4057, also decreased the numbers of metastatic foci with larger sizes but not the total numbers, respectively. Thus, LTB₄ can promote metastatic foci growth rather than tumor cell seeding to lungs.

There exist two distinct receptors for LTB₄, BLT1 and BLT2, which belong to the trimeric G-protein–coupled receptor family (37). BLT1 is a specific and high-affinity receptor for LTB₄, whereas BLT2 exhibits a low affinity for LTB₄ and can bind other eicosanoids. Accumulating evidence indicates that most cancer cells express both BLT1 and BLT2 (44). However, the roles of each LTB₄ receptor in cancer development and progression remain elusive. Ihara and colleagues demonstrated the involvement of the BLT1 pathway in colon cancer cell proliferation (45). Furthermore, the LTB₄–BLT1 axis could block TGF-β–induced cell growth inhibition (46). In contrast, some studies have indicated the crucial roles of BLT2 in LTB₄-mediated cancer cell proliferation (15), invasiveness (38), and the epithelial-mesenchymal transition (40, 41). Mouse and human HCC cell lines expressed both BLT1 and BLT2 mRNA although their levels varied among the cell lines, and a relatively high concentration of LTB₄ was required to enhance the in vitro HCC cell proliferation. Thus, it is more likely that BLT2, rather than BLT1, could mediate the LTB₄ activities in this model.

In breast cancer metastasis to lungs, a large number of neutrophils infiltrate into lungs and support the growth of metastasis-initiating breast cancer cells by providing LTB₄ (16). However, the present lung metastasis model exhibited only a marginal increase in intrapulmonary and intratumoral granulocyte numbers. On the contrary, macrophages were the predominant cell type present in metastatic foci, as compared with granulocytes or T lymphocytes. Moreover, consistent with a previous report (34), the 5-LOX mRNA and protein were highly expressed in AMs but to a lesser extent in IMs among lung macrophages. Furthermore, the 5-LOX protein was mainly detected in CD68⁺ macrophages in human lung tissues obtained from patients suffering from HCC lung metastasis. Thus, we cannot completely exclude the contribution of neutrophils to lung metastasis formation, but in the case of HCC, it is more likely that AMs were major source of LTB₄, which can promote HCC growth and eventually its metastasis to lung.

Macrophages in lungs can be classified into two subpopulations, CD11b^highCD11c⁻ IMs and CD11b^lowCD11c⁺ AMs, which reside in the interstitial and alveolar spaces, respectively (47). The origin of IMs under a steady state remains unclear, but blood inflammatory monocytes are recruited into lung tissues and acquire a CD11b^highCD11c⁻ IM-like phenotype under inflammatory conditions (48). Metastasis-associated macrophages are present in the interstitial space of metastatic lungs and are capable of promoting metastasis (5, 6, 31, 49). Lung metastasis can be promoted by immunosuppressive activities of AMs (11), but their roles in lung metastasis are still unknown. We observed that in addition to IMs, AMs accumulated in the metastatic lung parenchyma, including metastatic foci. Moreover, AMs expressed 5-LOX and secreted LTB₄ to a greater extent than IMs, and CLL-induced selective AM depletion decreased the metastasis formation, particularly when CLLs were administered at a later phase of lung metastasis. Thus, AMs can directly enhance the tumor cell growth in metastatic foci but not tumor cell seeding by secreting LTB₄. Lung metastasis enhanced the 5-Lox mRNA expression by AMs in mice, similar to the observation made on peripheral blood leukocytes under proinflammatory conditions (50). Moreover, 5-LOX and its metabolite, LTB₄, can activate a transcription factor, NF-κB, in human HCC cells (51). NF-κB activation plays a critical role in the expression of various proinflammatory cytokines and chemokines (52), i.e., the factors that can shape an inflammatory tumor microenvironment and promote subsequent tumor progression and metastasis (53).

AMs primarily originate from fetal liver monocytes and self-renew throughout the life cycle, with minimal replenishment from circulating monocytes in a steady state (8–10). However, under conditions when embryo-derived AMs are substantially lost, monocytes in the bloodstream can replenish them (30). We observed that AMs increased in metastatic lungs in a cell proliferation–independent manner. Moreover, accumulated AMs exhibited a gene signature similar to that of blood-borne AMs (30), suggesting that these AMs originated from the bloodstream. Several lines of evidence have implied the contribution of the CCL2–CCR2 axis to blood monocyte migration into the lung and subsequent differentiation into AMs (8, 54). In this lung metastasis model, CCL2 was produced by IMs, and AMs expressed CCR2. Moreover, CCR2 deficiency reduced the reconstitution of AMs in mice with lung metastasis. Furthermore, lung metastasis formation was attenuated in CCR2^−/− mice, together with reduction in both AM and IM accumulation. Thus, we propose that the CCL2–CCR2 axis–mediated interaction between AMs and IMs can crucially contribute to the formation of the tumor microenvironment to support lung metastasis progression (Supplemental Fig. 4).

We are grateful to Dr. Philip M. Murphy (National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD) for critical review of the manuscript.

The authors have no financial conflicts of interest.