Circulating Tumor Microparticles Promote Lung Metastasis by Reprogramming Inflammatory and Mechanical Niches via a Macrophage-Dependent Pathway

Despite metastasis being the leading cause of cancer-related mortality, the underlying mechanism of how primary tumor cells survive and colonize distant organs remains an enigma. Although genetic differences between primary tumors and metastases have been documented (1, 2), studies have also shown that the majority of genomic alterations present in the metastases were similarly present in the corresponding primary tumors (2–5), suggesting that nongenomic elements contribute to the metastatic process. Emerging evidence indicates that cancer metastasis can be facilitated by tumor-induced systemic environmental changes (6, 7). In line with the “seeds and soil” theory (8), it has been proposed that the primary tumor can influence the microenvironment of distant organs prior to metastasis, a process termed premetastatic niche formation (8, 9). Bone marrow–derived cells have been shown to be involved in this process (10). However, the detailed molecular and cellular mechanisms of premetastatic niche formation remain unclear.

In addition to releasing soluble factors into circulation, tumors can also communicate with distant organs, through the release of various microvesicles that mediate intercellular communications (11, 12). Exosomes are small, endosome-derived extracellular microvesicles (30–100 nm), delivering contents such as proteins, messenger RNAs and microRNAs to recipient cells (13). Tumor cell–derived exosomes contribute to premetastatic niche formation by means of their message delivery (14, 15). Tumor cells can also rearrange their cytoskeleton, leading to encapsulation of cytosolic contents within the cellular membrane to form large-sized vesicles, which are subsequently released into extracellular spaces (16). Such specialized subcellular vesicles of 0.1 to 1 μm sizes are termed microparticles (MPs; ref. 17). Studies from colorectal and pancreatic cancer patients have demonstrated that tumor cells release their MPs into the circulation (18, 19). Without disregarding current research on tumor exosomes in premetastatic niche formation, we propose that tumor cell–derived MPs (T-MPs) also contribute to this process. Capillary vessels are physiologically permeable to particles up to 5 to 12 nm in diameter (20), undoubtedly preventing circulating T-MPs (0.1–1 μm in sizes) from entering parenchymal tissues. However, one notable exception is the lungs, where the preexisting apertures in the basal lamina of alveolar capillaries and epithelium range from 0.3 to 3 μm in width (21), allowing MPs to freely cross membrane barriers. Therefore, circulating T-MPs may enter and deliver messages from primary tumors to the lung parenchyma, implying a possibility that primary tumors use T-MPs as a means to remodel the local lung microenvironment and pave the way for future lung metastasis.

Macrophages are the principal immune cells within the tumor microenvironment and are obligate partners for tumor cell migration and metastasis (22). Secondary lung tumors are one of the most frequent human metastatic lesions. Normal lung tissue contains an abundance of alveolar and interstitial macrophages (23). Whether and how lung macrophages are involved in the process of lung metastasis remains unclear. Our previous studies have shown that T-MPs polarized macrophages toward an M2 phenotype, leading to tumor growth and metastasis (24). In the present study, we provide evidence that lung macrophages effectively take up circulating tumor MPs and respond by creating an inflammatory and mechanical niche that promotes the survival of and colonization by immigrant tumorigenic cells, thus setting the stage for the formation of subsequent lung metastasis.

Female BALB/c and C57BL/6 mice (6–8 weeks old) were purchased from the Centre of Medical Experimental Animals of Hubei Province (Wuhan, China). CCR2^−/− mice were a gift from Dr. Lu Gao (Huazhong University of Science and Technology). All animal experiments were approved by the Animal Care and Use Committee of Tongji Medical College. B16-F10 melanoma, Lewis lung carcinoma (LLC), and 4T1 breast cancer cell lines were purchased from the China Centre for Type Culture Collection (CCTCC) and cultured according to the manufacturer's guidelines. (B16-F10 and LLC were purchased in 2014, and 4T1 were purchased in 2012). B16-F10 and LLC were maintained in DMEM supplemented with 10% FBS, and 4T1 cells were maintained in RPMI-1640 supplemented with 10% FBS. Cells were thawed from liquid nitrogen stocks and passaged 3 times before being used for in vivo or in vitro experiments. All cell lines were tested and determined to be free of mycoplasma and other rodent pathogens by the CCTCC.

Tumor cells or splenocytes were cultured in serum-free medium and hypoxic conditions for 24 hours after which the medium was collected for microparticle isolation as described previously (25). Briefly, supernatants were centrifuged at 1,000 × g for 10 minutes to remove whole cells and then centrifuged for 2 minutes at 14,000 × g to remove debris. The supernatant was further centrifuged for 60 minutes at 14,000 × g to pellet microparticles. The pellets were washed three times and resuspended in PBS for the subsequent experiments. The protein concentration of T-MPs was measured using a BCA kit, according to manufacturer's protocol (Pierce, Thermo Fisher Scientific). T-MPs size and particle number were analyzed using a NanoSight NS300 system.

To assess the T-MPs distribution in the lung, liver, and spleen, purified T-MPs were fluorescently labeled using PKH26 membrane dye (Sigma-Aldrich) according to the manufacturer's protocol. Labeled T-MPs (50 μg) were intravenously injected in mice. The lungs, liver, and spleen were collected and embedded in Tissue-Tek OCT (Sakura) for cryosections 2 hours later. Tissue cryosections (10 μm) were stained with DAPI and analyzed by two-photon fluorescent microscopy (Confocal Laser Scanning Microscope Leica TCS SP8; Leica).

Fifty micrograms of B16-MPs was injected by tail vein. Control mice were injected with 50 μg splenocyte-derived MPs (S-MPs) or PBS. Twenty-four hours after B16-MPs treatment, mice were injected with 2 mg of FITC-labeled dextran 70,000 MW (Sigma-Aldrich) by tail vein injection. Two hours after dextran injection, mice were euthanized and perfused with PBS. Lungs were fixed and inflated with a 50% vol/vol mixture of Tissue-Tek OCT for cryosections.

For the generation of spontaneous lung metastatic models, 6- to 8-week-old C57BL/6J female mice were injected in the flank with 5 × 10⁵ B16-F10 cells. After 5 days, 50 μg of B16-MPs was injected intravenously every other day for a total of 10 treatments within 3 weeks. Lungs were collected at days 21 and 35. For paraffin sections, before removal, lungs were perfused and fixed with 10% buffered formalin. Lungs were excised and fixed in formalin overnight. Paraffin-embedded lungs were systematically sectioned and stained with hematoxylin and eosin (H&E) staining, and images were taken. Metastasis index (the percentage of metastasis tumor areas to total lung areas based on calculation from 10 slides) and the number of metastatic foci (averaged number of metastasis sites per square millimeter of lung area) were used to describe the metastatic lesions. A subset of mice was treated with 60 μL clodronate liposome or CSF1R inhibitor GW2580 (Med Chem Express) intranasally to deplete lung macrophages. Clodronate liposomes were purchased from www.clodronateliposomes.org (Vrije University, Amsterdam, the Netherlands).

Tumor volume was measured by calipers and calculated according to the formula: length × width²/2. To assess the number of circulating tumor cells (CTC), whole blood was collected and centrifuged at 2000 × g × 5 minutes; buffy coat was isolated and cultured in DMEM (10% heat inactivated FBS, 1% penicillin/streptomycin and 2 mmol/L l-glutamine). Tumor colonies were counted on day 10 of culture and the number of CTCs was calculated as the colonies present in the dish. To analyze the CTCs by flow cytometry, buffy coat was collected and stained with APC-conjugated anti-CD45 antibody (clone OX1, eBioscience). CD45-negative cells were measured.

For the 4T1 breast cancer model, 5 × 10⁵ 4T1 breast cancer cells were injected into the second mammary pad of female BALB/c mice. Five days following this, 50 μg of 4T1-MPs was injected intravenously for a total of 10 treatments within 3 weeks.

To analyze the role of T-MPs education in tumor metastasis, 6- to 8-week-old C57Bl/6 female mice preeducated with 50 μg B16-MPs every other day for 20 days, followed by intravenous injection of 1 × 10⁵ B16-F10 cells. Mice were sacrificed 21 days later, and the black melanoma nodules on the lungs were assayed. A subset of mice was treated with clodronate liposome, CSF1R inhibitor GW2580, anti-IL6 (clone MP5-20F3, BioLegend), or plasmin (Sigma-Aldrich), where indicated.

Bone marrow cells were harvested by flushing the femurs and tibias of CCR2^−/− mice or wild-type (WT) littermates. Recipient C57BL/6 mice were irradiated with a single dose of 9 Gy, and all mice received 5 × 10⁶ bone marrow cells from either CCR2^−/− mice or WT littermates within 4 hours. After 4 weeks, the reconstituted C57BL/6 mice were used in subsequent experiments.

Lungs tissues were collected, cut into small pieces (∼1–2 mm²), and digested at 37°C for 45 to 60 minutes with an enzyme cocktail (collagenase A and DNase I, Sigma-Aldrich). Single-cell suspensions were filtered through a 40-μm cell strainer. Cell suspensions were washed in PBS and 1% FBS and incubated with the following fluorochrome-conjugated antibodies: CD11b (M1/70), CD11c (clone N418), F4/80 (clone BM8), Ly-6C (clone HK1.4), CD3 (clone 17A2), CD19 (clone 6D5), NK1.1 (clone PK136), CD206 (clone MR6F3), CD45 (clone OX1). For intracellular cytokine staining, lung single cells were treated with Fix/Perm (eBioscience) solution following staining with a surface marker, and restained with CCL2 (clone 2H5), IL6 (clone MP5-20F3), IL10 (clone JES5-16E3), TNFα (clone MP6-XT22), or Ki-67 (clone 16A8) antibody.

For isolation of macrophages from lung tissues, single-cell suspensions from lungs were stained with CD45 (clone OX1) and F4/80 (clone BM8) antibodies and then were sorted using a BD FACS Aria III Cell Sorter (BD Biosciences) with purities of >95%.

F4/80⁺CD11b⁺Ly6C⁻ macrophages were isolated from lung tissues of mice treated or not with B16-MPs (10 injections) using BD FACS Aria III Cell Sorter. After additional 24 hours culture in vitro, supernatants of lung macrophages were collected for following experiments.

For the analysis of antibodies induced T-cell proliferation, splenic CD8⁺ T cells were isolated by Magnetic-activated cell sorting using a CD8⁺ T-cell isolation kit (Miltenyi). T cells were labeled with CFSE (Sigma) and activated by anti-CD3/28 beads (Thermo Fisher Scientific) in the presence of supernatants from macrophages. T-cell proliferations were examined by CFSE dilution assay 72 hours later using flow cytometry.

For the analysis of antigen-induced T-cell proliferation in vitro, CD8⁺ T cells purified from the spleen of OT-I mice were labeled with CFSE and then cultured with dendritic cells (DCs) which pulsed with OVA_257-264 at a T-cell/DC ratio of 10:1 in the presence of supernatants from macrophages. T-cell proliferations were examined by CFSE dilution assay 72 hours later using flow cytometry.

For the analysis of antigen-induced T-cell proliferation in vivo, 1 × 10⁶ CD8⁺ T cells purified from the spleen of OT-I mice were labeled with CFSE and adoptive transferred into B16-MP or PBS treated mice. Next day, mice were infected with 5 × 10⁵ CFU of listeria monocytogenes expressing ovalbumin (OVA; LmOVA). T-cell proliferations in lung were examined 72 hours later.

For immunofluorescence, before removal, lungs were perfused with PBS and inflated with a 50% vol/vol mixture of Tissue-Tek OCT. The perfused lungs were then removed and embedded in Tissue-Tek OCT, following by freezing on dry ice. OCT tissue cryosections (10 μm) were stained with antibodies against F4/80 (1:200 dilution, Abcam) and fibrin (1:200 dilution, Abcam). Secondary antibodies conjugated to Alexa Fluor 488 or 594 were used. All experiments were performed according to the manufacturer's protocol and visualized by two-photon fluorescent microscopy fibrin expression was quantified using ImageJ software by determining the ratio between the areas of fibrin and DAPI staining and described as arbitrary units (a.u.).

Three-dimensional (3D) fibrin gel cell culture of tumor cells was conducted according to our previously described method (26). In brief, tumor cells were detached from a conventional rigid plate by 0.25% trypsin digestion and suspended in DMEM (10% FBS). Cell density was adjusted to 1 × 10⁴ cells/mL. Fibrinogen (Searun Holdings Company) was diluted into 2 mg/mL with T7 buffer (pH 7.4, 50 mmol/L Tris, 150 mmol/L NaCl). Fibrinogen and cell solution mixture (1:1) was made, resulting in 1 mg/mL fibrinogen and 5,000 cells/mL in the mixture. Cell/fibrinogen mixtures (250 mL) were seeded into each well of 24-well plate and mixed well with preadded 5 mL thrombin (0.1 U/μL, Searun Holdings Company) for culture under 37°C condition.

Total RNA was extracted from cells with TRIzol reagent (Invitrogen). Real-time PCR was performed with 2 μg of cDNA as a template, using Fast SYBR Green PCR master mix (TOYOBO) on a CFX96 Touch real-time PCR detection system (Bio-Rad). mRNA levels were normalized to β-actin. The primer sequences are shown as follows: S100A8, 5′-AAATCACCATGCCCTCTACAAG-3′ (sense) and 5′-CCCACTTTTATCACCATCGCAA-3′ (antisense); S100A9, 5′-ATACTCTAGGAAGGAAGGACACC-3′ (sense) and 5′-TCCATGATGTCATTTATGAGGGC-3′ (antisense); SAA3, TGCCATCATTCTTTGCATCTTGA-3′ (sense) and 5′-CCGTGAACTTCTGAACAGCCT-3′ (antisense); CCL2, TTAAAAACCTGGATCGGAACCAA-3′ (sense) and 5′-GCATTAGCTTCAGATTTACGGGT-3′ (antisense); IL6, CCAAGAGGTGAGTGCTTCCC-3′ (sense) and 5′-CTGTTGTTCAGACTCTCTCCCT-3′ (antisense); β-actin, GGCTGTATTCCCCTCCATCG-3′ (sense) and 5′-CCAGTTGGTAACAATGCCATGT-3′ (antisense).

CCL2, IL6, and VEGF levels were assessed using the murine ELISA kit (Dakewei) for mouse cells and lung homogenate according to the manufacturer's protocol.

B16-MPs or macrophages protein were extracted and analyzed by Western blot using anti-mouse HIF1-α (1:1,000 dilution, Cell Signaling Technology), anti-mouse VEGF-A (1:500 dilution, Abcam), anti-phospho-TBK (Ser172; 1:1,000 dilution, Cell Signaling Technology), anti-TBK (1:1,000 dilution, Cell Signaling Technology), anti-phospho-STAT6 (Tyr641; 1:500 dilution, Abcam), anti-STAT6 (1:1,000 dilution, Abcam), or anti-mouse β-actin (1:1,000 dilution, Cell Signaling Technology), followed by secondary horseradish peroxidase–coupled antibodies incubation and visualized by enhanced chemiluminescence according to the manufacturer's protocol (ECL kit; Pierce).

All experiments were performed at least three times. Results are expressed as mean ± SEM and analyzed by unpaired two-tailed Student t test. The Kaplan–Meier method was used to estimate overall survival, and the differences in survival were analyzed using the log-rank test. Differences were considered to be statistically significant when the P value was <0.05. The analysis was conducted using the GraphPad Prism version 6 (GraphPad Software).

Although lung metastasis is a common phenomenon for various types of solid tumors, whether primary solid tumor cells use a common pathway for their lung metastasis remains elusive. Tumor cells are capable of releasing MPs into the extracellular space after exposure to environmental cues. Hypoxia is ubiquitously present in solid tumor microenvironments. We found that hypoxia (1% O₂) caused more than a 10-fold release of T-MPs in tumor cell lines (B16-F10 melanoma, 4T1 breast or LLC lung tumor cells), compared with normoxia, as evidenced by the NanoSight system (Fig. 1A). The quantification of MP-contained proteins also showed a 10-fold increase in proteins (Fig. 1B). However, these hypoxia-induced T-MPs did not display a difference in the mean size or size distribution between hypoxia-induced and normoxia-formed T-MPs (Fig. 1C and D; Supplementary Fig. S1A—S1C). We analyzed hypoxia-inducible proteins, such as HIF-1α and VEGF-A by Western blot. We found that the level of HIF-1α, but not VEGF-A, exhibited a 5-fold increase in the hypoxia-MPs, compared with normoxia-MPs (Supplementary Fig. S1D). In addition to hypoxia, various exogenous and endogenous signals are also able to induce tumor cells releasing MPs, which are capable of entering the circulation (27–29). Given the large permeability of alveolar capillaries (0.3–3 μm), we hypothesized that circulating T-MPs might enter the lung parenchyma leading to metastasis of the primary tumor to the lungs. To test this hypothesis, C57BL/6 mice were subcutaneously inoculated with 1 × 10⁵ B16-F10 melanoma cells. Five days later, we injected hypoxia-induced, B16-F10-derived MPs (B16-MPs) to the mice every other day for a total of 10 injections. Normal splenocyte-derived MPs (S-MPs) were used as control MPs. We found that compared with the control, mice treated with B16-MPs showed a greater lung metastasis burden (metastasis index and number of metastatic foci) as measured by H&E staining on day 21 (Fig. 1E). On day 35, macroscopic melanoma nodules were visible in the B16-MPs group (Fig. 1E). Consistently, the intravenous injection of 4T1 breast cancer cell–derived MPs (4T1-MPs) also promoted breast cancer lung metastasis in mice (Fig. 1F). To verify whether the repeated injection of T-MPs is necessary for tumor cell metastasis, we conducted 3, 7, or 10 injections of B16-MPs to B16-F10 tumor-bearing mice. We found that 3 injections of B16-MPs did not result in the increase of B16-F10 metastasis. However, 7 and 10 injections both increased B16-F10 metastasis (Supplementary Fig. S1E). Collectively, these data suggest that circulating T-MPs are capable of promoting primary tumor cell lung metastasis.

Next, we investigated the underlying mechanism through which circulating T-MPs promote lung metastasis. Exosomes are smaller extracellular microvesicles (30–100 nm), which have been known to stimulate tumor growth (14). However, we found that the above described intravenous injection of T-MPs did not promote the growth of the primary melanoma or breast tumors (Fig. 2A), nor did it increase the number of CTCs (Fig. 2B; Supplementary Fig. S2A), suggesting that the main effect of T-MPs is not directly on the primary tumor but rather on the lungs where they promote lung metastasis. In line with this notion, pretreatment of animals with B16-MPs increased lung metastases following intravenous injection of circulating B16-F10 tumor cells (Fig. 2C), leading to significantly shortened survival of the mice (Fig. 2D). A similar result was also obtained from preinjection of 4T1-MPs (Supplementary Fig. S2B and C), suggesting that different T-MPs have a similar ability to influence the lung microenvironment. As further reinforcement, fluorescence-labeled, primary tumor cell–released MPs were found to access to the lungs via the circulation (Fig. 2E; Supplementary Fig. S3A). Those T-MPs in the lungs could be taken up by both F4/80⁺CD11c⁺ alveolar macrophages and F4/80⁺CD11c⁻ interstitial macrophages (30, 31), as evidenced by flow cytometry and immunostaining (Fig. 2F; Supplementary Fig. S3B). To dissect whether those macrophages are involved in T-MP–promoted lung metastasis, we used clodronate liposomes, a widely used macrophage-depleting agent to deplete lung macrophages (32). We found that macrophage depletion significantly blunted the metastasis-promoting effect of B16-MPs (Fig. 2G). In addition to the clodronate liposome, we also used a CSF1R inhibitor, GW2580, to disrupt the macrophages' effect in B16-MP–treated tumor-bearing mice. Consistently, the metastasis-promoting effect of B16-MPs was impaired (Supplementary Fig. S3C). These data suggest that lung macrophages mediate T-MP–promoted primary tumor cell lung metastasis.

Proinflammatory microenvironments are known as a “promoting force” that drives metastasis (33). In parallel with being taken up by macrophages, injection of B16-MPs resulted in the enhanced endothelial permeability in the lungs, as evaluated by the extravasation of FITC dextran (Fig. 2H), a typical feature of inflammation and inflammatory microenvironment (34). Consistently, the expression of S100A8, S100A9, and serum amyloid A3 (SAA3), three inflammatory mediators that have been known to remodel the lung microenvironment for primary tumor cell metastasis (35, 36), was upregulated in the lungs after the injection of B16-MPs (Fig. 2I). However, if we depleted macrophages in advance, neither vascular permeability nor the expression of S100A8, S100A9, and SAA3 was influenced by the injection of T-MPs (Supplementary Fig. S4A and S4B). Together, these results suggest that, upon uptake, T-MPs confer to macrophages the ability to build an inflammatory microenvironment in the lungs, leading to primary tumor cell metastasis.

Next, we investigated how the inflammatory microenvironment was transformed by lung macrophages after taking up T-MPs. Recruitment of myeloid cells is an essential step during the process of inflammation. To examine this, we analyzed immune cells in the lung 48 hours following an intravenous injection of T-MPs. We found that although proportions of T cells, B cells, and NK cells were not changed in the lungs (Supplementary Fig. S5A and S5B), the proportion of CD11b⁺Ly6C^high inflammatory monocytes (37) was increased ∼4-fold in the lungs (Fig. 3A). These cells can express myeloid chemokine receptor CCR2 that is attracted by CCL2 (34, 37). Injection of B16-MPs resulted in the upregulation of CCL2 in the lungs (Fig. 3B). Further investigation revealed that CCL2 was mainly produced by lung macrophages that had taken up T-MPs, evidenced by (i) CCL2 mRNA which was upregulated in macrophages following injection of T-MPs (Supplementary Fig. S6A); and (ii) macrophage depletion (chlodronate) or inhibition (GW2580) which blocked the increase of CCL2 levels and the recruitment of the inflammatory monocytes (Supplementary Fig. S6B and S6C). To directly confirm the CCL2 induction in lung resident macrophages by B16-MPs, we isolated the lung macrophages (CD11b⁺F4/80⁺ Ly6C⁻) for CCL2 detection. We found that CCL2 expression was increased in lung macrophages treated with B16-MPs (Supplementary Fig. S6D and S6E). Therefore, circulating T-MPs can enter the lung parenchyma where they are taken up by lung macrophages to induce the production of CCL2 in order to attract inflammatory monocytes. To provide additional support for this hypothesis, we injected B16-MPs into CCR2^−/− mice, and found the recruitment of CD11b⁺Ly6C^high inflammatory monocytes to the lungs by B16-MPs was abrogated (Fig. 3C). In parallel, B16-MPs neither promoted primary tumor cells lung metastasis in CCR2^−/− mice (Fig. 3D), nor shortened the overall survival of the mice (Fig. 3E). Then, we investigated the pathway through which T-MPs activate macrophages to produce CCL2. Previous studies showed that MPs from apoptotic tumor cells contain genomic and mitochondrial DNA fragments that can activate the cGAS-STING-TBK1 pathway, leading to type I IFN production in dendritic cells (27) and STAT6 activation in macrophages (24). Here, we also found that B16-MP–induced CCL2 expression in macrophages is dependent on the activation of TBK1/STAT6 (Supplementary Fig. S7A–S7C). Moreover, the use of STING or cGAS siRNA consistently abrogated the expression of CCL2 in macrophages (Supplementary Fig. S7D and S7E), indicating that DNA fragments within hypoxic tumor cell-MPs contributed to CCL2 upregulation in macrophages via the cGAS/STING/STAT6 pathway. Together, these data suggest that CCL2 production by T-MP–affected lung macrophages is critical for the formation of a prometastatic inflammatory microenvironment.

The above data showed that entry of T-MPs into the lungs resulted in the attraction of CD11b⁺Ly6C^high monocytes via the CCL2/CCR2 signaling pathway. However, this population of CD11b⁺Ly6C^high monocytes was not persistently maintained and even showed a decreasing trend (Supplementary Fig. S8A), regardless of continuous injection of B16-MPs and high levels of CCL2 in the lungs (Supplementary Fig. S8B). In contrast, the proportion of F4/80⁺CD11b⁺Ly6C⁻ macrophages significantly increased after 10 injections with B16-MPs, as evaluated by flow cytometry (Fig. 4A) and further confirmed by immunostaining of F4/80 (Fig. 4B). Macrophages are normally considered nonproliferating cells. Using the proliferation marker Ki-67, we did not find that B16-MPs treatment caused the proliferation of F4/80⁺CD11b⁺Ly6C⁻ cells in the lungs (Supplementary Fig. S8C–S8E), suggesting that the recruited CD11b⁺Ly6C^high monocytes differentiated into F4/80⁺CD11b⁺Ly6C⁻ macrophages. To confirm this, CD45.2⁺ mice were injected with B16-MPs for 6 hours, followed by an adoptive transfer of CD45.1⁺CD11b⁺Ly6C^high monocytes. We found that the transferred CD45.1⁺ monocytes migrated to the lungs and maintained the monocyte phenotype of CD11b⁺Ly6C^high during the first 24 hours, but most of these switched to a F4/80⁺CD11b⁺Ly6C⁻ macrophage phenotype 96 hours later (Supplementary Fig. S8F). These monocyte-transformed macrophages were seemingly capable of taking up 5-fold more B16-MPs than monocytes (Supplementary Fig. S8G). In addition, inhibition of the recruitment of CD11b⁺Ly6C^high monocytes by a CCL2-neutralizing antibody abrogated the number of T-MP–increased macrophages in the lungs (Fig. 4C). Such transformed macrophages appeared to have an inflammatory phenotype because high levels of proinflammatory cytokines, TNFα and IL6, were found in those cells, as evidenced by the intracellular flow-cytometric analysis (Fig. 4D). We also found that M2-related markers including CD206 and IL10 were upregulated in F4/80⁺CD11b⁺Ly6C⁻ macrophages in the lungs after 10 injections with B16-MPs (Fig. 4E). Such macrophages also highly expressed arginase 1, a typical marker of the M2 phenotype (Supplementary Fig. S8H). In line with the immunosuppressive trait of M2 macrophages, we found that these F4/80⁺CD11b⁺Ly6C⁻ macrophages effectively inhibited CD3/CD28 antibodies-stimulated T-cell proliferation as well as OVA_257-264 peptide-stimulated OT-1 T-cell proliferation in vitro (Fig. 4F; Supplementary Fig. S9A). Pretreatment with B16-MPs inhibited the proliferation of OT-1 T cells in vivo, which, however, could be rescued by the administration of a CSF-1R inhibitor GW2580 (Supplementary Fig. S9B). Therefore, CCL2-attracted monocytes are transformed to a new class of macrophages with both inflammatory and immunosuppressive traits. To clarify the role of these newly differentiated macrophages in lung metastasis, mice, pretreated with 10 injections of B16-MPs, were subjected to a combination of intravenous and intranasal administration of clodronate liposomes to deplete macrophages, followed by an intravenous injection of B16-F10 tumor cells. As expected, depletion of F4/80⁺CD11b⁺Ly6C⁻ macrophages by clodronate liposomes or inhibition of macrophages by GW2580 blocked the prometastatic effect of B16-MPs (Fig. 4G; Supplementary Fig. S9C). Together, these data suggest that T-MP–triggered production of CCL2 attracts monocytes which can mature into F4/80⁺CD11b⁺Ly6C⁻ metastasis-promoting macrophages with both inflammatory and immunosuppressive phenotypes.

Given the important role of inflammation in metastasis and that the above F4/80⁺CD11b⁺Ly6C⁻ macrophages highly express the key inflammatory cytokine IL6, we further investigated IL6. We found that the concentration of IL6 in the lungs gradually increased in line with the multiple intravenous injections of B16-MPs (Fig. 5A). However, this increase of IL6 was blocked by either macrophage depletion or CCL2 neutralization in B16-MP–treated mice (Fig. 5B). To clarify the role of IL6 in the above lung metastasis, we then used an IL6 antibody to neutralize IL6. As a result, B16-MP–promoted lung metastasis was blocked (Fig. 5C) and the mice had a prolonged survival (Fig. 5D), suggesting that T-MP–promoted lung metastasis is mediated through IL6 signaling. Next, we investigated how IL6 promoted lung metastasis.

Stem cell–like tumorigenic cells are thought to be the key cell population that repopulates a metastatic tumor in distant organs. We have previously established a highly effective soft 3D fibrin gel culture system to generate tumorigenic cells, termed tumor-repopulating cells. These soft fibrin gels correspond to 90 Pa in elastic stiffness and the cells were individually trapped inside allowing for colony formation (26). Such tumor-repopulating cells not only showed spheroid-like morphologic changes resembling stem-like cells, but also as few as 10 gel-selected B16 tumor-repopulating cells were able to repopulate a lung metastatic tumor in immunocompetent mice. Here, we found that B16-F10 tumor-repopulating cells had an increased ability to survive and colonize in the lungs, compared with differentiated B16-F10 cells (Fig. 6A). The survival and colonization of B16-F10 tumor-repopulating cells in the lungs could be further enhanced by the injection of B16-MPs (Fig. 6B), leading to a greater lung metastatic burden (Fig. 6C) and shorter overall survival (Fig. 6D), suggesting that stem-like tumor-repopulating cells are a major contributor to lung metastasis. IL6 has been associated with induction of tumor cell stemness (38), which prompted us to hypothesize that IL6 contributes to lung metastasis by regulating tumor-repopulating cells. To test this, we used B16-MP–conditioned macrophage supernatants to treat B16-F10 tumor-repopulating cells, and we found an increase of the size (∼86%) and number (∼57%) of B16-F10 tumor-repopulating cells colonies in soft 3D fibrin gels, which, however, was abrogated by anti-IL6 or anti-IL6R treatment (Fig. 6E and F). Although the injection of T-MPs promoted tumor-repopulating cells survival and colonization in the lungs of mice, IL6 neutralization impaired this process (Fig. 6G). Together, these results suggested that macrophages promote the growth of tumor-repopulating cells via IL6 signaling in order to enhance lung metastasis.

The above data suggest that T-MP–triggered lung inflammation results in the production of IL6, leading to tumor-repopulating cell colonization and growth. However, in addition to chemical signaling, stem cells also require an extracellular matrix-generated mechanical niche for their growth and survival (39). Collagen is a main stromal component in the lungs; however, tumor-repopulating cells only grow well in fibrin gels but not in collagen matrices (26), implying that lung-infiltrating T-MPs may build a fibrin(ogen) physical microenviroment for tumor-repopulating cells growth and survival. In line with this possibility, we found that the deposition of fibrin in the lungs was increased after an injection of B16-MPs (Fig. 7A). Such fibrin deposition could be the result of the enhanced lung endothelial permeability. When we injected fluorescence-labeled fibrinogen to mice following an injection of B16-MPs, after 24 hours, abundant fibrinogen was observed in the lungs (Fig. 7B). However, B16-MP–caused fibrin deposition could be degraded by an intranasal administration of plasmin (Supplementary Fig. S10A). Under this condition, we found that the survival of tumor-repopulating cells in the lungs of B16-MP–pretreated mice was impaired, evidenced by the decreased tumor-repopulating cell numbers (Fig. 7C). Consistently, plasmin-mediated fibrin degradation resulted in blunting the prometastatic effect of B16-MPs in the lungs (Fig. 7D). We isolated tumor cells from the lungs of the above treated mice and seeded them into soft 3D fibrin gels. The result showed that there were fewer colonies in the plasmin group as compared with the control (Fig. 7E), suggesting that fibrin deposition is essential for the survival and colonization of tumor-repopulating cells in the lungs. In addition, we found that either depleting (chlodronate) or inhibiting (GW2580) macrophages or neutralizing CCL2 resulted in decreases in fibrin deposition in T-MP–treated mice (Fig. 7F; Supplementary Fig. S10B), suggesting that fibrin deposition in the lungs is caused by T-MP–triggered macrophages. Fibrin deposition occurs during inflammation due to the increased permeability of blood vessels. VEGF plays a critical role in vascular permeability, which might lead to fibrin deposition into the lungs. Here, we found that B16-MPs treatment resulted in the upregulation of VEGF expression in the lungs, which was abrogated by GW2580 administration (Supplementary Fig. S10C). In addition, when we used B16-MPs to treat bone marrow–derived macrophages or macrophages isolated from the lungs, the expression of VEGF was also upregulated (Supplementary Fig. S10D and S10E). Furthermore, VEGF neutralization inhibited the leakage of fibrinogen in the lungs (Supplementary Fig. S10F), suggesting that T-MP–induced VEGF by macrophages contributes to fibrin deposition in the lungs. Unlike VEGF, IL6 neutralization did not affect T-MP–triggered fibrin deposition in the lungs (Fig. 7F). Thus, IL6 signaling and fibrin deposition are two parallel biological events that promote lung metastasis. Together, these data suggest that circulating T-MPs can cause fibrin(ogen) deposition in the lungs which supports tumor-repopulating cell growth and consequently promotes lung metastasis.

Despite the abundant literature on tumor cell metastasis, the manner by which primary tumor-derived signals help the tumor cell to survive and colonize distant organs is a poorly defined process. In the present study, we provide evidence that primary tumor cells form lung metastasis via a series of biological events triggered by circulating T-MPs and illustrated in Fig. 7G: (i) T-MPs enter the lungs via circulation; (ii) lung macrophages take up T-MPs and release CCL2; (iii) CCL2 recruits monocytes to the lungs where they are transformed to macrophages; (iv) macrophages release IL6 which facilitates the growth and survival of highly tumorigenic tumor-repopulating cells in the lungs; (v) in parallel, T-MP–conditioned macrophages induce lung inflammation which increases endothelial permeability, leading to fibrin deposition in the lungs; (vi) fibrin deposition provides mechanical signals for tumor-repopulating cell growth and survival; and (vii) tumor-repopulating cell growth results in the formation of metastatic lesions in the lungs.

The presence of T-MPs in tumor microenvironments is a general pathophysiological phenomenon (40). Such continuously produced T-MPs in the interstitial fluid of extracellular space may enter lymph or papillary vessels. After entering the draining lymph node through lymphatic circulation, T-MPs can be cleared by local macrophages. However, T-MPs may circulate to the lungs if they enter blood vessels due to their abnormally large apertures in the basal lamina of alveolar capillaries and epithelium (21).

This study shows that T-MP–reprogrammed lung macrophages to both M1 and M2 phenotypes. Although immunosuppression conferred by macrophages undoubtedly favors tumor cell immune evasion and survival in the lungs; the proinflammatory phenotype provides inflammatory signals that facilitate the growth and survival of tumor-repopulating cells in the lungs. IL6, a pleiotropic inflammatory cytokine, not only mediates the activation of the MAPK and PI3K/Akt signaling pathways for tumor cell growth, but also classically activates STAT3 which can maintain the stemness of tumor cells (41). Previously, we developed a mechanics-based method to select and amplify a subpopulation of cancer cells that are particularly tumorigenic can grow very round spheroids in 3D soft matrices and are called tumor-repopulating cells (26, 42). Although 3D fibrin matrices are an artificial culture system, we found, using immunohistochemical staining, that fibrin(ogen) was present in tumor tissues. CD45⁻ALDH⁺ tumor cells are reported to be stem cell–like tumor cells (43); however, we have previously shown that only ALDH⁺ rather than CD45⁻ breast cancer cells isolated from MMTV-PyMT mice grow into colonies in soft 3D fibrin gels (44). In addition, we have demonstrated that only CD133⁺ tumorigenic but not CD133⁻ B16 melanoma cells effectively grow into colonies in soft 3D fibrin gels (45). Findings presented in the current study suggest that tumor-repopulating cells are tumorigenic in vivo and fibrin is likely required for tumorigenic cell development. In the present study, we demonstrated that stem-like tumor-repopulating cells are important players for lung metastasis. We further confirmed that IL6 contributes to lung metastasis by promoting tumor-repopulating cell growth and survival. Thus, T-MP–reprogrammed macrophages can facilitate lung metastasis via releasing the inflammatory signal molecule IL6.

Cells apply contractile forces to sense the physical microenvironment and respond accordingly by binding of extracellular matrix (ECM) proteins such as collagen and fibrin to integrins, leading to mechanotransduction along clustered integrins to focal adhesions (46). In this study, we found that inflammatory signals by T-MP–reprogrammed macrophages caused an increased permeability of lung capillaries, leading to fibrin(ogen) deposition. Consequently, deposited fibrin provided mechanical signals for tumor-repopulating cell survival and growth in the lungs. We propose that T-MPs act as a pathway that primes the lungs for metastasis by creating an immunosuppressive, inflammatory, and physical mechanical microenvironment that is permissive to the growth of metastatic tumor cells. Our findings also evoke concerns regarding current EGFR-targeted therapy and high-dose chemotherapies. Although such treatments cause widespread tumor cell death in a short time, apoptotic tumor cells may be triggered to produce large amounts of T-MPs, which may promote lung metastasis or worsen primary tumors by facilitating tumor-repopulating cell growth and survival. This possibility has implications for current therapies.

In summary, the data presented in this study demonstrate that T-MPs can initiate a pathway for the generation of premetastatic niche-associated macrophages in the lungs, which can reprogram the lung immune, inflammatory, and mechanical microenvironments and thus promote tumor-repopulating cell growth and lung metastasis. Our findings provide mechanistic insights regarding premetastatic niche formation and organ site–specific tropism of metastasis. Further research on the metastatic axis of T-MPs may reveal effective targets to prevent and treat lung metastasis.

No potential conflicts of interest were disclosed.

Conception and design: B. Huang

Development of methodology: H. Zhang, Y. Yu, J. Ma, D. Chen, B. Huang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Zhang, Y. Yu, L. Zhou, K. Tang, P. Xu, T. Ji, X. Liang, T. Zhang, J. Xie

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Zhang, J. Ma, T. Zhang

Writing, review, and/or revision of the manuscript: H. Zhang, Y. Liu, B. Huang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Ma, P. Xu, T. Ji, J. Lv, Y. Liu

Study supervision: Y. Liu, B. Huang

Other (assisted in conducting experiments): W. Dong

This work was supported by the National Natural Science Foundation of China (81788101, 81661128007, 81530080, 81701544), the Chinese Academy of Medical Sciences Initiative for Innovative Medicine (2016-I2M-1-007), and China Postdoctoral Science Foundation funded project (2017M610478).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.