Lung mesenchymal cells elicit lipid storage in neutrophils that fuel breast cancer lung metastasis

Abstract

Acquisition of a lipid-laden phenotype by immune cells has been defined in infectious diseases and atherosclerosis but remains largely uncharacterized in cancer. Here, in breast cancer models, we found that neutrophils are induced to accumulate neutral lipids upon interaction with resident mesenchymal cells in the premetastatic lung. Lung mesenchymal cells elicit this process through repressing the adipose triglyceride lipase (ATGL) activity in neutrophils in prostaglandin E2-dependent and -independent manners. In vivo, neutrophil-specific deletion of genes encoding ATGL or ATGL inhibitory factors altered neutrophil lipid profiles and breast tumor lung metastasis in mice. Mechanistically, lipids stored in lung neutrophils are transported to metastatic tumor cells through a macropinocytosis–lysosome pathway, endowing tumor cells with augmented survival and proliferative capacities. Pharmacological inhibition of macropinocytosis significantly reduced metastatic colonization by breast tumor cells in vivo. Collectively, our work reveals that neutrophils serve as an energy reservoir to fuel breast cancer lung metastasis.

Main

Metastatic disease remains the major cause of cancer-related death. Among the vital organs to which solid tumors metastasize, the lung is one of the most common sites. In the past two decades, substantial advances in our understanding of lung metastasis have revealed intricate interactions between disseminated tumor cells (DTCs) and the lung-resident immune microenvironment that are essential for the development of metastatic lung lesions1,2. Within the lung immune microenvironment, bone marrow (BM)-derived neutrophils have been reported as an indispensable component that facilitates solid tumor metastasis1,2. Accumulating evidence suggests that neutrophils regulate lung metastasis as tumor cells colonize the lung, where neutrophils function to suppress antitumor immunity, accelerate DTC extravasation and proliferation and awaken dormant DTCs via neutrophil extracellular traps (NETs)3,4,5,6. Through these effects, neutrophils act in concert with other organ-resident stromal cells, contributing to formation of the premetastatic and metastatic niches1,2. However, the DTC supportive effects of neutrophils and other lung-resident stromal cells were mostly studied at transcriptional and protein levels, and the metabolic crosstalk between the lung microenvironment and DTCs is largely undefined.

In fact, solid tumor metastasis is a highly inefficient process7. DTCs need to undergo metabolic alterations to adapt to the new environment and successfully survive and colonize the metastatic sites8,9. Recent studies showed that DTCs can fully benefit from local resources, including the metabolic energetics of the organ microenvironment10,11. In ovarian cancer metastasis to omentum, ovarian cancer cells are capable of acquiring lipids from omental adipocytes to support their proliferation through accelerated β-oxidation11. In colon cancer liver metastasis, colon cancer cells take up extracellular phosphocreatine in the liver microenvironment to generate adenosine triphosphate for their metastatic survival10. Therefore, the organ microenvironment serves to metabolically support DTCs during metastasis; however, a metabolic-based regulatory role for immune cells in this context has not been systematically examined.

In this study, we show that lung-infiltrating neutrophils function as a nutrient source to fuel DTCs during lung metastasis of breast cancer in mouse models. Neutrophils potently accumulate neutral lipids starting from the premetastatic stage, which is stimulated by the lung-resident mesenchymal cells (MCs). Once metastatic breast tumor cells colonize the lung, they take up lipids from lung neutrophils and acquire elevated levels of survival and proliferation. Our results therefore revealed an unrecognized role of neutrophils to metabolically regulate breast cancer lung metastasis.

Results

Neutrophils have a lipid-laden phenotype in the lung

Preceding tumor metastasis, in response to tumor- and host-derived factors, organ-infiltrating neutrophils are generated in hematopoietic tissues and organs such as BM and spleen12. While neutrophils have been extensively studied for their metastasis-modulating effects, it remains largely unknown whether these effects are intrinsic or whether they are acquired as neutrophils transit specific tissue environments. Using the mouse 4T1 orthotopic breast tumor model, we compared transcriptional profiles of neutrophils isolated from BM, peripheral blood (PB) and lung by RNA sequencing (RNA-seq). At the premetastatic stage, a fundamental difference in gene expression was detected between lung neutrophils and those isolated from BM or PB (Fig. 1a and Extended Data Fig. 1a,b).

Fig. 1: Neutrophils acquire a lipid-laden phenotype in the premetastatic lung.
figure1

a, Volcano plots showing fold change and P value for the comparison of lung neutrophils (neu) versus PB neu on the basis of RNA-seq data from the 4T1 orthotopic model (n = 3). b, Intracellular lipids in neutrophils detected under confocal microscopy (top: scale bars, 5 μm) or transmission electron microscopy (bottom: scale bars, 0.5 μm). Arrowheads indicate LDs (representative of n = 3 mice). c, Immunostaining of lung section of MMTV-PyMT mice showing in situ LDs (BODIPY) in neu (Ly6G). Scale bars, 5 μm (representative of n = 3 mice). d, Lipid quantification in neu by flow cytometry (n = 6). MFI, mean fluorescence intensity. e, Cellular TG contents in neu (n = 5). f, Relative levels of the indicated types of lipids in PB and lung neu measured by mass spectrometry (n = 5; Extended Data Fig. 2a–h). CE, cholesteryl ester; PE, phosphatidylethanolamine; PC, phosphatidylcholine; PS, phosphatidylserine; PI, phosphatidylinositol; Cer, ceramides; PG, phosphatidylglycerol. g, Diagram showing ATGL and its regulatory factors in TG hydrolysis. h, Relative mRNA expression (as compared to the control gene Rps18, encoding ribosomal protein S18) of the indicated genes in neu determined by RT–qPCR (n = 5). i, The indicated protein expression in neu detected by immunoblotting, with glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as a loading control (n = 3). j, Correlation between expression of human LD-associated genes and that of neu signature genes in human breast cancer lung metastasis samples (GSE14018, n = 16), analyzed by linear regression with Pearson’s correlation (Methods). k, Relative TG hydrolase activity in neu (n = 6 and 5 for PB and lung, respectively). l, Cellular lipase activity in neu (n = 5). Neu were isolated from the indicated tissues from tumor-bearing MMTV-PyMT mice at the premetastatic stage (b,d,e,h,i,k,l). In all cases, n represents biologically independent animals except for j, where it represents independent human patients. Data are mean ± s.e.m. and P values in animal experiments were determined by one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test, except for a and k, which used unpaired two-tailed Student’s t-tests. NS, not significant.

Source data

Two lipid droplet (LD)-associated genes, Hilpda and G0s2 (refs. 13,14), were listed among the top differential genes in lung neutrophils versus BM neutrophils (Extended Data Fig. 1a) or PB neutrophils (Fig. 1a), whereas their expression levels were comparable between BM and PB neutrophils (Extended Data Fig. 1b). Further inspection of the genes involving in the main lipid metabolic pathways15 (Extended Data Fig. 1c) revealed that a proportion of genes associated with lipid absorption and LD formation, but not LD degradation and β-oxidation, are expressed at higher levels in lung neutrophils in comparison to their BM and PB counterparts (Extended Data Fig. 1d). Therefore, neutrophils tend to upregulate lipid storage genes in the premetastatic lung.

Consistent with gene expression profiles, neutrophils isolated from or residing in premetastatic lungs, were indeed found to contain a greater abundance of LD-like structures than those from BM and PB in both the orthotopic 4T1 model (Extended Data Fig. 1e,f) and the genetically engineered MMTV-PyMT metastatic mammary cancer model (Fig. 1b,c). Quantitatively, total lipids in lung neutrophils were also much higher than in neutrophils isolated from other tissues, including BM, PB, spleen and primary tumors in both the orthotopic 4T1 model (Extended Data Fig. 1g,h) and the MMTV-PyMT model (Fig. 1d). The lung-associated lipid-laden phenotype of neutrophils was not only detected under tumor-bearing conditions, but it also existed in tumor-free naive mice, although the contrast observed between lung neutrophils and their counterparts in other tissues was less marked (Extended Data Fig. 1h). Hence, lung-resident neutrophils intrinsically tend to accumulate lipids, which is further reinforced during breast cancer progression.

As the major intracellular organelles storing neutral lipids, especially triglyceride (TG), LDs are mostly studied in adipocytes and have recently been recognized as a critical metabolic player in other types of cells, such as macrophages in infectious diseases and atherosclerosis16,17. In accordance with their high LD content, lung neutrophils contain higher levels of TG than their BM and PB counterparts in both the MMTV-PyMT and the orthotopic 4T1 models (Fig. 1e and Extended Data Fig. 1i). Additional mass spectrometric analysis revealed that all TG species were detected at much higher levels in lung neutrophils than in PB neutrophils, whereas lipid types other than TG did not show such a notable difference (Fig. 1f and Extended Data Fig. 2a–h). Thus, neutrophils acquire a TG-enriched lipid-laden phenotype in the premetastatic lung.

By probing into the TG metabolic genes from the RNA-seq data, the genes governing TG hydrolysis were found to differ greatly between lung neutrophils and BM or PB neutrophils (Extended Data Fig. 1d). Hydrolysis of TG is primarily catalyzed by ATGL, which is regulated by its activator protein α-β hydrolase domain-containing protein 5 (ABHD5) and its inhibitory factors HILPDA, CIDEC and G0S2 (refs. 18,19; Fig. 1g). At both transcriptional and protein levels, all three ATGL inhibitory factors were each more highly expressed in lung neutrophils compared to those isolated from BM or PB in both the MMTV-PyMT (Fig. 1h,i) and the 4T1 models (Extended Data Fig. 2i,j). In contrast, the genes encoding ATGL and its activator ABHD5 did not show such a tissue-specific difference in expression (Fig. 1h and Extended Data Fig. 2i). By analyzing a transcriptome dataset (GSE14018) derived from human breast cancer lung metastasis samples20, the expression of HILPDA, CIDEC and G0S2 was also found to be positively correlated with that of neutrophil signature genes21 (Fig. 1j), suggesting that human neutrophils may undergo similar metabolic changes during lung metastatic progression of human breast cancer.

As a result of the high expression of ATGL inhibitory factors, the TG hydrolase activity of lung neutrophils was expectedly lower than that in PB neutrophils (Fig. 1k for the MMTV-PyMT model; Extended Data Fig. 2k for the 4T1 model). A similar pattern was also shown for total lipase activity (Fig. 1l and Extended Data Fig. 2l). Therefore, the lipid-laden phenotype in lung neutrophils is likely caused by their elevated expression of ATGL inhibitory factors, which in turn repress their TG hydrolysis, leading to intracellular TG accumulation.

Lung-resident mesenchymal cells drive neutrophils to accumulate lipids

To directly determine whether the lung is a unique site for neutrophils to accumulate lipids, we intravenously implanted fluorescent dye-labeled, BM-isolated neutrophils. In both tumor-free and tumor-bearing (premetastatic stage) recipient mice, the implanted neutrophils that infiltrated the lung, but not other tissues, substantially increased their lipid levels (Fig. 2a). We then ex vivo co-cultured various types of lung stromal cells with BM-isolated neutrophils to identify whether certain lung stromal cells drive neutrophils to accumulate lipids. Among the lung stromal cells examined, the CD140a+ MCs were found to significantly upregulate neutrophil lipids (Fig. 2b). By immunostaining of premetastatic lung sections from MC-reporter CD140a-GFP mice, we observed a close proximity of in situ neutrophils to lung MCs within the lung interstitium (Fig. 2c). Hence, when neutrophils infiltrate the lung interstitium, they could be elicited by their proximal resident MCs to start the process of lipid storage.

Fig. 2: Lung CD140a+ MCs drive lipid accumulation in neutrophils.
figure2

a, Comparison of lipid levels in implanted neutrophils (neu) in the indicated organs or tissues of recipient naive or 4T1 tumor-bearing mice (Methods; n = 6). b, BM neu were monocultured or co-cultured with the indicated stromal cells sorted from naive BALB/cJ mouse lungs. Lipid levels of neu were quantitated by flow cytometry 16 h later. c, Immunostaining of lung section showing the localization of MCs and neu in AT3-g-csf orthotopic tumor-bearing CD140a-GFP mice (premetastatic stage) (representative of n = 4 mice). Receptor for advanced glycation end products (RAGE) was also stained to indicate the lung epithelium. Scale bar, 20 μm. d, Relative mRNA expression of the indicated genes in BM neu after their monoculture or co-culture with CD140a+ lung MCs or lung MC-conditioned medium. e, An upstream regulator (soluble proteins or chemicals) prediction was performed using ingenuity pathway analysis on the basis of differentially expressed genes between lung and BM neu, using the RNA-seq data as in Extended Data Fig. 1a. CSF2, colony stimulating factor 2; HGF, hepatocyte growth factor; TNFSF11, tumor necrosis factor ligand superfamily member 11; CD40LG, CD40 ligand; PGE2, prostaglandin E2; IL, interleukin; IGF2, insulin-like growth factor 2; TGF-β1, transforming growth factor-β1. f, Lipid levels of BM neu stimulated with the indicated candidate regulators, as measured by flow cytometry (Methods). g, Relative mRNA expression of the indicated genes in BM neu stimulated with PGE2 (10 ng ml−1). h, Lipid levels of BM neu co-cultured with CD140a+ MCs in the presence and absence of the PGE2 receptor (EP) antagonists were quantitated by flow cytometry (Methods). i, Measurement of lung neu lipid levels in 4T1 tumor-bearing mice without and with administration of the EP2 antagonist (PF-04418948, 10 mg kg−1; n = 6). The BM neutrophils used in b,d and fh were isolated from 4T1 tumor-bearing BALB/cJ mice (n = 4, except g, with n = 3). Values of n represent biologically independent animals. Data are mean ± s.e.m. and P values were determined by one-way ANOVA with Tukey’s multiple comparisons test, except for g and i, which used unpaired two-tailed Student’s t-tests. NS, not significant.

Source data

As HILPDA, CIDEC and G0S2 were among the most upregulated TG-regulating factors in lung neutrophils, we speculated whether lung MCs trigger neutrophils to accumulate lipids through regulating these ATGL inhibitory factors. Indeed, lung MCs were able to stimulate the expression of all three genes encoding ATGL inhibitory factors, whereas the lung MC-conditioned medium could only upregulate Hilpda expression in neutrophils (Fig. 2d). Therefore, lung MCs modulate lipid metabolic gene expression in neutrophils through both soluble factor(s) and cell–cell contact-dependent mechanisms.

Through a screening for the possible lung MC-derived candidate soluble factors (Fig. 2e), as predicted by ingenuity pathway analysis of the RNA-seq data from lung versus BM neutrophils in the 4T1 model, PGE2 was found to be superior to other candidate factors in elevating the neutrophil lipid level (Fig. 2f). Consistently, PGE2 was able to upregulate Hilpda but not Cidec or G0s2, confirming that PGE2 is at least one soluble factor mechanism underlying the lung MC-triggered neutrophil lipid-laden phenotype (Fig. 2g). PGE2, a bioactive lipid belonging to the family of eicosanoids, has been extensively reported to play multifaceted roles in inflammation and cancer22. PGE2 exerts its effects through binding to PGE2 receptors (EP1–4) expressed on the target cell surface23. By further ex vivo and in vivo work, EP2 was determined as a primary PGE2 receptor to respond to lung MCs (Fig. 2h,i). Taken together, PGE2 was revealed as a possible lung MC-derived soluble factor that modulates the lipid profiles in neutrophils.

A PGE2–HIF1α–HILPDA axis is revealed for lung MC-driven neutrophil lipid accumulation

To better understand the role of PGE2 in endowing the lung-associated neutrophil lipid-laden phenotype, we next characterized PGE2 production by different tissue-derived MCs. Lung MCs isolated from naive mice were shown to intrinsically produce a considerable amount of PGE2 (1–4 ng per 105 cells in 24 h), whereas this capacity was further increased under tumor-bearing conditions (Fig. 3a). Furthermore, lung MCs produced apparently higher levels of PGE2 than those isolated from the other tissues examined, including bone, mammary gland, liver and heart (Fig. 3b). This differential pattern in PGE2 production among tissue-specific MCs, as expected, corresponded well with their abilities to stimulate neutrophils for lipid accumulation (Fig. 3c).

Fig. 3: A critical role of the PGE2–HIF1α–HILPDA axis in lung MC-triggered lipid storage in neutrophils.
figure3

a, The PGE2 levels in lung MCs (105 cells ml−1 in 24 h) from naive and AT3-g-csf tumor-bearing CD140a-GFP mice (n = 6). bd, CD140a+ MCs were freshly isolated from the indicated tissues of naive CD140a-GFP mice (n = 3). Their PGE2 secretion (b), abilities to stimulate lipid accumulation in BM neutrophils (neu) (c) and relative mRNA expression (as compared to Rps18) of the indicated genes (d) were assessed. MG, mammary gland. eg, Lung MC-specific COX2 in eliciting the lipid-laden neu phenotype. In vitro, BM neu were monocultured or co-cultured with lung MCs isolated from WT or Cox2-cKO (CD140a-Cre; Cox2flox/flox) mice. Then lipid levels (e) and relative mRNA expression of the indicated genes (g) in neu were determined. The BM neu were isolated from the AT3-g-csf tumor-bearing mice (n = 4). In vivo, the lipid levels in lung neu isolated from the AT3-g-csf tumor-bearing WT and Cox2-cKO mice (n = 5) were compared (f). h, Relative mRNA levels of Hilpda in lung neu isolated from AT3-g-csf tumor-bearing WT and Hif1a-cKO (S100a8-Cre; Hif1aflox/flox) mice (n = 5). i, Lipid levels in lung neu from AT3-g-csf-bearing WT (n = 3) and Hif1a-cKO (n = 4) mice were detected by confocal microscopy (left; scale bars, 5 μm) or flow cytometry (right). j,k, BM neu isolated from AT3-g-csf tumor-bearing WT or Hif1α-cKO mice (n = 4) were monocultured or co-cultured with lung MCs (j) or stimulated with exogenous PGE2 (10 ng ml−1) (k). Then lipid levels in neu were determined. l,m, Lung MCs were cultured with the indicated conditioned medium (CM) or IL-1β (20 ng ml−1), with and without anti-IL-1β (1 µg ml−1). PGE2 levels (l) and the relative mRNA expression of the indicated genes (m) were measured. Lung MCs were isolated from naive CD140a-GFP mice (n = 4). Values of n represent biologically independent animals. Data are mean ± s.e.m. and P values were determined by one-way ANOVA with Tukey’s multiple comparisons test, except for a,f,h,i, which used unpaired two-tailed Student’s t-tests. NS, not significant.

Source data

PGE2 is synthesized from arachidonic acid via the actions of several key enzymes, including cyclooxygenase 2 (COX2), cyclooxygenase 1 (COX1) and prostaglandin E synthases (PTGES)24. Among the three enzyme-encoding genes, Cox2 and Cox1, but not Ptges, in lung MCs were expressed at higher levels than in those isolated from other tissues examined (Fig. 3d). In light of the essential functions of COX2 in inflammation and cancer22,25, we next characterized the role of mesenchymal cell-specific COX2 in the lung premetastatic niche. In vitro, Cox2 ablation in lung MCs indeed reduced their capacity to stimulate lipid accumulation in neutrophils (Fig. 3e). In vivo, conditional knockout (cKO) of Cox2 in CD140a+ MCs significantly alleviated the lipid-laden phenotype of lung neutrophils (Fig. 3f). Thus, lung MC Cox2-PGE2 signaling is critical for inducing the neutrophil lipid-laden phenotype.

Among the three ATGL inhibitory factors, only Hilpda was found to be dependent on lung MC Cox2-PGE2 signaling (Fig. 3g). HILPDA, also known as hypoxia-inducible gene 2 (HIG2), was recently reported as a hypoxia-induced endogenous inhibitor of ATGL in cancer cells and macrophages26,27. The cKO of hypoxia-inducible factor 1-α (Hif1a), the master transcriptional regulator of cellular response to hypoxia28 in neutrophils (S100a8-Cre; Hif1aflox/flox)29 led to diminished expression of Hilpda (Fig. 3h) and mitigated the lipid-laden phenotype in lung neutrophils (Fig. 3i). Ex vivo, loss of Hif1a in neutrophils abated their lipid-accumulating capacity in response to lung MCs or exogenous PGE2 (Fig. 3j,k). Collectively, a COX2-PGE2–HIF1α–HILPDA axis was revealed as a mechanism linking lung MCs to lipid-laden lung neutrophils.

Last, we sought to determine how tumor-bearing conditions upregulate PGE2 production in lung MCs. It is well known that tumor cells themselves and host immune cells can remotely regulate organ stromal cells via soluble factors2. Indeed, tumor cell and lung neutrophil-derived conditioned medium were able to upregulate Cox2 gene expression and PGE2 production in lung MCs (Fig. 3l,m). To identify such tumor cell- and/or neutrophil-derived soluble factor(s), we focused on inflammatory cytokines that are known to be pivotal in formation of the premetastatic niche1,2. Re-examination of the neutrophil RNA-seq data (Extended Data Fig. 1a) revealed that Il1b was one of the most highly expressed inflammatory cytokine genes in lung neutrophils versus BM neutrophils. Neutralization of IL-1β indeed reduced neutrophil or tumor cell conditioned medium-induced Cox2 expression and PGE2 production in lung MCs ex vivo (Fig. 3l,m). Addition of exogenous IL-1β was similarly effective to the conditioned media in induction of Cox2 and PGE2 in lung MCs (Fig. 3l,m). Therefore a neutrophil-IL-1β → lung MC-PGE2 → neutrophil-lipids amplifying loop serves to reinforce the neutrophil lipid-laden phenotype in the lung premetastatic niche.

Lipids stored in neutrophils function to promote lung metastasis of breast cancer

LD formation has been documented in macrophages in infectious diseases and atherosclerosis in which LDs serve as an energy source for pathogens, elicit inflammatory responses by producing eicosanoids and mediate atherosclerotic plaque formation17,30. To understand the functional consequence of LD formation in premetastatic lung neutrophils, we specifically deleted LD-associated genes, including Atgl and the ATGL inhibitory genes Hilpda and Cidec, in neutrophils. Upon crossing S100a8-Cre driver mice targeting neutrophils29,31 with genetically modified mice that have loxP-flanked Atgl, Hilpda or Cidec, the resultant cKO strains (S100a8-Cre; Atglflox/flox; S100a8-Cre; Hilpdaflox/flox; and S100a8-Cre; Cidecflox/flox) were exploited in the orthotopic AT3-g-csf tumor model. The AT3-g-csf line is based on the murine breast cancer cell line AT3 originated from a MMTV-PyMT tumor in the C57BL/6J background32, and further constructed to overexpress granulocyte colony-stimulating factor (G-CSF) for induction of the host inflammatory condition6 at a similar level to the 4T1 model.

Corresponding to the elevated lipid level in lung neutrophils (Fig. 4a), more robust spontaneous lung metastases were detected in Atgl-cKO mice, compared to their wild-type (WT) controls, whereas the primary tumor growth remained unaltered (Fig. 4b,c). In contrast, genetic ablation of either Hilpda or Cidec in neutrophils was not sufficient to influence the spontaneous lung metastases although each deficiency did cause a significant reduction of lipid levels in lung neutrophils (Extended Data Fig. 3a–f). We reasoned that the three ATGL inhibitory factors in neutrophils may compensate each other for their tumor-modulating effects when a single factor was defective, whereas loss of Atgl in neutrophils caused fundamental alterations of their lipid-laden phenotype and metastasis-regulating ability.

Fig. 4: Genetic ablation of Atgl in host neutrophils leads to more aggressive lung metastasis of breast cancer.
figure4

a, Representative images showing LDs in lung neutrophils from AT3-g-csf-bearing WT and Atgl-cKO mice (left). Scale bars, 5 μm. Lipid levels of neutrophils from above mice (n = 6) were quantitated by flow cytometry (right). b,c, WT (n = 15) and Atgl-cKO mice (n = 16) were orthotopically implanted with AT3-g-csf cells. The primary tumors were resected and weighed on day 15 (n = 10) (b). At the end point (day 30), spontaneous lung metastases were assessed. Representative histological images of the lung sections stained with hematoxylin and eosin (H&E) are shown and arrowheads indicate metastatic lesions (c, left). Scale bars, 1 mm. The number of metastatic nodules on the lungs were counted (c, right). df, Following the modified experimental lung metastasis model (Extended Data Fig. 3g and Methods), the lung metastatic progression of AT3-Luc cells in Atgl-cKO recipient mice (d) and Hilpda-cKO recipient mice (e), as well as E0771-Luc cells in Atgl-cKO recipient mice (f) was determined by ex vivo BLI, with their WT littermates as controls. Representative BLI images of the collected lungs are shown (df, left); n = 11 (WT) and n = 10 (Atgl-cKO) for d, n = 13 (WT) and n = 14 (Hilpda-cKO) for e and n = 10 per group for f. g, Comparison of the spontaneous lung metastases developed in MMTV-PyMT Atgl-cKO mice (n = 8) and their WT littermates (MMTV-PyMT mice; n = 9). Representative histological lung sections stained with H&E are shown and arrowheads indicate metastatic nodules. Scale bars, 1 mm. Values for n represent biologically independent animals. Data are mean ± s.e.m. and P values were determined by unpaired two-tailed Student’s t-test. NS, not significant.

Source data

Neutrophils play a critical role in the metastatic colonization step and are a key component of the lung premetastatic niche in breast cancer models1,2,3. To specifically assess the functional contribution of lipid-laden neutrophils in lung metastatic colonization and avoid potential confounding of similar effects that could occur in the primary tumor microenvironment, we next adopted a modified experimental lung metastasis model (Extended Data Fig. 3g). In this model, AT3-g-csf tumor cells were first orthotopically implanted in WT and the lipid metabolic gene cKO mice to develop host conditions with varying neutrophil lipid storage statuses. At the premetastatic stage for orthotopic tumors, the luciferase-labeled AT3 cells (AT3-Luc) were intravenously injected in the WT and cKO mice. By monitoring the injected tumor cells with bioluminescence imaging (BLI), the impact of host neutrophil lipid status on metastatic colonization can be quantitatively evaluated. As expected, the neutrophil lipid-high host condition (Atgl-cKO) apparently favored the AT3 tumor colonization and outgrowth in the lung (Fig. 4d). Conversely, metastasis colonization was significantly inhibited by the neutrophil lipid-low host environment (Hilpda-cKO) (Fig. 4e). Comparing this result to the insignificant effect of Hilpda-cKO in the spontaneous metastasis model (Extended Data Fig. 3c), it is plausible to speculate that HILPDA-associated lipid metabolism in neutrophils exerts a more prominent effect in certain metastatic step(s) such as colonization, but is insufficient to alter the entire process of metastasis in vivo. Furthermore, genetic ablation of Cidec in host neutrophils, also showed a tendency to suppress the metastatic colonization of AT3 cells (Extended Data Fig. 3h).

Using another syngeneic breast tumor cell line E0771, which is genetically distinct from the AT3 line, a similar metastasis-promoting effect was found when the neutrophil lipid-laden phenotype was reinforced by Atgl-cKO in the experimental metastasis model (Fig. 4f). Moreover, targeted deletion of neutrophil Atgl in the MMTV-PyMT transgenic strain (MMTV-PyMT; S100a8-Cre; Atglflox/flox) significantly augmented spontaneous lung metastasis (Fig. 4g). Thus, neutrophil-specific ablation of LD-associated genes modulated lipid storage in lung neutrophils and consequently led to altered lung metastasis of breast tumors.

Neutrophils transfer their stored lipids to metastatic tumor cells via release of vesicles

Lipid metabolism has been extensively studied in tumor cells themselves33, whereas the metabolic crosstalk between the tumor microenvironment and tumor cells remains less characterized. Limited evidence about such a crosstalk has been documented between tumor cells and adipose tissues at the primary and metastatic sites11,34,35,36. We suspected that the lipid-laden neutrophils might play an ‘adipose’ role in fueling DTCs based on our observation that early lung-colonizing tumor cells were exclusively embraced by Ly6G+ neutrophils (Fig. 5a). We first tested this possibility by ex vivo co-culturing tumor cells with neutrophils preloaded with fluorescence-labeled palmitic fatty acid. Following co-culture, the labeled lipids originally only present in neutrophils were detected in tumor cells, suggesting possible lipid transport from neutrophils to tumor cells (Extended Data Fig. 4a). We further compared the PB and lung neutrophils, derived from tumor-bearing mice, for their capacities to transfer lipids to tumor cells. As lung neutrophils contained intrinsically higher levels of total lipids and TGs than PB neutrophils, consistently, tumor cells acquired more abundant lipids and TGs after co-culture with lung neutrophils than those with PB neutrophils (Fig. 5b and Extended Data Fig. 4b,c).

Fig. 5: Lung-infiltrating neutrophils transfer their stored lipid to metastatic tumor cells.
figure5

a, Adjacent localization of early-seeded 4T1-mCherry cells and neutrophils (neu) (Ly6G) in the lung (Methods). Scale bars, 20 μm (representative of n = 3 mice). b, The cellular TG contents of 4T1-mCherry cells after monoculture or co-culture with PB or lung neu (Methods). c, Measurement of lipids in early lung-colonizing 4T1-mCherry cells in G-CSF-pretreated mice without and with anti-Ly6G-based neu depletion (Methods) (n = 5). d, Lipid transfer from BODIPY-loaded lung neu to tumor cells in vivo as determined by flow cytometry (Methods; n = 4). e, 4T1 cells were fed with the unseparated CM, small-sized (<100 kDa) or large-sized (>100 kDa) CM portion prepared from BODIPY-loaded lung neu. Then lipids incorporated into the tumor cells were detected 4 h later. Scale bars, 20 μm. f, Protein and TG levels in the large-sized lung or PB-derived neu CM portions (n = 6). g, Localization of the absorbed lipids from BODIPY-loaded lung neu with lysosomes (Lysotracker) in 4T1 tumor cells. Scale bars, 5 μm. h, The effects of EIPA, genistein and chlorpromazine in blockage of lipid transport from lipid-laden lung neutrophils to tumor cells (Methods). Scale bars, 20 μm. i, The impact of EIPA on the capacity of 4T1 cells to acquire lipids from lung neutrophils. j, 4T1 cells were co-cultured with lung neu-derived vesicles and imaged by transmission electron microscopy (Methods). Red arrowheads indicate the engulfment of the lipid-like vesicles and yellow arrowheads indicate macropinosome-like structures. Scale bars, 0.5 μm. The lung and PB neu used throughout were isolated from the premetastatic stage of 4T1 tumor-bearing mice. Values for n represent biologically independent animals. Data are mean ± s.e.m. for c and f and mean ± s.d. from three (b) or four (i) biologically independent cell cultures. For e,g,h and j, data are representative of three biologically independent experiments. P values were determined by one-way ANOVA with Tukey’s multiple comparisons test (b,c,i) or unpaired two-tailed Student’s t-test (f). NS, not significant.

Source data

To gain insight into the in vivo lipid transport from neutrophils to tumor cells, we pretreated naive mice with recombinant mouse G-CSF to induce a host condition with high neutrophil infiltration before infusing mCherry-labeled tumor cells. In both 4T1 and AT3 models, a higher lipid level was detected in tumor cells colonized in G-CSF-pretreated mouse lungs than those in control mice (Fig. 5c and Extended Data Fig. 4d). Such a lipid elevation in injected tumor cells was nearly abolished by neutrophil depletion, confirming a specific role of neutrophils as the lipid source fueling tumor cells in the lung microenvironment (Fig. 5c and Extended Data Fig. 4d). To obtain more direct in vivo evidence of neutrophil–tumor cell lipid transfer, neutrophils prelabeled with fluorescent lipids and tumor cells were successively infused into the recipient mice. By analyzing the freshly isolated lung tissues, the fluorescent lipids originally present in neutrophils were detected in tumor cells (Fig. 5d and Extended Data Fig. 4e). Therefore, lipid-laden neutrophils were indeed capable of transferring their stored lipids to metastatic tumor cells. Moreover, lipid transfer was further reinforced when neutrophils underwent apoptosis upon activation (Extended Data Fig. 4f). Considering that neutrophils are characterized by a short lifespan37, the rapid neutrophil turnover may expedite their lipid transfer to early arrived DTCs to satisfy their high energy demands7,38.

Free fatty acids (FFAs) have been regarded as the main form of lipid transfer from adipose tissues to tumor cells11,34,35,36. More recently, an alternative lipid transfer form, exosome-sized lipid-filled vesicles, was shown to mediate lipid transport from adipocytes to macrophages39. To determine the lipid transfer form from lung neutrophils to tumor cells in our study, we divided the BODIPY-labeled lung neutrophil-derived conditioned medium (CM) into a portion that would contain FFAs and most soluble proteins (<100 kDa) and a portion containing non-FFA vesicles and larger proteins (>100 kDa). Upon incubating these two CM portions with tumor cells, the large-sized CM portion was determined to be the primary lipid contributor for tumor cells (Fig. 5e). In the large-sized CM portion prepared from premetastatic lung neutrophils, there was a much higher level of TG, but not proteins, compared to that prepared from PB neutrophils (Fig. 5f). Therefore, the lipid transport from neutrophils to tumor cells is mainly via large-sized vesicles and not FFAs.

Engulfment of large-sized extracellular vesicles by mammalian cells is usually achieved via a process called macropinocytosis40. Under nutrient deprivation stress, tumor cells are able to exploit macropinocytosis to take up extracellular nutrients to support their growth using the lysosomal pathway41,42. In our study, upon lipid transport from neutrophils to tumor cells, the lipids were indeed detected as being localized within tumor cell lysosomes (Fig. 5g). Furthermore, selective inhibition of macropinocytosis by 5-(N-ethyl-N-isopropyl) amiloride (EIPA)42, but not inhibition of clathrin- or caveolae-dependent endocytosis43, completely abolished lipid acquisition in tumor cells from lipid-laden lung neutrophils (Fig. 5h,i). Under transmission electron microscopy, the membrane protrusions and retractions, and subsequent formation of large macropinosome-like vacuoles in tumor cells, could be readily detected when neutrophil-derived vesicles were fed to tumor cells (Fig. 5j). Taken together, these data suggested that a macropinocytosis–lysosome pathway governs lipid transport from lung neutrophils to metastatic tumor cells.

Tumor cells engulfing neutrophil-derived lipids acquire an augmented proliferative capacity

We next determined how tumor cell behaviors are altered in tumor cells after they ingest lipids from neutrophils. In vitro, tumor cells co-cultured with premetastatic lung neutrophils significantly upregulated their expression of key lipolysis genes—lipase E (Lipe), Atgl and lipase A (Lipa), as well as fatty acid oxidation-associated genes, including carnitine palmitoyltransferase 1B (Cpt1b), carnitine palmitoyltransferase II (Cpt2), enoyl-CoA hydratase 1 (Ech1) and acyl-CoA oxidase 1 (Acox1) (Fig. 6a). Corresponding to the upregulation of these lipid utilization genes, lung neutrophil-educated tumor cells acquired an elevated oxygen consumption rate (OCR), indicating increased oxidative phosphorylation, which was significantly suppressed by etomoxir, a CPT1 inhibitor (Fig. 6b and Extended Data Fig. 5a). Therefore, acquisition of neutrophil-derived lipids drives tumor cells to undergo a metabolic shift toward lipid utilization.

Fig. 6: Neutrophil-derived lipids enhance the proliferative capacity of metastatic tumor cells.
figure6

a, Relative mRNA expression of the indicated genes in 4T1 cells upon monoculture or co-culture with PB or lung neutrophils (neu). b, OCR of 4T1 cells, following monoculture or co-culture with PB or lung neu, were measured by Seahorse XFe96 Analyzer (left). The amount of OCR derived from fatty acid oxidation was quantified as the magnitude of the response to etomoxir treatment (right). c, Measurement of the proliferative capacities of 4T1-Luc cells without and with co-culture with PB or lung neu (Methods). d, Comparison of the prosurvival abilities of 4T1-Luc cells without and with co-culture with PB or lung neu, when de novo lipogenesis was inhibited by cerulenin or C75 (Methods). e, Comparison of the metastatic colonization potentials of 4T1-Luc cells without and without co-culture with PB or lung neu (Methods). Representative BLI images of the recipient mice and quantification of BLI signals within the lung areas are shown (n = 5). f, Detection of Ki67+ lung-colonizing 4T1-mCherry cells in control (high neu) and neu-depleted (low neu) 4T1-bearing mice (Methods). Representative microscopy images; scale bars, 50 μm (left). Percentages of Ki67+ cells in total mCherry+ 4T1 cells (right). Five pictures were chosen from each mouse lung section; n = 3 mice per group. Data are mean ± s.d. Neu throughout this figure were isolated from the premetastatic stage of 4T1 tumor-bearing mice. Data are mean ± s.e.m. for e and mean ± s.d. from four (a, c, d) or six (b) biologically independent cell cultures. P values were determined by one-way ANOVA with Tukey’s multiple comparisons test (a,b and e), two-way ANOVA with Šidák’s multiple comparisons test (c,d, monoculture versus lung neu co-culture) or unpaired two-tailed Student’s t-test (f). NS, not significant.

Source data

In accordance with the above-mentioned metabolic changes, tumor cells, upon co-culture with lung neutrophils, had a higher proliferative capacity under nutrient deprivation (Fig. 6c and Extended Data Fig. 5b–d) and a more potent prosurvival ability when de novo lipogenesis was inhibited in vitro (Fig. 6d), compared to tumor cells after co-culture with PB neutrophils. In vivo, tumor cells educated by lung neutrophils gained a higher potential for metastatic colonization than those educated by PB neutrophils (Fig. 6e and Extended Data Fig. 5e–g). In situ, the proportion of Ki67+ proliferating lung-colonizing tumor cells was significantly reduced when neutrophils were depleted (Fig. 6f). Therefore, uptake of lung neutrophil-derived lipids endowed tumor cells with more robust proliferative and survival capacities.

Lipid-laden neutrophils fuel breast cancer lung metastasis in vivo

Neutrophils are well known to nourish tumor cells via their secretome, including a variety of cytokines, chemokines, growth factors and extracellular matrix44. To exclude the possibility that nonlipid neutrophil-derived factors drive tumor cell modulation, we further compared the tumor-regulatory effects of neutrophils isolated from Atgl-cKO mice and their WT littermates. In line with Atgl loss-caused lipid-laden reinforcement (Fig. 7a and Extended Data Fig. 5h,i), both lung and PB neutrophils from the Atgl-cKO mice were superior to their respective WT counterparts in amplifying tumor cell colonization in the lung (Fig. 7b and Extended Data Fig. 5j). These data indicated a specific role of neutrophil-derived lipids to endow tumor cells with an increased capacity for metastatic colonization.

Fig. 7: Blockage of macropinocytosis inhibits metastatic colonization in vivo.
figure7

a, Transmission electron microscopy images of neutrophils (neu) from AT3-g-csf tumor-bearing WT and Atgl-cKO mice (representative of n = 4 mice). Arrowheads indicate lipid droplets. Scale bars, 0.5 μm. b, AT3-Luc cells, upon co-culture with PB or lung neu isolated from AT3-g-csf tumor-bearing WT or Atgl-cKO mice, were intravenously injected into NSG mice (Methods). Representative BLI images of the recipient mice and quantification of BLI signals within the lung areas are shown (n = 6). c, The impacts of host neutrophil changes in quantity and in their lipid-laden phenotype on the proliferation of lung-colonizing tumor cells, as measured by BrdU incorporation (Methods) (n = 5). d,e, The effects of EIPA treatment on lung colonization by AT3 tumor cells (d) and on primary tumor growth (e) were determined following the experimental design as in Extended Data Fig. 5k (Methods); n = 9 (vehicle group) or 10 (EIPA treatment group). Values for n represent biologically independent animals. Data are mean ± s.e.m. and P values were determined by one-way ANOVA with Tukey’s multiple comparisons test (b,c) or by unpaired two-tailed Student’s t-test (d,e). NS, not significant.

Source data

To further illustrate how host neutrophils quantitatively and qualitatively modulate colonized tumor cells in the lung microenvironment in vivo, AT3-g-csf and AT3 parental cell lines were employed to induce neutrophil-high versus neutrophil-low host conditions, while Atgl-cKO and WT mice were utilized for generating neutrophil lipid-high versus lipid-low host conditions. Upon successful colonization in the lung, tumor cell proliferation under the above-mentioned different host conditions was compared and it was clearly shown that an increase in either host neutrophil number or lipid content, was sufficient to accelerate tumor cell proliferation in the lung (Fig. 7c). Therefore, neutrophils quantitatively and qualitatively (metabolically) dominate tumor cell fates within the lung microenvironment.

As macropinocytosis was revealed as a main pathway for tumor cells to ingest neutrophil-derived lipids, we next explored whether pharmacological inhibition of macropinocytosis represents an effective approach to restrain metastatic colonization. In both 4T1 and AT3-g-csf tumor models, administration of macropinocytosis inhibitor EIPA in tumor-bearing mice significantly reduced tumor cell colonization in the lung (Fig. 7d and Extended Data Fig. 5k,l). Such EIPA-mediated blockage of lipid transport, however, did not alter primary tumor growth, suggesting that lipid-laden neutrophils may exert effects specifically at the metastatic sites (Fig. 7e and Extended Data Fig. 5m). Hence, targeting lipid metabolism of the lung microenvironment is indeed effective in controlling breast cancer lung metastasis.

Data from the mouse models described above revealed that lung MCs trigger formation of a lipid-rich premetastatic niche in the lung, and we therefore speculated that primary tumor cells with a relatively higher capacity to absorb and utilize lipids might adapt more favorably to the lipid-rich lung environment that leads to lung metastasis. To address this, we analyzed a transcriptome dataset (GSE2603) derived from human primary breast tumor samples in which patients’ lung relapse information was available45. Indeed, the primary tumors expressing higher levels of genes associated with lipid utilization pathways, including lipid catabolism, lipid transport and lipid oxidation, showed a high risk for developing lung metastases in patients with breast cancer (Extended Data Fig. 6). In addition, gene expression levels of neutrophil chemotaxis, migration and extravasation pathways were also shown to be related to lung metastasis of those patients (Extended Data Fig. 6). This result suggested that the capacity of primary tumors to elicit neutrophil trafficking to the organs, such as the lung, is critical for triggering the subsequent metastatic cascade that occurs in the lung. A further analysis on the lipid- and neutrophil-associated pathways will help develop signature markers with prognostic value for patients with breast cancer.

Overall, our work revealed a lung MC–neutrophil–tumor cell metabolic axis (Extended Data Fig. 7) that dominates the extent of metastatic colonization, the most rate-limiting step of the metastatic process in solid cancers7.

Discussion

In this study, we demonstrated that neutrophils serve as an energy source to fuel metastatic tumor cells. At the premetastatic stage of breast cancer, after their expansion in hematopoietic tissues, neutrophils infiltrate the peripheral organs and tissues. At the lung, resident MCs trigger lipid storage in infiltrating neutrophils. When DTCs arrive at the lung, lipid-laden neutrophils then transfer their stored lipids to DTCs for their survival and proliferation, leading to metastatic colonization and outgrowth.

Our results suggest that lung-resident stromal cells, specifically CD140a+ MCs, act as an initiator driving the metabolic dynamics in infiltrating neutrophils and DTCs. Although we have characterized PGE2 as a primary lung MC-derived soluble factor and PGE2→HIF1α→HILPDA as the subsequent molecular cascade in neutrophils, ongoing efforts are underway to further define the cell–cell contact mechanisms underlying CIDEC and G0S2 induction in neutrophils. In vivo, a single depletion of ATGL inhibitory genes in neutrophils only caused a mild or insignificant reduction in their metastasis-regulating capacity, whereas genetic ablation of neutrophil Atgl led to substantially elevated lung metastases of breast tumor cells. A better understanding of the molecular mechanisms behind induction of ATGL inhibitory factors will potentially allow targeting of multiple LD-associated factors that could provide more effective approaches for abolishing the neutrophil lipid-laden phenotype and its detrimental effects.

In mammals, TGs are superior to carbohydrates and proteins for energy storage due to their reduced and anhydrous characteristics46. In our study, TGs are stored as lipid droplets in neutrophils, a cell type with a fast turnover rate, in the lung microenvironment. This may facilitate a timely release of TGs from neutrophils to the early DTCs, which have high energy demands47. As macropinocytosis is known as a rapid nutrient-scavenging pathway in cancer cells independent of protein synthesis48, this endocytic pathway would further accelerate energy acquisition by DTCs from the lung microenvironment. Additional work is needed to further characterize the nature of lipid vesicles, the molecular basis of vesicle release by neutrophils and macropinocytosis by tumor cells.

In addition to energy storage, lipids also serve as precursors of various bioactive lipid mediators such as prostaglandins24,25. In a recent report, neutrophils were shown to take up lipids, particularly arachidonic acids, for synthesis of PGE2 within the primary tumor microenvironment and resulted in enhanced immunosuppressive effects31. We did compare the immunosuppression by WT and Atgl-cKO mice-derived lung neutrophils; however, they were found to be comparable in suppression of natural killer cells (data not shown), the reported major metastatic colonization-restraining immune cells6,49. It will be intriguing to further elucidate the roles of lung neutrophil-derived bioactive lipid mediators in the lung metastatic niche, as well as a possible lipid transfer from neutrophils to tumor cells within the primary tumor and other tissue microenvironments. Such work will help to develop a systemic understanding, from a metabolic perspective, of neutrophil–tumor cell crosstalk in solid tumor metastasis.

Neutrophils are known to have multifaceted functions in metastasis including both protumoral and antitumoral effects2,44. Through expression of immunoregulatory factors, growth factors, inflammatory mediators and proteinases, as well as by forming NETs, neutrophils have been mostly reported to support metastatic progression1,44. On the other hand, through production of reactive oxygen species, neutrophils were also indicated to be antimetastatic through a direct cytotoxic effect on tumor cells44. Our results add a new layer of complexity to the functions of neutrophils in metastasis by revealing a modulatory role for neutrophil metabolism and energy availability. Metabolic adaptation of DTCs in their metastasized organs is emerging as an increasingly important topic for understanding organ-tropic metastases38,50. Our findings will advance a deeper understanding of the metabolic regulation of DTCs by organ-resident stromal cells and point to the metabolism of organ-specific stroma as a new and promising therapeutic target for prevention and treatment of solid tumor metastases.

Methods

Animal studies

For all animal studies, mice of similar age were block-randomized in an unblinded fashion. Female mice aged 8–10 weeks were used unless indicated otherwise. Sample sizes were estimated based on pilot experiments and were selected to provide sufficient numbers of mice in each group for statistical analysis. Animal handing and experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee of The Jackson Laboratory. The mice were fed on a chow diet ad libitum and housed in a specific-pathogen-free facility in plastic cages at 22 °C and 40–50% humidity, with a daylight cycle from 06:00 to 18:00. The mouse strains used in this study were as follows: BALB/cJ, C57BL/6J, NOD-scid (NOD.Cg-Prkdcscid/J), NSG (NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ), MMTV-PyMT (B6.FVB-Tg(MMTV-PyVT)634Mul/LellJ), CD140a-GFP (B6.129 S4-Pdgfratm11(EGFP)Sor/J) and were obtained from the Jackson Laboratory. Cox2flox/flox (B6;129S4-Ptgs2tm1Hahe/J) and CD140a-cre+/− (C57BL/6-Tg(Pdgfra-cre)1Clc/J) mice were obtained from the Jackson Laboratory and were crossed to generate Cox2 cKO (CD140a-cre+/−Cox2flox/flox) mice. Atglflox/flox (B6N.129S-Pnpla2tm1Eek/J), Cidecflox/flox (B6.Cg-Cidectm1.1Gonz/J), Hilpdaflox/flox (Hilpdatm1.1Nat/J), Hif1aflox/flox (B6.129-Hif1αtm3Rsjo/J) and S100a8-cre+/− (B6.Cg-Tg(S100A8-cre,-EGFP)1Ilw/J) mice were obtained from the Jackson Laboratory. Each of the four floxed strains were crossed with S100a8-cre+/− transgenic mice to generate cKO mice. For example, Atglflox/flox mice were bred to S100a8-cre+/− mice to generate S100a8-cre+/-Atglflox/WT mice, which were backcrossed onto Atglflox/flox mice to generate S100a8-cre+/−Atglflox/flox mice (Atgl-cKO). Age-matched littermate Atglflox/flox mice were used as WT controls.

Male MMTV-PyMT mice were crossed with female S100a8-cre+/−Atglflox/flox mice to generate MMTV-PyMT; S100a8-cre+/−Atglflox/WT mice, which were backcrossed onto Atglflox/flox mice to generate MMTV-PyMT; S100a8-cre+/−Atglflox/flox mice (MMTV-PyMT; Atgl-cKO). Age-matched littermate MMTV-PyMT; Atglflox/flox mice were used as controls.

Spontaneous breast cancer metastasis models

  1. 1.

    Orthotopic models: cultured 4T1, AT3, E0771, AT3-g-csf or E0771-g-csf tumor cells (5 × 105 cells per mouse) were resuspended in 50 µl growth factor-reduced Matrigel (Corning) and injected into the fourth mammary fat pads of female syngeneic BALB/cJ or C57BL/6J mice. Lung metastases in these mice were monitored by histological analysis and macroscopic examination. Day 10 and earlier was determined as the premetastatic stage for the 4T1 model, whereas day 12 and earlier was the premetastatic stage for the AT3-g-csf and E0771-g-csf model. Unless otherwise stated, neutrophils used for ex vivo cultures and assays were isolated from the tissues (BM, PB, lungs, spleen and primary tumors) of tumor-bearing mice at the premetastatic stage in each model.

  2. 2.

    MMTV-PyMT transgenic model: this model was determined to be premetastatic before 5 months of age. At 5 months of age, MMTV-PyMT (C57BL/6 background) mice had spontaneously developed primary tumors (weight 0.5–1.0 g) but had no detectable lung metastases by histological analysis and macroscopic examination.

  3. 3.

    Spontaneous metastasis model in WT and the lipid metabolic gene cKO mice (Atgl-cKO, Cidec-cKO and Hilpda-cKO): mice were orthotopically implanted with AT3-g-csf cells. On day 15, primary tumors were surgically resected. On day 30, mice were killed and their lungs were excised, inflated with Bouin’s solution and the numbers of metastatic nodules were counted under a dissecting microscope.

  4. 4.

    Spontaneous metastasis model in MMTV-PyMT and MMTV-PyMT; Atgl-cKO mice: 6-month-old mice that had spontaneously developed primary tumors (2.5–3.0 g) were killed. Lungs were excised and inflated with Bouin’s solution and the numbers of metastatic nodules were counted under a dissecting microscope.

Experimental breast cancer metastasis model

C57BL/6J mice or lipid metabolic gene cKO mice were first orthotopically injected with AT3-g-csf (or E0771-g-csf) (5 × 105 cells) to generate the neutrophil-high host conditions. Luciferase-labeled AT3-Luc cells (or E0771-Luc) (5 × 105 cells per mouse) were then intravenously injected on day 10, a time point within the premetastatic stage. On day 15, primary tumors were surgically resected. At the end point (day 25), mice were killed and the lungs were rapidly excised for ex vivo BLI. Isolated lungs were placed in PBS buffer supplemented with d-luciferin (150 μg ml−1; Gold Biotechnology) in a 24-well plate and the bioluminescence signals were detected with the Xenogen IVIS system. Image exposure times were between 10 s and 30 s, depending on signal strength. Photon emission was quantified using Living Images software as the sum of all detected photon counts per second within the specified area, after subtracting background luminescence.

To explore the effect of EIPA using the experimental metastasis model described above, C57BL/6J (or BALB/cJ) mice were first orthotopically implanted with AT3-g-csf (or 4T1) to induce a neutrophil-high host condition. Luciferase-labeled AT3-Luc (or 4T1-Luc) cells were intravenously injected in the above mice at their premetastatic stage. Then EIPA (10 mg kg−1) or vehicle was given daily by intraperitoneal injection for 8 (C57BL/6J mice) or 6 (BALB/cJ mice) consecutive days. At the end point, the metastatic progression of AT3-Luc (or 4T1-Luc) cells in the lung was detected by ex vivo BLI.

Cell lines

4T1 cells and MCF7 cells were purchased from the American Type Culture Collection and E0771 cells were purchased from CH3 Biosystems. MDA-4175 cell line, originally generated from J. Massagué’s laboratory (Memorial Sloan-Kettering Cancer Center)45, was kindly provided by Y. Kang (Princeton University). AT3 cell line was kindly provided by S.I. Abrams (Roswell Park Comprehensive Cancer Center), which was originally generated from S.I. Abrams’s laboratory32. E0771 cells were maintained in RPMI-1640 supplemented with 10% fetal bovine serum (FBS) and all other cell lines were maintained in DMEM supplemented with 10% FBS. All cells were cultured in a 5% CO2 humidified incubator at 37 °C. No further authentication of these cell lines was performed. Cell morphology and growth characteristics were monitored during the study and compared to published reports to ensure their authenticity. All cell lines used in this study were determined to be negative for Mycoplasma before experiments, and were used within six passages after thawing.

Lentivirus-mediated tumor cell labeling or gene overexpression

For luciferase labeling, parental 4T1, AT3, E0771, MCF7 and MDA-4175 cells were infected with luciferase-expressing lentivirus (Addgene, 17477) to generate 4T1-Luc, AT3-Luc, E0771-Luc, MCF7-Luc and MDA-4175-Luc cell lines, respectively. Construct-positive cells were selected with G418 sulfate. For mCherry labeling, parental 4T1, AT3 and MDA-4175 cells were infected with mCherry-expressing lentivirus (Addgene, 36084) to generate 4T1-mCherry, AT3-mCherry and MDA-4175-mCherry cells, respectively. After 48 h, mCherry+ cells were sorted from infected cells by FACS. To overexpress G-CSF, AT3 and E0771 cells were infected with g-csf (Csf3)-expressing lentivirus (the vector was a gift from R.A. Weinberg, Massachusetts Institute of Technology). Construct-positive cells (AT3-g-csf and E0771-g-csf) were selected with G418 sulfate.

Tissue dissociation

Killed mice were perfused with 5 ml PBS through the right ventricle. Lungs were inflated with 1–2 ml protease solution (1.5 mg ml−1 collagenase type I (Thermo Fisher Scientific), 0.1 mg ml−1 elastase (STEMCELL Technologies), 5 U ml−1 dispase II (Thermo Fisher Scientific) and 0.1 mg ml−1 DNase I (Sigma) in HBSS), cut into small pieces (<2 mm2) and incubated in 2 ml protease solution for 30 min at 37 °C with frequent agitation. RPMI-1640 + 5% FCS was added and tissues were disrupted by pipetting and washed with RPMI-1640. After filtering through a 70-μm cell strainer, cells were washed with 5 ml RPMI-1640 + 5% FCS, incubated at room temperature for 1.5 min in 2 ml ACK red blood cell lysis solution, filtered through a 40-μm strainer, centrifuged and resuspended in RPMI-1640 + 5% FCS.

Primary tumors were dissected, minced and digested with enzyme mix (1.5 mg ml−1 collagenase I, 0.75 mg ml−1 hyaluronidase and 0.1 mg ml−1 DNase I) in HBSS for 1 h at 37 °C. Enzymatic reactions were stopped by addition of 5% FCS in DMEM and suspensions were dispersed through a 70-μm cell strainer; the spleen was removed into a Petri dish and smashed gently by full-frosted microscope slides, then the suspension was filtered through a 70-µm cell strainer. BM was collected from the tibias and femurs of both hind legs of the mice and flushed using RPMI-1640 + 5% FCS through a 70-μm cell strainer. Peripheral blood was collected by cardiac puncture with 5 mM EDTA in PBS as an anticoagulant. For flow cytometric analysis or further purification, single-cell suspensions of tumor, spleen, BM and PB were subjected to hypotonic lysis (ACK red blood cell lysis solution) to remove erythrocytes and washed with 1× PBS containing 0.5% BSA and 2 mM EDTA.

Purification and in vitro culture of neutrophils

Neutrophils were purified from the digested tissues by using anti-Ly6G MicroBead Kit (Miltenyi Biotec), following manufacturer’s instructions. The isolated neutrophils were analyzed by flow cytometry and the cells with >95% purity were used for the next procedure.

The purified neutrophils were cultured in RPMI-1640 supplemented with 10% FBS. To screen the lung MC-derived candidate soluble factors, BM neutrophils were stimulated by the candidate regulators for 16 h and then the lipid levels of neutrophils were detected by BODIPY 493/503 staining and flow cytometry. The final concentration of each reagent: CD40LG and IGF2, 100 ng ml1; CSF2, HGF, TNFSF11, IL-6, IL-4, IL-33 and PGE2, 10 ng ml1; TGF-β1, 1 ng ml1; and H2O2, 5 μM.

To determine the EP receptor(s) used by neutrophils to respond to lung MCs, selective EP receptor antagonists were added to BM neutrophils and lung MC co-culture system for 16 h, then the lipid levels of neutrophils were detected by BODIPY 493/503 staining and flow cytometry. Antagonists for EP inhibitors: EP1, SC-51089; EP2, PF-04418948; EP3, L-798106; and EP4, MF-498 were each at a final concentration of 1 µg ml−1.

Isolation and primary culture of CD140a+ MCs

The lung, liver, mammary gland tissue and heart were cut into small pieces (<1 mm3) and the tissue pieces were dissociated with protease solution (1.5 mg ml−1 collagenase type I, 0.1 mg ml−1 elastase, 5 U ml−1 dispase II and 0.1 mg ml−1 DNase I in HBSS) for 1 h. Cells were then resuspended, passed through a 70-μm cell strainer and finally plated into culture dishes with MC culture medium (MesenCult Expansion kit, catalog no. 05513, STEMCELL Technologies). For isolating mouse bone MCs, the BM cavities were flushed at least three times to thoroughly deplete hematopoietic cells. Then the compact bones were dissected into fragments of 1–3 mm3 and digested with collagenase type II (1 mg ml−1). The released cells were discarded and the digested bone fragments were cultivated in the MC culture medium. After 5 d of expansion, CD45CD140a+ cells (from BALB/cJ mice) or GFP+ cells (from CD140a-GFP mice) were sorted by FACS. The first to third passages of the primary MCs were used in the next co-culture experiments.

Flow cytometry and cell sorting

Prepared single-cell suspensions from mouse tissues or ex vivo cultured cells were incubated with fluorescently labeled antibodies (1:200 dilution in 2% FCS in PBS) directed against cell surface markers. Cell surface labeling was performed on ice for 30 min. All antibodies were purchased from BioLegend (molecule, fluorophore): CD45, Alexa Fluor 700; CD11b, BV650; Ly6c, BV570; Ly6G, Pacific Blue; CD326, APC/Fire 750; CD31, FITC; CD140a, Brilliant Violet 421. Cells were run on FACSymphony A5 cytometer (BD Biosciences) and data were analyzed by FlowJo (Tristar). Further information about the above-mentioned antibodies, as well as other aspects of research design, is available in the Nature Research Reporting Summary linked to this article.

For cell sorting, a low-pressure sort was performed on the BD FACSAria II cell sorter with a 100-μm nozzle. Dead cells were stained with 4,6-diamidino-2-phenylindole (DAPI) or propidium iodide.

Measurement of lipid levels in neutrophils by flow cytometry

For the measurement of cellular lipid contents, after cell surface labeling, cells were washed and resuspended in 0.5 μg ml−1 BODIPY 493/503 for 15 min at room temperature in the dark before the analysis.

For directly determining whether the lung is a unique site for neutrophils to accumulate lipids, neutrophils isolated from the BM of naive mice were labeled with a cell dye CMTMR, then intravenously infused into naive mice or premetastatic orthotopic 4T1 tumor-bearing BALB/cJ mice. After 4 h, lipid levels of the CMTMR-labeled neutrophils from different tissues were determined by BODIPY 493/503 staining and flow cytometry.

In vivo BrdU incorporation

WT and Atgl-cKO mice were first orthotopically implanted with AT3 or AT3-g-csf tumor cells to generate the neutrophil-low (AT3 implantation) or neutrophil-high host condition (AT3-g-csf implantation). At the premetastatic stage for orthotopic tumors, AT3-mCherry cells were intravenously injected. After 6 d, the in vivo tumor cell proliferation was determined using an FITC-BrdU Flow kit (BD Biosciences, 559619). Briefly, the mice received a single dose of BrdU (1 mg per mouse) by intraperitoneal injection. At 4 h later, the mice were killed and the single-cell suspensions of lungs were prepared for flow cytometry analysis following the manufacturer’s instructions.

Tissue staining, immunofluorescence and light microscopy

For preparation of paraffin-embedded sections, dissected tissues were fixed in 10% formalin for 24 h at room temperature, washed twice with PBS and then stored in 70% ethanol solution at 4 °C until they were embedded in paraffin and sectioned at 4 μm. H&E staining was performed according to standard histopathological procedures.

For preparation of cryosections, formalin-fixed dissected tissues were embedded in Tissue-Tek O.C.T. (Electron Microscopy Sciences) and frozen on dry ice. The frozen tissues were stored at −80 °C until they were sectioned at 8 μm. The cryosections were incubated with mouse FcR Blocking Reagent (Miltenyi) followed by incubation with specific primary antibodies and fluorescently labeled secondary antibodies (Thermo Fisher Scientific). The cell nuclei were stained with DAPI.

To detect the lipid droplets under light microscopy, the purified neutrophils were first stained with BODIPY 493/503 at 1 µg ml−1 in PBS for 20 min. Then the cells were cytospun onto glass slides and mounted for imaging.

To determine the localization of early-seeded metastatic tumor cells and neutrophils in the lung, NSG mice were orthotopically implanted with unlabeled 4T1 tumor cells and at their premetastatic stage mice received an intravenous injection of mCherry-labeled 4T1 cells (4T1-mCherry). On day 2 and day 6 following 4T1-mCherry cell injection, the lungs were dissected and sectioned for immunostaining of neutrophils (Ly6G), the early-seeded 4T1-mCherry cells and the epithelium (RAGE).

Unless otherwise stated, more than six images per sample were acquired on a Leica SP8 inverted confocal microscope with a ×40 or ×63 objective and Leica Application Suite X software platform. All images were taken with identical laser set-up and voltage settings.

Ki67 staining

NOD-scid mice were first orthotopically implanted with 4T1 cells (5 × 105 per mouse). After 5 d, the mice received daily intraperitoneal injections of anti-Ly6G (clone 1A8; Bio X Cell, 12.5 µg per mouse) or rat IgG2a isotype control (clone 2A3; Bio X Cell, 12.5 µg per mouse) to develop a neutrophil-low or neutrophil-high host condition, respectively. After an additional 5 d, 4T1-mCherry cells (5 × 105 per mouse) were intravenously injected in the same mice and 6 d after these injections the mice were killed and 8-μm cryosections of lung tissues were prepared. The cryosections were then incubated with mouse FcR Blocking Reagent (Miltenyi Biotec) followed by incubation with Ki67 monoclonal antibody (clone SolA15, 0.5 µg ml−1) and Alexa Fluor 488-labeled secondary antibody (Invitrogen, 0.1 µg ml−1). The cell nuclei were stained with DAPI. Images were acquired on a Leica SP8 inverted confocal microscope. The percentages of Ki67+ cells in total mCherry+ 4T1 cells were counted.

Transmission electron microscopy

Neutrophils were freshly isolated from BM, PB and lungs of tumor-bearing BALB/cJ, Atgl-cKO or MMTV-PyMT mice at the premetastatic stage. 4T1 tumor cells were co-cultured with neutrophil-derived vesicles for 4 h in delipidated medium (DMEM containing 10% delipidized FBS (Gemini Bio)). The neutrophils or tumor cells were first fixed with 4% paraformaldehyde. The fixed cells were then embedded, sectioned and stained with osmium tetroxide and uranyl acetate. Images were captured by JEOL 1230 transmission electron microscope with an AMT 2K digital camera.

RNA-seq analysis

To compare the gene expression of neutrophils residing in BM, PB and lung, 4T1 cells (5 × 105 cells) were resuspended in 50 µl growth factor-reduced Matrigel (Corning) and injected into the fourth mammary fat pads of female syngeneic BALB/cJ mice (8 weeks old, n = 3). On day 10 (premetastatic stage), the mice were killed and then CD45+CD11b+Ly6GhighLy6Cmed neutrophils from BM, PB and lungs were sorted by FACS.

Total RNA was isolated from purified neutrophils using the miRNeasy Mini kit (QIAGEN), according to manufacturers’ protocols, including the optional DNase digest step. Sample concentration and quality were assessed using the Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific) and the Total RNA Nano assay (Agilent Technologies). Libraries were prepared using the KAPA mRNA HyperPrep kit (KAPA Biosystems), according to the manufacturer’s instructions. Briefly, the protocol entails isolation of polyA containing mRNA using oligo-dT magnetic beads, RNA fragmentation, first- and second-strand complementary DNA synthesis, ligation of Illumina-specific adaptors containing a unique barcode sequence for each library and PCR amplification. Libraries were checked for quality and concentration using the D5000 assay on the TapeStation (Agilent Technologies) and qPCR (KAPA Biosystems), according to the manufacturers’ instructions. Libraries were pooled and sequenced, 100-bp paired-end on the HiSeq 4000 (Illumina) using HiSeq 3000/4000 SBS kit reagents (Illumina).

Paired-end, Illumina-sequenced stranded RNA-seq reads were filtered and trimmed for quality scores >30 using a custom Python script. The filtered reads were aligned to Mus musculus GRCm38 using RSEM (v.1.2.12) to launch Bowtie2 (v.2.2.0) (command: rsem-calculate-expression -p 12–phred33-quals–seed-length 25–forward-prob 0–time–output-genome-bam–bowtie2–paired-end). RSEM calculates expected counts and transcripts per million. The expected counts from RSEM were used in the edgeR package (v.3.20.9) to determine differentially expressed genes. Core analysis was performed using ingenuity pathway analysis (QIAGEN). The analysis was restricted to differentially expressed genes with absolute value of log(fold change) >5.

RNA isolation and RT–qPCR

Total RNA was isolated directly from cells using the Direct-zol RNA Miniprep Plus kit (Zymo). Reverse transcription was performed with High-Capacity cDNA Reverse Transcription kit (Thermo Fisher Scientific). Gene expression levels were quantified by RT–qPCR on the ViiA 7 Real-Time PCR system (Thermo Fisher Scientific). Threshold cycle values (Ct) were normalized to Rps18 (ΔCt) and these values across samples were compared (ΔΔCt) to quantify relative gene expression. Primer sequences are listed in Supplementary Table 1.

Immunoblotting

Total protein was extracted by lysing neutrophils in RIPA buffer containing protease inhibitor. Protein samples were separated by SDS–PAGE and were then transferred onto polyvinylidene fluoride membranes. After blocking with 5% nonfat milk in TBS-T, membranes were incubated with the primary antibody. These primary antibodies were used: G0S2 polyclonal antibody (Thermo Fisher Scientific, catalog no. PA5-97764); CIDEC polyclonal antibody (Novus Biologicals, catalog no. NB100-430); HILPDA polyclonal antibody (Antibodies-online, catalog no. ABIN1030425); and GAPDH monoclonal antibody (D16H11; Cell Signaling Technology, catalog no. 5174). Goat-anti-rabbit IgG conjugated to horseradish peroxidase (Cell Signaling Technology) was used as the secondary antibody. Protein signal was detected with Image Acquisition using Azure c600 (Azure Biosystems).

Determination of TG hydrolase activity in neutrophils

Purified neutrophils were washed twice in ice-cold PBS and then disrupted on ice by passing 20 times through a 22-G needle in 0.2 ml of cell extraction buffer (0.25 M sucrose, 1 mM EDTA, 1 mM DTT, and protease inhibitors, pH 7.0). The cells were clarified by centrifugation at 1,000g for 15 min. The TG hydrolase activity against 3H-labeled triolein was measured as described previously14. Briefly, 25 μl of cell extracts were mixed with 25 μl of substrate solution (418 µM triolein and 48 µM of phosphatidylcholine/phosphatidylinositol in a ratio of 3:1 was sonicated in PBS). Before using, 0.5 µCi [9,10-3H]-triolein (PerkinElmer) were added to the TG substrate and vortexed to completely mix and incubated in a water bath at 37 °C for 60 min. The reaction was terminated by directly adding 650 µl of methanol/chloroform/heptane (10:9:7) and 200 µl 0.1 M potassium carbonate, 0.1 M boric acid, pH 10.5. After centrifugation (800g, 20 min) the radioactivity in 200 µl of the upper phase was determined by liquid scintillation counting.

Determination of lipid content and total lipase activity

The TG content of freshly isolated neutrophils was determined by a Triglyceride Quantification Colorimetric/Fluorometric kit (Biovision) and total lipase activity of freshly isolated neutrophils was measured by a Lipase Activity Assay kit (Cayman Chemical), following the manufacturer’s instructions.

Determination of PGE2 levels in MC culture supernatants

MCs were seeded at 1 × 105 ml−1 medium in the absence or presence of exogenous stimulations and incubated for 24 h. PGE2 concentrations in supernatants were measured by using the a PGE2 ELISA kit (Cayman Chemical), according to the manufacturer’s instructions. The net concentration of PGE2 was calculated upon subtracting the quantity of PGE2 from medium control.

Liquid chromatography–mass spectrometry

The freshly isolated neutrophils were immediately frozen in liquid nitrogen and stored at −80 °C until use. Untargeted lipidomics was performed by Cayman Chemical. In brief, lipids were extracted from cell pellets using the Bligh and Dyer method and lipidomics analysis was conducted following the protocol described by Breitkopf et al.51 with some modifications, using an ultra-performance liquid chromatography system (UltiMate 3000, Thermo Fisher Scientific) coupled with a hybrid Quadrupole-Orbitrap mass spectrometer (Q‐Exactive Plus, Thermo Fisher Scientific). Analysis of liquid chromatography–mass spectrometry data was performed using Lipostar software (v.1.0.5; Molecular Discovery). Lipid identification was provided by the software using the Lipid Maps database as reference. The peak area of each feature was normalized to peak area per 1 million cells. The lipid types measured included: TG, CE, PE, PC, PS, PI, Cer and PG.

Tumor cell lipid uptake assay

4T1-mCherry or AT3-mCherry tumor cells were seeded in a six-well plate in 10% FBS/DMEM medium. Cells were allowed to adhere and grow to 40% to 60% confluence for 24 h. Then the cells were starved for 2 h in delipidated medium. The freshly isolated neutrophils were added to the seeded tumor cells (the ratio of neutrophils to tumor cells was 20:1) in delipidated medium in a direct cell–cell contact manner for 16 h. For determination of the TG content in tumor cells, mCherry+ tumor cells were sorted and TG content was measured by a Triglyceride Quantification Colorimetric/Fluorometric kit (Biovision). For lipid content analysis by flow cytometry, all the cells were stained with BODIPY 493/503 and anti-CD45 and analyzed by flow cytometry. The mCherry+CD45 cells were gated as tumor cells. The relative MFI of BODIPY 493/503 was quantified and calculated as fold change in co-cultured cells compared to monoculture controls.

To directly observe lipid transfer between neutrophils and tumor cells, the neutrophils were pre-incubated with fluorescently labeled palmitic fatty acid (BODIPY FL C16, Thermo Fisher Scientific, 5 µM) at 37 °C in RPMI-1640 medium for 1 h. Then the extracellular BODIPY was removed from the cells by washing four times with 1× HBSS with 0.2% fatty acid-free BSA. After washing, labeled neutrophils were added to the dishes, where 4T1-mCherry cells had been seeded and these cells were co-cultured in a direct cell–cell contact manner for 4 h. The cells were then fixed and imaged on a microscopy platform. When the endocytic inhibitors (50 µM EIPA, 50 µM genistein and 15 µM chlorpromazine) were used, they were separately added 1 h before neutrophil addition.

In vivo lipid transport

To detect lipid transport from lung neutrophils to tumor cells in vivo, NOD-scid mice were used so that a sufficient number of tumor cells could be seeded in the lung to meet the requirements of the lipid content analysis by flow cytometry, in the experimental metastasis model.

In one experiment designed to compare the lipid levels in implanted tumor cells in the absence and presence of host neutrophils, NOD-scid mice received daily intraperitoneal injection of recombinant mouse G-CSF (PeproTech, 2.5 µg per mouse) without and with intraperitoneal injection of anti-Ly6G (clone 1A8; Bio X Cell, 12.5 µg per mouse) to generate the neutrophil-high and neutrophil-low host conditions, respectively. One week later, 4T1-mCherry or AT3-mCherry cells (5 × 105 cells per mouse) were intravenously injected into the mice. After 24 h, the lungs were dissected, stained with anti-CD45, DAPI and BODIPY 493/503 and analyzed by flow cytometry. The lipid levels of tumor cells (CD45mCherry+) seeded in the lungs were quantitated as MFI of BODIPY 493/503.

In another experiment designed to provide a direct in vivo evidence for the lipid transfer from neutrophils to tumor cells, lung neutrophils were first prepared from the premetastatic stage of 4T1 tumor-bearing mice, and then incubated with fluorescently labeled palmitic fatty acid (BODIPY FL C16, 5 μM) in RPMI-1640 medium at 37 °C for 1 h. Extracellular BODIPY was then removed by washing the cells five times with 1× HBSS containing 0.2% fatty acid-free BSA. NOD-scid mice were then intravenously injected with these BODIPY FL C16-loaded lung neutrophils or unlabeled neutrophils (5 × 106 cells per mouse). At 2 h later, the mice received a second intravenous injection of mCherry-labeled tumor cells (4T1-mCherry, AT3-mCherry or MDA-4175-mCherry at 5 × 105 cells per mouse). After another 4 h, mouse lungs were dissected for measurement of the BODIPY FL C16 levels in tumor cells (CD45mCherry+) by flow cytometry.

Cell proliferation assay

Luciferase-labeled cells (4T1-Luc, AT3-Luc, MCF7-Luc or MDA-4175-Luc) were seeded in a 24-well plate in 10% FBS/DMEM medium. Cells were allowed to adhere overnight and were monocultured or co-cultured with freshly isolated neutrophils (neutrophil to tumor cell ratio was 20:1) in DMEM supplemented with 10% FBS in a direct cell–cell contact manner for 24 h. The cell culture medium was then changed to delipidated medium and luciferase activity was measured at different time points using a SpectraMax i3 plate reader (Molecular Devices). For drug treatment experiments, 4T1-Luc cells were seeded and co-cultured with neutrophils as mentioned above. The cell culture medium was then changed to fresh delipidated medium containing either DMSO, cerulenin (Cayman Chemical) or C75 (Cayman Chemical) at the indicated doses. The final concentration of all wells contained equivalent amounts of DMSO solvent (0.2%). After a 48-h treatment, cells were collected and luciferase activity was measured using a SpectraMax i3 plate reader. All values were normalized to the DMSO control sample and performed in triplicate for all cell lines.

Extracellular flux analysis of tumor cells

Fatty acid oxidation of tumor cells precultured with neutrophils was determined through real-time measurement of the OCR using a Seahorse XFe96 Analyzer (Agilent Technologies), according to the manufacturer’s instructions for the XF Fatty Acid Oxidation assay. Briefly, FACS-isolated tumor cells were seeded into XFp 96-well microplate wells at a density of 30,000 cells per well and spun for 5 min at 1,400 r.p.m. to settle cells. Cells were maintained in XF assay medium (XF DMEM medium, Agilent Technologies, supplemented with 1 mM pyruvate, 2 mM glutamine and 10 mM glucose) in a CO2-free 37 °C incubator for 30 min to 1 h. At 15 min before plate loading, etomoxir (or vehicle) was added to cells at a final concentration of 20 µM.

Neutrophil vesicle isolation and labeling

Neutrophils were incubated with fluorescently labeled palmitic fatty acid (BODIPY FL C16, 5 µM) at 37 °C in RPMI-1640 medium for 1 h. Then the extracellular BODIPY was removed from the cells by washing four times with 1× HBSS with 0.2% fatty acid-free BSA. Labeled neutrophils were incubated in RPMI-1640 medium without serum at 37 °C at a density of 1 million cells ml−1. After 16 h, CM was collected and centrifuged at 1,200g for 15 min to remove cell debris and apoptotic bodies. The resulting supernatant was then passed through a 0.8-μm syringe filter (Sigma) and then concentrated via 100-kDa centrifuge filters (Millipore Amicon Ultra 15-ml centrifuge filters). The filtrated medium (<100 kDa) and concentrated vesicle-enriched medium (>100 kDa, washed twice with RPMI-1640) were used to feed tumor cells at a concentration of 1 million cell equivalents ml−1. Tumor cells were fed for 4 h before imaging. Lysotracker (LysoTracker Red DND-99, Thermo Fisher Scientific) was added for the last 30 min of incubation to label lysosomes.

Measurement of the metastasis-initiating potentials of neutrophil-educated tumor cells

Luciferase-labeled tumor cells (AT3-Luc, MCF7-Luc, MDA-4175-Luc or 4T1-Luc) were monocultured or co-cultured with different neutrophils for 24 h in delipidated medium, then the tumor cells were purified by depleting neutrophils using anti-Ly6G microbeads. The purified tumor cells (5 × 104 cells per injection) were then intravenously injected into NSG mice. At 6 d later, the mice were anesthetized under 2.5% isoflurane and administered with d-luciferin (150 mg kg−1) by intraperitoneal injection. The bioluminescence signals were detected with a Xenogen IVIS system. At 4 h later, mice were killed and the lungs were rapidly excised for ex vivo BLI analysis as described above. Photon emissions from each lung tissue in ex vivo BLI analysis were quantified and shown.

Correlation analysis of LD-associated genes and human neutrophil signature genes using published microarray data

To evaluate whether the lipid gene expression is correlated with neutrophil signature genes in human lung metastasis samples, we analyzed a published microarray dataset (GSE14018 (ref. 20)). The LD signature genes (HILPDA, CIDEC and G0S2) as revealed in this study and the reported human neutrophil signature genes (CREB5, CDA, CHST15, S100A12, APOBEC3A,CASP5, MMP25, HAL, C1orf183, FFAR2, MAK, CXCR1, STEAP4, MGAM, BTNL8, CXCR2, TNFRSF10C and VNN3)21 were used for correlation analysis. For each signature, we averaged the expression level of the genes in each of the 16 individual samples. Then the Pearson correlation coefficient between LD signature and neutrophil signature gene expressions was calculated.

Survival analysis of genes related with lipid catabolism or neutrophil trafficking

To determine whether neutrophil gene signature and lipid catabolism gene signature are associated with lung metastasis-free survival, we utilized a microarray dataset profiled by Affymetrix Human Genome U133A Array from 99 patients with primary breast cancer (GEO accession number: GSE2603)45. After excluding those with incomplete clinical annotations, it resulted in 82 analyzable samples. First, a univariate Cox regression analysis was performed to evaluate the association between expression levels of genes and samples’ lung metastasis-free survival. Then, we calculated a risk score for each sample, which was defined as a linear combination of expression values of genes in one signature set weighted by their estimated Cox model regression coefficients. If the risk score for one sample was in the top 20th percentile of the risk scores, then it was classified into a high-risk group, otherwise into a low-risk group. The top 20th percentile was used as a threshold because approximately 20% of the cases were expected to eventually develop lung metastasis45. Finally, Kaplan–Meier survival curves and a log-rank test were applied to evaluate the differences in sample lung metastasis-free survival between the high-risk and low-risk groups.

Quantification and statistical analysis

Data are presented as mean ± s.e.m. or mean ± s.d. as indicated. P values were calculated using Prism 7.04 statistical software (GraphPad Software). The details of the statistical tests carried out are indicated in the respective figure legends and statistical tests were two-tailed unless otherwise indicated. Values were compared using an unpaired two-tailed Student’s t-test to compare two groups or ordinary one-way ANOVA with Tukey’s multiple comparisons test to compare the variance in three or more groups with one independent factor (for example treatment group). When there were effects of two factors (for example treatment and time) on a dependent variable, two-way ANOVA with Šidák’s multiple comparisons test was employed. For Kaplan–Meier analyses of patients with breast cancer (GSE2603), statistical differences in survival curves were calculated by log-rank (Mantel–Cox) test. Correlation analysis of gene expression values of each signature gene group within each sample was performed by linear regression with Pearson’s correlation coefficient (r). Unless indicated in the figure legends, three independent experiments were performed and similar results were obtained. A P value <0.05 was considered significant, whereas >0.05 was assigned as NS. Exact P values are provided in each figure.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

RNA-seq data are deposited in ArrayExpress and are available under accession code E-MTAB-9128. For lung metastasis-free survival analysis of patients with breast cancer, the published microarray dataset was used (GSE2603). For correlation analysis of signature genes in breast cancer lung metastasis samples, the published microarray dataset (GSE14018) was used. All other data supporting the findings of this study are available within the article and its supplementary information files and on reasonable request from the corresponding author. A Nature Research Reporting Summary for this article is available as a supplementary information file. Source data are provided with this paper.

References

  1. 1.

    Liu, Y. & Cao, X. Characteristics and significance of the pre-metastatic niche. Cancer Cell 30, 668–681 (2016).

    CAS  PubMed  Google Scholar 

  2. 2.

    Peinado, H. et al. Pre-metastatic niches: organ-specific homes for metastases. Nat. Rev. Cancer 17, 302–317 (2017).

    CAS  PubMed  Google Scholar 

  3. 3.

    Wculek, S. K. & Malanchi, I. Neutrophils support lung colonization of metastasis-initiating breast cancer cells. Nature 528, 413–417 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  4. 4.

    Coffelt, S. B. et al. IL-17-producing γδ T cells and neutrophils conspire to promote breast cancer metastasis. Nature 522, 345–348 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  5. 5.

    Albrengues, J. et al. Neutrophil extracellular traps produced during inflammation awaken dormant cancer cells in mice. Science 361, eaao4227 (2018).

    PubMed  PubMed Central  Google Scholar 

  6. 6.

    Spiegel, A. et al. Neutrophils suppress intraluminal NK cell–mediated tumor cell clearance and enhance extravasation of disseminated carcinoma cells. Cancer Discov. 6, 630–649 (2016).

    CAS  PubMed  PubMed Central  Google Scholar 

  7. 7.

    Massagué, J. & Obenauf, A. C. Metastatic colonization by circulating tumour cells. Nature 529, 298–306 (2016).

    PubMed  PubMed Central  Google Scholar 

  8. 8.

    Dupuy, F. et al. PDK1-dependent metabolic reprogramming dictates metastatic potential in breast cancer. Cell Metab. 22, 577–589 (2015).

    CAS  PubMed  Google Scholar 

  9. 9.

    Christen, S. et al. Breast cancer-derived lung metastases show increased pyruvate carboxylase-dependent anaplerosis. Cell Rep. 17, 837–848 (2016).

    CAS  PubMed  Google Scholar 

  10. 10.

    Loo, J. M. et al. Extracellular metabolic energetics can promote cancer progression. Cell 160, 393–406 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  11. 11.

    Nieman, K. M. et al. Adipocytes promote ovarian cancer metastasis and provide energy for rapid tumor growth. Nat. Med. 17, 1498–1503 (2011).

    CAS  PubMed  PubMed Central  Google Scholar 

  12. 12.

    Messmer, M. N., Netherby, C. S., Banik, D. & Abrams, S. I. Tumor-induced myeloid dysfunction and its implications for cancer immunotherapy. Cancer Immunol. 64, 1–13 (2015).

    CAS  Google Scholar 

  13. 13.

    Heckmann, B. L., Zhang, X., Xie, X. & Liu, J. The G0/G1 switch gene 2 (G0S2): regulating metabolism and beyond. Biochim. Biophys. Acta 1831, 276–281 (2013).

    CAS  PubMed  Google Scholar 

  14. 14.

    Padmanabha Das, K. M. et al. Hypoxia-inducible lipid droplet-associated protein inhibits adipose triglyceride lipase. J. Lipid Res. 59, 531–541 (2018).

    CAS  PubMed  PubMed Central  Google Scholar 

  15. 15.

    Munir, R., Lisec, J., Swinnen, J. V. & Zaidi, N. Lipid metabolism in cancer cells under metabolic stress. Br. J. Cancer 120, 1090–1098 (2019).

    PubMed  PubMed Central  Google Scholar 

  16. 16.

    Vallochi, A. L., Teixeira, L., Oliveira, K. D. S., Maya-Monteiro, C. M. & Bozza, P. T. Lipid droplet, a key player in host–parasite interactions. Front. Immunol. 9, 1022 (2018).

    PubMed  PubMed Central  Google Scholar 

  17. 17.

    Chistiakov, D. A., Melnichenko, A. A., Myasoedova, V. A., Grechko, A. V. & Orekhov, A. N. Mechanisms of foam cell formation in atherosclerosis. J. Mol. Med. 95, 1153–1165 (2017).

    CAS  PubMed  Google Scholar 

  18. 18.

    Missaglia, S., Coleman, R. A., Mordente, A. & Tavian, D. Neutral lipid storage diseases as cellular model to study lipid droplet function. Cells 8, 187 (2019).

    CAS  PubMed Central  Google Scholar 

  19. 19.

    Nielsen, T. S., Jessen, N., Jørgensen, J. O., Møller, N. & Lund, S. Dissecting adipose tissue lipolysis: molecular regulation and implications for metabolic disease. J. Mol. Endocrinol. 52, R199–R222 (2014).

    CAS  PubMed  Google Scholar 

  20. 20.

    Zhang, X. H. et al. Latent bone metastasis in breast cancer tied to Src-dependent survival signals. Cancer Cell 16, 67–78 (2009).

    CAS  PubMed  PubMed Central  Google Scholar 

  21. 21.

    Charoentong, P. et al. Pan-cancer immunogenomic analyses reveal genotype-immunophenotype relationships and predictors of response to checkpoint blockade. Cell Rep. 18, 248–262 (2017).

    CAS  PubMed  Google Scholar 

  22. 22.

    Nakanishi, M. & Rosenberg, D. W. Multifaceted roles of PGE2 in inflammation and cancer. Semin. Immunopathol. 35, 123–137 (2013).

    CAS  PubMed  Google Scholar 

  23. 23.

    Sugimoto, Y. & Narumiya, S. Prostaglandin E receptors. J. Biol. Chem. 282, 11613–11617 (2007).

    CAS  PubMed  Google Scholar 

  24. 24.

    Dennis, E. A. & Norris, P. C. Eicosanoid storm in infection and inflammation. Nat. Rev. Immunol. 15, 511–523 (2015).

    CAS  PubMed  PubMed Central  Google Scholar 

  25. 25.

    Wang, D. & Dubois, R. N. Eicosanoids and cancer. Nat. Rev. Cancer 10, 181–193 (2010).

    CAS  PubMed  PubMed Central  Google Scholar 

  26. 26.

    Zhang, X. et al. Inhibition of intracellular lipolysis promotes human cancer cell adaptation to hypoxia. Elife 6, e31132 (2017).

    PubMed  PubMed Central  Google Scholar 

  27. 27.

    Knight, M., Braverman, J., Asfaha, K., Gronert, K. & Stanley, S. Lipid droplet formation in Mycobacterium tuberculosis infected macrophages requires IFN-γ/HIF-1α signaling and supports host defense. PLoS Pathog. 14, e1006874 (2018).

    PubMed  PubMed Central  Google Scholar 

  28. 28.

    Palazon, A., Goldrath, A. W., Nizet, V. & Johnson, R. S. HIF transcription factors, inflammation, and immunity. Immunity 41, 518–528 (2014).

    CAS  PubMed  PubMed Central  Google Scholar 

  29. 29.

    Passegue, E., Wagner, E. F. & Weissman, I. L. JunB deficiency leads to a myeloproliferative disorder arising from hematopoietic stem cells. Cell 119, 431–443 (2004).

    CAS  PubMed  Google Scholar 

  30. 30.

    den Brok, M. H., Raaijmakers, T. K., Collado-Camps, E. & Adema, G. J. Lipid droplets as immune modulators in myeloid cells. Trends Immunol. 39, 380–392 (2018).

    Google Scholar 

  31. 31.

    Veglia, F. et al. Fatty acid transport protein 2 reprograms neutrophils in cancer. Nature 569, 73–78 (2019).

    CAS  PubMed  PubMed Central  Google Scholar 

  32. 32.

    Stewart, T. J. & Abrams, S. I. Altered immune function during long-term host-tumor interactions can be modulated to retard autochthonous neoplastic growth. J. Immunol. 179, 2851–2859 (2007).

    CAS  PubMed  Google Scholar 

  33. 33.

    DeBerardinis, R. J., Lum, J. J., Hatzivassiliou, G. & Thompson, C. B. The biology of cancer: metabolic reprogramming fuels cell growth and proliferation. Cell Metab. 7, 11–20 (2008).

    CAS  Google Scholar 

  34. 34.

    Pascual, G. et al. Targeting metastasis-initiating cells through the fatty acid receptor CD36. Nature 541, 41–45 (2017).

    CAS  Google Scholar 

  35. 35.

    Zhang, M. et al. Adipocyte-derived lipids mediate melanoma progression via FATP proteins. Cancer Discov. 8, 1006–1025 (2018).

    CAS  PubMed  PubMed Central  Google Scholar 

  36. 36.

    Wang, Y. Y. et al. Mammary adipocytes stimulate breast cancer invasion through metabolic remodeling of tumor cells. JCI insight 2, e87489 (2017).

    PubMed  PubMed Central  Google Scholar 

  37. 37.

    Adrover, J. M., Nicolas-Avila, J. A. & Hidalgo, A. Aging: a temporal dimension for neutrophils. Trends Immunol. 37, 334–345 (2016).

    CAS  PubMed  Google Scholar 

  38. 38.

    Schild, T., Low, V., Blenis, J. & Gomes, A. P. Unique metabolic adaptations dictate distal organ-specific metastatic colonization. Cancer Cell 33, 347–354 (2018).

    CAS  PubMed  PubMed Central  Google Scholar 

  39. 39.

    Flaherty, S. E. III, Grijalva, A., Xu, X., Ables, E., Nomani, A. & Ferrante, A. W. Jr. A lipase-independent pathway of lipid release and immune modulation by adipocytes. Science 363, 989–993 (2019).

    CAS  PubMed  PubMed Central  Google Scholar 

  40. 40.

    Lim, J. P. & Gleeson, P. A. Macropinocytosis: an endocytic pathway for internalising large gulps. Immunol. Cell Biol. 89, 836–843 (2011).

    CAS  PubMed  Google Scholar 

  41. 41.

    Commisso, C. et al. Macropinocytosis of protein is an amino acid supply route in Ras-transformed cells. Nature 497, 633–637 (2013).

    CAS  PubMed  PubMed Central  Google Scholar 

  42. 42.

    Kim, S. M. et al. PTEN deficiency and AMPK activation promote nutrient scavenging and anabolism in prostate cancer cells. Cancer Discov. 8, 866–883 (2018).

    CAS  PubMed  PubMed Central  Google Scholar 

  43. 43.

    Rejman, J., Bragonzi, A. & Conese, M. Role of clathrin- and caveolae-mediated endocytosis in gene transfer mediated by lipo- and polyplexes. Mol. Ther. 12, 468–474 (2005).

    CAS  PubMed  Google Scholar 

  44. 44.

    Shaul, M. E. & Fridlender, Z. G. Tumour-associated neutrophils in patients with cancer. Nat. Rev. Clin. Oncol. 16, 601–620 (2019).

    PubMed  Google Scholar 

  45. 45.

    Minn, A. J. et al. Genes that mediate breast cancer metastasis to lung. Nature 436, 518–524 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  46. 46.

    Spiegelman, B. M. & Flier, J. S. Obesity and the regulation of energy balance. Cell 104, 531–543 (2001).

    CAS  PubMed  Google Scholar 

  47. 47.

    Lehuede, C., Dupuy, F., Rabinovitch, R., Jones, R. G. & Siegel, P. M. Metabolic plasticity as a determinant of tumor growth and metastasis. Cancer Res. 76, 5201–5208 (2016).

    PubMed  Google Scholar 

  48. 48.

    Zhang, Y. & Commisso, C. Macropinocytosis in cancer: a complex signaling network. Trends Cancer 5, 332–334 (2019).

    PubMed  PubMed Central  Google Scholar 

  49. 49.

    Ohs, I. et al. Restoration of natural killer cell antimetastatic activity by IL12 and checkpoint blockade. Cancer Res. 77, 7059–7071 (2017).

    CAS  PubMed  Google Scholar 

  50. 50.

    Basnet, H. Flura-seq identifies organ-specific metabolic adaptations during early metastatic colonization. Elife 8, e43627 (2019).

    PubMed  PubMed Central  Google Scholar 

  51. 51.

    Breitkopf, S. B. et al. A relative quantitative positive/negative ion switching method for untargeted lipidomics via high resolution LC-MS/MS from any biological source. Metabolomics 13, 30 (2017).

    PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

We thank S.I. Abrams (Roswell Park Comprehensive Cancer Center) for providing the AT3 breast cancer cell line, Y. Kang (Princeton University) for providing the MDA-4175 breast cancer cell line and R.A. Weinberg (Whitehead Institute for Biomedical Research) for providing the lentiviralvector expressing mouse Csf3. This work was supported by grants from the National Institutes of Health (R00-CA188093 to G.R.; R37-CA237307 to G.R.; P30-CA034196 to G.R., L.D.S. and S.L.; R35−GM133562 to S.L.; and R24-OD026440 and R01-AI132963 to L.D.S.), the U.S. Department of Defense (W81XWH-18-1-0013 to G.R.), Leukemia Research Foundation New Investigator Grant (to S.L.) and The Jackson Laboratory Director’s Innovation Fund (19000-17-31 to S.L.). Q.L. is supported by the Pyewacket Fund at The Jackson Laboratory. We acknowledge critical comments from E. T. Liu, N. A. Rosenthal, K. Seburn and C. Robinett. We also thank G. Stafford for RNA-seq analysis and W. Schott for cell sorting, as well as The Jackson Laboratory Scientific Service for assistance.

Author information

Affiliations

Authors

Contributions

G.R., P.L. and M.L. conceived the project, designed the study and performed the data analysis. P.L., M.L., J.S., Z.G., L.H. and Q.L. performed the experiments. L.D.S. provided critical assistance on experimental design. X.H.-F.Z. provided essential experimental materials. B.L. and X.H.-F.Z. provided assistance on clinical dataset analysis and experimental design; X.C. and S.L. performed the clinical dataset analysis and provided assistance on statistical analysis. G.R., P.L., M.L. and L.D.S. interpreted the data and wrote the manuscript.

Corresponding author

Correspondence to Guangwen Ren.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Peer reviewer reports are available. Zoltan Fehervari was the primary editor on this article and managed its editorial process and peer review in collaboration with the rest of the editorial team.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Lung neutrophils acquire a lipid-laden phenotype in the orthotopic 4T1 tumor model.

a–d, Transcriptional profiles of BM, PB and lung neutrophils sorted from the orthotopic 4T1 tumor-bearing BALB/cJ mice (n = 3). Volcano plots showing fold-change and p-value for the comparison of lung versus BM neutrophils (a), and PB versus BM neutrophils (b) based on the RNA-seq data. P values were determined by unpaired two-tailed t-test and smaller than 0.05 were considered significant. A schematic diagram showing the major pathway and key factors in lipid metabolism (c). A heat map depicts expression levels of the major lipid metabolic genes (d). e, Upper: a diagram showing the pre-metastatic and metastatic stages in the orthotopic 4T1 tumor model; lower: Representative immunostaining of neutrophils (Ly6G) in lung sections of naïve and 4T1 tumor-bearing BALB/cJ mice (n = 4 from 2 independent experiments). Scale bars, 20 μm. f, Intracellular lipids in neutrophils detected by BODIPY 493/503 staining under microscope (n = 4 mice from 2 independent experiments). Scale bars, 5 μm. g, The flow cytometry gating strategy is shown: neutrophils were identified as the Ly6G+Ly6Clow cell population which was gated on CD45+CD11b+ cells. h, Measurement of total lipid contents in neutrophils isolated from naïve and 4T1 tumor-bearing mice by BODIPY 493/503 staining and flow cytometry (n = 6). i, Cellular TG contents in neutrophils (n = 5). In f and i, neutrophils were isolated from the indicated tissues and organs of orthotopic 4T1 tumor-bearing mice at the pre-metastatic stage. n represents biologically independent animals. Data are mean ± s.e.m. and P values were determined by one-way ANOVA with Tukey’s multiple comparisons test (h and i). ns, not significant. Source data

Extended Data Fig. 2 Up-regulation of genes encoding ATGL inhibitory factors led to reduced TG hydrolysis in pre-metastatic lung-infiltrating neutrophils.

a–h, Levels of total triglyceride (TG) (a), cholesteryl ester (CE) (b), phosphatidylethanolamine (PE) (c), phosphatidylcholine (PC) (d), phosphatidylserine (PS) (e), phosphatidylinositol (PI) (f), ceramides (Cer) (g) and phosphatidylglycerol (PG) (h) in PB and lung neutrophils isolated from the pre-metastatic stage of orthotopic 4T1 tumor-bearing mice (n = 5), were determined by liquid chromatography-mass spectrometry. Relative levels of these lipid were also shown in Fig. 1f. The exact values of specific chemical species for each lipid type were provided in Source Data file. i-l, Neutrophils were isolated from the BM, PB and lung tissues of orthotopic 4T1 tumor-bearing mice (n = 5) at the pre-metastatic stage and compared for relative mRNA expression (to Rps18) of Hilpda, Cidec, G0s2, Atgl and Abhd5 by qRT-PCR (i); HILPDA, CIDEC and G0S2 protein expression by Western blotting with GAPDH as a loading control (j); relative TG hydrolase activity (k); and cellular lipase activity (l). n represents biologically independent animals. Data are mean ± s.e.m. and P values were determined by one-way ANOVA with Tukey’s multiple comparisons test (i and l) or unpaired two-tailed t-test (k). ns, not significant. Source data

Extended Data Fig. 3 Neutrophil-specific genetic ablation of Hilpda or Cidec did not lead to significant changes on lung metastases of breast cancer in vivo.

af, The neutrophil-targeting cKO mice (a-c, Hilpda-cKO; df, Cidec-cKO), and their wild type littermates were orthotopically implanted with AT3-g-csf cells. On day 15, the relative lipid levels in lung neutrophils were determined by BODIPY 493/503 staining and flow cytometry (a and d) (n = 5 per group), and the resected primary tumors were weighed (b and e). At the end point (day 30), spontaneous lung metastases of WT and cKO mice were quantified (c and f). n = 13 (WT group) and 14 (Hilpda-cKO group) for b, c; n = 15 (WT group) and 14 (Cidec-cKO group) for e, n = 10 (WT group) and 12 (Cidec-cKO group) for f. g, A schematic diagram showing the modified experimental lung metastasis model employed in this study. Mice were first orthotopically implanted with non-labeled AT3-g-csf cells to induce a neutrophil-high host condition. Luciferase-labeled AT3-Luc cells were then implanted intravenously on day 10, a time point within the pre-metastatic stage. On day 15, the primary tumors were resected. At the end point, the lung metastatic progression of AT3-Luc cells was detected by ex vivo BLI. h, Following the above modified experimental lung metastasis model, the lung metastatic progression of AT3-Luc cells in WT and Cidec-cKO recipient mice was determined by ex vivo BLI. Representative BLI images of the harvested lungs are shown and red lines indicate blank wells without lung tissues (left) (n = 11). n represents biologically independent animals. Data are mean ± s.e.m. and P values were determined by unpaired two-tailed t-test. ns, not significant. Source data

Extended Data Fig. 4 Tumor cells absorb lipids from lipid-laden lung neutrophils.

a, Representative images from 2 biologically independent experiments showing lipid transfer from BODIPY FL C16-loaded neutrophils to 4T1-mCherry cells. Arrowheads indicate neutrophils. Scale bars, 20 μm. b, 4T1-mCherry or AT3-mCherry cells were mono-cultured or co-cultured with their respective tumor-bearing mice-derived neutrophils that were pre-loaded with BODIPY FL C16. Then the intensity of BODIPY in tumor cells was examined by flow cytometry. Representative of 3 biologically independent experiments. c, The cellular TG contents of AT3-mCherry cells after monoculture or co-culture with neutrophils. Data are mean ± s.d. from 3 biologically independent cell cultures. d, Measurement of the lipids in early lung-colonizing AT3-mCherry cells in G-CSF-pretreated mice without and with anti-Ly6G-based neutrophil depletion (see Methods) (n = 5). Data are mean ± s.e.m. e, The lipid transfer from BODIPY FL C16-loaded lung neutrophils to the indicated tumor cells in vivo as determined by flow cytometry (see Methods) (n = 4). Representative flow cytometry histograms are shown. f, As depicted in the left panel, AT3-mCherry or 4T1-mCherry cells were mono-cultured or co-cultured with PB or lung neutrophils. In one experimental group, lung neutrophils were pre-activated by phorbol 12-myristate 13-acetate (PMA) and hydrogen peroxide (H2O2). The total lipid levels in tumor cells were determined by BODIPY 493/503 staining and flow cytometry. Data are mean ± s.d. from 5 (4T1-mCherry) or 3 (AT3-mCherry) biologically independent cell cultures. Neutrophils used throughout this figure were isolated from the pre-metastatic stage of 4T1 tumor-bearing mice or AT3-g-csf tumor-bearing mice. n represents biologically independent animals. P values were determined by one-way ANOVA with Tukey’s multiple comparisons test. ns, not significant. Source data

Extended Data Fig. 5 Neutrophil-derived lipids enhance the proliferative capacity of metastatic tumor cells.

a, Oxygen consumption rates (OCR) of AT3 cells, upon mono-culture or co-culture with PB or lung neutrophils (left). The amount of OCR derived from fatty acid oxidation was quantified as the magnitude of the response to etomoxir (right). b–d, Measurement of the proliferative capacities of the indicated tumor cells without and with co-culture with PB or lung neutrophils (see Methods). e–g, Comparison of the metastatic colonization potentials of the indicated tumor cells (e, AT3-Luc; f, MCF7-Luc; and g, MDA-4175-Luc) without and with co-culture with PB or lung neutrophils (see Methods). Representative BLI images of the recipient mice and quantification of BLI signals within the lung areas are shown (n = 5 for e and n = 6 for f, g). h, i, PB and lung neutrophils were isolated from AT3-g-csf tumor-bearing WT or Atgl-cKO mice (pre-metastatic stage). The intracellular lipids in neutrophils were detected by BODIPY 493/503 staining under microscope (representative of n = 5 mice). Scale bars, 5 μm (h). The cellular TG contents of neutrophils were measured (n = 5) (i). j, 4T1-Luc cells, upon co-culture with PB or lung neutrophils isolated from AT3-g-csf tumor-bearing WT or Atgl-cKO mice (pre-metastatic stage), were intravenously injected into NSG mice (see Methods). Representative BLI images of the recipient mice and quantification of BLI signals within the lung areas are shown (n = 6). km, As depicted in k, the effects of EIPA treatment on lung colonization by 4T1 tumor cells (l) and on primary tumor growth (m) were determined (see Methods) (n = 10). Neutrophils throughout this figure were isolated from the pre-metastatic stage of 4T1 tumor-bearing mice (c, d, f, g) or AT3-g-csf tumor-bearing mice (a, b and e), except otherwise stated. n represents biologically independent animals. Data are mean ± s.d. from 6 (a) or 4 (b–d) biologically independent cell cultures, and mean ± s.e.m. for e-g, i, j, l, m. P values were determined by two-way ANOVA with Sidak’s multiple comparisons test (b–d, mono-culture versus lung Neu co-culture), one-way ANOVA with Tukey’s multiple comparisons test (a, e–g, i and j) or unpaired two-tailed t-test (l, m). ns, not significant. Source data

Extended Data Fig. 6 Both lipid- and neutrophil-associated gene expression signatures are related with lung metastasis in human breast cancer patients.

Kaplan–Meier plots of lung metastasis-free survival of breast cancer patients, stratified by expression of indicated gene signature sets in their primary tumors (GEO accession number: GSE2603, n = 82 patients). A risk score was calculated for each sample which was defined as a linear combination of expression values of genes in one signature set weighted by their estimated Cox model regression coefficients. If the risk score for one sample was in the top 20th percentile of the risk scores, then it was classified into high-risk group, otherwise into low-risk group. P values were calculated by a 2-sided log-rank test. The used gene signature sets are derived from Gene Ontology (GO): Biological Process of MsigDB v.7.1, including GO_POSITIVE_REGULATION_OF_LIPID_CATABOLIC_PROCESS (M14107, 25 genes); GO_POSITIVE_REGULATION_OF_LIPID_TRANSPORT (M11731, 61 genes); GO_LIPID_OXIDATION (M15880, 101 genes); GO_NEUTROPHIL_MIGRATION (M25402, 119 genes); GO_REGULATION_OF_NEUTROPHIL_CHEMOTAXIS (M29283, 30 genes); and GO_NEUTROPHIL_EXTRAVASATION (M24616, 13 genes).

Extended Data Fig. 7 A lung MC → neutrophil → tumor cell metabolic axis in the lung metastatic niche.

A schematic diagram depicts how lung neutrophils are stimulated by lung resident MCs to accumulate lipids, and in turn transport their stored lipids to metastatic tumor cells for survival and proliferation leading to accelerated lung metastasis in a breast cancer model.

Supplementary information

Source data

Source Data Fig. 1

Statistical source data

Source Data Fig. 1

Uncropped immunoblots

Source Data Fig. 2

Statistical source data

Source Data Fig. 3

Statistical source data

Source Data Fig. 4

Statistical source data

Source Data Fig. 5

Statistical source data

Source Data Fig. 6

Statistical source data

Source Data Fig. 7

Statistical source data

Source Data Extended Data Fig. 1

Statistical source data

Source Data Extended Data Fig. 2

Statistical source data

Source Data Extended Data Fig. 2

Uncropped immunoblots

Source Data Extended Data Fig. 3

Statistical source data

Source Data Extended Data Fig. 4

Statistical source data

Source Data Extended Data Fig. 5

Statistical source data

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Li, P., Lu, M., Shi, J. et al. Lung mesenchymal cells elicit lipid storage in neutrophils that fuel breast cancer lung metastasis. Nat Immunol 21, 1444–1455 (2020). https://doi.org/10.1038/s41590-020-0783-5

Download citation

Search

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing