Coordinating effect of VEGFC and Oleic Acid drives tumor lymphangiogenesis

In cancer, the lymphatic system is hijacked by tumor cells to escape from primary tumor and to metastasize to the sentinel lymph nodes. Tumor lymphangiogenesis is stimulated by the vascular endothelial growth factors-C (VEGFC) after binding to its receptor VEGFR-3. However, how VEGFC cooperates with other molecules to promote lymphatic neovessels growth is not fully determined. Here, we found that tumor lymphangiogenesis developed in tumoral lesions and in their surrounding adipose tissue (AT). Interestingly, lymphatic vessel density correlated with an increase of circulating free fatty acids (FFA) in the lymph from tumor-bearing mice. We found that adipocyte-released FFA are uploaded by lymphatic endothelial cells (LEC) to stimulate their sprouting. Lipidomic analysis identified the monounsaturated oleic acid (OA) as the major circulating FFA in the lymph in tumoral context. OA transporters FATP-3, -6 and CD36 were only upregulated on LEC in the presence of VEGFC showing a collaborative effect of these molecules. OA released from adipocytes is taken up by LECs to stimulate the fatty acid β-oxidation, leading to increase adipose tissue lymphangiogenesis. Our results provide new insights on the dialogue between tumors and adipocytes via the lymphatic system and identify a key role for adipocyte-derived FFA in the promotion of lymphangiogenesis, revealing novel therapeutic opportunities for inhibitors of lymphangiogenesis in cancer.


SUMMARY
In cancer, the lymphatic system is hijacked by tumor cells to escape from primary tumor and to metastasize to the sentinel lymph nodes. Tumor lymphangiogenesis is stimulated by the vascular endothelial growth factors-C (VEGFC) after binding to its receptor VEGFR-3. However, how VEGFC cooperates with other molecules to promote lymphatic neovessels growth is not fully determined. Here, we found that tumor lymphangiogenesis developed in tumoral lesions and in their surrounding adipose tissue (AT). Interestingly, lymphatic vessel density correlated with an increase of circulating free fatty acids (FFA) in the lymph from tumor-bearing mice. We found that adipocyte-released FFA are uploaded by lymphatic endothelial cells (LEC) to stimulate their sprouting. Lipidomic analysis identified the monounsaturated oleic acid (OA) as the major circulating FFA in the lymph in tumoral context. OA transporters FATP-3, -6 and CD36 were only upregulated on LEC in the presence of VEGFC showing a collaborative effect of these molecules. OA released from adipocytes is taken up by LECs to stimulate the fatty acid β-oxidation, leading to increase adipose tissue lymphangiogenesis. Our results provide new insights on the dialogue between tumors and adipocytes via the lymphatic system and identify a key role for adipocyte-derived FFA in the promotion of lymphangiogenesis, revealing novel therapeutic opportunities for inhibitors of lymphangiogenesis in cancer.

INTRODUCTION
Lymphangiogenesis develops on the edge of solid tumors to provide routes for cancer cells to metastasize to the sentinel lymph nodes and then to distant organs (1). The activation and proliferation of the lymphatic network in the tumor microenvironment is an orchestrated process that involves cooperation between growth factors and adhesion molecules (2). However, tumor stage responsible for the starting signal of the peritumoral lymphangiogenesis remains unclear. The modifications observed in solid tumors during the progression from preneoplastic lesions to adenocarcinoma are sustained by microenvironmental changes including matrix and cells remodeling (3). Among these tissues, the adipose depots constitute the major source of energy for primary tumor and produce various (lymph)angiogenic growth factors, adipokines, hormones, and cytokines that regulate the local microenvironment (4). Vascular endothelial growth factor -C (VEGFC) was first identified as a multifaceted factor promoting the stimulation of tumor lymphangiogenesis (5). Then, other growth factors and bioactive peptides were found to participate to that process including VEGFD, VEGFA, FGF2, HGF and apelin (6)(7)(8). Most of them are synthesized by adipocytes and are well-known to regulate vessel growth (9). In addition to protein released from adipocytes, free fatty acids (FFA) generated by adipocyte lipolysis can modulate blood endothelial cell (BEC) function (10,11). FFAs derive from phospholipase-mediated hydrolysis and are composed by Saturated FA (SAFA), Monounsaturated FA (MUFA) and Polyunsaturated FA (PUFA) (12,13). FFAs are divided into three types depending upon their amino acid chain lengths: short-, medium-, and long-chain FFAs. Among them, the long-chain saturated palmitic acid (C16:0), the monounsaturated oleic acid (C18:1), and the polyunsaturated linoleic acid (C18:2) comprise the majority of FFA found in serum and in lymph where they are assumed to be the cause of leaky lymphatic vessels in obese subjects (14,15).
FFA are well-known risk factors of cardiovascular diseases (16) and are closely related to the events of metabolic syndromes such as obesity and type 2 diabetes (17,18). Palmitic acid (PA) impairs endothelial oxidative metabolism, promotes reactive oxygen species (ROS) formation, and decreases cell viability (19).
Despite these detrimental aspects, there are many clinical studies reporting the protective effects of oleic acid (OA) on flow-mediated dilatation and other blood endothelial markers (20). OA increases lymphatic endothelial cells (LEC) permeability (21) and was proposed as solute for lymphatic delivery of nanoparticles (22) and poorly water-soluble drugs (23). In cancer patients, high FFA level is observed in the serum and may represent an indicator of cancer suggesting their role in tumor progression (24). In particular, OA strongly decreases oxidative stress in melanoma (25). FFAs are transported into cells by fatty acid transporter proteins composed by 6 isoforms (FATP1-6) and the fatty acid translocase/cluster of differentiation 36 (FAT/CD36). After being uploaded by cells, FFAs are converted into metabolites, which participate in a variety of cellular regulatory mechanisms as second messengers. Recent findings highlighted that the beta-oxidation of intracellular fatty acids (FA) is crucial for LEC metabolism, but nothing is known about the role of lymph-circulating Free FA on LEC function (26).
In this study, we found that lymphangiogenesis develops in preneoplastic lesions and in the peritumoral adipose tissue. This lymphangiogenesis is dependent on VEGFR-3 signaling and is associated with an increase of circulating FFA in the lymph. We found that adipocyte-released FFA can be unloaded by LECs to stimulate their sprouting independently of growth factors. We identified the monounsaturated OA as a key regulator of the lymphangiogenesis in cancer. OA in association with VEGFC stimulated the expression of the transporters FATP-3 and -6 and CD36 at the surface of the lymphatic endothelial cells to activate their metabolism and the production of reactive oxygen species (ROS). In the past decade, studies have hinted at important connections between adipose depots and lymphatic vessel function, but clear molecular links have not been established. Here, we report that tumor lymphangiogenesis is not only promoted by the canonical growth factor/tyrosine kinase receptor pathway, but involves a cooperation with lipids, in particular adipocyte-released FFA to improve the lymphatic vessel growth. We demonstrated that VEGFC and OA work together to stimulate the peritumoral lymphatic growth thus providing a reliable network for metastases. Altogether, this work provides new critical insight for the development of anti-lipolysis therapies to reduce tumor lymphangiogenesis and lower the risk of metastasis.

Lymphangiogenesis develops in preneoplastic lesions and in peritumoral adipose tissue.
To study the lymphangiogenesis in the tumor microenvironment, we first investigated the lymphatic system in human pancreatic ductal adenocarcinoma (PDAC) characterized by a strong desmoplastic reaction that prevents the development of blood vessels (27). However, we identified an extensive infiltration of lymphatic vessels into the stroma, which is in accordance with the fact that lymphatics develop in collagenrich tissues (28). The lymphangiogenesis started in preneoplastic lesions (PanIN) (Fig 1A) extending into the vicinity of peritumoral adipose tissue (AT) (Fig 1B) and in draining lymph nodes (Supplementary Fig   1A and B). To confirm the expansion of the lymphatic network in peritumoral AT, we used three mouse models of cancer, including mice with transgenic pancreatic adenocarcinoma (Pdx1Cre; LSLKras G12D Ink4a +/-; PKI mice) (29), melanoma B16 (B16) and Lewis Lung Carcinoma (LLC) (Fig 1C-H) (30). As observed in human tissues, we found a stimulation of lymphangiogenesis in PKI mice starting in preneoplastic lesions (PanIN3) and developing in peritumoral AT (Fig 1C and D). Interestingly, we also found lymphangiogenesis in both tumors (Fig 1E and F) and peritumoral AT (Fig 1G and H) from melanoma B16 and LLC tumor-bearing mice. This was associated with increased systemic inflammation in LLC and B16 tumor models as previously described ( Supplementary Fig 2) (31).

Blocking preneoplastic lesion lymphangiogenesis improves survival.
To study the molecular mechanisms that control the peritumoral lymphangiogenesis, we used a blocking antibody against VEGFR-3 (mF4-31C1) and VEGFR-2 (DC101) to inhibit lymphangiogenesis and angiogenesis respectively (Fig 2) (32). We generated Kaplan-Meier survival curves for the antibody-treated PKI mice (Fig 2A). Without treatment these mice have a median survival of 9 weeks (29). We found that the anti-VEGFR-3 treatment improved the survival of PKI mice significantly, when compared to anti-VEGFR-2 or an isotype control IgG-treated mouse (Fig 2A). As PKI mice rapidly develop ADK, we next used heterozygous mice that have a median survival of 4 months (29) and thus allowed the treatment at preneoplastic stages (17 to 20 weeks). We observed a significant improvement compared to PKI using the blocking VEGFR-3 as survival was three folds increased, thus showing that blocking lymphatic vessel growth in preneoplastic lesions may be pivotal to improve survival rates ( Fig 2B). We next treated B16and LLC bearing-mice with VEGFR-3 blocking antibody. We observed an inhibition of tumor lymphangiogenesis (Fig 2C and D) that was associated with a decrease of tumor metastasis into the sentinel lymph nodes (Fig 2E-G) confirming that blocking peritumoral lymphangiogenesis inhibits metastasis.
Blocking peritumoral AT lymphangiogenesis decreases lymph circulating lipid levels.
Next, peritumoral AT from LLC and B16 tumor-bearing mice treated with the anti-VEGFR-3 blocking antibody were collected and analyzed. We observed an inhibition of peritumoral AT lymphatic density showing that the inhibition of lymphangiogenesis is not restricted to tumors (Fig 3A-C). In parallel, to investigate the role of peritumoral AT lymphatic vessels, dye fluorescent lipid (Bodipy) was injected into the lymphatic system (Fig 3D-E). The fluorescent lipid leaked out of the lymphatics into the peritumoral AT showing the local plasticity and permeability of lymphatics. Interestingly, fluorescent lipids spread to distant adipose depots demonstrating an increased lipid circulation in the lymph from tumor bearing mice compared to wild type mice (Fig 3E and F). This was reduced in the presence of the blocking antibody revealing an active control of lymphatic vessels in the peripheral lipid transport during tumor progression ( Fig 3E and F). Here, we found that lipids are conveyed from tumor proximal to distant adipose depots through the lymphatic system by an active mechanism that can be inhibited by the VEGFR-3 blocking antibody.

Tumor-bearing mice exhibit increased level of circulating Free Fatty Acids in lymph.
To identify protein biomarkers that contribute to the peritumoral lymphangiogenesis, we performed a comprehensive proteomic analysis of the lymph and serum proteomes of the PKI mice (Fig 4). Lymph and serum exhibited different migration profile on acrylamide gels ( Fig 4A). No significant difference in total protein amount was found between control and tumor-bearing mice. However, we identified 3 to 5 times more proteins in the lymph compared to serum, mostly due to the identification of degraded peptides from extracellular space that are drained by lymphatics ( Fig 4B). The majority of proteins identified in lymph compared to serum were involved in lipid metabolism suggesting modifications in surrounding adipose tissue (Fig. 4C). We identified three times more changes in protein amount involved in FFA metabolic processes in lymph ( Fig 4D) compared to serum ( Fig 4E) in tumor-bearing mice.

Oleic acid released by adipose tissue drive LEC sprouting.
We next identified which FFA transport was modified in tumor-bearing mice lymph (Fig 5). Adipose tissue (AT) is a key organ in the regulation of total body lipid homeostasis, which is responsible for the storage and release of FFA according to metabolic needs including Saturated FA (SAFA), Monounsaturated FA (MUFA) and Polyunsaturated FA (PUFA) (12,13). To determine which FFAs were conveyed by lymphatic vessels, lipidomic analysis of the lymph from B16-and LLC tumor-bearing mice treated with the VEGFR-3 blocking antibody was performed (Fig 5A-C). We found that lymph levels of long chain saturated, monoand poly-unsaturated lipids, which are major substrates for mitochondrial β-oxidation, were increased in lymph from tumor-bearing mice and reduced in the mice after treatment with VEGFR-3 blocking antibodies (Fig5 B and C) (33). The most abundant FFAs in the lymph from tumor-bearing mice were oleic acid (OA) and palmitic acid (PA). Their levels were decreased by the anti-VEGFR-3 treatment suggesting that the decrease in lymphatic vessel density in the adipose tissue reduces the FFA transport ( Fig 5C).
We next examined which FFA transporters are expressed on LECs. We confirmed that LECs express FA transport proteins (FATPs) and FA translocase (CD36) (Fig 5D and E, Supplementary Fig 3) (34). In particular, we identified a significant upregulation of FATP3 and FATP6 after OA+VEGFC stimulation suggesting a cooperative effect of these molecules on FFA transporter expression ( Fig 5D). Interestingly, only VEGFC was able to stimulate CD36 expression in LECs ( Fig 5E). Altogether, these data suggest that VEGFR-3 signaling is necessary for the FFA upload by LECs.

Lymphatic FFAs stimulate LEC function
To determine whether AT-released FFAs could control lymphangiogenesis in association with tumor cellsreleased growth factors, we next performed proliferation and tube forming assays on LECs incubated with conditioned media (Fig 6). Pancreatic tumor cells (Fig 6A and B) as well as B16 and LLC (Fig 6C and D) promoted both LEC proliferation and sprouting whereas adipocyte conditioned media (AM) only promoted branch point formation suggesting distinct signaling pathway stimulation (Fig 6E and F). To avoid any effect due to lymphangiogenic growth factors released from adipocytes, heat-inactivated adipocyte conditioned media (heated-AM) were produced. Heated-AM stimulated LEC branch point formation, but not proliferation in the same extent as AM suggesting that non peptidic molecules released from adipocytes can stimulate the LEC function (Fig 6G and H).
The main function of adipocyte is to store lipids and triglycerydes that are hydrolized by lipase to generate the release of free fatty acids (FFA) and glycerol. To investigate the role of lipolysis-generated FFA on LEC, adipocyte lipolysis was induced by beta-adrenergic receptor stimulation using isoproterenol (ISO) and forskolin (FK) (Fig 6I-L). FFA and glycerol release from adipocytes was measured as a control of lipolysis ( Supplementary Fig 4A and B). To determine whether FFA could activate LECs, we incubated cells with bodipy, a neutral lipid that serves as tracer for other lipids. In presence of heated-AM, bodipy accumulated into LECs more than in VEGFC-or AM-stimulated LECs showing the uptake of FFA by LECs (Fig 6J and K). The FFA uptake was increased in the presence of lipolytic media (Fig 6I and J) as well as LEC sprouting (Fig 6K). The effect of FFA on LEC was confirmed as LEC sprouting was reduced in the presence of conditioned media from lipolytic adipocytes treated with hormone sensitive lipase (BAY), and non-selective lipase (E600) inhibitors showing the selective and modular effect of FFA on LEC ( Fig 6L, Supplementary Fig 4C and D).

Exogenous oleic acid stimulates LEC beta-oxidation.
To confirm that FFA uptake stimulates lymphangiogenesis, we next performed lymphatic sprouting assay.
The uptake of OA by LECs stimulated LEC sprouting (Fig 7A). In contrast, PA induced LEC cell death (data not shown) (35), hence suggesting that the beneficial effect of FFA on the lymphatic endothelium is restricted to OA (FA chain length C18). These results were validated by FATP3 and FATP6 siRNA transfection that significantly decreased OA-induced branch point formation in LECs, confirming the crucial role of these transporters in LEC function during lymphangiogenesis induced by VEGFC+OA ( Fig  7A). Lymphatic development is dependent on intra-cellular fatty acids beta-oxidation (26). To investigate whether extra-cellular FFA could also participate to that metabolic process after being uptaken by LECs, we next performed in vitro blockade of mitochondrial FA entry with the CPT1a inhibitor, etomoxir (Eto).
Eto inhibited OA+VEGFC-induced LEC branch formation (Fig 7B), thus confirming that exogenous FFAs could also stimulate the LEC metabolic function. To analyze if the bioenergetic status of LECs is involved in FFA-induced cell functions, we characterized mitochondrial ATP production and oxygen consumption rates (OCR) of OA and/or VEGFC-stimulated LECs (Fig 7 C-E). We found that ATP-linked OCR is increased by OA and this is reduced in the presence of etomoxir or anti-VEGFR-3 (Fig 7 C-E). Interestingly, quantification of mitochondrial reactive oxygen species (mtROS) revealed increased levels of mtROS in VEGFC+OA-stimulated LECs, which was inhibited by the VEGFR-3 blocking antibody ( Fig 7F) and the N-acetyl cysteine (NAC) scavenger of ROS (Fig 7G), showing that the stimulation of LEC function by OA+VEGFC is dependent on the activation of LEC mitochondrial activity. Importantly, the effect of exogenous OA is dependent on VEGFR-3 activation by VEGFC to be uploaded and to stimulate LEC sprouting.

DISCUSSION
The development of solid tumors is often preceded, both in humans and experimental animal models, by the appearance of lesions referred to as preneoplastic. Here, we observed the development of lymphatic vessels in preneoplastic lesions suggesting that tumors sustain an early favorable environment for metastasize to distant loci. Also, we found that AT supports this lymphangiogenesis that is consequently independent of the molecules produced by the adenocarcinoma itself. We confirm the importance of this early lymphangiogenesis as blocking VEGFR-3 pathway in preneoplastic lesions and surrounding AT significantly improved survival and decreased metastases. Adipocytes produce both angiogenic and lymphangiogenic molecules. In human, tumor masses are significantly larger in obese patients and the tumor vessel density inversely correlates with the adipose area leading to a resistance of obese patients to anti-VEGF therapy (36). Multiple factors modulate the-complex interplay between the vascular system-and adipocytes, then targeting the adipose lymphatic vasculature may provide new therapeutic options not only for treatment of obesity and metabolic disorders, but for treatment of cancer (37). VEGFC and -D are elevated in adipose tissue during obesity (38) and an overexpression of VEGFD in adipose tissue resulted in de novo lymphatics and improved overall metabolism in mice (38). Adipocytes also produce peptidic molecules that are able to stimulated the lymphatic endothelial function such as apelin (7). Secreted molecules are then collected by the lymphatic system making the lymph composition representative of the environmental metabolic status. The lymph composition is dependent on the ultrafiltration of plasma proteins as well as proteins and molecules derived from the metabolic and catabolic activities of surrounding tissues (39). Therefore, the lymph protein and lipidic composition conveys major information on environmental tissue metabolic activity. Here we found that unsaturated long-chain FA are prominent compounds of the lymph in cancer and their transport is in part regulated by the lymphatic activity. Among them, OA and PA represent the largest FFA identified. Most of FFA have deleterious effect on the endothelium function (40). Their levels are increased in subjects with obesity-and type 2 diabetes, playing detrimental roles in the pathogenesis-of atherosclerosis and cardiovascular diseases (40). In blood endothelium, FATP3 and FATP4 are required for FFA transport across the vascular endothelial barrier (41).
Here, we observe a difference in blood and lymphatic endothelium as LEC expressed both FATP3 and 6 initially described to mediate fatty acid uptake in cardiomyocytes (42). The mechanism of how FFAs modulate LECs remains unclear. LEC metabolism involves intracellular FA beta-oxidation. Here we found that circulating FFA also promote this process, suggesting a new way for adipose tissue to control lymphatic endothelial physiology. Conditioned media from lipase inhibitor-treated adipocytes significantly reduced LEC sprouting and migration, demonstrating the crucial role of FFAs in the stimulation of lymphatic function. Here, we confirmed the deleterious effect of PA (palm oil) on the endothelium, whereas we observed an opposite effect of OA (olive oil) on lymphatic compared to blood endothelium (40). Whereas OA is known to diminish ATP-induced mobilization in bovine aortic endothelial cells (BAECs)(43), we observed an increase in ATP production and oxygen consumption rates in LEC showing the importance to selectively study the lymphatic transport of FFA in tumoral conditions. Lymphatic vessels are rare in the normal AT but often more conspicuous in tumor-surounding AT. However, the understanding of the mechanisms controlling fat transport and drainage of peripheral AT by the lymphatic system has remained limited to the study of intestinal villi. Our findings provide evidence for a new level of crosstalk between AT and lymphatic drainage to promote tumor lympohangiogenesis. Our work also provides novel therapeutic perspectives for the use of VEGFR-3 blocking antibodies in the regulation of circulating FFA levels. It suggests that targeting the lymphatic system could represent a promising strategy for treating metabolic diseases in which FFA release is stimulated such as obesity-associated type 2 diabetes.

Human tissues
In total, 15 primary human pancreatic adenocarcinoma specimens and their associated lymph nodes were collected. Samples were obtained from archival paraffin blocks of pancreatic cancer from patients treated Etomoxir is an irreversible inhibitor of the CPT1 enzyme that decreases β oxidation in the mitochondria.

Labeling of the lymphatic compartment in AT.
To assess the lymphatic vasculature in peritumoral AT, LLC-and B16-bearing mice treated with control

Lymph collection
Lymph nodes were excised and capsule was carefully dissected in 50L PBS. After gentle centrifugation to do not destruct immune and stromal cells, supernatant (lymph) was retrieved and frozen at -80°C.

ProteoMiner Protein Equalization and analysis of peptides by LC-MS/MS. For proteomic analysis,
we used shotgun proteomics that combines technologies to identify peptides produced by proteolytic digestion of proteins (45) coupled to ProteoMiner (Bio-Rad) that provide random generation of hexapeptides and creates a substantial ligand library for which proteins can selectively bind (46). Briefly, 700 μL lymph and serum (n=9 mice) protein homogenate was concentrated before application to the ProteoMiner low-yield enrichment kit followed by elution and loading 1-D SDS-PAGE gel (Bio-Rad Laboratories Ltd). The resulting peptides were extracted from the gel and analysed by nano-LC-MS/MS using an Ultimate 3000 system (Dionex) coupled to an LTQ-Orbitrap Velos mass spectrometer or LTQ-Orbitrap XL (Thermo Fisher Scientific). Bioinformatic analysis was performed using Ingenuity pathway analysis software. Results have been normalized per mg of total proteins.

Free Fatty Acid Methyl Ester (FAME) analysis.
Lipids corresponding to 10-50 µg of lymph's proteins (from control, B16 and LLC mice) were extracted according to Bligh and Dyer (47) in dichloromethane/methanol/water (2.5 :2.5 :2.1, v/v/v), in the presence of the internal standards heptadecanoate acid (2µg). The lipid extract was directly methylated in boron trifluoride methanol solution 14% (SIGMA, 1ml) and heptane (1ml) at RT for 10min. After addition of water (1ml) to the crude, FAMEs were extracted with heptane (3ml), evaporated to dryness and dissolved in ethyl acetate. FAME were analysed by gas-liquid chromatography(48) on a Clarus 600 Perkin Elmer system using a Famewax RESTEK fused silica capillary columns (30 m x 0.32 mm i.d, 0.25 um film thickness). Oven temperature was programmed from 110°C to 220°C at a rate of 2°C per min and the carrier gas was hydrogen (0.5 bar). The injector and the detector were at 225°C and 245°C respectively.

Sprouting assay.
To perform tube formation assay, twenty-four well plates were coated with 300µl of growth factor reduced matrigel (Corning, 354230) and incubated for 30 minutes in 37°C to allow gelling. 5x10 4 HDLEC were seeded per well in the appropriated medium at a ratio 1:1 with EBM2 containing 0,5% of FBS with or without etomoxir (50-300µM). At least 5 pictures per well were taken after 8 and 24h hours with phasecontrast optical microscope using a 5x objective (Leica, DMi8). The number of branch points was quantified using ImageJ software (NIH).

Heated adipocyte medium.
To eliminate growth factors from adipocyte media, media were heated for 10 min at 95°C.

Bodipy and red oil uptake assay.
HDLECs were seeded on coverslip and cultured for 12h in EGM2-MV medium. To explore the effect of adipocytes conditioned medium on lipid uptake, cells were incubated with the tested medium at a 1:1 ratio

MitoSOX Red
Mitochondrial Superoxide Indicator (Thermofisher) was incubated according to manufacturer instructions.
Briefly, LECs were incubated 15 min at 37°C in the presence of Oleic acid and Palmitic actid stimulation (300μM).