Male and female Prox1^+/– (Prox1^+/LacZ ), Prox1-tdTomato and WT mice (NMRI background) were kindly donated by Dr. Guillermo Oliver (Northwestern University, USA). Six- to eight-week-old mice were used in this study. Animals were housed in temperature- and humidity-controlled rooms, maintained on a 12-h light/12-h dark cycle (lights on at 7:00 hours). All animal procedures and experiments were performed according to animal protocols approved by the Institutional Animal Care and Use Committee at Universidad Autónoma de Chile.

Colitis was induced in mice by adding DSS (3% w/v) (MP Biomedicals) to drinking water for 8 days. Control groups (without DSS) only received drinking water. The general condition and health of the mice was monitored by a routine measurement of body weight and periodic observation. Animals were sacrificed by transcardial perfusion with phosphate-buffered saline 1X (PBS) at different time points (days 2, 5 and 12) and the small intestine and colon were carefully harvested. All tissues were measured, photographed and analyzed as described below.

Colons were isolated from mice, opened longitudinally, washed thoroughly in PBS with 100 U/ml penicillin G and 100 µg/ml streptomycin (Gibco), weighed, and ~0.5 cm sections were cultured in 500 µl RPMI-1640 (supplemented with 10% fetal bovine serum penicillin, streptomycin, and gentamicin) for 24 h at 37°C. Supernatants were then collected and analyzed by ELISA.

Freshly isolated small intestine and colons were washed thoroughly with cold PBS and fixed in 4% ice-cold paraformaldehyde overnight (Sigma-Aldrich, St. Louis, MO). Fixed samples were then washed in PBS and dehydrated on a sucrose gradient with 15% sucrose in PBS, followed by 30% sucrose overnight. Next day, tissues were embedded in Tissue-Tek OCT (Sakura Finetek, Torrance, CA) and stored at -20°C before sectioning. For tail analysis, mice were euthanized, and their tails were obtained, sliced, fixed in 4% ice cold paraformaldehyde overnight (Sigma-Aldrich, St. Louis, MO), and decalcified using 3% EDTA (Loba Chemie Pvt Ltd., Mumbai, India) for 3 weeks, under gentle agitation on a shaker. Fixed samples were then washed in PBS and treated with 15% sucrose in PBS for 5 hours, followed by 30% sucrose overnight. Next day, tissues were embedded in Tissue-Tek OCT (Sakura Finetek, Torrance, CA) and stored at -20°C before sectioning. Samples were sectioned at 30 µm with a cryostat and processes for immunofluorescence. Briefly, samples were slightly permeabilized in PBS +0.5% Triton X-100 for 20 minutes, blocked with a mix of 5% donkey serum (Jackson ImmunoResearch) +1% bovine serum albumin +0.05% sodium azide and 0.1% Triton X-100 in PBS, and incubated overnight at room temperature with anti-LYVE-1 (Cat No. AF2125, RyD Systems) at 1:1000 dilution. The next day, samples were washed with PBS + 0.1% Triton X-100, followed by secondary antibody conjugated with Alexa 488 (Molecular Probes, #A21208). Immune cell infiltration was evaluated by using a mAb against F4/80 (Cat No. 35-4801-U100, TONBO) and CD206 (Cat No. 141712, BioLegend) both at 1:250 dilution, followed by a secondary antibody conjugated with Alexa 488 (Cat No. A21208, Molecular Probes). Samples were washed with PBS + 0.1% Triton X-100; mounted with VECTASHIELD antifade mounting medium with DAPI (Vector Laboratories) and sealed with nail polish. Images were acquired either in an Olympus BX51 epifluorescence Microscope (Olympus, Melville, NY) or in a Leica Stellaris 5 Confocal microscope (Leica Microsystems, Wetzlar, Germany). Lymphatic vascular area and cell count were manually quantified in a blinded manner using ImageJ software. The number of F4/80⁺ and CD206⁺ cells were quantified from the photos obtained in the confocal microscope at a magnitude of 60X after performing a z-stack and a maximum projection of the full stack. Cells were quantified by field and including photographs throughout all the cell layers of the colon (mucosa, submucosa and muscularis).

Serum samples were obtained the day before DSS treatment (day -1) and after DSS treatment (days 2, 5 and 12). For IL-6 detection, capture mAb (Cat No. 504502, BioLegend) and detection mAb (Cat No. 504602, BioLegend) were used following manufacturer’s protocol. IL-1β levels were measured by using antibodies specific for the mouse cytokine (capture mAb: Cat No. 503502 and detection mAb: Cat No. 515801, BioLegend), following manufacturer’s protocol. Samples were measured on an Autobio PHOmo microplate reader in a blinded manner.

Total RNA was isolated from the distal colon using TRIzol reagent according to manufacturer’s instructions (Thermo Fisher Scientific). RNA (1 µg) was reverse transcribed using the High-Capacity RNA to-cDNA Kit (Applied Biosystems) according to the manufacturer’s instructions. Quantitative reverse transcription PCR was performed using SYBR Green Real-Time PCR Master Mix (Thermo Fisher Scientific). The PCR primers used in this study were: IL-6 (forward, 5’-CTGCAAGAGA CTTCCATCCAGTT-3’, reverse 5’-GAAGTAGGGAAGGCCGTGG-3’), IL-1b (forward, 5’-GCAACTGTTCCTGAACTCAACT-3’, reverse 5’-ATCTTTTGGGGTCCGTCAACT-3’), and GAPDH (forward, 5`-GAGGCCGGTGCTGAGTATGT-3’, reverse 5’-GGTGGCAGTGATGGCATGGA-3’). Each complementary DNA sample was analyzed in duplicate or triplicate for quantitative assessment of RNA amplification, and the results are expressed relative to the housekeeping gene GAPDH.

The chyle was obtained directly from the abdominal cavity of Prox1^+/- pups that present visible accumulation of chyle in the first 2-5 days after birth. Chyle was collected from the abdominal cavity of newborn Prox1^+/- pups immediately after euthanasia. The chyle was centrifuged at 300 g for 5 mins and the supernatant was stored at -20°C until use. All the aliquots of chyle obtained from several Prox1^+/- pups were pooled into a single stock and filtered on a 0.2 μm filter. 2 μl of this pooled chyle were used for the in vitro assays (described below). Peritoneal macrophages were obtained as described before(29). Briefly, cells from the abdominal cavity of mice were harvested by repeated lavage with 7 mL cold PBS/FBS 3%, using a 21-G needle. Cell viability was evaluated by the trypan blue dye exclusion method. Then, peritoneal cells were cultured in DMEM alone for 90 min, extensively washed with pre-warmed DMEM medium, and then enriched macrophages were cultured in DMEM medium with 10% FBS, supplemented with penicillin/streptomycin and glutamine overnight. The next day, macrophages were polarized to M0, (complete medium alone), M1 (100 ng/mL IFNγ and 20 ng/mL LPS), or M2 (20 ng/ml IL-4) with or without chyle or vehicle (2 µL) for 24 and 48 hours. Macrophage polarization was assessed by flow cytometry. M1 macrophages were defined as F4/80⁺CD11b⁺CD11c⁺CD206^- or F4/80⁺CD11b⁺ CD11c⁺ CD301^- and M2 macrophages were defined as F4/80⁺CD11b⁺CD11c^-CD206⁺ or F4/80⁺ CD11b⁺CD11c^-CD301⁺ as previously described (30, 31). Bone marrow cells were isolated from 6- to 8-week-old wild-type C57BL/6 mice. 2.5 × 10⁵ cells per well were cultured for 7 days in DMEM supplemented with 10% heated-inactivated FBS plus penicillin/streptomycin in the presence of 10 ng of M-CSF. After 7 days, cultures contained >97% macrophages as assessed by CD11b and F4/80 staining. Polarization was performed as described above in the presence of chyle or vehicle and analyzed 24 hours later.

White adipose tissue from the abdominal cavity was obtained as previously described (32). Briefly, adipose depots were isolated, rinsed and minced in KRB solution (HBSS diluted 1x in PBS; 2% BSA, 12.5 mM HEPES) with 1 mg/ml Collagenase II (Sigma) and 0.2 mg/ml DNAse I (Sigma) at 37°C with high-speed shaking. Cell suspension was filtered through a 100 µM nylon sieve and centrifuged at 500×g at room temperature for 5 minutes. Red blood cells were lysed with erythrocyte lysis (ACK) buffer, centrifuged and pellet SVF cells were staining for FACS analysis.

Acquired lymphedema was surgically induced in the tails of C57BL/6 mice as previously described (33, 34). Briefly, a full-thickness circumferential incision of the skin was made 15 mm distal to the base of the mouse tail. Lymphatic trunks through controlled limited cautery application (lymphedema) or just skin incision without lymphatic cautery (Sham) were performed. Mice were daily monitored to record the lymphedema progression.

Tumors were induced by subcutaneous injection of 3x10⁵ B16/F10 cells on the back of Prox1^+/- and littermate control mice. Tumor growth was recorded every two days with a caliper and calculated as V = 0.52 x d2 x D (D, long diameter; d, short diameter). After 3 weeks, the mice were sacrificed, and the tumors were excised for further analysis.

For macrophage polarization, cells were stained with a FITC-conjugated anti-F4/80 (Cat No. 35-4801-U500, TONBO), PerCP-conjugated anti-CD11b (Cat No. 65-0112-U100, TONBO), PE-conjugated anti-CD301 (Cat No. 145704, BioLegend), APC-conjugated anti-CD11c (Cat No. 20-0114-U100, TONBO), PE-conjugated anti-CD11 (Cat No. 12-0114-82, eBioscience) and Alexa Fluor-647-conjugated anti-CD206 (Cat No. 141712, BioLegend). Intratumoral M2 macrophages were determined by staining tumor cells with a FITC-conjugated anti-F4/80 and Alexa Fluor-647-conjugated anti-CD206. Samples were analyzed by flow cytometry using a FACS Calibur instrument and BD FACSDiva software (version 6.1.1; BD Biosciences, San Jose, CA, USA).

The expression of the gene signatures ( Supplementary Table 1 ), between Macrophages (M1 or M2) and LV signatures were correlated. Furthermore, Kaplan-Meier curves (K-M curves) for each gene signature were generated. For additional methods, see Supplementary Information .

Statistical analysis was performed using Prism (GraphPad Software, USA). The data are expressed as the means ± standard deviation of the mean (SD) or standard error (SEM). Data distributions were tested for normality using Shapiro-Wilk normality test (https://www.statskingdom.com/320ShapiroWilk.html). The significant differences between different data were calculated by unpaired two-tailed t-test (for two groups) and one-way ANOVA (for more than two groups), followed by Tukey’s multiple comparison test. All P values < 0.05 were considered significant. *: P < 0.05; **: P < 0.01; ***: P < 0.001; ****: P < 0.0001.

LV remodeling and dysfunction, especially in initial lymphatics in the villi of the small intestine, called lacteals, has been observed in humans and animal models suffering from IBD (35, 36), although how these LV alterations impact disease progression is not fully understood. We first decided to evaluate LV remodeling during IBD by using the Dextran sodium sulfate (DSS) colitis model, that induces a strong intestinal inflammation, especially in the colon, with symptoms resembling those observed in human ulcerative colitis (37). The exact mechanism whereby DSS causes inflammation is not fully explained; however, it induces reproducible colitis in rodents, and it is an established model of IBD. Although DSS is reported to have its main site of action in the large intestine, it is also known to affect the small intestine (38, 39). DSS treatment in drinking water on Wild Type (WT) mice in NMRI background (herein WT mice), showed dramatic weight loss and colon length shortening, two classical symptoms of colitis ( Supplementary Figures 1A, B ). Lacteal remodeling and regression were evident after 9 days of DSS-treatment ( Supplementary Figure 1C ), consistent with previous studies (38). Since DSS treatment induces inflammation mainly in the colon, we decided to also evaluate lymphatic structure in this tissue, and we observed dramatic remodeling as early as 5 days after DSS-treatment ( Supplementary Figure 1D ). We next analyzed the impact of a defective LV over colitis development by using the Prox1^+/– mouse model, characterized by dysfunctional lymphatic vessels including leakage of chyle present in the peritoneal cavity; reduced clearance and widespread lymphatic vascular leak (6, 14). Surprisingly, body weight loss and colon shortening were less severe in Prox1^+/– mice compared to WT littermate controls ( Figures 1A, B ). To assess whether Prox1^+/– mice have reduced colonic inflammation after DSS treatment, we evaluated the mRNA levels of two proinflammatory cytokines, IL-6 and IL-1β, at different days after treatment. The results showed significant increase in the mRNA levels of these two cytokines in the WT mice, indicating local inflammation, while Prox1^+/– mice did not show local inflammation in response to DSS treatment ( Figures 1C, D ). To confirm these results, we directly measured these proinflammatory cytokines released from colonic explant cultures of WT and Prox1^+/– animals harvested at different days after DSS treatment. The results showed reduced IL-6 and IL-1β production by colon explants from Prox1^+/- mice compared to WT littermate controls after DSS treatment, suggesting a reduced colonic inflammation in this mouse model ( Figures 1E, F ). Moreover, serum levels of pro-inflammatory cytokines were reduced in LV-defective animals treated with DSS, suggesting that systemic inflammation is also reduced ( Figures 1G–J ). We then evaluated whether morphological changes occur in the small and large intestine during DSS treatment by staining LV. When we analyzed LV remodeling, we observed reduced LV morphological changes in both the small and large intestine in Prox1^+/– mice compared to WT littermate controls ( Figures 1K, L ). The most striking finding in the small intestine was the reduced lacteal length in WT and Prox1^+/- mice by DSS treatment; however, the Prox1^+/- mouse is significantly more resistant to this morphological change compared to its WT littermate ( Figure 1K ). Moreover, a substantial increase in Lyve1/Prox1 staining area was observed in the large intestine of WT and Prox1^+/- mice after DSS administration, indicating a proliferating lymphangiogenic response ( Figure 1L ). This growth of the LV area in response to inflammation is much more exacerbated in the WT mouse than in Prox1^+/- . Even in the WT samples, swelling was observed in the mucosal and submucosal areas, which is accompanied by a large increase in LV density ( Figure 1L , white arrows). This is not observed in the Prox1^+/- mouse after treatment with DSS. Quantification of Lyve1⁺/Prox1⁺ staining area further confirmed that Prox1^+/- mouse is significantly more resistant to inflammation with DSS compared to its WT littermate. Together, these results suggest that defective LVs protect animals from DSS-induced colitis and chronic inflammation.

Because of our observation that Prox1^+/– mice are protected from DSS-induced colitis, we next decided to evaluate the immune cell type responsible for this effect. We focused on the colon, since DSS induces inflammation specifically in this tissue. Thus, we evaluated macrophage infiltration by confocal microscopy. Our results showed that after DSS-treatment, an increased number of macrophages were observed in both WT and Prox1^+/– mice, localized in the colon very close to lymphatics, suggesting cell infiltration ( Figures 2A, B ). When we dissected the type of infiltrating macrophage in the colon of DSS-treated mice, we found that Prox1^+/– mice showed increased numbers of M2 macrophages close to lymphatics structures ( Figures 2A, B , higher magnifications and quantifications), suggesting that this anti-inflammatory macrophage subpopulation could be playing a protective role in the DSS-colitis model.

Since the increased amount of M2 macrophages close to lymphatics observed in the colon of DSS-treated Prox1^+/– mice, and because Prox1^+/– animals have LV dysfunction and lymph leakage, we decided to evaluate if chyle can help to polarize macrophages into M2 phenotype. Chyle corresponds to the lipid-rich fluid transported by lymphatic vessels of the intestine. Disruption and/or obstruction of the intestinal lymphatic vasculature results in chylous ascites, due to the leakage of this fluid into the abdominal cavity. While 70% of Prox1^+/- embryos display edema and die at birth due to chylothorax and chylous ascites (6), we extracted chyle from the abdominal cavity of surviving Prox1^+/- pups and used it in our in vitro study. It has been previously shown that lymph extracted from WT or Prox1^+/– mice exert the same adipogenic effect in vitro (14). Additionally, WT and Prox1^+/– lymph were analyzed by gas chromatography and mass spectrometry and no significant differences were found in their composition (14). Thus, peritoneal macrophages from WT mice were cultured under M2 conditional medium in the presence or absence of chyle from Prox1^+/– pups. Our results showed stronger M2 polarization, defined as F4/80⁺CD11b⁺CD11c^-CD206⁺ or F4/80⁺ CD11b⁺CD11c^-CD301⁺, in macrophages cultured in the presence of chyle ( Figures 3A–D ). Similar results were obtained with bone-marrow derived macrophages ( Supplementary Figure 2 ). Next, we decided to evaluate if chyle from Prox1 ^+/- mice can also polarize macrophages towards a M2 phenotype in vivo. It has been previously described that leaky lymphatic in the Prox1^+/- mice is more severe in the abdominal cavity and mesenteric vessels (6), so we analyzed the M1/M2 macrophages population in stromal vascular fraction isolated from abdominal adipose tissue from WT and Prox1^+/– young animals. We observed an increased percentage of M2 macrophages in the stromal vascular fraction in mice with defective LV (Prox1^+/- ) compared to WT animals ( Supplementary Figure 3 ), suggesting that the lymph that leaks out of the LV stimulates M2 macrophage polarization in vivo. To directly evaluate the role of the lymph in promoting M2 macrophage polarization in vivo in a different experimental model, we decided to use a mouse model of tail lymphedema, where thermal ablation of deep collecting lymphatic vessels leads to lymph accumulation and swelling of the tissue (40). In line with our previous results, we observed increased numbers of M2 macrophages in the lymphedematous tissue compared to the sham control group ( Supplementary Figure 4 ), suggesting that lymph accumulation favors M2 macrophage polarization in vivo in this mouse model.

We next evaluated in a different inflammatory context, the role of defective lymphatics in inflammation. Although LV has been implicated in tumor metastasis (41–43), less is known about the role of LV function and dysfunction in local tumor growth. Thus, tumors were induced by subcutaneous injection of B16/F10 cells, a murine melanoma cell line, on the back of Prox1^+/– and WT littermate controls mice. Our results showed that Prox1^+/– mice developed larger tumors than their WT littermate controls ( Figures 4A, B ). Moreover, tumor-infiltrating macrophages from Prox1^+/– mice showed increased M2 macrophages proportion compared to WT littermate controls ( Figures 4C, D ). These results suggest that LV dysfunction and lymph leakage promotes a local anti-inflammatory environment by inducing M2 macrophage polarization that allows accelerated tumor growth.

Finally, we set out to determine whether there is evidence of LV dysfunction and macrophage polarization in human cancers. By using previously described gene signatures for LV, M1 and M2 macrophages, and detailed in Supplementary Table 1 (44, 45), we analyzed tumor transcriptomic data of patients with different cancers available within The Cancer Genome Atlas (46). Independent of the type of cancer evaluated, we found a striking positive correlation between LV and M1 macrophages signatures, and a negative correlation between LV and M2 macrophages signatures ( Figure 5A ). Since reduced LV signature can be related with reduced LV function and lymphedema (47–50), our finding is in keeping with our previous results that defective LV exacerbates M2 macrophage polarization in different inflammatory context and suggest a similar process in human cancers. Moreover, LV signature differentially correlates with patient’s survival depending on the cancer type analyzed and is also in line with the positive or negative association of M1 and M2 signature and cancer survival ( Figure 5B ). In fact, LYVE1, VEGF-C and ITGA9 are among the most down-regulated genes from the low LV signature patients among all the different cancer types analyzed ( Supplementary Figure 5 ). All together, these results suggest that a similar dysfunctional LV modulates M1/M2 macrophage polarization within the tumor microenvironment in different human cancers.