VEGF‐C is required for intestinal lymphatic vessel maintenance and lipid absorption

1Lymphatic vessels regulate tissue fluid homeostasis, immune cell trafficking, and dietary fat absorption, and their malfunction leads to chronic edema and impaired immune responses (Cueni & Detmar,; Alitalo,; Koltowska et al, ). Lymphangiogenesis occurs during pathological processes such as inflammation and tumor metastasis and inhibitors of lymphangiogenic growth factors and their receptors are currently in clinical trials in human cancer patients (Alitalo, ). The development of the lymphatic vasculature is guided primarily by VEGF-C-mediated activation of VEGFR-3, which is the main VEGF receptor expressed by lymphatic endothelial cells (Makinen et al,; Karkkainen et al, ). In the absence of VEGF-C, the development of lymphatic vessels is arrested during their initial spouting from embryonic veins (Karkkainen et al, ). VEGF-D, the second VEGFR-3 ligand, cannot compensate for the absence of VEGF-C during development, but it induces lymphangiogenesis when overexpressed and its deletion during development results in mild lymphatic vessel atrophy in the skin (Rissanen et al,; Karkkainen et al,; Alitalo,; Paquet-Fifield et al,; Astin et al, ).

2During the postnatal period, the lymphatic vessels continue to expand, and in the intestine, the lacteal vessels grow into the intestinal villi to facilitate lipid absorption from the fat-rich milk (Kim et al, ). Lipid absorption allowed Gaspare Aselli to discover lymphatic vessels by their content of milky fluid in the 17th century (Dixon, ). Recent studies have shown that the lacteal vessels are actively involved in lipid transport from the small intestinal epithelium to the lymphatic system and further to the blood circulation, although the detailed mechanisms have not been elucidated (Dixon, ). In several genetic mouse models of impaired lymphatic development, including Vegfr3 and Vegfc hypomorphic mice, and Chy mice that have a missense point mutation in Vegfr3, lipid-rich chylous ascites develops after birth but resolves before weaning (Karkkainen et al,,; Haiko et al, ). However, the possible function of lymphangiogenic growth factors in the normally developed adult intestine is not known.

3Here we have studied the role of VEGF-C in lymphatic vessel growth, maintenance, and function in neonatal and adult mice by using a mouse model that allows effective Vegfc gene deletion by the Cre-Lox system (Aspelund et al, ). We demonstrate that VEGF-C has a crucial role in the maintenance of the intestinal lymphatic vessels and dietary fat absorption.

4In Vegfc gene-deleted embryos, lymphatic vessel sprouting from the major embryonic veins at embryonic day (E) 10.5 is arrested and the embryos die between E15.5 and E17.5 (Karkkainen et al, ). To study how the loss of VEGF-C affects lymphatic vessel development during the last trimester of fetal development, we crossed the Vegfcflox/flox mice (Aspelund et al, ) with mice expressing the universal deletor R26Cre-ERT2 (Ventura et al, ). To delete Vegfc in the R26Cre-ERT2;Vegfcflox/flox embryos, pregnant females were injected with 4-OH tamoxifen at E12.5 and E13.5. Analysis at E18.5 indicated that the Vegfc-deleted (VCiΔR26) embryos lack mesenteric lymphatic vessels either completely (data not shown) or exhibit lymphatic vessel fragments and blind-ended lymphatic stubs that extend toward the intestinal wall (Fig1A and B). Deletion of Vegfc at E14.5 blocked the maturation of the mesenteric collecting lymphatic vessels. The developing vessels lacked valves and were much thinner than in the wild-type (WT) Vegfcflox/flox embryos (Fig1C and D). In addition, the lymphatic vessels in the intestinal wall failed to develop in the VCiΔR26 embryos, whereas the blood vessels were not affected (Fig1E–H).

5Lacteal lymphatic vessels develop after birth and are involved in the absorption of dietary lipids from milk (Kim et al, ). In order to determine whether VEGF-C is necessary for the formation of lacteal vessels, we deleted Vegfc by daily 4-OH tamoxifen injections to VCiΔR26 and WT pups from postnatal day 1 (P1) to P4. Upon necropsy at P6, chylous ascites was observed in the VCiΔR26 pups, but not in the WT littermates (Fig1I and J). Immunofluorescence staining of intestinal cross sections at P6 revealed that the lacteal vessels failed to develop in the VCiΔR26 pups (Fig1K and L and Appendix Fig S1A and B). Lymphatic vessels of the intestinal wall, which develop during the intrauterine period, failed to expand upon Vegfc deletion in postnatal period (Appendix Fig S1C–H). The lymphatic vessel density in the skin was also reduced in the VCiΔR26 pups when compared to their WT littermates (Appendix Fig S1I and J), whereas no changes were observed in the blood vessel density (Appendix Fig S1K and L). Taken together, these results indicate that VEGF-C is required for developmental and early postnatal lymphangiogenesis.

6The VCiΔR26 mouse model allowed us to determine the role of VEGF-C in the maintenance of lymphatic vessels in adult mice. To induce Vegfc gene deletion, we administered tamoxifen to the mice starting at 8 weeks of age. Real-time PCR analysis of several tissues indicated that Vegfc mRNA levels in VCiΔR26 mice were reduced to 2–15% of the levels in WT mice 3 months after tamoxifen treatment (Appendix Table S1). Notably, whole-mount LYVE1 staining revealed that the lymphatic vessels of the intestinal wall undergo a gradual atrophy in the VCiΔR26 mice (Fig1M–U). Shorter lacteals were observed already 3 weeks after Vegfc gene deletion (Appendix Fig S2A), and after 3 months, the lacteals were strikingly thinner and shorter, containing a reduced number of lymphatic endothelial cells, suggesting that vessel atrophy was due to cell loss (Appendix Fig S2B). Thus, the lymphatic vessel atrophy started from lacteals and then progressed to the lymphatic network in the intestinal wall. In contrast, even when Vegfc was effectively deleted for 6 months, there were no changes in lymphatic vessel density in the lymph nodes, ears, or trachea (Appendix Fig S3A–E).

7The development of the lymphatic vasculature relies principally on VEGF-C/VEGFR-3 signaling, while VEGF-D, the other known ligand for VEGFR-3, cannot compensate for the lack of VEGF-C signaling in Vegfc−/− embryos (Karkkainen et al, ). Yet, VEGF-D was found to induce lymphangiogenesis when overexpressed in mouse tissues (Veikkola et al,; Baluk et al, ). Constitutive deletion of Vegfd had no effect on lymphatic vasculature in the adult mouse intestine, in agreement with previous reports (Fig1U) (Baldwin et al, ). Surprisingly, however, conditional deletion of Vegfc in the Vegfd−/− mice resulted in more severe atrophy of the intestinal lymphatic vessels than Vegfc deletion alone (Fig1M–U). These data indicate that VEGF-C is the primary trophic signal for the maintenance of the intestinal lymphatic vessels, but VEGF-D can partially compensate for the absence of VEGF-C. Because we did not observe changes in Vegfd mRNA expression levels upon Vegfc deletion (WT = 1.02 ± SEM 0.14 vs. VCiΔR26 = 0.98 ± SEM 0.16), it is likely that increased bioactive VEGF-D or increased VEGF-D binding to VEGFR-3 partially compensates for the loss of VEGF-C.

8Next, we sought to identify the cell types that express VEGF-C in the adult intestine. Previous work from our laboratory showed that VEGF-C is expressed in arterial smooth muscle cells (SMCs) both in mice and humans (Partanen et al,; Paavonen et al,; Karkkainen et al, ). To define the cells expressing VEGF-C and its receptors, we performed β-Gal staining of intestines from Vegfc/LacZ (Karkkainen et al, ), Vegfr3/LacZ (Dumont et al, ), and Vegfr2/LacZ (Shalaby et al, ) mice. VEGF-C staining in the villi was weak in comparison with VEGFR-3 and VEGFR-2 stainings, in lymphatic and blood vessels, respectively (Fig2A and B). Higher resolution analysis combined with immunohistochemistry demonstrated VEGF-C expression in SMCs, in the inner circular muscle layer of the intestinal wall, in arterial smooth muscle, and in a subset of the SMC fibers in the villus (Fig2C). In the villi, the VEGF-C β-Gal signal was most prominent adjacent to the LYVE1-counterstained lymphatic vessels (Fig2D). The intestinal wall of the Vegfc/LacZ mice showed a prominent arterial β-Gal staining pattern, which was further analyzed by PECAM-1 counterstaining of cross sections (Fig2E), which confirmed VEGF-C expression in the arterial SMCs. Whole-mount confocal microscopy showed a close contact between the lacteal vessels and the SMC fibers in the basal part of the villus where also β-Gal staining of VEGF-C was detected (Fig2D and F). These results suggest that SMCs in the villi and in the intestinal wall provide an important source for VEGF-C, which is required to maintain the lymphatic vessel architecture in the intestine. It should be noted that SMC contractility in the villi has been suggested to be important for dietary lipid absorption and that VEGF-C can induce contraction of the SMCs around the collecting lymphatic vessels in normal and pathological conditions (Hosoyamada & Sakai,,; Breslin et al,; Gogineni et al, ).

9To determine whether the effect of VEGF-C on intestinal lymphatic vessels is dependent on VEGFR-3, we analyzed the Rosa26CreERT2;Vegr3flox/flox (R3iΔR26) mouse model (Haiko et al, ). As expected, deletion of Vegfr3 for 3 months induced lacteal vessel regression, similar to VEGF-C deletion (Appendix Fig S4A and B). However, lymphatic vessel density in the intestinal wall was not altered, suggesting that VEGF-C can signal via VEGFR-2 to stabilize the lymphatic plexus. We have previously shown that the tyrosine kinase inhibitor cediranib inhibits lymphangiogenesis induced by adenoviral VEGF-C delivery into adult mouse skin (Heckman et al, ). However, administration of the tyrosine kinase inhibitor sunitinib at doses that block VEGFR-2 and VEGFR-3 (Ki: VEGFR-3 17 nM/VEGFR-2 9 nM) (Faivre et al, ) had no effect on the lacteal vessels, although sunitinib significantly reduced blood vessel density in the intestinal villi (Appendix Fig S4C–E), in line with a previous report (Kamba et al, ). Thus, lacteal vessels appear to be more resistant than blood vessels in the intestinal villus toward VEGFR tyrosine kinase inhibition.

10To determine whether VEGF-C is important for lipid absorption by the intestinal lymphatic vessels, we performed an oral fat tolerance test in mice in which VEGF-C had been deleted 3 months earlier. The clearance of triglycerides from plasma was blocked by injection of Triton WR 1339, an inhibitor of lipoprotein lipase activity (Otway & Robinson, ). Analysis of triglyceride levels in serum after the administration of an oil bolus showed that the VCiΔR26 mice have impaired lipid absorption when compared to WT mice (Appendix Fig S5A). Although these results correlate with the lacteal regression, we cannot exclude the involvement of possible other effects of Vegfc deletion.

11We further studied whether the reduction in dietary lipid absorption observed in the VCiΔR26 mice has an impact on diet-induced obesity in mice fed high-fat diet (HFD). Initial experiments in the 129SV/C57Bl/6J mixed genetic background did not reveal major differences in body weight, but indicated that the VCiΔR26 mice have an improved glucose metabolism compared to WT mice (Appendix Fig S5B and C). Interestingly, in the mixed background, the Vegfc-deleted mice had reduced serum cholesterol levels, whereas fecal cholesterol and free fatty acid (FFA) levels were increased in the VCiΔR26 mice, indicating impaired dietary lipid absorption (Appendix Fig S5D). We further performed HFD feeding experiments in the pure C57Bl/6J background, an established model of diet-induced obesity. We deleted Vegfc in 8-week-old male mice and started HFD feeding 4 weeks later. The VCiΔR26 mice gained significantly less weight and had better glucose tolerance than their WT littermates, independently of concurrent Vegfd deletion (Fig3A–C and Appendix Fig S5E and F). At necropsy after HFD, very low amounts of chyle were detected in one out of 16 Vegfc-deleted mice and in two out of five Vegfc- plus Vegfd-deleted mice, indicating mild lymphatic leakage. Body composition analysis showed that the VCiΔR26 mice had a significant reduction in total fat weight and fat percentage, but no changes in lean weight in comparison with WT littermates (Fig3D and Appendix Fig S5G). The changes in fat accumulation could not be explained by reduced caloric intake, since food consumption was similar between the VCiΔR26 and WT mice (Fig3E). As expected on the basis of our results from the mixed background, Vegfc deletion induced intestinal lymphatic vessel atrophy and increased lipid excretion into the feces also in the C57Bl/6J background (Fig3F–H). No difference in body weight was observed between WT and Vegfc-deleted mice on regular diet in which the majority of calories are derived from carbohydrate. This further indicates that the reduced body weight of the Vegfc-deleted mice on HFD is a result of reduced dietary lipid absorption.

12Use of the newly established Vegfc gene targeted mouse model allowed us to determine the effect of chronic VEGF-C deficiency in adult mice, where lymphatic vasculature is normally in a quiescent state (Aspelund et al, ). The results of this study show that intestinal lymphatic vessels in adults require trophic signals from VEGF-C and that VEGF-D can only partially compensate for the loss of VEGF-C to maintain their structure and function. The unexpected finding that VEGF-C blockade affects only intestinal lymphatic vasculature and lipid absorption may provide new therapeutic opportunities. The specific VEGF-C/VEGFR-3 inhibitors that are currently in phase I clinical trials for cancer treatment could provide additional benefit for the treatment of obesity and cardiovascular disease by reducing the absorption of excess dietary lipids. Further studies should address lacteal vessel atrophy and dietary fat absorption in clinical trials employing VEGF-C/VEGFR-3 blocking therapeutics.

13National Animal Experiment Board in Finland approved all experiments involving the use of mice.

14The mouse lines Vegfcfl/fl (Aspelund et al, ), Vegfr3fl/fl (Haiko et al, ), Rosa26-CreERT2 (Ventura et al, ), Vegfd (Baldwin et al, ), Vegfc/LacZ (Karkkainen et al, ), Vegfr3/LacZ (Dumont et al, ), and Vegfr2/LacZ (Shalaby et al, ) have been described previously. We used VCiΔR26 mice in the mixed C57Bl/6J and 129SV background or after backcrossing to the C57Bl/6J strain for at least 7 generations. For induction of Cre-mediated recombination in embryos, the mother was injected at the indicated days with two consecutive intragastric doses of 4-OH tamoxifen (4-OHT) (Sigma) (25 mg/ml dissolved in 100 μl ethanol/olive oil). In the neonatal VCiΔR26 or WT mice, the Cre-mediated recombination was induced between P1 and P4 by daily intragastric administration of 2 μl 4-OHT (25 mg/ml dissolved in ethanol). Recombination in adult mice (7–8 weeks old) was done by intragastric tamoxifen (Sigma, dissolved in corn oil at 2 mg/ml, 100 μl) administration during five consecutive days. Detailed animal experiment information can be found in the Appendix Supplementary Materials and Methods.

15Quantitative data were compared between groups by two-tailed unpaired t-test or one-way ANOVA followed by Bonferroni post hoc test for multiple comparisons. Values are expressed as mean ± SEM. P-value < 0.05 was considered significant.