Report

Lacteal junction zippering protects against diet-induced obesity

See allHide authors and affiliations

Science  10 Aug 2018:
Vol. 361, Issue 6402, pp. 599-603
DOI: 10.1126/science.aap9331

Zipping up obesity

Chylomicrons are specialized particles that carry dietary fats from the intestine to the bloodstream for absorption into the body. Lacteals are lymphatic vessels that act as the highway for chylomicron transport, but it is unclear how passage occurs. Zhang et al. report that two endothelial cell receptors, neuropilin-1 (NRP1) and vascular endothelial growth factor receptor 1 (VEGFR1, also known as FLT1), are required to convert the entry spaces between lacteals from open junctions to closed, zipped structures (see the Perspective by McDonald). Mice that were fed a high-fat diet were subsequently rendered resistant to weight gain if NRP1 and FLT1 were inactivated.

Science, this issue p. 599; see also p. 551

Abstract

Excess dietary lipid uptake causes obesity, a major global health problem. Enterocyte-absorbed lipids are packaged into chylomicrons, which enter the bloodstream through intestinal lymphatic vessels called lacteals. Here, we show that preventing lacteal chylomicron uptake by inducible endothelial genetic deletion of Neuropilin1 (Nrp1) and Vascular endothelial growth factor receptor 1 (Vegfr1; also known as Flt1) renders mice resistant to diet-induced obesity. Absence of NRP1 and FLT1 receptors increased VEGF-A bioavailability and signaling through VEGFR2, inducing lacteal junction zippering and chylomicron malabsorption. Restoring permeable lacteal junctions by VEGFR2 and vascular endothelial (VE)–cadherin signaling inhibition rescued chylomicron transport in the mutant mice. Zippering of lacteal junctions by disassembly of cytoskeletal VE-cadherin anchors prevented chylomicron uptake in wild-type mice. These data suggest that lacteal junctions may be targets for preventing dietary fat uptake.

Dietary fats are absorbed by enterocytes and incorporated into triglyceride-rich lipoproteins, called chylomicrons. Nearly all dietary lipids are transported in chylomicrons from the intestine to tissues via the lymphatic system. Chylomicrons enter the lymphatics through lacteals, specialized lymphatic capillaries located in the center of the intestinal villi. From there, they are transported through the mesenteric lymphatic vessels into the thoracic duct, which drains into the venous circulation, and become metabolized by the liver (13).

Preventing lacteal growth in mice by conditional deletion of the genes encoding VEGF-C or Delta-like 4 (DLL4) renders mice resistant to high-fat diet (HFD)–induced obesity, providing experimental evidence that lacteals could be targeted to prevent obesity (4, 5). The cellular mechanisms controlling chylomicron entry into the lacteals are poorly understood; some studies suggested that chylomicrons enter through junctions between lymphatic endothelial cells (LECs) lining the lacteals (57), but other studies have shown that they pass through LECs by transcytosis (3). The molecular mechanisms controlling lacteal chylomicron uptake are currently unknown. Here we show that chylomicron uptake is controlled by VEGF-A signaling, through modulation of lacteal junctions. Lacteals respond to high VEGF-A levels by zippering up their junctions, which impairs chylomicron passage and renders mice resistant to diet-induced obesity.

FLT1 and NRP1 are VEGF-A receptors that were previously implicated in metabolism regulation, but how they function in endothelium in this context remained undefined (fig. S1). We therefore generated pan-endothelial deletions of Nrp1 and Flt1 by crossing Nrp1fl/fl and Flt1fl/fl mice to Cdh5-CreERT2 mice, where gene deletion can be induced by tamoxifen (TAM) in cells expressing the vascular endothelial cadherin promoter (hereafter referred to as Nrp1;Flt1iECko mice). Adult 5-week-old males and Cre-negative littermate controls received TAM to induce efficient gene deletion in the endothelium, as confirmed by Western blot and real-time quantitative polymerase chain reaction on purified endothelial cells (fig. S2). HFD feeding was initiated at week 8 and continued for 16 weeks (wk) (fig. S2). Control mice doubled their body weight after 16wk of HFD feeding, whereas Nrp1;Flt1iECko mice gained little weight (Fig. 1, A to C). No weight differences were observed in mice on a normal chow diet (Fig. 1D). The weight of Nrp1iECko mice was similar to that of littermate controls (Fig. 1E), whereas weight gain in Flt1 single mutants was slightly reduced, and attributed to increased adipose tissue angiogenesis (8, 9) (Fig. 1F), as described previously.

Fig. 1 Nrp1;Flt1iECko mice are resistant to diet-induced obesity.

(A) Nrp1;Flt1iECko and Cre-negative littermate control (Ctrl) mice after 16-week (wk) HFD. (B and C) Growth curves (B) and weight gain (C) of Ctrl (n = 8) and Nrp1;Flt1iECko (n = 11) mice after 16wk HFD feeding. Data are mean ± SEM. ***p < 0.001. Mann-Whitney U test. (D to F) Weight gain of normal chow (NC)- or HFD-fed mice (n = 4 to 11 per group). Data are mean ± SEM. ***p < 0.001. Mann-Whitney U test. (G) Body composition of mice (n = 7 to 8 per group) after 8wk NC or HFD. Data are mean ± SEM. *p < 0.05, ***p < 0.001. Mann-Whitney U test. (H) Quantification of BODIPY staining intensity of liver cryosections (left) and liver triglyceride content (right) from 8wk HFD-fed mice (n = 4 to 5 per group). Data are mean ± SEM. *p < 0.05, ***p < 0.001. Mann-Whitney U test. (I) Calorie intake, energy expenditure, and activity of Ctrl (n = 7) and Nrp1;Flt1iECko (n = 5) mice after 2wk on HFD. Data represent 24 hours (hrs), light hrs (7:00 a.m. to 7:00 p.m.) and dark hrs (7:00 p.m. to 7:00 a.m.) average, normalized to lean body mass (LBM). Error bars: SEM. (J) Intraperitoneal glucose tolerance test in HFD-fed Ctrl (n = 10) and Nrp1;Flt1iECko (n = 6) mice. Data are mean ± SEM. *p < 0.05; **p<0.01. Mann-Whitney U test.

HFD-fed Nrp1;Flt1iECko mice had reduced fat and increased lean body mass ratios when compared to control littermates, and their body composition resembled that of lean control mice on a normal chow diet (Fig. 1G and fig. S3). In contrast to controls, Nrp1;Flt1iECko mice had reduced liver triglyceride content and did not develop hepatic steatosis after 8wk of HFD feeding (Fig. 1H and fig. S4). Nrp1;Flt1iECko and control mice showed similar food and water consumption, physical activity, O2 consumption, CO2 production, and energy expenditure after 2wk and 8wk of HFD feeding (Fig. 1I and fig. S5). Glucose tolerance in Nrp1;Flt1iECko mice was improved compared to controls (Fig. 1J).

Plasma triglycerides, total cholesterol, and high-density lipoprotein cholesterol (HDL) concentrations were reduced in Nrp1;Flt1iECko mice when compared to controls after 16wk on a HFD (Fig. 2A), indicating that resistance to diet-induced obesity was either due to defective lipid uptake into the bloodstream, reduction in very-low-density lipoprotein (VLDL) production, or enhanced VLDL and chylomicron catabolism. To determine if lipid uptake was affected, we measured plasma triglyceride uptake after intragastric gavage with olive oil in the presence or absence of the lipoprotein lipase (LPL) inhibitor TritonWR1339. As expected, circulating triglyceride concentrations increased in controls in both conditions, but the effect was abrogated in Nrp1;Flt1iECko mice (Fig. 2, B and C). Similar results were obtained when we injected the mice with the LPL inhibitor Poloxamer 407 and fed them via gavage with oil mixed with [3H]-triolein (Fig. 2D), indicating that absence of NRP1 and FLT1 attenuates fat absorption. Bomb calorimetry showed increased calorie content in the feces of Nrp1;Flt1iECko mice, suggesting that part of the lipid is not absorbed but rather is cleared in the feces (Fig. 2E). Histological analysis revealed the presence of oil-red-O–labeled lipid particles in the colon of Nrp1;Flt1iECko mice, whereas less lipid was present in the colon of control mice (fig. S6). We confirmed reduced intestinal lipid uptake in Nrp1;Flt1iECko mice by analyzing postnatal day 7 (P7) mesenteries, after neonatal gene deletion. Chylomicrons were present in the mesenteric lymphatic vessels of control mice and single Nrp1 or Flt1 knockout mice, but few lymphatic vessels containing milky chyle could be seen in Nrp1;Flt1iECko pups (Fig. 2F and fig. S7). Neonatal Nrp1;Flt1iECko mice had decreased weight gain (fig. S8), which is consistent with lipid malabsorption.

Fig. 2 Endothelial Nrp1;Flt1 deletion prevents lacteal chylomicron absorption.

(A) Plasma lipid profile in 6-hours fasted Ctrl (n = 6) and Nrp1;Flt1iECko mice (n = 6) after 16wk HFD. Data are mean ± SEM. **p < 0.01. Mann-Whitney U test. (B and C) Plasma triglyceride content in NC-fed adult mice after gavage with 200 μl of olive oil. All mice (n = 8 to 10 per group) received 6 hours of fasting, and mice in (C) received Triton WR1339 (0.5 g/kg) intraperitoneally (i.p.) 30 min before gavage. Data are mean ± SEM. **p < 0.01, ***p < 0.001. Mann-Whitney U test. (D) Plasma 3H CPM (counts per minute) in NC-fed adult mice after gavage with 3H-triolein–containing lipid. Mice (n = 5 per group) were fasted for 6 hours and injected i.p. with poloxamer 407 (1 g/kg) 30 min before gavage. Data are mean ± SEM. *p < 0.05; **p < 0.01, ***p < 0.001. Mann-Whitney U test. (E) Fecal bomb calorimetry analysis of 2wk HFD-fed mice (n = 8 per group). Data are mean ± SEM. ***p < 0.001. Mann-Whitney U test. (F) Quantifications of chyle-filled lymphatics in mesenteries of P7 mice after injection with TAM (3 × 100 μg) at P2 to P4. Each symbol represents one mouse (n = 17 to 24 per group). Error bars: SEM. ***p < 0.001. Mann-Whitney U test. (G) Transmission electron microscopy of jejunum central lacteals in P13 mice. LL: lacteal lumen; LEC: lymphatic endothelial cell; CM: chylomicron. JNC: junction. The closed LEC junction and empty lacteal lumen are apparent in Nrp1;Flt1iECko mice.

Expression of key enzymes and components required for chylomicron assembly, including Mtp, ApoB, Sar1b, Plagl2, and fatty acid transporters (CD36 and Fabps), was similar in the jejunum of Nrp1;Flt1iECko and control mice, indicating that the enterocytes devoid of both Nrp1 and Flt1 can assemble and secrete chylomicrons (fig. S9). Transmission electron microscopy (TEM) showed similar morphology of intestinal enterocytes and the presence of chylomicrons within enterocytes and in the interstitial space between capillaries and lacteals regardless of genotype (Fig. 2G and fig. S10). TEM also revealed the presence of lacteal LECs but notably, the lacteal lumen was almost completely lacking chylomicrons in Nrp1;Flt1iECko animals, whereas numerous chylomicrons were present within the lacteal lumen in controls (Fig. 2G and fig. S10). High-magnification views showed many open junctions between LECs in controls, but more closed LEC junctions in Nrp1;Flt1iECko mice (Fig. 2G and fig. S10). These data are consistent with the concept that altered lacteal junctions result in defective lacteal chylomicron uptake, which limits weight gain in the absence of both NRP1 and FLT1.

Immunostaining of lacteal junctions with VE-cadherin confirmed that Nrp1;Flt1iECko mice exhibited fewer button-like LEC junctions and developed more impermeable zipper junctions compared to controls (Fig. 3, A and B). Lacteals devoid of both Nrp1 and Flt1 were of similar length, but showed reduced width compared to those lacking either Nrp1 and Flt1 alone, or control animals (fig. S11). VE-cadherin staining also revealed enlarged villus capillaries and disrupted blood endothelial cell (BEC) junctions, leading to leakage of intravenously injected fluorescent dextran in Nrp1;Flt1iECko mice (Fig. 3, C and D, and fig. S12). Postnatal or adult deletion of both Nrp1 and Flt1 caused altered vascular morphology and villus edema, which was surprisingly well tolerated and did not compromise vascular or enterocyte ultrastructure even after prolonged HFD feeding (fig. S10). Staining for pericytes, smooth muscle cells, and macrophages and analysis of inflammatory marker expression revealed no changes between controls and Nrp1;Flt1iECko mutants (fig. S13), suggesting a primary effect upon villus vasculature by the absence of Nrp1 and Flt1.

Fig. 3 Increased VEGF-A signaling alters villus endothelial junctions.

(A) VE-Cad and LYVE-1 staining of whole-mounted jejunum lacteals from P13 to P18 mice after postnatal TAM administration. (B) Quantification of zipper-like lacteal junctions in (A). Each symbol represents one lacteal (7 mice per group). Data are mean ± SEM. ***p < 0.001. Mann-Whitney U test. (C) VE-Cad staining of whole-mounted jejunum BEC junctions from P7 mice. (D) Quantification of dextran leakage in P11 mice after Rhodamine-dextran and Alexa 647-IsoB4 intravenous (i.v.) injection for 10 min. Each symbol represents one villus (5 to 6 mice per group). Data are mean ± SEM. ***, p < 0.001. Mann-Whitney U test. (E) Western blots and quantifications of VEGFR2 phosphorylation in P7 jejunum lysates. n = 5 to 6 mice per group. Data are mean ± SEM. **p < 0.01. Mann-Whitney U test. (F and G) VE-Cad staining of villus BEC junctions and quantification of zipper-like lacteal junctions in jejunum from P18 to P21 wild-type mice 30 min after injection of growth factors or phosphate-buffered saline (PBS). Each symbol represents one lacteal (4 to 6 mice per group). Data are mean ± SEM. n.s., not significant; **p < 0.01, ***p < 0.001. Mann-Whitney U test.

Loss of junctional VE-cadherin staining in BECs is a hallmark of increased VEGF-A signaling (10), and FLT1 is a known VEGF-A decoy receptor (11). To test if loss of NRP1 and FLT1 increased intestinal VEGF-A signaling, we probed lysates from control and Nrp1;Flt1iECko tissues with antibodies against phosphorylated and total VEGFR2. Activation of both tyrosine-1173 (Y1173) and Y949 signaling was significantly increased in the absence of both Nrp1 and Flt1 when compared to control littermates or single Flt1iECko mice (Fig. 3E and fig. S14), indicating that NRP1 enhances FLT1 decoy function. NRP1 also acts as a VEGFR2 co-receptor to activate downstream ERK signaling (12), and ERK activation was accordingly deficient in most Nrp1;Flt1iECko tissues (fig. S14).

Loss of Flt1 has been shown to induce adipose tissue browning (8, 9), and we observed increased vascular density in white adipose tissue (WAT) but not in brown adipose tissue (BAT) in Nrp1;Flt1iECko mice (fig. S15). Expression levels of the mitochondrial uncoupling protein UCP1 were only slightly increased in Nrp1;Flt1iECko epididymal WAT and were unchanged in subcutaneous WAT and BAT (fig. S15). Thus, adipose tissue browning is unlikely to account for resistance to diet-induced obesity in Nrp1;Flt1iECko animals.

VEGF-A is known to increase blood-vascular permeability (13, 14), and lacteals express VEGFR2 (fig. S16), but effects of VEGF-A on lacteal junctions have previously not been investigated, to the best of our knowledge. Intravenous VEGF-A injection disrupted BEC junctions after 30 min as expected, but also increased the number of zipper-like junctions in lacteals (Fig. 3, F and G). Intravenous injection of the lymphatic growth factor VEGF-C also increased junction zippering, but to a lesser extent than VEGF-A, whereas a mutant VEGF-C-156S protein that cannot bind VEGFR2 (15) had no effect (Fig. 3G and fig. S16). In vitro, treatment of starved confluent LECs with VEGF-A and VEGF-C, but not VEGF-C-156S, promoted the appearance of straighter VE-cadherin–lined junctions and decreased actin stress fiber anchoring to perpendicular arranged VE-cadherin (fig. S16). These data suggested that increased VEGF-A–VEGFR2 signaling in Nrp1;Flt1iECko villi might cause defective chylomicron uptake via lacteal junction zippering.

Intestinal villi show high levels of Vegf-a expression, lower levels of Vegf-c, and little expression of Vegf-b (fig. S17). Expression of Flt1 is increased at birth, concomitantly with the expression of Mtp and ApoB, a known inducer of Flt1 (16) (fig. S17). RNA sequencing showed that mature intestinal BECs express four times as many copies of Nrp1 when compared to Flt1 (fig. S17), consistent with the concept that chylomicrons induce an increase of FLT1, which functions together with NRP1 to prevent excessive VEGF-A signaling, thereby allowing lacteal junction maturation and chylomicron absorption.

Cdh5-CreERT2 mediates gene deletion in both intestinal BECs and LECs, raising the possibility that NRP1;FLT1 function might be required in both compartments in a cell-autonomous manner. We therefore generated LEC-specific Nrp1 and Flt1 deletions using Prox1-CreERT2(BAC) mice. Prox1-CreERT2(BAC) specifically recombined mesenteric and lacteal LECs, but loss of Nrp1;Flt1 in LECs had no detectable effects on postnatal mesenteric chyle uptake or on weight gain in adult HFD-fed mice (fig. S18). Lymphatic button-like junctions were present and responded to VEGF-A by forming zippers (fig. S18). Together with highly enriched expression of Nrp1 and Flt1 in intestinal BECs compared to LECs (17), these data suggest that NRP1 and FLT1 function in villus BECs to antagonize VEGF-A–VEGFR2 signaling on lacteal LECs.

The effect of Nrp1;Flt1 deletion on LEC junctions was tissue-specific, as the junction morphology in initial lymphatics in the skin and diaphragm of Nrp1;Flt1iECko mice was similar to that of controls (fig. S19). Intravenous VEGF-A, but not VEGF-C or VEGF-C-156S, modestly increased dermal initial lymphatic junction zippering (fig. S19). However, intradermal injection of VEGF-A or VEGF-C robustly increased dermal LEC junction zippering, whereas VEGF-C-156S did not (fig. S19). High VEGFR2 signaling may therefore be a general inducer of lymphatic junction zippering, which is antagonized by NRP1;FLT1 decoys in the small intestine and via other mechanisms in other tissues.

If enhanced VEGF-A signaling accounted for defective chylomicron uptake in Nrp1;Flt1iECko mice, blocking VEGFR2 function should rescue this process. Indeed, intraperitoneal injection of the VEGFR2 blocking antibody DC101 (18) decreased VEGFR2 phosphorylation, and increased uptake of mesenteric chyle in neonates 6 hours after treatment (Fig. 4, A and B, and fig. S20). Dilation of villus capillaries was reduced, and width of lacteals increased (fig. S20). DC101-treated double mutants showed less dextran leak from intestinal capillaries (fig. S20), had more LEC button-like junctions and fewer zippers, and had higher plasma lipid concentrations in the bloodstream when compared to untreated Nrp1;Flt1iECko animals (Fig. 4C-D).

Fig. 4 Lacteal zipper junctions prevent chylomicron absorption.

(A to D) DC101 rescues fat absorption in Nrp1;Flt1iECko mice. (A) Experimental timeline of TAM injection and DC101 treatment (30 μg/g, i.p., for 6 hours) and mesenteries of P7 Nrp1;Flt1iECko mice with or without DC101. (B) Quantification of chyle-filled lymphatics in (A). Each symbol represents one mouse (n = 11 to 24 per group). Data are mean ± SEM. n.s., not significant; ***p < 0.001. Mann-Whitney U test. (C) Quantification of zipper-like lacteal junctions in P13 to P18 Nrp1;Flt1iECko jejunum with or without DC101. Each symbol represents one lacteal (4 to 7 mice per group). Data are mean ± SEM. n.s., not significant; ***p < 0.001. Mann-Whitney U test. (D) Plasma triglyceride content in NC-fed adult mice after 6 hours of fasting and lipid gavage. All mice (n = 5 to 8 per group) received TritonWR1339 (0.5 g/kg, i.p., 30 min before gavage) and some received DC101 (30 μg/g, i.p., 4 hours before gavage). Data are mean ± SEM. *p < 0.05, **p < 0.01. Mann-Whitney U test. (E and F) BV13 rescues chylomicron uptake in Nrp1;Flt1iECko mice. (E) Quantification of chyle-filled lymphatics in P7 mice with or without BV13 treatment (10 μg/g, i.p., for 4 hours). Each symbol represents one mouse (n = 5 per group). Data show mean ± SEM. n.s., not significant; **p < 0.01. Mann-Whitney U test. (F) VE-Cad and LYVE-1 staining of jejunum lacteals from P13 Nrp1;Flt1iECko mice with or without BV13. (G to I) ROCK inhibition prevents chylomicron uptake in wild-type mice. (G) Quantification of chyle-filled lymphatics in mesenteries of P10 mice treated with Y27632 (20 μg/g, i.p., for 4 hours) or PBS. Each symbol represents one mouse (n = 5 per group). Data are mean ± SEM. *p < 0.05. Mann-Whitney U test. (H) Quantification of zipper-like lacteal junctions in P13 to P15 mice treated with Y27632 or PBS. Each symbol represents one lacteal (4 mice per group). Data are mean ± SEM. ***p < 0.001. Mann-Whitney U test. (I) TEM analysis of lacteals in P13 mice with or without Y27632 treatment. LL: lacteal lumen; LEC: lymphatic endothelial cell; CM: chylomicron. JNC: junction. (J) Model of NRP1;FLT1 effects on BECs and LECs in intestinal villi. VEGF-A binding to NRP1;FLT1 on BECs limits VEGF-A bioavailability for VEGFR2, resulting in continuous and discontinuous cell junctions in BECs and LECs, respectively. Discontinuous button-like LEC junctions allow lacteal chylomicron uptake. NRP1;FLT1 are highly expressed only on BECs. Increased VEGF-A concentrations or Nrp1;Flt1 deletion in BECs result in excessive VEGFR2 activation, which disrupts BEC junctions while zippering up lacteal LEC junctions, thereby preventing chylomicron uptake. This phenotype can be rescued by inhibition of VEGFR2 signaling with DC101.

We next determined if changes in lacteal junctions were sufficient to impair chylomicron uptake. Button-like junctions develop postnatally by transformation of zipper-like junctions in a poorly characterized mechanism that involves angiopoietin2 (ANGPT2) and changes in plasma membrane localization of VE-cadherin and LYVE-1 (1921). LYVE-1 immunostaining and protein abundances of LYVE-1 and ANGPT2 were comparable between Nrp1;Flt1iECko mutants and controls, and inhibition of matrix metalloproteinase (MMP)–mediated LYVE-1 cleavage (22) had no effect on mesenteric chylomicron uptake (fig. S21). To test VE-cadherin function, we used the BV13 antibody, which disrupts endothelial adherens junctions in BECs and LECs (19). Intraperitoneal injection of BV13 (10 μg/g) into postnatal mice rescued lacteal chylomicron uptake into mesenteric lymphatics in Nrp1;Flt1iECko mutants and disrupted LEC junctions (Fig. 4, E and F, and fig. S22). Notably, BV13 treatment disrupted capillary BEC adherens junctions but did not affect chylomicron uptake in control mice (fig. S22). This result indicates that disruption of BEC junctions per se has no effect on chylomicron transport and that lacteal LEC junctions, but not capillary BEC junctions, control chylomicron uptake.

In a complementary approach, we treated Nrp1;Flt1iECko mice with dexamethasone, a drug that induces zipper–to–button-like junction transition in tracheal lymphatics (21). Short-term dexamethasone treatment decreased the number of zipper-like junctions and increased chyle uptake in mesenteries of a subset of Nrp1;Flt1iECko mice (fig. S23). Nrp1;Flt1iECko mutant BEC junctions were unaffected by dexamethasone, but treated mutant animals exhibited fewer zipper-like junctions (fig. S23), confirming that LEC but not BEC junctional changes correlate with chylomicron uptake.

We reasoned that junctional changes between continuous VE-cadherin–lined zippers and discontinuous buttons could depend on cytoskeletal VE-cadherin anchoring, which can be inhibited in BECs by antagonizing Rho guanosine triphosphatase (GTPase) signaling (23). Indeed, the ROCK inhibitor Y27632 induced straighter junctions of cultured LECs in vitro (fig. S24), and short-term treatment of wild-type mice with Y27632 reduced chylomicron transport to mesenteric lymphatics and increased lacteal junction zippering, without affecting BEC junctions (Fig. 4, G and H, and fig. S24). TEM showed open lacteal junctions and chylomicrons in the lacteal lumen of untreated mice, but closed junctions and almost no chylomicrons in the lacteal lumen in Y27632-treated mice (Fig. 4I).

Together, our findings demonstrate that high VEGF-A signaling has opposing effects on blood and lymphatic vessels: opening capillary cell-cell junctions via well-studied pathways (10, 24), but closing lymphatic junctions via transformation of buttons into zippers. We show that LEC VEGFR2 activation inhibits VE-cadherin cytoskeletal anchoring, which can be mimicked by ROCK inhibition in vitro and in vivo. Cytoskeletal VE-cadherin anchoring might generate pulling forces that help maintain lacteal button junctions open. Notably, we show that short-term effects of VEGF-A gain of function (over hours instead of days) (25) is sufficient to functionally switch junction morphology and lipid uptake. This reveals an intestinal-specific lacteal barrier that could be closed on demand to prevent lipid uptake (Fig. 4J). Thus, NRP1 and FLT1 appear to function collaboratively as a double decoy receptor system in intestinal BECs to limit VEGF-A signaling (Fig. 4J). VEGF-A–VEGFR2 signaling might provide a LEC growth signal together with VEGF-C–VEGFR3 (21, 26). In addition, VEGFR2 signaling acts as a physiological inhibitor of lacteal maturation, whose activity must be dampened to allow acquisition of transport function. This mechanism may also apply to other tissues, with implications for edema prevention. Full knowledge of the mechanism responsible for LEC junction zippering may hold promise for identifying molecular targets for the treatment of obesity. Of note, ROCK inhibitors are clinically approved for treatment of cerebral vasospasm in some countries and have been shown to improve metabolism and decrease obesity in rodent models (2730). Further studies of the effects of such drugs in obesity prevention, as well as of potential adverse effects of lacteal junction zippering on intestinal health, may be warranted.

Supplementary Materials

www.sciencemag.org/content/361/6402/599/suppl/DC1

Materials and Methods

Figs. S1 to S24

Table S1

References (3151)

References and Notes

Acknowledgments: We thank Yale Mouse Metabolic Phenotyping Center, Yale Electron Microscopy Facility, and Cincinnati Mouse Metabolic Phenotyping Center for assistance; Y. Zhao for ANCOVA analysis; and K. Alitalo, J.-L. Thomas, and S. Lee for discussion. Funding: This study was supported by grants from NHLBI (1R01HLI125811), NEI (1R01EY025979-01, P30 EY026878), and Agence nationale de la recherche (ANR Brainwash) to A.E. F.Z. was supported by a T32 grant from NHLBI (2T32HL7950-16A1). G.Z. was supported by an EMBO Long-Term fellowship (ALTF 87-2016) and a T32 grant from NHLBI (2T32HL007974-16). A.D. and R.O. were supported by AHA postdoctoral fellowships 14POST20380207 and 15POST25560114, respectively. M.S. was supported by NIH grants R01HL053793 and P01HL107205. Author contributions: F.Z., G.Z., J.H., J.L., A.D., R.O., G.G., K.B., P.M., S.E.K., J.P.C., and A.K.S. performed experiments and analyzed data. A.E. and F.Z. designed experiments, analyzed data, and wrote the manuscript. G.-H.F., M.S., P.T., C.F.H., W.C.S., and G.I.S. provided reagents or expertise. Competing interests: A.E. and F.Z. are inventors on a patent application (U.S. Provisional Application no. 62/686,140) submitted by Yale University that covers compositions and methods for preventing dietary lipid uptake and obesity. The authors declare no additional competing interests. Data and materials availability: All data are available in the manuscript or the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article