|Home | About | Journals | Submit | Contact Us | Français|
Dietary saturated fat is linked to numerous chronic diseases, including cardiovascular disease. Here we show that the lipoprotein lipase inhibitor Angptl4 protects against the pronounced pro-inflammatory effects of dietary saturated fat. Strikingly, in mice lacking Angptl4, dietary saturated fat induces a severe and ultimately lethal phenotype characterized by fibrinopurulent peritonitis, ascites, intestinal fibrosis, and cachexia. These abnormalities are preceded by a massive acute phase response induced by saturated but not unsaturated fat or medium-chain fat, originating in the mesenteric lymph nodes (MLNs). MLNs undergo dramatic expansion and contain numerous lipid laden macrophages. In peritoneal macrophages incubated with chyle, Angptl4 dramatically reduced macrophage foam cell formation, inflammatory gene expression, and chyle-induced activation of the ER stress pathway. The data reveal a novel mechanism in which induction of macrophage Angptl4 by fatty acids serves to reduce postprandial lipid uptake from fatty chyle into MLN-resident macrophages by inhibiting triglyceride hydrolysis, thereby preventing macrophage activation and foam cell formation and protecting against progressive, uncontrolled dietary saturated fat-induced inflammation.
Abundant evidence indicates that elevated saturated fat consumption is associated with increased risk for chronic diseases, including cardiovascular disease and type 2 diabetes. However, the underlying mechanisms and why specifically saturated fat is harmful largely remains unknown. Consequently, there is a need to better understand the physiological and molecular mechanisms that govern the response to dietary (saturated) fat ingestion.
After digestion of dietary fat, absorbed long-chain fatty acids are incorporated into chylomicrons as triglycerides (TG) and released into the circulation after passage through the intestinal lymphatics. Hydrolysis of chylomicron-TG is catalyzed by the enzyme lipoprotein lipase (LPL), which is anchored to the capillary endothelium via heparin–sulphate proteoglycans and represents a key determinant of cellular fatty acid uptake (Merkel et al., 2002). LPL is expressed at high levels in tissues that depend on fatty acids as fuel (heart, skeletal muscle), or synthesize fats for storage or secretion (adipose tissue, mammary tissue), but high expression is also found in macrophages (Ostlund-Lindqvist et al., 1983; Wang and Eckel, 2009).
Activity of LPL is governed via numerous mechanisms that act primarily at the posttranscriptional and posttranslational level. One important modulator of LPL activity is Angiopoietin-like protein 4 (Angptl4) (Yoshida et al., 2002). Angptl4 was discovered as transcriptional target of the peroxisome proliferator activated receptor alpha and gamma and is expressed in numerous cell types including adipocytes, hepatocytes, (cardio)myocytes and endothelial cells (Kersten et al., 2000; Yoon et al., 2000).
Studies using different transgenic mouse models of Angptl4 over-expression or deletion show that Angptl4 potently raises plasma TG levels by suppressing LPL-mediated clearance of plasma TG-rich lipoproteins (Koster et al., 2005; Mandard et al., 2006; Xu et al., 2005). Angptl4 inhibits LPL by promoting the conversion of catalytically active LPL dimers into catalytically inactive LPL monomers (Lichtenstein et al., 2007; Sukonina et al., 2006). The anti-lipolytic effect of Angptl4 may be suppressed as a result of binding of LPL to GPIHBP-1 (Sonnenburg et al., 2009).
Recently, it was shown that deletion of other members of the Angiopoietin-like protein family influences the development of obesity-related complications in the C57Bl/6 mouse high fat-induced obesity model. Specifically, it was shown that surviving Angptl6−/− mice developed marked obesity, ectopic fat storage and insulin resistance. In contrast, mice over-expressing Angptl6 were leaner and had improved insulin sensitivity (Oike et al., 2005). Deletion of Angptl2 ameliorated adipose tissue inflammation and insulin resistance in obese mice, whereas Angptl2 over-expression resulted in elevated adipose tissue inflammation and systemic insulin resistance (Tabata et al., 2009). Given the role of Angptl4 in plasma clearance of dietary TG, we set out to study the effect of Angptl4 deletion in the context of chronically elevated dietary fat intake via high fat feeding.
To examine the effect of Angptl4 on diet-induced obesity and its metabolic consequences, WT and Angptl4−/− mice were fed a high fat diet (HFD) containing saturated fat-rich palm oil and compared with mice fed low fat diet (LFD) (Table S1). As expected based on its ability to inhibit LPL, Angptl4−/− mice had decreased plasma TG (Fig. 1A) and showed faster initial weight gain (Fig. 1B) (Voshol et al., 2009). Remarkably, bodyweights of Angptl4−/− mice fed HFD reached a plateau after around 12 weeks and declined thereafter (Fig. 1B). The decrease in bodyweight was related to anorexia, which was noticeable after about 10 weeks of HFD (Fig. 1C). If left undisturbed, all Angptl4−/− mice fed HFD ultimately die anywhere between 15 and 25 weeks. The cause of death was identified by an animal pathologist as severe fibrinopurulent peritonitis connected with ascites.
Large amounts of fibrin exudate covered the abdominal organs in Angptl4−/− mice fed HFD (Fig. 1D). Other macroscopic abnormalities included intestinal fibrosis (Fig. 1E), a compressed liver (Fig. S1A), hyperplasic spleen, and subcutaneous hyperemia (data not shown). No such abnormalities were observed in Angptl4−/− mice fed LFD, even at advanced age (>1.5 years). Routine clinical tests were performed on the ascites fluid, which varied in color from purulent white to purulent red (Fig. 1D, inset). Ascites white blood cell count was extremely high in all animals (25.5–34.1*109/L, diagnostic threshold: <0.5*109/L), as was the endotoxin concentration (50–120 EU/mL, zero threshold), strongly suggesting bacterial peritonitis. Ascites fluid of some animals tested positive for E. Coli. The high protein concentration (3.43–4.28 g/dL, diagnostic threshold: >2.5 g/dL) and low serum-ascites albumin gradient (SAAG, 0.11–0.34 g/dL, diagnostic threshold: <1.1 g/dL) indicated exudative ascites, thereby excluding portal hypertension. The ascites TG concentration was highly variable but clearly elevated (4.8–75.5 mM, diagnostic threshold: ~1.25 mM). Analysis of chylous ascites fluid by lipoprotein profiling indicated an abundance of TG-rich lipoproteins representing chylomicrons, as shown by immunostaining for apoB (Fig. 1F), suggesting leakage of chyle from lymphatic vessels. Additionally, significant vascular leakage occurred, as shown by the much lower protein concentration in chyle (1.38–1.83 g/dL) compared to the ascites fluid (3.43–4.28 g/dL).
Plasma endotoxin levels were significantly elevated in Angptl4−/− mice fed HFD for 19 weeks (Fig. 1G). Microscopic examination indicated that the fibrin exudate contained an abundance of foam cells, polynuclear giant cells, and various other leukocytes (data not shown). The same cells as well as focal lymphocyte infiltrates were observed in the small intestine (Fig. 2A, inset) and mesenteric fat (Fig. 2B), in the former encapsulated by collagen (Fig. 2C). Intestinal lymph vessels were significantly dilated, suggesting mesenteric lymphatic obstruction (Fig. 1H). Epididymal adipose tissue had a red appearance (Fig. S1C), and exhibited coagulation necrosis and steatitis as shown by presence of lymphocytes, granulocytes and other leukocytes, mainly at the periphery (Fig. S1D, inset).
Livers of Angptl4−/− mice fed HFD for 19 weeks were not fibrotic but resembled an ischemic liver. Portal triads, cords and sinusoids were poorly visible, and clumping of nuclei was seen, indicating collapse of liver (Fig. S1A,B). Focal infiltrates of neutrophils, eosinophils and macrophages were observed (Fig. S1B, inset), as were rod-shaped bacteria. Liver fat was almost absent, whereas it was elevated in Angptl4−/− mice on LFD (Fig. S1E). Weights of liver and epididymal fat pads were significantly lower in Angptl4−/− mice after 19 weeks of HFD (Fig. S1F,G), while epididymal fat pads were heavier in Angptl4−/− mice on the LFD. From these data it is evident that Angptl4−/− mice chronically fed a HFD develop a severe phenotype characterized by anorexia, cachexia, intestinal inflammation and fibrosis, chylous ascites, and fibrinopurulent peritonitis, ultimately leading to the death of the animal.
Angptl4−/− mice on a mixed genetic background were reported to die shortly after birth due to defective separation of the intestinal lymphatic and blood microvasculature (Backhed et al., 2007). Although we did not find these abnormalities in Angptl4−/− mice on pure C57Bl/6 background and adult mice in the proper Mendelian ratios were obtained, there might be an underlying primary weakness in intestinal lymphatics that becomes manifest when chyle flow is increased as with HFD, causing leakage of chyle into intestinal lumen and peritoneal cavity. However, none of the clinical abnormalities including chylous ascites were observed in Angptl4−/− mice fed a safflower oil-based HFD rich in polyunsaturated fat (Table S2) (data not shown, see below). Furthermore, if lymphatic vessels are intrinsically more permeable, ascites should develop upon starting the HFD, which was not observed. Additionally, inconsistent with loss of protein, fat and water from lymph into the intestinal lumen, no diarrhea was observed and surprisingly fecal fat excretion was markedly decreased in Angptl4−/− mice, indicating more efficient fat absorption (Fig. S2A). An acute intestinal lipid absorption test using 3H-triolein and 14C-palmitic acid failed to show any differences in rate of appearance of either label in blood between WT and Angptl4−/− mice (Fig. S2B,C), suggesting that chylomicron formation and release into the bloodstream is similar between WT and Angptl4−/− mice. In contrast, in all parts of the intestine, accumulation of both labels five hours after the lipid load was markedly higher in the Angptl4−/− mice (Fig. S2D,E). The similarity in results between 3H-triolein and 14C-palmitic acid argue against an effect of Angptl4 inactivation on TG digestion but instead suggest enhanced fatty acid uptake into enterocytes. This is supported by elevated expression of target genes of PPARa in small intestine of Angptl4−/− mice, indicating enhanced gene regulation by fatty acids (Fig. S2F,G). Overall, these data argue against a primary lymph vessel defect forming the basis for the severe pathology, suggesting a different and more progressive type of etiology.
To investigate the cause of the severe pathology, WT and Angptl4−/− mice were studied before onset of anorexia and cachexia at 8 weeks of HFD (Fig. 1A,B). Importantly, no ascites or other macroscopic abnormalities were observed. Strikingly, after 8 weeks of HFD, plasma levels of serum amyloid A (SAA) and other inflammatory markers were dramatically increased in Angptl4−/− mice (Fig. 3A,B). Whereas IL-6 was undetectable in plasma of WT mice, levels averaged 63±26 pg/mL in Angptl4−/− mice (normal values <15 pg/mL). These changes were paralleled by massive induction of hepatic mRNA for serum amyloid 2 and other acute phase proteins haptoglobin and lipocalin 2 (Fig. 3C). Furthermore, increased expression of macrophage/Kupffer cell marker Cd68 (Fig. 3D) and enhanced Cd68 immunostaining was observed (Fig. 3E). Consistent with these data, serum levels of negative acute phase protein albumin were decreased (Fig. 3F). These data indicate that Angptl4−/− mice fed HFD exhibit systemic inflammation and a massive acute phase response several weeks prior to development of ascites and other clinical symptoms.
Chronic HFD is known to induce adipose tissue inflammation, characterized by adipose infiltration of macrophages (Neels and Olefsky, 2006). However, no signs of enhanced macrophage or other leukocyte infiltration were observed in epididymal fat of Angptl4−/− mice after 8 weeks of HFD (Fig. S3A,B), suggesting that the enhanced systemic inflammation does not originate in the adipose tissue.
Interestingly, a trend towards increased plasma SAA levels in Angptl4−/− mice was already visible after one week of HFD (Fig. 4A). HFD has been proposed to lead to inflammatory stress via changes in intestinal microflora and/or increased release of LPS (Cani et al., 2007), which might contribute to the increase in plasma inflammatory markers in Angptl4−/− mice fed HFD. However, inconsistent with this scenario, portal LPS levels were lowest in Angptl4−/− mice fed HFD (Fig. 4B). Importantly, the increase in plasma SAA levels in Angptl4−/− mice by HFD was unaffected by chronic oral antibiotic treatment (Fig. 4C), which effectively reduced intestinal bacterial counts (Fig. 4D, Fig. S4A) and resulted in a severely enlarged caecum (data not shown). These data suggest that induction of systemic inflammation in Angptl4−/− mice by HFD is independent of the intestinal microbiota and is caused by dietary fat itself.
Remarkably, we noticed that the mesenteric lymph nodes (MLN) were dramatically enlarged in Angptl4−/− mice already after 5 weeks of HFD, indicating massive mesenteric lymphadenopathy/lymphadenitis (Fig. 4E,F). The inflammation extended to the mesenteric fat which exhibited mesenteric panniculitis (Fig S4B). High fat feeding is known to promote intestinal lymph flow and formation of chylomicrons, which pass through the MLN as chyle before reaching the circulation. Hence, MLN are exposed to extremely high TG concentrations, which we found could reach 55 mM in rats fed HFD (mean 35±11 mM). To investigate whether increased chyle flow is required for induction of the acute phase response and MLN enlargement, for 5 weeks mice were fed a diet rich in medium chain triglycerides (MCT) (Table S2), which are not processed via the lymph but directly enter the portal vein as free fatty acids. Remarkably, induction of plasma SAA and mesenteric lymphadenitis were absent in Angptl4−/− mice fed MCT (Fig. 4G,H), suggesting the response is mediated by chylomicrons. A safflower oil-based HFD, which as mentioned previously did not provoke a clinical phenotype, also did not promote inflammation in Angptl4−/− mice, whereas a lard-based high saturated fat diet give similar results as the palm-oil based HFD (Fig. 4G,H). Use of a palm oil-based diet with intermediate fat content revealed a clear correlation between the absolute saturated fat content of the diet and plasma SAA level in Angptl4−/− mice (Fig. 4I). Consistent with a direct pro-inflammatory effect of saturated fat via chylomicrons, there was a clear trend towards increased expression of several inflammatory mediators in MLN of Angptl4−/− mice already after one day of HFD (Fig. 4J, Fig. S4C). These data indicate that in Angptl4−/− mice, a diet rich in saturated fat rapidly causes severe mesenteric lymphadenitis and associated mesenteric panniculitis, which is mediated by chylomicrons, and which leads to a massive hepatic acute phase response via the connecting portal circulation.
MLN are densely packed with numerous immune cells including macrophages. Microscopic examination of MLN from Angptl4−/− mice but not WT mice fed HFD for 5 weeks showed an abundance of multinucleated Touton giant cells (Fig. 5A,B), which originate from fusion of aberrant lipid-laden tissue macrophages, as verified by F4/80 immunostaining (Fig. 5C). Formation of Touton cells is known to occur as a reaction to lipid material in lymph nodes and is characteristic of lipid lymphadenopathy (Aterman et al., 1988). Accumulation of neutral lipids in Touton cells was confirmed by Oil red O (Fig. 5D) and Sudan Black staining (Fig. 5E). Importantly, Touton giant cells were observed in Angptl4−/− mice already after 1 day of HFD (Fig. S4D). These data suggest a major role of MLN macrophages in initiating the inflammation in Angptl4−/− fed HFD.
Previously, TG-rich VLDL particles were shown to stimulate foam cell formation and provoke release of inflammatory cytokines by mouse peritoneal macrophages by serving as a source of pro-inflammatory saturated fatty acids (Gianturco et al., 1982; Saraswathi and Hasty, 2006). This effect required lipoprotein lipase (LPL), which is highly expressed in macrophages (Babaev et al., 1999; Ostlund-Lindqvist et al., 1983; Skarlatos et al., 1993) and lymph nodes (http://biogps.gnf.org/) (Lattin et al., 2008). Microarray analysis indicated that LPL was among the top 25 of genes with the highest microarray expression signal in peritoneal mouse macrophages (Table S3). Since Angptl4 is also expressed in macrophages, is dramatically induced by chyle (Fig. S5A), and can inhibit macrophage LPL (Fig. S5B), we hypothesized that Angptl4 may minimize lipolysis of chylomicrons by MLN-resident macrophages and accordingly suppress uptake of pro-inflammatory saturated fatty acids. To test this hypothesis, peritoneal macrophages from Angptl4−/− mice were incubated with chyle obtained from the mesenteric lymph duct of rats fed the palm oil-based HFD. Strikingly, chyle dramatically increased lipid storage in macrophages leading to formation of foam cells, which was strongly reduced by the LPL inhibitor orlistat (Fig. S5C). Increased lipid uptake was verified by increased expression of the PPAR-LXR target ABCA1 and decreased expression of SREBP targets, as shown by whole genome expression profiling (Fig. 6A) and qPCR (Fig. S5D). Importantly, increased lipid uptake and storage was associated with pronounced induction of numerous inflammation and immune-related genes, as shown by significant over-representation of Gene Ontology classes corresponding to those pathways and by Ingenuity Canonical Pathway analysis (Fig. S6A,B). Induction of inflammation, exemplified by Cxcl2, Gdf15,oncostatin M, Ptgs2 (COX-2) and numerous other genes, was almost entirely blunted by orlistat, indicating that lipid loading of macrophages by chyle and associated induction of inflammation are lipolysis and LPL-dependent (Fig. 6A). Some overlap in macrophage gene regulation was observed between chyle and LPS, but overall effects were mostly divergent (Fig. S6C). Specifically, typical targets of LPS such as IL-1β were not induced by chyle, while numerous inflammatory genes upregulated by chyle were not induced by LPS, including Gdf15 and Vegfa.
To examine whether Angptl4 can mimic the effect of orlistat on macrophage inflammation, macrophages were loaded with chyle in the presence of recombinant Angptl4. Angptl4 did not influence cell viability, which was equally high in Angptl4- and PBS-treated cells (Ϩ5%). Similar to orlistat, and at a concentration that was previously shown to cause maximal inhibition of LPL (Lichtenstein et al., 2007), Angptl4 prevented lipid uptake from chyle (Fig. 6B) and markedly reduced the inflammatory response (Fig. 6A,C), as shown by strongly blunted induction of inflammatory markers Ptgs2, Cxcl2, Ccr1 and Gdf15. The inhibitory effects of Angptl4 and orlistat on chyle-elicited changes in inflammatory gene expression were highly similar and support a common mechanism of action (Fig. 6A). No clear differences in foam cell formation upon chyle loading were observed between WT and Angptl4−/− macrophages (Fig. S5C), likely because induction of Angptl4 protein by chyle in WT macrophages and the subsequent feedback inhibition of lipid uptake via LPL lag behind the extremely rapid rate of lipid uptake.
To explore the potential mechanism that mediates the pro-inflammatory effect of chyle, peritoneal macrophages were incubated in the presence of different chemical modulators including an inhibitor of LPS-induced TLR4 signaling (Polymyxin B), sphingolipid biosynthesis (Myriocin), and fatty acid translocase/Cd36 (Sulfosuccinimidyl oleate) (Fig. 6D). None of the compounds showed any repressive effect on chyle-induced inflammation, although the LPS inhibitor polymyxin B augmented expression of Gdf15 and Cxcl2.
Since ER stress had emerged as one of the Canonical Pathways significantly altered by chyle (Table S6B), and has been intimately linked to activation of inflammation (Hotamisligil, 2010), we further focused our investigation on the ER stress pathway. The mammalian ER stress pathway consists of three major branches: IRE1a, PERK and ATF6. Upon ER stress activation, IRE1a and PERK undergo autophosphorylation and initiate downstream targets. IRE1α mediates the splicing of XBP1 mRNA while PERK phosphorylates eIF2α, leading to attenuation of global translation and induction of expression of Atf4 and CHOP (Ddit3). Remarkably, similar to the effect of known ER stress inducers thapsigargin and tunicamycin, chyle significantly increased total IRE1α protein and IRE1α phosphorylation (Fig. 7A), and dramatically stimulated XBP1 splicing (Fig. 7B). Moreover, chyle stimulated phosphorylation of PERK and its target eIF2α, and markedly increased CHOP protein. At the gene expression level, many ER stress target genes including XBP1s, CHOP, and Atf4 were significantly induced by chyle (Fig. 7C). Importantly and consistent with its anti-lipolytic role, the effects of chyle on ER stress marker genes were entirely abolished by Angptl4 (Fig. 7D). Thus, it can be concluded that chyle induces ER stress in macrophages, which may account for the pronounced activation of inflammation.
As mentioned above, chyle dramatically increased Angptl4 mRNA in wildtype peritoneal macrophages (Fig. 7E). Similarly, individual free fatty acids markedly increased Angptl4 mRNA in mouse peritoneal and human U937 macrophages (Fig. 7F). Compared to unsaturated oleic and linoleic acid, the saturated palmitic acid was significantly less potent in inducing Angptl4 expression. In contrast, expression of IRE1α, XBP1s, CHOP, Atf4,and Gdf15 and to a lesser extent Cxcl2 was specifically stimulated by palmitic acid (Fig. 7G). Palmitic, oleic and linoleic acid represent 95% of the fatty acid present in the various diets used, excluding the MCT diet.
Finally, to study which specific PPAR isotype may be involved in Angptl4 regulation by fatty acids, peritoneal mouse macrophages, mouse RAW264.7 macrophages and human U937 macrophages were treated with synthetic agonists for PPARα, PPARδ, and PPARγ. No induction of Angptl4 expression was observed with the PPARα agonist Wy14643 (Fig. S7A). The PPARδ agonist GW501516 consistently induced Angptl4 mRNA in all three cell types, whereas the PPARγ agonist rosiglitazone increased Angptl4 mRNA in peritoneal macrophages and to a minor extent in U937 cells. These data suggest that PPARδ most likely mediates the effect of chylomicron-derived fatty acids on Angptl4 expression, which is consistent with the established role of PPARδ as transcriptional sensor of the effect TG-rich lipoproteins in macrophages (Chawla et al., 2003).
Taken together, these results put forward a novel mechanism by which macrophages located in mesenteric lymph nodes are protected by Angptl4 from uncontrolled lipid accumulation after high fat feeding, thereby preventing lipid-induced ER stress and consequent inflammation.
After a saturated fat-rich meal, MLNs are exposed to extremely high concentrations of chylomicrons via the chyle, which might lead to generation of large amounts of pro-inflammatory saturated fatty acids upon TG lipolysis. Our data indicate that MLN and specifically resident macrophages are protected from the pro-inflammatory effect of saturated fatty acids via expression of Angptl4, which is strongly induced by chyle and fatty acids and which via inhibition of LPL prevents lipolysis of chylomicron-TG. In the absence of this protective autocrine mechanism, feeding a diet rich in saturated fat rapidly leads to enhanced lipid uptake into MLN-resident macrophages, triggering foam (Touton) cell formation and a massive inflammatory response characterized by severe mesenteric lymphadenitis. The concomitant induction of numerous cytokines leads to a massive hepatic acute phase response via the connecting portal circulation, which further evolves into a progressive, uncontrolled inflammation that culminates in fibrinopurulent peritonitis, chylous ascites, intestinal fibrosis and cachexia. The data thus show that Angptl4 is for a key player in the protection against the severe pro-inflammatory effects of dietary saturated fat. Based on our data in mice, it can be hypothesized that human subjects homozygous for the E40K mutation in Angptl4, which has reduced ability to inhibit LPL and is associated with lower plasma triglycerides (Romeo et al., 2007; Shan et al., 2008; Yin et al., 2009), may be particularly sensitive to the pro-inflammatory effects of dietary saturated fat.
According to our microarray analysis, LPL was among the most highly expressed genes in mouse peritoneal macrophages. The ability of macrophage LPL to facilitate lipid uptake into macrophages is well recognized (Babaev et al., 1999; Ostlund-Lindqvist et al., 1983). The locally released fatty acids may serve as energy source for active macrophages (Yin et al., 1997), but may also constitute a potential pro-inflammatory stimulus. Consistent with this notion, fatty acids offered to macrophages as VLDL-TG are taken up and engage MAPK-mediated inflammatory pathways along with increased expression of several pro-inflammatory cytokines (Saraswathi and Hasty, 2006). Our data indicate that exposure of macrophages to elevated yet physiologically relevant concentrations of chylomicrons containing saturated fatty acids unleashes a vast inflammatory response characterized by marked induction of numerous chemokines and other inflammation-related genes, which is entirely dependent on TG-lipolysis. Thus, lipolysis of TG-rich lipoproteins by macrophages is an important process that regulates intracellular fatty acid accumulation and contributes to initiation of pro-inflammatory signaling cascades. We propose that expression of Angptl4 in macrophages and its potent induction by chylomicron-derived fatty acids are part of a feedback mechanism aimed at protecting MLN-resident macrophages against post-prandial lipid overload and associated inflammation.
Ablation of Angptl4 is associated with decreased plasma TG levels caused by increased peripheral LPL activity (Koster et al., 2005). Recent data indicate that endothelium-bound LPL is stabilized by the protein GPIHBP1, which partially prevents LPL inhibition by Angptl4 (Sonnenburg et al., 2009). It can be hypothesized that the almost complete blockage of lipid uptake by Angptl4 in macrophages as opposed to its more modest effect in muscle and adipose tissue may be explained by the minimal expression of GPIHBP1 in macrophages (Fig. S7B) (Sonnenburg et al., 2009). Future studies will have to address this issue in more detail.
In our study, feeding Angptl4−/− mice a diet rich in polyunsaturated fatty acids did not elicit an inflammatory response, which is consistent with the data in peritoneal macrophages showing a lack of induction of Gdf15 and Cxcl2 by oleic and linoleic acid. In contrast, oleic and linoleic acid were much more potent inducers of Angptl4 expression compared to palmitic acid, suggesting that the Angptl4-mediated feedback inhibition of LPL-dependent fatty acid uptake and consequent suppression of inflammation is only weakly activated by saturated fatty acids.
An important question is how chyle induces inflammation in macrophages. Use of specific chemical inhibitors indicated that the response is not mediated by LPS, is not dependent on Cd36-mediated fatty acid transport, and does not require sphingolipid synthesis. Strikingly, we observed that chyle caused pronounced activation of different branches of the ER stress pathway in macrophages. It has been shown that ER stress can promote inflammation by various mechanisms, including via IRE1a-mediated activation of stress kinases such as the c-Jun N-terminal kinase (Urano et al., 2000), and via PERK-mediated activation of NF-κB (Jiang et al., 2003). We found that chyle stimulated IRE1α phosphorylation to promote XBP1 splicing, and activated PERK, eIF2α and their downstream targets. Activation of ER stress in peritoneal macrophages could be reproduced by free palmitic acid but not oleic acid or linoleic acid, suggesting the response to chyle is mediated by saturated fatty acids.
The mechanism by which saturated fatty acids induces ER stress has been the subject of recent investigations. Palmitate but not palmitoleate induced ER stress in pancreatic beta cells (Diakogiannaki et al., 2008). In liver cells saturated fatty acids induced ER stress independently of ceramide synthesis (Wei et al., 2006). Stimulation of ER stress by palmitate may occur via increasing the saturated lipid content of the ER membrane phospholipids and triglycerides, leading to compromised ER morphology and integrity and impaired function of protein-folding chaperones (Borradaile et al., 2006). Data also point to an important role for aP2 (Fabp4) in linking saturated fatty acids to ER stress in macrophages via alterations in lipid composition (Erbay et al., 2009).
Several studies have attributed the pro-inflammatory effect of saturated fatty acids to activation of TLR4 (Lee et al., 2001; Shi et al., 2006; Suganami et al., 2007). Recently, interplay between TLR4 (and TLR2) and the ER stress pathway was demonstrated. Specifically, IRE1α was shown to be a positive regulator of the inflammatory response to TLR activation in macrophages, while the PERK pathway was not induced by TLR signaling (Martinon et al., 2010). These data hint at a possible role for TLR signaling in the response to chyle in macrophages. However, unlike TLR signaling, chyle treatment dramatically induced ER stress as evidenced by the activation of ER stress sensors IRE1α and PERK as well as their downstream targets. Additionally, systematic whole genome analysis of gene regulation by chyle versus the TLR4 agonist LPS revealed some overlap, but chyle clearly did not mimic the effects of LPS, which is illustrated by the differential response of the classic LPS/TLR4-target IL-1β. Although these data do not rule out a role for TLR signaling in mediating the inflammatory effects of chyle, induction of ER stress seems to be a more plausible mechanism.
A previous report briefly eluded to the development of chylous ascites in Angptl4−/− mice after 20 weeks of HFD (Desai et al., 2007). In the same study it was found that repeated injections of WT mice fed HFD with a monoclonal antibody directed against the N-terminal portion of Angptl4 recapitulated the phenotype of Angptl4−/− mice. Since the antibody is directed against the N-terminal portion of Angptl4 and abolishes its ability to inhibit LPL, these data support the notion that the clinical abnormalities in Angptl4−/− mice fed HFD are related to altered LPL activity, and are independent of the signaling function of the C-terminal fragment of Angptl4.
(Chylous) ascites is a rare phenotype among transgenic mouse models. It has been observed in mice heterozygous for the transcription factor Prox1 as well as in mice lacking Angiopoietin-2. Both proteins are essential for development of the lymphatic vasculature (Gale et al., 2002; Harvey et al., 2005). Accordingly, it is tempting to hypothesize a similar role for Angptl4. However, it should be emphasized that the Prox1+/− and Angiopoietin2−/− mice develop chylous ascites shortly after birth, reflecting a severe developmental defect. In contrast, Angptl4−/− mice do not show any changes in lymphatic endothelial integrity and do not exhibit ascites unless challenged with HFD for at least 12 weeks. Rather, the data suggest that the ascites was secondary to progressive inflammation originating in the MLN macrophages, leading to massive lymphadenitis and consequent obstruction in mesenteric lymph flow, which in turn caused dilation of intestinal lymphatic vessels. Furthermore, inflammation of mesenteric lymp nodes and mesenteric fat led to increased local lymphatic and vascular permeability, as shown by chylous ascites and low SAAG, respectively, which is indicative of exudative ascites. The more than two-fold higher protein concentration in ascites fluid compared to chyle supports an important contribution of vascular leakage next to leakage from chyle. Increased circulatory leakage caused fibrinogen extravasation, which after clotting accumulated as fibrin and covered abdominal organs. Chronic inflammation likely gave rise to impaired intestinal barriers function and translocation of enteric bacteria, causing peritonitis which ultimately caused the death of the animals.
In conclusion, we demonstrate that Angptl4 protects against the severe pro-inflammatory effects of dietary saturated fat in MLN by inhibiting macrophage LPL, thereby reducing lipolytic release of fatty acids, macrophage foam cell formation, ER stress, and initiation of a marked inflammatory response. The data provide a clear illustration how the unique anatomy of intestinal lymphatic system, in which immune cells residing in mesenteric lymph nodes are exposed to excessive postprandial TG concentrations, requires the activation of an effective cellular mechanism that serves to protect against elevated lipid uptake and its complications. It can be speculated that the inability to effectively recruit this mechanism may contribute to pro-inflammatory changes related to elevated saturated fat consumption.
All animal studies were done using pure-bred WT and Angptl4 −/− mice on a C57Bl/6 background (Koster et al., 2005). In study 1, male 11-week old mice were fed a LFD or HFD for 8 or 19 weeks, providing 10 or 45% energy percent as triglycerides (D12450B or D12451, Research Diets, Inc., Table S1) (Research Diets Services, Wijk bij Duurstede, The Netherlands), after a 3 week run-in (adaptation period) with LFD. In study 2, male 10–18 week old mice were fed LFD or HFD for 5 weeks, after a 2 week run-in with LFD. The fat source of the HFD was either palm oil (standard HFD used in study 1, 3 and 4, Table S1), lard, MCT oil, or safflower oil (Table S2). Blood was collected from the tail vein at weekly intervals. In study 3, male 12-week old mice were fed standard HFD for 5 weeks, after a 2 week run-in with the standardized low fat diet AIN93G (see http://testdiet.purina-mills.com/PDF/57W5.pdf and Table S2). The following antibiotics were provided in the drinking water: ampicillin (1g/L), neomycin (1g/L), metronidazole (0.5g/L). Blood was collected from the tail vein at weekly intervals. In study 4, mice were fed low fat AIN93G or standard HFD for 24 hours, after a one week run-in with AIN93G. In the latter studies low fat AIN93G was chosen instead of D12450B to achieve minimal dietary saturated fat intake. The composition of the various diets used is provided in Table S1 and S2. At the end of each study, mice were anaesthetized with a mixture of isoflurane (1.5%), nitrous oxide (70%) and oxygen (30%). Blood was collected by orbital puncture into EDTA tubes. The mice were killed by cervical dislocation, after which tissues were excised and directly frozen in liquid nitrogen or prepared for histology.
Wistar rats were fed a palm-oil based high fat diet (D12451) overnight. The next morning, animals were anesthetized using isoflurane and their mesenteric lymph ducts were cannulated. Chyle was collected for 1–2 hours and stored at −20°C until usage. Chyle triglyceride concentrations averaged at 35 ± 11 mM as determined by an enzymatic assay (Instruchemie, Delfzijl, the Netherlands). The animal studies were approved by the Local Committee for Care and Use of Laboratory Animals at Wageningen University.
Two terminally ill animals were transferred to the Small Animal Pathology laboratory of the Faculty of Veterinary Medicine at Utrecht University for a formal autopsy by a licensed animal pathologist. A detailed macroscopic, microscopic and cytologic report was prepared.
U937 human monocytes were differentiated into macrophages by 24 hour treatment with phorbol myristate acetate (10 ng/µL). U937 macrophages were subsequently incubated for 6 hours with individual fatty acids (C16:0, C18:1, C18:2) coupled to fatty acid free BSA to a final concentration of 500 µM as previously described (de Vogel-van den Bosch et al., 2008).
To obtain peritoneal macrophages, wild-type and Angptl4−/− mice were injected intraperitoneally with 1 mL 4% thioglycollate. Three days later, the animals were anesthesized with isoflurane, bled via orbital puncture, and their peritoneal cavities washed using 10 mL ice-cold RPMI medium supplemented with 100 U/mL penicillin, and 100 µg/mL streptomycin (Lonza, Verviers, Belgium). Cell pellets were incubated with RBC lysis buffer on ice for 5 min and subsequently washed with RPMI medium supplemented with 10% fetal bovine serum (FBS) (Lonza, Verviers, Belgium) and antibiotics, re-pelletized and seeded at a density of 3×105 cells/cm2. Two hours later, the cells were washed twice with PBS to remove non-adherent cells and provided with medium. Two days later, the cells were exposed to chyle at a triglyceride concentration of 2 mM for 6 hours preceded by pre-incubation with either 20 µM Orlistat (Sigma Zwijndrecht, The Netherlands) or 2.5 µg/mL mouse recombinant Angptl4 (R&D Systems, Abingdon, United Kingdom) for 1.5 hours. To explore potential mechanisms through which chyle exert its effects, Angptl4−/− macrophages were pre-incubated with 10 µM myriocin, 10 µg/ml Polymyxin B or 0.5 mM Sulfosuccinimidyl oleate (SSO) for 30 min followed by exposure of the cells to either PBS or chyle at a triglyceride concentration of 2 mM for 6 hours. In ER stress experiments, Angptl4−/− macrophages were exposed to 100 nM thapsigargin, 2.5 µg/ml tunicamycin or chyle (triglyceride concentration 2 mM) for 6 hours. Analysis of ER stress in peritoneal macrophages was carried out as described (Yang et al., 2010).
Haematoxylin and Eosin staining of sections was performed using standard protocols. For detection of macrophages, immunohistochemistry was performed using antibody against Cd68 (liver, adipose tissue) or F4/80 (lymph nodes) (Serotec, Oxford, UK). Paraffin-embedded sections were pre-incubated with 20% normal goat serum followed by overnight incubation at 4°C with the primary antibody diluted 1:50 in PBS/ 1% Bovine Serum Albumin (BSA). After incubation with the primary antibody, a goat anti rat IgG conjugated to horseradish peroxidase (Serotec) was used as secondary antibody. Visualization of the complex was done using AEC Substrate Chromogen (Cd68) or 3,3'-Diaminobenzidine (F4/80). Negative controls were prepared by omitting the primary antibody.
For Sirius Red staining paraffin-embedded sections of the small intestine were mounted on Superfrost microscope slides. These sections were dewaxed in xylene and rehydrated in a series of graded alcohols. Slides were stained in picrosirius red 0.1% picric acid for 90 min and rinced in acidified H2O 0.5% acetic acid.
Oil red O stock solution was prepared by dissolving 0.5g Oil red O (Sigma, #O0625) in 500 mL isopropanol. Oil red O working solution was prepared by mixing 30 mL Oil red O stock with 20 mL dH2O, which was subsequently filtered. 5µm sections were cut from frozen MLNs embedded in OCT. Sections were air dried for 30 min, rehydrated in dH2O, and fixated for 10 min in formal calcium (4% formaldehyde, 1% CaCl2). Sections were immersed in Oil red O working solution for 10 min, followed by two rinses with dH2O. Haematoxylin Nuclei staining was subsequently carried out for 5 min followed by several rinses with dH2O. Sections were mounted in aqueous mountant (Imsol, Preston, UK).
For Sudan Black staining, 0.5 g Sudan black (Sigma, #86015) was dissolved in 100 mL warm 70% ethanol and subsequently filtered. Sections were fixed and rehydrated as above. After two 3 min rinses with 50% ethanol and two quick rinses with dH2O, sections were immersed in Sudan black solution for 10 min, followed by two rinses with dH2O. Sections were mounted in aqueous mountant (Imsol, Preston, UK).
We would like to thank Shohreh Keshtkar, Karin Mudde, Jenny Jansen, Mechteld Grootte-Bromhaar, Rinke Stienstra and Els Oosterink for laboratory analyses, Dr. Ben Witteman for clinical consultations, and Dr. Mark Boekschoten for assistance with microarray analysis. This work was financially supported by the Nutrigenomics Consortium, TI Food and Nutrition, the Netherlands Heart Foundation (2007B046), The Netherlands Organisation for Scientific Research (40-00812-98-08030), NIH (R01DK082582) and the American Diabetes Association 7-08-JF-47.
LL, FM, NdW, GH, YH and SK collected experimental data. LL, FM, NdW, GH, RvdM, YH, LQ, AK,JT, NST, MM, and SK participated in the design and interpretation of the study. LL and SK wrote the initial draft of the manuscript, which was corrected and approved by all authors.