|Home | About | Journals | Submit | Contact Us | Français|
Adult GPIHBP1-deficient mice (Gpihbp1−/−) have severe hypertriglyceridemia; however, the plasma triglyceride levels are only mildly elevated during the suckling phase when lipoprotein lipase (Lpl) is expressed at high levels in the liver. Lpl expression in the liver can be induced in adult mice with dietary cholesterol. We therefore hypothesized that plasma triglyceride levels in adult Gpihbp1−/− mice would be sensitive to cholesterol intake.
After 4–8 weeks on a western diet containing 0.15% cholesterol, plasma triglyceride levels in Gpihbp1−/− mice were 10,000–12,000 mg/dl. When 0.005% ezetimibe was added to the diet to block cholesterol absorption, Lpl expression in the liver was reduced significantly, and the plasma triglyceride levels were significantly higher (>15,000 mg/dl). We also assessed plasma triglyceride levels in Gpihbp1−/− mice fed western diets containing either high (1.3%) or low (0.05%) amounts of cholesterol. The high-cholesterol diet significantly increased Lpl expression in the liver and lowered plasma triglyceride levels.
Treatment of Gpihbp1−/− mice with ezetimibe lowers Lpl expression in the liver and increases plasma triglyceride levels. A high-cholesterol diet had the opposite effects. Thus, cholesterol intake modulates plasma triglyceride levels in Gpihbp1−/− mice.
Triglyceride-rich lipoproteins (chylomicrons and very low density lipoproteins) undergo lipolytic processing by lipoprotein lipase (LPL) within the lumen of capillaries, mainly in heart, skeletal muscle, and adipose tissue.1–3 A deficiency of LPL causes severe hypertriglyceridemia (chylomicronemia), both in humans and mice.2, 4, 5 Efficient lipolysis also depends on two other proteins, apolipoprotein (apo-) CII and GPIHBP1.2, 6, 7 Apo-CII is a cofactor for LPL,8 while GPIHBP1 is required for transporting LPL from the subendothelial spaces into the capillary lumen, where lipolysis occurs.7, 9–11 Defects in GPIHBP1 lead to chylomicronemia in both humans and mice.5–7, 12
Studies with Lpl and Gpihbp1 knockout mice (Lpl−/− and Gpihbp1−/−, respectively) have shown that LPL and GPIHBP1 are equally important for triglyceride hydrolysis in adult mice.4, 7 However, the phenotypes of Lpl−/− and Gpihbp1−/− mice differ significantly during the suckling phase.4, 7 Lpl−/− mice die within 24 h of birth with plasma triglyceride levels >10,000 mg/dl4 (unless they are rescued by transient production of LPL with an Lpl adenoviral vector13, 14). In contrast, suckling Gpihbp1−/− mice are healthy and display only mild increases in plasma triglyceride levels.7 By ~10 weeks of age, however, Gpihbp1−/− mice manifest plasma triglyceride levels of 3000–5000 mg/dl,7 even on a chow diet, similar to the plasma triglyceride levels in rescued adult Lpl−/− mice.13 The fact that plasma triglyceride levels in adult Gpihbp1−/− mice are similar to those in Lpl−/− mice implies that very little LPL-mediated lipolysis occurs in adult mice in the absence of GPIHBP1.
Why are the plasma triglyceride levels in Lpl−/− and Gpihbp1−/− mice so different during the suckling phase yet similar later in life? As noted by Beigneux et al.7, the mild phenotype of Gpihbp1−/− mice during the first few weeks of life is likely due to the fact that suckling mice synthesize large amounts of LPL in the liver.15 Beigneux et al.7 found that Lpl expression in suckling mouse livers is ~50-times higher than in livers of adult mice. The appearance of significant hypertriglyceridemia in Gpihbp1−/− mice coincides with the post-suckling decrease in Lpl expression in the liver.7
The finding that LPL in the liver might be catalytically active against plasma lipoproteins in the absence of GPIHBP1 is not particularly surprising. The fenestrated capillaries of the liver would allow access of plasma lipoproteins to LPL,16, 17 even when GPIHBP1 is absent. In other tissues, GPIHBP1 is required for transporting LPL from the interstitial spaces into the capillary lumen (where it hydrolyzes triglycerides in lipoproteins).11
Lpl expression in the livers of adult mice is extremely low, but Lpl transcripts can be detected with sensitive techniques (e.g., RT-PCR or northern blots).15, 18 Interestingly, Lpl transcripts in the mouse liver can be increased by more than 5-fold by feeding a high-cholesterol diet.18 The cholesterol-mediated upregulation in Lpl expression is mediated by the LXR nuclear hormone receptors. This regulation is negligible in mice lacking LXRα and is absent in mice lacking both LXRα and LXRβ.18
Because hepatic Lpl expression in adult mice can be modulated with cholesterol,18 we predicted that the plasma triglyceride levels in adult Gpihbp1−/− mice would be sensitive to changes in cholesterol intake. We hypothesized that reducing cholesterol uptake with ezetimibe would lower hepatic Lpl expression and increase plasma triglyceride levels, and further hypothesized that high levels of cholesterol in the diet would increase hepatic Lpl expression and lower plasma triglyceride levels. In the current study, we tested these hypotheses.
Gpihbp1−/− mice (>90% C57BL/6, <10% 129/Sv) have been described previously.7 Mice were housed in a specific pathogen–free barrier facility with a 12-h light/dark cycle. Genotyping was performed by PCR. The mutant Gpihbp1 allele was identified by amplifying a 208-bp DNA fragment with forward primer 5′–TCGCCTTCTTGACGAGTTCT–3′ and reverse primer 5′–GTTGAGGAGAGAGGAAGGCC–3′. The wild-type allele was identified by amplifying a 105-bp fragment with forward primer 5′–GAATAAACTTGAATGTCGTTTGCC–3′ and the identical reverse primer. All experiments were approved by UCLA’s Animal Research Committee.
Mice were maintained on a rodent chow diet (Purina, LabDiet #5001) containing 0.02% cholesterol. For some experiments, mice were fed a 20% fat (anhydrous milk fat), 0.15% cholesterol “western” diet, with or without 0.005% ezetimibe (Research Diets, New Brunswick, NJ). In other experiments, mice were fed western diets containing 21.2% fat (anhydrous milk fat) and either high (1.3%) or low (0.05%) amounts of cholesterol (TD #96121 and TD #05311, respectively, from Harlan Teklad, Madison, WI). All diets were stored at 4° C in vacuum-sealed containers.
Blood samples were obtained from anesthetized mice, and plasma lipid levels were measured with enzymatic kits. Neutral lipids in liver tissue were also measured with enzymatic kits. These procedures are described in the Supplemental Methods.
Whole-body lean tissue, fat, and free fluid mass were determined with a Bruker Optics (The Woodlands, TX) Minispec nuclear magnetic resonance (NMR) system, according to the manufacturer’s protocol. The lipid content of liver samples was determined with an EchoMRI 3-in-1 composition analyzer by Echo Medical Systems (Houston, TX). Hematoxylin and eosin–stained liver sections were reviewed in a blinded fashion by a veterinary pathologist in UCLA’s Department of Laboratory Animal Medicine.
Mice were individually housed in sealed chambers to monitor metabolic activity (Oxymax, Columbus Instruments, Columbus, Ohio). Animals were acclimated to the chambers for 8 h before data collection (consisting of three 12-h light/dark cycles). Mice were allowed free access to water by touch-activated sipper tubes, and food was available in a hopper attached to a scale. Physical activity was measured by infrared lasers in two dimensions. Data were compiled with OxyMax software and analyzed with Microsoft Excel by averaging readings taken every 20 min during each light/dark cycle.
Measurements of gene expression from RNA samples prepared from liver biopsies were performed by quantitative RT-PCR, as described in the Supplemental Methods.
Adenoviral vectors were prepared by ViraQuest (North Liberty, IA). A V5-tagged LPL adenovirus was generated by cloning the V5-tagged LPL cDNA into a shuttle plasmid (pVQAd CMV K-NpA), which was used to generate an E1A-deficient adenovirus under the control of a CMV promoter. The integrity of the LPL open reading frame in the adenovirus was confirmed by DNA sequencing. A V5-tagged β-galactosidase (LacZ) adenovirus was generated with the same techniques. All mice were injected intraperitoneally with gadolinium chloride (20 mg/kg) 24 h before the injection. Transaminase levels in the plasma four days after the adenovirus were measured in the clinical laboratory of UCLA’s Department of Laboratory Animal Medicine.
Expression of LPL and β-galactosidase was assessed by western blotting of liver extracts with an antibody against the V5 tag. Livers were homogenized in RIPA buffer; V5-tagged proteins were immunoprecipitated with a mouse monoclonal anti-V5 antibody (Invitrogen) and protein G–agarose beads (Roche). Following immunoprecipitation, proteins were separated by SDS–PAGE and then transferred to a nitrocellulose membrane for western blotting with a goat anti-V5 antibody (Abcam) followed by an IRDye800-labeled donkey anti-goat IgG secondary antibody (Rockland). Antibody binding was detected with an Odyssey infrared scanner (LiCor). Expression of β-galactosidase in the liver was also assessed by staining with an X-gal solution, as described in the Supplemental Methods.
Results are reported as mean ± standard error of the mean. Statistical significance was determined with a Student’s t-test (Microsoft Excel), ANOVA, or a paired t-test (Graphpad Prism). A Mann-Whitney test was used to analyze gene-expression data.
On a chow diet, adult Gpihbp1−/− mice have plasma triglyceride and cholesterol levels of ~3000–5000 mg/dl and ~300–500 mg/dl, respectively.7 When Gpihbp1−/− mice were switched from the chow diet to a “western” diet containing 20% milk fat and 0.15% cholesterol, the hyperlipidemia worsened. After one week on the western diet, the plasma triglycerides were 20,000–25,000 mg/dl, more than fourfold higher than baseline levels on a chow diet (Fig. 1A). Also, the plasma cholesterol levels increased to >1,700 mg/dl (Fig. 1B). At that time point, the whole blood of Gpihbp1−/− mice was pink; the plasma was cream-colored; and the blood vessels on the surface of the heart were white (Fig. I). After consuming the western diet for 4–8 weeks, the plasma triglyceride levels in Gpihbp1−/− mice fell to ~10,000–12,000 mg/dl (Fig. 1A).
Despite the severity of the hyperlipidemia, >90% of Gpihbp1−/− mice on the western diet appeared healthy, although they gained less weight than wild-type littermates (Gpihbp1+/+) (Fig. IIA) and exhibited reduced adiposity (Fig. IIB). These phenotypes were not due to ill health. Food intake was actually higher in Gpihbp1−/− mice (Fig. IIC), and activity levels were also increased (Fig. IID). A small percentage of Gpihbp1−/− mice stopped consuming food and water during the first 14 days on the western diet and were euthanized; those mice invariably had typhlitis (Fig. III).
To test the impact of cholesterol intake on plasma triglyceride levels and Lpl expression in the liver, we fed 8-week-old Gpihbp1−/− mice a western diet containing 0.005% ezetimibe (a dose that inhibits intestinal cholesterol absorption in mice19). We predicted that reduced cholesterol absorption would lower Lpl expression in the liver and lead to higher plasma triglyceride levels. Indeed, the plasma triglyceride levels were significantly higher in ezetimibe-treated Gpihbp1−/− mice (P = 0.0034, as judged by a repeated measures paired t test) (Fig. 1A). The plasma cholesterol levels were significantly lower in ezetimibe-treated Gpihbp1−/− mice at the 1-week time point (P < 0.001) and tended to be lower at the subsequent time points (Fig. 1B). At each time point, the ratio of triglycerides to cholesterol in the plasma was significantly higher (P = 0.0046) in ezetimibe-treated mice than in untreated mice (Fig 1C). The plasma triglycerides in wild-type (Gpihbp1+/+) mice consuming the western diet for 4 weeks were very low (~15 mg/dl); as expected, ezetimibe treatment had no effect on triglyceride levels (Fig. IV).
The higher triglyceride levels in ezetimibe-treated Gpihbp1−/− mice were accompanied by significantly lower levels of Lpl expression in the liver (measured after 13 weeks of diet) (Fig. 1D); however, ezetimibe had no effect on Lpl transcript levels in heart, skeletal muscle, and adipose tissue (Fig. 1E). As expected from earlier studies,20 ezetimibe treatment increased HMG-CoA reductase (Hmgcr) expression in the liver (Fig. 1D). In addition to reducing Lpl expression in the liver, ezetimibe reduced expression of three additional LXR-regulated genes in the liver—Abca1, Abcg5, and Abcg8 (Fig. 1F).
Our presumption was that LPL expression in the liver would lower plasma triglyceride levels, even in the absence of GPIHBP1, because the fenestrations in the sinusoidal capillaries would allow access of LPL to lipoproteins in the bloodstream. To test this idea, we injected either an adenovirus encoding either V5-tagged LPL (LPL–V5) or V5-tagged β-galactosidase (LacZ–V5) into Gpihbp1−/− and control mice. Within 2 days of injection, the plasma triglyceride levels in Gpihbp1−/− mice injected with the LPL–V5 adenovirus fell by 90% (P < 0.0001 compared to mice given the LacZ–V5 adenovirus) (Fig. 2A). This difference was also evident 4 h after delivering 75 μl of soybean oil by gavage (P < 0.0001). The plasma triglyceride levels in wild-type mice given the LPL–V5 adenovirus tended to be lower, but the changes were not statistically significant (Fig. 2B). In control experiments, we found that both adenoviruses yielded high levels of protein expression (Fig. VA–B). Neither adenovirus increased transaminase levels in the plasma.
To gain further insights into the relationship between cholesterol intake and plasma triglyceride levels in Gpihbp1−/− mice, we examined plasma lipid levels in 8-week-old Gpihbp1−/− mice consuming western diets containing either low (0.05%) or high (1.3%) levels of cholesterol. The plasma triglyceride levels in mice on the low-cholesterol diet were 20,000–25,000 mg/dl, higher than on the high-cholesterol diet (P = 0.0015). On the high-cholesterol diet, the triglyceride levels initially increased to ~15,000 mg/dl but then fell to ~6000–7000 mg/dl (Fig. 3A). At the 1-week time point, the plasma cholesterol levels in mice on the high-cholesterol diet were higher than in mice on the low-cholesterol diet (Fig. 3B). However, at the 2-, 3-, and 4-week time points, the cholesterol levels were lower in mice on the high-cholesterol diet (P = 0.0129, as judged by a repeated measures paired t test) (Fig. 3B). In wild-type mice, the high-cholesterol diet had little or no effect on plasma triglyceride levels (Fig. 3C) but resulted in slightly higher plasma cholesterol levels (P = 0.0086) (Fig. 3D).
The lower plasma triglyceride levels in Gpihbp1−/− mice on the high-cholesterol diet were accompanied by significantly higher Lpl expression levels in the liver after 5 weeks on the diet (Fig. 4A). Dietary cholesterol did not affect Lpl expression in heart, skeletal muscle, or adipose tissue (Fig. 4B). The high-cholesterol diet also increased expression of Abca1, Abcg5, and Abcg8 in the liver (Fig. 4C). The expression of Cyp7a1, an LXR-regulated gene21 that controls bile acid synthesis, was also increased on the high-cholesterol diet, but two FXR target genes, Bsep and Shp, were unaffected (Fig. VIA). The expression of Apoe was slightly but significantly increased in the livers of Gpihbp1−/− mice fed the high-cholesterol western diet (Fig. VIB).
We suspected that changing mice from a chow to a western diet would gradually increase Lpl expression levels in the liver [explaining the fall in plasma lipids that occurs in Gpihbp1−/− mice after one week on the western diet with 0.15% cholesterol (Fig. 1A) and the western diet with 1.3% cholesterol (Fig. 3A)]. Indeed, after three days on the western diet, the hepatic Lpl expression levels are higher than baseline levels on the chow diet (Fig. 5A). The hepatic Lpl expression levels were even higher after 21 days on the western diet (Fig. 5A).
We predicted that higher levels of Lpl expression in livers of Gpihbp1−/− mice on the high-cholesterol western diet might lead to increased lipid stores within the liver. Indeed, after only 5 weeks on diet, liver mass was significantly higher (P < 0.001) in Gpihbp1−/− mice on the high-cholesterol diet than in Gpihbp1−/− mice fed the low-cholesterol diet (Fig. 5B). Furthermore, as judged by nuclear magnetic resonance, the lipid content of the liver in Gpihbp1−/− mice on the high-cholesterol diet was significantly higher than in mice fed the low-cholesterol diet (Fig. 5C). In line with those findings, the livers of Gpihbp1−/− mice on the high-cholesterol diet contained higher levels of triglycerides (Fig. 5D) and cholesterol (Fig. 5E). The high-cholesterol diet was not accompanied by significant increases in the expression of acyl-CoA carboxylase (Acc), fatty acid synthase (Fas), and sterol regulatory element binding protein 1C (Srebp-1c), either in Gpihbp1−/− or wild-type mice (Fig. 5F).
Also, livers of Gpihbp1−/− mice maintained on the 1.3% cholesterol western diet for 5 weeks exhibited steatosis, as judged by hematoxylin and eosin staining (Fig. 6A). In contrast, mice fed the same western diet with 0.05% cholesterol had no evidence of steatosis (Fig. 6B). Gpihbp1−/− mice maintained on the 0.15% cholesterol western diet for three months also exhibited hepatic steatosis (Fig. 6C), but mice maintained on the same diet with 0.005% ezetimibe had no steatosis (Fig. 6D).
Hypertriglyceridemia in Gpihbp1−/− mice is mild during the suckling phase, when Lpl expression in the liver is high, but is severe in adult mice when Lpl expression in the liver is low. This observation led us to hypothesize that it would be possible to modulate the degree of hypertriglyceridemia in adult Gpihbp1−/− mice by altering hepatic Lpl expression levels. To test this hypothesis, we took advantage of an observation by Zhang and coworkers18—that Lpl expression in the mouse liver is sensitive to cholesterol intake. We began our studies by reducing cholesterol intake with ezetimibe, a drug that inhibits cholesterol absorption in the intestine.22–24 Compared with untreated Gpihbp1−/− mice, ezetimibe-treated Gpihbp1−/− mice had lower levels of Lpl expression in the liver and higher plasma triglyceride levels. Conversely, a high-cholesterol diet increased Lpl expression in the liver and lowered plasma triglyceride levels. Thus, reducing cholesterol absorption with ezetimibe increased plasma triglyceride levels in Gpihbp1−/− mice, while increasing cholesterol in the diet had the opposite effect.
Zhang et al.18 showed that cholesterol’s impact on hepatic Lpl expression is modulated by the LXR transcription factors. They found that Lpl transcripts in wild-type mice are increased by LXR agonists and dietary cholesterol, but that cholesterol had no effect on hepatic Lpl expression in mice lacking LXRα and LXRβ. Our results were consistent with LXR-mediated changes in Lpl expression. In addition to Lpl, we found that three other LXR-responsive genes in the liver (Abca1, Abcg5, and Abcg825, 26) were downregulated by ezetimibe treatment. As expected, ezetimibe increased transcripts for HMG-CoA reductase, a cholesterol biosynthetic enzyme.27, 28 In contrast, Lpl, Abca1, Abcg5, and Abcg8 expression levels were increased by the high-cholesterol diet in Gpihbp1−/− mice. The effects of ezetimibe and dietary cholesterol on Lpl expression were confined to the liver—there were no significant changes in Lpl expression in adipose tissue, heart, or skeletal muscle.
The studies by Zhang et al.18 focused on the effects of cholesterol and LXR agonists on Lpl transcript levels and did not explore the impact of altered hepatic Lpl expression levels on plasma lipids. This is understandable; discerning the effects of altered hepatic Lpl levels would be difficult in wild-type mice, where the vast majority of LPL-mediated lipoprotein processing occurs in extrahepatic tissues. We believe that Gpihbp1 deficiency constitutes a sensitized setting for uncovering physiologic consequences of altered Lpl expression in the liver. In the setting of GPIHBP1 deficiency, LPL secreted by myocytes and adipocytes remains mislocalized in the interstitial spaces and never reaches the capillary lumen.11 Thus, the LPL in peripheral tissues of Gpihbp1−/− mice has no access to triglyceride-rich lipoproteins in the bloodstream (explaining the severe hypertriglyceridemia in those mice). However, the liver has fenestrated capillaries,16, 17 so the LPL produced by hepatocytes has ready access to plasma lipoproteins—even when GPIHBP1-mediated transendothelial LPL transport is absent.
We observed no effect of dietary cholesterol on plasma triglyceride levels in Gpihbp1+/+ mice, even though Lpl expression was similarly induced in those animals. The lack of a change in plasma triglyceride levels in the wild-type mice is not surprising. In Gpihbp1+/+ mice, lipolytic processing of triglyceride-rich lipoproteins is robust in peripheral tissues, making it virtually impossible to detect the effects of LPL activity in the liver. The experiments with adenoviral expression of LPL supported this interpretation. The LPL adenovirus did not have a significant impact on triglyceride levels in wild-type mice, where lipolysis is robust in peripheral tissues, but markedly lowered plasma triglyceride levels in Gpihbp1−/− mice, where lipolysis in peripheral tissues is minimal or absent.
In Gpihbp1−/− mice, Lpl expression levels in the liver were inversely correlated with plasma triglyceride levels. Low hepatic Lpl expression levels during ezetimibe therapy were associated with higher plasma triglyceride levels, while higher Lpl expression levels on the high-cholesterol diet were associated with lower triglyceride levels. It is likely that this association was causal in nature because high hepatic Lpl levels led to hepatic steatosis, while low hepatic Lpl expression levels protected the liver from steatosis. Thus, Lpl expression in the liver of Gpihbp1−/− mice reduces triglycerides in the plasma at the expense of increased triglyceride accumulation in the liver. An alternative explanation would be that the hepatic steatosis in Gpihbp1−/− mice on the high-cholesterol diet was due to an LXR-driven increase in hepatic lipogenesis. Against that possibility, however, was our inability to detect increased SREBP-1c, Fas, or Acc expression in the livers of Gpihbp1−/− mice on the high-cholesterol diet.
In summary, we show that modulating cholesterol intake in Gpihbp1−/− mice changes Lpl expression levels in the liver, resulting in significant changes in plasma triglyceride levels. The effects of hepatic Lpl expression were quite apparent in Gpihbp1−/− mice, where LPL entry into capillaries of peripheral tissues is compromised. These dietary effects could be important in humans with chylomicronemia due to GPIHBP1 mutations5, 6, 12 or with acquired defects in GPIHBP1-mediated LPL transport.
Sources of Funding
This work was supported by a Scientist Development Award from the American Heart Association’s National Office (0735026N, to APB), a research grant from Merck-Schering Plough, a Ruth L. Kirschstein NRSA T32HL69766 (to MMW), a postdoctoral fellowship awards from the American Heart Association, Western States Affiliate (to BSJD and PG), 5P01HL090553 (to SGY), HL087228 (to SGY), and HL094732 (to APB), and Texas AgrilLife Research Project # 8738 (to RLW).
We thank Miklos Peterfy for the use of the NMR spectrometer.