|Home | About | Journals | Submit | Contact Us | Français|
Niemann-Pick C1-like 1 protein (NPC1L1), a transporter crucial in intestinal cholesterol absorption, is expressed in human liver but not in murine liver. To elucidate the role of hepatic NPC1L1 on lipid metabolism, we overexpressed NPC1L1 in murine liver utilizing adenovirus-mediated gene transfer. C57BL/6 mice, fed on normal chow with or without ezetimibe, were injected with NPC1L1 adenovirus (L1-mice) or control virus (Null-mice), and lipid analyses were performed five days after the injection. The plasma cholesterol levels increased in L1-mice, and FPLC analyses revealed increased cholesterol contents in large HDL lipoprotein fractions. These fractions, which showed α-mobility on agarose electrophoresis, were rich in apoE and free cholesterol. These lipoprotein changes were partially inhibited by ezetimibe treatment and were not observed in apoE-deficient mice. In addition, plasma and VLDL triglyceride (TG) levels decreased in L1-mice. The expression of microsomal triglyceride transfer protein (MTP) was markedly decreased in L1-mice, accompanied by the reduced protein levels of forkhead box protein O1 (FoxO1). These changes were not observed in mice with increased hepatic de novo cholesterol synthesis. These data demonstrate that cholesterol absorbed through NPC1L1 plays a distinct role in cellular and plasma lipid metabolism, such as the appearance of apoE-rich lipoproteins and the diminished VLDL-TG secretion.
Dyslipidemia has proved to be a major risk factor for cardiovascular diseases, and many lipid-lowering agents have been demonstrated to reduce the incidence of cardiovascular diseases. Ezetimibe is a recent lipid-lowering agent known to inhibit cholesterol absorption from the intestine (1). The main clinical benefit of ezetimibe is to lower LDL cholesterol levels (2–5); however, there have been reports of other beneficial effects of ezetimibe, such as lowering plasma triglycerides levels (3, 5, 6).
Niemann-pick C1-like 1 protein (NPC1L1) (7, 8), a target protein of ezetimibe, exists in the plasma membrane and early endosome and plays a key role in the intestinal absorption of free cholesterol (9–11). The cholesterol entering the plasma membrane is detected by NPC1L1 and internalized in the endocytic recycling compartment. Interestingly, the difference in the expression levels of NPC1L1 among various tissues differ among species; in humans, the level of NPC1L1 expression in the liver is almost the same as that of the intestine; however, there is little hepatic NPC1L1 expression in rodents, especially in mice (7). Previous work by Temel et al., which utilized transgenic mice stably expressing NPC1L1 in the liver, demonstrated that hepatic NPC1L1 played a critical role in the absorption of cholesterol from the bile (12). This observation enriched our knowledge regarding the sources of cholesterol in the human liver; besides the de novo synthesis of cholesterol and cholesterol supply from the sinusoidal side of the cells through lipoproteins, human hepatocytes acquire cholesterol from the bile duct canaliculi as well.
Several lines of evidence have shown that hepatic cholesterol derived from different sources has undergone different metabolic pathways; for example, cholesterol taken up through scavenger receptor class B, type I (SR-BI) from HDL particles is believed to be preferentially secreted in the bile (13, 14), whereas cholesterol newly synthesized or acquired via LDL receptors is secreted into the blood circulation or stored in the cells. Thus, it is plausible to hypothesize that cholesterol acquired through NPC1L1 from the canalicular side of the hepatocytes undergoes a different pathway from that of other origins and may possess novel metabolic properties on lipid and lipoprotein metabolism.
In this study, we examined the influence of hepatic NPC1L1 expression on lipoprotein metabolism utilizing adenovirus-mediated gene transfer. Stable inbred expression of a target protein utilizing transgenic animals would result in the secondary metabolic changes to maintain homeostasis and, in doing so, obscure the genuine metabolic effects of the target protein. On the other hand, abrupt and adventitious protein expression using an adenoviral-vector enabled us to examine the metabolic effect of the expressed protein with minimal secondary metabolic changes. Through this study, we found two interesting effects on lipoprotein metabolism with hepatic NPC1L1 expression; namely, we observed the emergence of apoE-rich lipoprotein (ERL) as well as a decrease in serum VLDL-triglyceride (TG). Besides this observation, we could clarify that cholesterol absorbed through NPC1L1 might possess physiological properties that are distinct from cholesterol derived from other origins.
Human NPC1L1 and LSS cDNAs were cloned from the cDNA library of human liver (Takara Bio Inc., Shiga, Japan) and peripheral blood mononuclear cells, respectively. Human NPC1L1 adenovirus (Ad-L1) and human LSS adenovirus (Ad-LSS) were constructed with AdEasy system (Stratagene, La Jolla, CA) and purified through CsCl gradient centrifugation. Control adenovirus (Ad-Null), which lacks encoded cDNA, was also constructed.
For the in vivo analyses, adenoviral vectors were administered with a dose of 1 × 1011 particles (6.5 × 109 pfu) via the tail vein into mice. For the in vitro experiments, the cells were infected with adenoviruses at the MOI of 25.
HepG2 cells, purchased from American Type Culture Collection (ATCC; Manassas, VA), were cultured in DMEM (Sigma-Aldrich Co.) supplemented with 10% fetal bovine serum (FBS; Gibco BRL, Eggstein, Germany) and 1% penicillin/streptomycin (Gibco BRL).
To analyze the effects of NPC1L1 expression in HepG2 cells, 2 days after the viral infection, the medium was replaced with serum-free, bile micelle-containing DMEM. Bile micelle was prepared as described previously (15); sodium taurocholate (Sigma-Aldrich Co.) dissolved in 96% ethanol, egg-yolk phosphatidylcholine (Sigma-Aldrich Co.) dissolved in methanol, and cholesterol (WAKO Pure Chemical Industries, Osaka, Japan) dissolved in ethanol were briefly mixed together, evaporated under nitrogen gas flow, stirred in DMEM for 6 h at 37°C, and filtered before use. The final concentrations of sodium taurocholate, egg-yolk phosphatidylcholine, and cholesterol were 2 mM, 40 μM, and 0.1 mM, respectively.
For the analysis of the effects of cellular de novo cholesterol synthesis on the FoxO1 protein level, the medium was replaced with serum-free DMEM containing 10 μM pitavastatin (Chemtech Labo., Inc., Tokyo, Japan) with or without 30 mM mevalonate (Sigma-Aldrich Co.). The cells were subjected to analyses 24 h after changing the medium.
C57BL/6 mice and apoE-deficient mice were purchased from CLEA Japan (Tokyo, Japan) and Sankyo Lab Service Co. (Tokyo, Japan), respectively. Ezetimibe-containing chow (10 mg/100 g food) (16) was prepared by mixing Zetia tablets (Schering, Kenilworth, NJ) dissolved in carboxymethyl cellulose (Sigma-Aldrich Co.) with normal chow (Oriental Yeast Co, Tokyo, Japan). Six-week-old mice were fed with normal chow containing ezetimibe [EZ(+)] or vehicle [EZ(−)] for 3 weeks, after which they were injected with adenoviral vectors. Animal experiments were approved by the animal committee within the University of Tokyo.
Liver samples of mice infected with Ad-L1 or Ad-Null were fixed with 4% paraformaldehyde, sectioned into 5 μm thick slices, and processed for immunohistochemistry utilizing rabbit anti-NPC1L1 antibody (Cayman Chemical Co, Ann Arbor, MI) as a primary antibody and peroxidase-labeled polymer of goat anti-rabbit Ig antibody (Envision system, Dako, Glostrup, Denmark) as the secondary antibody. Color development was performed with DAB, and the samples were counterstained by hematoxylin.
The bile was harvested from the gall bladders of anesthetized mice and mixed with methanol and chloroform. The aqueous phase was utilized to assess total bile acids using enzymatic methods (WAKO Pure Chemical Industries); the organic phase was designed to assess phospholipids through enzymatic methods (WAKO Pure Chemical Industries) and cholesterol using HPLC as described previously (17).
Five days after the injection of adenoviruses, the mice were fasted for 6 h, and blood samples were collected. Total and free cholesterol, triglycerides, and phospholipids were measured with enzymatic methods (WAKO Pure Chemical Industries). The values of esterified sterol were calculated by subtracting free sterol values from total sterol values and then multiplying by 1.67. The plasma samples were also subjected to Western blot analyses with anti-apoA-I antibody, anti-apoE antibody (Chemicon International Inc., Temecula, CA), and anti-apoB antibody (Calbiochem Inc., La Jolla, CA), after which plasma apolipoprotein levels were calculated with densitometric analyses of the blots.
Plasma samples were pooled and separated by FPLC utilizing Superose 6 column. The pooled plasma and several FPLC fractions were further separated by agarose electrophoresis with universal gel/8 (Helena Laboratory, Saitama, Japan). The agarose gels were stained with Fat Red 7B (Sigma-Aldrich Co.) to analyze the lipoprotein distribution; for the analyses, proteins on the gels were transferred to nitrocellulose membranes, followed by immunoblotting as described below.
Cellular and hepatic lipid was extracted with methanol and chloroform, and the cholesterol and triglycerides levels were measured with enzymatic methods. The levels of cholesterol and triglycerides were adjusted with the cellular or hepatic protein levels.
For the preparation of whole cellular proteins, murine livers and HepG2 cells were homogenized in RIPA lysis buffer (Santa Cruz Biotechnology, Santa Cruz, CA). For the preparation of membranous protein, murine livers and HepG2 cells were homogenized in 5 mM Tris-HCl (pH 7.5) solution containing 1 mM PMSF and protease inhibitor cocktail (Roche, Mannheim, Germany) and centrifuged at 700 g for 15 min three times. The supernatants were centrifuged at 444,000 g for 60 min, and the pellets were suspended in 50 mM Tris-HCl (pH7.5) solution containing 1% TritonX-100, 5 mM EDTA, 10 mM EGTA, and 1 mM PMSF. The nuclear fractions were prepared as described previously (18).
Protein analyses with Western blot were performed using the following antibodies; anti-Erk and anti-phospho Erk antibody (BD Bioscience, San Jose, CA); anti-FoxO1 (Ser256) antibody (GenScript Co., Piskataway, NJ); anti-β-actin antibody (MBL, Nagoya, Japan); anti-NPC1L1 and anti-LDL receptor antibody (Cayman Chemical Co., Ann Arbor, MI); anti-ABCA1 antibody (Novus Biologicals, Inc., Littleton, CO); anti-pan cadherin antibody (Thermo Fisher Scientific Inc., Fremont, CA); and anti-SREBP-1, SREBP-2, and laminA/C antibody (Santa Cruz Biotechnology).
Total RNAs extracted from murine livers with GenElute mammalian total RNA miniprep kit (Sigma-Aldrich) were subjected to reverse transcription with Superscript II enzyme (Invitrogen Co.). Real-time quantitative PCR was performed with LightCycler system (Roche Diagnostics, Basel, Switzerland). Hybridization probes and primers were obtained from Nihon Gene Research Laboratories, Inc. (Sendai, Japan). The expression levels of the genes of interest were adjusted to those of the endogenous control GAPDH mRNA.
Five days after the adenovirus injection, mice were fasted for 5 h, and then injected with a 500 mg/kg body weight dose of tyloxapol (Sigma-Aldrich Co.) (19). Blood samples were collected at 0, 40, 80, and 120 min and measured for triglycerides.
The results were expressed as mean ± SEM. Differences between two groups were evaluated with Student t-test, and the differences among more than two groups were assessed using one-way ANOVA, followed by multiple comparison tests. P-values less than 0.05 were deemed as statistically significant.
To demonstrate that NPC1L1 expression with human NPC1L1 adenovirus (Ad-L1) increased the uptake of cholesterol in the bile micelle, HepG2 cells infected with Ad-L1 or control adenovirus (Ad-Null) were loaded with bile micelles, and the cellular cholesterol contents were evaluated. As shown in Fig. 1, cellular infection of Ad-L1 increased NPC1L1 protein contents (Fig. 1A) and resulted in a significant increase in cellular free cholesterol, which was inhibited by ezetimibe treatment (Fig. 1B). Incubation of the cells for a longer period resulted in a significant increase in cellular cholesteryl ester content as well, which would be a consequence of the esterification of free cholesterol (supplementary Fig. I). These results indicated that NPC1L1, expressed with Ad-L1, absorbed free cholesterol from the bile micelles.
We overexpressed NPC1L1 protein in C57BL/6 mice and evaluated its effect on the lipid contents in the bile. Western blot analysis of the liver samples confirmed that the administration of Ad-L1 into C57BL/6 mice resulted in the increase of NPC1L1 protein in the liver (Fig. 1C). Furthermore, immunohistochemical analysis of the liver specimens showed that the overexpressed NPC1L1 protein was mainly localized to the bile canalicular surface (Fig. 1D, E), which was concordant with the NPC1L1 localization in human liver.
Analysis of the lipid and bile acid contents of the bile showed a significant and drastic decrease in the biliary cholesterol levels with no changes in concentration of bile acids (Fig. 1F), indicating that overexpressed NPC1L1 protein localized to the canalicular side had absorbed cholesterol from the bile. Concordant with this observation, hepatic cholesterol contents increased almost in parallel with hepatic NPC1L1 expression (supplementary Fig. II). The biliary phospholipid concentrations also decreased by as much as 50% (Fig. 1F) with the overexpression of NPC1L1.
We next analyzed the effects of hepatic NPC1L1 expression on plasma lipoprotein metabolism. As shown in Fig. 2A, hepatic NPC1L1 expression dramatically increased plasma total cholesterol levels (96 mg/dl versus 155 mg/dl, P < 0.01); this increase was partially inhibited by ezetimibe treatment [EZ(+)]. Plasma phospholipids were also increased by hepatic NPC1L1 expression (supplementary Fig. III-A). As for plasma apolipoprotein levels, hepatic NPC1L1 expression increased plasma apoE, apoB100, and apoB48; however, no difference was observed in plasma apoA-I (Fig. 2B, supplementary Fig. III-B, C).
To analyze more extensively the lipoprotein changes induced by NPC1L1 overexpression, pooled plasma samples were subjected to fast protein liquid chromatography (FPLC) analysis. As shown in Fig. 2C, the increase in plasma cholesterol levels found in mice administered Ad-L1 (L1-mice) was due to the increase in those of LDL fractions (fractions 30–34) and of fractions whose particle sizes ranged between those of LDL and HDL (fractions 36–40). The increase in cholesterol levels in these fractions was partially reversed by ezetimibe treatment; however, in mice injected with Ad-Null (Null-mice), treatment with ezetimibe did not affect the FPLC cholesterol profile, suggesting that the lipoprotein changes found in L1-mice emerged as a result of hepatic NPC1L1 expression. Changes in cholesterol levels were not observed in VLDL fractions (fractions 22–24) and HDL fractions (fractions 41–45). Analysis of free cholesterol levels among FPLC fractions revealed that fractions 36–40 of L1-mice were markedly rich in free cholesterol; specifically, about 50% of cholesterol was free cholesterol (supplementary Fig. III-D). The phospholipid contents of fractions 36–40 of L1-mice increased as well (supplementary Fig. III-E). Western blot analyses of FPLC fractions demonstrated that apoE protein levels increased markedly in fractions 36–40, indicating that these fractions were apoE-rich lipoproteins (ERL) (Fig. 2D). ApoE protein was also abundant in fractions 30–34, implying that fractions 30–34 were composed of not only LDL particles but also ERLs.
Previous studies have suggested that several classes of lipoproteins, such as pre-β1, pre-β2, or α-migrating lipoproteins (20, 21), possess the potency to represent ERL. Thus, to characterize the ERLs found in L1-mice, plasma and FPLC samples were subjected to agarose electrophoresis. Agarose electrophoresis of the plasma revealed that the lipid contents of pre-β lipoproteins were decreased in L1-mice (Fig. 2E), suggesting decreased neutral lipids in VLDL fractions. FPLC fractions 30–32 and fractions 41–43 showed β- and α-mobility on agarose electrophoresis, respectively, confirming that they were LDL and HDL fractions (Fig. 2F). Lipoproteins of fractions 37–38, which were abundant in apoE, showed α-mobility (Fig. 2F); furthermore, Western blot analysis of the agarose gel electrophoresed with fractions 37–38 revealed the existence of apoE exclusively in the α-position, with no detection of apoA-I and apoB proteins (Fig. 2G). Thus, these results demonstrated that ERLs found in L1-mice were homogeneous and had α-mobility on agarose electrophoresis, which could be classified as large HDL particles.
To elucidate the mechanism behind the appearance of ERL in L1-mice, we first examined whether Ad-L1 administration affected the intestinal sterol absorption. As shown in supplementary Fig. IV, absorption of sterol from intestine of L1-mice was almost the same as that of Null-mice, suggesting that the appearance of ERL was not due to the modulation in intestinal cholesterol absorption in L1-mice. Ezetimibe treatment in L1-mice understandably suppressed sterol absorption significantly (supplementary Fig. IV), which might be associated with the reduced plasma total cholesterol levels in EZ-treated L1-mice. Second, to clarify that ERLs were produced directly from hepatocytes, the media of HepG2 cells treated with Ad-L1 together with bile micelles were subjected to FPLC analyses. As shown in supplementary Fig. V, apoE protein in the media was distributed in fractions 36–40 as well as in other lipoprotein fractions, confirming that ERLs were directly produced from L1-expressing HepG2 cells. Furthermore, ezetimibe treatment efficiently suppressed apoE levels in fractions between LDL and HDL, suggesting that cholesterol absorbed through NPC1L1 would be secreted dominantly as ERL.
We next analyzed the factors believed to be relevant to ERL production in the liver. One of the candidates is the increased ATP binding cassette transporter A1 (ABCA1) protein levels, which has been proposed in two different lines of studies (12, 22); however, ABCA1 protein level was not altered in Ad-L1 mice (supplementary Fig. VI-A). Upregulation of hepatic apoE, which has been shown to be associated with increased ERL production in HepG2 spheroids when treated with a LXR agonist (18), was the other candidate; however, hepatic apoE protein and mRNA levels were unaltered (supplementary Fig. VI-B, C). Thus, hepatic ABCA1 and apoE expression levels were not the key factors behind ERL production.
As shown in Fig. 2H, the protein levels of LDL receptor decreased in L1-mice, which would have led to an increase in plasma apoB100 levels due to the decreased clearance of LDL particles (supplementary Fig. III-C). Furthermore, the clearance of chylomicron remnants, analyzed by retinol palmitate test, was also retarded in L1-mice (supplementary Fig. VII), suggesting the reduced levels of remnant receptors. Thus, it is plausible that the clearance of ERLs might also be retarded due to the reduced levels of receptors associated with lipoprotein clearance, which might contribute to the appearance of ERLs.
To investigate the role of apoE in the production of large HDL and elucidate the physiological effect of ERL, hepatic expression of NPC1L1 was performed in apoE-deficient mice. Overexpression of NPC1L1 in apoE-deficient mice resulted in a significant increase in plasma free cholesterol (Fig. 3A) and phospholipid levels (data not shown) as was observed in C57BL/6 mice. FPLC analysis of plasma samples, however, showed a distinct difference from that found in wild-type mice; large HDL particles, which corresponded to fractions 36–40, did not emerge in apoE-deficient mice (Fig. 3B, C), suggesting that the apoE protein was indispensable for the production of apoE-rich large HDL particles found in wild-type mice. In addition, FPLC analysis revealed that the increased free cholesterol was mainly due to the increased free cholesterol in VLDL fractions.
The analyses of hepatic lipids in apoE-deficient mice revealed a significant increase in both free cholesterol and cholesteryl ester contents with NPC1L1 expression (Fig. 3D), whereas those increases were not significant in wild-type mice. In addition, increased plasma glutamic pyruvic transaminase (GPT) levels and prominent infiltration of neutrophils in the liver specimens were found in apoE-deficient mice with NPC1L1 expression (Fig. 3E, supplementary Fig. VIII), findings not observed in C57BL/6 mice. These results suggested that ERL facilitates the secretion of free cholesterol into the circulation, thereby reducing the hepatic injury that can result from excessive accumulation of free cholesterol in the liver.
As described above, the neutral lipid content of pre-β lipoproteins was reduced in wild-type mice expressing NPC1L1 (Fig. 2E), whereas no reduction in VLDL cholesterol was observed (Fig. 2C). Thus, we next analyzed the changes in triglyceride metabolism induced by NPC1L1 expression. As shown in Fig. 4A, plasma triglyceride levels were decreased in L1-mice, and FPLC analysis revealed that this decrease was attributed to a reduction in VLDL-TG contents (Fig. 4B). These triglyceride changes were also observed in apoE-deficient mice (supplementary Fig. IX-A, B).
To illustrate the underlying mechanism behind the decrease in VLDL-TG levels, a hepatic VLDL-TG production assay was performed utilizing tyloxapol, and the reduced production of VLDL-TG was confirmed in L1-mice (Fig. 4C). One of the key molecules involved in the maturation and secretion of VLDL is the microsomal triglyceride transfer protein (MTP); consequently, we analyzed hepatic MTP expression. Compared with Null-mice without ezetimibe treatment [EZ(−)], the expression of MTP decreased by 67% in L1-EZ(−), and MTP mRNA levels in Null-EZ(+) mice were also significantly suppressed (Fig. 4D). In apoE-deficient mice infected with Ad-L1, hepatic MTP expression was also suppressed (supplementary Fig. IX-C). The decreased VLDL-TG secretion might increase hepatic triglyceride content; however, as shown in Fig. 4F, no significant difference was observed in this content. Concordant with this observation, the expression of fatty acid synthase (FAS), an important enzyme for lipogenesis, was suppressed in L1-mice (Fig. 4E).
Another plausible mechanism that affects hepatic VLDL production is the increment of hepatic endoreticulum (ER) stress; a previous study indicated that hepatic ER stress destabilized apoB (23), leading to a decreased hepatic production of VLDL. Furthermore, loading of free cholesterol in some cells has been known to evoke ER stress. However, in our study, examination of X-box binding protein-1 (XBP-1) splicing and Grp78 expression, which should be modulated by hepatic ER stress, did not show any changes in these markers with NPC1L1 expression (Fig. 4G, H).
Because previous studies have indicated that hepatic MTP expression was negatively regulated by SREBP-1, SREBP-2 (24), and ERK activation (25) and positively regulated by forkhead box protein O1 (FoxO1) protein (26), we analyzed these factors. As shown in Fig. 5A, B, the nuclear protein level of SREBP-1 was not altered and that of SREBP-2 tended to be reduced in L1-mice, and no appreciable difference was observed in the extent of ERK phosphorylation between L1-EZ(–) mice and Null-EZ(–) mice. These results indicated that SREBPs and ERK might not be involved in the suppression of MTP in L1-mice. The levels of ERK phosphorylation tended to increase in mice treated with ezetimibe, suggesting that ERK activation may be one of the factors suppressing MTP in mice treated with ezetimibe. On the other hand, the FoxO1 protein level was markedly reduced in L1-mice, and this reduction was partially reversed by ezetimibe treatment (Fig. 5B), suggesting that decreased VLDL-TG production with hepatic NPC1L1 expression can be attributed to the suppression of FoxO1 protein levels.
As described above, hepatic expression of NPC1L1 resulted in the appearance of ERL and decreased VLDL-TG production. Next, we investigated whether these metabolic changes would also be observed in experimental models of enhanced hepatic de novo cholesterol synthesis by overexpressing lanosterol synthase (LSS), an enzyme involved in cholesterol synthesis. Overexpression of LSS increased hepatic cholesterol ester contents, plasma apoB levels, and LDL cholesterol levels, accompanied by the decreased LDL receptor levels; however, ERLs were not found and no changes were observed in FoxO1 protein levels and VLDL-TG production rates (Fig. 6A, B, supplementary Fig. X).
We also examined whether the cholesterol absorbed through NPC1L1 possessed different properties compared with that synthesized de novo, utilizing HepG2 cells as in vitro models. As shown in Fig. 6C, NPC1L1 expression reduced FoxO1 protein levels in HepG2 cells, a result in accordance with that found in in vivo analysis. To analyze the effect of de novo synthesized cholesterol, we treated HepG2 cells with pitavastatin alone (p-Hep) or with pitavastatin together with mevalonate (m-Hep). As shown in Fig. 6D, E, m-Hep accumulated more cholesterol in the cells than p-Hep; however, no difference was observed in FoxO1 protein levels between p-Hep and m-Hep. These results suggested that cholesterol absorbed by NPC1L1 possessed distinct physiological properties from that derived from de novo synthesis.
In this study, we investigated the effects of hepatic NPC1L1 expression on lipoprotein metabolism and analyzed the difference in effects between cholesterol absorbed from the bile and cholesterol synthesized de novo. A previous study utilizing liver-specific NPC1L1 transgenic mice proved that hepatic NPC1L1 played a role in the absorption of cholesterol from the bile (12); however, inbred expression of NPC1L1 might provoke secondary modifications in the regulation of several genes to maintain homeostasis, thereby obscuring the direct effects of NPC1L1 expression. Thus, in this study, utilizing an adenoviral vector, we adventitiously expressed NPC1L1 in mice liver to elucidate the genuine effects caused by hepatic NPC1L1 expression. As shown in the results, the NPC1L1 protein expressed with adenoviral vector was localized on the side of the bile duct canaliculi, leading to the reduced cholesterol contents in the bile. With this model, we found two notable metabolic changes: the appearance of ERL in the plasma and the decrement in VLDL-TG levels.
ERLs, with particle sizes between those of LDL and HDL, would be classified as large HDL based on their mobility on agarose electrophoresis and the lack of apoB protein in the particles. Treatment with ezetimibe reduced the amount of ERLs, and the expression of LSS, an enzyme that increased cellular cholesterol content through a mechanism other than NPC1L1, did not result in the emergence of ERLs. These results indicated that hepatic NPC1L1 expression and subsequent uptake of cholesterol from the bile were essential in the appearance of ERLs. The selective uptake of free cholesterol by NPC1L1 (27) would have conferred these ERLs a unique property that they were rich in free cholesterol; this property would have been augmented by the inefficient lecithin-cholesterol acyltransferase (LCAT) function on these ERLs due to their little abundance of apolipoprotein A-I, an activator of LCAT. Furthermore, NPC1L1 expression in apoE-deficient mice did not culminate in the appearance of large HDL, suggesting that apoE protein was indispensible in the formation of ERLs. Neither the upregulation of ABCA1 protein levels, which was assumed to be responsible for the appearance of ERLs in high-cholesterol-fed NPC1L1 transgenic mice (12), nor the increased apoE expression were observed in our NPC1L1 overexpressed mice. There remains a possibility that some other factors might have contributed to the formation or the appearance of ERLs. For example, the clearance of ERLs might be retarded, considering the decreased protein level of LDL receptor and the decreased chylomicron clearance; or, the activities of lipases might be modulated, resulting in the increased plasma ERL levels. However, in light of the findings that the free cholesterol content of ERLs was much higher on day 3 (supplementary Fig. XI) than on day 5 and that ERLs were produced by HepG2 cells overexpressing NPC1L1, it can be speculated that ERLs were formed as a direct consequence of the increased cholesterol uptake from the bile.
ERLs are usually not detected in the plasma of normal humans who express NPC1L1 in the liver. However, expression of NPC1L1 in mice liver resulted in the emergence of these unique lipoproteins. The possible explanation for the appearance of ERLs in L1-mice may be that the mice are deficient in cholesteryl ester transfer protein (CETP). In fact, it was reported that subjects with CETP deficiency have apoE-rich HDL (28) and that pharmacological inhibition of CETP resulted in the increment of plasma apoE levels, especially in HDL fractions (29). Furthermore, expression of CETP in SR-BI-deficient mice, the animal model which possesses ERLs in the plasma, resulted in the reduction of ERL levels (30, 31). Of course, the extent to which hepatic ERL production contributes to the increase of apoE-rich HDL under CETP inhibition is still unclear; further analyses are awaited to clarify the association of CETP and the appearance of ERLs.
ApoE-deficient mice, which did not produce ERLs due to the lack of apoE, manifested hepatic injury upon NPC1L1 expression. Thus, one of the physiological roles of ERLs is to ameliorate hepatic injury that results from excessive free cholesterol taken up by the bile. In addition, as demonstrated in previous reports, HDLs abundant in apoE, regardless of its isoform, facilitated reverse cholesterol transport, thereby possibly exerting an antiatherogenic function (32, 33). Future investigations are necessary to test whether the ERLs produced directly from the liver possess antiatherogenic functions should CETP inhibitors become promising antidyslipidemic agents (34, 35).
The other consequence of hepatic NPC1L1 expression on lipid metabolism was the drop in plasma triglyceride levels due to the suppression of VLDL-TG production. This decrease in VLDL-TG production was, at least in part, attributed to the reduced expression of MTP, and the decreased MTP expression was assumed to be the consequence of the reduction in FoxO1 protein levels. Because FoxO1 protein is a transcriptional factor that plays an important role not only in lipid metabolism but also in glucose metabolism (26, 36, 37), further investigation on glucose metabolism in our present model might be fascinating.
The expression of NPC1L1 in the liver facilitates cholesterol uptake from the bile, whereas ezetimibe treatment inhibits cholesterol uptake from the bile. Thus, our observation of plasma triglycerides levels in L1-mice appears to be inconsistent with the previous reports demonstrating that administration of ezetimibe ameliorated triglyceride metabolism in rats (38) and decreased plasma TG levels in humans (3, 5, 6). A probable explanation for this discrepancy is that the levels of hepatic NPC1L1 expression in our model were so high that the delicate physiological balance between hepatic and intestinal NPC1L1 functions had been disturbed. In fact, as shown in Fig. 4D, ezetimibe treatment in Null-mice did decrease hepatic MTP expression. Taking account of these results and speculations, selective inhibition of intestinal NPC1L1 might be more beneficial in reducing plasma TG levels.
Through this study, we found that, depending on its source of origin, cholesterol exerts a differential effect on the intracellular metabolism. Both the NPC1L1 expression and LSS expression resulted in increased intracellular cholesterol levels; however, the former resulted in the plasma appearance of ERLs and suppression of FoxO1, while the latter did not. One possible explanation for this difference is the different localization of cholesterol originating from different sources: cholesterol taken up by NPC1L1 is delivered to the early endosome after being incorporated to the plasma membrane, whereas cholesterol synthesized de novo is localized on the ER membrane. It is easy to speculate that this different localization of cholesterol in the cellular organelle induces the subsequent differential physiological effect, thereby affecting the plasma lipoprotein metabolism. Considering that cholesterol in apoB-containing lipoproteins is taken up through lipoprotein receptors and undergoes lysosomal processing, it is possible that cholesterol derived from lipoproteins may possess unique properties. Future investigation of the intracellular transport pathways of cholesterol derived from different sources and their metabolic effects would enrich our understanding of the hepatic cholesterol metabolism and perhaps culminate in the development of a better treatment strategy for dyslipidemia and atherosclerosis.
In summary, the hepatic expression of NPC1L1 in mice resulted in the plasma appearance of ERLs and the decrement of VLDL-TG secreted from the liver. These findings were not found in mice with increased hepatic de novo cholesterol synthesis, suggesting that cholesterol, depending on its source, exerted distinct properties in cellular and plasma lipid metabolism.
This research was supported by Grants-in-Aid 20591079 (K. Tsukamoto) and 23591330 (K. Tsukamoto) from the Japan Society for the Promotion of Science.
[S]The online version of this article (available at http://www.jlr.org) contains supplementary data in the form of 11 figures.