Proprotein convertase subtilisin/kexin type 9 (PCSK9) posttranslationally promotes the degradation of the low-density lipoprotein receptor (LDLr) in hepatocytes and increases plasma LDL cholesterol. It is not clear, however, whether PCSK9 plays a role in the small intestine. Here, we characterized the patterns of variations of PCSK9 and LDLr in fully differentiated Caco-2/15 cells as a function of various potential effectors. Cholesterol (100 μM) solubilized in albumin or micelles significantly downregulated PCSK9 gene (30%, P < 0.05) and protein expression (50%, P < 0.05), surprisingly in concert with a decrease in LDLr protein levels (45%, P < 0.05). Cells treated with 25-hydroxycholesterol (50 μM) also displayed significant reduction in PCSK9 gene (37%, P < 0.01) and protein (75% P < 0.001) expression, whereas LDLr showed a decrease at the gene (30%, P < 0.05) and protein (57%, P < 0.01) levels, respectively. The amounts of PCSK9 mRNA and protein in Caco-2/15 cells were associated to the regulation of 3-hydroxy-3-methylglutaryl-CoA reductase and sterol regulatory element binding protein-2 (SREBP-2) that can transcriptionally activate PCSK9 via sterol-regulatory elements located in its proximal promoter region. On the other hand, depletion of cholesterol content by hydroxypropyl-β-cyclodextrin upregulated PCSK9 transcripts (20%, P < 0.05) and protein mass (540%, P < 0.001), in parallel with SREBP-2 protein levels. The addition of bile acids (BA) taurocholate and deoxycholate to the apical culture medium lowered PCSK9 gene expression (25%, P < 0.01) and raised PCSK9 protein expression (30%, P < 0.01), respectively, probably via the modulation of farnesoid X receptor. Furthermore, unconjugated and conjugated BA exhibited different effects on PCSK9 and LDLr. Altogether, these data indicate that intestinal PCSK9 is highly modulated by sterols and emphasize the distinct effects of BA species.
- bile acids
- farnesoid X receptor
familial hypercholesterolemia is characterized by elevated plasma low-density lipoprotein (LDL) and deposition of LDL in arteries, leading to premature cardiovascular events (15). This genetically dominant hypercholesterolemia is related to molecular aberrations in the LDL receptor (LDLr) or its ligand apolipoprotein B. Recently, mutations in a third gene, proprotein convertase subtilisin/kexin type 9 (PCSK9), were associated to this disease (1). Human PCSK9 gene is ∼22 kb long, comprising the promoter region and 12 exons, and it is located on chromosome 1p32 (49). The gene produces an mRNA of 3,636 bp encoding a 692-aa protein. Although it is proposed that secreted PCSK9 interacts with LDLr, enters the endocytic recycling pathway with LDLr, and impairs LDLr pathway by increasing the degradation of the receptor (5, 37), the precise molecular basis has yet to be determined.
PCSK9 is mainly expressed in the liver. It is thought to be involved in liver regeneration, neuronal differentiation/apoptosis, and cortical neurogenesis (49), but its primary function apparently consists in negatively regulating LDLr, likely by a posttranscriptional manner, e.g., degradation of LDLr in acidic subcellular compartments, possibly endosomes/lysosomes (23, 30, 49, 50). Even though PCSK9 has been found expressed substantially in intestine (49), studies have been mostly restricted to the liver (23, 30, 50). However, maintaining cholesterol homeostasis in the body requires accurate metabolic cross talk between hepatic and intestinal processes to adequately cope with large fluctuations in dietary cholesterol intake (58), whereas imbalance may lead to elevated LDL cholesterol levels and increased risks for coronary heart disease, the main cause of death in Western societies (56). Notably, the intestine plays a key role in cholesterol balance in animals and humans (55), constitutes the only site for absorption of dietary sterols (57), quantitatively represents the single active location for cholesterogenesis (53, 59), and remains the second most important organ for the uptake and degradation of circulating LDL (52). Whether PCSK9 reduces the number of LDLr in the enterocyte basolateral membrane is not known. This information is crucial since LDLr behavior in absorptive cells does not necessarily reflect that in hepatocytes.
Poor information is available about the regulation of PCSK9 even though it seems to be an attractive target for lowering LDL cholesterol (1). In mice, PCSK9 is downregulated by dietary cholesterol. Conversely, PCSK9 gene expression is upregulated in mice overexpressing sterol regulatory element binding proteins (SREBP)-1α and SREBP-2 (two transcription factors activated by low levels of intracellular cholesterol) (24). The nutritional status also modulates PCSK9 expression. For example, 24-h fasting in mice dramatically decreased hepatic PCSK9 mRNA and protein levels, which were progressively restored by carbohydrate refeeding (9). Similarly, statins have been shown to lower LDL by inducing SREBP-2, which enhances LDLr expression. Moreover, the peroxisome proliferator-activated receptor (PPAR)-α agonist fenofibrate downregulates PCSK9 gene expression and protein synthesis in controls and not, as expected, in PPAR-α-deficient mice (31). Most of the work on PCSK9 regulation has been performed in relation to the liver, but it must be extended to the small intestine, an important site for cholesterol homeostasis. Therefore, the purpose of the present study was to examine the regulation of PCSK9 in an established Caco-2/15 enterocyte cell line. An additional aim of the present work was to determine the expression profile of LDLr and various nuclear transcription factors.
MATERIALS AND METHODS
Caco-2/15 cells (American Type Culture Collection, Rockville, MD) were grown at 37°C with 5% CO2 in minimal essential medium (MEM) (GIBCO-BRL, Grand Island, NY) containing 1% penicillin-streptomycin and 1% MEM nonessential amino acids (GIBCO-BRL) and supplemented with 10% decomplemented fetal bovine serum (FBS) (Flow, McLean, VA). Caco-2/15 cells (passages 30–40) were maintained in T-75-cm2 flasks (Corning Glass Works, Corning, NY). Cultures were split (1:6) when they reached 70–90% confluence, by use of 0.05% trypsin-0.5 mM EDTA (GIBCO-BRL). For individual experiments, cells were plated at a density of 1 × 106 cells/well on 24.5-mm polycarbonate Transwell filter inserts with 0.4-μm pores (Costar, Cambridge, MA), in MEM (as described above) supplemented with 5% FBS. The inserts were placed into six-well culture plates, permitting separate access to the upper and lower compartments of the monolayers. Cells were cultured for various periods, including 21 days, at which the Caco-2/15 cells are highly differentiated and appropriate for lipid synthesis (16, 34, 48). The medium was refreshed every second day.
Cholesterol and 25-hydroxycholesterol (Sigma-Aldrich Canada, Oakville, ON) were dissolved in chloroform and dried under nitrogen. Acetone (0.5 ml) was then added to dissolve the dried sterol. This solution was slowly added with continuous stirring to 3 ml of an albumin solution (250 mg of fatty acid-poor BSA in 3 ml of 10% BSA pH 7.4). This mixture was placed under a stream of nitrogen until the odor of acetone was no longer detectable. Micellar cholesterol was prepared as described previously (45). The mixed bile salt micelle, as a vehicle of cholesterol and 25-hydroxycholesterol, contained 4.8 mM sodium taurocholate and 0.3 mM monoolein. The micellar solution was warmed to 37°C and stirred vigorously before use. Albumin- or micelle-bound sterols were mixed to MEM without FBS (18 h) before being added to the apical compartment. Hydroxypropyl-β-cyclodextrin (Sigma-Aldrich Canada) was directly dissolved in MEM at a 25 mM concentration and then added to the apical compartment. Cholate (4.8 mM), chenodeoxycholate (250 μM) and conjugated forms (Calbiochem, Gibbstown, NJ), as well as deoxycholate (250 μM) (ICN Biomedicals, Aurora, OH) and conjugated forms (Calbiochem, Gibbstown, NJ) were dissolved in 95% ethanol and evaporated under nitrogen before being mixed with MEM at their final concentrations.
Antibodies against PCSK9.
The specificity of the Abs (particularly PCSK9) (43, 46) was evaluated by various methods, including Western blotting following the incubation of the Abs in the presence or absence of specific antigens, the omission of the primary PCSK9 Ab in Western blot, and the identification of the human PCSK9 sequences following immunoprecipitation and SDS-PAGE.
To assess the presence and regulation of PCSK9, LDLR, and SREBP-2, Caco-2/15 cells were homogenized and adequately prepared for Western blotting as described previously (33). The Bradford assay (Bio-Rad) was used to estimate protein concentration. Proteins were denatured in sample buffer containing SDS and DTT (Thermo Scientific, Rockford, IL), separated on a 7.5% SDS-PAGE and electroblotted onto Hybond-C extra nitrocellulose membranes (Amersham Biosciences, Piscataway, NJ). Nonspecific binding sites of the membranes were blocked with defatted milk proteins followed by the addition of primary antibodies directed against the targeted proteins: β-actin (Sigma-Aldrich Canada), LDLr (Research Diagnostic, Flanders, NJ), PCSK9 (amino acids 31-454) (kind gift of G. Dubuc and J. Davignon, Clinical Research Institute of Montreal, University of Montreal, Montreal, QC, Canada), and SREBP-2 (Cayman, Ann Arbor, MI). The relative amount of primary antibody was detected with species-specific horseradish peroxidase-conjugated secondary antibody. Even if identical protein amounts of tissue homogenates were applied, the β-actin protein was used as a reference protein. Molecular size markers were simultaneously loaded on gels (data not shown on the figures). Blots were developed and the mass of proteins were quantified by using an HP Scanjet scanner equipped with a transparency adapter and software. Importantly, the utilization of differential loading and quantification suggested a dynamic range of densitometric measurements.
Experiments for mRNA quantification as well as for GAPDH (as a control gene) were performed by using the Eppendorf Mastercycler Gradient PCR machine (Eppendorf, Westbury, NY) as reported previously (40, 48). Approximately 30–40 cycles of amplification were used at 95°C for 30 s, 58°C for 30 s, and 72°C for 30 s. The following primer sequences were used: farnesoid X receptor (FXR) forward (5′-CATGCGAAGAAAGTGTCAAGAGTGTCG-3′) and reverse (5′-CTTTGTTGTCGAGGTCACTTGTCGCA-3′); GAPDH forward (5′-GTCCACTGGCGTGTTCACCA-3′) and reverse (5′-GTGGCAGTGATGGCATGGAC-3′); 3-hydroxy-3-methylglutaryl (HMG)-CoA reductase forward (5′-ACCCTTAGTGGCTGAAACAGATACCC-3′) and reverse (5′-AACTGTCGGCGAATAGATACACCACG-3′); LDLr forward (5′-TGAGAGGACCACCCTGAGCAAT-3′) and reverse (5′-TTACGGCTGTGGAGCTGACCTTTA-3′); and PCSK9 forward (5′-AGGACTGTATGGTCAGCACACT-3′) and reverse (5′-CGGGATTCCATGCTCCTTGACTTT-3′). Amplicons were visualized on standard ethidium bromide-stained agarose gels. Under these experimental conditions relative to RT-PCR, 34–36 cycles corresponded to the linear portion of the exponential phase. Pilot and previous studies in our laboratories showed that the RT-PCR, under our experimental conditions, is quantitative employing different amounts of RNA. Fold induction was calculated by using GAPDH (glyceraldehyde 3-phosphate dehydrogenase) as the reference gene.
All values were expressed as means ± SE. The data were evaluated by ANOVA, where appropriate, and the differences between the means were assessed by the two-tailed Student's t-test.
The first issue addressed by our studies was to explore whether a difference could be noted in the expression of PCSK9 under the influence of sterols. The incubation of fully differentiated Caco-2/15 cells with various concentrations of cholesterol solubilized in albumin resulted in a decrease in the gene and protein expression of PCSK9. At 100 μM cholesterol, decreases of 25% (P < 0.001) and 55% (P < 0.05) were significantly recorded in mRNA and protein mass levels, respectively (Fig. 1A). When the regulation of LDLr was investigated, a lower protein expression (∼60%, P < 0.001) along with significantly unchanged mRNA levels was noted (Fig. 1B). Since bile acid (BA)-facilitated cholesterol absorption occurs in the small intestine in physiological conditions, solutions containing either micelles alone or micelles together with cholesterol were added to the apical media. Again, a micellar cholesterol-mediated drop (30–50%) was recorded in the gene and protein expression of PCSK9 (Fig. 2A), accompanied with a similar reduction in LDLr protein expression (Fig. 2B). Finally, the substitution of cholesterol by 25-hydroxycholesterol produced a similar results on PCSK9 (Fig. 3A) and on both mRNA and protein levels of LDLr (Fig. 3B).
In view of these findings, we assessed protein expression of nuclear SREBP-2 with an antibody that distinguished the truncated active nuclear form from its membrane-linked precursor. Also, since LDLr and HMG-CoA reductase, the limiting enzyme in cholesterol synthesis, are sensitive to SREBP-2 and sterol in the small intestine, we determined HMG-CoA reductase transcription. As illustrated in Fig. 4A, the administration of micellar cholesterol or 25-hydroxycholesterol to Caco-2/15 cells lowered HMG-CoA reductase transcripts (30 and 55%, respectively). Moreover, measurement of SREBP-2 revealed a lessening in its protein expression (Fig. 4B).
We next evaluated how depletion of cholesterol cell content modulates PCSK9 and LDLr. The extraction of cholesterol was achieved with hydroxypropyl-β-cyclodextrin that has been shown to selectively eliminate cholesterol from the plasma membrane, in preference to other lipids (28). As illustrated in Fig. 5, A and B, hydroxypropyl-β-cyclodextrin markedly increased the gene and protein expression of PCSK9 and LDLr. The expression of HMG-CoA reductase and SREBP-2 was upregulated as well (Fig. 5C). To determine whether this impact is mainly due to cholesterol extraction, Caco-2/15 cells were supplemented with a combination of hydroxypropyl-β-cyclodextrin and micellar cholesterol. The mixture largely restored the control levels.
Since BA play a significant role in cholesterol metabolism and are increasingly being appreciated as metabolic integrators and signaling factors, we subsequently tested their specific effects on PCSK9. Taurocholate (4.8 mM) reduced mRNA level of PCSK9 by 25% without affecting LDLr expression (Fig. 6). On the other hand, the addition of deoxycholate (250 μM) enhanced the protein expression of PCSK9 by 30% (Fig. 7A). LDLr was more responsive to deoxycholate by showing a ∼70% and twofold increase in gene and protein expression, respectively (Fig. 7B). As noted in Fig. 8, taurocholate slightly diminished the transcripts of HMG-CoA reductase without affecting SREBP-2 and FXR gene expression. On the other hand, deoxycholate decreased FXR gene expression, enhanced the protein expression of SREBP-2, and remained without effect on HMG-CoA reductase (Fig. 8).
Given the different properties of unconjugated and conjugated BA, we have examined their effects on the expression of PCSK9 and LDLr. As noted in Table 1, cholate was more powerful than taurocholate in reducing PCSK9 mRNA, whereas glycocholate significantly enhanced it. No significant modifications were observed in PCSK9 protein mass and LDLr gene expression following the incubation of Caco-2/15 cells with cholate, taurocholate, and glycocholate. Cholate was also able to downregulate LDLr protein mass. Although some alterations were noted in FXR and SREBP2, they never reached statistical significance. We also assessed the influence of chenodeoxycholate, taurochenodeoxycholate, and glycochenodeoxycholate on the expression of PCSK9, LDLr, FXR, and SREBP2 and could not record significant alterations, except for glycochenodeoxycholate, that increased LDLr and FXR mRNA. Finally, we evaluated the impact of the two conjugated forms of deoxycholate. Taurodeoxycholate and glycodeoxycholate were both capable of reducing PCSK9 mRNA levels. However, only taurodeoxycholate was able to downregulate LDLr gene expression. In fact, deoxycholate was more effective in modulating gene and protein expressions of PCSK9, LDLr, and SREBP-2.
Despite valuable advances, additional studies are clearly warranted to understand the complex molecular mechanisms that orchestrate cholesterol homeostasis in the small intestine. A tremendous opportunity is offered by the recent discovery of PCSK9, which displays various functions, including hepatic LDLr degradation. In view of the small intestine's high capacity to absorb lipids and elaborate most of the major lipoprotein classes and considering the well-known actions of a number of nutrients and hormones on lipid metabolism and transport at the intestinal level, the lack of knowledge about the modulation of PCSK9 in the gut is puzzling. For the first time, the present work attempted to detail the modulation of PCSK9 in intestinal cells. According to our data, there is a discrete regulation of PCSK9 from stimuli, originating from apical media, such as sterols, BA, and cyclodextrin. Apparently, PCSK9 and LDLr are stimulated or downregulated in a similar fashion via SREBP-2 transcription factor. Furthermore, in our experiments, we have supplemented Caco-2/15 cells with sterols or caused the depletion of cholesterol cell content using hydroxypropyl-β-cyclodextrin, an approach that is both rapid and highly efficient (18, 19). In addition, hydroxypropyl-β-cyclodextrin is entirely surface acting, so, apart from removing cholesterol, it has a minimal effect on cell membranes (25). Our findings clearly showed that provision of sterols to enterocytes downregulated the gene and protein expression of PCSK9. Conversely, the lessening of cholesterol cell content enhanced PCSK9 transcription. In similar fashion, PCSK9 expression was highly downregulated by dietary cholesterol in mouse liver (38). This is also consistent with the demonstration that PCSK9 expression was strongly induced by statins in a dose-dependent manner in HepG2 cells and in human primary hepatocytes, likely a result of the cholesterol-lowering effect of the drug, and that this induction was efficiently reversed by mevalonate (11). Altogether, these studies pointed out that 1) the regulation of PCSK9 expression is dependent on the presence or absence of sterols not only in hepatocytes, but also in intestinal epithelial cells, and 2) the modulation of PCSK9 by sterols, which are supplied either exogenously or endogenously, is achieved at the transcription level.
It has been reported that gain-of-function mutations in PCSK9 have a 23% decreased level of cell surface expression of LDL receptors and a 38% decrease in internalization of LDL, whereas loss-of-function mutations are associated with a 16% increased level of cell surface LDLr and a 35% increased internalization of LDL (7). Apparently, the catalytic activity of PCSK9 appears not to be required for LDLr degradation, but it is essential for activation and secretion of PCSK9 (10, 22, 39). In our work, the protein expression of LDLr showed similar changes to PCSK9 in Caco-2/15 cells with sterol supplementation or under the sterol-depleted conditions. Considering that the function of PCSK9 is to promote the degradation of LDLr (23), as noted before (3), the parallel change in LDLr to PCSK9 by sterols is surprising. However, previous studies have reported a similar behavior of the two proteins in HepG2 cells (26). One might consider that these cells lack a component or machinery necessary for the coordination of the expression of these proteins, but other research groups have not found a reciprocal regulation of PCSK9 and LDLr. For example, inhibition of squalene synthase upregulates PCSK9 and LDLr expression in rat liver (4) and there was no reverse relation of PCSK9 and LDLr in rat liver cell line (17). PCSK9 was identified as one of the genes that are regulated by SREBPs (24, 38). The SREBPs are members of the basic helix-loop-helix leucine zipper family of transcription factors that regulate the expression of the target genes by binding to the sterol-regulatory elements (SRE) in their promoter regions (6). Dubuc et al. (11) reported that the transcription of PCSK9 was increased by statins, most likely through SREBP-2 activation in primary human hepatocytes, and proposed the importance of conserved SRE in the promoter region of mouse, rat, and human PCSK9 genes. Thereafter, the expression of PCSK9 was found dependent on the absence or presence of sterols via SRE in the minimal promoter region of the human PCSK9 gene by both SREBP-1 and SREBP-2 in HepG2 cells, and it was proposed that the predominant transcriptional regulator of PCSK9 by cholesterol is SREBP-2 in vivo (26). Accordingly, our data on intestinal epithelial cells showed that the administration of sterols reduced SREBP-2 transcripts while decreasing PCSK9 gene and protein expression. On the other hand, depletion of cholesterol content in Caco-2/15 cells by cyclodextrin raised the levels of mRNA and protein mass of PCSK9 and LDLr. As expected, the changes of SREBP-2 and PCSK9 in Caco-2/15 cells paralleled the alterations in mRNA encoding HMG-CoA reductase, the key cholesterol biosynthetic enzyme. These results might suggest that PCSK9 would be under the control of SREBP-2 in Caco-2/15 cells, as is the case for LDLr and HMG CoA reductase (12–14).
BA are amphiphilic molecules synthesized from cholesterol exclusively in the liver, represent the major route for removal of excess cholesterol from the body, and are essential for effective transport of dietary fat (29). In addition to their traditional role in dietary lipid absorption and cholesterol homeostasis, it has become clear now that BA constitute versatile signaling molecules endowed with systemic endocrine functions. In fact, BA are ligands for G protein-coupled receptors and modulate several nuclear hormone receptors (27, 35, 36). Through activation of these diverse signaling pathways, BA have been shown to regulate not only their own synthesis and enterohepatic recirculation, but also triglyceride, cholesterol, and glucose homeostasis (51). Given the broad relationship of BA with multiple pathways and their recent link with PCSK9 in hepatocytes (32, 44), we decided to investigate their impact on PCSK9 in the intestine. In the present experiments, we employed taurocholate and deoxycholate, two powerful BA. According to our data in Caco-2/15 cells, taurocholate brought about a decrease in the gene expression of PCSK9, whereas deoxycholate enhanced that of PCSK9 and LDLr along with a lessening of FXR mRNA. The distinct effects of BA on PCSK9 gene expression were also noted in another study on human hepatocytes that showed a repression by chenodeoxycholic acid, an induction by deoxycholic acid, and no influence with cholic acid and ursodeoxycholic acid (32). Further work is needed to explore whether these separate impacts on PCSK9 may be due to the potential activation by BA of other nuclear receptors such as pregnane X receptor, constitutive and rostane receptor, and vitamin D receptor in addition to FXR (54, 60).
Chenodeoxycholate and cholate are the two primary BA in humans and are conjugated mainly to glycine and taurine (21). Given their hydrophilic-hydrophobic properties, unconjugated and conjugated BA may differ in their various actions, including membrane permeability, lipid solubilization, intracellular signaling, and DNA synthesis (2, 8, 20, 35, 41, 47). We have therefore assessed the modulation of unconjugated and conjugated BA to taurine or glycine. Our findings indicate the high ability of cholate, taurodeoxycholate, and glycodeoxycholate to alter the gene expression of PCSK9. Cholate and taurodeoxycholate also diminished LDLr protein content and mRNA, respectively, in Caco-2/15 cells. The effects of cholate were superior to those of taurocholate and quite divergent from those of glycocholate. On the other hand, chenodeoxycholate and taurochenodeoxycholate were ineffective in modulating PCSK9 and LDLr. Only the glycochenodeoxycholate displays a capacity to upregulate LDLr and FXR mRNA. However, we were not able to show LDLr mRNA upregulation by chenodeoxycholate, previously reported by Nakahara et al. (42), presumably because of the difference in experimental techniques. Additional studies are necessary to uncover novel mechanisms by which specific BA modulate PCSK9 and LDLr. This in turn may have far reaching consequences in the control of the overall body metabolism homeostasis and the multiorgan implications.
This study was supported by the Canadian Institutes of Health Research (Grant MOP 49433 to E. Levy and 36496 to N. G. Seidah).
The authors thank Schohraya Spahis, Carole Garofalo, Émilie Grenier, and Geneviève Lalonde for helpful technical assistance.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2009 the American Physiological Society