BA (bile acid) formation is considered an important final step in RCT (reverse cholesterol transport). HDL (high-density lipoprotein) has been reported to transport BAs. We therefore investigated the effects of monogenic disturbances in human HDL metabolism on serum concentrations and lipoprotein distributions of the major 15 BA species and their precursor C4 (7α-hydroxy-4-cholesten-3-one). In normolipidaemic plasma, approximately 84%, 11% and 5% of BAs were recovered in the LPDS (lipoprotein-depleted serum), HDL and the combined LDL (low-density lipoprotein)/VLDL (very-low-density lipoproteins) fraction respectively. Conjugated BAs were slightly over-represented in HDL. For C4, the respective percentages were 23%, 21% and 56% (41% in LDL and 15% in VLDL) respectively. Compared with unaffected family members, neither HDL-C (HDL-cholesterol)-decreasing mutations in the genes APOA1 [encoding ApoA-I (apolipoprotein A-I], ABCA1 (ATP-binding cassette transporter A1) or LCAT (lecithin:cholesterol acyltransferase) nor HDL-C-increasing mutations in the genes CETP (cholesteryl ester transfer protein) or LIPC (hepatic lipase) were associated with significantly different serum concentrations of BA and C4. Plasma concentrations of conjugated and secondary BAs differed between heterozygous carriers of SCARB1 (scavenger receptor class B1) mutations and unaffected individuals (P<0.05), but this difference was not significant after correction for multiple testing. Moreover, no differences in the lipoprotein distribution of BAs in the LPDS and HDL fractions from SCARB1 heterozygotes were observed. In conclusion, despite significant recoveries of BAs and C4 in HDL and despite the metabolic relationships between RCT and BA formation, monogenic disorders of HDL metabolism do not lead to altered serum concentrations of BAs and C4.
- bile acid (BA)
- high-density lipoprotein
- reverse cholesterol transport
RCT (reverse cholesterol transport) is considered an important anti-atherogenic mechanism that involves the efflux of cholesterol from peripheral tissues, notably the lipid-laden macrophages of the arterial wall, and the subsequent delivery of cholesterol to the liver for its excretion from the body. HDL (high-density lipoprotein), a heterogeneous class of serum lipoproteins characterized by a density between 1.063 and 1.21 g/ml, are considered as the major players in RCT [1–4]. HDL particles are composed of protein and lipid components, of which ApoA-I (apolipoprotein A-I), the major protein constituent of HDL, is synthesized both by the liver and the intestine. After secretion into plasma, ApoA-I is lipidated with cellular phospholipids and cholesterol via the ABCA1 (ATP-binding cassette transporter A1) [2,3]. The resulting nascent HDL particles are converted into mature HDL by the acquisition of additional lipids from cells and lipoproteins and esterification of free cholesterol with the sn−2 fatty acid of phosphatidylcholine by LCAT (lecithin:cholesterol acyltransferase) . Functionally relevant mutations in the genes APOA1 (encoding ApoA-I), ABCA1 or LCAT cause gene-dosage-dependent decreases of HDL-C (HDL-cholesterol) plasma concentrations so that carriers of one defective allele have approximately half-normal HDL-C levels, whereas carriers of two defective alleles are virtually HDL-deficient [5,6].
Cholesteryl esters are removed from mature HDL particles mostly by the liver, either directly by selective uptake or indirectly after transfer to VLDL (very-low-density lipoprotein) or LDL (low-density lipoprotein). The first process involves SR-BI (scavenger receptor class B1; encoded by SCARB1) . The second process is mediated by CETP (cholesteryl ester transfer protein), which exchanges cholesteryl esters from HDL against TAGs (triacylglycerols) from VLDL . The TAGs received by HDL are rapidly hydrolysed by HL (hepatic lipase) . Functionally relevant mutations in the genes SCARB1, CETP or LIPC (encoding HL) lead to increased plasma levels of HDL-C [9–11].
Following hepatic uptake, cholesterol may be utilized for the de novo synthesis of lipoproteins and membranes or excreted into the bile, either directly or after conversion into BAs (bile acids). The biosynthesis of BAs involves several enzymes and leads to the formation of primary BAs, namely CA (cholic acid) and CDCA (chenodeoxycholic acid) in humans. A relatively stable metabolite, C4 (7α-hydroxy-4-cholesten-3-one), has been promoted as a biomarker of BA synthesis . Prior to their secretion into the bile, primary BAs are conjugated with either glycine or taurine . Further complexity of BAs is generated by the action of intestinal bacteria, which give rise to secondary BAs by deconjugation and/or 7α-dehydroxylation. Secondary BAs include DCA (deoxycholic acid), LCA (lithocholic acid) and UDCA (ursodeoxycholic acid). The mixture of primary and secondary BAs is extensively reabsorbed from the intestine and returned to the liver via the bloodstream in order to be secreted again into the bile, thereby completing the enterohepatic circulation .
In the bloodstream, albumin acts as the major transporter of BAs [15,16]. In addition, both conjugated and unconjugated CA and CDCA have been recovered in HDL . BAs, hence, have at least two relationships with HDL metabolism: they are final products of RCT and they are partially transported by these lipoproteins. In addition, as activating ligands of the nuclear receptor FXR (farnesoid X receptor), BAs may also contribute to the regulation of HDL metabolism, as FXR activation represses the expression of APOA1  SCARB1  and LIPC . In the present study, we set out to investigate whether these, at least triple, relationships between BA and HDL metabolism translate into altered serum concentrations of the 15 major human BAs as well as the biosynthesis marker C4 in patients with mutations in HDL genes.
MATERIALS AND METHODS
Fasting blood samples were collected in the morning, after a minimum of 10 h fast, from 48 Dutch patients with functionally relevant mutations in APOA1, ABCA1, LCAT, SCARB1, CETP or LIPC. Control samples were obtained from 45 unaffected family members of these patients. In addition, four Danish patients with mutations in the genes encoding APOA1 or CETP as well as five age- and sex-matched controls were included. The mutations of these individuals as well as their demographic, anthropometric and lipoprotein traits have been described previously (; for details see Supplementary Tables S1 and S2 at http://www.clinsci.org/cs/122/cs1220385add.htm). Serum was prepared by centrifugation of the whole blood at 2000 g for 10 min at 25°C after having allowed clotting for at least 30 min. Aliquots were immediately frozen at −80°C until analysis.
The Medical Ethics Committee of the AMC (Academic Medical Center) in Amsterdam, The Netherlands, as well as the Danish Ethics Committee for Copenhagen and Frederiksberg, Denmark, approved all genetic and phenotypic studies described and all participants signed an informed consent to join the study.
Lipoprotein fractions and lipoprotein-depleted plasma from healthy volunteers were isolated after sequential ultracentrifugation at 59000 rev./min at 15°C on a Beckman Coulter Optima L90K ultracentrifuge using a procedure described previously . The starting material consisted of 450 ml of pooled plasma from two healthy blood donors (Zürcher Blutspendedienst). The purity of the different fractions was evaluated by electrophoresis in a SDS/PAGE (10% gel) and proteins were stained with Coomassie Brilliant Blue according to standard protocols.
Lipoprotein fractions were also isolated from 0.9 ml of plasma from 15 SCARB1 heterozygotes and 15 controls. In this case, an Optima MAX bench-top ultracentrifuge (Beckman Coulter) was used. The samples were ultracentrifuged at 100000 rev./min at 10°C for 2 h, after which the fractions were collected using a tube slicer. The data from a patient of the wild-type group was excluded from the statistical data analysis, because this patient had a missing terminal ileum, which was expected to influence BA concentrations.
The 15 major human BAs as well as C4 were quantified using an LC–tandem MS method described previously by our group . For the healthy blood donors, quantifications were performed on either 100 μl of total serum or on 300 μl of the lipoprotein fractions isolated from 450 ml of plasma since initially very low BA concentrations were expected in most fractions.
Because the starting plasma volume (0.9 ml) was much smaller for the fractions obtained after ultracentrifugation of plasma from family members of SCARB1 mutation carriers, we used the total fraction volume that was obtained. The volume of ammonium carbonate buffer 100 mM, pH 9.3 added before SPE (solid-phase extraction) was adapted in order to correspond to the sample volume and all samples were completed to an identical sample volume with purified water. After the analysis, the calculated concentrations of these samples were corrected for the larger sample volume used. The BA concentrations corresponding to the lipoprotein fractions in total plasma were calculated by a rule of three using the concentration of the analysed fraction, the final volume of the analysed fractions after ultracentrifugation and the volume of starting material.
Concentrations of 27OHC (27-hydroxycholesterol) in total and ApoB (apolipoprotein B)-depleted serum were determined as described previously . Serum concentrations of total cholesterol, TAGs, HDL-C, LDL-C (LDL-cholesterol), ApoB and ApoA-I levels were measured with commercial kits (Wako and Randox) on a Cobas Mira autoanalyser.
Statistical analyses were performed using IBM® SPSS® Statistics, version 19. Because the data did not follow a Gaussian frequency distribution, univariate statistics were performed using the non-parametric Mann–Whitney or Kruskal–Wallis tests when appropriate. A Bonferroni correction was used to compensate for multiple comparisons. Results were considered to be statistically significant when the P value was below a threshold obtained by dividing 0.05 by the number of statistical tests per experiment, which corresponds to an α value of 0.05. Significance of correlations was calculated using the Spearman rank test.
Concentrations of BAs and C4 in lipoprotein fractions
To confirm previously reported observations in the presence of CA and CDCA in lipoprotein fractions, we quantified 15 BA species in lipoprotein fractions that were isolated by sequential ultracentrifugation of plasmas from healthy volunteers (Figure 1). By using SDS/PAGE and subsequent Coomassie Brilliant Blue staining, we ruled out any significant contamination of the HDL and LDL fractions with albumin as well as any contamination of LDL with ApoA-I and of HDL with ApoB (Figure 1A). In order to simplify the analysis, we grouped the BAs as primary against secondary, unconjugated against conjugated and total BAs. Based on the concentrations measured in total serum, the recoveries for C4 and the different BAs in the summarized density fractions amounted to 58% and 78–82%, respectively.
Figures 1(B)–1(E) show the distribution of the individual unconjugated, glycine- and taurine-conjugated BAs as well as the summarized BAs among the various lipoprotein fractions of normal plasma. Relative to total plasma, approximately 84%, 11% and 5% of total BAs were found in the LPDS, HDL and pooled LDL/VLDL fractions respectively. As shown in Figure 2, the proportions of conjugated as well as primary BAs were slightly increased in the HDL fraction relative to the respective proportions in other fractions and total plasma. Indeed, conjugated BAs represented 87% of all BAs in HDL as opposed to 79–81% of all BAs in other fractions and total plasma. Similarly, primary BAs represented 78% of all BAs in HDL as opposed to 71–73% in other fractions or total plasma. As the consequence, unconjugated and secondary BAs were under-represented in the HDL fraction (Figure 2). This effect seemed to be due mainly to GCDCA (glycochenodexycholic acid) (55% of all BAs in HDL against 29–47% of all BAs in other fractions and plasma) and TCDCA (taurochenodeoxycholic acid) (7% of all BAs in HDL against 4% of all BAs in plasma). GCDCA was by far the most abundant BA in all fractions, followed by GCDCA and GCA. As another interesting difference, GCA represented 25% of all BAs in the combined LDL/VLDL fraction but only 8–10% of all BAs in the other fractions and plasma. On the other hand, GCDCA represented only 29% of all BAs in the combined LDL/VLDL but 47–55% of all BAs in the other fractions and plasma.
By contrast with BAs, only a minority of C4, namely 23%, was recovered in the LPDS fraction. The majority of C4 was associated with LDL (41%), followed by HDL (21%) and VLDL (15%) (Figure 1F).
Serum levels of BAs and C4 in patients with inborn errors of HDL metabolism
In order to simplify the initial statistical analysis and to avoid too many statistical tests in the relatively small subcohorts with the different genotypes, the 15 individual BAs were grouped into primary, secondary, unconjugated, conjugated and total BAs. In addition, after correction for multiple testing, primary BA levels were higher in males than in females (P=0.008), mainly due to higher levels of CA, CDCA, GUDCA and GCDCA. There were no statistically significant correlations of age with either grouped or individual BAs, except for LCA. However, this correlation of LCA with age (r=0.285, P=0.004) must be taken with circumspection, because a significant proportion (65%) of LCA concentrations was below the LOQ (limit of quantification=0.1 μmol/l) (results not shown).
Tables 1 and 2 summarize the medians and ranges for the five BA groups as well as C4 in patients with mutations leading to low HDL-C and high HDL-C respectively, as well as all unaffected family members. To test whether HDL-C-lowering or HDL-C-increasing mutations affect BA concentrations independently of the specific genetic origin, we also evaluated the summarized data of patients carrying mutations in APOA1, ABCA1 or LCAT (Table 1) and of patients carrying mutations in SCARB1, CETP or LIPC (Table 2). Because of gender differences for primary BAs, the summarized data were further stratified by gender. Compared with all controls, neither mutations in single genes nor summarized mutations causing low HDL-C were associated with any significant difference in BA and C4 concentrations (Table 1). Of note also carriers of two defective alleles in either ABCA1 or LCAT did not present with significantly altered BA or C4 levels. A comparison of BA levels in carriers of mutations in CETP, SCARB1 or LIPC as well as unaffected controls by Kruskal–Wallis tests revealed a significant difference for conjugated BAs (P=0.034), which were lower in SCARB1 mutation carriers and higher in CETP and LIPC mutation carriers compared with controls. However, after correction for multiple testing, this difference was no longer significant (Table 2). Moreover, single comparisons of the mutation carriers with unaffected controls did not reveal any statistically significant difference. Secondary BAs were found to be decreased in SCARB1 mutation carriers compared with all controls (P=0.045) as well as in summarized male patients with HDL-C-increasing mutations compared with male controls (P=0.043). However, after correction for multiple testing these differences also did not remain statistically significant. C4 levels did not differ between mutation carriers and controls.
We also evaluated correlations between BAs and several lipid parameters, namely total cholesterol, HDL-C, non-HDL-C, LDL-C, TAG, ApoA-I, ApoB, total 27OHC and HDL-27OHC either in the entire cohort (Table 3) or in mutation-free controls only (Supplementary Table S3 at http://www.clinsci.org/cs/122/cs1220385add.htm). Only a few BA species showed statistically significant correlations with lipids, lipoproteins or apolipoproteins. All of them were moderate, with r2 values ranging between 0.04 (CDCA against cholesterol, P<0.05) and 0.12 (TCDCA against ApoB, P<0.01). In addition, of note, the majority of correlations were inverse. With the exception of significant positive correlations of C4 with ApoA-I (r2=0.077) and HDL-27OHC (r2=0.045) (both P<0.05), none of the BA parameters showed any significant correlation with any HDL-related parameter. Although ApoB containing lipoproteins transport the majority of C4 (Figure 1F), C4 did not show any significant correlation with ApoB, LDL-C, non-HDL-C, or TAG (Table 3).
BA levels in LPDS and HDL fractions from heterozygous SCARB1 mutation carriers and mutation-free controls
Owing to the presence of BAs in HDL and due to the trends of decreased levels of conjugated BAs and increased levels of secondary BAs in SCARB1 mutation carriers, we differentiated BA levels in plasma, LPDS and lipoprotein fractions from 15 heterozygous SCARB1 mutation carriers as well as 14 mutation-free controls from the same families. Concentrations of most BAs in the LDL and VLDL fractions were below LOQ (results not shown). Table 4 shows the medians and ranges for primary, secondary, unconjugated, conjugated and total BAs in both total plasma, HDL and LPDS. No difference was observed between mutation carriers and controls for any of the BA groups, either in total plasma or in HDL or in LPDS. Moreover, we analysed the composition of the BA pools associated with either fraction or total plasma as a percentage of the total BAs found in these fractions (Figure 3). The BA distribution was similar in mutation carriers and controls. In addition, BAs were similarly distributed between plasma (Figure 3A), the LPDS (Figure 3B) and HDL (Figure 3C) fractions. Primary as well as conjugated BAs constitute approximately 70–80% and 65–75% respectively of the BA pool of each fraction. Moreover, GCDCA was the most predominant BA in all cases. DCA showed dramatically increased concentrations in the fractions, but not in plasma, which was most probably due to an analytical artefact. Therefore DCA concentrations are not reported here. Moreover, some BAs were present in very low concentrations, mainly in plasma and in the HDL fractions. BAs for which more than 50% of the values were below LOQ are not reported here.
In the present study, the majority of BAs was found to reside in the LPDS fraction of plasma, thereby confirming albumin as the main transporter of BAs in the circulation [15,16]. Nevertheless, we, like others, recovered significant amounts of BAs in lipoprotein fractions, notably HDL . However, we recovered smaller proportions of BAs in HDL, namely 11% instead of 15–18% reported previously . As a new finding and surprisingly, about three-quarters of plasma C4 was retrieved in the lipoprotein fractions. However, other than for BAs, ApoB containing lipoproteins rather than HDL was identified as the main transporters of C4. Owing to their structural similarity to cholesterol, it is not surprising that BAs and C4 are associated with lipoproteins. The much higher proportion of C4 relative to BAs transported in lipoproteins may be of either physicochemical or metabolic origin: Owing to the presence of a carboxy group in unconjugated and glycine-conjugated BAs or a sulfate group in taurine-conjugated BAs, BAs are more hydrophilic than C4. In addition, C4 is exclusively of hepatic origin, whereas, due to the enterohepatic circulation, the majority of BAs in plasma are of intestinal origin [12,13]. This different origin may also explain the different preponderance of C4 in VLDL/LDL, which also are of exclusively hepatic origin, as opposed to the preponderance of BAs in HDL, which are both of hepatic and intestinal origin [25,26]. Like C4, 27OHC, which is another BA precursor, is also mainly transported by ApoB-containing lipoproteins .
Because of the association of BA with lipoproteins and because BA production is the final step in RCT, we expected BA levels to be influenced by gene defects in HDL metabolism. However, this was not observed. A few and small statistically significant differences were observed, namely for secondary BAs (between heterozygous SCARB1 mutation carriers and controls as well as between male carriers of HDL-C-increasing mutations and male controls) as well as for conjugated BAs (between all carriers of HDL-C-lowering mutations and controls) were lost after correction for multiple statistical testing. In a previous study, we observed a significant association of the metabolic syndrome with serum levels of C4, but not BAs, mainly because of the positive correlations of C4 serum concentrations with BMI (body mass index) and TAG levels . However, both in our previous study of metabolic syndrome patients and in the present study of patients with inborn errors of HDL metabolism , we did not find any significant correlation between BA parameters and HDL-related biochemical parameters. This is very much in contrast with our previous findings on 27OHC plasma levels, which are strongly influenced by functionally relevant mutations in HDL genes and strongly correlated with HDL-related biomarkers [21,27].
There are several possible explanations for the lack of associations and correlations of BA and C4 levels with disturbances of HDL metabolism and HDL-related biomarkers respectively.
First, BA concentrations show high interindividual variation, so that our study may be under-powered to find any significant association.
Secondly, BA and C4 levels fluctuate throughout daytime in response to food intake (conjugated BAs) and circadian clocks (unconjugated BAs and C4) . We paid attention to collecting blood samples in the morning and in the fasting state where most BAs and C4 reach trough levels . Nevertheless the inter-individual variation independently of genotype remained high. In this regard, it may be interesting to measure BA and C4 under provoked conditions, for example after fat load or treatment with BA sequestrants , to demonstrate any effect of disturbed HDL metabolism on BA metabolism.
Thirdly, only a low percentage of BAs are associated with HDL. Therefore the effect of mutations affecting HDL might not be sufficient to observe any significant change on total plasma levels of BAs. However, even in the case of heterozygous SCARB1 mutation carriers, who showed the largest difference in plasma levels of conjugated and secondary BAs, we could not unravel any difference in HDL-associated BAs. Moreover, increases and decreases of HDL-C do not affect all HDL subclasses. Because BA and C4 concentrations in HDL are lower than HDL particle concentrations (<1 μmol/l against approximately 20 μmol/l) only a minority of HDL particles carry these metabolites. The concentration of these subclasses may be unchanged by a given mutation.
Fourthly, genetic interferences with several HDL genes in mice were reported previously to have no effect on hepatic BA production and faecal sterol and BA excretion . By contrast, in a model that specifically records the faecal excretion of radiolabelled cholesterol from peritoneal macrophages, knockouts, transgenic expression or pharmacological interferences with several HDL genes were found to modulate the faecal excretion of both BAs and cholesterol (reviewed in [30–33]). In the light of these findings, concentrations of BAs or C4 in serum or HDL may not be able to record this specific fraction of reverse transport of macrophage-derived cholesterol.
Fifthly, the repressive effects of BAs on the expression of several HDL genes via FXR activation [18–20] is not reflected by any correlation with HDL-C or ApoA-I concentrations, either in the present study of primary defects in HDL metabolism or in our previous study of secondarily disturbed HDL metabolism in Type 2 diabetes and the metabolic syndrome . Apart from the limitations of statistical power resulting from the high degree of intra- and inter-individual variation discussed above, this lack of correlation may be due to the divergent effects of APOA1, SCARB1 and LIPC repression [18–20] on HDL-C concentrations: HDL-C levels are expected to decrease upon down-regulation of APOA1 but increase upon down-regulation of SCARB1 and LIPC.
Our findings in the family with SCARB1 mutations deserve some specific remarks, because cholesterol taken up by cells from HDL via SR-BI has been postulated to be channelled to oxidation by cytochrome P enzymes and, hence, steroidogenesis . In fact, SCARB1 mutation carriers were found previously to have decreased urinary corticosteroid excretion . The tendency of lower conjugated BA levels in SCARB1 mutation carriers relative to all controls (Table 3) may be taken as an indication that SCARB1 is a limiting factor for BA production. However, BA concentrations in plasma and HDL did not differ between mutation carriers and unaffected controls from the same family. Therefore and because C4 levels, which are considered as biomarkers of BA synthesis, did not differ either, it is unlikely that heterozygous SCARB1-deficiency limits BA production, at least under basal conditions. The lacking effect of SCARB1 mutations on C4 as a biomarker of BA synthesis is also in agreement with the finding of normal biliary BA secretion in Scarb1-knockout mice [35,36].
In conclusion, 10–20% of BAs and approximately 75% of C4 in human serum are transported by lipoproteins. Nevertheless, although being partially transported by HDL and although being an end-product of RCT, fasting serum concentrations of BAs and C4 are not changed by primary disturbances of HDL metabolism. Serum levels of BAs and C4 hence do not appear to be suitable biomarkers for measuring HDL-mediated RCT.
This work was mainly supported by the European Commission within the Sixth Framework Programme (‘HDLomics’) [grant number LSHM-CT-2006-037631]. The laboratories of J.A.K., A.T.-H. and A.v.E. are members of the European COST action ‘HDLnet’ [grant number BM0904]. A.G.H. is supported by the Netherlands Organization for Scientific Research (NOW) [project grant number 021.001.035].
Carine Steiner acquired the data and performed the statistical analyses, interpreted the results and drafted the paper. Ratna Karuna and Lucia Rohrer acquired the data and made critical revision of the paper for important intellectual content. Adriaan Holleboom, Mohammad Motazacker, Jan Albert Kuivenhoven, Ruth Frikke-Schmidt and Anne Tybjaerg-Hansen provided samples as well as demographic, clinical chemical and genetic data of the patients enrolled into the clinical study, handled the ethics and made critical revision of the paper for important intellectual content. Katharina Rentsch supervised the analytical work and made critical revision of the paper for important intellectual content. Arnold von Eckardstein conceived and designed the research, handled the funding, interpreted the results and finalized the paper.
Abbreviations: ABCA1, ATP-binding cassette transporter A1; ApoA-I, apolipoprotein A-I; ApoB, apolipoprotein B; BA, bile acid; C4, 7α-hydroxy-4-cholesten-3-one; CA, cholic acid; CDCA, chenodeoxycholic acid; CETP, cholesteryl ester transfer protein; DCA, deoxycholic acid; FXR, farnesoid X receptor; GCDCA, glycochenodeoxycholic acid; HDL, high-density lipoprotein; HDL-C, HDL-cholesterol; HL, hepatic lipase (encoded by LIPC); LCA, lithocholic acid; LCAT, lecithin:cholesterol acyltransferase; LDL, low-density lipoprotein; LDL-C, LDL-cholesterol; LOQ, limit of quantification; LPDS, lipoprotein-depleted serum; 27OHC, 27-hydroxycholesterol; RCT, reverse cholesterol transport; SR-B1, scavenger receptor class B1 (encoded by (SCARB1); TAG, triacylglycerol; TCDCA, taurochenodeoxycholic acid; UDCA, ursodeoxycholic acid; VLDL, very-low-density lipoprotein
- © The Authors Journal compilation © 2012 Biochemical Society