Charge Heterogeneity of LDL in Asymptomatic Hypercholesterolemic Men Is Related to Lipid Parameters and Variations in the ApoB and CIII Genes
Abstract—This study was carried out to examine the relationship between the charge on low density lipoproteins (LDLs) and lipid and clinical parameters in 104 asymptomatic dyslipidemic men and to identify biochemical and genetic factors that could contribute to the charge variability of LDL. LDL charge heterogeneity was evaluated by relative electrophoretic mobility (REM) on preformed 0.5% agarose gels and by chromatographic quantification of a minor electronegative LDL subfraction designated LDL(−). The mean REM value for LDL was 0.147±0.016 and the mean LDL(−) subfraction percentage was 5.6±2.8%. Both were positively correlated with common atherosclerotic risk factors, especially total cholesterol [for REM, r=0.27, P<0.005; for LDL(−), r=0.28, P=0.008] and LDL cholesterol [for REM, r=0.27, P=0.007; for LDL(−), r=0.26, P=0.01)] levels, and REM was positively correlated with triglycerides (r=0.27, P<0.005) and negatively with apoAI levels (r=−0.30, P<0.002). The variations in LDL charge were not due to oxidation, as measured by the lag phase and binding to the LDL receptor. The results of the 2 methods used to measure LDL charge were significantly correlated and had some identical characteristics (eg, association with LDL apoCIII content and plasma triglyceride levels in borderline and IIb dyslipidemic subjects); these methods reflect different specific features of LDL charge. The percentage of LDL(−) was correlated positively with the LDL sialic acid content (P<0.0001), whereas the REM was related to at least 2 distinct chromosomal loci. Multiple logistic analysis showed that individuals carrying minor alleles of BsrDI (P<0.05), apoCIII/SacI (P<0.01), as well as the frequent allele of XbaI (P<0.05) at the apoB and CIII gene loci had high REMs. This result suggests that LDL charge heterogeneity, which is positively correlated with the atherogenic lipid profile, is influenced by both genetic and biochemical factors.
- Received March 7, 1997.
- Accepted May 8, 1998.
Elevated serum concentrations of LDL cholesterol are associated with an increased risk of coronary heart disease.1 2 3 Several studies have demonstrated that LDL particles vary in size,4 density,5 chemical composition,6 and electric charge.7 Hypercholesterolemic subjects may also have an altered electrophoretic mobility of their LDL on agarose.7 8 An electronegative subfraction of LDL can be isolated by ion-exchange chromatography.9 10 This subfraction may be mildly oxidized.9 We have used fast protein-liquid chromatography (FPLC) to describe a procedure based on multistep NaCl gradient elution11 to separate a minor electronegative LDL subfraction designated LDL(−).12 Although LDL(−) is not oxidized, it differs markedly from the bulk of plasma LDL and is cytotoxic to endothelial cells.12 FPLC was also used by Holvoet et al13 to characterize malondialdehyde-modified LDL isolated from acute myocardial infarction patients and seems particularly suitable for investigations of LDL charge heterogeneity.
Charge heterogeneity of LDL may also affect the structure of the apoB protein, through substitutions of differently charged amino acids or the relative orientation of charged groups. Many forms of the apoB gene are known,14 15 16 but no causal relationship between amino acid changes in apoB and the LDL charge heterogeneity has yet been found.17 Genetic variation may directly cause the charge change, as with the MspI polymorphism that reflects the replacement of arginine by glutamine at position 3611. The 3 apolipoproteins encoded by the apoAI-CIII-AIV gene cluster include apoCIII, a component of VLDL and HDL, which is the most likely to influence the metabolism of apoB-containing lipoproteins. ApoCIII inhibits lipoprotein lipase activity in vitro18 and therefore reduces hydrolysis of triglyceride-rich particles, whereas plasma apoCIII levels are correlated with elevated triglycerides.19 Our previous studies12 20 have shown that LDL electronegativity is associated with the apoCIII content, so that changes within and around the apoCIII gene may influence LDL charge.
This study was carried out to evaluate LDL charge heterogeneity in asymptomatic dyslipidemic men, to investigate any correlations with their lipid and clinical parameters, and to look for changes in LDL that might account for the charge heterogeneity, such as those due to oxidation. Last, we have examined the biochemical and genetic factors that could influence LDL charge by determining the LDL apoE, CIII, and sialic acid contents and identifying changes in the apoB and CIII gene sequences.
Selection of Subjects
The 104 hypercholesterolemic subjects included in this study were selected by a cholesterol screening program conducted at the workplaces of several Paris companies by a group of occupational health physicians (the PCVMETRA Group: Prévention Cardiovasculaire en Médecine du Travail). The subjects selected were nondiabetic men who had total cholesterol levels >5.2 mmol/L and triglycerides <4.5 mmol/L and who had never been treated with lipid-lowering drugs. Their dyslipidemia was divided into 3 subtypes21 : borderline hypercholesterolemia (n=33), with total cholesterol levels of 5.2 to 6.19 mmol/L and triglycerides <2 mmol/L; type IIa hypercholesterolemia (n=46), with total cholesterol levels of 6.2 mmol/L or above and triglycerides <2 mmol/L; and type IIb hypercholesterolemia (n=25), with total cholesterol levels of 6.2 mmol/L or above and triglycerides of 2 mmol/L or above. Each subject was assessed for a history or symptoms of cardiovascular disease by a complete clinical examination and careful questioning. Those with cardiovascular disease were excluded from the study.22 None of the included subjects had ever undergone any coronary investigation.
Isolation of LDL
Blood was collected after a fast of at least 14 hours into EDTA-containing Vacutainers and was immediately centrifuged for 15 minutes at 2000g at 4°C to separate the plasma. LDLs (d=1.019 to 1.063 g/mL) were separated from fresh plasma (8 mL) by sequential ultracentrifugation in a Beckman L90 ultracentrifuge (Beckman Instruments) equipped with a Beckman NTV 90 rotor. LDLs were dialyzed for 24 hours at 4°C against 4 changes of 100 volumes of 10 mmol/L Tris-HCl buffer, pH 7.4, containing 1 mmol/L EDTA and were stored at 4°C in the dark.
Serum lipids (total cholesterol, triglycerides, HDL cholesterol, and LDL cholesterol) were determined by standard enzymatic methods.23 24 ApoAI and B levels were measured by immunonephelometry using a BNA analyzer and nonspecific antibodies (Behring). Lipoprotein(a) [Lp(a)] in serum and apoCIII and E in LDL were determined by noncompetitive ELISA on a Biomek 1000 analyzer (Beckman Instruments).20 25
The mobility of LDL was determined by electrophoresis on preformed 0.5% agarose gels (Beckman, Paragon Lipo kit), as described by Sparks and Phillips.26 The LDL concentration was normalized to a protein concentration of 1 mg/mL in 10 mmol/L Tris-HCl buffer, pH 7.4, containing 1 mmol/L EDTA, and fatty acid–free bovine albumin (final albumin concentration, 4 g/L) was added as a migration reference. Samples (5 μL) were applied to the gel wells and electrophoresed (100 V, Paragon power supply) for 30 minutes in Tris-barbital buffer, pH 9.2. The gels were then fixed for 5 minutes in ethanol/acetic acid/water (60:10:30, vol/vol/vol), dried, stained for 5 minutes with 0.07% Coomassie blue in ethanol/water, destained for 10 minutes in ethanol/water (45:55, vol/vol), and scanned with a densitometer (Preference, Sebia). Relative electrophoretic mobilities (REMs) were determined as the LDL to albumin migration distance ratio.
Chromatography was performed11 12 on an anion-exchange column (mono QHR 5/5) and an LCC-500 programmer controlling 2 P-500 pumps. The buffers were A, consisting of 10 mmol/L Tris-HCl, pH 7.4, containing 1 mmol/L EDTA, and B, consisting of 1 mol/L NaCl in buffer A. Buffers were degassed before use, and the system was operated at 4°C. The LDL sample (0.5 g/L LDL protein) was filtered (0.2-μm filter, Satorius), introduced into the system via a 0.5-mL loop, and eluted at a rate of 1 mL/min by a linear gradient of 0% to 20% buffer B for the first 20 minutes, followed by 30% buffer B for 5 minutes. The eluate was monitored by a single-path UV monitor at 280 nm, and the areas under the peaks were integrated with LLC-500 software. Two LDL subfractions, designated native LDL and LDL(−), were separated. The amount of LDL(−) was expressed as a percentage of total LDL. The major peak of native LDL was eluted with a theoretical NaCl concentration of 0.2 mmol/L at a retention time of 11 minutes, and a second minor peak, LDL(−), was eluted at 17 minutes with a theoretical NaCl concentration of 0.3 mmol/L.
LDL Sialic Acid Content
Susceptibility of LDL to Oxidation
The susceptibility of LDL to oxidation was determined as described elsewhere.29 LDLs were dialyzed for 24 hours against 10 mmol/L phosphate buffer, pH 7.4, containing 160 mmol/L NaCl and 10 μmol/L EDTA, and LDL was diluted 10-fold with EDTA-free phosphate buffer (final EDTA concentration, 1 μmol/L). Oxidation was initiated by adding freshly prepared CuCl2 · 2H2O solution to a final concentration of 5 μmol/L. The kinetics of LDL oxidation were monitored by the change in absorbance at 234 nm at 30°C in a spectrophotometer (Uvikon 930, Kontron) equipped with a 10-position automated sample changer, so that as many as 10 samples could be monitored simultaneously. Absorbance was recorded every 4 minutes for 4 hours. The lag phase was calculated from the oxidation profile of each LDL preparation by drawing a tangent to the slope of the propagation phase and extrapolating it to intercept the initial-absorbance axis. The lag phase represented the length of the antioxidant-protected phase during LDL oxidation by Cu2+ in vitro.
LDL Receptor Binding Assay
LDL was labeled with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI).30 The fibroblasts were human fetal lung fibroblasts (MRC5; Bio-Merieux, Marcy l’Etoile, France) cultured as monolayers in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FCS. Experiments were performed on confluent monolayers of fibroblasts (2×105 cells per 60-mm dish) plated out 5 days previously. The medium was replaced with DMEM containing 10% lipoprotein-deficient serum 2 days before the assay. The cell monolayer was washed twice with PBS, and 2 mL of DMEM containing normal human DiI-labeled LDL (5 μg LDL protein per mL) was added, together with increasing concentrations (0, 2.5, 5, 10, 20, and 50 μg/mL) of each patient’s LDL for lipoprotein competition assays. Cells were incubated at 4°C for 2 hours, washed 4× with PBS containing 0.4% BSA at 4°C and once with PBS alone; isopropanol (1 mL) was added to each well, and the plates were gently shaken for 15 minutes. Fluorescence in the isopropanol extract was determined with a spectrofluorometer (American Instrument Co), with excitation and emission wavelengths of 520 and 578 nm, respectively. Cells were dissolved in 0.2N NaOH for protein determination. The concentration of each patient’s LDL (IC50) that inhibited 50% of the DiI-LDL binding to the cell surface was calculated as the log concentration (μg protein per mL) versus probits. Probits were read from a probit transformation table.
Total protein was measured by Peterson’s method31 with BSA as the standard.
Genomic DNA was extracted from proteinase K–treated crude buffy coats by a salting-out procedure.32 All forms of polymorphism were typed by the polymerase chain reaction (PCR). DNA was analyzed for apoB restriction fragment length polymorphisms (RFLPs) with the endonuclease XbaI, EcoRI, MspI, RsaI, and BsrDI sites, all located in the coding region of the apoB gene,14 15 and with the endonuclease SacI, HphI, MspI, and PstI sites in the 5′ and 3′ noncoding regions of the apoCIII and apoAI genes.33 34 35 A 9-bp insertion/deletion (I/D) variant in the signal-peptide coding region of the human apoB gene36 and a variable number of tandem repeats (VNTRs) at the 3′ end of the apoB gene16 were also analyzed. All PCR reactions were carried out in a final volume of 20 μL containing 100 ng genomic DNA, 0.2 U Thermus aquaticus DNA polymerase (ATGC Biotechnologie), 200 μmol/L dNTP (Boehringer Mannheim), 10 pmol of each primer, and 2 μL of the amplification buffer recommended by the manufacturer (10× buffer: 500 mmol/L KCl, 15 mmol/L MgCl2, 100 mmol/L Tris-HCl, 0.1% [wt/v] gelatin; pH 9.0). The mixture was overlaid with 50 μL mineral oil (Sigma). The primer sequences and conditions used to type the 7 variants of the apoB gene and the 3 variants of the apoAI-CIII gene cluster have been reported elsewhere.15 34 35 37 38 39 The following primers were used to amplify the polymorphic sites at nucleotides −625 and −482 in the apoCIII promoter region: −625/apoCIII, 5′ (sense, A): TCA TCC TGC CAT CCC TGC TG; 5′ (antisense, B): GCC CCC ACC CTG TGT GCC TC; −482/apoCIII, 5′ (sense, C): TGC TGG GAG GGG CTG TGA GA; and 5′ (antisense, D): GCT GGT GAG AGG GGA AAT GG.
Amplification was carried out with the following cycles: initial denaturation (95°C, 5 minutes), followed by 35 cycles of denaturation (95°C, 1 minute), annealing (62°C, 1 minute with primers A and B and 54°C, 1 minute with primers C and D), and primer extension (72°C, 1 minute).
PCR was done in an automated Stratagene Robocycler gradient 96. Digestion with 10 U of XbaI, EcoRI, MspI, RsaI, PstI, and SacI and 1 U of HphI restriction nucleases was performed at 37°C overnight with 20 μL of each amplification product in the buffer recommended by the supplier (Biolabs) and with 2 U of BsrDI restriction nuclease at 60°C. The restriction products were electrophoresed in 1.5% agarose gels (Eurobio) at 80 V for 1 hour. The I/D variants of the apoB gene and the restriction products of the polymorphic site at nucleotide −625 in the apoCIII promoter region were analyzed by visual inspection after electrophoresis of 15 μL of amplification product in a 3% Resophor agarose gel (Eurobio) at 80 V for 3 hours. The VNTR 3′ alleles are referred to by their number of repeat units and were scored after electrophoresis of 20 μL of amplification product in a 2% agarose gel at 60 V overnight. The nomenclature used is that suggested by Boerwinkle et al.16
Statistical analysis was performed on an Apple Macintosh computer with JMP (SAS Institute) software. Quantitative variables are expressed as mean±SD and range. The normal distributions of electronegative charge parameters [REM and LDL(−)] were checked with the Shapiro-Wilk test. All nonnormally distributed variables were logarithmically transformed to a normal distribution, but untransformed means are shown in the Tables. LDL charge was analyzed according to the quartile values of age. Linear correlation coefficients were used to describe relationships between electronegative charge parameters and quantitative variables. Stepwise regression analysis with backward selection (ie, all terms entered into the model and the least significant terms removed until the remaining terms were significant, with a probability to enter=0.25 and a probability to leave=0.10) was used to identify links between electronegative charge parameters, taken as dependent variables, and lipid and clinical parameters, as independent variables. Gene frequencies were determined by gene counting. The χ2 test was used to test for Hardy-Weinberg equilibrium. Differences in means were examined by ANOVA. Multiple logistic analysis was used to identify the independent contributions of minor alleles of the variants in the apoB and apoCIII genes to electronegative charge parameters of LDL. The covariates included combinations of age, body mass index, diastolic blood pressure, LDL cholesterol, apoAI, triglycerides, LDL apoCIII, LDL apoE, and LDL sialic acid content. A P value <0.05 was considered statistically significant.
Table 1⇓ shows the mean frequencies of risk factors in the study group. Thirty-two percent of patients had borderline-high cholesterol concentrations, 44% had type IIa dyslipidemia, and 24% had type IIb dyslipidemia.
LDL Charge Heterogeneity and Lipid and Clinical Parameters
Figure 1⇓ shows the electrophoretic mobility of LDL isolated from the plasma of dyslipidemic patients. Within-run (n=8) and day-to-day (10 days, n=8) reproducibilities of REM determinations were 1.95% and 2.5%, respectively. The REM distribution was normal (Figure 2A⇓). The mean±SD (n) REM value was 0.147±0.016 (104), with a range of 0.112 to 0.195. The distribution of the percentage of LDL(−) was skewed (Figure 2B⇓). The mean±SD LDL(−) value was 5.6±2.8%, with a range of 1.5% to 17.0%. The percentage of LDL(−) was positively correlated with the REM value (r=0.35, P=0.007).
Figure 3⇓ shows LDL charge parameters for each subtype of dyslipidemia. REM values were significantly higher in type IIa and IIb dyslipidemias than in the borderline type, whereas the percentage of LDL(−) was higher in type IIa dyslipidemia than in type IIb and the borderline type. Hence, the REM values were positively correlated with the total cholesterol (r=0.27, P<0.005), LDL cholesterol (r=0.27, P<0.007), triglyceride (r=0.27, P<0.005), and apoB (r=0.23, P<0.02) concentrations and negatively correlated with the apoAI concentration (r=−0.30, P<0.002) (Table 2⇓). The only lipid parameters that were positively correlated with the percentage of LDL(−) were total cholesterol (r=0.28, P=0.008) and LDL cholesterol (r=0.26, P=0.014). The REM values were lower in hypertensive subjects (P<0.03), and the percentage of LDL(−) was lower in smokers (P<0.03) (data not shown).
When the 3 dyslipidemic populations were considered separately, the only significant associations were between REM and apoAI (r=−0.39, P<0.008) in group IIa and REM and triglycerides (r=0.52, P<0.007) in group IIb (Table 2⇑).
The parameters influencing the electronegative charge of LDL were identified by examining the links between charge and LDL sialic acid, apoE, and apoCIII contents. REM was correlated weakly with the LDL apoCIII content (r=0.20, P<0.05), and a similar trend was observed with LDL apoE content (r=0.19, P<0.06) for the whole population (Table 2⇑). These relations also held for borderline subjects for apoCIII and in IIb dyslipidemic subects for apoE (Table 2⇑). There was a strong, positive correlation between the percentage of LDL(−) and the LDL sialic acid content in borderline and type IIa dyslipidemic subjects.
The stepwise regression analysis with backward selection (Table 3⇓), with LDL charge heterogeneity as the dependent variable and lipid and clinical parameters as independent variables, showed that the electronegative parameters of LDL were significantly associated with the LDL apoCIII content in borderline subjects and significantly associated with the plasma triglyceride levels in type IIb dyslipidemic subjects. There was also a significant correlation between the percentage of LDL(−) and the LDL sialic acid content in the overall population and in each dyslipidemic group. The independent association of LDL charge heterogeneity with LDL cholesterol and apoAI levels within the whole population was not consistent when each lipid group was analyzed.
LDL Charge Heterogeneity and Changes due to Oxidation
We looked for a positive association between LDL charge heterogeneity and change due to oxidation. The mean±SD lag phase was 144±20 minutes (range, 105 to 195). There was no correlation between the length of the lag phase and the electronegative parameters of LDL, except in type IIa dyslipidemics, for whom there was a trend with the percentage of LDL(−) (r=−0.29, P=0.08). This correlation was lost in multivariate analysis. The electronegative parameters of LDL were then compared with the binding affinity for the LDL receptor of MRC5 fibroblasts. LDL binding studies were performed on samples from 23 hypercholesterolemic subjects representative of the whole population. The result, expressed as the IC50, was a mean±SD of 10.68±3.47 μg/mL (range, 3.22 to 16.36). There was no link between the IC50 and REM or the percentage of LDL(−) in univariate analysis.
LDL Charge Heterogeneity and Genetic Factors
Table 4⇓ shows the estimated allele frequencies for the variants of the apoB, AI, and CIII genes. The distribution of each diallelic genotype was in Hardy-Weinberg equilibrium. There were 10 alleles of the 3′ VNTR region in the apoB gene, ranging from 31 to 51 repeats of the 15-bp sequence and displaying a bimodal frequency distribution (maxima at 37 and 49 and minimum at 43 repeat units). The 3′ VNTR alleles were divided into small (S≤43 repeats) and large (B>43 repeats) alleles.16 The distributions of the small and large alleles were in Hardy-Weinberg equilibrium (Table 4⇓).
We next examined the influence of selected genetic markers on the charge heterogeneity of LDL. Table 5⇓ shows the LDL charge parameters for each genotype at the apoB and CIII loci. There was a statistically significant association between elevated REM values and the presence of MspI, BsrDI, I/D, and 3′ VNTR B/S minor alleles and the absence of the XbaI cutting site at the apoB locus. There was also a statistically significant association between elevated REM levels and the minor alleles of polymorphic sites −625 (P=0.04) and SacI (P=0.0001) at the apoCIII locus. There was a weak correlation between the percentage of LDL(−) and the I/D polymorphism of the apoB gene. The hypercholesterolemic men homozygous for the I allele had higher levels of LDL(−) (P=0.04) than did individuals homozygous or heterozygous for the D allele.
Logistic regression analysis of variants of the apoB and CIII genes with the electronegative charge parameters of LDL was performed by using the variables that were significantly different in the univariate analysis. REM values were significantly elevated in carriers of the XbaI (−/−) allele (P=0.05) and in carriers of the minor alleles BsrDI (P=0.05) and apoCIII/SacI (P=0.01). The association between the I/D polymorphism of the apoB gene and the percentage of LDL(−) observed in univariate analysis was lost after adjustment for age, body mass index, LDL cholesterol, and LDL sialic acid content.
The relationships between XbaI, BsrDI, and apoCIII/SacI site variants and REM were further analyzed by construction of several combinations. Carriers of a combination defined by the XbaI (−/−) allele and the minor alleles BsrDI and apoCIII/SacI had significantly higher levels of REM (P<0.0001, univariate analysis) than did carriers of combinations with 1 or no rare alleles. This relation was also statistically significant in a multivariate model after adjustment for LDL cholesterol, triglyceride, apoAI, LDL apoCIII, and LDL apoE levels (P=0.005).
LDL electronegativity has often been proposed as a feature of atherogenic LDL, but the various methods used to assess charge heterogeneity do not all provide the same type of information. We have evaluated LDL charge heterogeneity in hypercholesterolemic asymptomatic subjects by electrophoretic mobility on agarose gel and by anion-exchange chromatography based on FPLC.
An increased LDL electronegativity was associated with an atherogenic lipid profile. REM was positively correlated with the total cholesterol, LDL cholesterol, apoB, and triglyceride contents and negatively with apoAI in the whole population, whereas the percentage of LDL(−) was correlated with total and LDL cholesterol. As previously reported,7 there were significant differences in the charge characteristics of the LDL in different lipid groups of human subjects. The importance of triglyceride concentration for determining LDL charge was confirmed in type IIb dyslipidemia. The triglyceride enrichment might alter the lipid core and surface of LDL and affect apoB conformation.40 These alterations could then influence the net electric charge of LDL, as reported for HDL particles.41 42
Early in our analysis, we looked for changes in LDL due to oxidation that might account for LDL charge heterogeneity. We measured the susceptibility of LDL to oxidation by copper, as this modification is easy to perform and gives reproducible results. In agreement with our previous studies,12 20 oxidation does not appear to be the main change responsible for differences in LDL charge. However, the measuring of lipid peroxidation products (hydroperoxides and cholesterol oxides) may have revealed slight oxidation of LDL, which might account for some of the net electronegativity. Sevanian et al9 found that LDL(−), isolated according to Cazzolato et al,44 contained more lipid peroxides than did normal LDL from normolipidemic subjects. This apparent difference could be due to the preparation, handling, and storage of lipoproteins20 but remains to be explored. The increased negative charge of LDL particles does not appear to alter their binding to LDL receptors. However, a more detailed analysis of receptor binding affinity by Scatchard analysis for each of the dyslipidemic groups is needed, and this result must be viewed with caution.
The increased electronegativity of LDL could be due to a higher sialic acid content. In agreement with LaBelle et al,45 we found that the increased mobility of LDL on agarose gels was not influenced by the sialic acid content. However, there was a strong association between the percentage of LDL(−) and the LDL sialic acid content, both in the whole population and in each of the 3 dyslipidemic groups, strengthening our earlier observations.12
Because the apoCIII in plasma exists in several sialylated isoforms,46 47 the independent association between the LDL apoCIII content and the LDL charge may be 1 explanation for the increased sialic acid content of LDL from subjects with moderate hypercholesterolemia. The enrichment of LDL(−) in sialic acid content in other dyslipidemic groups could also be explained by the transport of different amounts of gangliosides, which contain 1 or more residues of sialic acid. Gangliosides are found in lipoproteins and predominantly in the LDL fraction in serum.48 It is unlikely that variations in the Lp(a) content contributed to the LDL charge differences, because there was no correlation between the percentage of LDL(−) and the Lp(a) content.
The charge heterogeneity of LDL could also be due to variations in the genes coding for their protein components. We have found evidence for an association between the electrophoretic behavior of LDL and changes at distinct chromosomal loci, the apoB and apoCIII genes. Because apoB is the principal structural apolipoprotein of LDL, variations in the apoB gene may result in differences in the electric charge of LDL between individuals. Many studies have investigated the contribution of variations in the apoB locus to plasma lipid levels and coronary heart disease,49 50 51 52 53 but they have given differing results.54 55 We have found that LDL REM was significantly higher in carriers of the XbaI (−/−) allele and the BsrDI minor allele. The XbaI RFLP showed a codominant relationship with REM. Because the base change that generates the XbaI RFLP occurs in the third base of codon 2488, this “silent” polymorphism per se cannot explain the change in the electrophoretic behavior of LDL.56 However, it could be in linkage disequilibrium with a causal mutation that could alter the tertiary structure and relative orientation of charged groups. The minor allele of BsrDI RFLP was also associated with an elevated REM. This polymorphism was recently described by Ilmonen et al15 and causes an AAT to AGT change in codon 1887 (resulting in the substitution of arginine for serine). The loss of a basic amino acid could lead to changes in the apoB surface, including a charge change, as shown by the impaired immunoreactivity of LDL apoB with Mab D7.2.15
The LDL REM was also associated with a variation at the apoCIII locus. There was a significant increase in LDL REM in carriers of the apoCIII SacI minor allele. Because it lies in the 3′ untranslated region, the SacI site is itself unlikely to be of functional significance, but it may be in disequilibrium with an as-yet-unknown functional mutation. The promoter region, a primary site of transcriptional control, is an excellent candidate site for this mutation. Dammerman et al57 identified 5 polymorphic sites in strong disequilibrium with each other and with the SacI site. We examined polymorphic sites at nucleotides −625 and −482 in the apoCIII promoter region to determine whether any of these markers affected LDL charge heterogeneity. A significant linkage disequilibrium between these 5′ sites and the 3′ SacI site was found, but none of them was significantly associated with LDL charge heterogeneity in multivariate analysis. Consequently, the relationship between the electrophoretic mobility of LDL and the apoCIII SacI site is not attributable to functional variations in the 5′ regulatory region of the apoCIII locus.
We also investigated the contribution of 2 other variants in the apoAI-CIII-AIV gene cluster to the electrophoretic mobility of LDL: a guanine to adenine substitution in the promoter of the apoAI gene at position −78 bp35 and a PstI site in the intergene region of apoAI and apoCIII.34 No association was found between the electrophoretic mobility of LDL and any of the apoAI variants analyzed. Hence, the 3′ region of apoCIII could influence protein synthesis, as suggested by Surguchov et al.57A The association between the SacI site and hypertriglyceridemia (P<0.04, data not shown) is widely supported.58 59 Genetic changes may influence triglyceride metabolism, as shown by studies in vitro and in transgenic mice. Increased apoCIII may also impair lipolysis of triglyceride-rich lipoproteins60 61 and interfere with receptor-mediated clearance of lipolytic remnant particles.62 ApoCIII transgenic animals have lipoproteins enriched in apoCIII, and their elimination is impaired,61 63 64 resulting in the accumulation of triglyceride-rich remnant particles and hypertriglyceridemia. The apoCIII gene could affect LDL metabolism by causing a change in lipid content. Alterations in lipid composition can alter the exposure of charged amino acid residues of apoB40 and also the diameter relative to negative charge, resulting in an increased surface charge density.45
This study thus provides evidence that the electronegative charge of LDL is associated with an atherogenic lipid profile. Although the results from the 2 methods used to measure LDL charge were correlated, these methods reflect different specific features of LDL. The REM of LDL particles is at least partly associated with several genetic loci, in particular the 3′ SacI site of the apoCIII gene, whereas the percentage of LDL(−) is mainly influenced by the sialic acid content of the particles. This difference might reflect differences in the charge assessed by these 2 methods. Electrophoretic mobility on agarose gel estimates the surface charge density of LDL, whereas anion-exchange chromatography based on FPLC reflects the number of negative charges on the particle surface. The origin of the influences of biochemical and genetic factors on LDL charge remains to be established.
The authors thank the staff of the PCVMETRA Group for their help: P. Segond (Chairman), D. Badet, C. Baylac-Lebot, P. Bonneau, A. de Bonnières, A. Borie, M.F. Bourillon, J. Boursier, S. Bressler, M. Bru, M. Chenet, P. Corteel, C. Coulange, C. Delmotte-Devocelle, B. Demure, M.T. Douguet, M. Dubost, T. Drumare, D. Estève, M. Fragny, O. Galamand, A.M. Giard, R. Gitel, C. Guilbert, H. Hafe, F. Kiesgen, E. Lamothe, C. Lanoiselée, M.L. Leblanc, N. Le Chevanton, I. Leprince, A. Marty, D. Miara, B. Millet, J. Oziel, A. Parini, M.C. Pasteau, M Picard, M.M. Pupponi, C. Quinio, F. Raulet, M.L. Rocca, F. Szabason, P. Taine, M.C. Tardieu, C. Tarin, A. Touati-Lumbroso, and L. Troudet. We also thank Dr V. Atger for assistance and Dr O. Parkes for checking the English text.
Castelli WP, Garrison RJ, Wilson PWF, Abbott RD, Kalousdian S, Kannel WB. Incidence of coronary heart disease and lipoprotein cholesterol levels: the Framingham study. JAMA. 1986;265:2835–2838.
Shen MS, Krauss RM, Lindgren FT, Forte TM. Heterogeneity of serum low density lipoproteins in normal human subjects. J Lipid Res. 1981;22:236–244.
Ghosh S, Basu MK, Schweppe JS. Charge heterogeneity of human low density lipoprotein (LDL). Proc Soc Exp Biol Med. 1973;142:1322–1325.
Hoeg JM, Papadopoulos NM, Gregg RE, Brewer HB. Heterogeneity of lipoprotein electrophoretic patterns in patients with type IIa hyperlipoproteinemia. Clin Chem. 1983;29:1459–1462.
Sevanian A, Hwang J, Hodis H, Cazzolato G, Avogaro P, Bittolo-Bon G. Contribution of an in vivo oxidized LDL to LDL oxidation and its association with dense LDL subpopulations. Arterioscler Thromb Vasc Biol. 1996;16:784–793.
Shimano H, Yamada N, Ishibashi S, Mokuno H, Mori N, Gotoda T, Harada K, Akanuma Y, Murase T, Yazaki Y, Takaku F. Oxidation-labile subfraction of human plasma low density lipoprotein isolated by ion-exchange chromatography. J Lipid Res. 1991;32:763–773.
Vedie B, Myara I, Pech MA, Maziere JC, Maziere C, Caprani A, Moatti N. Fractionation of charge-modified low density lipoproteins by fast protein liquid chromatography. J Lipid Res. 1991;32:1359–1369.
Demuth K, Myara I, Chappey B, Vedie B, Pechamsellem MA, Haberland ME, Moatti N. Cytotoxic electronegative LDL subfraction is present in human plasma. Arterioscler Thromb Vasc Biol. 1996;16:773–783.
Holvoet P, Perez G, Zhao Z, Brouwers E, Bernar H, Collen D. Malondialdehyde-modified low density lipoproteins in patients with atherosclerotic disease. J Clin Invest. 1995;95:2611–2619.
Priestley L, Knott T, Wallis S, Powell L, Pease R, Simon A, Scott J. RFLP for the human apolipoprotein B gene: I-V. Nucleic Acids Res. 1985;18:6789–6793.
Ilmonen M, Helio T, Butler R, Palotie A, Pietinen P, Huttunen JK, Tikkanen MJ. Two new immunogenetic polymorphisms of the ApoB gene and their effect on serum lipid levels and responses to changes in dietary fat intake. Arterioscler Thromb Vasc Biol. 1995;15:1287–1293.
Boerwinkle E, Xiong W, Fourest E, Chan L. Rapid typing of tandemly repeated hypervariable loci by the polymerase chain reaction: application to the apolipoprotein B 3′ hypervariable region. Proc Natl Acad Sci U S A.. 1989;86:212–216.
Humphries SE. Life style, genetic factors and the risk of heart attack: the apolipoprotein B gene as an example. Biochem Soc Trans. 1993;21:569–582.
Wang CS, McConathy WJ, Kloer HU, Alaupovic P. Modulation of lipoprotein lipase activity by apolipoproteins: effect of apolipoprotein C-III. J Clin Invest. 1985;75:383–390.
Schaeffer EJ, Levy RI. Pathogenesis and management of lipoprotein disorders. N Engl J Med. 1985;20:1300–1310.
Megnien JL, Sene V, Jeannin S, Hernigou A, Plainfosse MC, Merli I, Atger V, Moatti N, Levenson J, Simon A, PCVMETRA group. Coronary calcification and its relation to extracoronary atherosclerosis in asymptomatic hypercholesterolemic men. Circulation. 1992;85:1799–1807.
Giral P, Filitti V, Levenson J, Pithois-Merli I, Plainfosse M-C, Mainardi C, Gold A, Simon A, PCVMETRA group. Relation of risk factors for cardiovascular disease to early atherosclerosis detected by ultrasonography in middle-aged normotensive hypercholesterolemic men. Atherosclerosis. 1990;85:151–159.
Cambillau M, Simon A, Amar J, Giral Ph, Atger V, Segond P, Levenson J, Merli I, Megnien JL, Plainfosse MC, Moatti N, PCVMETRA group. Serum Lp(a) as a discriminant marker of early atherosclerotic plaque at three extracoronary sites in hypercholesterolemic men. Arterioscler Thromb. 1992;12:1346–1352.
Sparks DL, Phillips MC. Quantitative measurement of lipoprotein surface charge by agarose gel electrophoresis. J Lipid Res. 1991;33:123–130.
Chappey B, Myara I, Giral P, Kerharo G, Plainfosse MC, Levenson J, Simon A, Moatti N, PCVMETRA group. Evaluation of the sialic acid content of LDL as a marker of coronary calcification and extracoronary atherosclerosis in asymptomatic hypercholesterolemic subjects. Arterioscler Thromb Vasc Biol. 1995;15:334–339.
Warren L. The thiobarbituric assay of sialic acids. J Biol Chem. 1959;234:1971–1975.
Kleinveld HA, Hak-Lemmers HLM, Stalenhoef AFH, Demacker PNM. Improved measurement of low-density lipoprotein susceptibility to copper-induced oxidation: application of a short procedure for isolating low-density lipoprotein. Clin Chem. 1992;38:2066–2072.
Stephan ZF, Yurachek EC. Rapid fluorometric assay of LDL receptor activity by DiI-labeled LDL. J Lipid Res. 1993;34:325–330.
Miller SA, Dykes DD, Polesky HF. A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res. 1988;16:1215.
Juvonen T, Savolainen MJ, Kairaluoma MI, Lajunen LHJ, Humphries SE, Kesaniemi YA. Polymorphisms at the apoB, apoA-I, and cholesteryl ester transfer protein gene loci in patients with gallbladder disease. J Lipid Res. 1995;36:804–812.
Xu C, Tikkanen MJ, Huttunen JK, Pietinen P, Bütler R, Humphries S, Talmud P. Apolipoprotein B signal peptide insertion/deletion polymorphism is associated with Ag epitopes and involved in the determination of serum triglyceride levels. J Lipid Res. 1990;31:1255–1261.
Soria LF, Ludwig EH, Clarke HRG, Vega GL, Grundy SM, McCarthy BJ. Association between a specific apolipoprotein B mutation and familial defective apolipoprotein B-100. Proc Natl Acad Sci U S A. 1989;86:587–591.
Pagani F, Sidoli A, Giudici GA, Barenghi L, Vergani C, Baralle FE. Human apolipoprotein A-I gene promoter polymorphism: association with hyperalphalipoproteinemia. J Lipid Res. 1990;31:1371–1377.
Dallinga-Thie GM, Bu XD, Trip MV, Rotter JI, Lusis AJ, Debruin TWA. Apolipoprotein A-I/C-III/A-IV gene cluster in familial combined hyperlipidemia: effects on LDL-cholesterol and apolipoproteins B and C-III. J Lipid Res. 1996;37:136–147.
McNamara JR, Small DM, Li Z, Schaefer EJ. Differences in LDL subspecies involve alterations in lipid composition and conformational changes in apolipoprotein B. J Lipid Res. 1996;37:1924–1935.
Masson D, Athias A, Lagrost L. Evidence for electronegativity of plasma high density lipoprotein-3 as one major determinant of human cholesteryl ester transfer protein activity. J Lipid Res. 1996;37:1579–1590.
Davidson WS, Sparks DL, Lund-Katz S, Phillips MC. The molecular basis for the difference in charge between pre-β and α-migrating high density lipoproteins. J Biol Chem. 1994;269:8959–8965.
Deleted in proof.
LaBelle M, Blanche PJ, Krauss RM. Charge properties of low density lipoprotein subclasses. J Lipid Res. 1997;38:690–700.
Ito Y, Breslow JL, Chait BT. Apolipoprotein C-IIIo lacks carbohydrate residues: use of mass spectrometry to study apolipoprotein structure. J Lipid Res. 1989;30:1781–1787.
Mahley RW, Innerarity TL, Rall SC, Weisgraber KH. Plasma lipoproteins: apolipoprotein structure and function. J Lipid Res. 1984;25:1277–1294.
Friedl W, Ludwig EH, Paulweber B, Sandhofer F, McCarthy BJ. Hypervariability in a minisatellite 3′ of the apolipoprotein B gene in patients with coronary heart disease compared with normal controls. J Lipid Res. 1990;31:659–665.
Dammerman RM, Sandkuijl LA, Halaas JL, Chung W, Breslow JL. An apolipoprotein CIII haplotype protective against hypertriglyceridemia is specified by promoter and 3′ untranslated region polymorphisms. Proc Natl Acad Sci U S A. 1993;90:4562–4566.
Surguchov AP, Page GP, Smith L, Patsch W, Boerwinkle E. Polymorphic markers in apolipoprotein C-III gene flanking regions and hypertriglyceridemia. Arterioscler Thromb Vasc Biol. 1996;16:941–947.
Rees A, Stocks J, Shoulders CC, Galton DJ, Baralle FE. DNA polymorphism adjacent to human apolipoprotein A-I gene: relation to hypertriglyceridaemia. Lancet. 1983;26:444–446.
Ito Y, Azrolan N, O’Connell AS, Walsh A, Breslow JL. Hypertriglyceridemia as a result of human apo CIII gene expression in transgenic mice. Science. 1990;249:790–793.
Windler E, Havel RJ. Inhibitory effects of C apolipoproteins from rats and humans on uptake of triglyceride-rich lipoproteins and their remnants by perfused rat liver. J Lipid Res. 1985;26:556–565.
de Silva HV, Lauer SJ, Wang J, Simonet WS, Weisgraber KH, Mahley RW, Taylor JM. Overexpression of human apolipoprotein C-III in transgenic mice results in an accumulation of apolipoprotein B48 remnants that is corrected by excess apolipoprotein E. J Biol Chem. 1992;269:2324–2335.
Aalto-Setala K, Fisher EA, Chen X, Chajek-Shaul T, Hayek R, Zechner R, Walsh A, Ramakrishnan R, Ginsberg HN, Breslow JL. Mechanism of hypertriglyceridemia in human apolipoprotein C-III transgenic mice. J Clin Invest. 1992;90:1889–1900.