A Common Mutation in the Lipoprotein Lipase Gene Promoter, −93T/G, Is Associated With Lower Plasma Triglyceride Levels and Increased Promoter Activity In Vitro
Abstract Single-strand conformational polymorphism analysis of the lipoprotein lipase promoter identified a T→G transition at position −93. The frequency in healthy white men was 3.4% (n=1575). There was an 83% allelic association between −93T→G and Asp9→Asn (D9N); all N9 mutations occurred on a −93G allele, but not all −93G mutations occurred on an N9 allele. It was thus possible to assess the effect on plasma triglyceride (Tg) levels of the rare −93G mutation in the presence of the wild-type D9. Carriers of the −93G, with genotype TG/DD, had significantly lower Tg levels than TT/DD individuals (1.36 versus 1.78 mmol/L, P=.01); carriers of both mutations (TG/DN) had the highest Tg levels (1.93 mmol/L). When the group was stratified above and below the sample mean for body mass index (BMI), carriers of the −93G on a D9 allele (TG/DD) were “protected” against the Tg-raising effect of obesity, as assessed by BMI. In Afro-Caribbeans (n=91), the carrier frequency of −93G was 18-fold higher (63%), with weaker (17%) allelic association between −93G and N9. In vitro, the −93G promoter had 24% higher activity than the −93T in a rat smooth muscle cell line and 18% higher activity in a human adrenal cell line. A protein identified by band-shift assays bound to the −93G but not to the −93T allele, which may explain the lower Tg levels in −93G carriers.
- Received October 14, 1996.
- Accepted January 15, 1997.
There is now strong support from a number of epidemiological studies that elevated levels of plasma Tg act as an independent risk factor for coronary heart disease.1 2 3 4 LPL plays a central role in Tg metabolism; it is synthesized in parenchymal cells, primary adipose tissue, and skeletal muscle and transported to the vascular endothelium, where it binds to heparan sulfate proteoglycans and hydrolyzes Tg to free fatty acids and glycerol. The involvement of LPL in both Tg hydrolysis and uptake of lipoproteins by the LDL receptor–related protein is now well documented.5 Thus, variation in the LPL gene, either in the coding region or regions that regulate LPL expression, could directly influence plasma Tg levels and lipoprotein catabolism.
More than 50 mutations in the coding region, or splice mutations in the LPL gene, have been identified in individuals with LPL deficiency, giving rise to familial LPL deficiency (type I hyperlipoproteinemia).6 7 8 This is a rare recessive disorder, which usually presents in childhood and occurs at a frequency of 1 in 106. There is some evidence that heterozygous family members may have elevated Tg levels and present with the lipid profile of familial combined hyperlipidemia.9 However, since these mutations are rare, with a carrier frequency of 1 in 103, their impact at a population level is small.
Recently we and others have identified two common mutations in the LPL gene, an aspartic acid to asparagine change at amino acid 9, D9N, and an asparagine to serine change at amino acid 291, N291S.10 11 12 Unlike the mutations that are associated with familial chylomicronemia, which greatly reduce or abolish LPL activity, these mutations only partially reduce LPL activity. In the UK and other countries in Europe, both mutations occur at a frequency of 1% to 4% in healthy individuals; N9 occurs at increased frequency in patients with combined hyperlipidemia or coronary artery disease.10 11 Both mutations are associated with higher Tg and lower HDL levels in carriers, and this problem is compounded in individuals with a raised BMI.10 12 13 14
The aim of the present study was to identify common mutations in the promoter of the LPL gene that may be associated with modulation of LPL gene expression. The promoter of the LPL gene has been well characterized,15 16 and deletion mapping has identified a number of transcription factor binding sites detailed in Fig 1⇓. Recently (1995), Tanuma et al17 identified a number of LSEs in the region −225 to −81, and their data suggest that these elements might play a role in suppressing LPL transcription. Thus, within the promoter, there are a number of potential sites where sequence differences could modulate LPL gene expression.
Using SSCP analysis, a common T→G transition was found at nucleotide −93 of the LPL promoter. High-throughput screening was used to identify carriers of this mutation in healthy middle-aged men participating in NPHSII and in a sample of healthy Afro-Caribbeans and Gujarati Indians. The frequency and impact of the −93G mutation on plasma Tg levels and its effect on LPL gene expression in vitro are presented.
All groups of individuals examined in this study have been described in detail elsewhere. The initial search for mutations using SSCP screening was carried out on DNA samples from healthy whites (NPHSII),14 some of whom had been used previously in the identification of mutations in the coding region of the LPL gene,10 and from Afro-Caribbeans and Gujarati Indians (Brent and Harrow Study).18
Specific screening for the −93T/G mutation was undertaken in DNA samples obtained through two separate studies: (1) a total of 1575 healthy male subjects aged 40 to 64 years, all of whom had been recruited as part of the NPHSII project and were free of coronary artery disease at the time of entry into the study, as assessed by questionnaire and electrocardiography14 and (2) a total of 91 Afro-Caribbeans (51 men and 40 women) for genotype analysis and 70 Gujarati Indians (38 men and 36 women), recruited as part of the Brent and Harrow Study.18
Cholesterol and Tg concentrations were determined by standard colorimetric methods.14
Five overlapping oligonucleotide pairs (numbered 1 through 5) were designed for PCR amplification of the LPL gene from nucleotide +226 to −489 (a 715-bp fragment), presented in Fig 1⇑. PCR was carried out on a Hybaid Omnigene machine. The optimum amplification conditions and the sequence of the amplification primers (Genosys) are given in Table 1⇓. The reactions were carried out in a standard buffer, supplied by GIBCO-BRL, with 100 ng of each primer, 0.05% W-1, and 0.5 U Taq polymerase (GIBCO-BRL) per reaction, with a final MgCl2 concentration as presented in Table 1⇓.
Detection of the −93T/G mutation was carried out by PCR with oligonucleotides LPLPRO 3 and 4. For screening of the mutation, DNA, predried onto the microtiter plate, was used, producing a PCR fragment of 203 bp. The −93G introduces an Hae III site, and after digestion, fragment sizes are 153 bp and 50 bp. These fragments were separated by 7.5% polyacrylamide gel electrophoresis, using MADGE.21
Amplification of exon 2 and detection of the D9N mutation by Taq I digestion was performed as previously described.10 Subsequently, a replacement 3′ oligonucleotide (5′AGGGCAAATTTACTTGCGATG3′) was designed to produce a smaller PCR fragment and eliminate the constant Taq I site, thus producing Taq I digestion fragments of 76 bp in the presence of the N9 mutation and 52 bp and 24 bp fragments in the absence of the mutation (D9). The smaller digestion products make it possible to resolve the fragments on a 7.5% polyacrylamide MADGE gel and thus speeded up the screening process.
Previously, 773 individuals from NPHSII had been screened for the D9N mutation.10 An additional 802 were now screened in addition to the Brent and Harrow Afro-Caribbean and Gujarati Indian group.
SSCP analysis was carried out essentially as described by Gudnason et al,22 except that 0.25 μL of [32P]α-dCTP at 10 μCi/μL (3000 mCi/mmol; Amersham) was added to each sample to be amplified and the dCTP was reduced to 0.02 mmol/L. Direct sequencing of variants detected by SSCP was carried out with one of the primers used in the amplification reaction and a second biotinylated oligonucleotide, thus allowing the purification of single-stranded DNA with streptavidin-coated beads (Dynal) and sequencing according to the Sequenase protocol (USB).
A10 cells, a rat smooth muscle cell line (a kind gift from Dr O. Yalkinoglu, Bayer AG, Wuppertal, Germany) were grown at 37°C in 5% CO2 in Dulbecco’s modified Eagle’s medium containing 10% FCS, 100 U/500 mL penicillin, 100 U/500 mL streptomycin, 25 mmol/L HEPES, 1 g/L glucose, 2 mmol/L sodium pyruvate, and 1 mmol/L l-glutamine. The medium was changed every 3 days. NCIH295 cells, a human adrenal cell line expressing LPL, were maintained in RPMI-1640 medium supplemented with 10 μL/mL insulin-transferrin sodium selenite, 1 nmol/L hydrocortisone, and 1 nmol/L β-estrogen as described previously.23
The luciferase gene was used as a reporter to test the promoter activity of the −93T/G variants. A 357-bp fragment (from the translation start site to −161 bp) of the human LPL promoter was amplified from individuals homozygous for either the −93T or −93G allele, using oligonucleotides designed to give Sac I (5′) and HindIII (3′) cutting sites. The digested fragment was cloned immediately upstream of the luciferase coding region into Sac I/HindIII digested luciferase enhancer vector, pGL3-enhancer (Promega). After cloning, integrity of the insert was tested by sequencing on an ABI Prism 377 DNA sequencer (Applied Biotechnology Inc).
Transient transfections of A10 cells were carried out by using the Lipofectin liposomal reagent (GIBCO-BRL) with 5 μg pGL3-enhancer vector containing promoter insert and 1 μg of a β-galactosidase–expressing control vector. NCIH295 cells were transiently transfected in the same manner, except 2.5 μg of pGL3-enhancer was used. A10 and NCIH295 cells were harvested after a further 48 and 72 hours, respectively, with Promega cell-culture lysis reagent. The extract was stored at −70°C until assayed for luciferase activity, β-galactosidase activity, and total protein.
The Promega luciferase assay employing coenzyme A as an alternative substrate for luciferase was used. Luciferase assay reagent and cell extract were allowed to equilibrate to room temperature for 30 minutes of nonuse. To assay for luciferase, 20 μL of luciferase assay reagent was added to 4 μL of cell extract and mixed by pipetting. Luminosity was measured over a 10-second period in a Turner Designs TD-20 luminometer (Promega).
Cell extract (150 μL) was mixed with 150 μL of assay buffer (120 mmol/L Na2HPO4, 80 mmol/L NaH2PO4, 2 mmol/L magnesium chloride, 100 mmol/L β-mercaptoethanol, and 1.33 mg/mL o-nitrophenyl β-d-galactopyranoside), vortexed briefly, and incubated overnight at 37°C. Five hundred microliters of 1 mol/L sodium carbonate was added to stop the reaction, and 1 mL of purified water was added to dissolve the precipitate formed due to the presence of cell-culture lysis buffer. β-Galactosidase activity was measured as A420 and used as a measure of transfection efficiency.
Diluted cell extract (100 μL) was used in 5 mL of assay reagent (for 1 L: 100 mg Coomassie blue G250, 50 mL 95% ethanol, 100 mL 85% phosphoric acid, and 850 mL distilled water). A595 was used as a measure of protein concentration, which was obtained by comparison with a BSA standard curve. Protein concentration was used as a measure of cell number in the extraction.
Nuclear extracts used in the bandshift assay were prepared by snap-freezing a pellet of A10 cells (2×106 cells) at −70°C and lysing in five times the volume of whole-cell extraction buffer containing 10 mmol/L HEPES (pH 7.9), 0.4 mol/L NaCl, 1.5 mmol/L MgCl2, 0.1 mmol/L EGTA, 0.5 mmol/L DTT, 5% glycerol, and 0.5 mmol/L PMSF. The protein concentration of the extract was determined by the Bradford assay and adjusted to 1.5 μg/μL by dilution with whole-cell extraction buffer. Aliquots were stored at −70°C until required.
Oligonucleotides for the T or G probe were as follows: −93T AGTGAATTTAGGTCCCTCCCCCCAA and −93G AGTGAATTTAGGGCCCTCCCCCCAA.
Complementary oligonucleotides were also synthesized. Oligonucleotides were labeled using Ready-To-Go T4 polynucleotide kinase (Pharmacia Biotech) in a 50-μL reaction.
Bandshift assays were performed as described in the Pharmacia LKB BandShift kit (Pharmacia Biotech).
Computation and Statistical Analysis
Nuclear factor binding sites were screened for using the website http://bimas.dcrt.nih.gov/molbio/signal. Biometric data previously obtained for each of the studies were used for comparison of carriers with noncarriers. The gene-counting method with a χ2 test with Yates’ correction was used to compare the frequency of the −93G variant allele between the different groups. Linkage disequilibrium between the two variant sites (−93T/G and D9N) was estimated by using the correlation coefficient Δ.24 All other tests and transformations were performed with the SPSS/PC+ statistical package. A maximum-likelihood algorithm method was used to determine the most likely “phase” of the −93T/G and D9N mutations in those individuals heterozygous for both.25 To test differences in Tg concentrations, values were adjusted for age (for NPHSII) and log transformed before statistical analysis. The method of Welch et al,26 which tests for equality of means without assuming equal variances in each group, was used in preference to an ANOVA, since there was a significant difference in the variance in Tg levels among the three genotype groups in NPHSII (P=.03). Since the subjects contributed differently to the overall mean according to how many Tg measurements they had had, the subjects’ overall Tg mean was weighted by the number of measurements. Statistical significance was considered to be P=.05.
SSCP screening for mutations in the promoter of the LPL gene (from the start of exon 1 to nucleotide −489) was performed, using five overlapping PCR fragments (Fig 1⇑) in healthy individuals of white, Afro-Caribbean, and Gujarati Indian origin. A common variant band was identified in two SSCP reactions, from overlapping PCR products (2 and 3). The fact that the variant band was seen in the same samples in both reactions suggested that the mutation occurred in the region of overlap. One individual was homozygous for the SSCP. Direct sequencing of DNA from the homozygous individual and an individual heterozygous for the SSCP, using an internal sequencing primer for both PCR products, identified that the mutation, a T→G substitution at position −93, was found in both PCR fragments, in the overlapping region of the two PCR products. This T→G transition creates an Hae III cutting site and also destroys an Ava II site.
High-throughput genotyping, using small-scale PCR and Hae III digestion, resolved by MADGE, was used to identify carriers of the −93T/G mutation. The frequency of this mutation was investigated in the healthy control subjects from the NPHSII study and in the healthy Afro-Caribbeans and Gujarati Indians participating in the Brent and Harrow study. These results are presented in Table 2⇓. The carrier frequency of the −93G allele was 3.4% in the NPHSII healthy men. In Gujarati Indians, the frequency (2.9%) was not significantly different from the NPHSII group. However, in Afro-Caribbeans, the −93G allele occurred in 57 of the 91 individuals, a frequency of 62.6%. Furthermore, among Afro-Caribbeans, there were 14 individuals who were homozygous for the mutation.
What became evident on identifying carriers of the −93T/G mutation was that within the group of individuals who had been screened previously for the D9N mutation,10 all individuals who were carriers of N9 were also carriers of the −93G. Those NPHSII samples not previously genotyped for D9N, and the Brent and Harrow samples, were therefore screened for the N9 mutation to estimate the allelic association between these two variant sites, and the observed distribution of the genotype classes are presented in Table 3⇓. In the NPHSII sample, 45 of 54 −93G carriers were also carriers of the N9, representing 83% allelic association (Δ=.91, P=.0001). In the Afro-Caribbeans, the carrier frequency of the −93G (57 of 91) was 18-fold higher than in the NPHSII white group; in total, 14 of the 57 carriers were homozygous for the −93G, with the genotype GG/DN or GG/DD. Furthermore, of the 57 −93G Afro-Caribbean carriers, only 10 were carriers of N9, giving 17% allelic association (−=.32, P=NS). The carrier frequency of the N9 mutation (10 of 91;10.9%) was also higher in this group compared with NPHSII (45 of 1575; 2.8%, P=.03). In the Gujarati Indians, the carrier frequency of the −93G was 2.9%; no N9 carriers were detected in this group, although N9 carriers have been identified in other Asian Indian samples (unpublished data, P Talmud and A Panaloo, 1995).
In the 1575 NPHSII men, 43 were doubly heterozygous TG/DN and 9 were carriers of the −93G (TG/DD); however, we did not identify a single individual in this sample nor in the Afro-Caribbeans (Table 3⇑) or Gujarati Indians who carried the N9 mutation on a −93T allele. Using the maximum-likelihood algorithm,25 it is highly likely that all N9 mutations are always on a −93G allele, and thus the effect of N9 without the −93G cannot be assessed. The effect on plasma lipid levels could be assessed for those detected variants, −93T/D9 (TT/DD), −93G/N9 (TG/DN), and −93G/D9 (TG/DD), and in the NPHSII study baseline, and five annual measures of lipids were available. It was therefore possible to look at effects on Tg levels over the 6-year period (Fig 2⇓), and the overall results are presented in Table 4⇓. Taking the mean Tg levels over 6 years compared with those of the individuals who had only the wild-type alleles, TT/DD, and who had mean plasma Tg levels of 1.76 mmol/L, carriers of the −93G/D9 with genotype TG/DD had significantly lower plasma Tg levels (1.32 mmol/L; P=.01). By contrast, those individuals who were carriers of both mutations (TG/DN, including 2 homozygotes, GG/NN) had higher plasma Tg levels (1.89 mmol/L; P=.31, NS; overall test for the difference between the three genotypes, P=.01). Thus, the −93G/D9 allele was associated with a 31% lowering of Tg levels compared with wild-type, whereas the −93G/N9 allele was associated with an 8.4% increase in Tg levels (P=NS). The Tg variance in the TG/DD group was 0.07 compared with 0.18 in the TT/DD group and 0.27 in the TG/DN groups, which suggests that in TG/DD individuals, the −93G/D9 allele is having a homeostatic effect. For the sample of Afro-Caribbeans, full data on lipid levels and BMI were unavailable, so no meaningful analysis could be carried out.
To test whether there was an interaction between BMI and Tg according to genotype, individuals were stratified according to the mean of the sample BMI (< or >26.5 kg/m2; Fig 3⇓). Individuals with the common genotype (TT/DD) and low BMI had lower Tg levels than those TT/DD individuals with high BMI (1.67 versus 2.08 mmol/L, respectively P<.0001; 25% increase), characteristic of the Tg-raising effect of high BMI. In individuals with the genotype TG/DN, this BMI-associated increase in Tg was much larger (1.57 versus 2.33 mmol/L, P=.03; 48% increase). This N9-BMI interaction has previously been reported.10 13 However, in carriers of the −93G with D9 (TG/DD), the BMI effect was absent, and the group of individuals with a BMI above the sample mean had a lower mean Tg level of 1.15 mmol/L (27.2% lowering), but this difference was not statistically significant.
To test whether the −93T→G mutation was itself functional, a 357-bp fragment of the LPL promoter from the translation start site to bp −161, with either the T or G at position −93, was cloned into pGL3-enhancer, a luciferase reporter gene vector. the two constructs, pGL3/-93T and pGL3/-93G, were then used to transfect a rat smooth muscle cell line, A10, and a human adrenal cell line, NCIH295.23 The mean results from four repeat experiments, each representing 7 to 10 transfected plates for each construct, are given in Fig 4⇓. These results are corrected for transfection efficiency and protein content. Compared with the T allele, given as 100%, the G allele consistently showed significantly higher luciferase activity, with a mean of 124% (±2.4%, P=.05) in the A10 cells and 118% (±16%, P=.097, NS) in adrenal NCIH295 cells.
To analyze further the interaction of the −93T→G with nuclear factors, two labeled 25-mer double-stranded oligonucleotides, from nucleotides −81 to −104, with either a T or G at position −93, were used in a bandshift assay with cellular extracts from A10 cells (Fig 5⇓). A DNA-protein complex was observed with −93G probe (lane 9) but not with −93T probe (lane 3). This interaction was not competed out with increasing amounts of unlabeled T competitor (lanes 10 and 11). However, with increasing amounts of unlabeled G competitor (lanes 12 through 14), the intensity of the band was reduced, thus diluting the effect of the labeled probe binding to the protein extract.
Using SSCP, we have identified a common variant within the LPL promoter, a T→G transition at position −93. This mutation was previously reported by Yang et al27 to occur in an individual who had a T-to-C substitution at nucleotide −39 of the LPL gene, but on the other chromosome. In a sample of healthy white men from the UK, the carrier frequency of this mutation is 3.4%. Several of the DNA samples used in the SSCP analysis were from individuals of Afro-Caribbean origin, and from the initial SSCP screening, it was obvious that the −93G allele was much more common in this group. Genotyping a sample of 91 Afro-Caribbeans showed that the −93G carrier frequency was 18-fold higher than in the whites (62.6% versus 3.4%).
Since the NPHSII sample had been genotyped for both mutations, it became apparent that there was very strong allelic association between −93G and N9 (83%; −=.91, P=.00001), with 17% of −93G alleles being on the wild-type D9 background (−93G/D9). We did not identify a single individual who carried the N9 allele without the −93G (genotype TT/DN), and since the healthy individuals are drawn from an outbreeding general population, it is highly unlikely that those individuals heterozygous for both mutations are a genetically distinct group from those heterozygotes for either mutation alone. That is, the phase of the two mutations (haplotype) found in those individuals heterozygous for both can be accurately inferred from the phase of the unambiguously observed haplotypes for those heterozygous for either mutation alone. In the Afro-Caribbeans, the allelic association was much weaker, with only 17% allelic association (−=.32, P=NS) between −93G and N9. These results suggest that the −93T→G is the older of the two mutations and that the N9 mutation occurred on a −93G. The strong allelic association that exists in whites suggests that little recombination has occurred between the two variant sites. The difference in allelic association of the two variable sites in Afro-Caribbeans compared with whites may be explained by the increased frequency of the −93G in Afro-Caribbeans.
Carriers of the N9 mutation have been shown to have higher plasma Tg levels than D9 carriers.10 However, it was evident that all N9 carriers were also carriers of −93G, suggesting the possibility that the elevated Tg levels associated with N910 were in fact due to the functional effect of the −93G. Since it was possible to identify carriers of the −93G distinct from N9, the effect on lipid levels of the promoter mutation on a D9 background could be assessed. From the NPHSII study, at baseline and over the subsequent 5-year period, the mean Tg level in carriers of −93G (genotype TG/DD) was consistently lower than in noncarriers (mean of 31%, which was statistically significant, P=.01). We do not know why there is an increase in Tg in years 4 and 5 in TG/DN individuals. It could be that year 3 was a low estimate, but whatever the reason, the increase in plasma Tg levels from year 3 (1.72 mmol/L) to year 4 (2.11 mmol/L) of 0.39 mmol/L was not statistically significant.
Several previous studies have shown that the LPL N9 mutation shows an interaction with BMI;10 13 therefore, carriers of the mutation with BMI in the top half or tertile of BMI for the group have Tg levels significantly higher than noncarriers within the same BMI range. Both we and others have shown that the D9N mutation produces defective LPL with residual activity of 70%, though specific activity is close to normal.10 28 Evidence from in vitro studies on N9 suggest that the secretion rate is reduced, which accounts for the reduced activity.10 28 Thus, a hypothesis to explain this Tg-BMI interaction is that in the presence of increased production of VLDL associated with increased body weight, the defective N9-LPL is overwhelmed, resulting in decreased Tg catabolism rate and thus elevated Tg levels. Results from the present study (Fig 3⇑) show that in the healthy white men with a BMI below the sample mean, there was no difference in Tg levels among the three genotype groups. In those with BMI in the top half compared with those with the common allele (TT/DD), the TG/DN carriers had a 12% higher mean Tg level, which was not statistically significant. However, in the group with the genotype TG/DD, mean Tg levels at the higher BMI range were lower than in those TG/DD carriers with BMI below the sample mean (−27%, P=.05). It is therefore possible that the −93G allele may “protect” against the lipid-raising effect of the increased production of Tg-rich particles.
In whites, the −93G allele occurs in 83% of carriers on the same chromosome as the N9 variant, which in vitro results in a moderately lower (≈70% of wild-type) level of LPL secretion.28 It is likely that in an individual heterozygous for both mutations, upregulation of this secretion-defective LPL would result in a larger-than-expected proportion of secretion-defective heterodimer LPL being produced. Therefore, the overall effect of carrying the two mutations is to raise plasma Tg levels, with the −93G/N9 allele being a predisposing factor for hypertriglyceridemia, interacting with other genetic or environmental factors, such as obesity.
To date, only one other naturally occurring mutation in the promoter of the LPL gene has been identified. The mutation, a T→C substitution at −39, occurs in the Oct-I binding site, which results in a reduction by 85% of promoter activity compared with wild-type, and was identified in a patient with familial combined hyperlipidemia.27 We screened for the −39 mutation in 116 combined hyperlipidemic patients, using allele-specific oligonucleotides (results not shown) and found no carriers of this mutation. Yang et al27 reported the identification of the −93T/G mutation in the same patient, but reported no effect on promoter activity. A10 cells, although of rat origin, have been used previously in in vitro studies of human gene expression.29 Functional assays presented here, using the luciferase gene as a reporter, show 24% increased promoter activity of the −93G compared with the T allele. Using a human adrenal cell line, NCIH295, which has been shown to secrete LPL,23 a similar (18%) increase in luciferase activity was seen with the G allele. For a common mutation with small effect, this is the order of increase that would be predicted. This observation suggests a mechanism for the −93G/D9 allele-associated reduction in plasma Tg levels in vivo, by resulting in increased LPL mRNA and protein and thus increased catabolism of Tg-rich particles. Since the −93 nucleotide lies within the region of LSE-2, identified by Tanuma et al17 as a potential LSE, it is possible that the T-to-G change at −93 results specifically in the disruption of LSE-2, leading to increased expression of LPL. However, evidence from the bandshift assay does not support this hypothesis, since the T allele did not bind a nuclear protein under the assay conditions. In contrast, the increased promoter activity of the −93G-luciferase fusion protein suggested that the −93G has a motif that is recognized by a nuclear factor that does not bind (or binds very poorly) to the T allele; hence, the T competitor could not compete out the G protein interaction, while increasing amounts of unlabeled G probe did. Therefore, results from both the luciferase and bandshift assays clearly demonstrate that the G allele results in increased promoter activity, and the interaction with protein extract suggests this activity is due to binding of a nuclear factor to the G allele, which enhances promoter activity.
The 18-fold higher carrier frequency in the −93G in Afro-Caribbeans compared with Caucasians suggests that −93G is in fact the ancestral form of the LPL gene, providing a nuclear binding recognition sequence that could promote LPL expression. Support of this hypothesis comes from DNA sequencing of the LPL promoter from gorilla, orangutan, and chimpanzee; all three species have a G at position −93 (unpublished data, S Hall, 1996). If the “out-of-Africa” hypothesis is correct, it could be speculated that migrating individuals carried the rarer −93T allele, which then became the common allele in whites. Our data suggest that the G-to-T change at −93 leads to the loss of a recognition sequence for a nuclear protein that can upregulate LPL expression in the face of increasing Tg and that this ability has been lost in −93T white carriers. Identification of this endogenous protein will be necessary to confirm our hypothesis.
Thus, the transfection data support a direct effect of the −93G variant on LPL transcription, and its association with lower Tg levels might be due to higher levels of LPL mRNA in muscle and adipose tissue and thus higher levels of secretion of LPL. LPL activity measures of −93G carriers may clarify this possibility, but it is likely that owing to the insensitivity and large coefficient of variation of the postheparin LPL assay, a moderate difference of 24% predicted from the A10 transfection data (18% in NCIH295 cells) may not be easily detectable. In white populations, the impact of −93G on lipid levels will not be large, because of the low frequency of the mutation. However, in people of Afro-Caribbean origin, in whom the frequency is dramatically higher and most of the −93G mutations occur independently of the N9, the impact at the population level might be greater, which might directly contribute to the lower plasma Tg levels and thus the reduced coronary artery disease risk reported in Afro-Caribbeans.30 31 Studies are under way to test this hypothesis.
Selected Abbreviations and Acronyms
|BMI||=||body mass index|
|LSE||=||LPL silencer element|
|MADGE||=||microtiter array diagonal gel electrophoresis|
|NPHSII||=||Northwick Park Heart Study II|
|PCR||=||polymerase chain reaction|
|SSCP||=||single-strand conformational polymorphism|
This work was supported by grant PG15 from the British Heart Foundation, London. S. Hall was supported by a UCLMS Clinical Science Studentship. NPHSII was supported by the Research Medical Council, London, the National Institutes of Health, Bethesda, Md (grant No. NHLBI 33014), and DuPont Pharma, Wilmington, Del. The rat smooth muscle cell line A10 was a kind gift from Dr O. Yalkinoglu, Bayer AG, Wuppertal, Germany.
Bainton D, Miller NE, Bolton CH, Yarnell JW, Sweetnam PM, Baker IA, Lewis B, Elwood PC. Plasma triglyceride and high density lipoprotein cholesterol as predictors of ischaemic heart disease in British men: the Caerphilly and Speedwell Collaborative Heart Disease Studies. Br Heart J. 1992;68:60-66.
Bengtsson C, Bjorkelund C, Lapidus L, Lissner L. Associations of serum lipid concentrations and obesity with mortality in women: 20-year follow up of participants in prospective population study in Gothenburg, Sweden. BMJ. 1993;307:1385-1388.
Beisiegel U, Weber W, Bengtsson-Olivecrona G. Lipoprotein lipase enhances the binding of chylomicrons to low density lipoprotein receptor–related protein. Proc Natl Acad Sci U S A. 1991;88:8342-8346.
Babirak SP, Brown BG, Brunzell JD. Familial combined hyperlipidemia and abnormal lipoprotein lipase. Arterioscler Thromb. 1992;12:1176-1183.
Mailly F, Tugrul Y, Reymer PW, Bruin T, Seed M, Groenemeyer BF, Asplund-Carlson A, Vallance D, Winder AF, Miller GJ, Kastelein JJP, Hamsten A, Olivecrona G, Humphries SE, Talmud PJ. A common variant in the gene for lipoprotein lipase (Asp9→Asn): functional implications and prevalence in normal and hyperlipidemic subjects. Arterioscler Thromb Vasc Biol. 1995;15:468-478.
Fisher RM, Mailly F, Peacock RE, Hamsten A, Seed M, Yudkin JS, Beisiegel U, Feussner G, Miller G, Humphries SE, Talmud PJ. Interaction of the lipoprotein lipase asparagine 291→serine mutation with body mass index determines elevated plasma triacylglycerol concentrations: a study in hyperlipidemic subjects, myocardial infarction survivors, and healthy adults. J Lipid Res. 1995;36:2104-2112.
Mailly F, Fisher R, Nicaud V, Luong L, Evans AE, Marques-Vidal P, Luc G, Arveiler D, Bard JM, Poirier O, Talmud P, Humphries SE. Association between the LPL-D9N mutation in lipoprotein lipase gene and plasma lipid traits in myocardial infarction survivors from the ECTIM study. Atherosclerosis. 1996;122:21-28.
Previato L, Parrott CL, Santamarina Fojo S, Brewer HB Jr. Transcriptional regulation of the human lipoprotein lipase gene in 3T3-L1 adipocytes. J Biol Chem. 1991;266:18958-18963.
Enerback S, Ohlsson BG, Samuelsson L, Bjursell G. Characterization of the human lipoprotein lipase (LPL) promoter: evidence of two cis-regulatory regions, LP-alpha and LP-beta, of importance for the differentiation-linked induction of the LPL gene during adipogenesis. Mol Cell Biol. 1992;12:4622-4633.
Tanuma Y, Nakabayashi H, Esumi M, Endo H. A silencer element for the lipoprotein lipase gene promoter and cognate double- and single-stranded DNA-binding proteins. Mol Cell Biol. 1995;15:517-523.
Miller SA, Dykes DD, Polesky HF. A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res. 1988;16:1215.
Kunkel LM, Smith KD, Boyer SH, Borgaonkar DS, Wachtel SS, Miller OJ, Breg WR, Jones HW Jr, Rary JM. Analysis of human Y-chromosome–specific reiterated DNA in chromosome variants. Proc Natl Acad Sci U S A. 1977;74:1245-1249.
Gudnason V, Mak YT, Betteridge J, McCarthy SN, Humphries S. Use of the single-strand conformational polymorphism method to detect recurrent and novel mutations in the low-density lipoprotein receptor gene in patients with familial hypercholesterolemia: detection of a novel mutation Asp200→Gly. Clin Investig.. 1993;71:331-337.
Staels B, Martin G, Martinez M, Albert C, Peinado-Onsurbe J, Saladin R, Hum DW, Reina M, Vilaro S, Auwerx J. Expression and regulation of the lipoprotein lipase gene in human adrenal cortex. J Biol Chem. 1996;271:17425-17432.
Templeton AR, Sing CF, Kessling A, Humphries S. A cladistic analysis of phenotype associations with haplotypes inferred from restriction endonuclease mapping, II: the analysis of natural populations. Genetics. 1988;120:1145-1154.
Welch W, James J, Brown-Forsythe BF. Robust alternatives to traditional analysis of variance. In: Reed JF, Stark DB, eds. Computer Methods and Programs in Biomedicine. Amsterdam, Netherlands: Elsevier Science; 1988:233-237.
Yang WS, Nevin DN, Peng R, Brunzell JD, Deeb SS. A mutation in the promoter of the lipoprotein lipase (LPL) gene in a patient with familial combined hyperlipidemia and low LPL activity. Proc Natl Acad Sci U S A. 1995;92:4462-4466.
Ye S, Erikksson P, Hamsten A, Kirkinen M, Humphries SE, Henney AM. Progression of coronary atherosclerosis is associated with a common genetic variant of the human stromelysin-1 promoter results in reduced gene expression. J Biol Chem. 1996;271:13055-13060.
Fontbonne A, Papoz L, Eschwege E, Roger M, Saint Paul M, Simon D. Features of insulin-resistance syndrome in men from French Caribbean Islands: the Telecom Study. Diabetes. 1992;41:1385-1389.