DETECTION OF THE LOW DENSITY LIPOPROTEIN RECEPTOR GENE PVUII INTRON 15 POLYMORPHISM USING THE POLYMERASE CHAIN REACTION : ASSOCIATION WITH PLASMA LIPID TRAITS IN HEALTHY MEN AND WOMEN

V. GUDNASON*, T. ZHOU', K. THORMAR*, S. BAEHRING', J. COOPER', G. MILLER, S.E. HUMPHRIES*", H. SCHUSTER' *Divisioll of Cardiovascular Genetics, Department of Medicine, UCL Medical School, The Rayne Institute, University Street, London WC IE 6JJ 'FranzVolhard-Clinic, Max-Delhriik Centre for Molecular Medicine, Willhergstrasse 50, 13122 BerlinBuch, Germany ,'MRC Epidemiology a/ld Medical Care Unit, Wolfson Institute of Preventive Medicine, SI Bartholomews and the Roml London school of Medicine. CharteriJouse Square, London EC I M 6BQ


INTRODUCTION
The LDL-receptor plays a major role in the removal of LDL-cholesterol particles from the blood (Goldstein and Brown, 1989), and many mutations in the LDL receptor gene have been reported that destroy receptor function and cause the disorder Familial Hypercholesterolaemia (FH) (Hobbs et al., 1992, Day et ai., 1997).Patients who are carriers for such mutations have plasma cholesterol levels that are elevated twofold or more above normal levels (Goldstein and Brown, 1989) and are at elevated risk of developing atherosclerosis and Ischaemic Heart Disease (IHD) at an early age (Betteridge et ai., 1991).Carriers for such mutations are believed to occur at a frequency of 1/500 members of the general population (Goldstein and Brown, 1989).
It is plausible that there may also be mutations in the gene that have only a mild effect on receptor function, and thus cause only a moderate difference in plasma levels in carriers.These mutations could be at any level of receptor regulation or activity.If such mutations were common they would be having a significant effect on determining plasma lipid levels ofthe overall popUlation.An example of common mutations of mild effect on plasma lipid levels are those producing the isoforms of the apoE protein (Weisgraber et al., 1981), which explain up to 5% of popUlation variance in LDLcholesterol (Davignon et al., 1988).
So far, only one common mutation altering an amino acid in the LDL receptor has been reported.This is an Alanine to Threonine change at position 370 (A370T), with the T370 allele occurring at a frequency of roughly 0.05 in European Caucasian populations (Kotze et al., 1986, Taylor et al., 1988, Gudnason et aI., 1995).This mutation does not cause FH, but there is some evidence that it may be associated with a small effect on plasma LDL cholesterol levels (Gudnason et al., 1989).Four studies have reported that a PvuII polymorphism located in intron 15 of the LDL receptor gene is associated with differences in LDL cholesterol levels (Pederson et al., 1988, Pederson et aI., 1989, Schuster et al., 1990, Humphries et al., 1991).The frequency of the Pvull cutting allele is roughly 0.27 in Caucasian popUlations and those homozygous for the cutting allele, had 10-20% lower levels of LDL cholesterol in studies of non-FH individuals from Norway (Pederson et al., 1988, Pederson et al., 1989), Germany (Schuster et al., 1990), and Italy (Humphries et al., 1991), although, no effect was seen in a group of individuals from the UK (Taylor et al., 1988) nor from Denmark (Klausen et ai., 1993).The PvuIi polymorphism is not likely to be functional in itself, but possibly serves as a marker linked to another variation in the gene that either structurally alters the receptor and its activity, or alters its function in a regulatory manner.
Until recently, the Pvull polymorphism has only been detectable using Southern blotting methods, although a PCR-based method has been devised for long PCR (du Plessis and Kotze, 1996), but this is inconvenient for examination in larger samples.Intron 15 is large (roughly 5kb) and has been difficult to amplify.Only a partial sequence of intron 15 is available (Hobbs et al., 1985), but from Southern blotting experiments (Humphries et ai., 1985) the location of the Pvull site was known to be about 900bp from the 3' end of exon 16.We report here a partial sequence of the intron which has enabled an easily applied PCR method for detection of this polymorphism to be developed.

Subjects
Three groups of individuals were studied, the first from Iceland comprising 318 healthy men and women (mean age 45 years) selected from a dietary survey (Sigurdsson et ai. , 1992).Individuals reporting evidence of IHD or with FH, diabetes or taking lipid lowering medication were excluded, and full details of the characteristics of this group have been published previously (Sigurdsson et ai., 1992).The second and third groups were healthy middle-aged men (mean age 56 years) selected from a general practice in the South of England (Camberley) and another in Scotland CSt.Andrew ' s) .All were free of symptoms ofIHD and their mean (± SO) body mass index (BMI) was 26.5 ± 3.7 kg/m2 (Miller et at., 1996).Plasma lipids were measured using standard techniques as reported (Sigurdsson et ai. , 1992, Miller et aI., 1996).Serum apolipoprotein B (apoB) was measured by an automated immunoturbidometric method with reagents from Incstar/Sorin (Wokingham Berks, UK) for the UK sample and with reagents from Roche MA 30: Cobas Mira for the Icelandic sample.The different source of reagents and standards explains the difference in mean levels of apoB between the two samples.

Long Chain PCR
The entire intron 15 was amplified with the following primers previously published (Hobbs el ai., 1992): 5'-exon 15 primer: 5'-GAA GGG CCT GCA GGC ACG TGG CAC T-3' and 3' exon 16 primer: 5'-CGC TGG GGG ACC GGC CCG CGC TTA C-3'.50 ng of genomic DNA was used in a PCR reaction containing 200 M of each dNTP, 10 pmol of each primer, 5 U Taq polymerase (Perkin-Elmer, USA) and 5 U Taq extender (Stratagene, USA) in a total volume of 100 III of the extender buffer provided by the manufacturer.The PCR conditions were as follows : hot start 94 'c for 2 min and subsequently 30 cycles of 94 'c for IS seconds, 6YC for 30 seconds, 70' C for 5 min and finally 72' C for 10 min.Digestion of PCR was performed as described in the genotyping protocol.

Anchored PCR and Sequencing
To amplify a smaller 3' subfragment containing the PvulI restriction site, 30 different primers available in the laboratory from other projects were used as forward primers, while the 3' -exon 16 primer was used as reverse primer, in separate reactions .I ng of the expected 5 kb long chain PCR amplification product comprising exon 15, intron 15 and exon 16 was used under low stringency conditions to get amplification products using the same PCR reaction mix as described for the genotyping protocol.With the 3'-exon 13 LDL receptor primer (5'-GTT TCC ACA AGG AGG TTT CAA GGT T-3',) (Hobbs et al 1992) a 1.4 kb fragment was detected after electrophoresis in a 0.8 % agarose gel.After purifying the amplification product from the gel, a booster PCR was performed using the same conditions to confirm specificity of the amplification product.Subsequently, the existence of the Pvull site was confirmed by restriction analysis.Direct sequencing was performed using a standard protocol as described before (Schuster et aI., 1995).

DNA isolation and genotyping
DNA was isolated using a salting out method (Miller, 1988).Pvull genotype was determined amplifying a part ofintron 15 of the LDL receptor gene using 10 pmol of each of the following primers: 5'primer: 5'-TCC CCT TCA AAA TGC CCT CTT-3' and 3' -primer: 5'-AGC CAC CGA GCC CAG CCT AAG AA-3' with 50 ng of genomic DNA and 0.75 U of Taq polymerase (Gibco, UK) in a total volume of 20 III of the buffer recommended by the manufacturer containing 4.1 mM MgCI 2 200 M each dNTP.The PCR conditions were as follows : 94°C for 5 min , 63 °C for 3 min and 70°C for 3 min once and subsequently for 30 cycles of 94°C for I min, 63C for I min and at 70 °C for 2 and a half min , and finally at 70°C for 10 minutes.Digestion of PCR product was performed by addition of 5 U of Pvull (Boehringer) to 10 III of the PCR product in the buffer recommended by the manufacturer and incubated at 37"C for 4 hours.The product was run on a 2% agarose gel and the bands visualised by ethidium bromide and viewing on a UV trans-illuminator.

Statistical analysis
The relationship between genotype and levels of plasma lipid traits was examined by ANOV A after adjustment for age and body mass index (BMI) using multiple linear regression.The characteristics (weight, height, age and plasma lipids) of the two groups of men from the UK were very similar (none significantly different) and analysis was carried out in the combined group .The frequency of the genotype groups compared to that expected for Hardy Weinberg properties was analysed by a X 2 test.To analyse potential interaction with BMI, individuals were coded 0, I, or 2 according to their terti Ie of BMI and this was included as a factor in a two-way ANOV A, after adjustment for age.Analysis was carried out using SPSS PC analytical package.Statistical significance was taken at a level of p=0.05.

RESULTS
Initial attempts were made to amplify the entire intron using a forward primer flanking exon 15 at the 5' end and a reverse primer flanking exon 16 at the 3' end using a long chain PCR protocol (data not shown).Several random primers available in the laboratory were then separately used as forward primers in conjunction with the exon 16 reverse primer to amplify DNA from individuals homozygous for the lack of the cutting site and heterozygous for the cutting site, as determined by previous Southern blotting experiments.Figure 1 a illustrates the location of the forward primer with sufficient sequence homology to intron sequence to allow a successful amplification of an anchored PCR fragment (upper panel).As shown in Figure 1 b, this resulted in the amplification of a fragment approximately 1400 bp, which was cleaved into two fragments of roughly 900 bp and 500 bp.
Further seq uencing of this 1400 bp fragment was carried out to determine the exact location and sequence change creating the Pvull site.Figure 2 shows the sequence determined from an individual heterozygous for the cutting site, with a C to T tran sition within the sequence CAGCCG which would create the cutting site 3' CAGCTG.No other sequence differences were seen in this fragment of over 900 bases.The part of the sequence obtained was used to synthesise a reverse primer, which was used for subsequent genotyping.As shown in Figure 3, the result is amplification of a fragment of roughly 800 bp which is cleaved by Pvull digestion into fragments of around 600 bp and 200 bp.The figure shows that the presence of the cutting site (T allele) is inherited in the expected manner in a small nuclear family .The genotypes obtained with this method were consistent with genotypes previously obtained by Southern blotting (not shown).

GATe
Sequence of the part of intron 15 containing the PvuII site from an individual heterozygous for the site.
Genotype was next determined in three groups of healthy men and women from Iceland, England and Scotland.For all samples genotype distribution was not different from that expected for Hardy Weinberg proportions.The estimated frequencies of the T allele and 95% confidence intervals of the estimates are shown in Table I.These range between 0.21-0.24,and none are significantly different from each other.
The relationship between CIT genotype and levels of total cholesterol, apo Band plasma triglycerides was examined separately in the men and women from Iceland and in the combined groups of men from the UK.As shown in Table 2, plasma cholesterol, triglyceride and apo B levels were consistently lower in those homozygous for the T allele except for cholesterol in Icelandic women, compared to other genotype groups,  and this effect reached statistical significance in the combined UK group.Data were also analysed in men stratified by tertiles of BMI, and as shown in Figure 4 the genotype effect on cholesterol and apoB was strongest in men below the lower tertile of BMI.This genotype x BMI interaction was statistically significant for the trait of plasma apoB levels.In the group of Icelandic men and women the lowering effect associated with the T allele was also larger in those with lower BMI, but this interaction was not statistically significant.For example, in men, compared to the C/C group, cholesterol levels for those with one or more T allele were 0.29 mmollilower in those with a BMI below the lowest tertile « 23.3 kg/m2) but 0.21 mmolll higher in those above the highest tertile (> 25.8 kg/m2).

DISCUSSION
Although previous studies (Pederson et aI., 1988, Pederson et aI., 1989, Schuster et aI., 1990, Humphries et al., 1991) have shown an association between the Pvull + cutting site (the Intron 15-T allele) and lower levels of plasma cholesterol, studies from the UK (Taylor et al., 1988) and Denmark (Klausen et aI., 1993) failed to detect a significant effect.Several possible reasons for this discrepancy can be proposed.Firstly, the original three studies could be showing an association by chance alone, but this is statistically unlikely.Secondly, the original studies used unaffected relatives (mainly spouses) of patients with FH, where Pvull genotyping had been carried out as part of co-segregation studies.Such samples are clearly not truly representative of the general population and there may have been some underlying genetic or environmental bias in the sample.For example, it has been suggested (Klausen et aI., 1993) that the group of relatives lacking the cutting site (genotype C/C) may have included some individuals with FH, and the higher levels of plasma cholesterol of these individuals would thus elevate the estimate of the whole genotype group, gi ving a larger difference between the CIC and TIT groups.
Alternatively, the relatives of patients with FH may have adopted a lipid-lowering diet or life style which would magnify any small effect associated with the Pvull genotype.Such environmental effects would be absent in a representative sample of the general population, which was examined in the Danish study.
In this study we have detected a small but consistent association between the LDLreceptor intron IS PvuII cutting allele (T) and lower levels of plasma lipid traits in samples of healthy men and women representative of the general population in Iceland and in the UK.There was no gene-dosage effect of the PvuII polymorphism on any of the traits seen in the UK men, which is similarto that seen in some other studies of the PvuII polymorphism (Pederson et aI., 1989), and the effect is apparent only in those homozygous for the T allele, who, compared to those with the CC genotype, have levels (in men) of total cholesterolS-6% lower, triglycerides 12-1S % lower and apoB 3-S % lower.In the sample of women , although non significant, the effects were of similar size, being 2.7 %, IS .3% and 8% lower respectively.The size of the lowering effect associated with the TIT genotype in these samples of healthy individuals is roughly half that reported previously, and was only significant in the large sample of men from the UK.In part because of the rapid PCR-based genotyping method, the size of the present sample is at least twice as large as has been examined previously (eg 170 men in the Danish study) and allowed the detection of a small effect.Roughly 5% of individuals in Iceland and the UK will be homozygous for the T allele and thus at a population level the plasma lipidlowering effect associated with this genotype is small (R2 for cholesterol in the UK men was < 1 %, not shown).However, the effect was larger in healthy individuals who were thin, (BMI < 25 kg/m2) and this may explain in part the failure to detect the association in the Danish study, where the mean BMI was 25kg/m2 (Hansen et al., 1993) and no stratification for BMI was used.It is possible that the reason for the lack of a significant effect of the PvuII genotype on plasma lipoprotein levels in the Icelandic sample may in part be related to the average BMI in the Icelandic group, which is 25.4kg/m2, in men .Although the difference in plasma cholesterol and apoB levels was not statistically significant between those homozygous for a C allele and those homozygous for a T allele, the trend was observed in those below a BMI of 25kg/m2 (+ 2.8 SC and + 4.6% respectively).
Berg and colleagues have reported a potential interaction between variation in the apo E gene (E2 , E3 and E4 isoforms) and the LDL receptor Pvull genotype in determining plasma lipid levels (Pederson et al., 1989).In the samples examined here apo E genotype was strongly associated with plasma lipid levels as expected for the Icelandic sample (Gudnason et ai., 1993), and for the UK samples (Day, Bolla, Humphries and Miller, unpublished).However, because the LDL-receptor genotype lipid lowering effect was only seen in the smallest TIT group (n=45 in the UK men) no analysis for interaction with apoE was attempted.
Although first reported in 1988 (Pederson et al., 1988), the mechanism of this association is currently unknown.The intron 15 CIT change is unlikely to be offunctional significance itself, but is probably an allelic marker for a functional sequence elsewhere at the gene locus with the T allele being a marker for a sequence change that results in greater numbers of LDL-receptors on the cell surface of a higher LDL receptor, activity or stability.In those with the genotype TIT this would have the result of improved clearance of both apoE (eg VLDL remnants) and apoB (eg LDL) containing lipoproteins, leading to lower plasma levels of triglycerides, cholesterol and apoB.The observation that this genotype effect is confined to those homozygous for the T allele, and is enhanced in those with low BMI needs confirmation in a larger study, and the possible molecular mechanisms of these effects is unclear.
The availability of a convenient PCR method for determining PvuII genotype will enable further studies to be carried out to confirm and extend these observations, and cell and molecular biology studies are also warranted to determine the mechanism of the lowering effect.The polymorphism will also be of use in tracing the LDL receptor gene in families with FH.
Gudnason by a grant from the Helen Eppel Foundation.The support of the 1celandic Heart Association and the 1celandic Council of Science is appreciated.We thank Manjeet Bolla and Ian Day for access to unpublished apoE genotype data in the sample of the UK men.
Figure I. a) Map of the intron 15-16 region of the LDL-receptor gene.showing the appropriate location of the Pvull site and the primers used.b) 2'1r.Agarose gel showing M .marker of primers used DNA digested with Hindlll.Pvu +/and Pvu -/-.Amplified DNA from individuals of known Pvull genotype.digested with Pvull and run as described in Materials and Methods.

Figure 3 ,
Figure3, P,'ull genotype from four individuals from a small nuclear family, using conditions as described in Materials and Methods,

Figure 4 .
Figure 4. Mean chole sterol (mmol /I) and apoB (g/I) le vel s in the U K men w ith different LDL-reee ptor P\'ull genot y pe , di vided by te rti Ie of Bod y Mass Inde x

Table I .
Estimated T allele frequency and 95% confidence intervals in the samples studied.

Table 2 .
Plasma lipid traits (mean ±(SE» in subjects with different Pvull genotypes