genetic variation at the plasminogen activator inhibitor-1 locus is associated with altered levels...

S Dawson, A Hamsten, B Wiman, A Henney and S Humphriesaltered levels of plasma plasminogen activator inhibitor-1 activity

Genetic variation at the plasminogen activator inhibitor-1 locus is associated with

1524-4636 Copyright © 1991 American Heart Association. All rights reserved. Print ISSN: 1079-5642. Online ISSN:

7272 Greenville Avenue, Dallas, TX 72514Arteriosclerosis, Thrombosis, and Vascular Biology is published by the American Heart Association.

1991, 11:183-190Arterioscler Thromb Vasc Biol

http://atvb.ahajournals.org/content/11/1/183located on the World Wide Web at:

The online version of this article, along with updated information and services, is

http://www.lww.com/reprintsReprints: Information about reprints can be found online at [email protected]. E-mail: Kluwer Health, 351 West Camden Street, Baltimore, MD 21202-2436. Phone: 410-528-4050. Fax: Permissions: Permissions & Rights Desk, Lippincott Williams & Wilkins, a division of Wolters http://atvb.ahajournals.org//subscriptions/is online at Subscriptions: Information about subscribing to Arteriosclerosis, Thrombosis, and Vascular Biology

by guest on July 10, 2011http://atvb.ahajournals.org/Downloaded from

http://atvb.ahajournals.org/

183

Genetic Variation at the PlasminogenActivator Inhibitor-1 Locus Is Associated

With Altered Levels of Plasma PlasminogenActivator Inhibitor-1 Activity

Sally Dawson, Anders Hamsten, Bjorn Wiman, Adriano Henney, and Steve Humphries

Plasminogen activator inhibitor-1 (PAI-1), a rapid inhibitor of tissue-type plasminogenactivator, has been shown to be an independent risk factor for recurrent myocardial infarction(MI) at a young age. To investigate whether genetic variation in the PAI-1 gene is affectingplasma PAI-1 levels, a sample of 145 patients with an MI before the age of 45 years wasgenotyped for two polymorphisms at the PAI-1 locus, together with a sample of 95 healthyindividuals of a similar age. All individuals were measured for plasma PAI-1 levels as well asfor other fibrinolytic and metabolic risk indicators. A HindlU restriction fragment lengthpolymorphism (RFLP) was used in this study in conjunction with a previously unreportedeight-allele dinucleotide repeat polymorphism at the PAI-1 locus. The dinucleotide repeatpolymorphism and HindUl RFLP were in strong linkage disequilibrium. There was nodifference in the frequency of alleles of either polymorphism between patient and controlgroups. However, the smaller dinucleotide repeat alleles were significantly associated (p=0.03)with higher plasma PAI-1 levels in the patient sample. This association was also apparent inthe control sample but not at significant levels. Differences in regression coefficients for theeffect of triglycerides on plasma PAI-1 levels suggest that triglyceride regulation of PAI-1 isgenotype specific Our data suggest that genetic variation at this locus contributes tobetween-individual differences in the level of plasma PAI-1, which is important in fibrinolysisand the pathogenesis of MI. (Arteriosclerosis and Thrombosis 1991;ll:183-190)

The conversion of plasminogen to proteohyti-cally active plasmin by plasminogen activa-tors is a crucial step in the fibrinolytic pro-

cess.1 Plasminogen activator inhibitor-1 (PAI-1) is afast-acting inhibitor of tissue-type plasminogen acti-vator (t-PA), the major proteolytic activator of plas-minogen in vivo.2"3 PAI-1 is a glycoprotein with amolecular weight of 50,000 d and is a member of theserine protease inhibitor family, with an arginineresidue at the reactive center.2-4 PAI-1 exists in two

From the Charing Cross Sunley Research Centre (S.D., A.H.,S.H.), Hammersmith, London, U.K.; the Departments of InternalMedicine (A.H.) and Clinical Chemistry (B.W.), Karolinska Hos-pital, and King Gustaf V Research Institute (A.H.), Stockholm,Sweden.

Supported by grants from Research Into Ageing (9/85S), theBritish Heart Foundation (RG5 and F137), the Charing CrossSunley Research Trust, the Swedish Medical Research Council(05193, 08691, and 19P-08152), the Swedish Heart-Lung Founda-tion, and King Gustaf V 80th Birthday Fund.

Address for correspondence: Sally Dawson, CXSRC 1 LurganAvenue, Hammersmith, London W6 8LW, UK.

Received March 29, 1990, revision accepted September 17,1990.

forms in plasma, with the active form existing as acomplex bound to vitronectin.5-6

In the past few years, there have been numerousCToss-sectional studies of patients with angina pecto-ris or previous myocardial infarction (MI). Thesestudies have consistently shown a decreased fibrino-lytic activity in patients compared with controls, adecrease that is due mainly to elevated plasma levelsof PAI-1.7-11 Strong support for a cause-and-effectrelation between PAI-1 elevation and risk of MI hasbeen obtained from a longitudinal cohort study of109 unselected men who had survived a first MIbefore the age of 45 years. This study indicated thata high plasma concentration of PAI-1 activity wasassociated with reinfarction, along with dyslipopro-teinemia, poor left ventricular performance, andmultiple-vessel coronary artery disease.12 In addition,increased plasma PAI-1 activity has been suggestedto have pathogenic importance in patients with deep-vein thrombosis13 and in diabetics.14

The PAI-1 gene has been cloned and localized toq21.3-q22 of chromosome 7.15 Studies with thePAI-1 cDNA probe in vitro have shown that PAI-1



184 Arteriosclerosis and Thrombosis Vol 11, No 1, January/February 1991

TABLE 1. Characteristics of Patients and Controls

Measurement

Age (yr)Weight (kg)tHeight (cm)'Choi (mmol/1)'Trigly (miao\]\)tVLDL trigly (mmol/l):tLDL chol (mmol/l)tHDL chol (mmol/1)SAOGTTtPAI (AU/ml)*

Patients

39.89 ±0.3579.50 ±1.29

174.72±0.787.28±0.142.62 ±0.201.89±0.174.86±0.121.20±0.03

8,268 ±74317.59±1.09

Controls

40.42±0.4180.57±0.95

180.89±0.666.10±0.121.48±0.140.92±0.124.00±0.101.42 ±0.03

4,439 ±20913.82±1.04

Chol, cholesterol; trigly, trigrycerides; VLDL, very low densitylipoprotein; LDL, low density lipoprotein; HDL, high densitylipoprotein; XAOGTT, incremental insulin during oral glucosetolerance test; PAI, plasminogen activator inhibitor; AU, arbitraryunits.

Values are mean±SEM.•p<0.05, tp<0.01, $p<0.001.

synthesis is regulated by a number of different agents,including glucocorticoids, cytokines, lipopolysaccha-rides, and insulin.16-18 It is unclear whether thisregulation takes place exclusively at the transcrip-tional level or also at the posttranscriptional level.PAI-1 is produced by a variety of cells in culture,including endothelial cells, hepatocytes, smooth mus-cle cells, and platelets.

In this study, we used two polymorphisms at thePAI-1 locus to investigate the possibility that ge-netic variation at the PAI-1 locus contributes toaltered plasma levels of PAI-1 and therefore todisease incidence.

MethodsSamples

The sample of 145 patients and 95 controls wasrecruited from Stockholm County, Sweden. All pa-tients suffered a first MI before the age of 45 years,and the clinical procedures used to assess cardiovas-cular disease were those as described by Hamsten etal.12 The control sample consisted of randomly se-lected, age-matched, healthy residents of StockholmCounty. Of the 145 patients, 38 were excluded fromthe analysis due to familial hypercholesterolemia ordiabetes mellitus, leaving 107 patients, 92 men and 15women, who took part in a follow-up study based onserial coronary angiograms. Analysis was done withand without data from the 15 female patients. Linkagedisequilibrium analysis on the two polymorphisms alsoincluded data from the other 38 individuals. All indi-viduals in the study are unrelated and are of Swedishorigin, except for 10 patients who had one or moreparent and/or grandparent born in Finland.

Many hemostatic and metabolic variables were mea-sured for each individual,8 and a summary of several ofthese is shown in Table 1. Serum lipoproteins weredetermined by a combination of preparative ultracen-trifugation and precipitation of apolipoprotein (apo)B-containing lipoproteins, followed by lipid analyses in

the major lipoprotein fractions.19 Oral glucose toler-ance was assessed after ingestion of glucose in a dose of1.75 g/kg body wt.20 Glucose was measured in wholeblood by the glucose oxidase method,21 and plasmainsulin was measured by a double-antibody radioimmu-noassay.22 Plasma PAI-1 activity was determined byadding a certain amount of t-PA to diluted plasma andmeasuring residual t-PA activity.2-23 The family used tostudy the inheritance of the dinucleotide repeat se-quence was initially identified from the RheumatologyClinic at Guy's Hospital for a study of joint hypermo-bility syndrome.

DNA AnalysisA 2.9-kilobase (kb) PAI-1 cDNA probe described

previously24 was used in hybridization analysis andwas kindly provided by Tor Ny, University of Umea,Sweden. The Hindlll polymorphism was that firstdescribed by Klinger et al.15 DNA was prepared fromblood using standard techniques.25 A P-labeledprobe was generated using the method of Feinbergand Vogelstein,26 and Amersham International cyti-dine-5'-[a-32P]triphosphate (aCTP), Amersham,Buckinghamshire, U.K. Standard Southern blottingwas performed using Amersham Hybond N, andfilters were washed to a final concentration ofl x standard saline citrate and 0.1% sodium dodecylsulfate after hybridization to a 32P-labeled PAI-1cDNA probe at 65°C.25 The filters were then exposedto Amersham Hyperfilm for 24 hours at -70°C.

Dinucleotide Repeat PolymorphismGenBank databases (GenBank, Los Alamos,

N.M.) with DNAstar computer software were used insequence analysis.27 Amplification of the (C-A)n re-gion in the PAI-1 gene was achieved with two 26-meroligonucleotides synthesized on a Pharmacia GeneAssembler, Pharmacia, Uppsala, Sweden, by theBiochemistry Department, Charing Cross and West-minster Medical School (Figure 1). Conditions forthe polymerase chain reactions (PCRs) were stan-dard28-29 using 1 /xg genomic DNA and 250 ng of eachprimer in a volume of 50 jtl. Each PCR reactionmixture contained 200 /imoles each of thymidine-5'-triphosphate, guanosine-5'-triphosphate, and aCTP,100 Aimoles each oATP plus 100 ^moles [a-^S]aATP, 10 mM Tris HC1 (pH 8.3), 1.5 mM MgCl2, 50mM KC1, 5fi\ dimethyl sulfoxide, and 1 unit Perkin-Elmer Taq polymerase. Samples were overlaid withparaffin oU and underwent one cycle of 3 minutes at90°C, of 1 minute at 55°C, and of 1 minute at 72°C,and then 50 cycles of 1 minute at 90°C, of 1 minute at55°C, and of 1 minute at 72°C. Half the resultingPCR species was mixed with 4 /il formamide runningbuffer after ethanol precipitation and was run on adenaturing polyacrylamide gel for approximately 4hours. The gel was then dried and exposed to Amer-sham Hyperfilm /3-max for 2-3 days. BacteriophageM13 mplO vector DNA was used as a template toproduce a dideoxy sequencing ladder, which wasused to size the PCR species.



Dawson et al Genetic Variation and PAI-1 Levels 185

PAI (CA)n Repeat

1 2 3 4 6 7 8 9

AA6CT6A66C 666A68AACA TTTSAACC66 ATTC86A66C T6CA6T6A6C TAT6ATT6CA CCACT6C8CT

7741Oligo 1 A TTTGAACCGG ATTCGGAGGC TGC

CCA6TCT6TB TBACABT6A8 ACCCT6TCTC TT|ACACACAC ACACACACAC ACACACAC6C ACACACACAB

ABABAAATTA 6AA6ATACTB AATTB6CABA A8A6AA886A AATABAAATT AAAATACT6A ATAB86BABC

^ Oligo 2 CT TATCCCCTCG

ABTBAACA66 66ATACCCAA AABCCAABAB

TCACTTGTCC CC 7S80

Statistical Analysis

X1 analysis was used to determine whether therewas any significant difference in allele or genotypefrequencies between control and patient samples. Astepwise regression of the PAI-1 value was per-formed separately in both patient and control sam-ples after adjusting for age, body mass index (BMI),very low density lipoprotein triglycerides (VLDL-TG), and oral glucose tolerance test (OGTT). Analysiswas performed with and without the 10 Finnishpatients, and results were not significantly different.These adjusted PAI-1 values were then used in

FIGURE 1. Schematic showing the di-nucleotide repeat polymorphism. Theupper panel of Figure 1 shows theposition of the (C-A)n repeat region inthe PAI-1 gene and the oligonucleo-Mes used to amplify this region. Posi-tions ofexons (1-9) are represented byblack boxes. The numbers relate to thenumber of bases from the start of thesequence in the GenBank database.The lower panel demonstrates the (C-

A)n repeat polymorphism with a bacte-riophage Ml 3 sequencing reaction as asize marker. PAI-1, plasminogen acti-vator inhibitor-1.

standard analysis of variance (ANOVA) and t teststatistical tests to assess whether genotype had astatistically significant effect on plasma PAI-1 levels.Regression analysis was also performed separately ineach genotype class of the HindlW RFLP to comparethe regression slopes. One individual with an ex-tremely high PAI-1 value was excluded from the finalanalysis on the basis that this value was extreme(defined as a value more than three interquartileranges above the 75th percentile) and thereforehaving a disproportionately large effect on the statis-tical analysis of the data. An "average excess" calcu-




oz/z + 6

•aZ+2/Z+10

Dz / z

6z/z + 10

z+6/z+10

•z/z + 10 z/z + 6

Z+6/Z+10

6 6Z-4/Z+10

oZ - 4 / Z + 1 0

FIGURE 2. Pedigree showing mendelian inheri-tance of the dinucleotide repeat polymorphism, o,Females; o, males.

lation was performed for each allele in the dinucle-otide repeat polymorphism as described by Sing andDavignon30 and Templeton.3i

ResultsThe (C-A)n repeat polymorphism used in the study

is shown in Figure 1. In the previously reportedHindlU RFLP,1' the "2" allele represents the pres-ence of an additional Hindlll site compared with the"1" allele. The (C-A)n repeat sequence within thePAI-1 gene identified from GenBank databases withDNAstar computer software was found to be a lengthpolymorphism with eight alleles detected in the sam-ple. Each allele differed from another by the numberof -CA- dinucleotides in the repeat sequence, that is,by multiples of two base pairs. This concurs with theresults of Weber and May,27 who found similar poly-morphisms at many other loci. Nomenclature of thedinucleotide repeat alleles was as suggested by Weberand May, the most frequent allele being designated"z" and the other alleles designated by their base-pairdifferences from z, that is, "z+2, z+4. ... ". As withother dinucleotide repeat polymorphisms, the PAI-1polymorphism was found to be inherited in a mende-lian fashion in an extended family (Figure 2). Individ-uals in this family had five different alleles, and thefragment sizes observed were constant over threegenerations (six informative meioses).

The patient and control samples were genotyped forboth the Hindlll RFLP and the dinucleotide repeatpolymorphism. As shown in Table 2, these two poly-morphisms are significantly correlated (p<0.001),suggesting that they are in strong linkage disequilib-rium, the larger C-A repeat alleles being in disequilib-rium with the Hindlll 2 allele. There was no signifi-cant difference between the frequency of any allele inthe patient and control samples. Both polymorphismswere in Hardy-Weinberg equilibrium.

The characteristics of the patient and control sam-ples are summarized in Table 1. Significant differencesbetween the two samples have been discussed previ-ously, among them the reduced fibrinolytic activity inpatients due to increased levels of plasma PAI-1.8

Stepwise regression analysis of plasma PAI-1 levelsagainst BMI, VLDL-TGs, and incremental insulin

during the OGTT showed that in the patient sample,only VLDL-TGs and BMI had a significant indepen-dent effect on plasma PAI-1 levels, and these wereincluded in the regression equation. In the controlsample, BMI was not included in the regressionequation, as it was found not to have a significantindependent effect on PAI-1 levels. VLDL-TG levelsexplained 18.2% of the PAI-1 variation in patients and40.0% of the variation in controls; BMI explained anadditional 23.4% of PAI-1 variation in the patients.

Analysis based on these adjusted PAI-1 valuesshowed an association with Hindlll genotype (Table 3)

TABLE 2. linkage Disequilibrium Between HindUl RFLP andDinucleotide Repeat Polymorphism

C-A repeatgenotype

z-4/z-2

z-2/z-2z-2/zz-2/z+Wz/zzlz+2z/z+4zJz+6z/z+8z/z+10z+2/z+2z+2lz+4z+2/z+8z+2/z+lOz+4lz+4z+4/z+6z+4/z+8z+4/z+lOz+6/z+6z+8/z+8z+8/z+lO

1/1

1

3

23

6

3

1

3

2

1

1

1

Hindlll genotype

1/2

1

1

9

34

33

2125351

4

2

1

2/2

1

476

224

134

4175

21

RFLP, restriction fragment length polymorphism.Data are shown for all 221 individuals in the patient and control

samples who were genotyped for both polymorphisms plus somedata from additional individuals. Pearson's #>=0.4377,p<0.0001.




TABLE 3. Association of HindlU RFLP and Plasma PAI-1 Levels

Hin&lllgenotype

Controls(ANOVA-0.52)

Patients(ANOVA=0.19)

111

112

212

15.26±1.81(/i»15)13.47±1.02(n=33)13.79±2.13 (n=20)

19.56±1.95 (n=20)17.84±1.44(n=45)15.13±132(n=30)

RFLP, restriction fragment length polymorphism; PAI-1, plas-minogen activator inhibitor-1; OGTT, oral glucose tolerance test;BMI, body mass index; VLDL, very low density lipoprotein.

PAI-1 values (mean±SEM) are those regressed for incrementalinsulin during OGTT, BMI, age, and VLDL triglyceride levels.ANOVA is a standard one-way analysis of variance on the PAI-1values with genotype.

within both patient and control samples, with the HIgenotype associated with higher levels of plasma PAI-1than the 212 genotype. In the patient sample, mean

PAI-1 levels for the 111 group were 29% higher thanthe 212 group (p=0.068 in a / test).

One-way ANOVA on the (C-A)n repeat data (Fig-ure 3) showed a significant association between(C-A)n repeat genotype and PAI-1 levels in thepatients (/?=0.03), with the shorter alleles associatedwith higher levels. A similar trend was observed inthe control sample, although not at a significant level(p=0.14). Due to the large numbers of alleles pres-ent in the sample, we have chosen to use the averageexcess calculation to estimate the effect on PAI-1levels associated with this polymorphism.30"31 Aver-age excess is an estimate of the average effect eachallele has on PAI-1 levels within the sample. Resultsof the calculation of average excess for each (C-A)nrepeat allele are shown in Table 4. In both patientsand controls, the z+2 allele and all smaller alleles

z+fl/z+10

i

2.

Patients.Controls.

f(3

Z-4/Z-2

10 20 FIGURE 3. Bar graphs showing plasma PAI-1 levelsand frequency of dinucleotide repeat genotypes. Theupper panel shows the frequency of the (C-A)D

repeat genotype in both samples. The lower panelshows the average PAI-1 level in each (CA)n repeatgenotype. PAI-1, plasminogen activator inhibitor-1.

20 30

Adjusted PAJ-1 Level




TABLE 4. Calculation of Average Excess for Alleles of theDinucleotJde Repeat Polymorphism

Average excess on PAI-1 level

Dinucleotiderepeat allele

z-2zz+2z+4z+6z+8z+10

Patients(No. of alleles)

+0.35 (2)+ 1.75(82)+ 1.09(36)+0.98 (37)-9.14 (5)-4.76 (14)-6.33 (9)

Controls(No. of alleles)

+0.70 (4)+0.01 (60)+ 1.18(28)-1.24(25)

+ 1.98(8)-5.97 (6)

PAI-1, plasminogen activator inhibitor-1.Calculations were performed as described by Templeton.31

(i.e., z, z-2) are associated with a higher plasmaPAI-1, and those alleles of a greater size than z+4are generally associated with a lower plasma PAI-1.

Next, we investigated whether the effect on plasmaPAI-1 levels of increasing plasma triglycerides, incre-mental insulin during OGTT, and BMI was the samein individuals with different genotypes. Regressionanalysis was performed separately in each HindlUgenotype class, and the regression slope for eachvariable was compared (Table 5). These results showa different slope in each genotype class for VLDL-TGs. PAI-1 levels for the 1/1 genotype class increaseat a greater rate with increasing VLDL than the 1/2and 2/2 classes in both patient and control samples.The data also show that insulin response to oralglucose challenge has a significant effect only onindividuals of the HindlH 2/2 genotype in bothpatient and control samples (p<0.01). Analysis ofthis type was not possible on the (C-A)n repeatpolymorphism due to the small number of individualsin each of the 17 genotype classes.

DiscussionThe use of nonfunctional sequence variations (e.g.,

RFLPs) as markers for other functionally importantchanges in the same gene relies on these two differ-ences being in linkage disequilibrium,32 where thepresence of one allele is indicative of the presence ofthe other at a frequency higher than that expected bychance alone. This principle has been used widely inpopulation studies with RFLPs to identify candidate

genes that are important in the development ofpolygenic disease.33-34 One way to do this is tocompare the frequency of polymorphic alleles incontrol and patient populations. This requires theuse of very large samples to detect significant differ-ences in frequency because it relies on an associationbetween a genetic marker and disease incidence.This is particularly difficult with a multifactorialdisease, where there are many genetic and otherfactors involved in any one individual's presentationas a patient. Studies on the association betweenpolymorphisms in a candidate gene and the plasmalevel of its gene product do not require such largenumbers to detect an association, and they suggest afunctional variation in sequence in the same gene.This technique has been used in the past to identifypotential functional changes in gene sequence at thefibrinogen35 and apo B loci,25-26-37 among others.

The positive and statistically significant relationbetween plasma PAI-1 levels and both VLDL-TGand plasma insulin concentrations (in the 2/2 class)that we have observed here confirms observations byothers8-38-41 and raises the possibility that VLDL andinsulin might be major physiologic regulators ofPAI-1 activity in plasma. Recent experiments onhepatoma cells in vitro also suggest a regulatory rolefor insulin.43 Our data suggest that this regulation ofPAI-1 by VLDL and insulin is genotype specific:Patients and controls of //indlll genotype 1/1 show atwofold to threefold greater increase in plasma PAI-1with an increase in VLDL-TGs than do individuals ofgenotype 2/2 (Table 5), whereas the OGTT data onlyhave a significant effect on 2/2 individuals. It ispossible that the Hindlll RFLP is in linkage disequi-librium with a base change at a site of functionalimportance in the regulation of PAI-1 by VLDL/insulin. One such site would be a sequence in thePAI-1 gene involved in the binding of a transcriptionfactor whose level is indirectly or directly altered byVLDL levels or VLDL composition.

The data show an association between Hindlllgenotype and plasma PAI-1 levels although this wasnot significant at conventional levels. The Hindlll 1allele in both patient and control samples was asso-ciated with higher levels (Table 3). The position ofthzHindlU polymorphism has not yet been identifiedalthough we have tentatively mapped it to the 3'-

TABLES.

Hindmgenotype

Total111112212

Regression

N or n

93194430

Analysis on Three Hindm RFLP

Patients

Slope

3.034.234.071.59

SEM (slope)

0.61.4132.2

GenotypesVLJDL triglycerides

Sig0.010.010.000.06

N or n

68153320

Slope

4.028.986.373.27

Controls

SEM (slope)

0.65.42.33.3

Sig0.000.130.010.00

RFLP, restriction fragment length polymorphism; VLDL, very low density lipoprotein."Slope" is the slope of regression B, given with its standard error. "Sig" is the t significance and is an estimate of the

significance of the fit of the regression line.




flanking region of the gene (S. Dawson et al, unpub-lished data).

In both the control and patient samples analyzedhere, a clear association between the smaller (C-A)n

repeat genotype and raised plasma PAI-1 levels wasfound. This association is most evident within thesample of young male MI survivors, but the sametrend is also apparent within the control sample.There is recent evidence that purine-pyrimidine re-peats of this kind may be sites for topoisomerase IIactivity, which could well be important in transcrip-tion.42 Such a direct effect is unlikely in this casebecause the polymorphism is in intron 3 of the gene,but it is instead more likely to be acting as a markerfor a functional change elsewhere in the PAI-1 gene.

The dinucleotide repeat polymorphism appears tobe in linkage disequilibrium with the Hindlll RFLP(Table 2). This association of the shorter alleles withthe Hindlll 1 allele and with higher PAI-1 levelssuggests that the dinucleotide repeat polymorphismarises through a slippage mechanism rather than anyrecombination event and that the longer the stretchof dinucleotide repeats, the more likely this slippageevent is to occur. This hypothesis is supported by thefact that there were only two alleles smaller than thepredominant z allele in the samples genotyped andfive that were larger. The distribution of allele fre-quency in the two samples was found to be bimodal,with the antinode around the z+6 allele (Figure 3,upper panel). Other studies with dinucleotide repeatpolymorphisms have also found bimodal distributionsof allele frequency, but the significance of this isunknown.44

We have demonstrated that genetic variation inthe PAI-1 gene is a significant factor in determininglevels of PAI-1 in plasma in two independent sam-ples. It is most likely that the (C-A)n repeat polymor-phism and Hindlll RFLP are in linkage disequilib-rium with a sequence variation elsewhere in the genethat influences the expression of the PAI-1 gene.Regulation of the gene by several different factors invitro has previously been shown to be important incontrolling synthesis of PAI-1.15-17 These experi-ments indicate that regulation of PAI-1 at the tran-scriptional level is a significant component in deter-mining the amount of active PAI-1 in plasma. SincePAI-1 is an important component of the fibrinolyticsystem, identification of a sequence variation in thePAI-1 gene that affects its expression would be usefulto identify an individual predisposed to arterial andthrombotic disease.

AcknowledgmentsWe thank Tor Ny and Leif Strandberg for gener-

ously providing the PAI-1 cDNA probe.

References1. Castellino FJ: Biochemistry of human plasminogen. Semin

Thwmb Hemost 1984;10:18-23

2. Chmielwska J, Ranby M, Wiman B: Evidence for a rapidinhibitor to tissue plasminogen activator in plasma. ThwmbRes 1983;31:427-436

3. Kruithof EKO, Tran-Thang C, Ransijn A, Bachmann F:Demonstration of a fast-acting inhibitor of plasminogen acti-vators in human plasma. Blood 1984;64:907-913

4. Pannekoek H, Veerman H, Lambers H, Diergaarde P, VerweijC, van Zonneveld A, van Mourik J: Endothelial plasminogenactivator inhibitor (PAI): A member of the serpin gene family.EMBO J 1986^:2539-2544

5. Dederck P, De Mol M, Alessi M, Baudner S, Pagnes E,Priessner K, Muller Berghaus G, Collen D: Purification andcharacterization of a plasminogen activator inhibitor-1-bindingprotein from human plasma./BJo/C/KJn 1988;263:15454-15461

6. Wiman B, Almquist A, Sigurdardottir O, Lindahl T: Plas-minogen activator inhibitor 1 (PAI) is bound to vitronectin inplasma. FEBS Lett 1988;242:125-128

7. Aznar J, Esters A, Tormo G, Sapena P, Tormo V, Blanch S,Espana F: Plasminogen activator inhibitor activity and otherfibrinolytic variables in patients with coronary artery disease.Br Heart J 1988^59:535-541

8. Hamsten A, Wiman B, De Faire U, Blomback M: Increasedplasma levels of a rapid inhibitor of tissue plasminogenactivator inhibitor in young survivors of myocardial infarction.NEngiJMed 1985^313:1557-1563

9. Johnson O, Mellbring G, Nilsson T: Defective fibrinolysis insurvivors of myocardial infarction. Int J Cardiol 1984;6:380-382

10. Paramo JA, Colucci M, Collen D, van de Werf F: Plasminogenactivator inhibitor in the blood of patients with coronary arterydisease. Br Med J 1985;291:573-574

11. Hamsten A, Blomback M, Wiman B, Svensson J, Szamosi A,De Faire U, Mettinger L: Haemostatic function in myocardialinfarction. Br Heart J 1986^5:58-66

12. Hamsten A, De Faire U, Walldius G, Dahlen G, Szamosi A,Landou C, Blomback M, Wiman B: Plasminogen activatorinhibitor in plasma: Risk factor for recurrent myocardialinfarction. Lancet 1987;2:3-9

13. Wiman B, Ljungberg B, Chmielewska J, Urden G, BlombackM, Johnson H: The role of the fibrinolytic system in deep veinthrombosis. Lab CUn Med 1985;105:265-270

14. Auwerx J, Bouillon R, Collen D, Gebeors J: Tissue typeplasminogen activator antigen and plasminogen activatorinhibitor in diabetes mellitus. Arteriosclerosis 1988;8:68-72

15. KUnger K, Winqvist R, Riccio A, Andreasen PA, Sartorio R,Nielson LS, Stuart N, Stanislovitis P, Watkins P, Douglas R, etal: Plasminogen activator inhibitor type 1 gene is located atregion q21.3-q22 of chromosome 7 and genetically Linked withcystic fibrosis. Proc NatlAcad Sci USA 1987;84:8548-8552

16. Riccio A, Lund L, Sartorio R, Lania A, Andreasen P, Dan0 K,Blasi F: The regulatory region of the human plasminogenactivator inhibitor type-1 gene. Nucleic Acids Res 1988;16:2805-2824

17. Emeis J, Kooistra T: Interleukin-1 and lipopolysaccharideinduce a fast acting inhibitor of tissue-type plasminogenactivator in vivo and in cultured endothelial cells. / Exp Med1986;163:1260-1266

18. Alessi M, Juhan-Vague I, Kooistra T, Declerck P, Collen D:Insulin stimulates the synthesis of plasminogen activator inhib-itor 1 by the human hepatocellular cell line HepG2. ThrombHaemost 1988;60:491-494

19. Carlson K: Lipoprotein fractionation. / CUn Pathol 1973;5(suppl 26):32-37

20. Efendic S, Wajngot A, Cerasi E, Luft R: Insulin release,insulin sensitivity, and glucose intolerance. Proc NatlAcad SciUSA 1980-,77:7425-7429

21. Hugget ASG, Nixon DA: Use of glucose oxidase, percwridaseand O-dianisidine in determination of blood and urinaryglucose. Lancet 1957;2:368-370

22. Hales CN, Randle PJ: Immunoassay of insulin with insulin-antibody precipitate. Biochem J 1963;88:137-146

23. Wiman B, Chmielewska J, Ranby M: Inactivation of tissuetype plasminogen activator in plasma: Demonstration of a



190 Arteriosclerosis and Thrombosis Vol 11, No 1, January IFcbmaiy 1991

complex with a new rapid inhibitor. / Bid Chem 1984;259:3644-3647

24. Ny T, Sawdey M, Lawrence D, Millan J, Loskutoff D: Cloningand sequencing of a cDNA coding for the human B-migratingendothelial-ceU-type plasminogen activator inhibitor. ProcNatlAcad Sd USA 1986;83:6776-6780

25. Talmud PJ, Barni N, Kessling A, Carlsson P, Darnfors C,Bjursell G, Gatton D, Wynn V, Kirk H, Hayden M, HumphriesS: Apolipoprotein B gene variants are involved in the deter-mination of serum cholesterol levels: A study in normo- andhyperlipidaemic individuals. Atherosclerosis 1987;67:81-89

26. Feinberg A, Vogelstein B: A technique for radiolabellingDNA restriction endonuclease fragments to high specificactivity. Anal Biochem 1983;132:6-13

27. Weber J, May P: Abundant class of human DNA polymor-phisms which can be typed using the poh/merase chain reac-tion. Am J Hum Genet 1989;44:388-396

28. Saiki R, Gelfand D, Stoffel S, Scharf S, Higuchi R, Horn GT,Mullis K; Primer directed enzymatic amplification of DNAwith a thermostable DNA polymerase. Science 1988;239:487-491

29. Mullis K, Faloona F: Specific synthesis of DNA in vitro via apolymerase-catalyzed chain reaction. Methods Enzymol 1987;155:335-350

30. Sing CF, Davignon J: Role of the apolipoprotein E polymor-phism in determining normal plasma lipid and lipoproteinvariation. Am J Hum Genet 1985^7:268-285

31. Templeton AR: The general relationship between averageeffect and average excess. Genet Res 1987;49:69-70

32. Bodmer W, Bodmer J, Ihde D, Adler S: Genetic and serolog-ical association analysis of the HL-A leukocyte system, inMorten NE (ed): Computer Applications in Genetics. Honolulu,University of Hawaii Press, 1969, pp 117-127

33. Humphries SE: DNA polymorphisms of the apolipoproteingenes—Their use in the investigation of the genetic compo-nent of hyperlipidaemia and atherosclerosis. Atherosclerosis1988;72:89-108

34. Cooper DN, Schmidtke J: Diagnosis of genetic disease usingrecombinant DNA: Second edition. Hum Genet 1989;83:307-334

35. Humphries SE, Cook M, Dubowitz M, Stirling Y, Meade TW:Role of genetic variation at the fibrinogen locus in determi-

nation of plasma fibrinogen concentrations. Lancet 1987;1:1452-1454

36. Berg K: DNA polymorphism at the apolipoprotein B locus isassociated with lipoprotein level. Clin Genet 1986;30:515-521

37. Law A, Powell L, Brunt H, Knott T, Altaian D, Rajput J,Wallis S, Pease R, Priestley L, Scott J, Miller G, Miller N:Common DNA polymorphisms within coding sequence ofapolipoprotein B gene associated with altered lipid levels.Lancet 1986;l:1301-1303

38. Juhan-Vague I, Valadier J, Alessi M, Aillaud M, Ansaldi J,Philip-Joel C, Holvolt P, Serradinugni A, Collen D: DeficienttPA release and elevated PA inhibitor in patients with spon-taneous or recurrent deep vein thrombosis. Thromb Haemost1987^7:67-72

39. Mehta J, Mehta P, Lawson D, Saldeen T: Plasma tissueactivator inhibitor levels in coronary artery disease: Correla-tion with age and serum trigryceride concentrations. J Am CollCardiol 1987;9:263-268

40. Juhan-Vague I, Alessi MC, Badier C, Valadier J, Aillaud MF,Atlan C: Relationship between plasma insulin, triglyceride,body mass index and plasminogen activator inhibitor-1. Dia-bete Metab 1987;13:331-336

41. Juhan-Vague I, Alessi M, Jory P, Thirion X, Vague P, DederckP, Serrodinugni A, Collen D: Plasma plasminogen activatorinhibitor 1 in angina pectoris: Influence of plasma insulin andacute phase response. Arteriosclerosis 1989;9:362-367

42. Kooistra T, Bosma PJ, Tons HAM, van den Berg AP, Meyer P,Princen HMG: PAI-1: Biosynthesis and mRNA level areincreased by insulin in cultured human hepatocytes. ThrombHaemost 1989;62:723-728

43. Spitzner JR, Chung IK, Muller MT: Eukaryotic topoisomeraseII preferentially cleaves alternating purine-pyrimidinerepeats. Nucleic Acids Res 1989;18:1-11

44. Ludwig EH, Friedl W, McCarthy BJ: High-resolution analysisof a hypervariable region in the human apolipoprotein B gene.Am J Hum Genet 1989;45:458-464

KEY WORDS • plasminogen activator inhibitor-1 • myocardialinfarction • fibrinolysis • restriction fragment lengthpolymorphisms • dinucleotide repeat polymorphisms • geneticvariation • DNA



genetic variation at the plasminogen activator inhibitor-1 locus is associated with altered levels...

Documents