|Home | About | Journals | Submit | Contact Us | Français|
Platelet endothelial aggregation receptor-1 (PEAR1) is a recently identified platelet transmembrane protein that becomes activated by platelet contact. We looked for novel genetic variants in PEAR1 and studied their association with agonist-induced native platelet aggregation and with aspirin's inhibitory effect on platelets.
We genotyped PEAR1 for 10 single nucleotide polymorphisms (SNPs), selected for optimal gene coverage at a density of 4kb, in 1486 apparently healthy individuals from two generations of families with premature CAD. Subjects had a mean age of 45 years; 62% were white and 38% African American. Platelet aggregation to collagen, epinephrine, and ADP was measured in platelet rich plasma, at baseline and after 2 weeks of aspirin (ASA, 81 mg/day), and genotype-phenotype associations were examined separately by ethnicity using multivariable generalized linear models adjusted for covariates. The C allele of SNP rs2768759 [A/C], located in the promoter region of the gene, was common in whites and uncommon in African Americans (allele frequency 70.2% vs 17.7%). The C allele was generally associated in both ethnic groups with increased aggregation of native platelets to each agonist. Following ASA, the associations were stronger and more consistent, and remained significant when post ASA aggregation was adjusted for baseline aggregation, consistent with a relationship between the C allele and reduced platelet responsiveness to ASA. The PEAR1 SNP explained up to 6.9% of the locus specific genetic variance in African Americans and up to 2.5% of the genetic variance in whites following ASA.
PEAR1 appears to play an important role in agonist-induced platelet aggregation and in the response to ASA in both whites and African Americans.
Platelets initiate coronary artery thrombosis by aggregating on ulcerated atherosclerotic plaques.1 Although aspirin is widely used to inhibit platelets and prevent arterial thrombosis, incident acute coronary artery disease (CAD) events may still occur if platelet aggregation is not sufficiently suppressed.2 Considerable heterogeneity is known to exist both in native platelet aggregation and in the magnitude of platelet suppression by aspirin.3-7 The reason for this phenotypic heterogeneity is not completely known, but genetic factors are thought to play an important role. Native platelet aggregation, measured by a number of in vitro tests, has been shown to be heritable in the Framingham Offspring Study8 and also in a large study in families with premature CAD, consistent with a genetic influence on platelet function.9 Polymorphisms in the platelet surface membrane adhesion receptors for collagen and fibrinogen (integrins α2 and β3, respectively) have been associated with platelet reactivity,10-11 and a polymorphism in the gene encoding the beta subunit of G proteins (GNB3) predicts abnormally increased platelet aggregation in healthy subjects.12
Aspirin acetylates and irreversibly inhibits platelet cyclo-oxygenase-1 (COX1), resulting in decreased conversion of arachidonic acid to thromboxane A2. Because maximal platelet aggregation is dependent in part on thromboxane A2-associated amplification, aspirin treatment results in inhibition of platelet aggregation initiated by a variety of agonists. Several studies have suggested that persons with greater residual platelet aggregability while taking aspirin have a higher risk of incident CAD.4,7,13,14 As with native platelet function, the inhibitory effect of aspirin is heritable, consistent with a genetic influence.3 Genetic variants in COX1 itself are uncommon and account for only a small portion of platelet responsiveness to aspirin15. Variants in the pathways indirectly involving COX1, such as aggregation initiated by collagen or epinephrine, likely also play a role in platelet response to aspirin.3 A polymorphism in platelet GPIIIa (PlA2) has been associated with less inhibition by aspirin following collagen stimulation.16
Another type of indirect pathway is activated by platelet-to-platelet contact.17,18 During a recent search for platelet receptors that become tyrosine phosphorylated after platelet contact, several novel transmembrane molecules, including “platelet endothelial aggregation receptor-1” (PEAR1)17,18 and ephrins/Eph kinases,19 were identified which interact with adjacent platelets to send signals that enhance and stabilize platelet thrombi (a type of “outside-in signalling”).20 Following activation by collagen, PEAR1 becomes phosphorylated on certain cytoplasmic tyrosine and serine residues, in a manner dependent on functionality of GPIIb/IIIa, and involving association of the ShcB adaptor protein.17 Although the ligand for PEAR1 is unknown, inherited PEAR1 variations that alter expression or function of this platelet signalling molecule could modify agonist-induced aggregation in native platelets. In addition, a genetic variant in PEAR1 could be an important determinant of residual platelet function during aspirin treatment, since the COX1/thromboxane A2 pathway will be strongly inhibited by aspirin, and maximal aggregation will then be dependent on other secondary signalling pathways. However, no previous studies have examined whether PEAR1 does in fact play a role in agonist-induced platelet aggregation or in the responsiveness to aspirin.
In this study, we found that a novel variant in the PEAR1 gene (rs2768759) is associated with enhanced agonist-induced platelet aggregation and explains a significant portion of the genetic variance associated with both native platelet aggregation (0.22 to 5.3%) and residual platelet aggregation following aspirin treatment (1.3 to 6.9%). Despite a markedly higher allele frequency in white than African American individuals (70.2% vs. 17.7%), the associations between the PEAR1 variant and increased platelet aggregation are similar in the two ethnic groups.
Participants (n=1486) were recruited from a study of the genetic determinants of platelet responsiveness to aspirin in healthy siblings and offspring of 427 probands with CAD < 60 years of age.21 In all subjects, platelet aggregation to collagen, epinephrine, ADP, and arachidonic acid (see below) were measured at baseline and again after 14 days of aspirin, 81 mg/day. All subjects were aspirin-free for 2 weeks before baseline measurements, and drugs that affect platelet function were proscribed during the trial.21 Environmental exposures (e.g., diet, exercise, smoking) known to affect platelet function were avoided or protocolized to provide consistency for the baseline and post aspirin platelet function measurements. The study was approved by the Johns Hopkins Institutional Review Board and monitored by an external Data Safety and Monitoring Board constituted by the National Heart, Lung and Blood Institute. Written informed consent was obtained from all participants.
A medical history and physical examination, including blood pressure measurement, were performed. Hypertension was defined as a mean blood pressure ≥ 140/90 mm Hg and/or on anti-hypertensive medication. Current smoking was defined as any self-reported smoking within the previous 30 days or an expired carbon monoxide level of >8ppm. Plasma levels of glucose, total cholesterol, and triglycerides were measured using standard methods (Cholestech Corporation, Hayward, CA) after a 12 hour overnight fast. Low-density lipoprotein cholesterol (LDL-C) was estimated using Friedewald's formula.22 Diabetes was defined as a fasting plasma glucose level ≥ 126 mg/dl or the use of a hypoglycemic agent. Weight and height were measured, and body mass index (BMI) was calculated by dividing the weight in kg by the square of the height in m. Plasma fibrinogen was measured using an automated optical clot detection device (Behring Coagulation System; Dade-Behring, Newark, DE).
Ten single nucleotide polymorphisms (SNPs) were selected to cover the PEAR1 gene sequence at a density of 4kb, from 10kb before the start coordinate to 5kb after the stop coordinate as obtained from the UCSC genome browser (www.genome.ucsc.edu). SNP selection was made using in-house custom software (available at http://mendel.som.jhmi.edu/pub/bioinformatics/SHolder.html) derived from the “SNP cherry picker” approach described by Harris et al.23 The method is a stochastic optimization of the number of covered bases with the maximum probability of the selected SNPs being informative. SNP selection was based on the calculated probability that a given SNP would be real and polymorphic, and included information from the HapMap data base (www.hapmap.org), the dbSNP data base (http://www.ncbi.nlm.nih.gov/projects/SNP/), and proprietary information from Illumina (San Diego, CA) concerning the design score of a putative SNP of interest. The probability of a SNP being real and polymorphic was calculated as p = 0.3L + 0.2H + 0.2S + 0.1M + 0.1V, where L = Illumina score, H = heterozygosity (from dbSNP), S = success rate (from dbSNP), M = 1 if present as tag SNP in the HapMap, or zero if not, and V = the number of validation sources/10. The final selection involved a balance between the value of p and the adequacy of gene coverage (all p values were > 0.2 and expected coverage was 97.1% at a 4kb density; actual coverage turned out to be 78.9% since some of the SNPs were not polymorphic or successfully genotyped). Genotyping of SNPs was performed by the Johns Hopkins Genetics Core using the Illumina BeadArray platform. A commercially available 32-STR marker ancestry panel was genotyped by deCODE Genetics (Reykjavik, Iceland).24
Platelet function testing was performed at baseline and after aspirin therapy.3 Blood was obtained by venipuncture and collected into tubes containing either EDTA (for blood cell counts) or 3.2% sodium citrate (for platelet function testing and fibrinogen levels). Optical aggregometry in platelet rich plasma (PRP, 200,000 platelets per microliter) was performed in a 4-channel PAP-4 Aggregometer (Horsham,PA) after stimulating samples with collagen (2 and 5 μg/mL) (Chronolog Corp, Havertown, PA), epinephrine (2 and 10 μM), ADP (10 μM), and arachidonic acid (1.6 mM). Percent maximum aggregation within 5 minutes of agonist stimulation was recorded. The intra-assay coefficient of variation was 16% for collagen, 20% for epinephrine, 9% for ADP, and 7% for arachidonic acid. Collagen and epinephrine were chosen because of their biological importance as initiators of platelet aggregation, the known variability of these pathways to the inhibitory effects of aspirin (consistent with a genetic influence), and the moderate to high heritability of baseline and post-aspirin aggregation responses to both agonists.3,9 In addition, collagen was one of the agonists shown by Nanda et al to result in PEAR1 phosphorylation.17 ADP was chosen because of its clinical interest, high heritability of both baseline and post-aspirin platelet aggregation,3,9 and the thought that platelet aggregation might be particularly dependent on ADP following aspirin-induced inhibition of thromboxane A2. Arachidonic acid was selected because of its direct involvement in the pathway inhibited by aspirin.
Variable normality was confirmed by the Wilk-Shapiro test, and log-transformed where necessary. Demographic characteristics between ethnic groups were compared by t-test or X2 as appropriate. Multivariable linear regression analyses were adjusted for cardiac risk factors (age, sex, hypertension, current smoking, BMI, diabetes, LDL-C, and fibrinogen levels). Z scores were calculated for the residual values and were used in all multivariable genotype analyses. Phenotype-genotype associations were determined by multivariable linear regression analysis using the ASSOC method in S.A.G.E., which takes into account familial relationships (S.A.G.E.  Statistical Analysis for Genetic Epidemiology, Release 5.3). An additive genetic model was used. An aspirin response phenotype was created by linear regression of the post-aspirin aggregation values against the pre-aspirin values, utilizing the residuals and the intercept from the regression. Regression analyses for white and African American subjects were performed separately.
We calculated the genetic variance attributable to rs2768759 by obtaining the corresponding phenotypic values for each genotype group and the allele frequencies for each ethnic group. The locus specific variance (VL and the components (additive and dominant)) of the genetic variance were obtained using standard formulas.25
The study population consisted of 1486 apparently healthy individuals from 427 families with premature CAD. Study participants were siblings (n = 594) of probands who had documented CAD before age 60, adult offspring of the probands or the siblings (n = 743), or the co-parent of the offspring (n = 149). The average age of subjects was 45 ± 13 years (range 21 to 79); 56% were female; 62.4 % reported their ethnicity as white and 37.6% reported that they were African American. Self-reported ethnicity was verified as correct in 96% of cases by a 32 STR ancestry panel, based on >60% of genotypes in the appropriate ancestry cluster (CEU, YRI, or CHB+JPT).24 Of the 427 families, 264 were white and 163 were African American (mean family size 3.51 ± 2.7 and 3.43 ± 2.7, respectively). Compared to white subjects, African Americans were more often female, and had higher prevalences of hypertension, current smoking, and diabetes (Table 1). Mean body mass index and serum fibrinogen were higher, and LDL cholesterol lower in African Americans (Table 1).
At baseline, 6.4% of subjects had no measurable platelet aggregation to arachidonic acid. Although 14.4% of subjects were taking aspirin regularly prior to the study, all reported complete cessation for at least 2 weeks prior to the baseline measurement. However, we suspect that those with zero baseline platelet aggregation were either non-compliant or were unknowingly exposed to aspirin or non-steroidal anti-inflammatory agents. Based on questionnaires and pill counts, all subjects took aspirin for two weeks according to protocol, except that 12% of subjects admitted to missing one or two of the 14 doses. All participants had taken a dose in the 24 hours preceding the post-aspirin measurements.
Of the 10 SNPs genotyped in PEAR1, 5 SNPs in whites and 7 SNPs in African Americans were polymorphic. Linkage disequilibrium (LD) patterns and the positions of the selected SNPs are shown by ethnicity group in Figure 1. The genotype frequency of those SNPs successfully genotyped for each ethnic group are shown in Supplementary Table 1 (available on-line). A SNP in PEAR1, rs2768759, was found to be associated with nearly all platelet aggregation phenotypes. This SNP was not in LD with any of the other SNPs genotyped in PEAR1 (Figure 1). The SNP was in Hardy Weinberg Equilibrium within each ethnic group (whites, p=0.73, and African Americans, p =0.87). The prevalence of the homozygous CC genotype was more than 25 fold higher in whites (Figure 2). We tested whether the CC genotype in African Americans was related to European admixture using the deCODE 32 STR marker ancestry panel. European admixture in African Americans was on average 20%. A significantly greater percentage of European admixture was associated with the CC genotype, compared to the AC heterozygote and the AA homozygote in African Americans (40%, 22%, and 17% respectively, p<0.0001). European admixture is thus the likely origin of the C allele in African Americans.
In unadjusted analyses, genotype for the rs2768759 SNP in PEAR 1 was associated with maximal aggregation to collagen (in whites), epinephrine (in whites and African Americans), and ADP (in African Americans) prior to aspirin, and these associations were generally stronger and more significant after aspirin (Table 2A). There was no relation between genotype and arachidonic acid-induced aggregation at baseline. Removal from the analysis of subjects with zero aggregation to arachidonic acid at baseline (who presumably had known or inadvertent exposure to aspirin or non-steroidal anti-inflammatory agents) had no effect on the results except that the association between the genotype and aggregation to the lower dose of collagen and the lower dose of epinephrine were no longer significant in African Americans. In general, platelet aggregation prior to aspirin was highest in people with the CC genotype, least in those with the AA genotype, and intermediate in those with the AC genotype in both ethnic groups.
Post-aspirin, maximal platelet aggregation to collagen and epinephrine was reduced substantially for all genotypes in both ethnic groups. Maximal aggregation to ADP was also reduced, but by a lesser amount. However, residual aggregation to all three agonists was consistently higher with the CC genotype, intermediate with the AC genotype, and lowest with the AA genotype in both ethnicities (Table 2A). As expected, aggregation to arachidonic acid was markedly reduced after aspirin, with only 7.9% of subjects exhibiting measurable aggregation. However, the percent of subjects who did aggregate to arachidonic acid was significantly higher with the CC genotype in African Americans (aggregation occurred in 31.3% with CC, 9.1% with AC, 9.5% with AA, p=0.016). In whites, the trend was similar but the differences by genotype were not statistically significant (aggregation occurred in 8.2% with CC, 4.6% with AC, 6.9% with AA, p=0.13). In summary, these results are consistent with greater platelet reactivity both at baseline and after aspirin with the CC genotype.
To examine the effect of the PEAR1 genotype on aspirin responsiveness, we examined associations with a platelet response phenotype, expressed as post aspirin maximal aggregation adjusted for pre aspirin maximal aggregation, separately for each agonist and concentration. The associations of the rs2768759 genotype with aspirin responsiveness were significant for both collagen and epinephrine in both ethnicities, and borderline significant for ADP in whites. This analysis could not be done for arachidonic acid because of the dichotomous results for arachidonic acid-induced aggregation post aspirin. These results are consistent with an association between the genetic variant and reduced aspirin responsiveness, ie greater platelet aggregation post aspirin, even after adjustment for pre aspirin aggregation.
Consistent with the markedly higher C allele frequency we observed in whites, native platelets from white subjects demonstrated (in unadjusted analyses) greater aggregation to collagen, epinephrine, ADP, and arachidonic acid than platelets from African Americans (collagen 2μg/ml, 64.4 ± 29.2% vs 58.4 ± 33.2%, p<0.0001; collagen 5μg/ml, 82.2 ± 17.0% vs 78.9 ± 21.9%, p=0.062; epinephrine 2μM, 57.03 ± 33.5% vs 53.3 ± 36.1%, p=0.051; epinephrine 10μM, 72.9 ± 26.6% vs 66.3 ± 33.0%, p<0.0001; ADP, 80.1 ± 13.8% vs 78.2 ± 16.4%, p=0.028; arachidonic acid, 73.4 ± 24.7% vs. 67.0 ± 31.6%, p<0.0001). Post-aspirin, platelets from white subjects continued to be more aggregable to both collagen and epinephrine (but not ADP) than platelets from African Americans (collagen 2μg/ml, 13.7 ± 12.8% vs 12.9 ± 15.9%, p=0.55; collagen 5μg/ml, 28.7 ± 20.5% vs 25.8 ± 22.4%, p=0.051; epinephrine 2μM, 22.7 ± 12.3% vs 20.0 ± 14.3%, p=0.0002; epinephrine 10μM, 30.0 ± 14.6% vs 26.0 ± 16.3%, p<0.0001; ADP, 68.7 ± 12.9% vs 67.8 ± 13.6%, p=0.204).
In multivariable adjusted analyses, pre-aspirin phenotypes for collagen aggregation were not associated with the rs2768759 SNP in either ethnic group, but the post-aspirin phenotypes for collagen aggregation were significant and were similar by ethnicity (Table 2B). In contrast, epinephrine aggregation, both before and after aspirin, was significantly and independently associated with the variant in both ethnicities. For ADP aggregation, the variant was significantly associated with the pre-aspirin phenotype in African Americans but not in whites, while the opposite was true for the post-aspirin phenotype (significant in whites, not in African Americans). Finally, for arachidonic acid-induced aggregation only the baseline phenotype could be analyzed, and for this, the variant was independently associated with aggregation in African Americans, but not in whites. Thus, although there was more variability in baseline platelet aggregation, almost all post-aspirin platelet aggregation measures in both ethnicities were significantly associated with the rs2768759 SNP in PEAR1.
We also conducted variance-components analysis to estimate the genetic component of aggregation phenotypes to collagen, epinephrine, and ADP, pre and post aspirin (Table 3). The phenotypic variance was higher in African Americans than in whites, and higher for baseline aggregation. The pre-aspirin platelet aggregation phenotypes were generally less heritable than those post-aspirin. The portion of genetic variance explained by the rs2768759 locus in native platelet phenotypes among African Americans varied from 0.22% for the lower concentration of epinephrine, to 5.3% for the higher concentration of collagen. In whites, the proportion of the genetic variance explained by the locus in native platelet phenotypes also ranged widely from 0.43 to 3.7%. Following aspirin, however, a greater proportion of the genetic variance was explained by the locus, and this proportion was higher and more consistent in African Americans, ranging from 1.7 to 6.9%, while in whites the values ranged from 1.3 to 2.5%.
In their original description of PEAR1, Nanda et al17 demonstrated that platelet activation and initial contact induced by thrombin or collagen resulted in tyrosine phosphorylation of PEAR1, and that this could be inhibited by eptifibatide, a GPIIa/IIIb inhibitor. Platelet function was not assessed. Our results are the first to show that PEAR1 contributes significantly to maximal aggregation induced by collagen or epinephrine in native platelets, as well as to responsiveness of platelets to aspirin, in both whites and African Americans.
Our study demonstrates that the CC genotype of rs2768759 in the PEAR1 gene is associated with higher levels of residual platelet aggregation to collagen, epinephrine, and ADP following low dose aspirin, consistent with greater platelet reactivity. The CC genotype was also associated with greater native platelet aggregation, but the results were less pronounced and less consistent. Consistent with these results, the aspirin response phenotype (post-aspirin aggregation adjusted for pre-aspirin aggregation) was significantly associated with the CC genotype for most of the agonist/dose/ethnicity groups, consistent with reduced responsiveness to aspirin. The PEAR1 variant explained a surprisingly high proportion of the locus-specific genetic variance (up to 6.9% in African Americans and 2.5% in whites). By comparison, in the Framingham Heart Study, variants in the platelet glycoprotein IIIa and β-fibrinogen genes together accounted for <1% of the variance in epinephrine aggregation and collagen lag time, even though both of these phenotypes were highly heritable.8 The pronounced effect of the PEAR1 SNP on platelet aggregation in people taking aspirin, compared to aggregation of native platelets, suggests that PEAR1 may signal upstream of COX-1 or in pathways distinct from those that involve COX-1. Additional studies are required to determine if the PEAR1 activation pathway is a suitable target for anti-platelet therapy in individuals who have insufficient suppression of platelet function from aspirin alone. Alternatively, aspirin may acetylate a PEAR1 ligand, thereby modifying PEAR1-induced platelet signalling. A similar effect has been observed for the PEAR1 homolog SREC-I that binds acetylated low density lipoprotein.26
This is the largest study to investigate the relation between gene variants and platelet function, and the first to demonstrate that a variant in PEAR1 is associated with agonist-induced platelet aggregation and response to aspirin treatment. We recently reported that platelet aggregability before and after aspirin constitutes a set of heritable traits, but the specific gene variants that determine these traits remain uncertain.3,9 Previous studies involving relatively few individuals and inadequately controlling for genetic admixture, have reported inconsistent results regarding the relation between candidate gene SNPs and platelet function and/or response to aspirin. In our study, large samples of ethnically distinct whites and African Americans, analyzed separately, demonstrated a similar association between the PEAR1 gene variant and platelet function, which was particularly robust after aspirin exposure. This provides an internal replication of our findings, since the two ethnic groups were ascertained for the study in identical fashion, but were treated independently in the analysis. This replication was all the more remarkable given the 4-fold difference in C allele frequency between the two ethnicities. However, additional replication sets in other white and African American cohorts and other ethnic groups will be important to confirm our findings.
Although we identified a variant in the PEAR1 gene which is associated with agonist induced platelet aggregation and reduced aspirin responsiveness, we do not know whether the variant itself produces change in the expression or function of the translated protein, or whether it is merely in linkage disequilibrium with another SNP that may confer a change. Since the variant is located in the distal promoter region, approximately 10 Kb upstream from the transcription start site, and far from any exons (Figure 3), it is unlikely to be associated with a coding change in the protein. In addition, although we have considered rs2768759 to reside in the far upstream part of the PEAR1 promoter, it might also be seen as lying in the distal 3′-UTR of the next gene upstream, NTRK1 (neurotrophic tyrosine kinase receptor type 1). Although platelets do contain receptor tyrosine kinases, and we cannot exclude the possibility that rs2768759 is part of NTRK1, this receptor for neurotrophic growth factors has not been identified on platelets and has not been previously reported to modulate platelet function, as has PEAR1.
An increase in expression of PEAR1 associated with rs2768759 could potentially explain the increase in platelet aggregation that we observed by enhancing one or more of the amplification pathways involved in formation of platelet aggregates. However, we have no data on changes in expression of PEAR1 protein with the rs2768759 variant. In theory, there is a glucocorticoid receptor (GR) binding site in the region which includes rs2768759, which is present only when the sequence contains the C allele. The GR receptor is known to be expressed in megakaryocytic cell lines27 and recent evidence confirms expression of GR receptors in platelets.28 However, this potential relationship between the C allele and PEAR1 expression will require further investigation.
Our findings imply that white individuals, with a much higher frequency of the C allele, should have higher levels of agonist-induced native platelet aggregation or reduced responsiveness to low dose aspirin. We did in fact find greater aggregation of native platelets to all of the agonists tested in whites than in African Americans, as well as greater levels of aggregation following aspirin. Although PEAR1 explains a fairly high percentage of the genetic variance in platelet aggregation, most of the variance is still unexplained, making it difficult to conclude that the effect of ethnicity on these phenotypes is actually related to PEAR1. Further work will be required to better understand the biology underlying ethnic differences in platelet aggregation and their ultimate importance.
Supported by grants HL072518 from the National Heart, Lung and Blood Institute, and M01-RR00052 from the National Center for Research Resources, National Institutes of Health, Bethesda, MD, USA. Some analyses were performed using the program package S.A.G.E., which is supported by U.S. Public Health Service Resource Grant RR03655 from the National Center for Research Resources. McNeil Consumer and Specialty Pharmaceuticals supplied aspirin for the study.