Gene Variants of VAMP8 and HNRPUL1 Are Associated With Early-Onset Myocardial Infarction

【摘要】  Objectives- Identify gene variants associated with early-onset myocardial infarction (MI).

Methods and Results- We tested 11 647 single-nucleotide polymorphisms (SNPs) for association with early-onset MI in a case-control study (study 1 200 cases, 262 controls). To reduce the number of false positives among the 666 SNPs that were nominally associated with early-onset MI ( P <0.05) in study 1, we tested these SNPs in study 2 (434 cases, 504 controls). We found that 8 of the 666 SNPs were associated with early-onset MI in study 2 ( P <0.05) and had the same risk alleles as in study 1. These 8 SNPs were then tested for association with early-onset MI in study 3 (187 cases, 434 controls). We found that a VAMP8 variant ( P =0.025; odds ratio , 1.75; CI, 1.17 to 2.62) and an HNRPUL1 variant ( P =0.0043; OR, 1.92; CI, 1.28 to 2.86) were associated with early-onset MI (nominal P <0.05; false discovery rate <10%) and had the same risk alleles in all 3 studies.

Conclusions- Variants in 2 genes were associated with early-onset MI: VAMP8, which is involved in platelet degranulation, and HNRPUL1, which encodes a ribonuclear protein. The identification of these variants could improve understanding of disease mechanisms and suggest novel drug targets.

Genetic associations with early-onset myocardial infarction were identified by testing successively fewer SNPs in 3 consecutive case-control studies. Thus, 11 647 SNPs were tested in the first study, 666 in the second study, and only 8 SNPs were tested in the third study.

【关键词】  myocardial infarction genetics single nucleotide polymorphism risk factors

Introduction

Myocardial infarction (MI) is a complex disease with a strong genetic component. Gene variants are thought to affect traditional risk factors for MI such as hypertension, hypercholesterolemia, and diabetes. 1 In addition, family history of cardiovascular disease is a risk factor independent of traditional risk factors and is driven largely by genetic variation. 2,3 Genetic factors have been shown in twin studies to be particularly important for increasing the risk of early-onset MI. 3,4

Patients who experience early-onset MI may not be effectively identified by current cholesterol treatment guidelines such as those suggested by the National Cholesterol Education Program. 5 Because risk for MI can be reduced by lifestyle changes and by treatment of modifiable risk factors, better methods to identify patients at risk of early-onset MI could be useful for making treatment decisions.

Genetic markers for risk of early-onset MI could potentially be incorporated into individual risk assessment protocols. Genetic markers have the advantage of being easily detected at any age, yet there is currently no consensus on which genetic markers, if any, are suitable for identification of risk for early-onset MI. Thus, we asked whether genetic variants associated with early-onset MI could be identified by investigating 11 647 gene-centric, putatively functional single-nucleotide polymorphisms (SNPs) in case-control studies of early-onset MI.

Materials and Methods

Design and Strategy

To identify genetic polymorphisms associated with early-onset MI, we interrogated 3 case-control studies comprising cases with a history of early-onset MI and controls without a history of MI. We excluded young controls because they could experience an early-onset MI shortly after recruitment. The first 2 case-control studies (study 1 and study 2) identified SNPs nominally associated with early-onset MI. The hypotheses that these SNPs are associated with early-onset MI were tested in study 3. To increase the number of SNPs that could be tested in study 1 and study 2, we determined the allele frequency of each SNP in pools of case and control DNA 6 before determining the genotype of a smaller number of SNPs for all individual DNA samples.

SNP Selection

We determined the allele frequencies of 11 647 SNPs (in 7136 genes) in study 1. These SNPs are located in genes and have the potential to affect gene function or expression. The majority of these SNPs (69%) are missense or nonsense, or they modify acceptor and donor splice sites. Other SNPs (25%) are located in transcription factor binding sites or in untranslated regions of mRNA, which could affect mRNA expression or stability. Most 5% in whites.

Allele Frequency and Genotype Determination

DNA concentrations were standardized to 10 ng/µL using PicoGreen (Molecular Probes) fluorescent dye. 7 Allele frequencies in pooled DNA and genotypes of individual DNA samples were determined by kinetic polymerase chain reaction 8 (see the online supplement, available at http://atvb.ahajournals.org).

Subjects

Cases had a history of early-onset MI, and controls had no history of MI; all subjects gave informed consent and completed an institutional review board-approved questionnaire. Study 1 and study 2 are subsets of patient populations we described previously. 9 Participants in study 1 were patients of the Cleveland Clinic Foundation Heart Center who had undergone diagnostic or interventional cardiac catheterization between July 2001 and March 2003. The patients considered for inclusion in study 1 had a median age at enrollment of 63 for females and 58 for males. All cases in study 1 were below the median age for their sex at either the time of their enrollment or the time (if known, 103 cases) of their MI. Conversely, all controls were older than the median age for their sex.

Participants in study 2 and study 3 were enrolled between July 1989 and May 2005 by the University of California, San Francisco (UCSF) Genomic Resource in Arteriosclerosis. The cases in study 2 and study 3 had had their first MI before the age of 55 for males or 60 for females. Controls had no history of MI, diabetes, or symptomatic vascular disease. In study 2, male controls were 60 years of age; these controls had no known first-degree relative with a history of symptomatic coronary disease before age 65. In study 3, male 65 years of age. Participants in all 3 studies were white (see online supplement).

Statistical Analysis

Tagging SNPs were selected by pairwise tagging using Tagger 10 as implemented in Haploview. 11 We assessed association between continuous phenotypes and disease status by the Wilcoxon rank sum test, continuous phenotypes and genotype by the Kruskal-Wallis test, MI status and allele frequencies by 2-tailed 2 tests, diabetes status and MI risk allele by logistic regression (Wald test), MI status and genotype and by multiple logistic regression (Wald test), and MI status and haplotype by score statistics. 12 Potential interaction between each traditional risk factor and genotype was tested in separate logistic regression models that included the interaction (cross-product) term between the risk factor and the genotype; these models also included main effect terms for the genotype and the other risk factors. In study 3, because we tested a single, prespecified risk allele for each SNP, we present 1-sided P values and 90% CIs (for odds 1, there is 95% confidence that a true risk estimate is greater than the lower bound of a 90% CI). Similarly, when testing for association of diabetes with a prespecified risk allele, we present 1-sided P values. Because we considered 3 genetic models (dominant, recessive, and additive) for each SNP tested in study 3, we corrected these P values for multiple testing using a permutation test (see online supplement). False discovery rate (FDR) was calculated using the MULTTEST procedure 13 (SAS statistical package Version 9.1) using the permutation-based minimum P values.

Results

Study Populations

For all 3 studies, the prevalence of traditional risk factors was higher in the case groups than in the control groups (supplemental Table I). There were significantly more males in the case groups than in the control groups of study 2 and study 3. The mean age at enrollment was higher in the control groups of all 3 studies because of study design. In all 3 studies, the median age of MI was 53 years for females (53, 52, and 47 in studies 1, 2, and 3, respectively) and 50 years for males (50, 44, and 47 for studies 1, 2, and 3, respectively), reflecting the recruitment of cases who experienced MI at an early age.

Hypotheses Generation in Study 1 and Study 2

In study 1, the first hypotheses-generating study, we asked which of 11 647 SNPs were associated with early-onset MI by comparing the allele frequencies between cases and controls. We found 666 SNPs associated with early-onset MI in study 1 (unadjusted P <0.05; supplemental Table II). In study 2, the second hypotheses-generating study, we tested these 666 SNPs for association with early-onset MI in pooled DNA samples and selected the 16 SNPs that had the lowest P values ( P <0.022) for further study. We then determined the genotypes of these 16 SNPs in all the individual DNA samples of study 1 and study 2 to confirm the results obtained using pooled DNA. We found that 8 of these SNPs (in 7 genes) were associated with early-onset MI ( P <0.05) and had the same risk allele in both studies ( Table 1 ). Because the other 8 SNPs had a nominal P 0.05 in 1 of the 2 studies, they were not included in further analysis.

TABLE 1. Association of 8 SNPs With Early-Onset MI in Study 1 and Study 2

Hypotheses Testing in Study 3

In study 3, we tested the hypotheses formulated in study 1 and study 2, namely that a specific allele of each of the 8 SNPs is associated with early-onset MI. The genotype distributions for these 8 SNPs did not deviate from Hardy-Weinberg equilibrium ( P 0.13). We calculated the risk associated with each genotype of the 8 SNPs using the prespecified risk allele (supplemental Table III). Because the genotypic risk estimates suggested that specific genetic models might be appropriate for some of these SNPs, we tested each SNP in 3 genetic models (recessive, dominant, and additive) while adjusting traditional risk factors. After adjusting the resulting P values for testing multiple models, we calculated the FDR for each independent SNP ( Table 2 ). We considered only 7 of the 8 SNPs tested in study 3 to be independent because the 2 SNPs in AQP10 are only 322 bp apart and are in strong ( r 2 =0.995) linkage disequilibrium (LD). We found that 2 of these 7 SNPs were associated with early-onset MI in study 3 ( P <0.05; FDR <10%): 1 in HNRPUL1, which encodes a heterogeneous ribonuclear protein that participates in nucleocytoplasmic RNA transport, 14 and 1 in VAMP8, which is a vesicle docking protein that plays a role in platelet degranulation. 15

TABLE 2. Model-Specific Associations of 8 SNPs With Early-Onset MI in Study 3

Because we selected controls who did not have diabetes and who were older than cases, we were unable to adjust the risk estimates of these 2 SNPs for the effect of age or diabetes. However, we observed no significant association of the risk alleles of the VAMP8 or HNRPUL1 SNP with either age or diabetes. For the VAMP8 SNP, the median age of controls in study 2 was 75.6 years for carriers and 75.7 years for noncarriers and 70.2 and 70.8 years in study 3 ( P 0.2 for both studies). For HNRPUL1, the median age of controls in study 2 was 75.7 for carriers and 75.6 for noncarriers and 70.3 and 70.4 years in study 3 ( P 0.5 for both studies). When we tested the association of the prespecified risk alleles of these SNPs with diabetes in the control groups of study 2 and study 3, the OR was 0.66 in study 2 and 0.62 in study 3 for the dominant model of the VAMP8 SNP ( P 0.9 for both studies) and 1.22 in study 2 and 1.23 in study 3 for the recessive model of the HNRPUL1 SNP ( P 0.25 for both studies).

We did not observe significant interactions between the risk genotypes and traditional risk factors (sex, dyslipidemia, hypertension, smoking, and body mass index) in the fully adjusted model of study 3 (risk factor by genotype interaction all P 0.05); however, because some risk factors may be sex specific, we also determined the risk estimates of the VAMP8 and HNRPUL1 SNPs in male and female subgroups. To increase the power to detect differences between the subgroups, we conducted a combined analysis of the study 2 and study 3 subjects (all of the UCSF subjects) and found that the risk estimate of the VAMP8 variant was higher in males (OR, 1.97) than in females (OR, 1.09; P interaction=0.048). The risk estimates of the HNRPUL1 SNP did not differ between the male and female subgroups.

LD and Risk in the VAMP8 and HNRPUL1 Regions

We then investigated whether these 2 SNPs (in VAMP8 and HNRPUL1) were simply markers for other SNPs that were more strongly associated with early-onset MI. In study 2, the largest study, we investigated 11 SNPs in the &80-kb VAMP8 LD region ( Figure, A), including 4 potentially functional SNPs (rs1010, rs1058588, rs699664, and rs14242; supplemental Table IV). The VAMP8 region 2% (HapMap public release No. 19 16 ), and 59 of these were tagged (mean r 2 =0.95) by the SNPs we investigated (9 of the 11 SNPs were tagging SNPs). We found that none of these 11 SNPs were more strongly associated with MI than the initially identified SNP (rs1010). However, 1 of these SNPs, rs1058588, was in perfect LD with rs1010 ( r 2 =1.0 in study 2), and thus, this pair of SNPs can be considered a single VAMP8 variant: all chromosomes carry both or neither risk allele for these 2 SNPs. Three of the 9 tagging SNPs trended ( P <0.1) toward association with MI; however, the risk estimates for these SNPs were markedly reduced after adjustment for the risk associated with rs1010 (supplemental Table IV). Conversely, the risk estimates of rs1010 were minimally affected when adjusted for any other SNP we tested in this region.

The black triangles represent the extent of the interrogated LD region in the VAMP8 (A) and HNRPUL1 (B) regions. LD was determined by the International HapMap project for 90 Utah residents with ancestry from Northern and Western Europe (CEPH) release No. 19 16 and displayed in Haploview. 11 Each square in the triangle represents the LD between 2 SNPs. Bright red represents D'=1, shades of pink represent levels of D'<1. Where there is no evidence for LD (LOD <2), squares are blue for D'=1 and white for D'<1. Red arrows indicate the initially identified SNPs. Black arrows indicate tagging SNPs. Black boxes indicate genes.

For the &110-kb HNRPUL1 LD region ( Figure, B), we investigated 6 tagging SNPs and 4 potentially functional SNPs (Chr19:46508042, rs6957, rs1800472, and rs1982073) in study 2 (supplemental Table IV). The HNRPUL1 region contains 44 SNPs that have allele frequencies 2% (HapMap public release No. 19 16 ), and 38 of these were tagged (mean r 2 =0.93). Two of these tagging SNPs were associated with MI: the initially observed rs11881940 and rs12981053, which had an OR of 1.35 ( P =0.027). When the risk estimates for these 2 SNPs were adjusted for each other, both risk estimates were dramatically reduced (supplemental Table IV). Thus, these 2 SNPs were equivalently associated with risk of MI in study 2, a result that is consistent with their similar allele frequencies (86% and 88%) and strong linkage ( r 2 =0.72 in study 2). The OR for the 2-SNP haplotype containing the risk alleles of both these SNPs was 1.30 ( P =0.047), which does not differ appreciably from the ORs of the individual SNPs (supplemental Table V).

Discussion

We identified 2 genetic variants associated with early-onset MI by conducting 3 association studies that included cases with 58 years of age. These are variants of VAMP8, a gene involved in platelet degranulation, 15 and HNRPUL1, a gene encoding a heterogeneous ribonuclear protein believed to be involved in mRNA processing and transport. 14

VAMP8 is on chromosome 2 and encodes a vesicle-docking protein required for platelet-dense granule secretion, 15 which leads to platelet activation, a process that contributes to MI. Coronary heart disease linkage studies have identified several peaks on chromosome 2, 17-19 1 of which is a broad 68-mbp peak that contains VAMP8. 17 Our analysis of tagging and putative functional SNPs in the VAMP8 region indicated that the VAMP8 variant most significantly associated with early-onset MI comprised rs1010 and rs1058588, 2 SNPs that were completely concordant in study 2. These SNPs are 111 bp apart in a portion of the 3'UTR of the VAMP8 transcript that also contains putative binding sites for 2 microRNAs 20 (miR96 and miR15). The VAMP8 risk variant may reduce the stability of a predicted stem-loop structure (based on predictions by the MFOLD program 21 ). Thus, one could speculate that destabilization of this stem loop might enhance miR96 binding, which could affect mRNA translation. However, functional studies of VAMP8 protein expression or activity are needed to further characterize this variant.

This VAMP8 risk variant may have different effects in different populations. In a combined analysis of study 2 and study 3, the risk estimate for early-onset MI was higher in males than in females; however, future studies would be needed to confirm this observation. Association between this VAMP8 variant and MI should also be investigated in nonwhite populations (see online supplement for frequency information in nonwhite populations).

We also found that an SNP in HNRPULL1, which encodes 1 of a family of heterogeneous nuclear ribonucleoproteins, was associated with early-onset MI in study 3. HNRPUL1 plays a role in RNA transport, processing, and transcriptional regulation. 14,22 Because the major allele of the HNRPUL1 SNP is associated with risk of early-onset MI, one could consider either the minor allele to be protective or the major allele to be contributing to risk in many members of the population. However, investigation of other SNPs in this region revealed that the HNRPUL1 SNP (rs11881940) and a tagging SNP (rs12981053) had similar association with early-onset MI. This tagging SNP is located between the MGC20255 and TGFB1 genes. Determining which of these SNPs is most strongly associated with early-onset MI would require a much larger association study because only &1% of the study 2 population carries a risk allele for 1 SNP but not the other. Alternatively, functional studies could indicate whether either of these 2 SNPs is causally associated with early-onset MI. For example, rs11881940 is in an intron of HNRPUL1 and alters a conserved adenine in a putative interferon regulatory factor-1 binding site. In a shorter alternative transcript of HNRPUL1 (AK021455 ), this SNP causes a coding change (Tyr41Phe). The associated tagging SNP in rs12981053 is intergenic, thus it seems less likely to be functional.

In association studies that test thousands of SNPs, multiple-hypothesis testing can lead to false-positive results. Therefore, we used 3 sequential studies to reduce the number of hypotheses tested in the last study and then estimated the effect of multiple testing on the results. We used the first 2 studies to reduce the number of hypotheses from the 11 647 SNPs tested in study 1 to the 7 independent hypotheses tested in study 3. Because testing even 7 hypotheses could generate false positives, we set 10% as the FDR criterion for accepting nominally significant associations. That is, we expected that <1 of 10 SNPs considered significant would be a false positive.

In study 3, 2 gene variants were associated with early-onset MI, and in a previous study, we reported 4 other gene variants that were associated with MI. 9 We included subjects from a previous report in the studies described in this report if the age of the subject was consistent with the constraints of our early-onset MI study design. The relationships between the subjects and SNPs studied in these 2 reports are as follows. Study 1 subjects in this report were subjects in study 2 of the previous report. Study 2 cases in this report were cases in the previous report (110 in study 1 and 324 in study 3). Most controls in study 2 of this report were controls in study 1 (28 controls) or study 3 (411 controls) of the previous report. In this report, we tested 11 647 SNPs (in 7136 genes) for association with early-onset MI. In the previous report, we tested 10 377 of these SNPs (in 6606 genes) for association with MI. The VAMP8 and HNRPUL1 SNPs described in this report were not tested in the previous report. Of the 4 SNPs found to be associated with MI in the previous report, the SNPs in ROS1 and palladin were among those tested in study 1 of this report, and they trended toward association with early-onset MI (0.05< P <0.1). The other 2 SNPs described in the previous report (in TAS2R50 and OR13G1) were not among the SNPs initially tested in study 1 of this report. However, we later genotyped these SNPs in study 1 and found that they were associated with early-onset MI ( P <0.05; data not shown), but this is not an independent confirmation of the association of the TAS2R50 and OR13G1 SNPs with MI because many subjects were in both studies. We also genotyped a THBS4 SNP in study 1 and study 2 because it had been reported to be associated with early-onset MI, 23 and we found that it was associated with early-onset MI in study 2 ( P <0.05) but not in study 1 ( P =0.77).

The case-control studies conducted in this investigation limit our conclusions. Because these retrospective studies did not include fatal cases of early-onset MI, genetic markers specifically associated with this severe phenotype would not have been identified. The 3 studies did not all use the same definitions of ethnicity or case and control status, and because control status was based on self-reported history, some controls could have been misclassified. Additionally, this study was designed to identify associations with early-onset MI, thus it may not have had sufficient power to detect potential interaction with traditional risk factors.

The variants we described in VAMP8 and the HNRPUL1 region are good candidates for genetic analysis in population-based studies of cardiovascular disease. If the association of the VAMP8 and HNRPUL1 region variants with early-onset MI is further confirmed, these variants could advance our understanding of disease mechanism, suggest targets for therapeutic intervention, and be useful in assessing genetic risk of early-onset MI.

Acknowledgments

The authors acknowledge the contributions of the Celera High Throughput Laboratory and Computational Biology group and thank John Sninsky and Thomas White for helpful comments on this manuscript.

Sources of Funding

This study was funded by the UC Discovery Grant Program, the Mildred V. Strouss Charitable Trust, the Joseph Drown Foundation, and a gift from Donald Yellon (J.P.K., M.J.M.) and the grant HL077107 (E.J.T.).

Disclosures

D.S., C.M.R., J.Z.L., M.M.L., L.A.B., J.I.B., B.A.Y., J.J.C., and J.J.D. have employment and ownership interests. S.G.E. has consulting fees. The remaining authors have no disclosures.

【参考文献】
  Lusis AJ, Mar R, Pajukanta P. Genetics of atherosclerosis. Annu Rev Genomics Hum Genet. 2004; 5: 189-218.

Shea S, Ottman R, Gabrieli C, Stein Z, Nichols A. Family history as an independent risk factor for coronary artery disease. J Am Coll Cardiol. 1984; 4: 793-801.

Marenberg ME, Risch N, Berkman LF, Floderus B, de Faire U. Genetic susceptibility to death from coronary heart disease in a study of twins. N Engl J Med. 1994; 330: 1041-1046.

Zdravkovic S, Wienke A, Pedersen NL, Marenberg ME, Yashin AI, De Faire U. Heritability of death from coronary heart disease: a 36-year follow-up of 20 966 Swedish twins. J Intern Med. 2002; 252: 247-254.

Akosah KO, Schaper A, Cogbill C, Schoenfeld P. Preventing myocardial infarction in the young adult in the first place: how do the national cholesterol education panel iii guidelines perform? J Am Coll Cardiol. 2003; 41: 1475-1479.

Sham P, Bader JS, Craig I, O?Donovan M, Owen M. DNA pooling: a tool for large-scale association studies. Nat Rev Genet. 2002; 3: 862-871.

Singer VL, Jones LJ, Yue ST, Haugland RP. Characterization of PicoGreen reagent and development of a fluorescence-based solution assay for double-stranded DNA quantitation. Anal Biochem. 1997; 249: 228-238.

Germer S, Holland MJ, Higuchi R. High-throughput SNP allele-frequency determination in pooled DNA samples by kinetic PCR. Genome Res. 2000; 10: 258-266.

Shiffman D, Ellis SG, Rowland CM, Malloy MJ, Luke MM, Iakoubova OA, Pullinger CR, Cassano J, Aouizerat BE, Fenwick RG, Reitz RE, Catanese JJ, Leong DU, Zellner C, Sninsky JJ, Topol EJ, Devlin JJ, Kane JP. Identification of four gene variants associated with myocardial infarction. Am J Hum Genet. 2005; 77: 596-605.

de Bakker PI, Yelensky R, Pe?er I, Gabriel SB, Daly MJ, Altshuler D. Efficiency and power in genetic association studies. Nat Genet. 2005; 37: 1217-1223.

Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005; 21: 263-265.

Schaid D, Rowland C, Tines D, Jacobson R, Poland G. Score tests for association between traits and haplotypes when linkage phase is ambiguous. Am J Hum Genet. 2002; 70: 425-434.

Benjamini Y, Hochberg YJR. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B. 1995; 57: 289-300.

Gabler S, Schutt H, Groitl P, Wolf H, Shenk T, Dobner T. E1B 55-kilodalton-associated protein: a cellular protein with RNA-binding activity implicated in nucleocytoplasmic transport of adenovirus and cellular mRNAs. J Virol. 1998; 72: 7960-7971.

Polgar J, Chung SH, Reed GL. Vesicle-associated membrane protein 3 (VAMP-3) and VAMP-8 are present in human platelets and are required for granule secretion. Blood. 2002; 100: 1081-1083.

The International HapMap Consortium. The International HapMap Project. Nature. 2003; 426: 789-796.

The BHF Family Heart Study Research Group. A genomewide linkage study of 1933 families affected by premature coronary artery disease: the British Heart Foundation (BHF) Family Heart Study. Am J Hum Genet. 2005; 77: 1011-1020.

Wang Q, Rao S, Shen G-Q, Li L, Moliterno DJ, Newby LK, Rogers WJ, Cannata R, Zirzow E, Elston RC, Topol EJ. Premature myocardial infarction novel susceptibility locus on chromosome 1P34-36 identified by genomewide linkage analysis. Am J Hum Genet. 2004; 74: 262-271.

Pajukanta P, Cargill M, Viitanen L, Nuotio I, Kareinen A, Perola M, Terwilliger JD, Kempas E, Daly M, Lilja H, Rioux JD, Brettin T, Viikari JS, Ronnemaa T, Laakso M, Lander ES, Peltonen L. Two loci on chromosomes 2 and X for premature coronary heart disease identified in early- and late-settlement populations of Finland. Am J Hum Genet. 2000; 67: 1481-1493.

Lewis BP, Burge CB, Bartel DP. Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell. 2005; 120: 15-20.

Zuker M. On finding all suboptimal foldings of an RNA molecule. Science. 1989; 244: 48-52.

Kzhyshkowska J, Rusch A, Wolf H, Dobner T. Regulation of transcription by the heterogeneous nuclear ribonucleoprotein E1B-AP5 is mediated by complex formation with the novel bromodomain-containing protein BRD7. Biochem J. 2003; 371: 385-393.

Topol EJ, McCarthy J, Gabriel S, Moliterno DJ, Rogers WJ, Newby LK, Freedman M, Metivier J, Cannata R, O?Donnell CJ, Kottke-Marchant K, Murugesan G, Plow EF, Stenina O, Daley GQ. Single nucleotide polymorphisms in multiple novel thrombospondin genes may be associated with familial premature myocardial infarction. Circulation. 2001; 104: 2641-2644.

作者单位：From Celera Inc. (D.S., C.M.R., J.Z.L., M.M.L., L.A.B., J.I.B., B.A.Y., J.J.C., J.J.D.), Alameda, Calif; Brigham Young University (C.F.S.), University Station, Provo, Utah; Cardiovascular Research Institute (C.R.P., M.J.M., J.P.K.), University of California San Francisco; Case Western Reserve Univer