Functional Analysis of the Chromosome 9p21.3 Coronary Artery Disease Risk Locus
Objectives— We have investigated the functional significance of conserved sequences within the 9p21.3 risk locus for coronary artery disease (CAD) and determined the relationship of 9p21.3 to expression of ANRIL and to whole genome gene expression.
Methods and Results— We demonstrate that a conserved sequence within the 9p21.3 locus has enhancer activity and that the risk variant significantly increases reporter gene expression in primary aortic smooth muscle cells. Whole blood RNA expression of the short variants of ANRIL was increased by 2.2-fold whereas expression of the long ANRIL variant was decreased by 1.2-fold in healthy subjects homozygous for the risk allele. Expression levels of the long and short ANRIL variants were positively correlated with that of the cyclin-dependent kinase inhibitor, CDKN2B (p15) and TDGF1 (Cripto), respectively. Relevant to atherosclerosis, genome-wide expression profiling demonstrated upregulation of gene sets modulating cellular proliferation in carriers of the risk allele.
Conclusion— These findings are consistent with the hypothesis that the 9p21.3 risk allele contains a functional enhancer, the activity of which is altered in carriers of the risk allele. 9p21.3 may promote atherosclerosis by regulating expression of ANRIL, which in turn is associated with altered expression of genes controlling cellular proliferation pathways.
The 9p21.3 risk locus, identified in several genome-wide association studies for coronary artery disease (CAD),1,2 spans 58 kb and encompasses multiple single nucleotide polymorphisms (SNPs) in tight linkage disequilibrium (LD). Approximately 25% of whites carry 2 copies of the risk allele and have a 1.5-fold increased risk for CAD.2 The increased risk is independent of all known risk factors including plasma lipids, hypertension, diabetes, obesity, and markers of inflammation. This implies a novel biological pathway relevant to atherosclerosis.
BLAST searches against the NCBI nucleotide sequence database and inspection of the UCSC Genome Browser revealed no annotated genes within the 58-kb interval. The localization of the risk locus to a region devoid of known protein coding genes implicates a novel gene or regulatory element that promotes atherosclerosis independently of established risk factors. It is notable that the 9p21 risk locus overlaps a newly annotated noncoding RNA (ncRNA), termed ANRIL, recently identified through a deletion analysis of a large French kindred with hereditary melanoma.3 ANRIL spans 126.3 kb, overlaps at its 5′ end with CDKN2B (p15), and consists of 20 exons subjected to alternative splicing. Full-length ANRIL (3834 bp), deposited under GenBank accession number DQ485353, encompasses the first 12 exons plus exons 14 to 20. At least 2 shorter variants of ANRIL that end with an alternative exon 13 have been reported, DQ485454 and EU741058. DQ485454 (2659 bp) comprises the first 12 exons and an alternative exon 13, whereas EU741058 (688 bp) encompasses exons 1, 5 to 7, and 13 (Figure 1). Noncoding RNAs such as ANRIL can alter expression of associated protein coding genes through multiple mechanisms that include RNA interference, gene silencing, chromatin remodeling, or DNA methylation.4 There is growing evidence of involvement of ncRNAs transcribed from the 9p21 locus in disease etiology. A natural p15 antisense, p15AS, triggers transcriptional silencing of p15 and shows increased levels of expression in acute myeloid leukemia cell lines.5 Broadbent and colleagues have demonstrated that ANRIL is expressed in many cell types known to be affected by atherosclerosis and suggested that this gene is a possible candidate gene at the 9p21 CAD risk locus.6 Despite the potential importance of ANRIL, limited information is available on its genetic architecture.
Here, we have investigated the functional significance of conserved sequences within the 9p21.3 risk locus and have determined the effects of genetic variation at the 9p21.3 locus on expression of ANRIL and other gene transcripts in whole blood RNA. Because peripheral blood mononuclear cell (PBMC) gene expression is altered in a number of disease states and in response to various medications, we first studied healthy control subjects and then confirmed our findings in patients with stable CAD.
As part of the Ottawa Heart Study, healthy control subjects with no history or symptoms of CAD, consisting of men aged >65 and women aged >70 years, were recruited as described.1 DNA samples were genotyped using the Affymetrix 500K or 6.0 array. For whole-blood RNA analysis of healthy control subjects, 124 nonsmoking subjects, in good health with no history, signs, or symptoms of cardiovascular and not on statin therapy were selected on the basis of homozygosity for the reference or risk alleles (Table). Individual transcripts were quantified by QRT-PCR for all subjects, and whole genome expression analysis was carried out in subset of 44 subjects. For expression analysis of CAD patients, a second group of 42 well-matched subjects homozygous for the reference allele and the risk alleles with stable angiographically defined CAD and no history of diabetes were selected (supplemental Table I, available online at http://atvb.ahajournals.org).
The study was approved by the Human Research Ethics Committee of the Ottawa Heart Institute and written informed consent was obtained from all participants.
Four putative regulatory elements identified by PIPMAKER analysis of the 9p21.3 region, denoted conserved noncoding sequences (CNSs): CNS1, CNS2, CNS3, and CNS4 (supplemental Figure II) were functionally characterized. To produce Luc reporter constructs, individual CNSs (Figure 2A) were amplified from both the 9p21 risk and reference haplotypes with proof-reading PfuUltra II Polymerase and then inserted upstream of β-globin minimal promoter into PGL4.23 expression vector (Promega, E8411). Integrity of the reporter constructs was confirmed by restriction digest and sequencing. Details of primers and amplification reactions are provided in supplemental Methods.
Transfection and Luciferase Assays
COS-7 cells and human aortic smooth muscle cells (AoSMC; Lonza Walkersville, Md., USA) were cultured in Dulbecco modified Eagle and SMC Basal Medium, respectively. COS-7 and AoSMC cells were plated in 6- or 24-well dishes 24 hours before transfection. Reporter plasmids bearing individual CNSs were cotransfected with a plasmid constitutively expressing Renilla luciferase (PGL4.74, Promega) using Lipofectamine (Invitrogen) following the manufacturers’ protocols. Luciferase assays were performed using the Dual-Luciferase Reporter Assay System (Promega) following a standard protocol.
Characterization of Functional Variation Within the CNS3 Regulatory Region
To identify functional sequence variations responsible for the observed differences in the regulatory activity of CNS3 from the reference and risk haplotypes, we screened the sequences for the presence of single nucleotide polymorphisms. We identified only one single bp difference corresponding to database entry rs1333045, which we then genotyped in 513 healthy controls and 483 patients with CAD. Details of primers and amplification reactions are provided in supplemental Methods.
Quantitative Real-Time RT-PCR
Whole-blood RNA was extracted using the PAXgene Blood RNA kit (PreAnalytiX). To exclude amplification from genomic DNA contamination, an additional off-column DNAseI digest was carried out for each RNA sample. Determination of relative amounts of short and long ANRIL transcripts, CDKN2B, TDGF1, KALRN, and GAPDH was carried out using quantitative real-time RT-PCR (QRT-PCR) of ≈150 bp regions of the corresponding transcript. Relative copy numbers of each transcript were normalized for expression levels of the constitutively expressed β-actin and cyclophilin (PPIA) genes. Normalization for both housekeeping genes yielded similar results, and data obtained from normalization for PPIA are shown. Details of primers and amplification reactions are provided in supplemental Methods.
Human Whole Genome Expression Analysis
Whole genome expression analysis of whole blood RNA was carried out using Agilent 4×44K Human Whole Genome microarrays as described in detail in supplemental Methods.
Statistical Analysis of Microarray Data
Testing for differential gene expression was performed using a 2-class, unpaired, unequal variance form of the t test. Results were filtered to identify probes exhibiting a probability value less than 0.05 and an absolute fold change greater than 1.5 and the filtered gene list was visualized using hierarchical cluster analysis. Gene Set Analysis (GSA) was used to access genotype-phenotype correlations of all genes within a gene set. The Molecular Signatures Database (MSigDB) provided the organization of genes into categories.7 GSA was run using the 2-class unpaired test, and gene sets were considered significant when the probability value was less than 0.005 (supplemental Methods). A positive GSA score reflects upregulation while negative score reflects downregulation of a gene set.
Identification of Putative Regulatory Elements Within 9p21.3 Locus
Phylogenetic footprinting analysis of the sequences including ANRIL transcription units from human, rhesus monkey, chimpanzee, mouse, rat, dog, and opossum (supplemental Methods) indicates that whereas exon sequences of ANRIL are highly conserved in primates, the sequence conservation is not apparent in orthologous sequences from dog, mouse, rat, and opossum (supplemental Figure I). Furthermore, several exons of ANRIL retain repetitive elements. Thus, exons 7 and 12 of ANRIL consist entirely of Alu repeats, exons 8 and 16 of Long Terminal Repeats (LTRs), exon 14 of the LINE2 element, and exons 4, 17, and 20 contain multiple MIR SINEs (data not shown). To determine whether the 9p.21 risk interval encompasses cis-acting regulatory elements, we then aligned the risk region sequences from human, chimpanzee, macaque, dog, and opossum. Using this approach, we identified 4 conserved noncoding sequences (CNS) within the 58-kb risk interval, designated CNS1, CNS2, CNS3, and CNS4, which may contain such regulatory elements (supplemental Figure II, supplemental Methods).
Luciferase Reporter Analysis of Putative Regulatory Elements Within the 9p21.3 Locus
We analyzed the activity of conserved noncoding sequences within the 9p21.3 locus by inserting 4 conserved sequences from the risk haplotype into PGL4.23 expression vector. Enhancer activity of each of the sequences was tested in dual luciferase assays where Renilla luciferase activity was used to normalize transfection efficiency. We first examined the ability of identified conserved regions from the risk haplotype to induce luciferase activity in COS-7 cells. The results demonstrated that one of the conserved noncoding sequences, designated CNS3, acted as an enhancer and resulted in a 22-fold increase of reporter gene expression when compared to transfection with PGL4.23 vector alone (Figure 2B). We then tested enhancer activity of the CNS3 sequences from individuals homozygous for the risk and reference 9p21 alleles in primary human AoSMCs. These experiments revealed that both the risk and the reference counterparts of CNS3 exhibited significant enhancer activity as compared to PGL4.23 vector (Figure 2C). Moreover, the CNS3 sequence cloned from a homozygote carrier of the 9p21.3 risk allele exhibited 73% higher enhancer activity compared to the same sequence cloned from a reference homozygote (Figure 2C).
Characterization of Sequence Variation in the CNS3 Regulatory Elements
By sequencing the CNS3 regions from the reference and the risk haplotypes, we demonstrated that the observed difference in the CNS3 regulatory activity is attributable to the presence of a single nucleotide polymorphism, rs1333045, which is in tight LD with the lead predictive SNP in the 9p21.3 locus, rs1333049 (r2=0.86). Moreover, rs1333045 has a virtually identical allele frequency to that of rs1333049 in premature angiographic CAD cases and healthy elderly controls (MAFcases=0.605 and 0.598; MAFcontrols=0.445 and 0.441, respectively) and a similar allele specific odds ratio for CAD (OR=1.28). These data are consistent with the hypothesis that rs1333045 represents a functional variant within the 9p21.3 risk locus and provides significant predictive information with regards to CAD risk.
ANRIL Transcripts Are Expressed in Multiple Tissue Types Relevant to Atherosclerosis
We determined the relative abundance of the 3 identified ANRIL transcripts, DQ485453, DQ485453, and EU741058, in several human tissue types relevant to atherosclerosis: whole blood, CD14+ and CD14− PBMCs, AoSMCs, AoECs, and human heart. Our data demonstrate that each of the ANRIL transcripts (DQ485453, DQ485454, and EU741058) is expressed in a subset of cells or tissues linked to atherosclerosis (supplemental Figure III).
9p21 Risk Allele Is Associated With Altered Expression of Splicing Variants of ANRIL
We next questioned whether expression levels of the 3 variants of ANRIL in whole blood differ in control subjects homozygous for the 9p21.3 risk alleles versus those homozygous for the reference alleles (supplemental Figure III). The QRT-PCR data demonstrate expression levels of the 2 short variants of ANRIL were increased ≈2.2-fold in 60 healthy subjects homozygous for the risk allele compared to 60 well-matched individuals homozygous for the reference allele (Figure 3A). Consistently, expression of the shortest ANRIL transcript, EU741058, was significantly greater in individuals homozygous for the risk allele (Figure 3B). In contrast, expression levels of the long variant were decreased ≈1.2-fold in subjects homozygous for the risk allele versus reference allele (Figure 3C). We also found a significant correlation between expression levels of the short and the long variants of ANRIL (r=0.47; P=0.0152; data not shown), suggesting the existence of a regulatory feedback loop between the long and short ANRIL variants.
Differential Expression of ANRIL in Subjects Homozygous for the Risk and Reference Allele Associate With Differences in Expression of Several Genes Implicated in Atherosclerosis
Because ANRIL overlaps the INK4/ARF locus, we determined whether differential expression profiles of the ANRIL variants in subjects homozygous for the risk and reference allele was associated with expression differences of CDKN2A and CDKN2B. CDKN2A and CDKN2B control cellular proliferation by preventing entry to the S-phase of the cell cycle. We also we assessed effects on expression of GAPDH because apoptotic stimulation causes S-nitrosylation of GAPDH which in turn binds to E3 ubiquitin ligase and triggers apoptosis.8
Subjects homozygous for the risk allele demonstrated a decrease in relative copy numbers of CDKN2A, CDKN2B transcripts (P=0.06 and P=0.156 respectively) and significant decreases in the expression of GAPDH, KALRN, a gene previously linked to CAD9 and TDGF1, which functions in cellular proliferation10 (supplemental Figure IVA through IVC). Furthermore, relative expression levels of CDKN2B and TDGF1 correlated significantly with that of the long variant and shortest variants of ANRIL respectively (Figure 3D; supplemental Figure IVD). These data suggest that CDKN2B and TDGF1 may be directly affected by alterations in ANRIL expression related to the 9p21.3 risk haplotype.
Upregulation of Gene Sets Involved in Cellular Proliferation in Homozygous Carriers of 9p21.3 Risk Alleles
To determine whether the allelic variation at the 9p21.3 locus can alter gene networks relevant to atherosclerosis, we performed whole genome expression analysis on a subset of healthy subjects used for QRT-PCR analysis. A total of 97 differentially expressed genes were identified at a relative expression level of >1.5 and a probability value cutoff <0.05 (supplemental Table II). A heat map was generated based on these 97 genes for visualization purposes (Figure 4A).
To identify biological pathways associated with the 9p21.3 allele in healthy individuals, gene set analysis was performed using the Molecular Signature Database (MSigDB) comparing homozygous risk allele carriers with homozygous reference allele carriers. Several annotated gene sets and pathways associated with increased cell proliferation, including CELL_CYCLE_KEGG and vascular endothelial growth factor signaling (VEGF-HUVEC), were found to be upregulated in homozygous 9p21.3 risk allele carriers (Figure 4B). Gene sets relevant to inflammation, including TNF/ MAP kinase pathway signaling in activated T cells (41BBPATHWAY) and interferon response genes (CMV_HCMV_TIMECOURSE_24HRS), were also differentially expressed (Figure 4B; supplemental Methods).
To confirm that the 9p21 risk allele is also associated with significant gene expression changes in patients with established but stable CAD, cDNA samples obtained from 42 matched patients homozygous for the reference allele and the risk alleles were similarly analyzed on Agilent human genome microarrays (supplemental Methods; supplemental Figure V, supplemental Table III). QRT-PCR analysis was performed on a subset of 29 matched patients. Consistent with allele-specific expression differences detected in healthy individuals, these experiments demonstrated upregulation of gene sets relevant to cellular proliferation (P<0.005) (supplemental Table IV), increased expression of short variants of ANRIL (P=0.041), and a reciprocal and decrease in expression of CDKN2B, GAPDH, and TDGF1 genes (supplemental Figure VI) in patients homozygous for the risk allele. Analysis of combined data from CAD cases and controls revealed a significant decrease in CDKN2B expression (P=0.038) in homozygous carriers of the risk allele (data not shown).
Taken together, these data support the hypothesis that the 9p21.3 risk region contains cis-regulatory elements responsible for allele-specific gene expression differences.
The association between the 9p21.3 locus and CAD risk is robust and extensively replicated but the molecular basis for a causal relationship has remained obscure. Our finding that the frequency of the risk allele was greatest in patients in the Ottawa Heart Study with severe premature atherosclerosis as compared to subjects with incident CAD events1 suggests that it acts by increasing the atherosclerotic burden rather than by promoting a thrombotic or electric event.
To explore the molecular basis of the effect of the 9p21.3 locus on gene sets involved in cellular proliferation, we focused on further characterization of ANRIL, a newly annotated noncoding RNA. Notably, the high-risk CAD haplotype overlaps exons 13 to 20 of ANRIL and ANRIL is expressed in many cell types known to be affected by atherosclerosis. In an attempt to refine the evolutionary origin of ANRIL, we carried out phylogenetic footprinting analysis of the sequences including ANRIL transcription units from multiple species. When present, functional units of genes such as exons and regulatory elements are usually conserved between different species, whereas irrelevant DNA sequences are randomized. Our data demonstrate that although exon sequences of ANRIL are highly conserved in primates, the sequence conservation is not apparent in orthologous sequences from dog, mouse, rat, and opossum. Furthermore, sequence analysis indicates that several exons of ANRIL retain repetitive elements. The presence of repetitive elements within the ANRIL gene is surprising because low complexity DNA sequences are not usually found within coding regions. A recent study suggests that the presence of repetitive elements is a common signature for novel genes originated from retroposition and exon shuffling, and degeneration of such elements is correlated with the increasing ages of the genes.11 Thus, ANRIL may have arisen relatively recently in mammalian evolution before the divergence of the primate lineage and may possess novel primate-specific functions. Alternatively, low sequence conservation of ANRIL may reflect high rates of primary sequence evolution suggested for long ncRNAs.12
We examined the structure and function of 4 conserved noncoding sequences present in the 9p21.3 locus and demonstrated that the CNS3 regulatory region acts as an enhancer and significantly increases reporter gene expression when tested in COS-7 and human AoSMCs. CNS3 derived from the 9p21.3 risk allele exhibited significantly higher levels of enhancer activity compared to the reference CNS3 sequence. Furthermore, the observed difference in the CNS3 regulatory activity is attributable to the presence of a single nucleotide polymorphism, rs1333045, in strong LD with representative SNPs of the previously defined 9p21.3 risk region (r2=0.86). These findings indicate that the 9p21.3 risk allele alters activity of at least 1 regulatory sequence which in turn may lead to changes in expression levels of ANRIL or other genes relevant to atherosclerosis. CNS3 may act as a general enhancer increasing the total levels of ANRIL expression. Given the structural similarity of ANRIL to other long ncRNA such as XIST and HOTAIR, ANRIL regulatory mechanisms are likely to rely on multiple transcriptional or splicing regulators.13,14 We hypothesize that, in addition to CNS3, other sequences within the 9p21.3 risk haplotype may modulate activity of ANRIL splicing enhancer(s) resulting in predominant expression of shorter ANRIL variants.
Noncoding RNAs such as ANRIL have been implicated in a variety of cellular pathways and stress responses and may influence gene transcription through multiple mechanisms that may include RNA interference, gene silencing, chromatin remodeling, or DNA methylation. For example, one of the best characterized large noncoding RNAs, XIST, initiates the process of X-inactivation by recruiting Polycomb-Repressive complexes to the X-chromosome.13 Similarly, Hox antisense transcript, HOTAIR, recruits Polycomb complex PRC2 to the HOXD locus and imposes a repressive chromatin state.15 Interestingly, several lines of evidence suggest that Polycomb repressor complexes are able to bind and alter expression of CDKN2A and CDKN2B transcripts at the INK4/ARF locus.16 Similarly to XIST, the ANRIL gene contains an unusually high number of repetitive elements and encompasses multiple binding sites for transcription factors that regulate transcriptional repression, such as Plzf17 (data not shown). It is thus plausible that ANRIL regulates expression of the INK/ARF transcripts by mechanisms analogous to that of XIST and HOTAIR.
A previous study demonstrated that ANRIL is expressed in tissues and cell types that are affected by atherosclerosis but based on the primers used, the data obtained reflect the combined expression levels of short (DQ485454) and long (DQ485453) ANRIL transcripts.6 We have extended these findings by documenting expression of each of DQ485453, DQ485454, and EU741058 in multiple human tissue types relevant to atherosclerosis (supplemental Figure III).
Importantly, we demonstrate that expression levels of short variants of ANRIL are increased and expression levels of the long variant are decreased in subjects homozygous for the risk alleles compared to those carrying two copies of reference alleles (Figure 3A through 3C). Previous studies indicated a positive correlation between expression levels of the long variant of ANRIL and genes controlling cellular proliferation, notably CDKN2A (p14, p16) and CDKN2B (p15). Similarly, we noted a positive correlation between transcript levels of the long variant of ANRIL and CDKN2B (Figure 3D). These findings suggest that SNP(s) within the 9p21.3 locus may act as a molecular switch resulting in reciprocal changes in expression levels of the short and long ANRIL transcripts. By reducing expression of the long ANRIL transcript or increasing expression of the short ANRIL transcripts, the risk allele may modify expression of cell cycle regulatory genes, thereby promoting a proliferative phenotype in arterial smooth muscle cells or other cell types relevant to atherosclerosis.
Consistent with this hypothesis, distinct gene expression signatures were noted for homozygous carriers of the risk versus reference alleles with evidence for highly significant upregulation of gene sets involved in cellular proliferation in healthy control subjects and also confirmed in patients with stable CAD. Furthermore, the top differentially expressed genes verified by QRT-PCR included KALRN and TDGF1, were previously linked to atherosclerosis or myocardial infarction.9,10 KALRN functions in the Rho GTPase signal tranduction pathway, which regulates a number of cellular processes including cell proliferation, migration, and adhesion. TDGF1 has potentially complex effects on cellular proliferation because it is required for signaling by some TGF-β family members such as Nodal but antagonizes others such as activin.18,19 Bioinformatics analysis indicated that the rs1333045 risk allele resides within a Smad-binding site, and Smad3 is apparently required for the antiproliferative effect of TGF-β.20 Consistent with allele-specific expression differences detected in healthy individuals, we also demonstrated upregulation of several cellular proliferation pathways as well as increased expression of short variants of ANRIL and reciprocally lower expression of CDKN2B, GAPDH and TDGF1 in CAD patients homozygous for the risk allele.
Overall, these studies suggest that SNP(s) within the 58-kb region orchestrate a series of molecular events which promote a proliferative response and are associated with altered expression of individual ANRIL transcripts. ANRIL is expressed in arterial smooth muscle cells, and vascular smooth muscle cell proliferation and migration are important events in both atherosclerosis and aneurysm formation.21 Relevant to these findings, the 9p21.3 locus has also been linked to abdominal and intracranial aneurysms.22 We demonstrate that a conserved sequence (CNS3) within the 9p21.3 locus has enhancer activity and that the risk allele of rs1333045 within CNS3 significantly increases reporter gene expression in primary aortic smooth muscle cells. We hypothesize that this or other allelic variants within the 9p21 risk locus mediates the increased predisposition to CAD by altering expression levels of short and long ANRIL transcripts, which in turn may affect cellular proliferation pathways. These risk allele-mediated pathways are potential new targets for therapeutic intervention in CAD and may provide expression markers that are associated with risk of developing premature CAD.
We thank the study subjects for their generous participation in this study. We are grateful to Melanie Belanger, Heather Doelle, Shailah Murji, Sabeen Mapara, and Angela Berkeley for their excellent technical assistance and Erinn B. Broshko and Dr Nathan Yoganathan for helpful discussions.
Sources of Funding
This study was supported by the Canadian Institutes of Health Research, Canadian Foundation for Innovation, the Heart & Stroke Foundation of Ontario (NA-6640), and the National Research Council Industrial Research Assistance Program. R.M. holds the Merck Frosst Canada Chair in Atherosclerosis Research.
BM, CB are employees of, and RC and JP are consultants for, MedBiogene Inc.
Received November 4, 2008; revision accepted June 17, 2009.
McPherson R, Pertsemlidis A, Kavaslar N, Stewart A, Roberts R, Cox DR, Hinds DA, Pennacchio LA, Tybjaerg-Hansen A, Folsom AR, Boerwinkle E, Hobbs HH, Cohen JC. A common allele on chromosome 9 associated with coronary heart disease. Science. 2007; 316: 1675–1684.
Schunkert H, Gotz A, Braund P, McGinnis R, Tregouet DA, Mangino M, Linsel-Nitschke P, Cambien F, Hengstenberg C, Stark K, Blankenberg S, Tiret L, Ducimetiere P, Keniry A, Ghori MJ, Schreiber S, El Mokhtari NE, Hall AS, Dixon RJ, Goodall AH, Liptau H, Pollard H, Schwarz DF, Hothorn LA, Wichmann HE, Konig IR, Fischer M, Meisinger C, Ouwehand W, Deloukas P, Thompson JR, Erdmann J, Ziegler A, Samani NJ. Repeated replication and a prospective meta-analysis of the association between chromosome 9p21.3 and coronary artery disease. Circulation. 2008; 117: 1675–1684.
Pasmant E, Laurendeau I, Heron D, Vidaud M, Vidaud D, Bieche I. Characterization of a germ-line deletion, including the entire INK4/ARF locus, in a melanoma-neural system tumor family: identification of ANRIL, an antisense noncoding RNA whose expression coclusters with ARF. Cancer Res. 2007; 67: 3963–3969.
Amaral PP, Dinger ME, Mercer TR, Mattick JS. The eukaryotic genome as an RNA machine. Science. 2008; 319: 1787–1789.
Broadbent HM, Peden JF, Lorkowski S, Goel A, Ongen H, Green F, Clarke R, Collins R, Franzosi MG, Tognoni G, Seedorf U, Rust S, Eriksson P, Hamsten A, Farrall M, Watkins H. Susceptibility to coronary artery disease and diabetes is encoded by distinct, tightly linked SNPs in the ANRIL locus on chromosome 9p. Hum Mol Genet. 2008; 17: 806–814.
Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, Pomeroy SL, Golub TR, Lander ES, Mesirov JP. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005; 102: 15545–15550.
Wang L, Hauser ER, Shah SH, Pericak-Vance MA, Haynes C, Crosslin D, Harris M, Nelson S, Hale AB, Granger CB, Haines JL, Jones CJ, Crossman D, Seo D, Gregory SG, Kraus WE, Goldschmidt-Clermont PJ, Vance JM. Peakwide mapping on chromosome 3q13 identifies the kalirin gene as a novel candidate gene for coronary artery disease. Am J Hum Genet. 2007; 80: 650–663.
Shani G, Fischer WH, Justice NJ, Kelber JA, Vale W, Gray PC. GRP78 and Cripto form a complex at the cell surface and collaborate to inhibit transforming growth factor beta signaling and enhance cell growth. Mol Cell Biol. 2008; 28666–72867.
Bracken AP, Kleine-Kohlbrecher D, Dietrich N, Pasini D, Gargiulo G, Beekman C, Theilgaard-Monch K, Minucci S, Porse BT, Marine JC, Hansen KH, Helin K. The Polycomb group proteins bind throughout the INK4A-ARF locus and are disassociated in senescent cells. Genes Dev. 2007; 21: 525–530.
Gray PC, Harrison CA, Vale W. Cripto forms a complex with activin and type II activin receptors and can block activin signaling. Proc Natl Acad Sci U S A. 2003; 100: 5193–5198.
Kelber JA, Shani G, Booker EC, Vale WW, Gray PC. Cripto is a noncompetitive activin antagonist that forms analogous signaling complexes with activin and nodal. J Biol Chem. 2008; 283: 4490–4500.
Boucher P, Gotthardt M, Li WP, Anderson RG, Herz J. LRP: role in vascular wall integrity and protection from atherosclerosis. Science. 2003; 300: 329–332.
Helgadottir A, Thorleifsson G, Magnusson KP, Gretarsdottir S, Steinthorsdottir V, Manolescu A, Jones GT, Rinkel GJ, Blankensteijn JD, Ronkainen A, Jaaskelainen JE, Kyo Y, Lenk GM, Sakalihasan N, Kostulas K, Gottsater A, Flex A, Stefansson H, Hansen T, Andersen G, Weinsheimer S, Borch-Johnsen K, Jorgensen T, Shah SH, Quyyumi AA, Granger CB, Reilly MP, Austin H, Levey AI, Vaccarino V, Palsdottir E, Walters GB, Jonsdottir T, Snorradottir S, Magnusdottir D, Gudmundsson G, Ferrell RE, Sveinbjornsdottir S, Hernesniemi J, Niemela M, Limet R, Andersen K, Sigurdsson G, Benediktsson R, Verhoeven EL, Teijink JA, Grobbee DE, Rader DJ, Collier DA, Pedersen O, Pola R, Hillert J, Lindblad B, Valdimarsson EM, Magnadottir HB, Wijmenga C, Tromp G, Baas AF, Ruigrok YM, van Rij AM, Kuivaniemi H, Powell JT, Matthiasson SE, Gulcher JR, Thorgeirsson G, Kong A, Thorsteinsdottir U, Stefansson K. The same sequence variant on 9p21 associates with myocardial infarction, abdominal aortic aneurysm and intracranial aneurysm. Nat Genet. 2008; 40: 217–224.