Background of the Invention Human coronary atherosclerosis is a multifactorial disease, which develops insidiously over decades progressing from quiescent to overt life- threatening coronary artery disease (CAD) clinically known as acute coronary syndromes-the spectrum of unstable angina, acute myocardial infarction and sudden coronary death (1,2). Pathogenically, acute coronary syndrome is the clinical manifestation of two concurrent pathological processes: a vascular pathology marked by"culprit"atherosclerotic plaques and myocardial pathology marked by ischemia and/or infarction (3-5), hence typically studied separately. Comparative analysis of emerging paradigms from these studies highlights parallel cellular events involving inflammation, apoptosis, matrix degradation, and neutrophil recruitment in both pathogenic hallmarks of overt- CAD (the destabilized"culprit"plaque (3) and post-infarction myocardium (4) ). These parallel changes reflect either an interesting, albeit inconsequential coincidence or a central interactive paradigm. The challenge therefore lies in the elucidation of the pathogenic framework involved in overt-CAD as a first step, and the subsequent delineation of mechanisms within and cross talk between vascular and myocardial pathological events. The notion of cross-talk seems intuitively pertinent given that circulating elevated inflammatory cytokines would affect both vascular and myocardial events. Critical first-step analysis in validated animal models of coronary artery disease becomes key.

Transcription profiling offers an investigative paradigm that can identify framework concepts of pathogenesis and specific gene pathways once confounding variables are regulated. The existence of confounding variables, however, is inherent in acute coronary syndromes given the characteristic heterogeneity of"culprit plaques" (5) and differential genetic susceptibility to atherosclerotic plaque formation (6). These features underscore an inherent complexity in the progression from quiescent-to overt-CAD and reiterate the need for first-line transcription profiling investigation in validated surrogate experimental systems, wherein confounders can be experimentally controlled while maintaining potential for integrated analysis. A priori, the presence of a reproducible transcriptional profile that distinguishes overt-CAD from quiescent-CAD would imply that an active transcriptional process is central to the pathogenesis of overt-CAD. Determining whether there is indeed a differential transcription profile in overt-CAD would provide critical insight into detection, intervention, and prevention pathways.

Summary of the Invention This invention features a method for diagnosing a human patient having quiescent or overt coronary artery disease (CAD). The method includes the steps of : (i) measuring the expression of genes from the cluster controlling lipid entry, matrix regulation, inflammation, the oxidative stress response, the contractile apparatus, proliferation, angiogenesis, the adrenergic system, cardiac ion channels and pumps, and cardiac energy metabolism; (ii) comparing the level of expression of the genes from the patient to the level of expression in a subject not having CAD; (iii) scoring each of the genes from the cluster controlling lipid entry, matrix regulation, inflammation, the oxidative stress response, the contractile apparatus, the TSC-22 gene, and the potassium channel protein gene, that are induced more than 50%;

(iv) scoring each of the genes from the cluster controlling proliferation, other than the TSC-22 gene, angiogenesis, the adrenergic system, cardiac ion channels and pumps, other than the potassium channel protein gene, and cardiac energy metabolism, that are deinduced more than 50%; and (v) diagnosing the patient as having CAD wherein a total of five or more genes are scored from step (iii) and step (iv). In preferred embodiments, cardiac tissue or neutrophils are used for gene cluster analysis.

This invention also features a method for identifying candidate compounds useful for treating or preventing CAD. The method includes the steps of : (i) providing a cell expressing five or more of the genes identified in Figure 2; (ii) contacting the cell with the candidate compound; (iii) measuring expression of the genes identified in Figure 2; (iv) comparing the expression of the genes with the expression in a cell that has not been contacted with the candidate compound; and (v) interpreting the comparison wherein a total of five or more genes either (a) from the cluster of genes controlling lipid entry, matrix regulation, inflammation, the oxidative stress response, the contractile apparatus, the TSC- 22 gene, or the potassium channel protein gene are induced greater than 50%, or (b) from the cluster of genes controlling proliferation, other than the TSC-22 gene, angiogenesis, the adrenergic system, cardiac ion channels and pumps, other than the potassium channel protein gene, and cardiac energy metabolism are deinduced greater than 50%, identify the candidate compound as being useful for treating or preventing CAD. In preferred embodiments, the genes include endothelial receptor for oxidized low density lipoprotein (LOX-1), tissue inhibitor of metalloproteinases-1 (TIMP-1), TIMP-2, lysyl oxidase, the membrane-type metalloproteinases, matrix G 1 a protein, Fc gamma receptor II, interleukin (IL) 1-beta, IL-18, thyrotropin releasing hormone, glutathione peroxidase-1, fast myosin light chain-3 (fast MLC-3), transforming growth

factor (TGF) -beta-stimulated clone-22 transcription repressor (TSC-22), activating transcription factor-3, early growth response-1 (EGR-1), insulin-like growth factor binding protein-3 (ILGF-bp3), cyclin Dl, vascular endothelial growth factor-A (VEGF-A), microvascular endothelial differentiation factor-1, adrenergic receptors (a 1 A and a 1 B), the G-protein P-subunit, a2-NaK- ATPase, the potassium channel protein, the r-ERG potassium channel, mitochondrial cytochrome B5, solute carrier family 20, pyruvate dehydrogenase kinsase-4, glucose transporter-4, branched chain a-ketoacid dehydrogenase, acyl Co-A dehydrogenase, glucagons receptor, cyclic adenosine monophosphate (cAMP)-activated protein kinase (PKA), and SNF- related kinase.

By"gene cluster"or"cluster of genes"is meant a group of genes which are involved in the co-ordinate, functional regulation of a metabolic pathway.

By"gene cluster controlling lipid entry"is meant the group of genes involved in cellular uptake of lipids from the circulation as well as genes controlling intracellular lipid trafficking. An example of a gene from the cluster controlling lipid entry includes the endothelial receptor for oxidized low density lipoprotein (LOX-1).

By"gene cluster controlling matrix regulation"is meant the group of genes that control the synthesis, stability, and degradation of the extracellular matrix. Examples of genes from the cluster controlling matrix regulation include tissue inhibitor of metalloproteinases-1 (TIMP-1), TIMP-2, lysyl oxidase, the membrane-type metalloproteinases, and matrix Gla protein.

By"gene cluster controlling inflammation"is meant the group of genes that regulate the inflammatory response associated with sclerotic atherosclerotic plaques and post-infarction myocardial inflammation.

Examples of genes from the cluster controlling inflammation include the Fc gamma receptor II, interleukin (IL) 1-beta, IL-18, and thyrotropin releasing hormone.

By"gene cluster controlling the oxidative stress response"is meant the group of genes that respond to oxidative stress, particularly those found in the post-infarction (post-ischemic) myocardium. An example of a gene from this cluster is glutathione peroxidase-1.

By"gene cluster controlling the contractile apparatus"is meant the group of genes that control myofibril shortening and cardiomyocyte contractility. An example of a gene from this cluster is fast MLC-3.

By"gene cluster controlling proliferation"is meant the group of genes that control cardiomyocyte, vascular smooth muscle cell, and fibroblast proliferation. Examples of genes in this cluster include TGF-beta-stimulated clone-22 transcription repressor (TSC-22), activating transcription factor-3, early growth response-1 (EGR-1), insulin-like growth factor binding protein-3 (ILGF-bp3), and cyclin D1.

By"gene cluster controlling angiogenesis"is meant the group of genes that promote the formation of blood vessels. Examples of genes from this cluster include vascular endothelial growth factor-A (VEGF-A) and microvascular endothelial differentiation factor-1.

By"gene cluster controlling the adrenergic system"is meant the group of genes that comprise and control the cardiac and vascular beta-adrenergic system. Examples of genes from this cluster include the adrenergic receptors (a lA and alB) and the G-protein (3-subunit.

By"gene cluster of cardiac ion channels and pumps"includes all cardiac and vascular ion channels, ion transporters (ATP-dependent or-independent).

Examples of genes from this cluster include a2-NaK-ATPase, the potassium channel protein, and the r-ERG potassium channel.

By"gene cluster controlling energy metabolism"is meant the group of genes that regulate or participate, either directly or indirectly, in oxidative phosphorylation in the post-infarction myocardium. Examples of genes from this cluster include mitochondrial cytochrome B5, solute carrier family 20,

pyruvate dehydrogenase kinsase-4, glucose transporter-4, branched chain a- ketoacid dehydrogenase, acyl Co-A dehydrogenase, glucagons receptor, cAMP-activated protein kinase (PKA), and SNF-related kinase.

By"cell-based expression system"is meant any in vitro cell culture system in which test cells express, either naturally or by artifice, one or more CAD-related genes, such as those identified in Figure 2. Cell-based expression systems useful for the present invention can consist of a single cell type expressing a plurality of CAD-related genes. Alternatively, the system may contain a plurality of cells wherein each cell expresses a single CAD-related gene of interest. In this case, each cell type expressing a CAD-related gene may be segregated from each other cell type expressing a different CAD- related gene. In an alternative format, the cells expressing different CAD- related genes are not segregated from each other (i. e. , cells are present in a mixed culture).

By"induced"is meant an increase of gene expression above control levels.

By"deinduced"is meant a reduction of gene expression below control levels.

Brief Description of Drawings FIGURE 1 is a series of scatter plots showing an overview of transcription profile changes in overt CAD versus quiescent CAD. Scatter plots of right ventricular transcription profiles reveals a significant number of genes with > 2-fold expression changes between (A) overt CAD [o-CADR] and quiescent CAD [q-CADR] ; (B) overt CAD [o-CADR] and attenuated quiescent CAD [aq-CADL] ; (C) Scatter plot comparing quiescent CAD [q-CADR] and attenuated CAD [aq-CADL] reveals minimal >2-fold gene expression changes.

Units 1-10, 000 = LoglO (signal-local background) intensity of fluorescence signal.

FIGURE 2 is a table showing functional gene-cluster specific differential regulation of turnkey genes. Based on known gene function, genes that exhibit consistent vectorial change (> 1.5 fold) in gene expression in overt CAD compared with both quiescent CAD control groups fall into the following functional clusters: lipid entry, matrix balance, inflammation, oxidative stress response, contractile apparatus, proliferation, angiogenesis, adrenergic system, channels-pumps, energy/metabolism. Induction of gene expression detected spans 1.5 to 40-fold ; deinduction spans 1.5 to 40-fold. Induced functional gene clusters are distinct from deinduced functional gene clusters. Pattern of expression changes were identified by ranking levels of expression in the following sequence: overt o-CAD, quiescent q-CAD and attenuated, quiescent aq-CAD as noted: gradient pattern (rank 321: 30-CAD >2>1 or rank 123: 1°-CAD <2 < 3); step-change pattern (rank 133: 1°-CAD <3=3, or rank 311: 30-CAD >1=1). ID#, unigene ID number; rl, r2, r3, r4: 8 month-old end-stage Tg53 rats analyzed individually, compared with quiescent-CAD (pooled n = 4); and attenuated, quiescent aq-CAD (pooled n = 4); x2, refers to duplicate spots on each array.

FIGURE 3 is a photomicrograph of TIMP-1 immunohistochemical analysis of coronary plaques in end-stage and non-end-stage Tg53 rats. (A) TIMP-1 expression (stained brown) is detected (-+) in end-stage Tg53 proximal right coronary"culprit"plaque that is foam-cell rich, with prominent extracellular lipid accumulations, and minimal to no fibrous cap. Scale bar = 50 microns. Immunostaining is also detected in endothelial cells of arterioles and capillaries (C). Vascular media and adventitia are not stained. (B) & (C) High magnification (scale bar = 25 microns) of plaques depicted in panel A, corroborating specificity of TIMP-1 immunostaining in culprit plaque foam cells and matrix. Endothelial cells in these lesioned areas () do not express TIMP-1, in contrast to expression in endothelial cells of intramyocardial arterioles/capillaries (C). (D) In contrast to proximal coronary"culprit" plaques, TIMP-1 immunostaining is not detected in distal coronary stable

plaque (+ ) in end-stage Tg53 RV. Scale bar = 10 microns. (E) In attenuated, quiescent-CAD, proximal coronary lesions are smaller and exhibit a cap (++). TIMP-1 immunostaining is detected in macrophage foam cells and overlying endothelium. Bar = 25 microns. (F) High magnification (bar = 10 microns) of panel E corroborates TIMP-1 immunostaining in plaque macrophage foam cells and endothelium ; no immunostaining is detected in the media or adventitia. (ad, adventitia; m, media; rbc, red blood cells) FIGURE 4 is a schematic diagram describing a framework of transcriptional profile changes in overt coronary artery disease. Based on known gene functions, interaction links between turnkey genes exhibiting vectorial changes in overt CAD compared with both quiescent CAD controls reveal a circuitous self-propagating interacting gene network that is concordant with emerging pathogenic paradigms of plaque destabilization and post- infarction myocardial injury. Interaction links with known correlates of acute coronary syndromes (shear stress, oxidized LDL, C-reactive protein) are noted.

Genes not implicated before in acute coronary syndromes, TRH and FcyII receptor, also fit nicely into this self-propagating interacting gene network.

Interactions are indicated by like-color and like-symbols: solid symbols induce target (open symbols) matched as to color and shape; +, induction ; deinduction; ACS, acute coronary syndrome; actv'n, activation; acyl CoA dehydrogenase; adrnrg rec, adrenergic receptor; AMP-act pr knse; adenosine- monophosphate-activated protein kinase; branched-chain alpha-keto dehydrogenase subunit E1 ; CRP, C-reactive protein; dysfx, dysfunction; EC, endothelial cell; Egr-1, early growth response-1; FcyIIr, Fc gamma-11 receptor; G-pr ß, G-protein ß subunit; gluthn perox-I, glutathione peroxidase-I ; hge, hemorrhage; IL-lß, interleukin 1-beta; IL-18, interleukin-18; ILGF-bp3, insulin-like growth factor binding protein-3; intraplq, intraplaque; LDL-IC, low density lipoprotein immune complexes; LOX-1, endothelial oxidized LDL receptor; glucose transporter 4 ; p+, macrophage; MT-mmp, membrane-type

metalloproteinase; mobilizn, mobilization; occlus'n, occlusion; oxLDL, oxidized LDL, PMN, polymorphonuclear leukocyte or neutrophil; pyruv dehydr knse, pyruvate dehydrogenase kinase; SNF, SNF-related kinase; solute carrier family 20 ; TSC-22, TGF-ß stimulated clone 22 transcription repressor ; TF, transcription factor; TIMP-1, tissue inhibitor of metalloproteinases-1 ; 9RH, thyrotropin releasing hormone; VEGF-A, vascular endothelial growth factor-A; VLDL-IC, very low density lipoprotein immune complexes.

Detailed Description This invention provides methods for diagnosing both quiescent and overt CAD in human patients. The invention also provides screening methods for identifying candidate compounds useful for treating quiescent and overt CAD and reducing the progression of CAD pathology.

We investigated the transcription profile of right ventricles with different stages of CAD in the inbred transgenic hyperlipidemic-polygenic hypertension Tg53 rat model (7). This model develops significant coronary atherosclerosis simulating key histological features of"culprit"coronary plaques in human acute coronary syndromes (8). We compared eight month old male Tg 53 rats having clinically overt CAD with those fed regular rat chow and having quiescent CAD with equivalent hyperlipidemia and hypertension (qCADR). Also compared were age-and gender-matched rats, fed low salt diets, having quiescent CAD and equivalent hyperlipidemia but reduced hypertension (aq CADL) (8). We selected the right ventricle (RV) as tissue source for transcription profiling because coronary atherosclerosis in this model is more robust in the RV than the left ventricle (LV) and, being thin- walled and non-hypertrophied compared to the LV, it provides a relative experimental enrichment of coronary artery disease pathology. Thus, the integrated analysis of both vascular and myocardial events in overt-CAD is

possible. Transcription profiling was done using a rat-specific array containing 1046 unique genes and 109 additional gene spots for negative, positive and quantitative controls.

Scatter plots of gene array data from right ventricular RNA samples provide an overview of comparative transcription profiles (Figure 1). Most genes are expressed above the array technical threshold which serves as a stringent parameter designed to eliminate false positives. Of 1046 unique genes on the array used, only a minority exhibit >2-fold vectorial change marked as outliers above or below the red-lined 2-fold cutoff (Figure 1).

Differential analysis of transcription profiles of pooled (n = 4) right ventricular total cellular RNA from Tg53 male rats on regular rat chow exhibiting overt CAD [o-CAD] R highlights 2-fold vectorial changes in gene expression compared with pooled (n = 4) right ventricular total cellular RNA from control age-and diet-matched Tg53 male rats with quiescent, empirically asymptomatic CAD [q-CAD] R (Figure 1A), as well as when compared with Tg53 rats with attenuated, quiescent-CAD on low salt diet (n = 4) (8), Tg53 [aq- CAD] L (Fig. 1B). As would be expected from an attenuated phenotype on low salt diet (8), more genes exhibit >2-fold induction when comparing overt-CAD in Tg53 [o-CAD] R rats with attenuated, quiescent-CAD Tg53 [aq-CAD] L rats (Fig. 1B), in contrast to the transcription profile comparison between overt-and quiescent (non-attenuated) CAD (Figure 1 A). Transcription profile comparison of both quiescent CAD controls, Tg53 [q-CAD] R vs Tg53 [aq- Cadi, reveals quite similar expression profiles (Fig. 1C), thus providing confidence in the detected differential transcription profile in overt-CAD compared with two different control groups. A priori, a differential transcription profile implies differential molecular framework in overt-CAD.

In order to gain insight into the differential gene pathways, we then dissected the differential expression profiles revealed in Fig. 1 for functional gene clusters. First, we filtered for consistency of vectorial individual gene expression changes detected in overt CAD compared to both quiescent CAD

controls. We defined consistency of gene expression changes as o-CAD > q- CADR 2 qa-CADL, or inversely aq-CADL 2 q-CADR > o-CAD. Second, we filtered for reproducibility in independent array experiments. Only changes that were detected reproducibly when RV samples were analyzed individually and when pooled (n = 4) are presented. Third, we filtered for robust detection signals. Gene expression changes are noted only if the normalized expression level (signal-local background) is greater than threshold (local background) in the overt CAD samples or in both quiescent CAD control groups. Genes consistently showing >1. 5-fold vectorial change in expression in overt CAD compared with both control groups for quiescent CAD are delineated.

Although the typical 2 2-fold change has to date become the emerging standard for accepting vectorial fold-changes in array analysis that minimizes spurious results, we report > 1.5-fold changes that were reproducible in independent experiments. Assignments were based on the pattern of gene expression levels where a gradient pattern implies worsening plaque changes and a step-change pattern implies association with cardiac compromise and/or plaque compromise. After this quantitative filtering, qualitative functional gene clustering was done based on known gene function or when unknown, based on homology with known genes.

Of the 1046 unique genes on the rat array used, 568 (54%) are expressed above array threshold in 8 month-old Tg53 rats with overt-CAD. Of these, only 2.3% are induced and 3.7% are deinduced in overt-CAD compared with age-matched Tg53 rats with quiescent-CAD. Assuming randomness of genes on the array, gene expression changes stratify into distinct functional gene clusters between induced versus de-induced genes (Figure 2). Interestingly, 69% of induced genes belong to functional gene clusters implicated in coronary plaque destabilization such as matrix regulation 4/13, inflammation 4/13,

oxidized LDL entry 1/13. These functional gene groups are distinct from deinduced gene clusters (Figure 2), which involve energy metabolism 9/21, proliferation 5/21, a-adrenergic system 3/21, angiogenesis 2/21.

In addition to stringent parameters for consistency and robustness of expression signals among study individual and pooled samples, within-array reproducibility of gene expression levels was also observed for both positive and negative controls, and for markedly increased genes such as TIMP1 and membrane-type matrix metalloproteinase (Figure 2). Moreover, immunohistochemistry with TIMP-1 antibody detects expression in macrophage-foam cells of proximal high-risk lesions rather than in the myocardium or more distal coronary stable lesions (Figures 3A-D). This observation confirms the TIMP-1 results and helps validate the results from the gene array experiments. Immunohistochemical analysis detects a deinduction of TIMP-1 expression in endothelial cells overlying lesions in overt-CAD (Figures 3A-C) compared with attenuated quiescent-CADL (Figures 3E-F).

Analysis of the functional clusters exhibiting differential gene changes identifies putative gene networks involved in overt-CAD (Figure 4). Based on published research, gene pathways can be assigned to either or both hallmark pathologies of overt-CAD; plaque destabilization and myocardial ischemia/infarction (Figure 4). Confidence in these publication-based assignments can be derived from concordance with lesion or lesion+cardiac assignments predicted by pattern analysis of differential expression (Figure 2).

Furthermore, analysis of known functional interrelationships of induced genes among themselves and their respective effects on events associated with plaque destabilization reveals a putative circuitous gene network detected in overt- CAD (Figure 4).

More specifically, pathway analysis suggests several putative paradigms (Figures 2 and 4). First, a paradigm reminiscent of tumor invasiveness brings coherence to observed changes in matrix degradation, inflammation, and apoptosis with concomitant induction of MT-MMP, TIMP-1, IL-1 beta, IL-18;

genes associated respectively with tumor invasiveness (10-14). A tumor- invasiveness paradigm would be consistent with plaque destabilization.

Secondly, a paradigm of decreased proliferation having decrease of growth factors, transcription factors, and cell cycle regulators, as well as an increase of proliferation suppressor TSC-22, supports the histological hallmark of fibrous cap weakness in high-risk plaques marked by decreased number of smooth muscle cells and fibroblasts. Thirdly, a paradigm of lipid-associated and lipid- immune complex triggers, with respective increased oxLDL endothelial receptor and macrophage/neutrophil Fc yII receptor, provides a plausible initiator of inflammatory cascade (s) leading to destabilization. A fourth paradigm represents changes in functional gene clusters that reflect an altered myocardial functional state: al adrenergic receptors, G-protein P subunit, K+ channels, a2 Na, K-ATPase, and genes associated with energy utilization and metabolism (Figures 2 and 4).

These paradigms could be united into a hypothesis of overt-CAD pathogenesis concordant with known associated events in acute coronary syndromes: initiating trigger paradigms with increased adrenergic stress and/or increased oxLDL/lipid-immune complex entry followed by plaque destabilization mimicking tumor invasiveness, decreased proliferation, and inflammation-facilitating mechanisms (Figure 2). A circuitous interaction through IL-1 and IL-18 induced macrophage, neutrophil and T-cell activation induces IL-1 and IL-18 mediating cytokines. Additionally, this circuitous interaction may be enhanced by induced lysyl oxidase gene expression which has been associated with increased monocyte motility (15), as well as by TRH- associated T-cell dependent immune response and increased macrophage superanion-induced oxidative stress.

Differentialfunctional gene cluster expression profiles The dichotomy of induced and de-induced functional gene clusters is quite compelling and provides confidence in the observations. Transcription profile analysis of aging skeletal muscle also detected a dichotomy of functional gene-cluster induction versus de-induction (16). Functional gene- clusters that are differentially regulated in overt-CAD compared with both quiescent-CAD controls are involved in overt-CAD pathogenesis, and that specific genes that are induced or deinduced in the clusters, most likely play putative turnkey roles in overt-CAD. These turnkey genes implicate upregulation of inflammation, matrix changes, oxidative stress response to a presumed oxidative stress, oxidized LDL entry, and a down regulation of proliferation, angiogenesis, adrenergic receptor system, metabolism, and ion channels and transporters which, altogether, assign a complex molecular framework of overt-CAD that is essentially concordant with known pathophysiological dysfunction in human acute coronary syndrome.

Fully cognizant that specific gene-based pathway determination is limited by the 1046-gene array, deduced framework concepts can assume more definitiveness. The data, whether through a 1000 or 10,000 array, demonstrates that transcriptional mechanisms are involved, thereby supporting the hypothesis that quiescent-to-overt CAD is an active transcriptionally- mediated pathogenic process and not just a secondary"wear-and-tear" disruption-thrombosis event. This is in marked contrast to the detected cardiac transcription profile from a mouse myocardial ischemia/infarction model wherein the left anterior descending (LAD) coronary artery was surgically ligated to zero flow (9). The only gene change in common is the induction of myosin light chain gene.

Threshold-players in the Onset of Overt CAD The identification of reproducible differential gene regulation from overt CAD in end-stage Tg53 rat hearts compared with either age-matched non-overt CAD (non-endstage Tg53-reg) and attenuated non-overt CAD (non-endstage Tg53-lsd) rat hearts supports the concept of putative threshold-gene players.

Because the array is limited to 1046 known genes, the full complement of threshold-gene players is not possible; larger arrays can be used to identify and characterize all of the important pathways and threshold gene players. A list of threshold-players (Figure 2) provides concordance with previous histological and immunohistochemical analyses (8). Of note, TIMP-1 induction aligns with plaque matrix stabilization and its robust induction is counterintuitive to plaque destabilization (17,18). TIMP-1, not the expected metalloproteinases, has been identified as a robust marker for invasive cancer (14), upregulated in premature coronary atherosclerosis (19). Thus, TIMP-1 may be a compensatory response and may serve as a biomarker for plaque destabilization. Likewise, lysyl oxidase is associated with stabilization of collagen fibers by crosslinking (21).

However, lysyl oxidase is also associated with enhanced monocyte motility (15), leading to activation of the macrophage-mediated plaque destabilization pathways, and with tumor invasiveness (22). The robust 25 to 29-fold induction of TRH in ventricular samples would implicate vascular adventitial and endothelial expression since it is not detected in ventricular cardiomyocytes by in situ hybridization (23). Because TRH has been shown to play a critical role in the T-cell mediated immune response (24) and the enhancement of superanion production of activated macrophages (25), we hypothesize that induced adventitial TRH represents a paracrine vascular pathway that could be expected to contribute to inflammatory cascades in plaque destabilization, as well as contribute to myocardial injury response. The induction of FcyII receptor suggests a possible role for lipoprotein-immune complex uptake and FcyII receptor activation-induced inflammatory

mechanisms in overt-CAD. This notion is supported by the detection of FcyII receptors in human atherosclerotic lesions (26). Additionally, we note that C- reactive protein, increasingly implicated as a coronary risk marker (27), binds to Fc yII receptors on monocytes and neutrophils (28). The induction of both IL-1 and IL-18 supports the hypothesis of a self-propagating upregulation of inflammatory response mediated by neutrophils, macrophages and/or T-cells (29-32), which contributes to acute plaque destabilization. Induction of both IL-1 and IL-18 also supports a tumor invasiveness paradigm of plaque destabilization (11, 12). These hypotheses are supported by the detection of IL- 1 and IL-18 in human atherosclerotic plaques and relation to plaque instability (33).

Transcriptionalframework of overt-CAD Having controlled genetic background, age, gender, diet, and environment, the transcriptional profile of overt CAD exhibits detectable distinct functional gene cluster regulation compared with quiescent CAD.

Although limited to a 1046-gene array, the robust and reproducible transcription profile results associated with overt-CAD in end-stage Tg53 rats, and the absence of change in LAD-ligated myocardial infarction transcription profile (7), as well as in a shear-stress vascular smooth muscle cell model (34) strongly support the specificity of our observations for overt CAD.

Altogether, the data demonstrate that transition to overt CAD involves a transcriptionally-active pathogenic framework with a set of turnkey genes which, based on known function and interaction, comprise a self-amplifying interrelated gene-network that can drive processes linked to both coronary plaque destabilization and post-ischemic myocardial destabilization including matrix degradation, inflammation, and lipid-entry, and de-induced proliferation and angiogenesis. These processes arise from alterations in metabolism, ion transporters, and the alpha-adrenergic receptor system.

Diagnosis of Coronary Artery Disease The present discovery of gene cluster regulation allows for more accurate diagnosis and staging of CAD in human subjects. Those skilled in the art will appreciate that simultaneously correlating changes in several CAD- associated genes provides the clinician with a more complete overview of the disease process and also provides multiple confirmatory endpoints, reducing the error caused by inter-individual variability. Thus, observing a co-ordinate regulation of multiple genes within and across disease-related pathways reduces the chance of misdiagnosis compared to single endpoint tests.

Diagnostic assays, according to the methods of this invention, measure the expression of at least five genes from CAD-associated clusters. The relevant gene clusters and exemplary genes from each cluster are provided in Figure 2. Typically, changes in gene expression (induction or deinduction) of 50% or more, having the directionality described in Figures 2 or 4, contribute to the onset of CAD. Identification of five or more genes with altered expression are indicative of CAD. Any assay that measures gene expression can be used as a diagnostic assay for the purpose of this invention. For example, RNA can be isolated from cardiomyocytes or blood neutrophils and quantified by Northern blotting. Optionally, reverse transcriptase PCR (RT- PCR) can be performed prior to quantification.

Nucleic acid chips (e. g. , Affymetrix chips) can also be used to rapidly and reliably quantify gene expression. A description of the use of nucleic acid <BR> <BR> chips can be found in, e. g. , U. S. Patent Nos. 6,344, 316,6, 340,565, 6,333, 155, 6,306, 643,6, 040,138, 5,695, 937,5, 445,934, and PCT Application Nos. WO 97/10365 and WO 92/10588. Thus, chips containing members of the CAD- associated gene clusters can be used for routine diagnostic purposes.

Furthermore, nucleic acid chip technology can be used to monitor the progression of CAD and/or responsiveness of the patient to treatment. For example, following treatment, another RNA sample from the same cell type is obtained and used to generate a second expression profile. Changes in the

profile, compared to pre-treatment expression levels, are indicative of the efficacy of treatment, or worsening of the CAD, while a lack of change can be indicative that the therapy has thus far been ineffective.

Identification of Candidate Compounds for Treatment or Prevention of Coronary Artery Disease A candidate compound useful for the treatment, prevention, or stabilization of CAD can also be identified using the methods of the present invention. A candidate compound can be identified by its ability to affect the expression of one or more genes of a CAD-associated gene clusters, such as those clusters identified in Figure 2. Specifically, candidate compounds that reverse CAD-related changes in expression of these identified genes can be used as therapeutics.

Screening assays known in the art can be adapted for measuring changes in the expression of CAD-associated genes. For example, expression of a reporter gene that is operably linked to a CAD-associated gene promoter can be used to identify such candidate compounds. A reporter gene may encode a reporter enzyme that has a detectable read-out, such as beta-lactamase, beta- galactosidase, or luciferase. Reporter enzymes can be detected using methods known in the art, such as the use of chromogenic or fluorogenic substrates for reporter enzymes and such substrates are known in the art. Such substrates are desirably membrane permeat. Chromogenic or fluorogenic readouts can be detected using, for example, optical methods such as absorbance or fluorescence. A reporter gene can be part of a reporter gene construct, such as a plasmid or viral vector, such as a retrovirus or adeno-associated virus. A reporter gene can also be extra-chromosomal or be integrated into the genome of a host cell. The expression of the reporter gene can be under the control of exogenous expression control sequences or expression control sequences within the genome of the host cell. Under the latter configuration, the reporter gene is desirably integrated into the genome of the host cell. Preferably, each

host cell contains more than one reporter gene wherein each reporter gene operably linked to a different CAD-associated gene promoter. In this configuration, detection methods for each reporter gene should be distinguishable from each other.

Alternatively, a gene product can be isolated from a host cell expressing one or more of the CAD-associated genes. Preferably, the host cell is a cultured cardiomyocyte and the gene product is RNA or cDNA created by RT- PCR from the host cell RNA. Standard techniques, such as Northern or Southern blotting, are used to quantify gene products.

Other Embodiments All publications and patent applications mentioned in this specification are herein incorporated by reference to the same extent as if each independent publication or patent application was specifically and individually indicated to be incorporated by reference.

While the invention has been described in connection with specific embodiments thereof, it will be understood that it is capable of further modifications and this application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure that come within known or customary practice within the art to which the invention pertains and may be applied to the essential features hereinbefore set forth.

What is claimed is: