Matters of the Heart Transcriptome

Conveying the history of the development of genomics and more specifically of cardiovascular genomics can be done in a number of ways. We could choose a single aspect of the matter—a single gene, a single manifestation of a disease, a single protein, or a single therapy—and then explore the developments that took place over the last decades. But because history is still being born, an impressionistic portrait of the global scheme recommends itself: a selection of events that, although recent, might be considered historic. That this new science of medicine has inspired artists (see frontispiece) is a telltale sign that its history has begun.

Like all history, the history of cardiovascular genomics is unfinished. Like most history, it has no specific, unequivocal beginning. Yet its trends are being defined, and those trends concern interdisciplinary studies, mathematical models, personalized medicine, and preventive health care.

From Genetics to Genomics

It is difficult to pinpoint when the science of genetics was born, or the practice of cardiology for that matter. It is common knowledge that ancient Egyptian doctors knew that the pulse was related to the heart. Two thousand years ago, in India, Susrut suggested that diabetes was “passed through the seed,” which is one of the 1st known references to genetic inheritance. ¹ From our perspective, there is little argument that Mendel and his peas turned genetics into a quantitative discipline. W. Sutton talked about chromosomes in 1904, and W. Johansen about genes in 1909. T. Morgan's work around 1910 with the common fruit fly Drosophila melanogaster was also seminal, as was the realization around 1944 that nucleic acids, and not protein, were the genetic material. ² However, what truly caught everyone's attention in the 2nd half of the 20th century was deoxyribonucleic acid, or DNA.

In a captivating book titled Medicine's 10 Greatest Discoveries, published in 1998, ³ M. Friedman and G.W. Friedland describe Erwin Chargaff's reaction when asked who deserved the most credit for the discovery of DNA: “It's Friedrich Miescher. Miescher discovered DNA as a chemical entity. Isn't that important to you?” While dutifully acknowledging Miescher's 1871 article “Uber die Chemische Zusammensetzung der Eiterzellen,” the authors nevertheless titled their chapter “Wilkins and DNA.” However, in a famous article that appeared in the 23 April 1953 issue of Nature, ⁴ Maurice Wilkins was the last of 6 references. The other 5 were Pauling, Furberg, Chargaff himself, Wyatt, and Astbury. That paper, of course, is “Molecular Structure of Nucleic Acids: A Structure for Deoxyribose Nucleic Acid,” and its authors were J.D. Watson and F.H.C. Crick. Their words “we wish to put forward a radically different structure for the salt of deoxyribose nucleic acid,” and the diagrammatic representation of one of the possible DNA double helixes that accompanies those words, seem scanty for the impact that the DNA helix would have in the next few decades. Incidentally, only once do they use the abbreviation D.N.A. in their paper, and then they do so with periods. None of the current familiarity with which we speak about the nucleic acid molecules was present in that 1-page communication. A history of the unraveling of the structure of DNA would later be published by Watson himself in the famous bestseller The Double Helix. ⁵

Between the elucidation of the structure of DNA in 1953—along with the realization of its genetic implications that same year ⁶—and the publication in 2001 of a draft of the human genome sequence simultaneously in Science ⁷ and Nature, ⁸ only 48 years elapsed. During those 48 years, knowledge was acquired about DNA, RNA, and proteins. But from the point of view of the birth of genomic medicine, when did the conceptual quantum leap occur?

One possible answer is that it happened in 1978, when Y. Kan and A. Dozy published their finding that near the sickle-cell gene was a variable stretch of DNA ⁹—in other words, a polymorphism. Subsequently, in 1980, Botstein and colleagues ¹⁰ published their suggestion that the location of genes could be predicted in given kindreds by their genetic linkage to a variant. This was tantamount to mapping. The advent of positional cloning in the 1980s gave a new boost to the task of identifying inherited candidate-disease genes. ¹¹ Positional cloning permitted the isolation and characterization of a given gene once its approximate location was known. Using this strategy, investigators have identified dozens of crucial disease genes: in 1986, the gene for Duchenne muscular dystrophy; in 1991, the gene for familial adenomatous polyposis; in 1993, the gene for Huntington's disease; and in 1995, 1 gene for Alzheimer's disease. Similarly, a number of cardiovascular disease-associated genes were beginning to be mapped to certain positions in the chromosomes and “found.”

Among the 1st genes to attain the limelight were the apolipoprotein (APO) genes. In their recently updated book Genome, ¹² J.E. Bishop and M. Waldholz recreate the discovery of APO A-I and APO C-III in 1982. Apparently, Dr. J. Breslow had in his hands an article in the New England Journal of Medicine ¹³ that gave credibility to the theory of synchronicity. Just at that time, he and his colleagues realized that APO A-I and APO C-III were located next to each other on the same chromosome. That article linked the 2 protein products in a case involving 2 sisters who lacked HDL cholesterol and had precocious coronary artery disease. In 1985, the APO-IV gene made its appearance just below the APO C-III gene. It turned out that APO E was related both to Alzheimer's disease and to ischemic heart disease ^14,15 and was located in 19q13.1, while at least 1 variant of APO-B was traced to chromosome 2 (2p24). Studies on the apolipoprotein genes and their protein products continued throughout the 1990s, and new types are likely to be disclosed in the future.

In 1994, Rockefeller University scientist Jeff Friedman isolated the mouse “obesity” gene, which codes for the protein leptin. This led to the subsequent conduction of a major study by Friedman's group of the genetic factors in obesity and Type II diabetes on the Micronesian island of Kosrae. ¹⁶ To carry out his study, which he conducted in collaboration with J. Breslow and M. Stoffel, Friedman analyzed blood obtained in 1994 for insulin, cholesterol, and other clinically relevant substances. The need for such geographically isolated populations was indeed an important factor in the conduction of genetic studies, and Iceland has become a rising star in the study of inherited genetic traits.

In Cracking the Genome, ¹⁷ K. Davies, founder and former editor of Nature Genetics, jokingly remarks that by the mid-1990s there arose an expectation that “one gene a week” would be discovered. Yet this bounty did not dampen the interest with which some genes were received by the medical community. In 1999, 3 teams of researchers simultaneously identified the Tangier gene, ¹⁸ whose protein product ABC1 immediately revealed a strong link to cholesterol metabolism. Later, heart specialists were thrilled to find that treating mice with a drug that boosted production of the Tangier protein inhibited cholesterol absorption. ¹⁹

Several Web sites, such as the National Library of Medicine's http://www.ncbi.nlm.nih.gov/disease/, provide information on the disease genes that have been mapped so far. That site does warn, however, that most of those diseases are associated with monogenic mutations (that is, mutations in a single gene). Disclosing the interlinked genetic and epigenetic changes of complex multifactorial diseases like atherosclerosis and hypertension will be entirely another matter, with linkage and association studies constituting a necessity. ²⁰

The Human Genome

Polymorphisms and Mapping.

The seeds for sequencing the entire string of human nucleotides had been planted. During the past decade, the mass media have introduced the public to events surrounding human genomic sequencing, as well as to discoveries of disease-associated genes. These discoveries on occasion took place amid frantic races to be first. Indeed, there was a rumor that whoever found a gene had the right to name it. If this is so, hepatic lipase, fibrinogen, platelet-derived growth factor, and intercellular adhesion molecule-1 gene exhibit a certain lack of creativity, for which ACE (angiotensin-converting enzyme) is small compensation. ²¹

It is difficult to ascertain who 1st had the idea to sequence, one by one, the 3 billion letters that make up human beings. Dr. Davies ¹⁷ offers insight into which people made the idea official, after it no doubt had crossed many scientific minds. Suffice it to say that the development of the automated ABI sequencer in the early part of the last decade was critical for converting that idea into a reality. In 1985, less than 20 years ago, a laboratory could sequence a mere 50 to 100 bases a day, at $10 apiece. That same year, Gennari and Fischer ²² studied the cardiovascular action of calcitonin-gene-related peptide (CGRP) in human beings by examining the effects of intravenous administration. At the rate then attainable, sequencing the genome in any reasonable amount of time was just not possible, especially when one considers that sequencing had to be repeated several times in order to yield reliable results. In the 1990s, the new automated analyzers, together with bold new sequencing strategies, enabled the genomes of small organisms to be sequenced and published in record time; the 1st genome to be sequenced was that of the influenza virus, in 1995. ²³ The genomes of other organisms, such as yeast (Saccharomyces cerevisiae), nematodes (Caenorhabditis elegans), and drosophila, soon followed. ^24–26

Because the human sequence comprises not only protein-encoding regions, but also introns (“junk” DNA), a team of investigators led by C. Venter proposed 10 years ago to first sequence clones from cDNA libraries (that is, expressed sequence tags, or ESTs), in an attempt to attain biologically relevant information as soon as possible. ²⁷ The majority of these mapped ESTs encoded proteins of unknown function. This insight might have an impact in cardiology as well as in other medical disciplines, because associations between variants in “anonymous” genes can also be tested in case control studies. In 1997, C. Liew and associates ²⁸ analyzed about 76,000 ESTs from different cDNA libraries of the cardiovascular system. So far, however, more information has been gathered from case-control studies that required prior knowledge of the gene or its protein product. For example, variants within renin, angiotensinogen, and nitric oxide synthase were tested for their association with hypertension. ²⁹ The renin-angiotensin system fast became the object of increased attention in order to identify genomic risk factors in hypertension. ³⁰ Only 14 years ago, in 1988, these genetic factors were considered “emerging issues” in the cellular biology of the heart, as insulin, angiotensin II, and other vasoactive factors began to be recognized and understood. ³¹

Meanwhile, systematic sequencing of the human genome was revealing the existence of single nucleotide polymorphisms (SNPs, pronounced “snips”) that accounted for differences among the genomic DNA of different individuals. ³² While a SNP is not always associated with disease (indeed the human genome sequence must be considered a reference sequence), specific SNPs were increasingly linked to specific manifestations of disease. What was the point at which a SNP was no longer a SNP but a disease-producing mutation? Although SNPs appeared to play an important role in the development of disease, ³³ scientists hastened to warn against oversimplification. For example, in 1998 D. Nickerson and colleagues ³⁴ published their findings on the DNA sequence diversity in a 9.7 kb region of the human lipoprotein lipase gene, an extremely important cardiovascular-disease-associated gene. The implications of this study were that if other human genes exhibited these same polymorphic characteristics, this pattern of sequence variation would present challenges in associating genes with their corresponding risk factors. Also in 1998, J. Kuivenhoven and coworkers ³⁵ published a study on the role of a common variant of the cholesteryl ester transfer protein gene in the progression of coronary atherosclerosis. It appears that a variant in CETP resulted in decreased plasma HDL levels and increased VLDL–LDL/HDL ratio, with an increased rate of progression to atherosclerosis. ³⁵

Thus, the quest for candidate disease genes, such as the atrial fibrillation gene, ³⁶ was on. While everybody was busy trying to find a gene, there was enormous uncertainty about the actual number of genes that constituted the human genome sequence. The one trend of which everybody was aware was that the estimated number was steadily decreasing. As early as 1994, the question was seriously launched in Nature Genetics: “How many genes in the human genome?” ³⁷ For most of the decade, researchers expected between 100,000 and 150,000. To make these estimates, they used the fact that the 3′ UTR (untranslated region) in mRNA has the greatest variation between similar but not identical genes. Few people were prepared for the announcement, in February 2001, that the actual number might be close to 30,000, with some investigators mentioning an astoundingly low 26,000 genes in 3 billion nucleotides. ³⁸ Since then, however, the gene roller-coaster ride has gone up again, and new publications are expected to report estimates of 40,000 genes or above.

Gene Therapy.

Whatever the final number turns out to be, the prospect of gene therapy in its various forms already has filled a great number of journal pages. One of the most spectacular reports, which might become truly historic, was that published by Mason and associates, who used gene transfer in utero to biologically engineer a patent ductus arteriosus in lambs. ³⁹ Progress in the field was sufficient to warrant the publication, in 1999, of at least 1 book on the molecular basis of cardiovascular disease, including information on gene transfer approaches to treatment. ⁴⁰ Expectations of a future for the field of gene therapy turned out to be warranted when, on 28 April 2000, Science published an article reporting the 1st truly successful treatment of human severe combined immunodeficiency–X1 (SCID-X1) with a retrovirus-based gene therapy. ⁴¹

In 1999, J.M. Isner, who until his recent death was a leader in biorevascularization with gene transfer PHVEGF 165, published an editorial in Circulation titled “Cancer and Atherosclerosis: The Broad Mandate of Angiogenesis.” ⁴² Judah Folkman had already introduced the notion that tumors were angiogenesis dependent, ⁴³ but considerably less was known about this subject in the cardiovascular field. In 1981, living tissue that had formed in vitro was accepted as “real,” ⁴⁴ and 5 years later a blood vessel model was constructed from collagen and cultured vascular cells. ⁴⁵ The literature that surrounds vascular endothelial growth factor (VEGF) could be the subject of a history in itself, ⁴⁶ and the statins were also attaining well-deserved fame. In 2000, simvastatin was used for the revascularization of ischemic limbs. ⁴⁷

The History of High-Throughput Genome Screening Tools

The fact that so much genomic information was available, coupled with the fact that interest was increasingly focused on polygenic diseases and signal pathways, prompted the development of high-throughput genome screening tools. Among these, perhaps the one that has attained the most recognition is the biochip. There is a certain amount of evidence that Stephen Fodor invented the biochip around 1988. After a postdoctoral fellowship at Berkeley, he worked at Affymax Research Institute until, in 1993, he and a few colleagues founded Affymetrix. Affymetrix pioneered the field of DNA chips, but more and more institutions tried to produce their own in-house microarrays (or chips) in order to customize them while reducing their cost and controlling their quality. ⁴⁸ These in-house microarrays were not identical to the original in-silico synthesized chips produced by Affymetrix. Minor differences notwithstanding, 2 kinds of biochips were conceived and developed during the 1990s: the DNA microarray and, more recently, the protein microarray.

Even though DNA arrays were already used in 1991 for hybridization studies, ⁴⁹ the true birth of DNA microarrays as genomic medicine tools took place in 1996 with their use to analyze gene expression patterns in human cancer. ⁵⁰ The main idea behind DNA chips was that thousands of transcripts could be assayed in a single experiment, giving quantitative and qualitative information about the genes expressed in a given cell or tissue at a given time. Furthermore, they enabled comparison of 2 different transcripts in the same experiment, from which gene expression profiles of normal versus diseased state could be obtained by calculating ratios of expression. ⁵¹ The idea behind a DNA chip was simple (once someone had thought of it): thousands of DNA fragments corresponding to genes or ESTs can be robotically deposited on a solid support, such as a membrane or a microscope's glass slide. Next, the researcher reverse-transcribes mRNA from a normal cell population incorporating a given label, for example a fluorophore like Cy5; then he or she does the same with mRNA from a diseased cell population using a different label, such as Cy3. Both labeled cDNAs are applied to the DNA microarray and allowed to hybridize. Later, the intensity of the signal produced by the fluorescent labels when excited at a given wavelength by a laser scanner is detected and quantified. This intensity is assumed to correspond to the relative level of expression of that gene in the cells studied. In other words, a genetic profile is obtained.

This DNA microarray technology quickly yielded quite impressive results in cancer research, with applications that range from molecular classification of cancers to the identification of differentially expressed transcripts, prospective markers, and drug targets. ⁵² Significantly fewer publications were produced in the field of cardiology. Only recently was the 1st large DNA microarray used to analyze the heart transcriptome. In February 2001, a publication by Barrans and coworkers ⁵³ presented the results of an experiment that used the largest “CardioChip” to date, containing 10,368 redundant and randomly selected expressed sequence tags. In a total of 38 transcripts, expression pattern differences were found between human fetal, adult, and hypertrophic heart samples; for example, certain cytoskeletal proteins, novel clones, and ESTs provided a “portrait of gene expression alterations in hypertrophic cardiomyopathy” ⁵³ (Fig. 1). In June 2001, another scientific publication reported a study, using DNA microarrays, of the divergent transcriptional responses to independent genetic causes of cardiac hypertrophy. ⁵⁴

Fig. 1 Barrans, Stamatiou, and Liew ⁵³ produced the 1st large DNA microarray (CardioChip) with 10,368 redundant and randomly selected expressed sequence tags.

(From: Barrans DJ, Stamatiou D, Liew C. Construction of a human cardiovascular cDNA microarray: portrait of the failing heart. Biochem Biophys Res Commun 2001;280:964–9. Reproduced by permission of the authors and publisher. Copyright ©2001 Academic Press, Harcourt Publishing Division.)

Another historic whole-genome screening tool emerged in the mid-1990s. In 1995, Velculescu and colleagues ⁵⁵ from the Johns Hopkins School of Medicine published a paper in Science describing a new method for high-throughput screening of the human genome that they called SAGE (serial analysis of gene expression). This method relied on the expectation that a 9- to 10-nucleotide sequence, a “tag,” would contain enough information to uniquely identify a gene. This technology required generation and subsequent sequencing of gene tag concatamers, on the assumption that the frequency with which a sequence appeared in the SAGE library would be related to the abundance of a particular transcript. As in all newly developed technologies, problems arose in its application; in the case of SAGE, these problems were (and are) mainly related to the extensive and labor-intensive sequencing required and to the fact that, in practice, only 10% to 20% of tags represent unique genes. In 2001, 6 years after the publication of Velculescu's paper, Anisimov and associates ⁵⁶ issued the first of 2 reviews that presented the initial, limited SAGE analysis of rodent heart gene expression and anticipated the manner in which SAGE data could be applied to the study and treatment of cardiovascular disease. Further, SAGE has been used to elucidate the functions of kringle proteins, which are strongly expressed in lung, heart, and skeletal muscle. ⁵⁷

On Men and Mice

The development and honing of tools for recombinant DNA technology has been nothing short of overwhelming in the last few decades. Each of the developed techniques—from PCR (the Polymerase Chain Reaction) to Southern, Northern, and Western blotting, to FISH (fluorescence in situ hybridization) and dozens of others—has contributed to science in a way that is difficult to summarize in a book, let alone an article.

From the perspective of medical history, one sees that physicians and scientists do not become truly excited about a new technology until it permits the understanding of phenomena in vivo. Two developments, micro-injection and transgenic/knockout mice, constituted such an event. Micro-injection permitted the transfer of genetic material in vivo, and the resulting transgenic animals have afforded a unique way of observing the real-life effects of such an alteration. Because virtually all human genes have mouse counterparts, and because the sequence of the mouse genome was almost finished, mice have been the transgenic animals most commonly used.

In 1981, a mere 2 decades ago, the 1st transgenic animal was born: it was a mouse with cDNA for human interferon developed in the laboratory of Dr. Frank Ruddle at Yale University. ⁵⁸ Since then, patent and animal-rights wars aside, thousands of design mice have been bred for a variety of diseases, including cardiovascular diseases. The technology involves either inserting a new gene in the mouse (a “transgene”), or mutating a target gene, which produces its functional knockout. ⁵⁹

One of the fields that has been revolutionized by the advent of cardiovascular genomics and by the parade of fancy design mice has been the study of atherogenesis and of plaque vulnerable to rupture. The molecular basis of atherogenesis has been reviewed in recent publications by Dr. A. Lusis ⁶⁰ and by Drs. R. Lefkowitz and J. Willerson, ⁶¹ as well as in a historical overview in Circulation. ⁶²

According to the ancient physician Alcmaeon of Croton, health requires the maintenance of a proper balance between heat and cold. ⁶³ Many centuries later, James Willerson and Ward Casscells discovered that atherosclerotic plaque exhibits heterogeneity of temperature. ⁶⁴ The study of atherosclerosis exemplifies the advances made possible by the use of transgenic mice. In 1992, the molecular mechanisms involved in atherogenesis began to be disclosed with the introduction of the 1st transgenic and knockout mice. ⁶⁵ That year, J.L. Breslow reportedly was the 1st scientist to use recombinant DNA technology and transgenic animals to study atherosclerosis. As early as 1994, 6 genes highly related to apolipoprotein (a) had been identified. ⁶⁶ Moreover, it was found in 1994 that atherosclerosis in APO E-deficient mice was associated with increased lipid peroxidation of plasma, LDL, and VLDL; further, these lipoproteins were more susceptible to lipid peroxidation under oxidative stress. ⁶⁷ Consequently, APO E-deficient mice revealed themselves to be extremely atherogenic animals.

Subsequent to the identification of various apolipoproteins and several variants of the apolipoprotein E, APO E knockout mice were bred and used to observe 2 of the 3 stages in atherogenesis: foam cells and fibrous plaques. ⁶⁸ In 1996, a team of researchers used genetically modified mice to study apolipoprotein B variants (APO B-48 and APO B-100) ⁶⁹ that are essential to mammalian lipoprotein metabolism. In December 2000, a paper published in Life Sciences claimed that the continuous administration of aspirin attenuated atherosclerosis in APO E-deficient mice. ⁷⁰ This was not surprising, in view of the known role that inflammation played in the initiation and progression of atherosclerotic lesions. Accordingly, A. Paul and colleagues found that aspirin slowed carotid plaque growth in a dose-dependent manner. ⁷⁰

Atherosclerosis-prone transgenic mice are but 1 example. Hundreds of design knockout mice have been born in the last 15 or so years: anxiety-prone mice, obese mice, and all sorts of mice. In the last 5 years, lines of knockout mice for α 2 A, α 2 B, and α 2 C, and for β₁- and β₂-adrenergic receptors, have been developed in order to study the role of the sympathetic nervous system in the pathogenesis of heart failure. ⁷¹ It is important to mention that polymorphisms in the β₂-adrenergic receptor have been associated with hypertension. ⁷² The rationale behind these experiments has been that patients with reduced cardiac function have a compensatory increase in sympathetic nervous system activity.

Transgenic mice have also been used (for example) to study stunned myocardium. Using mice that expressed the major proteolytic degradation fragment of troponin I, H. Kogler and colleagues ⁷³ found support for the proposition that calcium regulatory pathways and β-adrenergic signal transduction remain intact in isolated trabeculae from the transgenic animals.

Although it might not be possible to condense the history of the recombinant DNA technology that increased our knowledge of cardiovascular genomics, there is an important development that we need to mention. In 2000, an article in Nature Genetics ⁷⁴ dealt with the inability to achieve stable transmission of exogenous mitochondrial DNA into mice. This inability posed a problem because mitochondrial DNA defects are associated with disease. Inoue and colleagues ⁷⁴ were able to transmit mutant mtDNA maternally into mice, most of which died of renal failure.

The History of Personalized Medicine

Agatha Christie is said to have sold more books in the English language than any other writer. Also an excellent playwright, more than an amateur archaeologist, an opera singer, and a musician, she happened to work as a pharmacist's assistant in a hospital dispensary during World War I. ⁷⁵ Having witnessed therapeutic procedures, she described a scenario that might sound familiar:

The ignorant layman, or laywoman, as I suppose I ought to call myself, believes that the doctor studies your case individually, considers what drugs would be best for it, and writes a prescription to that effect. I soon found that the tonic prescribed by Dr. Whittick, and the tonic prescribed by Dr. Vyner were all quite different, and particular, not to the patient, but to the doctor. ⁷⁵

Christie would no doubt have been glad to live long enough to witness the advent of pharmacogenomics and personalized medicine. As gene expression profiles developed the potential to show the molecular basis for resistance or sensitivity, and demonstrated that they could aid in the establishment of personalized therapies that would elicit optimal responses from patients, patient-tailored therapies became le cri de guerre. This was increasingly coupled with a multifactorial approach to treatment, in order to focus upon the various steps of the cascade of pathophysiologic events ⁷⁶ leading to a particular state of disease.

Here are some examples in the recent history of pharmacogenomics. In 1998, J. Kuivenhoven and colleagues ³⁵ found that patients who presented with a variant of CETP (cholesterol ester transfer protein) responded better to prevastatin. In June 2001, researchers ⁷⁷ demonstrated that failure to suppress aldosterone, in spite of the administration of angiotensin-converting enzyme (ACE) inhibitor, was associated with the ACE DD genotype in chronic heart failure.

Toward the Cardiome

One of the positive effects of the advent of genomics was that it forced a collaboration among engineers, computer scientists, mathematicians, statisticians, and life-science professionals. An anticipated side effect of this collaboration was that the rate of development and refinement of new genomic and proteomic tools grew markedly. New materials, sensor surfaces, miniaturization, robotics, micro fluidics, and data analysis tools and software all came into play. An unanticipated side-effect was that the practitioners of each discipline had to learn at least the essential elements of their collaborators' languages.

One of the 1st needs that arose during the sequencing of the human genome, as well as during genomic and proteomic studies, was the necessity not only to store huge amounts of data, but also to annotate and interpret them. The ever-increasing number of rapidly growing databases forced the resuscitation and development of the science of bioinformatics. “Resuscitation” is prompted by J.B. Hagen's year 2000 review, ⁷⁸ which brought to attention the fact that bioinformatics, or some form of it, existed in the early 1960s. Despite the fact that DNA sequencing did not become feasible until the 1970s, protein biochemistry data were accumulating rapidly a decade earlier. ⁷⁹

Rapidly growing genomic data and computational biology gave researchers the next set of ideas for the cardiovascular system. If we review the history of our vision of the heart, we can see an evolution from the tangible toward the abstract, from the sensory toward the mathematical and computational. Accordingly, the physiome became an abstract model of the body, and the cardiome an abstract model of the heart, attempting to reproduce faithfully what the heart was and did—with the ultimate goal of enabling predictions about health and disease, and of playing a major role in preventive health care through patient-specific cardiome models.

Large-scale projects were launched in the mid 1990s to model the physiome both at the commercial (Physiome Sciences Inc.) and the academic (www.physiome.org) levels. In collaboration with other scientists, Raimond Winslow, founder of Physiome Sciences Inc., used the vast amount of biological, biochemical, and biophysical data accumulated over the preceding 30 years to create a virtual heart (www.bme.jhu.edu/∼rwinslow/recent/imgmvpvt.html/). Denis Noble (Oxford University), one of the main contributors to the project and also a member of Physiome Sciences, was among the 1st to develop mathematical models of the heart's electrical activity. R.F. Service ⁷⁹ published in Science a review of the computer simulation of heart failure that Winslow's virtual heart can perform. Computer simulations of the heart enable the prediction of what would happen should the production of proteins that help control the concentration of calcium ions in cells be altered.

In 1997, J.B. Bassingthwaighte ⁸⁰ declared the physiome project to be too ambitious and proposed a more “modest and achievable” goal: the cardiome. The initial goal of the project was to develop databases with genomic, biochemical, anatomic, and physiologic information. By 1999, these researchers ⁸¹ were already talking about a 3-dimensional computational model of excitation, metabolism, and contraction of the heart that should be available in 2002. The importance of these inventions is that they permit computational modeling with the ultimate goal of predicting behaviors.

Perhaps one of the most interesting aspects of these types of studies is the potential to discover elusive links. Heinrich Taegtmeyer, for example, has advanced the hypothesis that “energy substrate metabolism constitutes the essential link between cardiac gene expression on the one hand, and contractile function on the other hand.” ⁸²

The Nascent History of Proteomics

If the number of genes is 30,000, as tentatively set early in 2001, the number of human proteins must be at least 1 or 2 orders of magnitude greater, due to differently spliced gene products and post-translational modifications. Proteomic techniques, such as 2-dimensional (2-D) electrophoresis for the resolution and quantitation of complex protein mixtures, had been in use for years: to offer an example, a human myocardial 2-D electrophoresis protein database (http://userpage.chemie.fu-berlin.de/∼pleiss/) was made available in 1995. During the past few years, however, we have witnessed an exciting introduction to the world of protein arrays. In January 2001, Dr. Brian Haab and colleagues ⁸³ published their work on the application of a well established DNA microarray technology, in order to produce a protein microarray that is based on antigen-antibody interactions (Fig. 2). This publication was one of the very few that suggested that genomic technologies could potentially be adapted to proteomics.

Fig. 2 Protein chips are the next step. Brian Haab and colleagues ⁸³ succeeded in applying a variation of DNA microarray technology to proteins.

(From: Haab BB, Dunham MJ, Brown PO. Protein microarrays for highly parallel detection and quantitation of specific proteins and antibodies in complex solutions. Genome Biol 2001;2:RESEARCH0004. Reproduced by permission of the authors. Copyright ©2001.)

The fascination that the new science of proteomics is exerting on the scientific community has raised some alarm about its prematurity, for the study of the genome is far from finished. However, the impact that the development of new proteomic technologies will have on cardiovascular disease is likely to be even greater than that of vascular genomics. For example, a recent report in Science ⁸⁴ describes arrays of proteins with 100 to 300 nanometer features and reports undetectable nonspecific binding.

Concluding Remarks

The attempt to navigate our way through a maze of new and still conflicting molecular genetic information reminds us that the term cardiovascular disease itself might undergo redefinition in the not-so-distant future. Already genetic profiles are the basis for new molecular classifications of cancer, the other major cause of death in the United States. ⁸⁵ As of 2001, the World Health Organization classified cardiovascular disease into 11 groups for assembling mortality data, ⁸⁶ but there is a fair chance that new reclassifications and subclassifications will occur as genetic similarities among different manifestations of the disease are unveiled.

As the history of science, indeed of cardiology, moves towards a greater level of abstraction and towards the integration of many disciplines, the cardiome is going to become a mathematical and computational model. After Jean Baptiste de Sénac's publication of the 1st comprehensive textbook of cardiology in 1749, ⁸⁷ Leopold Auenbrugger made the observation (1761) ⁸⁸ that the condition of the heart could be estimated by the sound returned from tapping on the chest (percussion). Auenbrugger's procedure was subsequently refined by R.T.H. Laennec's introduction of mediate auscultation with the stethoscope (1819). ^{89, 90} Although it has become increasingly difficult to rely on sensory perception for diagnosis, one needs to remember that, over the long term, the highest places in the history of cardiology are reserved for compassionate, intuitive, and trustworthy physicians.

Acknowledgments

Wei Zhang, PhD, Director of the Cancer Genomics Core Laboratory at MD Anderson Cancer Center (Houston), introduced the author to the study of genomics and proteomics. Enrique Lahoz-Echeverría, MS (mathematics), MS (physics), provided a superb painting from his collection Matters of the Heart. Drs. Brian Haab and colleagues and Choong-Chin Liew and colleagues kindly granted permission to reproduce figures from their recent pioneering publications on protein arrays and CardioChips, respectively.

Figure. Frontispiece: Spanish scientist and artist enrique Lahoz-Echeverría conceives cardiovascular genomics as deep red branches of the sick heart inside a double helix, reaching out into a blue sea of hope.

(From the collection Matters of the heart, Granada, Spain, 2001. Reproduced by permission of the artist. Copyright ©2002 Enrique Lahoz-Echeverría.)

Glossary

biochip: A surface, usually miniaturized, by means of which large amounts of biological information can be obtained upon following an established assay protocol.

bioinformatics: The science of informatics applied to biology. For example, the generation, storage, and annotation of massive quantities of data obtained via the sequencing of genomes. Also, the development of mathematical algorithms that enable the meaningful analysis of such data.

CardioChip: A DNA microarray or biochip that focuses on obtaining gene expression profiles of normal versus diseased heart tissue, in an attempt to identify differences and thus targets of study.

cardiome: The ensemble of genes expressed in heart tissue.

concatamer: A number of DNA molecules or sequences of nucleotides that are covalently linked in series.

differentially spliced protein: DNA is transcribed into mRNA, which is then translated to yield a protein or gene product; mRNA can be processed (spliced) in several different ways to yield different protein products. In addition, proteins can undergo further (post-translational) modifications, such as the addition of chemical groups (phosphorylation, glycosylation, etc.). These facts, among others, account for the expectation that the proteome exhibits a much greater degree of complexity than the genome.

epigenetic: This is a term applied to those heritable changes in gene function that cannot be explained by a change in the DNA sequence alone.

expressed sequence tag (EST): A set of sequenced cDNAs from an mRNA population that can be obtained, for example, from a specific cell line. The fact that they are “expressed” means that they made it to the mRNA stage (that is, they were transcribed because they are not junk DNA).

genomics: The study of the human genome, particularly (from our point of view) in regard to its implications for health and disease.

microarray: Similar to biochip. The term array means that features on the chip (for example, ESTs) are present at fixed and previously known positions. This is essential because information derived from that position will be used after the assay to gather relevant biological information. The term micro is used because the chips are miniaturized: it is possible to assay thousands of ESTs or cDNA clones on the surface of a microscope's glass slide.

pharmacogenomics: The study of how genomic information can be used to maximize the therapeutic potential of drugs.

physiome: The integrated study of the organism as a whole, taking into account anatomic, biochemical, genomic, and proteomic data.

proteomics: The high-throughput study of proteins present in cells and tissues.

transcriptome: The set of genes that has been transcribed (that is, has reached the mRNA stage) in a given cell population.