1 Deputy Editor, Physiological Genomics
2 Department of Medicine, Harvard Medical School, Brigham and Womens Hospital, Boston, Massachusetts 02115
ONE OF THE EXCITING PROMISES of functional genomics is the ability to characterize specific developmental and disease states via expression profiling. It has become increasingly common for physiology researchers to employ microarrays to characterize differentially expressed transcripts. These research techniques are providing scientists with the opportunity to obtain novel information on molecular pathways and to test fundamental hypotheses of complex diseases such as cancer and heart disease. It is expected that as techniques are further refined, a comprehensive picture of the basic genetic programs underlying complex disease will emerge. One potentially powerful application of these techniques is to compare gene expression in multiple models (e.g., transgenics, knockouts, surgical, or drug-induced) of complex diseases. In theory, differences in genetic programs between models may reflect effects specific to a given model, while similarities may reveal fundamental common pathways.
In myocardial hypertrophy, a variety of mechanisms, including genetic and environmental factors, lead to an increase in the mass of myocardium. Surgical models, such as coronary artery ligation, pressure overload via aortic banding, and volume overload via A-V fistula, have been used for decades to study the classic and molecular physiology of hypertrophy. In addition, researchers have recently developed a number of transgenic mouse models in which specific genes implicated in hypertrophy are deleted or overexpressed under the control of the ß-myosin heavy chain promoter. Genetic manipulation of the expression of one or more of these genes modifies distinct pathophysiological pathways, yielding the common phenotypic outcome of hypertrophy, which has led workers in the field to hypothesize a common pool of transcripts responsible for the shared manifestations of hypertrophy.
In this release, Aronow et al. (Ref. 1; see page 19 in this release) report their microarray analysis of differential gene expression in four such transgenic lines. One hypothesis they tested was that there would be a group of shared transcripts that would be differentially regulated with respect to the control animals. Interestingly, their results demonstrate that a single, broad model may be inadequate to describe the molecular changes characteristic of each subset of this syndrome.
Specifically, the authors (1) compared differential levels of mRNA expression between four transgenic strains and their wild-type counterparts. The strains studied included three pathological models, in which the genes Gq, calcineurin (CN), and calsequestrin (CSQ) were overexpressed, as well as one transgenic model (
RACK) in which a synthetic octapeptide activates protein kinase C-
to bring about hypertrophy. Cardiac tissue was collected at a time when the heart was in a nonfailing state, as defined by a lack of pulmonary congestion. The authors then utilized Incyte mouse spotted cDNA microarrays to compare transcript levels in each of these lines, compared to their nontransgenic siblings.
Within each model, differentially expressed genes were defined as those that were at least 1.7-fold greater in either the recombinant mice or the control mice. Not surprisingly, using these criteria, it was shown that the more severe models (Gq, CN, and CSQ mice) exhibited greater numbers of differentially regulated genes than did the less severe model (
RACK). Their most striking finding was that there were no differentially expressed genes in common among the four models of cardiac hypertrophy; indeed, only atrial natriuretic peptide (ANP) was coregulated in the three pathological models (when using the 1.7-fold up- or downregulated criteria). Aronow et al. then performed clustering analysis, a more sensitive approach to defining patterns of expression, and were able to identify clusters of genes that appeared to correlate with the degree of severity of pathological hypertrophy across the different models.
Do these results mean that a conserved hypertrophic gene program does not exist? Not necessarily. While the results of \E Aronow et al. are important and, in fact, intriguing, there are several caveats that need to be considered. One must always interpret results from analysis of transgenic and knockout mice with a certain degree of caution. Abnormal upregulation of a protein from birth may induce compensatory changes that are specific for the pathway being manipulated and may bear little resemblance to clinical disease. Furthermore, to better define a common disease profile across multiple models, especially in a disease that is progressive, the disease state across these models needs to be appropriately phenotyped and staged. Comparing one model of early disease with a second model of more advanced disease might not be expected to yield the same profiles. Indeed, Aronow et al. showed that clusters existed that correlate well with disease severity. It would be informative to examine how the profiles observed in the different models compare during the progression of disease. Although the single gene perturbations studied do, in fact, produce hypertrophy, they are sufficiently different from physiological causes of hypertrophy that it is perhaps not surprising that their expression profiles have little in common with each other.
These caveats aside, the results from this study reemphasize the challenges of using comparative genomics to study complex, polygenic diseases with multiple environmental influences. Regardless of the initiating cause of the disease and regardless of whether a single gene alteration will predict the response to a more global stimulus such as pressure overload or myocardial infarction, it is remarkable that a cluster of genes that would define hypertrophy (as opposed to causing hypertrophy) was not found. Do these results mean that convergent pathological symptoms may not necessarily indicate a single fundamental genetic program? Does this emphasize the point that "hypertrophy" may be too general a term and that more specific definitions are required? Could profiling be used to define different classes or subsets of hypertrophy more specifically? Do all forms of cardiac hypertrophy follow the same course? Indeed, genotyping studies of human hypertrophic cardiomyopathy have documented that specific mutations of sarcomeric proteins can predict different clinical outcomes, suggesting the existence of divergent genetic/pathophysiological pathways, despite a comparable phenotype (2). As Aronow et al. note, this sets the stage for a second look at the inherent differences in models of cardiac hypertrophy.
FOOTNOTES
Address for reprint requests and other correspondence: S. B. Glueck, Physiological Genomics, Thorn 1324C, Brigham and Womens Hospital, 20 Shattuck St., Boston, MA 02115 (E-mail: sglueck{at}rics.bwh.harvard.edu).
REFERENCES
|
HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Visit Other APS Journals Online |