| COMMENTARIES |
Atherosclerosis a Network of Genes Driven By Environmental Pressures Filtered Through the Genetic Make-Up of the Individual
Johan Björkegren, M.D., Ph.D., Associate Professor of Molecular Medicine, Computational Medicine Unit, Department of Medicine, Center for Molecular Medicine, Karolinska Institutet, Karolinska University Hospital, Solna, 171 76 Stockholm, Tel: (+46) 733568181, Email: Johan.Bjorkegren@ki.se
In the genomics era, we are finally in a position to move from the assessment of individual candidate genes to the simultaneous assessment of all genes involved in the development of disease in organs and tissue [1]. New technologies developed in parallel with the Human Genome Project make it possible to capture close to all gene activities during the initiation and progression of disease. Traditional statistics can now be complemented with other analytical tools. In particular, systems biological tools can be used to infer networks or otherwise functionally associated genes in so-called modules from sets of whole-genome data and thereby determine their role in disease development [2].
Over the past century, atherosclerosis research has yielded important insights into genes and pathways that are central in atherogenesis [3]. One well-documented example is the LDL cholesterol pathway. The association between plasma levels of LDL cholesterol and the extent and rate of atherosclerosis development has been established in many studies, including LDL-lowering studies [4]. Changes in plasma LDL cholesterol levels seem to affect atherosclerosis development in both the early and very late stages, when clinical sign are manifest. With a systems biological approach to atherosclerosis, this kind of prior knowledge can be used to advantage. We recently identified a network of genes that responded to lowering of plasma cholesterol and prevented further progression of atherosclerosis [5].
Our understanding of how genes interact during the initiation and progression of atherosclerosis is most likely limited by bias inherent in hypothesis-driven approaches. Network-inference tools enable us to take a fresh look at the stages of atherosclerosis development without using prior knowledge. In fact, these tools allow us to view atherogenesis as a “black box.” Using silencing RNA experiments or a series of gene knockouts in animal and/or cell model systems, we can repeatedly perturb this black box. Then, by monitoring the reactions to different perturbations (e.g. by whole-genome measurements), responsive components (genes) and their interactions can be revealed. In this way, we can efficiently narrow down a network of thousands of genes by performing fewer than 100 perturbation experiments [6]. The generated networks and their wire diagrams serve as disease maps we can use to determine how established disease genes are related to genes previously unassociated with the disease in question [5].
By understanding molecular networks of atherosclerosis at the gene, protein, and metabolite levels, we can also address the question of how well-established atherosclerosis pathways, such as those governing inflammation, immune functions, and lipid accumulation, are co-regulated and thus how they influence each other. In this fashion, key regulators can be identified—some of which may prove to be suitable targets for therapeutic intervention. Networks can also be used to discover new pathways in atherosclerosis.
In addition, gene networks and modules will be important to improve our understanding of how genotype is coupled to phenotype [7]. Today, emphasis is given to large meta-analysis of association studies to uncover the genetics of coronary artery disease and other complex disorders. Two large association studies recently identified a locus on chromosome 9 that influences the risk of coronary artery disease/myocardial infarction by 31% [8,9]. Revealing the mechanism of this fascinating result is an important future challenge. However, large association studies are designed to reveal common genetic variants with little effect on risk of developing disease (10% to 50%). Introducing gene networks and/or modules as a middle step (genotype – networks – disease phenotype) will allow us to reveal combinations of rare variants that together will have a much larger effect on disease risk (several fold). This idea is built on the perception that the genetics of complex disorders such as atherosclerosis influence phenotypes differently, depending on the contexts of environmental pressures and ethnicity [10,11]. If so, important genetic variants can be identified in relation to specific contexts—groups of individuals who share genetic background and lifestyle (i.e. risk factors). A good example is the interaction between genetic variants and environment in rheumatoid arthritis [12]. Anti-CCP positive carriers of risk variants in HLA-DRB1 and PTPN22 have a > 20-fold higher risk of developing rheumatoid arthritis than nonsmoking carriers of only the HLA-DRB1 variant. Carriers of the HLA-DRB1variant who smoke have the same risk as carriers of both genetic markers. An indirect sign of the idea that the genetic makeup determines responses to environmental pressures (in this field, drug responses) in an individual fashion is the research field of pharmocogenomics. Statin-related myotoxicity serves as one example this [13].
We call for atherosclerosis research that moves away from studying individual atherosclerosis genes and pathways and instead seeks to identify regulatory gene networks that govern atherosclerosis development under the dynamic influences of genetics. Ideally, these networks will first be identified in humans, simply because humans carry the disease mechanisms and genetics of interest [14]. By collecting smaller but extremely well-characterized cohorts, including several whole-genome measurements per patient, it will be possible to identify groups of functionally associated genes (i.e. modules and networks), which in turn cause disease-relevant changes of phenotypes. Once the human networks have been established, their roles in disease development and their relations to established risk factors can be established in animal models of disease; the true biological architecture of the networks can be unraveled in relevant eukaryote cell model systems.
Focusing on gene networks instead of individual genes will eventually pave the way for the identification of markers and targets that can be used to develop more specific diagnostic tools and therapeutic regimens that are specific for the stage of disease and the genetic makeup of the individual. Such regimens would likely require particular combinations of drugs and dosages to maximize individual treatment effects and minimize side effects.
References