At A Glance
- Anthony Rosenzweig, MD
- Cardiovascular Researcher, Massachusetts General Hospital since 1994.
- Leads a team that uses somatic gene transfer and other methods to study the pathophysiology of numerous cardiovascular diseases, including atherosclerosis, transplant arteriosclerosis, and reperfusion injury.
- Predoctoral work with molecular genetics researchers Jonathan and Christine Seideman, who are at Harvard Medical School.
- MD Harvard Medical School
- Author of numerous articles on gene therapy and heart disease, as well as a new review article “DNA Microarrays: Implications for Cardiovascular Medicine,” published in the October 4 issue of the journal Circulation Research (91(7):559-64.)
While cancer researchers have seized upon the microarray as a 21st century tool for probing molecular patterns, this move has surprisingly not been followed in an area of research on a more common and deadlier killer than cancer: heart disease. But a few pioneers, such as Massachusetts General Hospital cardiology researcher Anthony Rosenzweig, have begun to explore the possibilities that microarrays may offer in providing them insights into the often mysterious etiology and outcomes of heart disease. In October, Rosenzweig and his colleague, Stuart Cook, published an article in the journal Circulation Research introducing microarrays to the cardiology world. Recently, BioArray News discussed with Rosenzweig this article, and the potential of microarrays for cardiological research.
Your microarray article in Circulation Research seems to be aimed at the generalist who is just trying to understand microarrays. What level of interest do you see among cardiology/cardiovascular medicine researchers in microarrays?
I think there’s a tremendous amount of interest among those who do basic cardiovascular research. There’s much less knowledge among the clinicians. Most don’t know what the microarray is. It hasn’t really infiltrated into the clinical arena. But I think there is a growing interest. What’s driving it is what’s driving [microarray research] in the cancer field — we might be able to identify clinical markers, or a phenotype that might give us additional information about diagnosis and prognosis or response to therapy. Right now in cardiology, we basically treat all heart failure — it is the No. 1 diagnosis of people over 65 admitted to the hospital — as the same thing, but we suspect that all these [cases] are composites of many different conditions. If we could get a better idea of what is going on, we could target therapy or at least know a little bit more about prognosis.
[Also], if we could get a handle on the underlying biology, it could be useful from a clinical perspective. In terms of understanding the basic pathology, microarrays could help us to pick up unanticipated connections between a pathway and phenotype.
Given that heart disease is even more common than cancer, why do you think that microarray gene expression studies have been so focused on cancers?
Almost always, with cancer you have tissue [because you do biopsies.] And I once heard [NIH director] Harold Varmus saying that all cancer is genetic. One step up, is that you look at transcripts that are going to be altered [in disease]. In contrast, in heart disease we don’t have tissue, in part because we don’t do biopsies [except in transplant cases] and you just get a small amount of tissue. It’s also an open question whether gene expression studies in heart disease are going to be as successful [as in cancer]. Because in contrast to cancer, it is less likely that you expect to see transcriptional changes in heart disease. [Given] posttranslational modifications of proteins, you might have to wait for a good way to do proteomics to get at these sorts of questions. But a lot of us are speculating that [microarrays] will be successful in heart disease because there’s a big genetic component to heart disease.
A group that published a recent article in the New England Journal of Medicine did RT-PCR to look at biopsies of people with heart failure. The researchers found that the expression of a key gene at the transcript level was significantly altered in heart failure. Interestingly, people who responded to standard therapy showed improvement in at lease a few of these genes, back toward the normal patterns.
So if you can’t always get heart tissues easily, how do you do microarray experiments in cardiology?
We spend a lot of time working in animal models. We have transgenic mice that are manipulated by somatic gene transfer in surgical models of heart disease. But in terms of access to heat tissue, a lot of people have made use of the heart failure population that comes to transplantation. They may not be representative of the entire population that gets [heart] disease, but they have tissues, and here and at other institutions, [researchers] have begun to amass in tissue banks [tissues] that could be used for that kind of study.
The other thing we’re particularly interested in, is that the heart is bathed in the blood that circulates throughout the body, and changes in the heart are mirrored in changes in the blood.
What specific types of study you see being important in microarray studies on heart tissue and blood?
In cardiac transplantation, people get nine or 10 biopsies in the first year [after the transplant]. It would be nice to spare people of this, so one of the things we are looking at is, are there predictors that would show up in the peripheral blood to show patterns that would [indicate] rejection? People have thought of this before, but until microarrays, people took their favorite gene [as a marker] and got variable and unimpressive results. But the hope is that if you can screen 22,000 genes, as on the current Affymetrix chips, you might find a gene that is unanticipated but might be a good marker. We are looking at the peripheral blood post-biopsy. If we look at the ones with signs of rejections, we will compare them to ones that don’t have rejection and will look at the serum samples over time to identify these peripheral markers of rejection.
Another study we’re doing is looking at transcript markers in patients who present to the cath lab with unstable angina. We compare them to those who come in for a transplant. In those who have unstable angina, a process of inflammation is though to contribute to the acute occlusion of the arteries. Our question is whether we can pick up the marker in the peripheral blood.
Wouldn’t protein microarrays be a potentially better tool in studies like this than DNA arrays — since, as you said earlier, the proteins present in the heart and blood could be detected directly, rather than through associated RNA transcript levels?
Most of us are intrigued by protein chips because those chips are closer to the effectors in the cells. Obviously, there are technical issues in the protein chips. Additionally, talking to friends who are doing proteomics in serum, there is a problem with very abundant proteins like albumin thin the serum and mess up your mass spec. But, in theory, you’d like to be able to screen on both levels. [When you are] able with protein chips to compare state A and state B, not in terms of just expression but also posttranslational modification, this will really be the holy grail of being able to get a snapshot of what’s going on in the patient.
In your own work, do you use Affymetrix or home-made cDNA, or oligo arrays?
We’ve used, almost exclusively, Affymetrix arrays. Early on we did some work with Incyte arrays, but virtually everything we’ve done has been through the Harvard Center for Genomic Research. And they have a research effort where people like Gavin MacBeath are looking at protein chips and more cutting edge technology, but also run standard assays as a core facility. From our perspective, there seemed to be a real advantage to going through people who focused on this rather than doing it in what we feared would be an amateurish way ourselves.
[The decision to] use Affymetrix partly fell into that, and partly [was because we] got good results with it. There is the notion that they had the perfect match and the mismatched oligo sets, and my understanding is that others have actually looked at this, and it’s not clear how much this adds to the analysis. Others have advocated not relying on the PM-MM because of nonspecific binding to the mismatch. There are some technical issues, but virtually everything that we picked up using the [Affymetrix] filtering array has been confirmed by RT-PCR. What we aren’t happy about is the price, which has improved a lot, but by the time you plan a well-controlled experiment with multiple time points, it turns out to be a bit intimidating. We are limited by the price and end up doing fewer replicates than we would like to do.