Department of Chemistry and Chemical Biology, Harvard University
Name: Gavin MacBeath
Title: Assistant Professor, Department of Chemistry and Chemical Biology, Harvard University
Professional Background: 2002-present, assistant professor, department of chemistry & chemical biology, Harvard University; 2002-2000, research fellow, Bauer Center for Genomics Research, Harvard University.
Education: 1997 — DPhil, macromolecular and cellular structure and chemistry, the Scripps Research Institute; 1991 — BS, genetics, University of Manitoba.
A paper published in the Nov. 6 issue of Nature details how a Harvard team used protein microarrays to give a genome-wide view of epidermal growth factor receptor signalling networks.
Entitled "A quantitative protein interaction network for the ErbB receptors using protein microarrays," the paper describes a quantitative protein interaction network and reveals greater insight into protein promiscuity, a biological activity often linked with some cancers.
To learn more about the results of this project as chronicled in the paper, BioArray News spoke with Gavin MacBeath, a co-author on the paper, last week.
Can you give me a quick history of the MacBeath lab and tell me what you are focused on at the moment?
Its existence really started in 2000. I originally started the lab as a research fellow at the Bauer Center for Genomics Research. This was a non-tenure track PI position at Harvard. I started the lab back in 2000, and on July 1, 2002, I got hired by the chemistry department and that's where I am now.
In terms of our interests, they are ultimately in systems biology, trying to understand how multiple proteins work together to control complex processes, particularly signal transduction processes. Protein microarray technology is certainly something that forms the backbone of the lab. But, in many senses it is only the tool we use to try to understand problems in systems biology.
There are two main areas of research in the lab. One is receptor tyrosine kinase-mediated signaling: using microarrays of SH2 and PTB domains in order to understand protein recruitment to receptor tyrosine kinases. This is what the work described in the Nature article is about. The other main effort in the lab is looking at PDZ domains and their role in synaptic signaling.
So what led you to study these activities?
It really started through our interest in receptor tyrosine kinase signaling. The first step in that process is recruitment of proteins to activated receptors. Protein recruitment is mediated by two domains that are known to bind to proteins that have been phosphorylated on tyrosine residues: SH2 and PTB domains. Since we are trying to get a broad, genome-wide perspective on these processes, we started by looking at the human genome to find all the SH2 and PTB domains. That way, we would have the full complement of every protein that gets recruited to these receptors when the receptors get activated.
What tools did you use to identify them?
In terms of identifying the domains, that was nothing special. This type of information is available on the web, so we used various existing databases that have identified these domains. In terms of getting them in hand, we went through the process ourselves of cloning, expressing, and purifying all of these domains.
We used high-throughput cloning methods in the lab to get our hands on the coding regions for these domains and then expressed them all in E. coli.
You said that there currently are drugs that target primary pathways rather than secondary pathways …
Yes, this focused in on one of the observations we made from the data in this paper. This paper focused on a single family of receptor tyrosine kinases; the ErbB receptors. There are four members of this family. What the data revealed was that these receptors differ in a particular property that really hasn't been described before: They differ in the extent to which they become more promiscuous when they are overexpressed.
What does that mean?
One of the key concepts in this paper is that we uncovered a quantitative protein interaction network. Many people have described protein interaction networks where they discover new protein-protein interactions. The difference with what we've done is that we didn't just discover interactions, we actually measured the strength of every single interaction. So we have dissociation constants for every binding event, which enabled us to look at just the high-affinity interactions and then see what happens as we add in the lower-affinity interactions. The high- affinity interactions are going to be important when the receptor is present at low levels. As the receptor concentration increases, the lower-affinity interactions also become important.
The paradigm in cancer right now is that a protein is an oncogene because, when you overexpress it or make it constitutively active, it does more of what it normally does. For example, EGFR normally signals for growth through certain pathways. If you overexpress it or make it constitutively active, it will continuously signal for growth. What we are proposing, based on the data we've collected, is that it is not just more signaling through the normal pathways, but the addition of new signaling pathways that makes EGFR and ErbB2 particularly oncogenic. As the proteins become more promiscuous, they recruit other proteins in addition to ones they normally recruit at lower levels. That is the concept we are proposing in this paper based on our large-scale quantitative data set.
From your perspective, is this quite novel or have you seen anything like this before?
I haven't seen anything specifically like this. This kind of data set has never existed before; this is the first large-scale quantitative protein interaction network. There are a lot of qualitative interaction networks out there — ones that have been produced by yeast two-hybrid screens, or mass spec. This is the first quantitative network and so gives us for the first time the opportunity to see how the network changes as you change affinity thresholds. The idea that only some proteins become dramatically more promiscuous when overexpressed is one that is only addressable experimentally through this type of study. In that sense, it is new.
What could be the ramifications of this study in terms of cancer treatment?
One of the ramifications I see is that right now many modern chemotherapeutic agents are directed at the receptor itself; the protein upstream in the process. What we are proposing, based on our data, is that a more selective way of targeting these types of cancers is not necessarily to go after the receptor itself, because it will induce negative side effects, but rather to go after the secondary signaling pathways that are only getting turned on when the receptor is being over produced. That way you are being more selective. New cancer therapeutic strategies would be to target the most critical secondary pathways that are only turned on when the receptors are over produced.
How will you channel those ideas to the right people? I saw on your CV that you are a co-founder of Merrimack Pharmaceuticals.
Merrimack is a company that is using systems biology approaches to study problems in cancer. So, how would these types of discoveries get channeled? Obviously the information contained in the paper is public domain, it's not patented. Anybody can go after it. There's no specific pipeline to Merrimack here. It really is open access. It is my hope that any company would pick up that type of idea and pursue it. It's not something that is necessarily patentable. Hopefully it will inspire people to redirect their focus.
I can certainly push it a little further in my own lab. I am pursuing this hypothesis further by trying to identify which are the most critical secondary pathways and if we can knock these things down with shRNAs. That's not going to be a therapeutic, but that's going to provide some validation for some of these targets. Then, it's really up to the pharmaceutical companies to pick up on that and go after it.