Name: Mark Molloy
Position: Director, Australian Proteome Analysis Facility; Associate Professor at Macquarie University
Background: Senior Research Scientist, Discovery Proteomics at Merck; PhD Biochemistry, Macquarie University
Mark Molloy is the director of the Australian Proteome Analysis Facility and an associate professor at Macquarie University. Birthplace of the term "proteomics", APAF provides proteomic services to clients throughout Australia and around the world, with a focus on differential protein expression analysis.
Recently, APAF purchased an AB Sciex TripleTOF 5600 mass spectrometer, making Molloy among the first researchers to have access to the instrument since its launch this May at the American Society for Mass Spectrometry's annual conference (PM 05/28/2010).
This week Molloy is attending the Human Proteome Organization's annual conference in Sydney. He spoke to ProteoMonitor about his impressions of this year's HUPO meeting and his lab's initial work with the TripleTOF 5600.
The following is an edited version of the interview.
The Institute for Systems Biology and the Swiss Federal Institute of Technology announced at HUPO this week that they've completed the first phase of building a mass spectrometry map of the human proteome. What is the significance of that for proteomics research?
Right, they're claiming that they've got 99-plus percent of the human proteome. Ruedi Aebersold [ISB co-founder and professor of systems biology at ETH] was saying that they have five peptides for each of the proteins in the human proteome. That's significant to have an SRM assay – essentially a look-up table for any protein in the human proteome – and to have your choice of five peptides for any of those proteins. What hasn't been done is to start to look at these [peptides] in real systems now – so in disease states and normal states and to see how well they correlate with reality. The other thing that hasn't been considered yet is the effect of modifications to the peptides, and we know that's going to be a common thing that will occur when you're analyzing real samples. These peptides they've made haven't accounted for those additional modifications. And I think they acknowledged that, they understand that, and I think that will be part of the next phase, to try and address that.
So this initial map just covers synthetically made peptides?
That's right. So they're just really cataloging what are the combinations of ions that one should look for for that peptide. And that's great if, in the real experiment, the protein is an unmodified protein that doesn't contain any other additions. Then those masses will be correct. But as soon as you stick a phosphate on a peptide, everything is going to be out of whack.
And post-translational modifications like phosphorylation are thought to be crucial to understanding things like disease states?
That's exactly correct. So I think for HUPO and the community at large to be able to say, Well, we do have assays for every protein in the human proteome,' that's an achievement. Whether we can detect those in real samples is another matter. And I guess that's what we're moving on to.
It's a starting point. With this [map], in theory, one could ask questions about any protein of interest that they had some concern about in any biological system. So, for instance, if you wanted to know what was happening to p53 in some cells that you're interested in, you now have a look-up table to tell you how to acquire data on your mass spectrometer to detect that protein of interest. Of course, that doesn't mean that you will detect it, because the level of expression isn't known. But I guess that's the goal of the experiment. And then also what are the background interfering ions that might contaminate the signal? All that will have to be assessed in real samples. What they've done should provide a shorter path to allow us to at least test those hypotheses.
Have there been any other presentations at HUPO that struck you as particularly interesting?
There have been some very nice talks on using stem cell proteomics to look at disease states. That's work that I haven't seen at other HUPO meetings, applying proteomics in the stem cell setting. That's like deep drilling mass spectrometry for comparative analyses. I haven't seen much in the past with stem cells and that level of coverage.
There was a good talk by Carl Borrebaeck from Lund University in Sweden using antibody microarrays for coming up with candidates for panels of biomarkers. The sensitivity and specificity looked pretty impressive.
Speaking of antibody arrays, the need for improved antibody standards has been an issue in proteomics. What's being discussed in that area?
Mathias Uhlen gave an update on his Human Protein Atlas, where he's building polyclonal antibodies and starting to make some monoclonals, as well. I don't remember the exact number [the Atlas currently contains information on more than 11,000 antibodies], but different antibodies from a whole range of different sources – vendor-supplied, investigator-supplied – with corresponding specificity data is available on that website. So I think people really will have the opportunity to see the quality of the antibodies that they're using.
[ pagebreak ]
With regard to the Human Proteome Project, there was debate at last year's HUPO conference about whether the project should take a gene-centric approach or a sample-centric approach. Which way do you think people are leaning on that issue?
Some individual speakers have put their opinions forward on the way it ought to be done. From the sessions I've attended it seems that a gene-centric approach is the way that will probably come out as the way forward, rather than a protein-centric approach. And the argument the speakers have put forward is that, at least by doing this we'll have a start and a finish to a minimum requirement. So we can measure progress. There are probably a lot of other attributes to it as well, but that is a good one where we can say, 'Well, we've begun the project and we can complete the project when we've done all the genes that are on the 23 chromosomes.'
Because with a sample-based approach you could theoretically always be adding more samples to the project from different sources?
Right. [Although] I suspect that some groups will be starting to do sample-oriented workflows in addition to that gene-centric approach.
Your lab recently purchased an AB Sciex 5600 TripleTOF mass spectrometer. What are your impressions of the machine so far?
One of the main workflows we have is the iTRAQ method, and we're interested in using the iTRAQ workflow on that instrument. The demo data that we've done looks pretty impressive. We certainly can acquire more peptides than what we do with our current instrumentation. It's a 30-percent to 40-percent increase in the number of peptides [compared to] other Q-TOF instruments that we have in our laboratory.
We're [also] interested in a label-free quant-based approach – going back and being able to extract ions associated with known sequences of peptides. We don't do that work currently, but from what we've seen, that will now become possible. So it will allow a new workflow for us. And we're particularly interested in doing it around phosphopeptides. This allows us to do time-course experiments in a label-free manner. There's nothing different about how the instrument works that changes anything for phosphopeptides, but it just means that we can acquire a tremendous depth of data and by doing comparative runs go back and mine that information and extract quantitative information from it.
We have a big program in colorectal cancer, and we're interested in coming up with prognostic markers and predictive markers for response to therapy. So we're going to be doing quite a large number of sample comparisons using that dataset.