Researchers from the Pacific Northwest National Laboratory and the Gene Therapy Institute at the Oregon Health & Science University presented a poster at last month’s Eighth International Symposium on Mass Spectrometry in the Health and Life Sciences in San Francisco on work they have done profiling the proteome using an integrated top-down and bottom-up strategy.
Their strategy is based on concurrent LC-MS analysis and fraction collection for comprehensive high-throughput protein profiling. High-resolution reversed-phase LC separation is coupled with a 12 Tesla Fourier transform MS to identify modified proteins, “using detected intact protein masses in conjunction with bare protein identifications from the bottom-up data of the same fraction,” the authors said in the abstract for their poster.
This week, ProteoMonitor spoke with two of the authors — Ljiljana Paša-Toliæ and Si Wu — about their research. Below is an edited version of the conversation.
Tell me what your approach entails and what you did?
SW: For our strategy we used concurrent LC-MS and fractionation. So we have LC-MS results, [and] at the same time we collected fractions corresponding to the LC-MS. So after we looked at the LC-MS, if we had some interesting proteins coming out, and we [went back] to the fraction and did either the bottom-up or top-down. That is basically the idea for the strategy.
For the bottom-up [work], basically for now, we just want to do it in a high-throughput fashion. We did each fraction and bottom up and got the result for protein identification from the bottom-up pipeline. After that, we compared it with top-down LC-MS intact protein mass, because we know that our intact protein mass is very accurate.
Because our mass accuracy off the LC-MS for intact proteins is less than 10 ppm it’s very accurate. For this accurate mass, we compared [them] with bottom-up results, and we can have tentative protein post-translational modification identification.
What kind of samples did you use?
SW: For the benchmark study, we used a standard protein mix. For this standard protein mix, we had eight standard proteins.
LP-T: The idea was to develop a method that would allow us to do high-throughput proteomics at the intact protein level. As you know, top-down proteomics is a really cool thing, but it’s not really compatible with high-throughput LC. You can’t really do MS/MS on these bead molecules on the LC timescale.
So in order to deal with that, we decided to couple our LC system using an Advion TriVersa to fractionate the proteins. While we are obtaining mass spec data on intact proteins, we are simultaneously collecting the fractions.
How many proteins were you able to identify with your bottom-up work and then how much were you able to identify with your top-down work?
SW: For the standard benchmark study, we were able to identify all eight proteins with both the bottom-up and the top-down strategy.
LP-T: The whole idea behind this is, in this particular poster, exemplified by the phosphoprotein beta-casein … we did both as a standard protein, so it should just be a single species.
And then we did the LC, we actually got three peaks. And from the bottom-up data, all three of them corresponded to the beta-casein. But based on the mass, just one of these corresponded to the expected mass of the beta-casein.
So we then did the MS/MS on all three isoforms and actually confirmed that the other two species are single nucleotide polymorphisms. From the bottom-up data, you cannot really deduce this because all three peaks point out to the beta-casein, so you just don’t have this information.
However, from the intact protein mass, we knew that there is something different with this species. And then using MS/MS we were able to find these single nucleotide polymorphisms and identify specifically each of the three isoforms.
In your poster, you said that by limiting the bottom-up analyses to individual fractions, you were able to limit the complexity of the protein mixture. Can you elaborate?
LP-T: The idea here is that from the LC-MS data, you have the masses, and then from the bottom-up data, for the fraction you have a list of proteins that are present in that fraction. Now, you can actually try to match these intact masses to these proteins allowing for certain post-translational modifications if you have high enough mass measurement accuracy.
And if your complexity is not too high, then you have a chance to identify these proteins or post-translational modifications.
What would it have meant if you didn’t try to limit the analysis to individual fractions?
LP-T: It would have [been] just too many possibilities. For instance, in an LC-MS run, we can have 500 different intact protein masses. And if we do a bottom-up analysis of that, we can have 500 different possible proteins.
On the one hand you have 500 masses, on the other you have 500 proteins, and you try to match these two. And at the same time you have to allow for phosphorylation, acetylation, different post-translation modifications or cleavages on the C- or N-terminus, and you can see that the number of possibilities that you can generate is going to be huge, even if your mass measurement accuracy is within 10 ppm.
You would have to have mass measurement accuracy well below one ppm to be able to do that for this complex mixture. So you have to eliminate, to minimize the complexity to be able to do these tentative assignments based on the mass only and bottom-up data.
There are other people who have combined a bottom-up and top-down strategy to their work.
SW: There are other people who have reported their work. But we are the first to do concurrent LC-MS and fractionation. At the same time we got the intact protein mass and the fractionation.
This is [one novelty of our approach]. Another is [that] our LC system is very high resolution and this allows us to separate our proteins quite well. For our beta-casein we were able to separate them into three different isoforms, into different fractions. This allowed us to be able to do identification on both the top-down and the bottom-up.
: The difference, for instance, from [Neil] Kelleher’s [at the University of Illinois] approach would be that they also do TriVersa, and then they reconstitute each fraction and do MS/MS on each peak in that fraction [See PM 07/05/07]. And that takes a lot of time. What we do is try to identify proteins doing the bottom-up analysis of that fraction. And then using the bottom-up data and the intact protein masses, we can do tentative identification. And then we can go and do targeted MS/MS, do MS/MS only on species that we were not able to identify or for which we are uncertain of their identity.
Overall, throughput is significantly increased because the identification does not necessarily require the MS/MS at the intact protein level.
Can you do a quantitative approach on this as well, or is this strictly for the detection and identification of proteins?
SW: Yes, this is the next step.
LP-T: The other thing is … just using LC-MS traces for two different conditions, you can overlay them and pull out species that differ significantly in abundance, and then you can go back to these particular fractions and do targeted MS/MS to identify these species that have changed in abundance between two conditions.
Have you done that yet, or is that the next step in your work?
LP-T: This is what we’re trying to do right now.
Is that part of what you’re trying to do with the vaccinia virus?
LP-T: It is part of the work we’re doing with [the virus]. And we’re also doing it with some other systems. For instance we are looking at hibernation in squirrels. It has been known that hibernation induces certain modifications that are reversible.
Can you describe your work with the vaccinia virus and what you’ve done and found so far?
SW: We just started this study and we were able to detect 500 proteins in the top-down fashion, and now we are positioning bottom-up studies for the fractions. And hopefully, after [we get] the bottom-up results, we’ll combine this top-down mass and bottom-up results … to find some novel modifications.
For the virus, nobody has ever done system studies for post-translational modifications, and nobody really knows what exactly can happen for the virus, what kind of post-translational modification can happen.
For us, we would like to stress on that first and then we’ll study different systems for the virus.