NEW YORK (GenomeWeb) – A new study demonstrates how large proteins can be analyzed with greater detail using an alternative mass spectrometry method where all protein fragments are analyzed simultaneously.
Led by first author Hayley Simon and senior author Peter O'Connor of the UK's University of Warwick, the scientists determined structural information about a type of collagen using two-dimensional mass spectrometry (2DMS), a technique that determines mass-to-charge ratio of both fragments and precursors from a single spectrum. They published their results in November in Analyst.
"We can take a terrible mixture of peptides and fragment all of them at the same time and figure out which peptide comes from which precursor," O'Connor told GenomeWeb. "It's unbiased for selections because we don't have to choose which ones to select. We'll sequence all of the peptides simultaneously."
The 2DMS method runs on Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometers, modulating ion signal intensities in a way which carries over into fragment ion signals, and therefore allowing researchers to correlate individual fragment ion signals with precursor ions.
"Since collagen has more than 400 individual peptide signals and whole cells have tens of thousands, this method saves a huge amount of time because there is no need to individually isolate and fragment each one serially," O'Connor said.
It's not only collagen than can be analyzed this way. Any large protein, including ones targeted by drugs or involved in cancer biology, could be analyzed more thoroughly. "It's something that's going to become a big deal in the next couple years," O'Connor hypothesized.
The technique is marginally faster than liquid chromatography tandem mass spectrometry, but the big benefit comes from being able to analyze everything at once; however, the informatics used to analyze the data sets are painfully slow and are the limiting factor to the technique. A computational bottleneck was the reason this one-and-done method never took off in the first place, when it was developed decades ago.
Tandem MS doesn't extract as much information, O'Connor said, because it's often picking the top ten peaks and fragmenting each of those. With 2DMS a researcher can see not only those top ten, but everything else as well, he said. "Fragmenting everything at the same time is easy, but making sense of which fragment comes from which precursor is difficult, and that's what the modulation allows us to do."
The 2DMS technique was inspired by two-dimensional nuclear magnetic resonance more than 30 years ago.
"In multidimensional NMR, what they do is modulate a signal and transfer that modulation and measure the signal itself," O'Connor said. "It has a main frequency and a modulated frequency and you can pull out the two different frequencies. In [2DMS], we modulate the cyclotron frequency and the cyclotron amplitude of our ions. The fragmentation will carry the same modulation as the precursor so we can pull out the parent ion frequencies and spread them out into the two frequency domains."
While several labs published papers on the 2DMS technique in the '80s and '90s, they could never get around problem of the computational bottleneck due to large data files and "serious" noise issues, O'Connor said. "But it is taking off now."
Much of the resurgence is because of the informatics work done by Marc-Andre Delsuc and Christian Rolando of the Université Lille 1 in France, O'Connor said. Delsuc and Rolando are co-authors on the Analyst paper. "They kind of resurrected it from the literature and made it work," partly by applying de-noising algorithms, he said.
Though a desktop computer can analyze 2DMS datasets, O'Connor said it is painstakingly slow. But the algorithms are easily portable to cluster and supercomputing platforms, so increased computing power will help make it a more viable analysis technique.
But the potential increase in detail also greatly increases the analysis time.
"The increase in the number of peaks we can handle is 1,000 fold, a vast improvement in the number of peaks we can handle," O'Connor said. "In mass spectrometry, the info is always in the peaks."
If 2DMS takes off, Bruker, which makes the FT-ICR instrumentation needed to run it, could stand to benefit. "They're the only people selling those anymore," O'Connor said, noting that there are some legacy instruments from Thermo Fisher Scientific and Varian, which Agilent Technologies acquired in 2010, but that both firms stopped making them about a decade ago.
But anybody with an FT-ICR instrument could easily begin using 2DMS. "You literally just get the pulse sequence from us and download the software package and run it," O'Connor said. "There are probably a few dozen people that could do it right today, that have the equipment and expertise."
In addition to obtaining more powerful computing resources, O'Connor said there are still more bioinformatics challenges the subfield needs to address.
"Once we can pick all the peaks, the charge data, and which isotope are we looking at, we can get the mass of all the peptides in one spectrum," he said, which is what's needed to do database searches and identify all the proteins present in the sample. So far, picking out the proteins has been entirely manual and very low-throughput, but O'Connor believes it can be done algorithmically.
"The data is in there to do much, much more," he said.
Once all the computational tools are established, others can join O'Connor and his collaborators in picking out interesting applications. One of his students is using 2DMS to find the exact molecules in the body that interact with the cisplatin class of anti-cancer drugs. Another is looking at islet amyloid polypeptide, a damaging protein that can accumulate in type 2 diabetes patients and kills beta-cells.
"Two-dimensional mass spectrometry can work for any biological system," O'Connor said. "It could be a general tool for all biological testing."