Using Thermo Fisher Scientific's new Orbitrap Fusion mass spectrometer, University of Wisconsin-Madison researchers have achieved comprehensive analysis of the yeast proteome in one hour.
That represents a roughly four-fold increase in speed compared to the field's previous best efforts and offers the prospect of potentially performing comprehensive human proteome analyses in as a few as three to four hours, UW-Madison researcher Josh Coon, the leader of the effort, told ProteoMonitor.
Thermo Fisher introduced the Fusion in June at the American Society for Mass Spectrometry annual meeting (PM 6/14/2013). Combining in one device a quadrupole for precursor selection with both an Orbitrap and ion trap mass analyzer, the instrument's unique architecture along with its high resolution and fast scan rates had early-access users, including Coon, predicting that it would enable significant improvements in proteome coverage.
Indeed, in an interview with ProteoMonitor preceding the new system's launch, Coon noted that, with the Fusion, his lab had been able to identify two to three times more peptides than they could using Thermo Fisher's previous top-of-the-line instrument, the Orbitrap Elite.
He added that he expected that within months of its launch the proteomics community would "see people detecting and quantifying more proteins than anyone has ever done before."
Detailed in a paper published this week in Molecular & Cellular Proteomics, the UW-Madison yeast study didn't so much go deeper into the yeast proteome than previous efforts, but, rather, dramatically cut down the time required for such an analysis.
According to Coon, prior to his lab's Fusion work, the fastest comprehensive analysis of the yeast proteome was a four-hour workflow performed last year by the lab of Max Planck Institute researcher Matthias Mann using a Thermo Scientific Q Exactive instrument.
That result, Coon noted, "was pretty remarkable" in and of itself. By way of comparison, in 2008 Mann's lab needed three months to complete the same yeast analysis using a Thermo Scientific LTQ Orbitrap machine.
Now, having cut the time for such an effort down to one hour, the UW-Madison team has turned its efforts to the human proteome. With studies having found that roughly 10,000 to 12,000 proteins appear to be expressed in a given human cell at any particular time, the researchers hope to complete characterization of a human proteome in around three to four hours, Coon said.
"So far we haven't gotten that, but we are working with [the Fusion] to get there," he said. "Right now we can clearly get several thousand human proteins in a few hours on the system, and I'm hopeful that with a few months of optimization we'll get there."
The yeast work, Coon said, stemmed from his lab's control procedures for its mass spec instruments.
"We generally run a yeast whole lysate digest as a control standard just to see how our instruments are performing," he said. "And we realized that with our one-hour control experiment [on the Fusion] we could do pretty well. Then, with some tweaks of the chromatography and the cell lysis [procedure] we were about to achieve 4,000 proteins in the one-hour control period, and we thought, 'That seems pretty significant. Let's write it up.'"
He compared his lab's work on the Fusion to its efforts using Thermo Fisher's previous flagship mass spec model, the Orbitrap Elite, with which they were able to identify around 3,000 yeast proteins in an hour.
"The Fusion has a sampling rate that is around twice as fast as the Elite, which means that in that short [one-hour chromatography] gradient, where there are so many peptides co-eluting, you can sample more and get much better coverage," he said.
Coon attributed this improvement both to the instrument's unique architecture and an enhanced control system that parallelizes functions including ion injection, precursor isolation and fragmentation, and mass analysis.
"In the geometry of the new system, the ion traps don't have to do isolation, they don't have to do fragmentation, they only have to mass analyze," he said. "So they are relieved of some of the functions that they have to do in the older system, which just makes them faster."
"On top of that, [Thermo Fisher] has redesigned the computer controls so [the instrument] can do things in parallel, which was not done that efficiently in the older systems," he added.
The MCP study, Coon noted, is an example of the move within the field to "single-shot" proteomics – the practice of characterizing whole proteomes without pre-fractionation.
Enabled by ongoing advancements in mass spec technology, the notion is "a pretty transformative idea," he said, adding that his lab's recent work with the Fusion "really does forecast that moving forward, extensive fractionation is probably not going to be necessary to get deep proteomic coverage."
Rapid "single-shot" analyses coupled with multiplexing mean that researchers can compare, for instance, 12 proteomes at full depth in one hour, Coon suggested. "This instrument combined with the idea of not fractionating means the rate at which we can do proteomics is going to be much faster."
In fact, he said, mass spec-based proteomics "might become quite competitive with transcriptomic technology."
Achieving the speed, coverage, reproducibility, and relative simplicity of transcriptomic experiments has emerged as a commonly cited goal for mass spec instrumentation. At the Human Proteome Organization's 12th annual meeting, a session on proteomics instrumentation led by Bruno Domon, head of the Luxembourg Clinical Proteomics Center, had as one of its stated goals the idea of developing a proteomics system analogous to RNA-seq platforms. Such an instrument, the participants determined, would be capable of identifying and semi-quantifying every protein in a cell at an average of roughly 50 percent sequence coverage and reproducibly measuring differences between two cell populations (PM 9/27/2013).
Using the Fusion for "single-shot" analyses is "absolutely" a step in this direction, Coon said.
"Fractionation typically requires an offline HPLC, a lot more sample, an expert user, and more instrument time," he noted. "If we can eliminate the need to do that, we make proteomics simpler and faster, which means it's going to be more accessible and cheaper. So I think it's going to be the way forward."
Furthermore, Coon suggested, continued advances in instrumentation will only further decrease the analysis time required. He predicted that machines capable of characterizing the whole human proteome in an hour would go on the market in the next several years.
The UW-Madison study "is a very impressive advance," said Swiss Federal Institute of Technology Zurich researcher Ruedi Aebersold, who was not involved in the work. "The [Fusion] is definitely an extremely attractive instrument, and it shows very high performance."
He noted, however, that in addition to speed and coverage depth, reproducibility would be key for any platform aiming to offer a proteomic version of transcriptomics.
Coon and his colleagues demonstrated "very deep coverage," Aebersold said. "Now it would be very interesting to see if [they] did this many times over, what is the technical variability, what is the biological variability, and how reproducible is this whole analysis?"
Aebersold added that he believed that targeted, data-independent acquisition mass spec methods like the SWATH workflow his lab developed for use on AB Sciex's TripleTOF 5600+ instrument would ultimately provide higher quantitative accuracy when comparing multiple samples.
In data-dependent acquisition – the method used by Coon and his colleagues in their yeast work – the mass spectrometer performs an initial scan of precursor ions entering the instrument and selects a sampling of those ions for fragmentation and generation of MS/MS spectra. Because instruments can't scan quickly enough to acquire all the precursors entering at a given moment, however, some ions – particularly low-abundance ions – are never selected for MS/MS fragmentation and so are not detected.
In DIA, on the other hand, the mass spec selects broad m/z windows and fragments all precursors in that window, allowing the machine to collect MS/MS spectra on all of them. Researchers then pull protein identification and quantitation data out of these spectra.
Using DIA on the TripleTOF 5600+, Aebersold's team has managed to detect roughly 3,000 yeast proteins in a one-hour, "single-shot" experiment. However, he said, he expected that their quantitative data would be more reproducible across different samples due to the higher number of measurements per peptide typically performed during DIA mass spec.
"I think clearly where the DDA analysis [such as that performed by Coon's lab in the MCP paper] is highly superior is in finding new things," Aebersold noted. "Basically any time that you are in discovery mode, that is the way to go – and there is still an enormous amount to be discovered about the proteome."
On the other hand, "in cases where you compare dozens or hundreds of samples as tightly as possible to find quantitative differences between them, we would assume that targeted [DIA] methods like SWATH would have the [advantage]," he said.
However, he added, "these comparisons really haven't been done yet, because everything is so new."