As is well known, trypsin digest-based bottom-up workflows have dominated proteomics since the field's inception. More recently, top-down, intact protein approaches have gained ground, driven by instrument advances and increasing interest in analysis of post-translational modifications and protein isoforms.
Now, researchers are hoping to split the difference with middle-down proteomics, a method that aims to combine the best of both worlds.
Like bottom-up methods, middle-down uses digestion to break target proteins into peptides of a size suitable for mass spec analysis. Middle-down digestion methods create longer peptides than conventional trypsin digestion, however, simplifying and improving analysis of features like PTMs and enabling more complete sequence coverage.
In a lecture last month at the Association of Biomolecular Resource Facilities' annual meeting, Albert Heck, chair of the Biomolecular Mass Spectrometry and Proteomics group at Utrecht University, shared what he called his dream of a protease that would digest proteins into 10 kD peptides.
At that length, he noted, the peptides would still be well under the 20 kD size at which mass spec analysis becomes difficult, but would be significantly larger – around tenfold – than the peptides created by trypsin digestion.
"If you can cut a protein in five pieces, then you only have to analyze these five stretches and then you have 100 percent sequence coverage," Heck elaborated in an interview last week with ProteoMonitor. "So you don't have to be Einstein to think that that is a better approach than cutting it into 50 pieces."
Additionally, Heck noted, the larger pieces will likely be more amenable to mass spec workflows than some tryptic peptides.
"If you can cleave [a protein] into five pieces, these pieces are probably soluble, they probably fly well [in a mass spec], and they probably fragment well," he said. "So you can actually get 100 percent sequence coverage."
Additionally, said Yury Tsybin, a professor at Ecole Polytechnique Fédérale de Lausanne and a leading middle-down researcher, middle-down should offer improved analysis of PTMs.
With larger peptides "I should have not just one modification [as is typical in bottom-up work], but two or maybe three or four modifications" on a single peptide, he told ProteoMonitor. This, he noted, should make it easier to determine the full complement of modifications on a protein.
"If you increase the size of the peptides, you can reduce the complexity of the analysis, so that is a big incentive," he said.
Despite such incentives, however, middle-down has struggled to move forward. The notion is not new, but it has been held back by a lack of tools – specifically the lack of good proteases for chopping proteins down to the desired size. Heck's dream of an enzyme that cuts proteins into 10 kD peptides remains, at the moment, just a dream.
Work is ongoing on discovering suitable proteases for such work. In 2012, Northwestern University researcher Neil Kelleher published a paper in Nature Methods on the use of outer membrane protease T, OmpT, in middle-down research. OmpT is a di-basic protease, meaning that it recognizes not one but two amino acids and will cut only at sites where both are present, which typically leads to production of longer peptides compared to trypsin.
Tsybin and his colleagues have also been exploring a di-basic protease for middle-down work. However, he said that neither his nor Kelleher's protease were entirely specific for di-basic sites.
Additionally, Tsybin said, a study his group published last year in the Journal of Proteome Research found that di-basic cleavages did not consistently create peptides small enough for middle-down analysis and also failed to provide complete proteome coverage.
"Di-basic is good, but it will not give you complete proteome coverage. [The coverage] will be worse than with trypsin." he said. "Also, some proteins don't have many di-basic sites, so you will have peptides that are much larger than 10 kD."
Other groups, Tsybin said, have tried using the protease Lys-C, but this enzyme typically gives peptides only 50 percent larger than those generated by trypsin. Another option is performing incomplete trypsin digestion. Such experiments are not reproducible, though, due to the fact that different proteins will be digested to different extents each time.
"That is the main limitation right now, there is no enzyme," Tsybin said. In fact, he added, perhaps the future of middle-down will not be enzyme-based but rather will rely on chemical methods of breaking proteins into peptides of the desired size.
It is well known that cyanogen bromide can be used for such purposes, Tsybin noted, but a number of other chemicals might also work, he said.
Such chemical methods also have drawbacks, though. As Kelleher told ProteoMonitor in an interview upon publication of the Nature Methods paper, "with chemical methods it's a nightmare of side chain reactions and oxidations."
Nonetheless, Tsybin said his group has had some success with chemical approaches and plans to present data from the work at the American Society for Mass Spectrometry annual meeting in June.
Beyond the challenge of identifying a suitable digestion method, middle-down also needs to work out informatics challenges. Specifically, Tsybin said, because mass spec software has been developed for bottom-up approaches, it isn't good at searching databases for longer, multiply charged peptides.
One issue, he noted, is that because there are so few middle-down datasets, there's not much data that can be used for training these search algorithms to identify such peptides.
"You can play with simulated data, but you really need real data to train the algorithm," he said. In an effort to rectify this problem, Tsybin and his colleagues have been collaborating with several bioinformatics groups, providing their middle-down data "as seeding data that they can use to develop software," he said.
In terms of instrumentation, Tsybin's lab typically uses a Thermo Fisher Scientific Orbitrap mass spec, but high-resolution time-of-flight instruments should work as well, he said. "If you take the flagship time-of-flight of any company, I don't think it should be a limitation."
He did suggest, though, that the combination of high speed and high resolution offered by recent Orbitrap instruments made those machines particularly well suited to middle-down.
"I think for them it is a very good niche to push the Orbitrap into because the Orbitrap provides you such good resolution in such short time," he said.
In general, though, Tsybin said he thought vendors would wait for discovery of a good middle-down enzyme before jumping in with products tailored to the area. "I think once you tell them you have an enzyme that works, they will go for it."
On the researcher side, meanwhile, the middle-down interest is relatively small but growing, Heck said.
"I know of 10 groups that are really working on this middle-down approach," he said. "And I mean 10 groups that are really on the forefront of proteomics."