Top-down MS picking up speed

Determining calgranulin phosphorylation levels by MudCAT. Broadband MS of human leukocytes shows unphosphorylated full-length (13,144.48 Da) and truncated (12,681.24 Da) calgranulin B as well as both phosphorylated species (13,224.44 and 12,761.28 Da, respectively).
View larger image

Although some prefer to focus on the forest, others like to start with the trees. The two MS-based approaches to proteomics—bottom-up, in which the proteins of interest are first digested into peptides and then separated before MS analysis, and top-down, which first detects the intact proteins and then forms protein pieces via gas-phase fragmentation—are no different. Usually, the two approaches provide complementary information.

Because it is easier to do, the bottom-up technique is more popular. Previously, most top-down proteomics experiments were performed off-line with nanospray. But advances in hybrid MS instrumentation are making the top-down method more feasible for high-throughput proteomics research. In two recent AC articles (2007, 79, 7984–7991; 2008, 80, 2857–2866), Neil Kelleher and his colleagues at the University of Illinois Urbana–Champaign describe their new method for top-down MS on a chromatographic timescale and its applications to yeast and human samples. It's called multidimensional protein characterization by automated top-down (MudCAT).

Reminiscent of multidimensional protein identification technology (MudPIT), MudCAT uses two types of HPLC for the first stage: off-line anion exchange and then on-line reversed-phase. By performing LC/MS/MS without the anion exchange step on a yeast lysate at 12 tesla, the researchers identified 22 proteins ranging from 14 to 35 kDa in one injection. After the team incorporated off-line anion exchange prior to the LC/MS/MS, the number of detected proteins increased to 231 with a mass range of ~8–45 kDa. "The surprise was that we could detect almost 300 proteins by just implementing anion exchange–reversed-phase [LC] without any tuning whatsoever," Kelleher says.

Merits of the top-down method include detection of posttranslational modifications, individually and in combination; high sequence coverage; and better discernment of proteins that have similar sequences. For LC/MS proteomics studies, the top-down method requires high mass accuracy and high resolving power at a fast scan repetition rate and is difficult to achieve in a robust and automated way. Until recently, the best results for top-down proteomics were achieved on custom FT ion cyclotron resonance (FTICR) instruments, but these instruments are complicated to use, have low throughput, and are not accessible to many labs. Kelleher and colleagues addressed some of those challenges by using hybrid linear ion trap–FTICR mass spectrometers. One advantage of using a hybrid instrument is that ion-trap instruments can detect more fragment ions on a chromatographic timescale, whereas FT instruments yield better match confidence with fewer fragment ions.

For the data analysis, the investigators presented new high-throughput software called cRAWler, which processes all of the high-resolution MS/MS data into a format for high-throughput database searching with ProSight software (a free web application for the identification and characterization of proteins by using top-down MS/MS data).

With earlier top-down platforms, the largest proteins detectable in discovery mode had been <50 kDa. By using MudCAT to analyze human leukocytes from blood donors, the team could detect a single nucleotide polymorphism in a 63 kDa protein. The group also determined allelic, truncation, and phosphorylation ratios of protein forms in different blood samples. In addition, Kelleher and colleagues combined MudCAT with the intact mass-tag approach. When proteins in the leukocyte sample were first identified by on- or off-line MS/MS and entered into a leukocyte-specific database, 108 protein forms from 53 genes could be reidentified by their accurate intact masses in ~1 week, including experiments and data analyses.

MudCAT does have some limitations. For example, it will not detect posttranslational modifications present at <5%. In addition, top-down proteomics does not find as many predicted proteins as bottom-up proteomics does. "We have less proteome coverage, but every identification is absolutely sure, and we have 100% coverage of the protein," says Kelleher.

Goals for the group's future research include effectively ionizing a wider mass range of intact proteins, detecting proteins with higher masses, and identifying proteins present at lower abundances. Another aim is to improve the LC/MS/MS fragmentation data on the more hydrophobic proteins. Finally, they want to continue fine-tuning the engineering to gain another order of magnitude in throughput. Kelleher says, "I think we're going to have a new tier of performance for top-down."

—Christine Piggee