Researchers at Lund University are exploring an immunoaffinity technique that could enhance the usefulness of antibody enrichment in mass spec-based discovery proteomics.
The method, which relies on antibodies that target short amino-acid sequences, or motifs, present in multiple proteins, could significantly improve the sensitivity of current biomarker-discovery approaches, said Carl Borrebaeck, chair of the Department of Immunotechnology at Lund and one of the leaders of the project.
Immunoenrichment is commonly used to isolate proteins prior to mass-spec analysis in order to address issues like sensitivity, resolution, reproducibility, and dynamic range. Because antibodies are designed to bind specific proteins, though, using them for enrichment requires selecting the analytes of interest beforehand – meaning such assays are largely confined to targeted mass spec work.
Borrebaeck and his colleagues have worked around this limitation by generating antibodies that bind not to sites specific to single proteins but to short amino-acid motifs present in up to several hundred different proteins.
Dubbed context-independent motif specific, or CIMS, antibodies, these reagents allow researchers to isolate proteins for mass-spec analysis in a semi-untargeted fashion, combining the enhanced sensitivity of targeted approaches with the breadth of a discovery workflow.
"If you have 100 antibodies that recognize between 50 and 100 peptides [each], that would allow you to interrogate maybe 5,000 to 10,000 different proteins," Borrebaeck told ProteoMonitor. "So if you multiplex this and have it as a bead array or a chip array, you can actually use it as a discovery tool."
The effort, Borrebaeck said, stemmed from his lab's research using antibody microarrays for biomarker-discovery work.
"We've been doing antibody microarrays for the last decade, and although we've been coming up with some very nice data on breast cancer and pancreatic cancer, there was still a bias based on what antibodies you put on the array," he said. "So we were trying to turn the microarray into a discovery platform and instead target those motifs that are present on more than one protein."
In June, the researchers detailed a proof-of-concept study using the method in a paper in press at Molecular & Cellular Proteomics. In it, they profiled extracts from human colon tissue, yeast cells lysate, and mouse liver tissue, capturing, enriching, and identifying distinct subpopulations of peptides from all three samples.
In mouse liver, four of the CIMS antibodies bound to between four and eight non-redundant peptides each, corresponding to 21 total unique mouse proteins. In colon tissue, 7 CIMS antibodies bound to between 20 and 63 non-redundant proteins each, identifying a total of 217 different human proteins. In yeast lysate, 6 CIMS antibodies bound to between 23 and 148 non-redundant peptides, corresponding to 251 different proteins.
In addition to demonstrating the CIMS antibodies' ability to pull out multiple proteins, the study highlighted the heightened sensitivity offered by the workflow as 54 percent of the mouse peptides identified had not previously been reported in MS/MS experiments, and 42 percent of the colon peptides and 23 percent of the yeast peptides had not previously been reported in PeptideAtlas.
"When you work with fewer peptides, the sensitivity of the mass spec increases dramatically, Borrebaeck said. "So you get a much higher sensitivity than if you just took the whole proteome and tried to do shotgun" proteomics without immunoenrichment.
In the case of yeast lysate the researchers were able to measure proteins at levels as low as 50 copies per cell, he added.
Since the MCP study, Borrebaeck's team has begun applying the technique to breast cancer research, looking for proteins that might be useful for staging the disease.
"We've seen that we can discover proteins that haven't been associated before with breast cancer," he said, noting that they hope to publish results of this research this fall. Next the researchers plan to investigate markers for pancreatic cancer as well as follow up with the new technique on previous work they've done with chip arrays on gastric and pancreatic cancer.
They also plan to refine their methods for generating and selecting CIMS antibodies and optimize them for working in more complex media like serum.
"We're not there yet, but I think we've done the hard part and actually shown the concept works in real life, so to speak," Borrebaeck said. "What we're doing now is getting more sophisticated in doing the bioinformatics regarding the motifs a little bit more carefully where we deliberately exclude the motifs that are in high-abundance proteins.
"The goal is to be able to use serum as a sample," he said. "It works nicely for those tissues [that the researchers have tried], but since we're interested in clinical applications, we're mostly interested in serum."
Borrebaeck's group has five patents covering the CIMS technology and, he said, would be potentially interested in collaboration or commercialization opportunities. However, they aren't currently pursuing anything concrete.
"We've been working with recombinant antibodies for decades," he said. "This is what we do, so we would be interested in collaborations, and we have five patents around this [technology], so there is also the possibility of commercialization."
His lab is not alone in working on antibodies targeting amino-acid motifs. In October 2010, a team led by University of Tübingen researcher Thomas Joos published a paper in Molecular & Cellular Proteomics detailing a similar immunoaffinity mass-spec platform.
"The concept is very similar," Borrebaeck said, noting that he and Joos had collaborated in the past on the basic concept of motif-targeting antibodies. "The major difference is that we use recombinant antibodies, so they are very easy to manipulate and engineer. The second thing is that we allow some of the amino acids in our motifs to 'wobble,' so they can be any [amino acid], which means that we get a broader coverage."
Joos was not available for comment in time for this article.
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.