BOSTON — A series of speakers here at this week’s IBC international protein arrays conference highlighted problems in working with the human plasma proteome, including a decline in success in finding proteins suitable for clinical diagnostic tests, a lack of validation following identification of potential biomarkers, and high costs for developing immunoassays.
“In the last 10 years the number of new proteins per year approved [by the US Food and Drug Administration] for clinical diagnostic tests has declined to almost zero,” said Leigh Anderson, the CEO of the Plasma Proteome Institute in Washington, DC, who gave a keynote speech at the conference. “We are discovering more proteins, so why is the curve going down?”
Anderson said that although proteomics researchers are finding markers, they’re not being validated in part because the cost for validation is too high. In addition, the role of validation has not been taken up by, or designated to, specific companies or institutions.
“Immunoassays would be good, but it takes $2 million to $4 million to make a sandwich assay, per protein,” Anderson said. “Who is going to take these candidate proteins and attempt to figure out which of the hundreds are useful markers?”
Keith Rose, the chief scientific officer of GeneProt, based in Switzerland, said he hopes Anderson is mistaken in his estimation of how much it takes to develop a sandwich immunoassay. He said his company’s industrial-scale work on discovery proteomics has resulted in the identification of 900 proteins in the human plasma, 70 of which are potential biomarkers that have been shown to be differentially expressed in diseased versus control samples.
Rose’s research group has begun validating these 70 potential biomarkers using 2D gels, but the gels are slow and only process one sample at a time. Rose’s group is looking to validate the proteins on a larger scale using antibodies against the proteins as capture agents.
“We have 70 proteins that need capture agents to hopefully validate them,” said Rose. “I’m hoping that someone will come up to me and say ‘You know what? It doesn’t cost that much money [to make capture agents]. We can do it for a lot less.’”
Once capture agents such as antibodies have been developed against the target proteins, they can be immobilized on wells, arrays or beads, and used against thousands of samples to try to validate if the protein is valuable as a diagnostic tool.
Both Rose and Anderson discussed the preliminary steps in analyzing plasma, and debated whether the most abundant proteins should be removed before analysis, and whether proteins should be fractionated first before digestion, or digested first and then separated.
Anderson pointed out that 10 proteins make up 90 percent of the protein mass in serum, and 22 proteins make up 99 percent. He showed that if albumin is not removed using affinity columns such as those manufactured by Agilent and Applied Biosystems, it shows up as a large smear on the 2D gel. If the protein is removed, the smear disappears and allows for other small spots representing lower-abundance proteins to be visible on the gel.
However, removing high-abundance proteins means that minor proteins that bind to high-abundance proteins may also be removed, Anderson and Rose cautioned.
“You may throw the baby out with the bathwater,” said Rose. “When you strip the columns, is it [removing] just albumin or are there other proteins in there?”
Despite the potential loss of important non-specific binder proteins, Rose said he favors removing high-abundance protein before analysis using 2D gels. He pointed out that good media now exist for capturing the proteins specifically, and that the depleted protein media can be examined to ensure that only the targeted high-abundance protein is present in the media.
In deciding whether to separate proteins first and then digest them, or to digest first and then separate, Rose pointed out that it is more work to separate proteins first, but said it is worth the effort because the origin of a peptide can be determined if proteins are separated first. Using the latter method, the origin of a peptide is unknown and modifications to a protein are less detectable, he said.
Standardizing methods for processing plasma is one of the primary initial goals of the Human Plasma Proteome Project, a Human Proteome Organization initiative that was started about a year ago. As part of the pilot phase, 12 reference specimens collected from three different ethnic populations were sent to 46 different labs in 14 different countries. Each of the labs used their own proteomics techniques to analyze the reference specimens.
Results from the 46 labs are expected to be presented at the HUPO conference in Beijing during the last week of October. Manuscripts of the results will be published in an edition of the journal Proteomics dedicated especially to the Plasma Proteome Project.
“We would like to learn a lot more about different technologies and to know which should be used for particular studies,” said Daniel Chan, the director of the Biomarker Discovery Center at John’s Hopkins University, and the chair of the specimen collection and specimen handling subcommittee for the HPPP.
Chan said he anticipates that results from the different HPPP reference laboratories will be different.
“We want to then investigate why different technologies give different results,” he said. “We would like to standardize technologies and try to get the same results among different laboratories.”