TAMPA, Fla. — Purification ranked as the top challenge in protein expression research, according to a survey performed by a new research group formed by the Association of Biomolecular Resource Facilities.
The survey was previewed at ABRF’s annual conference here this week and was one of 11 such surveys and research projects undertaken by ABRF and its research groups during the past year that covered both proteomics and non-proteomics topics.
Proteomics- and protein-specific projects included work on standards and quantitative methods and technologies.
For its first research project, the new group, the Protein Expression Research Group, chose to do a survey to get a bead on the concerns of researchers doing protein expression experiments.
ABRF formed PERG last year as a result of growing demand from researchers whose work was becoming throttled by the complexities of protein expression work, said Michael Doyle, a PERG member and group leader of the Protein Biochemistry Core Laboratory at Bristol-Myers Squibb Pharmaceutical Research.
"It's becoming a bigger and bigger bottleneck in the scientific industry at large, particularly in the pharmaceutical industry, but also in academia," he said.
In a presentation at the conference, John Hawes, a PERG member and assistant professor of chemistry and biochemistry at the University of Miami, Ohio, said that PERG’s goal is similar to all the other research groups of ABRF: to educate the organization's members and the scientific community about the tools and technologies available, in this case for "maximizing expression and purity of recombinant proteins."
In total, 33 laboratories responded to the PERG survey. When asked what process was the biggest problem in their laboratories, 44 researchers answered purification, making it the top problem by a 2-to-1 margin over mammalian expression, cited as the second-biggest problem.
PERG members did not expect the finding.
"I thought it was going to be more in expression," said Doyle. The results did not tease out why purification vexed so many researchers, but Doyle said it may “have to do with the strategy that people are taking.
“They may have a standard expression method that they do not optimize, and then they struggle purifying. Whereas a different strategy may be to optimize the expression before you even get to purification,” he said. “If you can express more of it, then it’s easier to purify.”
After mammalian expression, which received 22 votes, insect cell expression had 19 votes, and bacterial expression was fourth with 15 votes. Gene cloning and yeast expression each had three votes to complete the results.
Survey responders said with mammalian cell expression, low expression is the most frequent problem. With insect cell expression, the most cited issues were low expression and secretion into media. And with bacterial expression, the largest problem encountered is solubility and inclusion. Most respondents said that they encountered solubility issues in up to 75 percent of all their research.
For next year's ABRF conference, PERG may look at how researchers conduct their work.
"What we'd like to do is send out a system that people can [describe] what expression levels they can get using their favorite methods," Doyle said. "Not only to test the abilities of different labs on how they can do and to compare the different methods, but also as a way to teach people who are just entering the field: ‘Here [is] the recommended protocol, try it, see how you compare to all these other labs that have more experience.’”
Assessing Quantification Abilities
As a follow-up to its 2006 quantitative study, for which participants measured the relative abundances of eight proteins in two samples, PERG this year asked participants to identify and quantify proteins, this time from a more complex sample.
A total of 43 labs participated in the study. Participants were sent three samples, each with 100 micrograms of E. coli lysate and each spiked with a total of 12 proteins, 10 non-E coli proteins and 2 E. coli proteins. Two samples were identical to each other, but the third sample was spiked at different levels and ratios.
While some of the participants reported results that PERG found to be "excellent," which led it to conclude that a quantitative assessment of complex samples is achievable, other labs had trouble carrying out their tasks.
About one-third of the participants were able to identify and detect differences in half of the 10 non-E.coli added proteins. While a greater number of participants were able to detect proteins that were added in higher abundance and larger ratios, many proteins that were present at the same levels in different samples were incorrectly reported.
Also, different participants using the same techniques reported different findings.
Protein expression is ”becoming a bigger and bigger bottleneck in the scientific industry at large, particularly in the pharmaceutical industry, but also in academia."
"There is not enough data to draw sound conclusions" about the different technologies and methods used by the participants, said Michael MacCoss, a PERG member and assistant professor at the University Washington department of genome sciences. Rather, "experience [conducting such experiments] seemed to be the overwhelming factor" in determining a participant's success, he said.
Many participants also reported that it took them 10 or more days to run the experiment, even in labs with more than one person working on it, suggesting that quantitative proteomics work is still not routinely done, MacCoss said.
Among the difficulties voiced by participants was the complexity of the proteolytic dyes and long calculation times at several analytical steps. Some also complained about the number of replicates available, making it difficult to determine a reasonable error rate and whether a protein was, in fact, differentially expressed.
Still others said that they found it difficult to find the necessary resources, both in terms of manpower and money, to carry out the experiment. Some labs spent more than $1,000 to carry out the experiment and devoted multiple people to it.
Another group, the Proteomics Standards Research Group, or sPRG, set out to develop phosphoprotein standards and to evaluate the ability of core facilities to identify proteins and their phosphorylation sites in a simple sample.
For the study, which was also presented at the ABRF meeting, each lab received up to eight proteins, some containing one or more phosphoryated residues. Proteins in the samples included catalase, troponin T, osteopontin, and ovalbumin,
A total of 44 laboratories participated in the study. Because some labs did multiple analyses using different technologies, a total of 68 results were reported. Some of the questions asked by sPRG included whether the labs did any enrichment, and, if so, what type; what search engines were used; what other bioinformatics tools were used; and which databases other than SwissProt were used.
Not surprisingly, sPRG concluded that qualitative identification of phosphorylation sites remains a significant challenge for many labs; many participants said it took them anywhere between one to four days to complete the experiment, while some said it took more than a week.
While more than half of the responders found some known phosphorylation sites, some sites were identified by few or no labs, sPRG said. It said also that a lack of accepted statistical and spectra methods for identifying phosphopeptides made it difficult to compare results between labs.
Finally, the group said that the study demonstrates that creating a standard mixture of phosphoproteins remains a significant challenge.
Nonetheless, "We feel this was a good first effort and [it] gave us an idea where we can go from here," said Jeff Kowalak, staff scientist at the National Institute of Mental Health and a member of sPRG.
Results from each of ABRF’s research group projects will be posted on the ABRF website during the next few weeks.