This story originally ran on Nov. 18.
Swedish scientists released this week the seventh version of the Human Protein Atlas database, providing localization and expression data on 10,000 human proteins including 10 million high-resolution immunohistochemistry images.
With the release, the Protein Atlas project – led by Mathias Uhlen, professor of microbiology at Stockholm's Royal Institute of Technology – is now halfway to its goal of characterizing 20,000 human proteins.
Launched in 2003, the Protein Atlas includes, for each characterized protein, information on its subcellular localization; data on its expression patterns in 46 organs and 20 different types of cancer; and information on what cell lines it's expressed in. The open-access database is a tool for basic research, Uhlen told ProteoMonitor, providing "a knowledge base for people studying human proteins, to see what proteins are mitochondrial, what proteins are in the nucleus, and so on." It should also aid applied work like protein biomarker research and the discovery of potential new drug targets, he added.
Robert Moritz, director of proteomics at the Institute for Systems Biology, noted that ISB has been using the Protein Atlas in its work on protein biomarker signatures for indications like colon cancer, glioblastoma, and liver toxicity.
"The effort that [Uhlen] has put into not only characterizing the antibodies but also showing their functional use in terms of spatial detection of proteins and their tissue specificity is of great importance" to ISB's work, he told ProteoMonitor.
Because putting together the Protein Atlas requires antibodies for detecting each protein, the project has also become a valuable affinity agent resource, with Uhlen's team generating roughly 75 percent of the antibodies used in the effort and buying the rest from some 60 vendors around the world. In the course of the project, the KTH scientists have discovered that many commercial antibodies don't have the specificity required for their work, which has led them to develop rigorous validation processes for the agents used in the research.
"We have tested quite a lot of antibodies, and quite a few of them we could not approve for the applications that we wanted to do," Uhlen said. "Even a very specific antibody, if the target is low-abundance, can have cross-reactivity. Therefore we take great care to look at every antibody and validate them in different assays. We're constantly adding new validations to get a feeling for how reliable the antibodies are."
A database "where you have fully qualified antibodies that not only work in Western blots but also in immunohistochemistry, as well, is a golden resource," Moritz said. "It's not available anywhere else."
"There are a lot of antibodies out there that don't work," he added, citing the example of one vendor that, he said, is notorious for "gladly replacing" antibodies that don't work, "saying, 'Thanks for telling us. You've done our QC for us.'"
In addition to using the Protein Atlas for its biomarker work, ISB is interested in incorporating the antibodies compiled by Uhlen's team into its SRMAtlas, Moritz said, noting that he and Uhlen discussed this possibility in September at the Human Proteome Organization's annual meeting in Sydney. At that meeting, ISB and its collaborators at the Swiss Federal Institute of Technology announced the completion of an initial draft of the SRMAtlas, which currently contains more than 170,000 single reaction monitoring assays – one each for at least five proteotypic peptides for each of the 20,300 human genes currently annotated as protein-encoding (PM 09/24/2010).
"What we want to use [the antibodies] for is to produce affinity reagents to the individual peptides that we have defined within our human SRMAtlas – along the lines of a SISCAPA approach," he said.
Uhlen is collaborating on the Protein Atlas with Matthias Mann, director of the department of proteomics and signal transduction at the Max Planck Institute. In February 2009, Mann told ProteoMonitor that his lab was using mass spectrometry to generate data against which Uhlen could calibrate his antibody-based protein work (PM 02/12/2009).
"In some ways, we're trying to do similar things but with completely different technologies," Mann said. "And they should agree, or they should be reasonable. But when they don't agree… that's interesting by way of quality control."
The pair currently has a paper in press at Molecular Systems Biology describing the transcriptome and proteome in three different human cell lines, Uhlen said. That study is part of the European Union's Proteomics Specification in Time and Space, or PROSPECTS, program, which aims to annotate quantitatively the human proteome with respect to protein localization and dynamics.
Funding for the Protein Atlas comes primarily from the non-profit Knut and Alice Wallenberg Foundation, which has donated €100 million ($136 million) to the project. It employs 100 full-time staff at seven laboratories – four in Sweden and one each in South Korea, China, and India.
At their current rate of eight to 10 proteins a day, the researchers should reach their goal of mapping 20,000 human proteins sometime in 2015, Uhlen said, with the project possibly continuing on to map different protein isoforms after the initial draft is completed.
"We hope by using immuno-pullout mass spectrometry to come in a second wave and look at the [protein] isoforms and so on," he said. "But in this first case we aren't doing that. This is kind of similar to the Human Genome Project where you start by sequencing kind of a consensus genome and then you do the variations."
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.