NEW YORK – Scientists with the Human Protein Atlas (HPA) project have characterized the collection of proteins secreted by human cells.
Detailed in a study published last week in Science Signaling, the work provides a basic resource for biological research as well as a potential list of targets for drug and diagnostics development, said Mathias Uhlén, a professor at the Stockholm Royal Institute of Technology (KTH) and first author on the paper.
Uhlén is also one of the leaders of the HPA project, which launched in 2003 with the goal of mapping all the human proteins in cells, tissues, and organs, using various omics technologies. The HPA currently contains protein expression data from 44 human tissue types covering 15,313 gene products, nearly 80 percent of the predicted human proteome.
With their secretome analysis, the researchers aimed to investigate what proteins are secreted by cells as opposed to being held in the nucleus or cytoplasm, an inquiry that is of particular interest from a clinical standpoint as secreted proteins are important both as drug targets and diagnostic markers.
Despite secreted proteins' potential clinical importance, there have been relatively few efforts to characterize the secreted proteome, Uhlén noted. To tackle the challenge, he and his colleagues examined the roughly 20,000 predicted protein-coding genes, assessing whether these proteins and their isoforms were likely secreted based on their genetic sequence, with proteins lacking a transmembrane region and including a signal peptide deemed to be secreted. They also included proteins annotated in the UniProt database as secreted, resulting in a list of 3,513 genes that produced at least one protein isoform that was predicted to be secreted.
They then winnowed this list by eliminating proteins that had no entry in the Ensembl database along with predicted secreted isoforms with low evidence of existence, which led to a list of 2,641 secreted proteins.
Using a combination of bioinformatic analyses and literature searches, the researchers determined that 932 of those proteins remain within the cell (retained in the endoplasmic reticulum or Golgi or other intracellular compartments), 806 are locally secreted within the tissue of origin, 173 are secreted to an unknown location, and 730 are secreted into the blood.
This latter figure is, "I think for everyone in the field an astonishingly low number," Uhlén said.
He added that around 100 of these proteins have no annotated function, noting that these should be of significant interest to researchers. Additionally, 72 of the proteins are products or targets of US Food and Drug Administration-approved drugs.
For roughly half of the proteins predicted to be secreted into blood no immunoassay was available, while around 50 percent of these proteins have been measured by mass spectrometry in various studies.
Given these facts, Uhlén said that one obvious value of the study was giving the proteomics community a list of protein targets to develop immunoassays and targeted mass spec assays.
Uhlén noted that he and his colleagues did not look at what percentage of these proteins were covered by assays in SomaLogic's SomaScan platform, which is able to measure significantly more proteins in blood (around 5,000) than either current immunoassay or mass spec methods.
"I would presume that quite a few of these  proteins are included in the 5,000-[protein] panel from Somalogic," he said.
Uhlén said the findings have drawn significant interest from outside researchers, particularly within pharma firms interested in the role of blood proteins in immune-oncology as well as researchers working in the precision medicine space for whom blood-based protein markers are of great interest.
"People have been waiting for this kind of annotation for many, many years," he said.
While the study itself didn't require any new technology, improvements in proteomic technologies could make the findings more useful, Uhlén said, citing immunoassay approaches like the Olink proximity extension assay the researchers used in the work and advances in mass spec methods.
"The reason we think this is important right now is that we are starting to have very sensitive multiplex assays where we can start to think about analysis of all the blood proteins in one go using very little material to analyze thousands and thousands of proteins," he said, adding that newer approaches also offered much better quantitative accuracy and reproducibility, "which is very important when you are trying to fish for new markers for disease and also just all kinds of biological responses."
While plasma proteomics perhaps lost some of its luster after not living up to early hype, new instruments and new workflows are revitalizing interest in the field. Mass spec experiments are now able to reliably quantify around 500 proteins in plasma at throughputs as high as around 100 samples per day, while affinity reagent-based platforms from Olink and Somalogic can measure 1,161 and 5,000 proteins in plasma, respectively.
Uhlén said that the secretome published in the Science Signaling study was just an initial draft and that he and his colleagues plan to release an updated version on an annual basis.
"We are looking forward to getting input from the community and have already started to," he said, noting that while in the paper the authors determined that there were 730 proteins secreted into blood, the current version of the secretome included only 729.
"Already one of the proteins has been removed based on input from the community," he said. "And a more systematic revision of the list will be done every time we release the Protein Atlas in the future, which will be more or less once a year."