A cell is not just a bag of proteins and other molecules; organelles provide niches with specific chemical environments in which certain members-only proteins are allowed. “Knowing the exact subcellular locations provides good clues for protein function and can also be used as boundaries for protein–protein interactions in systems biology applications,” says Emma Lundberg of KTH Royal Institute of Technology in Stockholm. “Knowing the whole set of proteins in an organelle also facilitates the understanding of its morphology and function.” But despite the availability of numerous methods, identifying proteins' subcellular addresses has turned out to be challenging to do on a proteomic scale. As a result, most human proteins have not been mapped to specific organelles.

Lundberg, her KTH colleague Matthias Uhlén, and a large team of their colleagues now present the Cell Atlas, a comprehensive map of subcellular addresses of human proteins. This project, a part of the Human Protein Atlas (HPA) effort, complements the previously reported Tissue Atlas, a map of protein distribution across human tissues and organs. The data are freely available to the community and should be of great interest to the already hundred thousand or so monthly HPA users. The work also nicely complements other emerging initiatives to build whole human cell maps.

Immunofluorescence maps subcellular protein localization in the Cell Atlas. Credit: From Thul et al. Reprinted with permission from AAAS.

In building the Tissue Atlas, the HPA team generated a large number of antibodies against human proteins, which were used for immunohistochemistry-based mapping. “When we started the Cell Atlas effort 10 years ago we realized that in order to resolve not only the major organelles in the cell but also the fine intricate substructures we needed to work with high resolution at a large scale,” recalls Lundberg. The Cell Atlas team therefore applied a more sensitive approach combining antibody-based immunofluorescence and high-resolution confocal microscopy. Once they had established a robust, automated pipeline, they were able to analyze about 1,000 samples per week, says Lundberg.

Altogether, the researchers mapped the subcellular locations of more than 12,000 proteins to 30 subcellular structures across 22 human cell lines. The cellular addresses of about 47% of the proteins that the team mapped were previously unknown. Approximately one-third of the proteins they charted are expressed in all cells, indicating housekeeping functions. Interestingly, the researchers also found that more than half of the detected proteins localize to multiple subcellular structures. “Multilocalizing proteins may have context-specific function and 'moonlight' in different parts of the cell, thus increasing the functionality of the proteome and the complexity of the cell from a systems perspective,” notes Lundberg.

The validation of their immunofluorescence results was a crucial part of the project. The researchers used a four-tier scoring system based on the availability of supporting validation data to determine the reliability of their antibodies. To confirm a portion of their subcellular localization results with a completely orthogonal method, they used a mass-spectrometry-based mapping technique, hyperLOPIT, which assigns proteins to organelles based on their density gradient fractionation profiles.

The HPA team plans to continue work on the Cell Atlas to map the complete subcellular human proteome. “This means that we will need to work with more specialized cell types, such as primary cells and stem cells,” Lundberg notes. The team's ultimate goal is to develop a model of the entire proteome of a human cell over the course of one cell cycle and to understand how protein localization contributes to cell function.