NEW YORK – A team led by researchers at the Lunenfeld-Tanenbaum Research Institute at Toronto's Mount Sinai Hospital has used proximity labeling mass spectrometry to map the intracellular locations of 4000-plus proteins in HEK293 cells.
According to the authors, the study, published this week in Nature, provides protein localization data, with improved specificity, for areas of the cell not easily analyzed by previous approaches.
The researchers have published their full localization dataset at humancellmap.org, an open-access resource with tools for data analysis.
Proximity labeling typically uses a target protein to tag other nearby proteins with a molecule, often biotin, that allows them to be extracted from a cell and analyzed. The method has become popular with researchers who want to identify the protein components of specific subcellular compartments or identify interaction partners of proteins that are difficult to isolate via conventional immunoprecipitation mass spec experiments.
The Mount Sinai researchers employed the BioID proximity labeling method, which uses the enzyme BioID, a mutant form of the Escherichia coli biotin ligase that promiscuously labels proteins with biotin. In BioID experiments, researchers take proteins known to localize to a cellular region of interest and then express those "bait" proteins as fusion proteins that include the BioID enzyme. In the cell, the BioID enzyme labels all nearby proteins with biotin, and those proteins can then be pulled down and identified via mass spectrometry.
In the Nature study, the researchers used 192 different protein baits to establish the subcellular localization of 4,145 unique proteins. Anne-Claude Gringas, senior investigator at Lunenfeld-Tanenbaum and senior author on the study, said in an email that she and her colleagues used literature searches and data from the ProteinAtlas resource to select its set of protein baits, with the aim of using "several well-characterized markers for each organelle."
Comparing their localization data to existing data from microscopy and fractionation studies, they found that their data identified a similar number of known protein localizations as previous studies, but with significantly higher spatial resolution.
They also used immunofluorescence microscopy to test a number of their localization predictions, particularly in poorly studied proteins with little localization data. Tagging 65 proteins with green fluorescent protein, they found that for 56 of them, the microscopy-based data matched their BioID-based localization predictions.
The researchers also demonstrated the use of BioID to look at changes in protein localization in response to cell stimulus, collecting localization data on the bromodomain-containing protein BRD3 following treatment with the BET inhibitor JQ1, which is known to cause BRD3 to move to the nucleolus. They observed the expected relocalization, which, they wrote, "attests to the applicability of the [Human Cell Map resource] for the exploration of condition-dependent BioID datasets."
The authors noted that in the future, they hope to add to the resource by adding more bait proteins to improve the density of coverage, including analyses of different cell types, and by using different proximity labeling techniques including some with faster-acting enzymes to capture different protein interaction dynamics.
Gingras said that she and her colleagues have begun using the TurboID labeling technology developed by Stanford University researchers, noting that they have found it particular useful for "time-resolved analysis or condition-dependent interactions" and experiments where the stronger signal-to-noise provided by the TurboID approach allows them to use fewer cells per experiment."