Cell biologists view the terrain of organelles using dyes and microscopy, while biochemists try to unlock functional clues by smashing up organellar compartments via density gradient centrifugation. Proteomic scientists have further refined the recipe: take organelle-enriched fractions, then sequence peptides with mass spec. It’s been a successful tactic — databases are brimming with cytoplasmic and nuclear organelle data — but it’s not perfect.
“The trouble is that mass spectrometry is very sensitive. Even if your fraction is 99 percent pure, that remaining one percent will give you 200 proteins that have nothing to do with the organelle,” says Matthias Mann of the Max-Planck Institute for Biochemistry.
Incorrectly assigning contaminating proteins picked up by mass spec is a significant worry when generating proteome-wide catalogs.
Mann and his colleagues decided to take on this problem with a technique they pioneered a few years ago when studying human centrosomes. In that study, Mann’s team used mass spec data from centrosomal marker proteins to define a consensus profile through a density centrifugation gradient — a technique they termed protein correlation profiling.
What worked for centrosomes seems to work for the entire cell. Using PCP with mouse liver cell extracts, Mann’s latest investigation builds on the centrosome study by looking at mass spec data obtained from gradient-fractionated cells and compared with proteins known to localize to specific organelles. The result? A complete map of 1,404 proteins localized to 10 different cellular compartments.
Several proteins found by PCP appeared to home in on more than one compartment, so the MPI investigators opened up their cell biology toolkit to take a closer look. Using confocal microscopy and immunofluorescence to visualize overlapping PCPs, the researchers were able to distinguish subtle differences in localization patterns.
Mann readily acknowledges that highly specialized tools are needed to do protein correlation profiling. He says that high-resolution mass spec is integral to a study of this scope, considering the immense number of peptides to keep track of. Analysis time was also significant: it took a year of PC time to analyze the data generated from a single mouse liver cell, he says.
Yet the real strength in Mann’s method lies in the promise of integrating quantitative proteomics with classical biochemistry techniques and imaging tools. Applying this methodological über-mix to study the temporal aspects of organellar proteomics is next on Mann’s agenda. Let’s just hope the computers in Martinsreid can handle it.