WEBINAR REVIEW: Locating Proteins - Subcellular Tracking with Quantitative Mass Spectroscopy

Skip to Navigation


  • Published: Oct 28, 2015
  • Channels: Proteomics & Genomics / HPLC / Proteomics / Base Peak
thumbnail image: <b>WEBINAR REVIEW:</b> Locating Proteins - Subcellular Tracking with Quantitative Mass Spectroscopy

This webinar was presented by Kathryn S. Lilley on February 26, 2015 and is now available for download by visiting: http://www.spectroscopynow.com/details/webinar/14b0703b4d6/Robustclassification-of-stemcell-protein-localization-by-SPS-MS3.html.

Highlights from a webinar held by spectroscopyNOW.com reveal the biological importance of the location of proteins within subcellular structures and their interactions.

The location of proteins in subcellular structures reveals information about the mechanisms that drive cellular processes, especially during cell differentiation, development and disease. For example, the sets of proteins in a specific subcellular structure can suggest the biological processes that take place in that location. Nonetheless, most of today’s proteomic studies destroy the subcellular compartmentalization from the start, say with the addition of a detergent that dissolves the membranes. New techniques, however, preserve protein-localization information and even allow scientists to study subcellular protein dynamics.

In spectroscopyNOW.com’s webinar, “Robust Classification of Stem Cell Protein Localization by SPS-MS3,” Kathryn S. Lilley of the University of Cambridge described her long-term interest in developing quantitative mass spectroscopy (MS) methods to robustly locate proteins within cells. The following contains highlights from her talk.

The Challenge

In eukaryotic cells, a protein can be found in more than one of the many subcellular compartments. In addition, proteins interact to form functional subunits. During cellular processes, these proteins can also move from one subcellular location to another. These features make up key elements in unraveling subcellular mechanisms, and accurately measuring them requires new methods.

Shortcomings in the traditional methods of tracking proteins limit their value. A known protein, for example, can be tagged with green fluorescent protein (GFP) and located with microscopy, but the fusion protein—the protein of interest plus GFP—doesn’t necessarily move in the same way as the protein alone. Likewise, tagging a known protein with an antibody and tracking it with immunohistochemistry can produce aberrant results, such as showing false locations caused by cross-reactivity with the antibody.

The process gets even more complex when scientists look for unknown proteins. In such cases, a scientist could purify the subcellular niche of interest and use MS to identify the proteins. Nonetheless, most of the subcellular niches cannot be perfectly purified, which leaves proteins from other niches, thereby contaminating the sample.

Consequently, scientists need better methods for identifying and locating the proteins in particular subcellular fractions. As Lilley explained, she wants to know five things:

  • Where is a protein localized?
  • Is it present in multiple locations?
  • What partner proteins does the protein of interest interact with at different locations?
  • How does this information vary for different forms of a protein?
  • How can these questions be addressed dynamically?

Lilley also asked: “How do these change when cells are perturbed in a different way.” The perturbation could be, for instance, a different developmental stage or treatment with different drugs. She said, “We want to do this on a proteome-wide scale and a cell-wide scale with as much suborganelle resolution as we can get.”

Location from Correlation

To make the next step in locating proteins, Lilley relied on work that the late Nobel Laureate Christian de Duve started in the 1950s. Using density gradients, de Duve separated proteins and compared them to marker proteins from known organelles, and he concluded that proteins that behave in the gradient like the marker protein come from the same organelle. Eventually, he described this process as tissue fractionation (de Duve, 1971).

Other scientists used de Duve’s finding to develop gradient-based techniques for quantitative proteomics, such as LOPIT, or localization of organelle proteins by isotope tagging. In brief, LOPIT includes several steps: gently lyse the cells; centrifuge the cells in a density gradient, which puts the heavier organelles at the bottom and leaves the lighter ones toward the top; test fractions with Western blotting and antibodies attached to marker proteins to ensure that the gradient separated the organelles; precipitate the proteins from each fraction and convert them to peptides with trypsin; label the peptides from each fraction with a different isobaric tag, such as a tandem mass tag (TMT); and analyze the samples with liquid chromatography (LC) and MS/MS. The resulting spectra can be used to identify the peptides in the fractions.

As Lilley explained, one experiment creates hundreds of thousands of spectra—far too many to analyze manually. So Lilley and her colleagues perform principle components analysis (PCA) to compare every peptide with organelle-specific markers. So, for example, a peptide that lies in the same area of a two-dimensional PCA plot as a marker from the endoplasmic reticulum is probably a protein from that organelle.

Some of the peptides land in so-called no-man’s land, areas of a PCA plot with no defined connection to an organelle. Peptides in no-man’s land come from unannotated locations or occur in multiple locations.

Lilley’s lab has used LOPIT on a wide range of biological samples, including various cell lines and even some whole organisms, including Drosophila embryos. The heterogeneity of the embryos, however, limited the ability to distinguish some subcellular structures from others.

Like other proteomic methods, LOPIT suffers from some limitations. For one thing, it only captures a subset of the subcellular niches. The data also tend to be noisy, which reduces the resolution of this method. In combination, these two shortcomings limit LOPIT’s coverage of the proteome. As Lilley said, “We’ve been able to see some very nice information for a very few subsets of different subcellular niches, but we want to see everything—the global picture of the cell.”

Dynamic Data Analysis

To move LOPIT ahead, Lilley wanted to add a temporal dimension. As she said, “If you imagine that each PCA plot is a frame of a film, then if you carried out experiments where something was a bit different in each experiment you would get a moving image.” As an example, the first plot could be a sample before treatment with a drug, and the following plots could represent a time series after treatment. For such a dataset, said Lilley, “If we look at the position of the proteins, we would see those that were very consistent in their locations and those that moved around in response to this particular drug treatment.” She added in her webinar, “We may even be able to see sets of proteins that move in a concerted manner to identify functional units.”

To achieve this advance, Lilley’s team started with informatics. In particular, they used pattern recognition to get around the need for large sets of well-curated markers. The work of Laurent Gatto and Lisa Simpson—both at the Cambridge Centre for Proteomics—and Matthew Trotter—one of Lilley’s former colleagues at Cambridge—produced pRoloc, which is a set of tools that can be used to mine spatial proteomics data sets (Breckels, et.al., 2013; Gatto, Breckels, Wieczorek, et.al., 2014; Gatto, Breckels, Burger, et.al., 2014). Lilley said that pRoloc “contains many machine-learning methods from which you can infer protein subcellular localization.” For instance, the phenoDisco (phenotype discovery) algorithm in pRoloc can model LOPIT datasets by using labelled and unlabeled data. Clustered sets of proteins make up phenotypes that can be explored to find the biological reason for the connection. For instance, reanalyzing the Drosophila embryo data with phenoDisco revealed a phenotype that turned out to be proteins that make up a ribosomal subunit.

Hyperplexing LOPIT

To generate what Lilley called “a very good subcellular map of an embryonic stem cell,” she and her colleagues developed HyperLOPIT. It relies on four elements:

  • more complex cell fractionation
  • the Thermo Fisher Scientific TMT 10plex kit
  • synchronous precursor selection (SPS) and Thermo Fisher Scientific’s Orbitrap Fusion Tribrid MS (MS3)
  • multivariate analysis.

With this technique, scientists can sample 10 fractions in one experiment. Appling HyperLOPIT to mouse embryonic stem cells (ESCs), Lilley’s team revealed several benefits. For one thing, SPS-MS3 enhanced the quantitative results versus using MS2. For example, the SPS-MS3 data grouped sameorganelle proteins more tightly and provided more separation between different organelles. This technique also simultaneously tracks thousands of proteins, which it can connect to functional networks, such as signaling pathways. As Lilley explained in this webinar, HyperLOPIT can also be used to generate or test hypotheses. For instance, Lilley’s team asked if the various isoforms of leucine amino peptidase 3 exist in different subcellular locations, and HyperLOPIT showed that some of the isoforms function in the mitochondria and others appear in no-man’s land. As Lilley said, “This is giving us evidence not only that the different isoforms … are in different subcellular locations but actually in these positions they’re forming different interactions.”

Interested scientists can explore these data from Lilley’s team with pRolocGUI.

Dynamic LOPIT

From the start, Lilley wanted to dynamically track subcellular proteins. With her Dynamic LOPIT workflow, she grew ESCs in media that prevented differentiation and in media that encouraged differentiation to a state like the primitive streak. The scientists used HyperLOPIT to study differences in protein location and abundance. With TMT 10plex kit, Lilley’s team used two tags for protein abundance in the two conditions and four for each media type for spatial location—all on about 8,000 protein groups.

Although most of the proteins did not change in abundance, the ones that did made sense. For instance, proteins that control pluripotency decreased in the ESCs that differentiated.

Even though Lilley called these early data, she said that the results produced the “most comprehensive map of the mouse subcellular proteome to date.” In fact, she showed a three-dimensional PCA analysis that provided very good resolution of the marker proteins against different organelles. That is even more impressive when keeping in mind that this technology tracks thousands of proteins in one experiment.


Breckels, L.M., Gatto, L., Christoforou, A., Groen, A.J., Lilley, K.S., and Trotter, M.W.B. 2013. The effect of organelle discovery upon sub-cellular protein localization. Journal of Proteomics 88, 129–140, doi: 10.1016/j.jprot.2013.02.019.

De Duve, C. 1971. Tissue fraction-past and present. Journal of Cell Biology 50, 20d–55d, doi:10.1083/jcb.50.1.20d.

Gatto, L., Breckels, L.M., Burger, T., Nightingale, D.J.H., Groen, A.J., Campbell, C., Nikolovski, N., Mulvey, C.M., Christoforou, A., Ferro, M., and Lilley, K.S. 2014. A foundation for reliable spatial proteomics data analysis. Molecular & Cellular Proteomics 13, 1937–1952, doi:10.1074/mcp.M113.036350.

Gatto, L., Breckels, L.M., Wieczorek, S., Burger, T., and Lilley, K.S. 2014. Mass-spectrometry-based spatial proteomics data analysis using pRoloc and pRolocdata. Bioinformatics 30, 1322–1324, doi:10.1093/bioinformatics/ btu013.

Lilley, Kathryn S. Robust classification of stem cell protein localization by SPS-MS3. Webinar, February 26, 2015, http://www.separationsnow.com/details/webinar/14b0703b4d6/Robust-classification-of-stem-cell-protein-localization-by-SPS-MS3.html.

pRolocGUI: http://ComputationalProteomicsUnit.github.io/pRolocGUI/.

Webinar was presented online on February 26, 2015 with sponsorship by spectroscopyNOW and Thermo Fisher Scientific, Inc.

For research use only. Not for use in diagnostic procedures.

Prepared for Wiley and Thermo Fisher Scientific, Inc. by Webinar Review Editor, Mike May, Ph.D.

Copyright © 2015 Wiley Periodicals, Inc. All rights reserved.No part of this publication may be reproduced, stored or transmitted in any form or by any means without the prior permission in writing from the copyright holder.

Social Links

Share This Links

Bookmark and Share


Suppliers Selection
Societies Selection

Banner Ad

Click here to see
all job opportunities

Most Viewed

Copyright Information

Interested in separation science? Visit our sister site separationsNOW.com

Copyright © 2018 John Wiley & Sons, Inc. All Rights Reserved