Skip to main content
Premium Trial:

Request an Annual Quote

EMBL Team Automates Fluorescence Correlation Spectroscopy for Large-scale Protein Analyses


NEW YORK(GenomeWeb) – Researchers at the European Molecular Biology Laboratory, Heidelberg have developed an automation system that enables high-throughput fluorescence correlation spectroscopy (FCS).

Detailed in a paper published this week in Nature Biotechnology, the approach increases the speed of data collection by roughly 10-fold while also improving the technique's reproducibility, EMBL researcher Malte Wachsmuth, first author on the paper, told GenomeWeb.

These advances, Wachsmuth noted, make it amenable to large-scale protein studies, as he and his colleagues demonstrated by making 60,000 measurements across 10,000 living cells in a study of the dynamics of 53 nuclear proteins.

FCS uses confocal microscopy to study fluorescently labeled proteins in living cells. The technique offers high three-dimensional spatial resolution, enabling researchers to study proteins at the single molecule level.

Essentially, Wachsmuth said, FCS involves focusing the microscope on a portion of the cell of interest and waiting for the labeled target molecules to pass through this area of observation. Based on the number of molecules detected passing through this area and characteristics like their diffusion rates, researchers can determine the concentration of proteins in the cell (from which they can calculate absolute abundance numbers) and their binding states.

Additionally, by using multiple fluorophores at a time, they can track the behavior of two molecules at once, making the technique particularly useful for gathering spatial and temporal interaction data on proteins that have been identified as interactors via techniques like yeast two-hybrid screens or mass spec.

A major limitation of FCS, however, has been the labor-intensive nature of the work, Wachsmuth said.

"It is quite sensitive, but it can only address a single cell at a time, and only a single spot, and each measurement takes around 20 seconds," he said. "So if you want to address several points per cell, and hundreds of cells, which you need to get reasonable statistics, it is a very tedious job. And no one really can or wants to sit in front of a microscope and click to start [a measurement] every 20 seconds or something."

And so the EMBL researchers set out to automate the process, developing a system by which a computer instead of a person could determine the cells and portions of cells to be measured.

Primarily this was a software development challenge, Wachsmuth said, noting that the microscope system itself (in the case of the Nature Biotechnology study a fully motorized confocal Leica TCS SP5 AOBS SMD FCS instrument) is capable of automated readings.

What was needed, he said, was software trained to read an initial image and then instruct the microscope on which areas of the image to concentrate.

"We basically acquire an image of a cell, and then we then let the software identify the cells and the right spots to measure, and then it tells the microscope to take the measurement at these spots," he said. "And this can increase the throughput quite significantly."

"Instead of just a single pair of proteins or single experimental condition, we can put, say, a microtiter plate with 10 proteins of interest on the microscope, and then we can let it run overnight or 48 hours, and we can collect data from eight proteins or 10 proteins at a time targeting different cellular localizations," he added. "So in a reasonable amount of time we can collect interaction data on 50 or 100 proteins."

Additionally, Wachsmuth said, automating the technique improves reproducibility as the decisions of where to focus are being made according to the software as opposed to human judgment.

"When you do it manually it is always biased by what you think a good cell or sample looks like, whereas now it is really based on the quantitative analysis of the images," he said. "So even though it may be wrong at some point, it is at least reproducibly wrong, not just different every time that you do it. So that is a big plus."

Wachsmuth said the technology is a useful tool for gathering additional information on proteins identified as interacting by other methods like mass spec or yeast two-hybrid screens.

"It lets [researchers] identify when and where these molecules interact and to what degree," he said.

For instance, in the Nature Biotechnology paper, the researchers used the technique to look at two proteins – Aurora B-EGFP and INCENP-mCherry that in previous work had been shown to bind to one another. Use of the HT-FCS approach allowed them to gather temporal data on their binding patterns – specifically, they were able to determine that they had high correlation during the cell's M phase, then low correlation during G1-S transition, and then high correlation again in G2 and mitosis. From this, Wachsmuth and his colleagues were able to determine that the chromosomal passenger complex, which is key to the regulation of cell division, contains both proteins only when fully assembled.

They also used the approach to look at chromatin-binding proteins, investigating 53 nuclear proteins, 24 of which they identified as binding proteins. In this case, Wachsmuth said, the technique offered a complementary approach to chromatin immunoprecipitation assays.

A key advantage of the method, he noted, is that it allows researchers to look at protein behavior in living cells.

One limitation, on the other hand, is the technique's low multiplexing capabilities. Generally speaking, researchers can look at just two fluorophores at a time, Wachsmuth said, meaning that investigations of interactions between large groups of proteins must be done combinatorially.

"To scale up we have to do it in a combinatorial way – combine [protein] A with [protein] B, then A with C, then and B with C, and so on," he said.