Automatic tracking of single fluorophores
Software algorithms for total internal reflection microscopy developed
As they continue to further unravel the inner workings of cells, scientists are increasingly interested in the roles that proteins play in everything from regulating osmotic pressure in cells to excreting waste. Now, researchers at London’s MRC National Institute for Medical Research have developed a system for automatically detecting and tracking individual molecules that have been labeled with a fluorophore to better understand how they move in different environments and in different cells. The system involves two new computer algorithms for automatic detection and tracking.
“The emphasis of our recent work has been to try to understand how proteins move around in the outer membrane of living cells,” explained Justin E. Molloy, who with colleague Gregory I. Mashanov recently published the details of the new system in the January online issue of Biophysics Journal. The plasma membrane has proteins that transport, pump, bind and respond to ions, ligands and other biological molecules. For instance, receptor proteins bind hormones, and channel proteins respond to electrical changes so that many cells can work together within an intact organism.
Without a fluorescent label, the proteins are too small to see with a light microscope. Fluorescent labels, however, present their own problems, such as autofluorescence and fluorescence from labeled fluorophores not in the focal plane. Total internal reflection microscopy is one way of controlling out-of-focus fluorescence.
To help control out-of-focus fluorescence, researchers use total internal reflection microscopy to illuminate the sample. The resulting evanescent wave penetrates only about 100 nm into the sample, helping to limit fluorescence excitation to fluorophores on the cell membrane.
However, Molloy said that, in most studies, researchers have to analyze the data by hand. This not only creates an opportunity for observer bias, but it also is not feasible for sampling and comparing hundreds or thousands of molecules. So, Molloy and Mashanov developed two computer algorithms for single-fluorophore detection and for automatic single-particle tracking. Together they allow the experimenter not only to detect labeled proteins, but also to watch their behavior within living cells.
The algorithm for single-fluorophore detection uses three characteristics to decide whether an event is the fluorescence of a single fluorophore: diffraction-limited size, a known and constant emission rate and abrupt single-step photobleaching. The algorithm also allows researchers to export data about each object they detect, including fluorescence intensity, timing and the object’s X-Y coordinates. With simulated data, this algorithm allowed detection of more than 95 percent of real fluorescence events and generated less than 1 percent false events.
An automated system for detecting single fluorophores will allow researchers to analyze more fluorophores than manual detection alone. Images courtesy of Justin E. Molloy.
The algorithm for automated single-particle tracking locates single fluorophores in a video and tracks them between video frames using pattern matching of the point-spread function. It stores the center of the fluorescence and then compares the coordinates of these stored centers between frames. When the object tracks are identified, the algorithm refines them using original intensity data. Testing this algorithm on simulated data, the researchers detected 71 percent real events but had a false positive rate of 23 percent.
They also employed the system to study actual protein activity on the cell membrane. They used a total internal reflection setup with biological cells grown in culture and adhered to the microscope coverslip. The setup excites only the fluorophores on the cell membranes that lie close to the glass surface.
The total internal reflection system is built around a Zeiss Axiovert 135 microscope with a 488-nm Novalux laser, an image-intensified CCD camera from Photonic Science and a Solent Scientific temperature-controlled Perspex box.
Tracking the motion of fluorescently labeled proteins on the cell membrane can lead researchers to a better understanding of the protein’s function in cells and how it works in different conditions.
“If the fluorophores are present at a low copy number, then they appear as individual spots of light. Each spot reports the position of an individual protein,” Molloy explained. “The view is a frenzy of chaotic thermal motion. Proteins jostle about and bump into each other as they diffuse within the fluid membrane.”
The researchers use the chaotic motion to their advantage. By tracking the motions of individual spots of light, they obtain an estimate of the protein diffusion coefficient so that they have a real-time readout of the proteins’ movements. For example, two spots become one when two labeled proteins bind, or if a protein binds to an unlabeled one, a sudden drop in diffusion coefficient might be detected. And, a protein might appear and disappear as it exports and imports material to the cell, Molloy added.
Automating the detection allows the researchers to analyze the data using statistics, which, in turn, lets them examine rare or unusual events. The events, Molloy said, occasionally can yield genuine insights, but unless the researchers know the statistical properties of the bulk behavior, such observations can be highly misleading.
“The advantage of observing single molecules in a living system is that we can learn about the kinetics of biomolecular interactions directly,” Molloy added. He compared the effort to attempting to learn the rules of a soccer match by watching how the players move during the game. “Rare events, like a ‘red card’ offense, can determine the outcome of a match, but such events must be carefully studied in context. For instance, a player simply stopping to tie his boots would be a trivial and irrelevant rare observation.” He explained that the goal of learning the rules of biomolecular interactions inside living cells is to predict biological outcomes.
Molloy said that the system runs at only slightly faster than video rate, but in the future, they hope to make it faster. The group also is considering ways to detect other optical characteristics — such as polarization — that can reveal protein orientation or fluorescence lifetime that can report on the chemical environment. “[These] and other optical phenomena also can be exploited to give us more detailed information about the environment and properties of a protein while it is at work in the cell,” he said.
The work has a medical aspect, as well. Molloy said that it could help scientists understand how single genetic mutations of individual proteins can cause impaired cellular function.
“Many inherited and acquired diseases result from changes in the way that proteins and DNA interact with one another, and single-molecule approaches offer a useful tool to help unravel and understand these unwanted changes,” he said.
MORE FROM PHOTONICS MEDIA