Last October, about 120 researchers studying everything from biophysics to plant science met at Howard Hughes Medical Institute at Janelia Farm in Ashburn, Va., to discuss fluorescent proteins and biosensors. They'd been reading each other's papers for 15 years, but most had never met, and biophysicist Thomas Hughes of Montana State University remembers it as the most exciting meeting of his career. "I got the feeling from that conference that [biosensor development] really is going to be a field," he says. Indeed, techniques such as fluorescence resonance energy transfer (FRET) have become key tools in detecting real-time molecular interactions within living cells.
A FRET biosensor usually consists of one donor and one acceptor fluorescent protein, a substrate molecule that changes conformation upon binding the molecule of interest, and sometimes, a linker of a few amino acids that joins...
User: Thomas Hughes, Montana State University, Bozeman
Project: Making a FRET-based biosensor that can be inserted into the amino acid chain of a large signaling protein, such as a glutamate channel, to measure its activation in living cells.
Problem: Pinpointing a location within the channel that would allow the biosensor to function without interfering with channel activity
Solution: Hughes's group made a random library of transposons carrying fluorescent protein fragments, inserted these into 150 different places within the channels expressed on bacterial cells, and then did functional screens for fluorescence. They found that the biosensor could be inserted in the most unexpected places. In the glutamate channel, "there were insertions right next to transmembrane domains that no one in their right mind would've tried. They worked great," says Hughes. Once the group finds insertion points that work, they can start optimizing the sensor by tweaking the linker length or changing the fluorophores.
Considerations: Read up on whether others have tried adding biosensor elements to your protein of interest. For some proteins, a rational design works best when you have a detailed structural information in hand, says Hughes, though it may take months rather than weeks to do if you don't have high resolution (angstrom-level) structural data.
If you're not seeing the positive clones light up during your fluorescence screens, don't give up - small changes can be easy to miss, Hughes notes. Make sure you're using the correct focal plane and filters. If a clone doesn't express well, switch to objectives with higher apertures. Be systematic as you scan the plates.
User:Klaus Hahn, University of North Carolina at Chapel Hill
Project: Tracking molecular interactions of the protein RhoA, which regulates a cell's motility and morphology
Problem: Hahn's first stab at a RhoA biosensor relied on intramolecular FRET, in which the donor and acceptor fluorophores are held together by a linker protein even in their "off" state. That creates some residual energy transfer between the fluorophores, diminishing the difference in fluorescence upon RhoA binding.
Solution: Varying the length of the amino acid linker piece that joined the two fluorophores helped immensely in making the energy transfer between fluorophores detectable. A linker must maintain a distance of no more than 10nm between the two fluorophores in order to allow the energy to transfer. Within that distance, says Hahn, "We were very surprised that there was a very small range of linker space that worked." But Hahn also found he could raise the sensitivity further still by eliminating the linker piece, separating the two fluorophores completely, in a method known as intermolecular FRET.
Considerations:First, try to find the right linker length in your intramolecular FRET, Hahn says. If intramolecular FRET isn't working well, consider boosting sensitivity by removing the linker. However, know that this step can result in fluorescence bleedthrough, which requires extra corrections in image processing.
User: Vladislav Verkhusha, Albert Einstein College of Medicine, Bronx, New York
Project: Designing fluorescent proteins for biosensors that measure two different signaling events simultaneously within a cell
Problem: Detecting two different events in a cell requires two pairs of fluorescent protein sensors with non-overlapping wavelength spectrums of visibility. One pair, orange-red, worked well, but the blue protein of a blue-green pair was not photostable. "When you start to image, it would dim within a few seconds," says Verkhusha.
Solution: Verkhusha's group bumped up the brightness of their blue fluorescent proteins by taking red fluorescent proteins of different genetic backgrounds, such as TagRFP, and converting them into blue probes using rational, site-directed mutagenesis. Then they used random mutagenesis to tweak the blue variant of TagRFP. The resulting probe, which they call mTagBFP and which took about a year to develop, is more stable and has nearly twofold higher brightness than other blue fluorescent proteins (Chem Biol, 15:2008).
Considerations: Shop around for your fluorescent proteins by looking at the available literature on photostability, pH stability, specificity, and sensitivity, and remember to factor elements of your experiment into your choice. For example, Verkhusha prioritized brightness for use in widefield microscopy, but if you plan to do a lot of confocal microscopy, you might focus on photostability with a different protein such as EBFP2.
User: Yingxiao Wang, University of Illinois at Urbana-Champaign
Project: Wang's group uses a FRET biosensor to detect cell structure enzyme Src kinase to measure how laser-induced mechanical stimuli trigger biochemical changes in individual cells.
Problem: Researchers had developed a FRET-based biosensor for Src kinase activity in 2001, but it turned out to also respond to other kinases and growth factors. Wang needed a biosensor with better specificity.
Solution: It took Wang more than two years to achieve the specificity he needed, a process that involved redeveloping the segment of substrate peptide that binds to the Src kinase enzyme and triggers the transfer of energy between the two adjacent fluorophores. Taking a stab in the dark, he began replacing the original biosensor's synthetic substrate peptide with a naturally-occurring one. "We felt an endogenous protein would be better," Wang says. (True in some cases, though in others, naturally occurring proteins can create background noise by competing with endogenous proteins.)
The naturally-occurring substrate Wang's group chose had many possible fragments that could bind Src kinase, so they used structural data to select a few for in vitro kinase assays to find one that would be the most specific (Nature, 434:1040-5, 2005). They also extended the linker piece, a stretch of amino acids that joins the substrate peptides with the fluorescent proteins, by a few amino acids, further improving the specificity and sensitivity.
Considerations: As you're tweaking, you can model your biosensor and predict FRET efficiencies with biosensor development software called Fusion Protein MODeller, freely available by Kevin Truong's group at the University of Toronto.
Fluorescent proteins (FPs) or FP genes
DNA2.0 ($1.25 - $1.60 per base pair of DNA submitted for custom FPs
• Cost of custom FPs varies based on the complexity of sequence -- lots of repeats or high GC nucleotide content increases complexity
• Prices are somewhat negotiable
• Turnaround time is 3-15 business days
Clontech (~$500 for an FP but varies based on the constructs)
• Exhaustive list of available FP genes
• Pricey, but scientists can simply clone the constructs to share the product
If you're buying a new FRET-enabled microscope, large companies include FRET image processing in the software that accompanies their confocal and widefield systems. Leica Microsystems' Leica AF7000 costs between $200,000 - $400,000+, for example.
• Check what sort of FRET formulas they use -- most are configured for sensitized emission FRET. (Leica's package includes calculations for spectral FRET and acceptor photobleaching).
If you're buying software alone:
Molecular Devices's MetaMorph 7.5 - Software for controlling the microscope stage, camera, and other hardware components, as well as image processing, start at $15,000, while image precessing software alone starts at $3400
Andor Technologies' iQ Software ($3,900) - image analysis package specifically designed for live cell imaging
CircuSoft's PFRET ($1250) - a calculation module that can be integrated with other software to subtracts out background
(~$900-2,000, depending on whether the user already has some of the parts)
• Prices are comparable for standard filter sets, which consist of one excitation filter, one dichroic mirror and two emission filters
• Standard sets are available for the most common fluorophore pairs: blue-green, cyan-yellow, and green-red
• For custom-designed filters for newly developed fluorescent proteins, Chroma Technology is a popular choice
• Do your homework and choose FPs and filters carefully - they will affect your background calculations later