Fluorescent proteins are the most common and versatile class of genetically encoded optical probes. While structure-guided
rational design and directed evolution approaches have largely overcome early problems such as oligomerization,
poor folding at physiological temperatures, and availability of wavelengths suitable for multi-color imaging, nearly all
fluorescent proteins have yet to be fully optimized. We have developed novel methods for evaluating the current
generation of fluorescent proteins and improving their remaining suboptimal properties. Little is yet known about the
mechanisms responsible for photobleaching of fluorescent proteins, and inadequate photostability is a chief complaint
among end users. In order to compare the performance of fluorescent proteins across the visual spectrum, we have
standardized a method used to measure photostability in live cells under both widefield and confocal laser illumination.
This method has allowed us to evaluate a large number of commonly used fluorescent proteins, and has uncovered
surprisingly complex and unpredictable behaviors in many of these proteins. We have also developed novel methods for
selecting explicitly for high photostability during the directed evolution process, leading to the development of highly
improved monomeric orange and red fluorescent proteins. These proteins, most notably our photostable derivative of
TagRFP, have remarkably high photostability and have proven useful as fusion tags for long-term imaging. Our methods
should be applicable to any of the large number of fluorescent proteins still in need of improved photostability.