Abstract
Three-color single-molecule Förster resonance energy transfer (FRET) is a valuable tool to study conformational dynamics of macromolecules. In this work, we present a maximum likelihood method for analyzing three-color fluorescence bursts collected from freely diffusing molecules in confocal microscopy. In three-color single-molecule FRET measurements, the third dye with the longest wavelength typically has a much lower quantum yield than the other two dyes, which leads to significantly reduced brightness, particularly for molecular states with high energy transfer to the third dye. This results in biased detection of bursts and inaccurate estimation of kinetic parameters. We extend the previously developed two-color maximum likelihood method (burstML) to the analysis of three-color data, rigorously accounting for burst selection criteria and background noise. The analyses of both experimental and simulated data show that brightness, fractions of acceptor photons related to two-color FRET efficiencies, diffusivity, populations of different molecular states, and transition rates between them can be accurately determined from widely used single-molecule free diffusion experiments without immobilization. BurstML is especially important for the analyses of molecular states with unequal brightness or in the presence of high background noise, outperforming conventional methods that do not explicitly account for burst selection bias and background contributions.