H²MM-guided removal of dye blinking effects from single-molecule FRET burst data

Abstract

Background Confocal burst single-molecule FRET (smFRET) is a valuable technique for studying biomolecular dynamics over various timescales. Photon-by-photon recording and analysis approaches, such as multiparameter Hidden Markov analysis (mpH MM), exploit the full time resolution of the data and allow disentangling FRET-related signal changes from fluctuations caused by dye-related phenomena, such as blinking. However, the influence of blinking dynamics on quantitative mpH MM analysis has not been explored in detail. Methods Using simulated smFRET data, we characterized the impact of blinking dynamics on quantitative mpH MM analysis. We developed an mpH MM-guided approach to remove the subset of bursts affected by blinking. We systematically validated our approach with simulations and applied it to experimental confocal burst smFRET data of DNA hairpins. Results We demonstrated that standard-processed smFRET data from a dynamic DNA hairpin contain dye-blinking states. Since the true parameters in experimental data are unknown, we used simulations with different severities of blinking to characterize the impact of dye-blinking dynamics on mpH MM analysis. We showed that blinking causes mpH MM to systematically underestimate FRET state exchange rates and shift the our proposed mpH E− histogram. We applied MM-cleaning approach to simulated data with various degrees of blinking dynamics. Removing blinking-affected bursts corrected the blinking-induced bias in the mpH E− plot and mitigated the blinking bias of MM analysis. Finally, we demonstrated the effect of the filtering approach on experimental smFRET data of a dynamic DNA hairpin and observed similar effects to those in simulated data. Discussion We propose a method to minimize the impact of dye blinking on dynamic smFRET analysis by using the ability of mpH MM to identify short-lived dye blinking states. Removing blinking-affected bursts improved the accuracy of dynamic smFRET analysis and enabled accurate recovery of state exchange rates by subsequent mpH MM analysis, though at the cost of losing affected bursts. Our findings highlight the importance of considering dye blinking effects in qualitative and quantitative smFRET analysis and the critical need to combine alternating-excitation smFRET with photon-by-photon analysis.