smFRET Experiments Using TIRF.

All smFRET experiments were performed at 22 °C in buffer A with additions. For experiments with translocon ligands, RNCs [100 nM, dissociation constant (K_d) = 10 nM (26)], 70S ribosomes [100 nM, K_d = 20 nM (26)], or FtsY [2 µM, K_d = 0.2 µM (33)] and 5′-guanylyl imidodiphosphate (GDPNP; 0.5 mM) were added to the imaging buffer. Nanodiscs containing biotin-linked MSP1D1 protein were immobilized on biotin-polyethylene glycol–functionalized coverslips according to published protocols (56). TIRF imaging was performed on an IX 81 inverted microscope (Olympus) using 561-nm solid-state laser excitation (25 mW) and fluorescence time courses for donor (Cy3) and acceptor (Atto647N) were extracted as previously described (56). Anticorrelated fluorescence traces (correlation coefficient <−0.4) that contained single photobleaching steps for acceptor and then donor were selected for further analysis. The acceptor fluorescence was corrected for bleed-through of donor signal into the acceptor channel. Each trajectory was then smoothed once over three data points. FRET efficiency was corrected for relative quantum yields and detection efficiencies of donor and acceptor. FRET-histograms were fitted to Gaussian distributions using GraphPrism. The vbFRET software package (http://vbfret.sourceforge.net/) (32) was used for HMM analysis of the FRET data and stochastic rate constants were determined by dwell-time analysis of the idealized FRET traces (57).

All model-free analysis was performed in MATLAB. FRET histograms obtained from TIRF experiments were compared using the Kolmogorov–Smirnov test to quantify the largest difference in the empirical cumulative distribution functions obtained from two experiments (58). For significance testing, 1,000 simulated datasets were constructed for each experiment by empirical Monte Carlo simulation. Similarly, the range of the FRET peak for each experiment was determined by identifying the peak FRET value in each of the 1,000 simulated datasets.