Quantitative multi-color FRET measurements by Fourier lifetime excitation-emission matrix spectroscopy

Abstract

Förster resonant energy transfer (FRET) is extensively used to probe macromolecular interactions and conformation changes. The established FRET lifetime analysis method measures the FRET process through its effect on the donor lifetime. In this paper we present a method that directly probes the time-resolved FRET signal with frequency domain Fourier lifetime excitation-emission matrix (FLEEM) measurements. FLEEM separates fluorescent signals by their different phonon energy pathways from excitation to emission. The FRET process generates a unique signal channel that is initiated by donor excitation but ends with acceptor emission. Time-resolved analysis of the FRET EEM channel allows direct measurements on the FRET process, unaffected by free fluorophores that might be present in the sample. Together with time-resolved analysis on non-FRET channels, i.e. donor and acceptor EEM channels, time resolved EEM analysis allows precise quantification of FRET in the presence of free fluorophores. The method is extended to three-color FRET processes, where quantification with traditional methods remains challenging because of the significantly increased complexity in the three-way FRET interactions. We demonstrate the time-resolved EEM analysis method with quantification of three-color FRET in incompletely hybridized triple-labeled DNA oligonucleotides. Quantitative measurements of the three-color FRET process in triple-labeled dsDNA are obtained in the presence of free single-labeled ssDNA and double-labeled dsDNA. The results establish a quantification method for studying multi-color FRET between multiple macromolecules in biochemical equilibrium.

Fig. 1

Excitation emission matrix (EEM) representation of three-color FRET between fluorescein, Cy3 and Cy5. (a) Photon pathways in a three-color FRET process. Six possible exciter-to-emitter photon pathways are present. (b) EEM representation of the three-color FRET as a function of both excitation and emission wavelengths. Different photon pathways occupy different regions of the EEM. For each photon pathway, the excitation spectrum follows the exciter, and the emission spectrum follows the emitter.

Fig. 2

Time-resolved EEM analysis sequence of three-color FRET. The analysis is performed on a channel-by-channel basis, with each channel involving only at most one unknown lifetime parameter. EEM channels in the illustration are color-coded by their decay models. The analysis first obtains the longest wavelength acceptor (fluorophore No. 3) lifetime τ₃ in ê₃₃, then finds quenched lifetime ${\tau }_2^{123}$ of fluorophore No. 2 in FRET channel ê₂₃. The percentage of quenched fluorophore, P₂ is then calculated from fluorophore 2 EEM channel ê₂₂. The FRET channel ê₁₂ is next, which yields the quenched lifetime ${\tau }_1^{123}$of fluorophore 1. The percentage of quenched fluorophore No. 1, P₁ is extracted from the EEM channel ê₁₁, and finally the FRET channel ê₁₃ serves as a verification of the time-resolved EEM analysis.

Fig. 3

Structure of the triple-labeled dsDNA. The distances between fluorescein and Cy3, or Cy3 and Cy5 were 15 base-pairs or 5.1 nm.

Fig. 4

Schematic of the FLEEM system and data processing. A Michelson interferometer is used to modulate a multi-wavelength CW laser source. The modulated output of the interferometer then excites a fluorescent sample. The laser references and fluorescence emission signals at different excitation-emission wavelengths combinantions are digitized and cross-correlated to obtain frequency response of the sample as an EEM.

Fig. 5

Time-resolved EEM measurements on double-labeled dsDNA. (a) Modulation and phase of quenched Alexa488 (donor EEM channel ê₁₁), in comparison with unquenched Alexa488. The fluorescence lifetime of Alexa488 decreased from 4.1 ± 0.1 ns to 3.0 ± 0.1 ns due to FRET. (b) Modulation and phase of Aelxa546 (acceptor EEM channel ê₂₂). The fluorescence lifetime of Alexa546 was 3.4 ± 0.1 ns, same as pure Alexa546. The acceptor EEM channel is unrelated to FRET, thus the lifetime remains constant. (c) Modulation and phase of the Alexa488-Alexa546 FRET EEM channel ê₁₂. Experimental results in (c) were overlaid with the theoretical model (Eq. (8)). The phase delay in the FRET EEM channel exceeded π/2, which was a signature of FRET. (d) Spectral configuration of the EEM measurement. Error bars in (a~c) represent standard deviations of multiple 46-μs frequency sweeps.

Fig. 6

Frequency responses of bleedthrough corrected EEM channels from a mixture of two-color FRET complexes and free donors. (a) Cy3 EEM channel ê₂₂ fitted with single exponential decay. The acceptor lifetime was found to be τ_Cy3 = 1.5 ± 0.1 ns. (b) Fluorescein-Cy3 FRET EEM channel ê₁₂ fitted with the FRET frequency response model (Eq. (8)). The quenched lifetime of fluorescein was calculated as τ_fluo-quench = 1.0 ± 0.1 ns (c) Fluorescein EEM channel ê₁₁ fitted with a double exponential decay model, in which two lifetime, quenched and unquenched lifetime were fixed at knowing values. The percentage of quenched fluorescein was found to be P_fluo = 37 ± 2%. (d) Spectral configuration of the EEM measurement. Error bars in (a~c) represent standard deviations of multiple measurements with 1 ms integration time. Fitted curves of quenched/unquenched donor and acceptor response are plotted in (a~c) for reference.

Fig. 7

Frequency responses of bleedthrough corrected EEM channels for a three-color FRET mixture with incomplete FRET complexes and free fluorophores. (a) Cy5 EEM channel ê₃₃. The lifetime of the final acceptor Cy5 remained unchanged at τ_Cy5 = 1.8 ± 0.1 ns. (b) Cy3-Cy5 FRET channel ê₂₃. The lifetime of quenched Cy3 was measured as τ_Cy3-quench = 0.8 ± 0.1 ns. (c) Cy3 EEM channel ê₂₂. A double exponential fit found the percentage of quenched Cy3 is P_Cy3 = 80 ± 5%. (d) Fluorescein-Cy3 FRET channel ê₁₂. The quenched fluorescein lifetime was measured as τ_fluo-quench = 1.2 ± 0.1 ns. (e) Fluorescein EEM channel ê₁₁. The percentage of quenched fluorescein was measured as P_fluo = 85 ± 2%. (f) Fluorescein-Cy5 FRET channel ê₁₃. This channel contained signal from the two-step FRET (fluorescein-Cy3 then Cy3-Cy5). The measured frequency response was overlaid with the theoretical two-step FRET model, which was a product of individual frequency responses of quenched fluorescein, quenched Cy3 and Cy5. Fitted curves of quenched/unquenched donor and acceptor response were plotted for reference. Error bars represent standard deviations of multiple measurements with 1 ms integration time.