Accurate Single Molecule FRET Efficiency Determination for Surface Immobilized DNA Using Maximum Likelihood Calculated Lifetimes Joshua B. Edel, † John S. Eid, ‡ and Amit Meller* Department of Physics and Biomedical Engineering, Boston UniVersity, 44 Cummington Street, Boston, Massachusetts 02215 ReceiVed: October 4, 2006; In Final Form: January 12, 2007 Single molecule fluorescent lifetime trajectories of surface immobilized double-stranded DNA coupled with a tetramethylrhodmaine and Cy5 FRET pair were directly measured using time-tagged single-photon counting and scanning confocal microscopy. A modified maximum likelihood estimator (MLE) was developed to compensate for localized background fluorescence and instrument response. With this algorithm, we were able to robustly extract fluorescent lifetimes from their respective decays with as few as 20 photons. Fluorescent lifetimes extracted using an MLE were found to be highly dependent on background fluorescence. We show that appropriate factors are required to extract true lifetime trajectories from single fluorophores. Time-resolved detection of single fluorophores using the principles of time-correlated single-photon counting (TCSPC) and confocal scanning microscopy has become increasingly popular in the past few years. 1-6 The fluorescence lifetime of individual dyes is an intrinsic property of the molecule, affected only by its chemical environment. In contrast, the detected photon intensity is an extrinsic quantity, which depends on many experimental factors, such as excitation intensity, collection efficiency, and position of the dye with respect to the excitation beam, all of which do not report on the actual dye properties. Consequently, a quantitative comparison between single- molecule (SM) intensity information and bulk measurements is non-trivial. On the other hand, SM lifetime data can be directly compared with bulk (or ensemble-averaged) measurements, and especially in conjunction with SM intensity measurements, it can reveal information hidden or imbedded in SM intensity- based experiments, such as static versus dynamic quenching and molecular heterogeneity. 7 The most common experimental approach for SM lifetime is performed by observing diffusing single fluorophores through a detection probe volume defined by the confocal volume (commonly known as SM lifetime burst analysis). This tech- nique has the advantage that large SM statistics can be quickly acquired, due to the rapid diffusion of the molecules through the confocal volume. However, it is limited by the relatively short residence time of the molecule within the probe volume, typically no more than ∼1 ms, which precludes the detection of processes with slower timescales. Alternatively, SMs can be immobilized on a surface allowing long measurements (tens of seconds, limited only by photobleaching) of individual mol- ecules to be performed. This approach offers an essential dimension for probing biomolecular dynamics on time scales highly relevant for many biomolecular processes. The fundamental signal in the TCSPC experiment is the time delay between the excitation laser pulse and a single photon emitted by the fluorophore. This signal, however, contains a few artifacts: first, the time delay is convolved with the instrument response function (IRF) of the measuring apparatus. Second, the signal may contain photon contributions from background fluorescence as well as scattering, which contami- nate the pure fluorescence lifetime of the probed molecule. A number of numerical methods have been developed to solve these problems. In particular, Maus and co-workers have recently described a method for IRF deconvolution as well as background scattering determination, applied to burst analysis of diffusing molecules. In this paper, we extend these methods for the case of immobilized SM in conjunction with SM fluorescence resonance energy transfer (sm-FRET). We show that similar to the intensity-based sm-FRET measurements 8 the background fluorescence and background scattering may vary from molecule to molecule and therefore need to be treated independently. To this end, we develop a simple numerical approach to perform time-resolved SM lifetime determination, which takes into account scattering and fluorescence background on a per molecule basis. We demonstrate our method by measuring SM time-resolved lifetime from donor-acceptor FRET pairs conjugated to double-stranded DNA (dsDNA), which serves as a rigid scaffold for the dye. We find that our method is extremely robust and can provide reliable lifetime results with as few as 20 photons. Time-resolved single molecule FRET of immobilized molecules using lifetime probing opens the door to accurately measure the dynamics of individual molecules and to probe distances and distributions of nucleic acids and proteins at the single molecule level. 9-12 Single molecule decays were measured using a high through- put scanning confocal microscope similar to that described by Sabanayagam and co-workers. 8,13 In brief, a modified Zeiss Axiovert 200 microscope integrated with a DC stage and a piezo-driven nanopositioner (Physik Instrumente) was used for all measurements. An 80 MHz femtosecond Ti:Sapphire laser (Tsunami, Spectra Physics) operating at 1000 nm was frequency doubled using a lithium triborate crystal, with the resultant excitation wavelength at 500 nm. The laser was attenuated with a polarizer to reduce the power at the sample to 10 µW, and its polarization was made circular using a quarter wave plate. The * Corresponding author. E-mail: ameller@bu.edu. † Current address. Institute of Biomedical Engineering and Department of Chemistry, Imperial College London, South Kensington, SW7 2AZ, London, U.K. ‡ Current address. Pacific Biosciences, 1505 Adams Drive, Menlo Park, CA 94025. 10.1021/jp066530k CCC: $37.00 © xxxx American Chemical Society PAGE EST: 4.6 Published on Web 02/27/2007