Super-resolution microscopy reveals that energy transfer between fluorophores at distances <10 nm speeds up photoswitching kinetics

The study investigates whether photoswitching kinetics of fluorophores encode sub-10 nm interfluorophore distances and how such information can be extracted with single-molecule localization microscopy (dSTORM, DNA-PAINT) and time-resolved measurements. It aims to understand if energy transfer and related photophysical processes at <10 nm separations can repopulate the fluorescent on-state and how these effects manifest in measurable fingerprints (localization timing, antibunching) that report on nanoscale distances beyond the classical resolution limit. This has importance for structural biology and nanoscopy, enabling quantification of distances on the order of a few nanometers using standard super-resolution workflows.

Single-molecule localization microscopy analysis: All SMLM results were analyzed with rapidSTORM 3.3 and high-resolution reconstructions were generated with ThunderSTORM. Localization precisions were calculated following Mortensen et al. For photoswitching fingerprint analysis, only fluorescent spots containing >500 photons per frame (dSTORM) or >6,000 photons per frame (DNA-PAINT) were analyzed. To estimate localizations per fluorophore, the Kalman filter-based tracking function in rapidSTORM 3.3 was used to track fluorescent spots over the entire image stack (120,000 frames for dSTORM and 18,000 frames for DNA-PAINT) within a 200 nm tracking radius, saving tracked localization files. A custom Python script calculated: (i) number of frames of consecutive localizations per spot (on-time), (ii) number of frames between on-time events of the same spot within the 200 nm radius (off-time), (iii) average photons detected per frame, and (iv) number of on-time events per tracked spot.

Fluorescence lifetime intensity trajectories: Measurements were performed on a PicoQuant MicroTime200 time-resolved confocal setup with a FLIMbee galvo scanner and an Olympus IX83 microscope (60x, NA 1.45 oil). Emission was split to two Excelitas SPADs via a 50:50 beamsplitter; ET700/75M bandpass filters were placed before each SPAD to suppress afterglow and scattered/reflected light. Excitation used a white-light laser (NKT superK extreme) coupled via fiber (NKT SuperK FD PM). A 100 µm pinhole was used. Data were acquired and analyzed in SymPhoTime64 at ~0.5–2.5 kW cm^-2. Trajectories were recorded in T3 mode with 25 ps resolution; photon antibunching was measured in T2 mode (Sync cable replaced by SPAD 2 cable). Fluorescence decays were analyzed by least-squares deconvolution; model adequacy was judged by reduced χ^2 (~1) and residuals. Monoexponential decays were used for reference structures and 18-nm DNA origamis; multiexponential models (τ1, r1, α1, τ2) were used when needed.

Photon antibunching: Antibunching exploits that a single emitter cannot emit two photons within one excited-state lifetime. The ratio of photon pairs in the central peak at zero delay to the average of lateral peaks in interphoton-time histograms (Nc/Nl) indicates the number of independent emitters. For dSTORM, Nc/Nl < 0.20 indicated single-emitter dominance in the confocal focus with low background. Even if two fluorophores were simultaneously on, homo-energy transfer and singlet–singlet annihilation lead to single-emitter dominated emission. For determining Nc/Nl, the average of the nearest eight peaks (four on each side of zero) was used. Ensemble antibunching showed lateral peak counts are nearly constant at very short times and decrease at large interphoton times.

TCSPC bulk measurements: Performed in a 0.3 mm path-length cuvette (Hellma 105.251-QS) on a PicoQuant FluoTime 200 with a 635 nm pulsed diode laser via SepiaII module, PicoHarp300 TCSPC (80 MHz, 50 ps pulse length, 8 ps resolution), collecting up to 10,000 photons in the maximum channel. Fluorescence was detected at the magic angle (54.7°) to exclude polarization effects. Decays were analyzed with FluoFit v4.4.0.1 via least-squares deconvolution; fit quality judged by reduced χ^2.

Samples and constructs: DNA origami with one (reference) or four Cy5 dyes with interfluorophore distances of 18, 9, 6, and 3 nm were imaged by dSTORM and DNA-PAINT (four docking strands at the same spacings). For protein labeling, genetic code expansion and click chemistry were used to label GluK2 mutants (positions S343, S398, K494, S741) with H-Tet-Cy5 via ncAA TCO*A incorporation; control without ncAA showed no click labeling. Distances were estimated in PyMOL using PDB ID 5KUF.

• dSTORM and DNA-PAINT of DNA origami with four fluorophores at 18, 9, 6, and 3 nm spacings demonstrated that interfluorophore distances <10 nm are encoded in photoswitching kinetics. Bivariate histograms of localizations versus the time to reach 80% of localizations showed that for 3 nm and 6 nm spacings, 80% of all localizations occur within the first minute.
• Photon antibunching confirmed single-emitter-dominated emission in dSTORM conditions, despite increased photoactivation at <10 nm separations. The Nc/Nl,avg ratios (min, max, mean, 25th, 50th, 75th percentiles) for n = 10 single-molecule trajectories per construct in photoswitching buffer were: • Reference: 0.000, 0.116, 0.059, 0.025, 0.067, 0.090 • 18 nm: 0.030, 0.120, 0.077, 0.060, 0.072, 0.107 • 9 nm: 0.033, 0.110, 0.075, 0.050, 0.085, 0.100 • 6 nm: 0.100, 0.285, 0.202, 0.155, 0.207, 0.240 • 3 nm: 0.160, 0.520, 0.299, 0.230, 0.255, 0.360 These values indicate higher Nc/Nl at 3–6 nm consistent with enhanced energy transfer–mediated repopulation and interactions at very short distances.
• Lower irradiation intensity slows photoswitching kinetics. For 3 nm DNA origami, confocal fluorescence trajectories at 640 nm show progressively slower kinetics at 50% and 25% laser power relative to 100% (with 100% corresponding to 2.5 kW cm^-2; 1 ms binning shown at 25%).
• Photon antibunching data and kinetic analyses support that at interfluorophore distances <10 nm, the on-state is repopulated via energy transfer mechanisms; even when two fluorophores are transiently on, homo-FRET and singlet–singlet annihilation lead to emission dominated by a single emitter.

The findings demonstrate that photoswitching fingerprints—particularly the timing distribution of localizations and antibunching behavior—are sensitive reporters of sub-10 nm interfluorophore distances. For constructs spaced at 3–6 nm, rapid accumulation of localizations within the first minute indicates accelerated kinetics consistent with energy transfer–assisted on-state repopulation. Photon antibunching confirms that emission is effectively from a single quantum system under dSTORM conditions, validating the use of antibunching-based Nc/Nl ratios to quantify the number of independent emitters and the extent of interaction. Together, these results address the research question by showing that super-resolution trajectories and time-resolved photon statistics can encode and read out distances below the nominal optical resolution, enabling structural inference at the few-nanometer scale in designed DNA origami and click-labeled protein systems.

This study establishes that photoswitching kinetics and photon antibunching provide robust fingerprints of sub-10 nm interfluorophore distances. Using DNA origami with defined spacings and genetically encoded click labeling of proteins, the authors show that dSTORM/DNA-PAINT data combined with lifetime and antibunching analyses can infer nanoscale proximities below 10 nm due to energy transfer–mediated effects. Future work could extend this approach to complex cellular environments, multiplexed labels, different fluorophore chemistries, and quantitative calibration frameworks to translate kinetic signatures into absolute distance estimates across diverse biological targets.

Specific limitations are not explicitly stated in the provided text. The demonstrations rely on well-defined DNA origami standards and controlled labeling; translation to heterogeneous cellular environments may require additional calibration and controls. Photoswitching behavior depends on irradiation intensity and buffer conditions, which may affect generalizability.