Localization accuracy in single-molecule fluorescence microscopy

Abstract

The two prevalent methods to determine the position of individual fluorophores in the focal plane of a microscope rely either on the centroid of the fluorescence pattern produced by the fluorophore on a camera or on targeting the fluorophore's coordinate with a spatially controlled illumination minimum, as realized in MINFLUX. In both methods, (sub)nanometer localization precisions have been reached, but little is known about their accuracy, i.e., the deviation of the obtained coordinate from its ground truth. Here we report a theoretical and experimental comparison of the accuracy provided by these two strategies. We quantify substantial systematic errors in camera-based localization due to the interplay of two difficult-to-control parameters: the orientation of the fluorophore in space and its axial deviation from the focal plane (defocus). For example, within an axial range of {+/-}120 nm around the focus, the median distance between two fluorophores on a nanobody measured by camera-based localization varies by 26.5 nm. Under the same conditions, a commercial MINFLUX system yields a median variation in distance of 1.4 nm, indicating an improvement in accuracy by an order of magnitude. The accuracy of MINFLUX localization can be further improved to (sub-)[a]ngstrom levels by reducing the average distance between the coordinates targeted by the illumination minimum and the fluorophore. As it delivers unrivaled precision and accuracy for mapping fluorophores and their distances, we consider coordinate-targeted localization essential for studying labeled cells at the sub-10 nm level and for exploring structural and conformational changes of labeled biomolecules.