Statistics further pushes the limits of optical microscopy

Using off-the-shelf hardware and reagents, researchers have achieved resolutions on the order of angstroms with optical microscopes. An international team led by Ralf Jungmann of the Max Planck Institute of Biochemistry outlines the new fluorescence microscopy method—known as resolution enhancement by sequential imaging, or RESI—in a recent Nature study . The researchers demonstrate the resolution of targets at a distance of only one DNA base pair (about 0.8 nm), well beyond the capabilities of previous fluorescent microscopy techniques.

Because of the wavelike nature of light, the resolution of optical microscopy is limited to about half the wavelength of visible light, or approximately 200 nm. Superresolution techniques break the resolution barrier by binding fluorescent labels to target molecules, exciting those labels with clever combinations of lasers, and then analyzing photon detections to determine a target’s location (see Physics Today, December 2014, page 18 ). Such approaches routinely allow resolutions in the range of 15–20 nm, providing imaging capability on the scale of organelles and protein complexes. To map individual protein structures, however, higher resolution is needed.

RESI is built on a superresolution technique for single-molecule localization called DNA-PAINT, in which researchers use fluorescently labeled DNA strands that bind transiently to target molecules. When excited by lasers, the labels “blink” against a dark background, which enables a position measurement. The localization precision is limited to about 10–20 nm due to the number of photons collected per blinking event. If multiple target molecules are close enough to fall within that distance range, it becomes impossible to tell which localizations belong to which target.

RESI avoids the problem of crowded fluorescent labels through a process of stochastic labeling and sequential imaging, Jungmann says. Samples are prepared in a solution that contains equal concentrations of several distinguishable labels; each target molecule has a specific probability of binding with a specific label. As a result, there is a smaller chance that adjacent targets will fall within the localization distributions of each other or interfere due to photophysical effects.

After measuring localizations using DNA-PAINT methods, researchers rinse the sample and repeat the process, with the number of iterations depending on the spacing of targets and the desired resolution. Whereas the precision of previous superresolution methods is based on the point-spread function of the optical imaging system and the number of photons collected per blinking event, RESI offers a new precision scaling law in which resolution is increased not by collecting more photons but by acquiring more localizations per target.

“RESI can be considered a paradigm shift because it moves the attention from photons to localizations,” says Fernando Stefani of the University of Buenos Aires, who was not involved in the work. Whereas previous superresolution techniques calculate the standard deviation of individual localizations, RESI involves averaging sequences of localizations.

To illustrate the efficacy of RESI, Jungmann and colleagues used the method to image structural proteins of the nuclear pore complex, which is located on the membranes of cellular nuclei. Individual pairs of Nup96 proteins, which previously could not be resolved with optical microscopy, were resolved with RESI, yielding a sixfold improvement over the best existing methods.

The team then pushed below 1 nm resolution using a flat DNA origami structure that has six base pairs spaced 20 nm apart with the components of each pair at a distance of 0.8 nm. Previous superresolution techniques imaged the pairs as clouds, resolving the 20 nm spacing but not the ends of each base pair. RESI resolved the individual bases with two imaging rounds and an imaging time of 100 minutes.

RESI is especially useful for studying the structural biology of macromolecules or protein complexes in fixed cells, says Kristin Grußmayer of the Kavli Institute of Nanoscience Delft, who was not involved in the study. The image times needed for RESI, however, preclude its use in live imaging of biological processes or tracking of biomolecular pathways.

The new method places increased resolution within the reach of any lab already equipped for localizing single molecules with superresolution microscopy, using standard experimental setup and imaging tools. In contrast, recently developed methods such as MINFLUX require specialized apparatus (see Physics Today, May 2023, page 14 ). And because RESI resolution scales with number of localizations and not with photons gathered, it has the capacity for enhanced resolution with increased imaging times. The authors show that resolution scales of 0.1 nm should be possible with an imaging sequence that encompasses hundreds of localizations for each target. It might seem surprising that something as straightforward as a new averaging technique could provide such a substantial increase in imaging resolutions. Yet, as Stefani points out, “Isn’t that what happens with all great ideas?”