Super-resolution microscopy has provided groundbreaking views of biological structures. Yet issues such as phototoxicity and slow imaging speeds still practically limit most experiments to fixed cells or slow processes. Although these studies are informative, following the dynamics of subcellular structures at close to their true size and in real time remains an important unmet goal.

Actin (orange) and early endosomes (green) imaged with PA NL-SIM and TIRF-SIM, respectively. Reprinted from Li et al. (2015) with permission from AAAS.

Eric Betzig, Dong Li (postdoc in Betzig's laboratory at Janelia Research Campus) and a team of researchers sought to improve live-cell super-resolution imaging of dynamic structures. They started with a method called structured illumination microscopy (SIM). Conventional, or linear, SIM has many advantages that make it useful for live-cell imaging, including rapid image acquisition at low doses of light. However, the principles that underlie SIM also limit it to approximately twofold improvements over diffraction-limited resolution.

Betzig was inspired by his late colleague Mats Gustafsson, who pioneered many SIM approaches, to extend the utility of SIM. He recalls “falling in love with SIM because of its much lower intensities and higher speeds and the beauty of the data [Gustafsson] could get on live cells.” For this reason, his team sought to “find ways around the resolution limit without giving away more than we have to in terms of speed and intensity,” he says.

They developed two approaches for enhancing resolution in live-cell SIM. The first was relatively straightforward; they carried out SIM on a total internal reflection fluorescence (TIRF) microscope with a high–numerical aperture (NA) objective (called high-NA TIRF-SIM). Using this strategy, they were able to perform two-color, high-speed imaging at sub–100-nm resolution using a small fraction of the excitation light necessary for equivalent imaging with alternative approaches. However, the resolution was still limited.

To further improve the resolution achievable with SIM in live cells, the team developed a second approach, called patterned-activation nonlinear SIM (PA NL-SIM). Nonlinear SIM has been used to improve SIM resolution, but the increase in resolution is achieved by compromising imaging speed and light dosage, making it unsuitable for imaging processes in living cells.

PA NL-SIM makes use of photoswitchable fluorescent proteins. Prior to imaging, the fluorescent proteins are in a nonfluorescent state. Patterned light then switches a subset of them to the active state. This dynamic patterned activation means that every active molecule contributes useful photons to the image, allowing the researchers to achieve the same resolution as in conventional nonlinear SIM, but at higher speeds and with less light.

Developing PA NL-SIM was not without its challenges. One hurdle was finding an appropriate photoswitchable fluorescent protein. “You need a photoswitchable fluorescent protein that can take a licking and keep on ticking and undergo many switching cycles without bleaching,” notes Betzig. He also explains that photoswitchable FPs in their off state must be completely off, as “glow in the inactive state can create background that hurts performance.” He credits his collaborator Pingyong Xu for developing Skylan-NS, the protein that was used in their experiments to obtain super-resolution movies of structures such as the actin cytoskeleton with lateral resolution of ∼60 nm. The team also combined PA NL-SIM with lattice light-sheet microscopy for 3D live-cell imaging of mitochondria and Golgi dynamics.

These new methods represent a major step toward the goal of imaging biologically relevant processes at super-resolution, and they stress the importance of imaging living cells with nontoxic doses of light. But Betzig notes that improvements in both optical systems and protein-labeling strategies will be important for future breakthroughs.