Most optical imaging techniques are subject to the tyranny of the diffraction limit, where the optical properties of conventional objective lenses make it impossible to distinguish two objects separated by less than 180 nanometres in the focal plane.

But limits were made to be broken. In 1993, Stefan Hell of the Max Planck Institute for Biophysical Chemistry in Göttingen, Germany, developed the concept of stimulated emission depletion (STED). Fluorescent molecules are activated with a laser spot, which is then overlaid with a ring-shaped beam of low-energy photons that shrinks the effective area of excitation to an extent determined by the brightness of the ring. This means that STED can generate almost arbitrarily small excitation areas.

Stimulated emission depletion provides improved resolution for immunostained neurofilaments compared with confocal microscopy. Credit: R. MEDDA, G. DONNERT, S. HELL

The instrument lives up to its promise, achieving resolution of well under 100 nanometres along all axes, and representing the first 'super resolution' technique to break the diffraction limit. This platform has evolved rapidly — Leica Microsystems of Wetzlar, Germany, plans to release a commercial instrument in 2007 — and Hell has also integrated STED with his confocal 4Pi instrument to push the resolution limit further. “We've got a resolution of 15 to 30 nanometres in the focal plane,” he says. Proof-of-concept studies suggest that STED should be suitable for live-cell imaging, although imaging rapid events with small focal volumes could prove tricky.

Fluorophore concentration is a major concern in super-resolution imaging. Mats Gustafsson, of the University of California, San Francisco, has demonstrated the use of patterned light in a wide-field 'structured illumination' scheme that is effective for planar imaging at less than 50-nm resolution, but he feels that such techniques may be less ideal for smaller focal regions. “As your resolution volume shrinks, it holds a smaller and smaller number of molecules,” he says, “and the individual stochastic response of each molecule becomes more apparent. Eventually, you reach a point where it is better to exploit, rather than fight, this independent molecular behaviour. This is precisely what Eric Betzig, for example, has done beautifully.”

Betzig, of the Howard Hughes Medical Institute (HHMI) Janelia Farms campus in Virginia, cites the research of Hell and Gustafsson as fuelling his interest in super-resolution. His efforts recently yielded a technique for single-molecule imaging: photoactivated localization microscopy (PALM), which he developed in collaboration with Harald Hess of NuQuest Research in La Jolla, California. Subsets of individual photoactivatable fluorescent molecules are activated, and subsequently bleached, within a sample through laser exposure, and then imaged by total internal reflection fluorescence (TIRF) microscopy. This process is repeated, and the resulting images are superimposed into a stack that is computationally resolved into a fluorescent molecule 'map' with resolution below 10 nm.

The technique is not intended for live-cell work, but could offer a powerful super-resolution imaging tool for fixed cells. “There's good hope that it will be fairly easy to adapt for commercial TIRF,” says Betzig. “I have some more grandiose ideas about how to do three-dimensional PALM, but the first order of business is to get a more usable instrument for cell biologists.”

Imaging with stochastic optical reconstruction microscopy (inset) can reveal the detailed structure of RecA-DNA filaments. Credit: X. ZHUANG

HHMI investigator Xiaowei Zhuang, at Harvard University in Cambridge, Massachusetts, working with students Mark Bates and Michael Rust, recently described a similar technique, stochastic optical reconstruction microscopy (STORM). STORM emerged from the discovery by Zhuang's group that the commonly used Cy5 fluorophore has photoswitchable properties. “It can be switched in a controlled way, back and forth hundreds of times, between light and dark states,” says Zhuang. Like PALM, STORM involves cycles of selective excitation and imaging followed by computational reconstruction of the full image. As the fluorophores can be turned on and off, it is possible to do time-resolved imaging and faster imaging cycles. Resolution as low as 18 nanometres has been demonstrated, and Zhuang thinks this is just the beginning. Right now, her top priorities are exploring the potential for multicolour imaging, and improving imaging speed. “If we can bring STORM to one-second resolution, that's going to open up a big door and enable many different things,” she says.

Surveying the field, Hell views these various approaches as complementary tools that advance a common goal — forcing researchers to re-examine what is possible in imaging. “I think breaking the diffraction barrier is a fundamental step forward,” he says. “This is an idea whose time has been coming for a long time now.”