A new microscope can generate three-dimensional images of living specimens with a greater combination of depth and precision than before, according to a report in the August 13 Science. Researchers say the novel technique could prove useful for peering inside developing embryos, for example, to track gene and protein expression patterns over days.

"The direction we see for our instrument is to perform experiments with whole organisms that mimic physiological conditions to get a better picture of everything that's going on," researcher Ernst Stelzer, from the European Molecular Biology Laboratory at Heidelberg, told The Scientist.

Stelzer and colleagues sought to devise a method to map relatively large millimeter-sized samples in vivo. Optical projection tomography can image embryos at high resolution, but of fixed specimens only. Magnetic resonance imaging and optical coherence tomography feature noninvasive imaging, but do not easily provide specific contrasts. Laser scanning microscopy (LSM) can use green fluorescent protein (GFP) or similar labels as contrasting agents for high resolution imaging of protein localization patterns in living organisms, but suffers from limited penetration depth in heterogeneous samples.

The new method, selective plane illumination microscopy (SPIM), shines a sheet of laser light 2 to 8 microns thin through a slice of the sample and then systematically moves the specimen to capture images from each layer. Samples are embedded in a low-concentration agarose cylinder. Since any fluorescence imaging system suffers from scattering and absorption in tissue that can reduce image quality in large samples, the microscope can employ multi-view reconstruction, in which multiple three-dimensional data sets of the same object are collected from different directions and combined in post processing for an optimal representation of the specimen.

In experiments in which the researchers imaged embryos of the teleost fish Medaka (Oryzias latipes) of the transgenic Arnie line that expresses GFP in somatic and smooth muscles as well as in the heart, SPIM was capable of resolving the internal structures of an entire 4-day-old fixed specimen with better than 6-micron resolution as deep as 500 microns inside the fish.

"Confocal imaging can only image 100 to 200 microns deep into tissue. Five hundred microns is a real advantage with something like fish embryos, which are 500 microns in size," said Sean Megason of the California Institute of Technology in Pasadena, who did not participate in this research.

SPIM achieved greater penetration depth because it employs lower numerical aperture illumination than confocal imaging, Stelzer explained. This means SPIM optics receive more light, allowing for higher quality images.

Megason told The Scientist he would, however, "like to see SPIM improve in resolution—confocal can go down to 0.2 microns in the planar direction and about 1 micron in the axial direction."

"What's exciting about this microscope is this fills a niche, a middle ground between techniques," said Scott Fraser of the California Institute of Technology, who did not participate in this research. "With preparations a bit thinner than this, we're able to use 2-photon microscopy to look at events over time at much higher resolution, and using MRI microscopy, we can look at preparations slightly larger than this at lower resolution."

The investigators imaged all the muscles in vivo in Medaka as well as the inner surface of the Medaka embryo heart, a structure barely accessible by confocal LSM. Fast frame recording also allowed SPIM to image the heartbeat. Previously, similar imaging was only demonstrated when the heart was exposed and the embryo was cooled to reduce the heart rate, the researchers note.

"Recording speed is excellent, at up to six frames per second at 1300 by 1000 at 10 bits, but can still be improved," Stelzer said. Shrinking the frame size can increase speed up to 17 frames per second.

By using thin slices of light instead of illuminating the entire sample at once, SPIM greatly reduces photodamage to samples. The research team reported they routinely imaged live Medaka and Drosophila embryos over periods of up to 3 days, apparently without detrimental effects on embryogenesis and development.

The researchers suggest SPIM can be automated for large-scale studies of developing organisms and the systematic collection of expression data. Stelzer said coauthor Joachim Wittbrodt is examining spatial temporal expression patterns for some 15,000 genes. "We could do that in a year or so if well organized," Stelzer said.

"This is a powerful approach, well suited for looking at embryos," commented James Pawley of the University of Wisconsin–Madison, who was not involved in this study. Megason added, "You could imagine the goal of imaging the gene expression patterns from the moment of taking a fertilized egg, dropping it in the microscope, and actually following every single cell division and movement to reconstruct a digital embryo."

"What's exciting about this and other techniques that collect 4-D images—3D images over time—is that instead of guessing at what might have happened between two time points, you can follow a specimen, which makes it much easier to come up with proper interpretations," Fraser said. "It's surprising how often the ability to follow something over time reveals what we thought was happening wasn't what was really going on."

Stelzer acknowledged that although his team's instrument is far simpler to build than a confocal microscope, it would probably prove difficult for the average biologist to implement. He is currently negotiating with companies to commercialize the instrument within 2 years.