Issue Date: September 7, 2009
Superresolution Imaging Goes 3-D
We live in a three-dimensional world, and scientists would like to zoom in on even the smallest biological structures in 3-D.
In the past few years, several groups have developed superresolution fluorescence microscopy—techniques with sharpness exceeding the diffraction-limited resolution of light. Recently, they have also discovered, and now continue to refine, fluorescence microscopy methods that achieve superresolution in 3-D.
These 3-D advances were the focus of a significant portion of a symposium on fluorescence microscopy beyond the diffraction limit, sponsored by the Division of Physical Chemistry at last month's ACS national meeting in Washington, D.C.
Superresolution microscopy "is a combination of really sophisticated optical physics with photochemistry and with molecular biology," symposium organizer James C. Weisshaar, a chemistry professor at the University of Wisconsin, Madison, told C&EN. "Those tools are beginning to answer some interesting structural questions in cell biology."
The advances build upon 2-D superresolution fluorescence microscopy, which began in the 1990s with a technique called STED (stimulated emission depletion) microscopy, developed by Stefan W. Hell of Max Planck Institute for Biophysical Chemistry, in Göttingen, Germany, and coworkers (Opt. Lett. 1994, 19, 780). The field gained momentum in 2006, when three groups independently and in rapid succession reported superresolution methods that were substantially the same as one another but were given different names: STORM (stochastic optical reconstruction microscopy) by Xiaowei Zhuang of Harvard University (Nat. Meth. 2006, 3, 793); PALM (photoactivated localization microscopy) by Eric Betzig and Harald F. Hess of Howard Hughes Medical Institute's Janelia Farm Research Campus (Science 2006, 313, 1642); and FPALM (fluorescence PALM) by Samuel T. Hess of the University of Maine, Orono (Biophys. J. 2006, 91, 4258).
The methods pushed the x-y resolution of otherwise diffraction-limited fluorescence microscopy to 10–20 nm, but essentially the technique remained in "flatland," unable to provide 3-D information.
Microscopists have now thrown off the shackles of 2-D superresolution imaging that bound STORM, PALM, and FPALM. In the past couple of years, new 3-D approaches have come as fast and furious as three of the original 2-D techniques did in 2006. Where the 2006 batch of superresolution techniques differed in name only, the recent onslaught of 3-D improvements comes in many flavors.
STORM, PALM, and FPALM rely on accurately determining the location of sparse fluorescent molecules—either proteins or synthetic dyes—that can be toggled between "on" and "off" states with light. Only a small percentage of the fluorescent molecules are activated at any given time, so the images of their emission, also known as point spread functions (PSFs), don't overlap. By iteratively determining the precise location of many such fluorophores, microscopists, like pointillists, paint a picture dot by dot, with lateral resolution not achievable via conventional fluorescence microscopy.
Zhuang and colleagues achieved 3-D STORM by giving the microscope a bit of fuzzy vision. They make the microscope astigmatic through insertion of a cylindrical lens into the optical path, which leads to slightly different focal planes in the x and y directions (Science 2008, 319, 810). In this configuration, a fluorophore's PSF changes as its z position changes. At the average focal plane, the image is a circle. Above this focal plane, the image is more in focus in the y direction; below it, the x direction is sharper. The team fits the image with a 2-D elliptical Gaussian function to determine the x and y positions and then calculates the z coordinate. In this way, they can achieve resolution of 20 nm in the x and y directions and 50 nm in the z direction.
The group has used 3-D STORM for several biological applications. For example, they have studied clathrin-coated pits, which are cagelike structures that form on cell surfaces to facilitate the process of bringing cargo inside, said Bo Huang, formerly a postdoc with Zhuang and now at the University of California, San Francisco. The pit grows inward from the cell surface, forming a neck that must be cut to release a vesicle into the cell. Because this process occurs in the direction perpendicular to the cell surface, it can't be observed by 2-D imaging methods, Huang said. In addition, they have used 3-D STORM to image the entire mitochondrial network in monkey kidney cells (Nat. Meth. 2008, 5, 1047).
Meanwhile, Samuel Hess's group had been working on biplane FPALM in collaboration with Joerg Bewersdorf, then at the Jackson Laboratory, in Bar Harbor, Maine, and now in the department of cell biology at Yale School of Medicine. In this method, Hess described, the research team split the light into a short path and a long path to obtain simultaneous images in upper and lower focal planes that are ~350 nm farther from and closer to the microscope objective than the original focal plane (Nat. Meth. 2008, 5, 527). They can analyze these images to determine a fluorophore's vertical position as long as the fluorophore is close to one of the focal planes or between them. In this way, they achieve ~30-nm lateral resolution and ~75-nm vertical resolution.
Switching a microscope between biplane FPALM and 3-D STORM is relatively easy, Bewersdorf noted. All it takes is removing the mirror and beam splitter from biplane FPALM and adding a cylindrical lens for 3-D STORM. He took advantage of the experimental ease to compare the performance of the two methods, using a single algorithm to determine the fluorophores' positions from their PSFs.
Biplane FPALM achieved more homogeneous localization performance over a larger range of depths, Bewersdorf found. He could fit the 3-D localization over a 2-μm range in axial depth. In Bewersdorf's hands, the 3-D STORM did not have as extensive a depth range as biplane FPALM.
Another 3-D version of PALM comes from W. E. Moerner's group at Stanford University, in collaboration with Rafael Piestun's group at the University of Colorado, Boulder. "We want a PSF that changes dramatically," graduate student Michael A. Thompson said during the symposium. By adding two extra lenses and a spatial light modulator to the optical path, the team generated a PSF with two lobes; orientation of the lobes depends on the axial position of the emitter. The PSF is shaped like a double helix in three dimensions, giving rise to the name double-helix PALM (Proc. Natl. Acad. Sci. USA 2009, 106, 2995). Using this method, they localized molecules with 20-nm precision in the x, y, and z directions over a 2-μm depth of field.
Yet another approach for achieving 3-D resolution involves what Alipasha Vaziri, a member of Charles V. Shank's group at Janelia Farm, called "temporal focusing" (Proc. Natl. Acad. Sci. USA 2008, 105, 20221). In this method, multiple layers of superresolution images are generated by selectively activating fluorescent proteins within 1.6-μm-thick layers by compressing ultrafast laser pulses in the axial direction. They achieve this compression by first broadening the light pulse with a diffraction grating and then imaging the spot on the grating onto the sample with a telescope. This resulting light pulse is broadened everywhere except in the focal plane, where molecules are activated via a two-photon absorption process. With the method, they imaged the mitochondrial matrix within human fibroblast cells.
Shank's group is also working on another technique called virtual volume PALM, which Jianyong Tang described during the symposium. In this technique, a tilted mirror generates a side view of the sample. In that view, the former axial dimension becomes a lateral dimension, allowing it to be analyzed with conventional PALM. The team is imaging bacteria and neuronal dendrites with this method, Tang said.
Still another 3-D PALM method is interferometric PALM (iPALM), developed by Gleb Shtengel in collaboration with Harald Hess, one of the original inventors of PALM (C&EN, Feb. 9, page 8; Proc. Natl. Acad. Sci. USA 2009, 106, 3125). The researchers achieve 3-D superresolution with a combination of two microscope objectives and a specially designed three-way beam splitter. These adaptations allow the sample fluorescence beam to serve as its own reference beam for interferometry.
iPALM is the one 3-D superresolution method with better resolution in the axial direction (10 nm) than in the lateral direction (20 nm). It is limited, however, to thin samples, Shtengel noted, because the interferometry doesn't work well with thick specimens.
This high axial resolution allows iPALM to resolve biological structures on the protein length scale, Shtengel said. They have already used iPALM to reveal information about the stratification of proteins in focal adhesions—macromolecular assemblies involved in structural and signaling regulation in cells.
Advances in 3-D imaging have also been made in the fluorescence microscopy technique STED. These were described by Andreas Schönle of Hell's group. In 2-D STED, one laser excites fluorescence, and a second laser shuts it down everywhere but in a tightly confined spot, making it possible to localize emitters with high accuracy. Hell and coworkers achieve 3-D resolution by adding another laser beam that squeezes the spot in the axial direction by using two objectives placed opposite one another. The resolution is better than 45 nm in all three dimensions (Nat. Meth. 2008, 5, 539). They have used this method to image the inner folds, called cristae, in mitochondria, Schönle said (Nano. Lett. 2009, 6, 2508).
These many advances in superresolution microscopy give researchers a range of options for collecting 3-D information. The main challenge right now, Bewersdorf noted, is to develop new probes that can be used with these methods.
- Chemical & Engineering News
- ISSN 0009-2347
- Copyright © American Chemical Society