Molecular Biology of the Cell by Bruce Alberts, Alexander Johnson, Julian Lewis, David Morgan, Martin Raff, Keith Roberts, Peter Walter by by Bruce Alberts, Alexander Johnson, Julian Lewis, David Morg

552 Chapter 9: Visualizing Cells
excitation spot
STED beam
effective fluorescence
spot
(A) (B) (C)
200 nm
(D)
(E)
250 nm
Figure 9–37 Superresolution microscopy can be achieved by reducing
the size of the point spread function. (A) The size of a normal focused
beam of excitatory light. (B) An extremely strong superimposed laser beam,
at a different wavelength and in the shape of a torus, depletes emitted
fluorescence everywhere in the specimen except right in the center of the
beam, reducing the effective width of the point spread function (C). As
the specimen is scanned, this small point spread function can then build
up a crisp image in a process called STED (stimulated emission depletion
microscopy). (D) Synaptic vesicles in live cultured neurons, fluorescently
labeled and imaged by ordinary confocal microscopy, with a resolution of
260 nm. (E) The same vesicles imaged by STED, with a resolution of 60
nm, which allows single vesicles to be resolved. (F) Fluorescently labeled
replication factories in the nucleus of a cultured cell, imaged by ordinary
confocal microscopy. (G) The same replication factories imaged by STED.
Single, discrete replication sites can be resolved by STED that cannot be
seen in the confocal image. (A, B, and C, from G. Donnert et al., Proc. Natl
Acad. Sci. USA 103:11440–11445, 2006. With permission from National
Academy of Sciences; D and E, from V. Westphal et al., Science 320:246–
249, 2008. With permission from AAAS; F and G, from Z. Cseresnyes,
U. Schwarz and C.M. Green, BMC Cell Biol. 10:88, 2009.)
(F)
(G)
2 µm
MBoC6 n9.235/9.37
of adjacent fluorescent molecules, as we saw earlier, is that they each contribute
blurry, overlapping point spread functions to the image, making the exact position
of any one molecule impossible to resolve. Another way round this limitation
is to arrange for only a very few, clearly separated molecules to actively fluoresce
at any one moment. The exact position of each of these can then be computed,
before subsequent sets of molecules are examined.
In practice, this can be achieved by using lasers to sequentially switch on a
sparse subset of fluorescent molecules in a specimen containing photoactivatable
or photoswitchable fluorescent labels. Labels are activated, for example, by illumination
with near-ultraviolet light, which modifies a small subset of molecules
so that they fluoresce when exposed to an excitation beam at another wavelength.
These are then imaged before bleaching quenches their fluorescence and a new
subset is activated. Each molecule emits a few thousand photons in response to
the excitation before switching off, and the switching process can be repeated
hundreds or even thousands of times, allowing the exact coordinates of a very
large set of single molecules to be determined. The full set can be combined and
digitally displayed as an image in which the computed location of each individual
molecule is exactly marked (Figure 9–38). This class of methods has been variously
termed photoactivated localization microscopy (PALM) or stochastic optical
reconstruction microscopy (STORM).
By switching the fluorophores off and on sequentially in different regions
of the specimen as a function of time, all the superresolution imaging methods
described above allow the resolution of molecules that are much closer together
than the 200 nm diffraction limit. In STED, the locations of the molecules are
determined by using optical methods to define exactly where their fluorescence
will be on or off. In PALM and STORM, individual fluorescent molecules are
switched on and off at random over a period of time, allowing their positions to
be accurately determined. PALM and STORM techniques have depended on the