Biotechnology
Journal
DOI 10.1002/biot.200900024
Biotechnol. J. 2009, 4, 846–857
846
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
Introduction
Light microscopy was hampered until recently by a
resolution limit that in principle did not allow the
localization of single molecules within a cell, de-
spite the myriad claims to the contrary.This resolu-
tion limit of about 200 nm, based on the work of
Abbe and Rayleigh [1] was recently overcome with
the development of a series of new microscopy
techniques that allow the localization of molecules
down to about 10-nm precision.What are the most
tractable biological questions that can be ad-
dressed with this new set of techniques in the com-
ing years? Given the limitations of existing fluores-
cent proteins, fluorescent dyes and probes for “su-
per-resolution,” is it possible for the common life
scientist to deploy and profit from the available
technologies? Recently, a number of super-resolu-
tion techniques have emerged that are within reach
of laboratories possessing a total internal reflection
fluorescence (TIRF) or laser-based wide-field mi-
croscopy setup and experience in quantitative im-
age analysis. Breaching Rayleigh’s limit vastly en-
hances our capabilities to image single molecules
since all the molecular “players” in biology, DNA,
RNA and protein, exist at a scale at least an order
of magnitude below the diffraction limit. Recent
work offers tantalizing views of where far-field
nanoscopy can take us.
1.1
The optical microscopy diffraction limit
When observing a single fluorophore on a micro-
scope, the emitted light must traverse several dif-
ferent physical mediums until it reaches the detec-
tor.As a result, light scatters throughout the differ-
ent environments leading to an artificial spatial
broadening of the discrete point. In the 19th centu-
ry Abbe described analytically this hard limit on
optical microscopy by which any point source of
light smaller than the diffraction limit of the imag-
ing system would have a fixed observable size [1].
Thus, the minimum distance where two different
points are regarded as resolvable, the Rayleigh lim-
Review
PALM and STORM: What hides beyond the Rayleigh limit?
Ricardo Henriques1 and Musa M. Mhlanga1,2
1Gene Expression and Biophysics Unit, Instituto de Medicina Molecular, Faculdade de Medicina Universidade de Lisboa, Lisbon,
Portugal
2Gene Expression and Biophysics Group, CSIR Synthetic Biology, Pretoria, South Africa
Super-resolution imaging allows the imaging of fluorescently labeled probes at a resolution of just
tens of nanometers, surpassing classic light microscopy by at least one order of magnitude. Re-
cent advances such as the development of photo-switchable fluorophores, high-sensitivity micro-
scopes and single particle localization algorithms make super-resolution imaging rapidly accessi-
ble to the wider life sciences research community. As we take our first steps in deciphering the
roles and behaviors of individual molecules inside their living cellular environment, a new world
of research opportunities beckons. Here we discuss some of the latest developments achieved
with these techniques and emerging areas where super-resolution will give fundamental new “eye”
sight to cell biology.
Keywords: Fluorescent proteins · Microscopy · Nanoscopy · Single molecule · Super-resolution
Correspondence: Dr. Musa Mhlanga, CSIR-Synthetic Biology, Gene Ex-
pression and Biophysics, Box 395, Pretoria 0001, South Africa
E-mail: mhlanga@gmail.com
Fax: +27-12-841-3651
Abbreviations: FPALM, fluorescence PALM; PALM, photoactivation local-
ization microscopy; PSF, point spread function; STED, stimulated emission
depletion; STORM, stochastic optical reconstruction microscopy; TIRF,
total internal reflection fluorescence
Received
1 February 2009
Revised
15 May 2009
Accepted 15 May 2009

----!@#$NewPage!@#$----
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
847
Biotechnol. J. 2009, 4, 846–857
www.biotechnology-journal.com
it, equals the space where the diffraction maximum
of one point image coincides with the first mini-
mum of the other. The points become indistin-
guishable if this distance is smaller while they re-
main resolvable if the distance between the two ob-
jects if greater. The observable spatial profile of
such spots defines the point spread function (PSF)
of the microscope, also known by the Airy diffrac-
tion pattern.
Any two fluorescent molecules whose overlap-
ping PSFs are separated by a distance smaller than
the PSF width become difficult or impossible to re-
solve as separate objects. This distance known as
the Rayleigh Criterion is given approximately by
λ/(2NA) laterally (x-y) and 2λη/(2NA)2 axially (z),
where λ is the wavelength of the emitted light, η is
the index of refraction of the medium and NA the
numerical aperture of the objective lens [2]. In a
conventional fluorescence microscope using visi-
ble light (λ between 450 and 700 nm) and a high nu-
merical aperture objective (NA=1.4) the resolution
limit given by the Rayleigh Criterion is approxi-
mately ~200 nm in the focal plane (x,y) and
500–800 nm along the optic axis (z).
1.2
From micro-to-nano
For several years electron microscopy (EM) has
been the predominant technique for high-resolu-
tion nanoscopy of biological samples [3], and has
had an immense significance on our understanding
of biology. Nonetheless, this technique has several
limitations such as low labeling efficiency and la-
borious sample preparation methods that are in-
compatible with live-cell imaging. As a result the
vast majority of microscopy research in the life sci-
ences is still carried out with optical light mi-
croscopy [4].
By default, fluorescence microscopes are able to
detect and position single molecules with a high ac-
curacy if they present distinct spectral emissions or
if their associated PSFs do not spatially overlap ex-
tensively [5, 6]. Problem arise when neighboring
markers are excited simultaneously stimulating a
coincident emission, making the separation of their
overlapping PSFs virtually impossible (Fig. 1).
One of the earliest techniques that attempted to
tackle this problem was near-field scanning optical
microscopy (NSOM) where the detection of
evanescent light waves emanating from a very re-
stricted number of molecules was achieved
through scanning with a nanosized tip [7]. Howev-
er, this technique is restricted to the imaging of the
sample surface and not able to give information
about the interior and contents of cells.
Far-field fluorescence nanoscopy arose by clev-
erly making use of Abbe’s formulas; one such way
was to reduce the PSF size, e.g., by increasing the
angular aperture of the imaging optics. These con-
cepts have been appraised since the late 1970s,
such as in the case of the first 4Pi illumination ideas
from the Cremer brothers in Heidelberg [8], lead-
ing to the development of the 4Pi confocal micro-
scope [9–11] or the wide-field approach I5M [12].
Both techniques take advantage of opposing objec-
tive lenses leading to up to a sevenfold increase in
the z-axis resolution. However, using two objec-
tives creates a wave front that is still non-spherical
giving rise to the appearance of unwanted side
lobes on the focal spot that need to be reduced
mathematically in post-processing steps [13]. An-
other approach has been achieved by structured il-
lumination microscopy (SIM) [14–16], where a
wide-field periodically patterned illumination al-
lows for the expansion of the frequency space de-
tectable by the microscopy reducing the size of the
PSF and allowing a resolution increase of up to a
factor of two [17]. The combination of this method
with I5M allows for a 3-D image resolution of
around 100 nm [18]. By implementing saturated in-
tensities, saturated SIM (SSIM) is able to use arbi-
trarily high spatial frequencies to generate an exci-
tation pattern [19].This enables SSIM to achieve an
experimentally measured 50-nm lateral resolution
[20]. Stimulated emission depletion (STED) fluo-
rescence microscopy makes use of two lasers; while
a focused excitation laser beam pushes fluo-
rophores to their excited state, a second laser with
Figure 1. Scheme of the distortion created by a fluorescence microscope. Using the words PALM and STORM as an example, as light travels from the emit-
ting molecules through the optics of the imaging system, diffraction occurs and the spatial information of the fluorophores becomes blurred. When a light
detector acquires this information, noise is introduced into the resulting produced image.

----!@#$NewPage!@#$----
Biotechnology
Journal
Biotechnol. J. 2009, 4, 846–857
848
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
a doughnut-shaped intensity profile prompts a
stimulated emission of the excited fluorophores
surrounding the excitation spot impelling them to
the ground state [10, 21]. This feature confines the
fluorescence emission to the non-stimulated re-
gion inside the doughnut reducing the size of the
PSF and thus achieving measured resolutions as
high as 20 nm laterally and 30–40 nm axially,. Res-
olutions of 40–45 nm can be achieved in any of the
three dimensions if STED is combined with a 4Pi
setup [22,23].Recently,STED has been shown to be
able to produce video-rate images of synaptic vesi-
cles with a resolution of 60 nm in live neuronal cells
[24].Thus, Abbe’s basic insight remains true and is
only overcome by “PSF engineering”.
A concept observed since the 1980s comes out of
the fact that, although the size of observable parti-
cles is limited by the resolution of the microscope,
the center of the particle can be determined pre-
cisely if a sufficient number of photons are detect-
ed, [25, 26]. Thus, unlimited resolution could even
be achieved if this number tends to infinity [27].
Notwithstanding, two main constraining factors in
achieving accurate particle localization exist: first-
ly, the PSFs of neighboring particle cannot overlap
and, secondly, the localization accuracy depends
heavily on the signal to noise ratio of the sampled
image. Despite these limitation it has still been pos-
sible to conduct far-field microscopy single-parti-
cle tracking studies with an astonishing precision
as high as 1 nm, as is the case of the work of Gelles
et al. [28] in the movement of kinesin-coated beads,
or Yildiz et al. [29] on the movement of myosin V
over actin filaments. However, in most instances,
biological-imaging experiments will encounter the
visualization of densely labeled samples. In such
cases the overlapping PSFs due to the emission of
the densely packed fluorophores prevent their ac-
curate localization (Fig. 1).
Researchers have been able to find clever meth-
ods to circumvent these constraints. In 1995, Eric
Betzig [30] suggested that one might be able to
identify individual molecules with differing spectra
whose separation is smaller than the PSF width.
Several groups then later showed the application of
these ideas in single molecule spectral selection
sub-diffraction imaging [31–33]. Other ingenious
approaches have also been developed such as sin-
gle-molecule localization through sequential pho-
tobleaching [34, 35] or through the analysis of the
stochastic blinking of quantum dots [36].
One of the solutions to these problems that has
become popular in the research community
emerges through the sequential and stochastic
switching on and off of fluorophores, therefore
minimizing the probability that at any given time
two or more fluorophore light emissions (and thus
PSFs) spatially overlap. In each imaging cycle most
molecules remain dark but a small number are ran-
domly switched on, imaged and localized. This
process can then be repeated for numerous itera-
tions until the majority of molecules have been ac-
curately detected and positioned (Fig. 2). A side ef-
fect of using this method means that thousands of
images may need to be acquired to generate one
super-resolution dataset. This feature has become
the backbone of techniques such as photoactiva-
tion localization microscopy (PALM) [37], fluores-
cence photoactivation localization microscopy
(FPALM) [38], stochastic optical reconstruction mi-
croscopy (STORM) [39] and PALM with independ-
ently running acquisition (PALMIRA) [40–42]. For
the remainder of this review we focus on this
method and how it has emerged as one of the most
tractable approaches for any laboratory wishing to
generate super-resolved data.
Figure 2. PALM and STORM imaging scheme. Several images are acquired where very few emitting molecules are observed in each. Although ultimately
images become blurred and noisy during acquisition, the low probability of overlapping fluorophores (and PSFs) allows for the detection and localization
of the majority of emitting particles inside each image. As this information is integrated in time, the localization of all detected molecules can be matched
and a final super-resolution dataset or image is generated.

----!@#$NewPage!@#$----
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
849
Biotechnol. J. 2009, 4, 846–857
www.biotechnology-journal.com
2
Super-resolution localization by stochastic
activation of single molecules
Three key components are required to achieve su-
per resolution with stochastic activation. They are
firstly the microscope optics and the detector that
must be able to detect very few photons. Secondly,
the fluorescent dyes and proteins that enable
photoswitching, permitting stochastic activation.
Thirdly, and perhaps not fully appreciated, analysis
of thousands of images and their reconstruction
into a super-resolved image.
2.1
Detection and microscope optics
As isolated emitting diffraction-limited spots are
imaged, their centroid can be localized with an ap-
proximate precision of σ/√⎯⎯N, where σ is the stan-
dard deviation of the PSF and N is the number of
detected photons [5]. This relationship reveals the
importance of the imaging system and fluo-
rophores used.To achieve high localization accura-
cy, one would want to achieve a small PSF by using
a high NA objective and optical components that
minimize the unwanted diffraction. Yet one of the
main problems is the low photon-detection effi-
ciency in microscopes, leading to the need for high-
sensitivity and low-noise detectors as is the case
with electron-multiplying charge-coupled devices
(EM-CCD) cameras and low absorption optics. La-
beled biological samples, however, can present a
large number of fluorophores in a diffraction-lim-
ited region preventing precise single-molecule lo-
calization. A solution based on photo-switchable
fluorescent molecules – molecules that can be re-
versibly switched from non-fluorescent (off-state)
to fluorescent (on-state) by stimulation of light
with specific wavelengths – can be created through
the use of activation-deactivation imaging cycles.
Importantly the density of activated molecules in
each single image must be kept very low, such that
the images (in this case PSFs) of individual fluo-
rophores do not overlap, permitting non-overlap-
ping single-particle detection. In each step of this
imaging process a small number of molecules are
activated, imaged and deactivated; repeating this
cycle creates a sequence of images in which most
of the photo-switchable population will appear se-
quentially distributed over a large number of tem-
poral slices, allowing for individual spot detection
and localization.
Imaging by itself can be easily implemented on
a TIRF or wide-field based system that features a
high magnification and numerical aperture objec-
tive (100× 1.45 NA or better), high-efficiency optics
and detector, and an excitation system that is com-
patible with the photoswitching activation and ex-
citation of the chosen fluorophore (see Table 1) –
minor modifications to the setup might be needed
to achieve 3-D information such as the introduction
of astigmatic lenses or a simultaneous double-
plane detection [43, 44].
2.2
Fluorescent proteins and dyes
For PALM and STORM, the choice of the right pho-
toswitchable fluorophores is vital as they are criti-
cal factors that modulate the speed of acquisition
and localization accuracy of the techniques. Desir-
able fluorescent molecules should be extremely
bright to allow for the detection of the maximum
possible amount of photons per molecule (N); this
can be achieved by selecting dyes/fluorescent pro-
teins that have both a large extinction coefficient
(ε) and quantum yield (QY). It is important to have
a good ratio between the amounts of photons emit-
ted in the fluorescent activated state and the resid-
ual photons emitted in the non-activated state. Ac-
quisition speed is highly dependent on the activa-
tion and deactivation rates of the fluorophores and
experimental settings need to be optimized to pre-
vent unwanted PSFs overlapping (this issue comes
to the forefront in specific biological examples de-
scribed later).Thus, single-molecule super-resolu-
tion can be achieved by maintaining a good balance
between the small number of fluorescent mole-
cules that are activated in each acquired frame and
the amount that do not immediately deactivate be-
tween consecutive frames. Several different class-
es of photoswitchable fluorophores compatible
with super-resolution localization have been de-
scribed (see Table 1).
Irreversible photoactivatable fluorescent pro-
teins can only make a single transition from a dark
to an emitting state. Deactivation is done by irre-
versibly bleaching the molecule. One of the earliest
developed and most successful fluorophores of this
class is PA-GFP [45], leading to the early work of
the Diaspro group on restricting the activation of
PA-GFP in 3-D space through 2-photon activation
[46] or the recent work of Hess and colleagues [47]
with live-cell FPALM. Due to its low contrast ratio
the spatial resolution is severely limited.
Irreversible photoshiftable fluorescent proteins
are able to make an irreversible shift between two
fluorescent colors. These proteins are some of the
most commonly used proteins for super-resolution
imaging. mKiKGR has been recently engineered
based on the KiKGR tetramer, making it a more
suitable alternative for cellular protein imaging
[48]. Of this class, PS-CFP2 is the only green-emit-

----!@#$NewPage!@#$----
Biotechnology
Journal
Biotechnol. J. 2009, 4, 846–857
850
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Table 1. Photoswitchable fluorophores for super-resolution imaginga)
Fluorophore
Switch
λ(act)
λ(ex)
λ(em)
ε
QY
N
Fluorescence 
Oligom. State
Reference (s)
Increase
Irreversible photoactivation fluorescent proteins
PA-GFP (Pre)
GG
413
400
515
20.700
0.13
~70
5
Monomer
[53, 58]
PA-GFP (Post)
504
517
17.400
0.79
~300
PAmCherry (Pre)
GG
405
ND
ND
ND
ND
ND
5000
Monomer
[59]
PAmCherry (Post)
570
596
24.000
0.53
ND
Irreversible photoshiftable fluorescent proteins
PS-CFP2 (Pre)
GG
405
400
468
43.000
0.20
ND
>2000
Monomer
[53, 58, 60]
PS-CFP2 (Post)
490
511
47.000
0.23
~260
Kaede (Pre)
GG
350–410
508
518
98.800
0.88
ND
2000
Tetramer
[53, 58]
Kaede (Post)
572
580
60.400
0.33
~460
mKiKGR (Pre)
GG
405
505
519
49.000
0.69
ND
>2000
Monomer
[48]
mKiKGR (Post)
580
591
28.000
0.63
~970
mEosFP (Pre)
GG
400
505
516
67.200
0.64
ND
ND
Monomer
[53, 58]
mEosFP (Post)
569
581
37.000
0.62
~490
tdEos (Pre)
GG
405
506
516
84.000
0.91
ND
200
Tandem dimer
[59, 61, 62]
tdEos (Post)
GG
569
581
33.000
0.62
~750
mEos2 (Pre)
GG
405
506
519
56.000
0.84
ND
ND
Monomer
[61]
mEos2 (Post)
GG
573
584
46.000
0.66
ND
Dendra-2 (Pre)
GG
405 or 488
490
507
45.000
0.50
ND
300
Monomer
[53, 59, 60]
Dendra-2 (Post)
553
573
35.000
0.55
ND
Reversible photoactivatable fluorescent proteins
Photoswitchable EYFP
GGGG
405
514
527
137.000
0.61
ND
ND
Monomer
[63, 64]
Dronpa
GGGG
405
503
522
125.000
0.68
120
ND
Monomer
[50, 53, 58]
bsDronpa
GGGG
405
460
504
45.000
0.50
ND
ND
Monomer
[50]
Padron
GGGG
496
503
522
43.000
0.64
ND
ND
Monomer
[50, 53]
rsFastLime
GGGG
405
496
518
46.000
0.60
>2000
ND
Monomer
[41, 50]

----!@#$NewPage!@#$----
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
851
Biotechnol. J. 2009, 4, 846–857
www.biotechnology-journal.com
ting protein and has successfully been used in mul-
ticolor super-resolution imaging [49].
Reversible photoactivatable fluorescent pro-
teins are capable of shifting between active and
non-active states multiple times, even though the
wavelength of the emission peaks of bsDronpa, rs-
FastLime and Padron are near each other. An-
dresen et al. [50] have shown that taking advantage
of the time-spectral photoswitching characteristics
of these fluorophores can be differentiated in mul-
ticolor imaging.
Reversible photoactivatable fluorescent dyes
are not genetically encoded and are unable to un-
dergo several shifts between a florescent and non-
fluorescent state. In comparison to their analog
photoswitchable proteins, the photon output (N) of
some of these probes is extremely high leading to
lateral resolutions in the order of 20 nm, as
achieved in STORM [51]. Although some of these
dyes are able to photoswitch alone, e.g., Cy5 [52],
when paired with a secondary chromophore
switching might be improved. This feature greatly
increases the color range of STORM probes [51].
In the case of caged fluorophores, irradiation
with UV light causes the release of a protective
group leading to a fluorescence increase of the dye.
Caging brings a novel way to create new photo-
switchable probes based on pre-existing fluo-
rophores with desirable photophysical properties
[53].
Fluorophores with reversible photobleaching
have been recently developed. It was shown that
embedding samples in oxygen-depleted mediums
such as polyvinyl alcohol, glucose oxidase or even
standard media such as glycerol, allows conven-
tional fluorophores (such as EGFP, EYFP, ECFP,
Alexa 488,Alexa 568, fluorescein among many oth-
ers) to switch into dark states after strong illumi-
nation. This allows them to stochastically switch
back to an emitting state after a relaxation time
[54–56].This resulted in a 30-nm resolution by im-
aging Alexa 488 [57].
2.3
Single molecule analysis and reconstruction
Analysis and reconstruction are one of the major
parts of PALM and STORM. Post-acquisition or in-
acquisition processing infer the characteristics of
each molecule present in the large amount of im-
ages that constitute the dataset. In a typical exper-
iment 1000–10 000 images are generated in which
tens to hundreds of particles are present in each
frame. Although the acquisition by itself takes few
minutes, the data analysis can take up to several
hours and is highly dependent on the processing
power available to the user. Ironically, this means
Table 1. Continued
Fluorophore
Switch
λ(act)
λ(ex)
λ(em)
ε
QY
N
Fluorescence 
Oligom. State
Reference (s)
Increase
Reversible photoactivatable fluorescent dyes
Photochromic Rhodamine
GGGG
375
530
620
105.000
0.65
~750
ND
NA
[40, 53]
Alexa Fluor 647*
GGGG
532*
650
665
240.000
0.33
~6000
ND
NA
[51, 53]
Cy5*
GGGG
405 or 457 or 532*
649
664
250.000
0.28
~6000
ND
NA
[51, 53]
Cy5.5*
GGGG
532*
675
694
190.000
0.23
~6000
ND
NA
[51, 53]
Cy7*
GGGG
532*
747
767
200.000
0.28
~1000
ND
NA
[51, 53]
Photocaged fluorophores
Caged Q-rhodamine
GG
405
545
575
90.000
0.90
ND
ND
NA
[53]
Caged carboxyfluorescein
GG
405
494
518
29.000
0.93
ND
ND
NA
[53]
a) λ(act), photoactivation wavelength in nm; λ(ex), excitation wavelength in nm; λ(em), emission wavelength in nm; ε, extinction coefficient in M-1cm-1; QY, fluorescence quantum yield; N, emitted number of photons per
molecule, Oligom. State, oligomeric state.
* Pairing this fluorophores with a second activator chromophore greatly increases the photoswitching characteristics, literature described pairs: Cy3-Alexa Fluor 647, Alexa Fluor 405-Cy5, Cy2-Cy5, Cy3-Cy5, Cy3-Cy5.5, Cy3-
Cy7 [51].

----!@#$NewPage!@#$----
Biotechnology
Journal
Biotechnol. J. 2009, 4, 846–857
852
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
that the super-resolved image cannot be “seen” un-
til well after the end of the actual acquisition and
imaging of the sample. Localization accuracy de-
pends greatly on the capacity of the algorithms to
deal with detection noise, PSF overlapping and
particle motion. Fortunately, biology can inherit a
great deal of knowledge and solutions from astron-
omy, as astronomers have faced a similar problem
when observing stars that are in essence also dif-
fraction limited. For example, Egner and colleagues
[41] describe the usage of a modified version of the
1974 Hogbom’s CLEAN algorithm [65] in conjunc-
tion with a mask-fitting algorithm of the Airy dif-
fraction pattern to provide particle segmentation
and positioning.These tasks become more difficult
when particle tracking is needed as in the case of
following molecular dynamics.
The single-particle tracking (SPT) algorithm
needs to relate the different particles being detect-
ed at different time points. Such tasks face several
challenges, such as particles disappearing due to
blinking, moving into or outside of the region being
imaged, particle splitting, particle merging and de-
tection failure [66, 67]. Recently, both groups of
Jaqaman et al. [67] and Serge et al. [68] have pro-
vided the research community with freely available
Matlab (Mathworks)-based algorithms able to
achieve localization accuracies near the theoretical
limit and multi-particle tracking.
As PALM and STORM evolve to live-cell imag-
ing, the need for in-acquisition real-time analysis
increases in order to optimize and adapt the micro-
scope settings.This would be the case in data-driv-
en excitation and activation optimization to reduce
the needed imaging time and the high phototoxici-
ty of PALM and STORM experiments. Also, this
would provide biologists with tools that allow ex-
perimental decisions based on the observed super-
resolution information immediately. It is clear that
as the technique becomes more generally used in
the near future, so too will algorithm optimization,
multi-processor analysis and clustered computing
evolve to provide these features.
3
Stepping into 3-D super-resolution
Although PALM and STORM have been generally
used to improve 2-D image resolution, achieving 3-
D intracellular information has been one of the ma-
jor challenges for the techniques. Other nanoscopy
techniques like 4Pi or I5M have been able to sur-
pass this problem and achieve an axial resolution
below 100 nm [12,13,69].STED microscopy in a 4Pi
illumination geometry has been shown to reach
resolutions as high as 30–50 nm, although without
super resolution in the lateral dimensions [70].The
recent work of Lemmer and colleagues on super-
resolving cellular nanostructures through the com-
bination of spectral precision distance microscopy
(SPDM) with spatially modulated illumination
(SMI) has achieved 20-nm lateral and 50-nm axial
resolution [71]. PALM and STORM have been
mostly used under TIRF imaging due to the inher-
ent high-contrast optical sectioning achieved with
such systems limiting the z axis imaging depth gen-
erated by the evanescent field to around 150 nm. A
similar technique called highly inclined and lami-
nated optical sheet (HILO) microscopy makes use
of a highly inclined light beam to generate a thin
optical sheet that penetrates the sample at a shal-
low angle, HILO has effectively been used in sin-
gle-molecule fluorescence nanoscopy of the cell
nucleus with high signal/background ratios [72]
and should prove to be compatible with PALM and
STORM imaging, but no association has been re-
ported to the date. Two-photon activation-based
optical sectioning with PALM has been achieved
with photochromic rhodamine dyes [40]. Recently,
two elegant solutions circumventing optical-beam
scanning have been demonstrated that allow 3-D
PALM and STORM imaging.Firstly,by inducing op-
tical astigmatism on a wide-field setup where the
PSF shape becomes ellipsoid and dependent on the
Z-position of the fluorophore – a trait that has been
shown to allow 20–30-nm lateral and 50–60-nm ax-
ial resolution [43]. Secondly, the use of a biplane
(BP) detection scheme where a 3-D stack of two
slightly displaced Z-planes are acquired simulta-
neously allowing for accurate z-position determi-
nation, leading to a experimentally measured
30-nm lateral and 75-nm axial resolution [44]. One
of the main problems of this latter technique comes
from the need to split the incoming emitted light
into two images, leading to a decrease of the de-
tected signal to noise ratio.
4
Imaging of molecular structures
The super-resolution method PALM first distin-
guished itself in the labeling of lysosomal trans-
membrane protein CD63 fused to Kaede (see
Table 1) [37].The resolution improvements offered
by PALM were adeptly illustrated when a trans-
mission EM and a PALM image of the same thin
cryosection were overlaid, resulting in a near per-
fect fit. In the same work, Betzig and coworkers
went on to label dimeric Eos (Table 1) with a cy-
tochrome C oxidase localization sequence,thus tar-
geting the mitochondria. In a comparison with
transmission EM images of the same mitochondria

----!@#$NewPage!@#$----
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
853
Biotechnol. J. 2009, 4, 846–857
www.biotechnology-journal.com
they were able to achieve highly similar structural
information and images. The power of this tech-
nique is clearly the nanometer resolution that is
achieved in determining the distribution of a given
protein within the cellular ultrastructure in a live
sample.
Early experiments using PALM to study cellular
ultrastructure focused on fixed cells or cryosec-
tions. However, being able to observe the assembly
of such structures in real time with super-resolu-
tion opens up several areas of scientific investiga-
tion under in vivo conditions.These include visulis-
ing how other cellular factors contribute to the as-
sembly and development of these processes with
high spatio-temporal resolution. To achieve this
level of resolution the sampling interval must be
half the desired resolution (i.e., the Nyquist criteri-
on must be met). Other important criteria must be
met for live-cell super-resolution. As acquisition
cycles typically require high laser powers for fluo-
rophore activation and bleaching, a significant
amount of phototoxicity is generated,leading to ab-
normal cell behavior or death – this becomes a
greater problem when UV-lasers are used for fluo-
rophore activation due to the high energy of the
photons used.To compensate for molecular motili-
ty and ensure that a fair number of molecules re-
main detected, a high frame rate needs to be used,
potentially leading to a decrease in the detected
signal to noise ratio and thus compromising the lo-
calization accuracy and resolution. Another limita-
tion arises from the limited number of detected
photons.As the overall number of molecules able to
photoswitch is depleted due to irreversible photo-
bleaching, the detected molecular density per
frame will decrease over time. Notwithstanding
these challenges, at least two groups have been
able to successfully achieve super resolution in live
cells. To image the assembly of ultrastructures,
Schroff et al. [62] tagged the protein paxillin with a
dimer of Eos.They were able to image the adhesion
complexes or transmembrane cytoskeleton-sub-
strate attachment points central to the process of
cell migration. They overcome the inherent prob-
lems of live-cell imaging in PALM using a fast im-
aging speed or image acquisition rate, which must
be faster than the activity imaged. In addition, they
compensated for photon scarcity by using a dimer
of Eos for each target protein, thereby achieving a
dense labeling of paxillin. Combining all these ap-
proaches enabled them to observe the formation of
individual adhesion complexes for extended peri-
ods of time, measuring the number of individual
paxillin molecules entering and leaving the com-
plex in real time over 25 min. Further, they were
able to note that, during cell migration, adhesion
complexes assume a wide diversity of morpholo-
gies.
4.1
Host-pathogen interactions
Because of the use of TIRF microscopy, the study of
plasma membrane-associated proteins with PALM
is particularly suited to imaging viral entry. Hess
and coworkers [47] have used FPALM, (essentially
identical to PALM) to examine a clustered viral
membrane protein, influenza hemagglutinin (HA).
Influenza viral entry and membrane fusion is me-
diated by HA and its visualization until recently
was limited to classical resolution limits. To enter
its host, the influenza virus uses HA to open a fu-
sion pore in the endosomal membrane and then in-
serts its viral RNA into the host cytoplasm. Lipid
rafts are believed to play an important role in the
distribution of HA at the cell surface, although sev-
eral theories exist on the nature and dynamics of
these rafts. Hess and coworkers [47] went on to use
FPALM to discriminate between the various raft
theories, which all propose that lipid rafts exist at
resolutions well below the diffraction limit of light.
As in the earlier work of Shroff and coworkers, they
were able to perform these experiments in live cells
principally because the diffusion coefficient of HA
was low, and thus a single molecule of HA tagged
with PA-GFP (Table 1) could be localized. Essen-
tially, for such experiments in live cells, to success-
fully localize a single molecule, the mean-squared
displacement attributable to two dimensional dif-
fusion (during acquisition) must be smaller than
r0
2, where r0 is the 1/e2 radius of the PSF. Most re-
cently STED microscopy has been used to address
at least one of these theories.
4.2
Super-resolution particle tracking
Advances in the use of photoactivatable markers
have enabled the dynamics of individual molecules
to be determined and thus tracked with super-res-
olution at the single-molecule level. Manley and
coworkers [73] were able to track the dynamics of
thousands of vesicular stomatitis virus G (VSVG)
and Gag (HIV-1 structural protein) proteins
(tagged with Eos) in the plasma membrane using
PALM. To do so, the data acquisition rate and cell
viability had to be optimized and a very high nu-
merical aperture lens (1.65 NA) was utilized. Using
what they termed sptPALM (single-particle track-
ing PALM), they were able to observe a relatively
homogenous localization of VSVG and the multi-
merization of the Gag particles into virus-like par-
ticles.They also could track single particles for path
lengths of up to ∼1 µm determining the mean

----!@#$NewPage!@#$----
Biotechnology
Journal
Biotechnol. J. 2009, 4, 846–857
854
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
square displacement within given molecular envi-
ronments. Using this information they created a
spatially resolved map of single molecule diffusion
coefficients for VSVG and Gag. The potential for
the use of sptPALM in the exploration of the be-
havior of different subsets of individual molecules
based on their spatial organization and subcellular
dynamics is enormous. Indeed, such information
could be correlated with different viral strains or
mutations as well as different cell types. Many ex-
isting imaging tools and reagents exist that permit
the use of sptPALM in the study of host-pathogen
interactions.
4.3
STORMing onto the scene
While PALM has been used by several groups to
image structural features in fixed and living cells,
STORM has been used similarly to image cellular
organelles and microtubules. Zhuang and cowork-
ers [51] were able to achieve dramatic improve-
ments in the resolution of the microtubule network
in fixed cells. Microtubules [labeled with antibod-
ies conjugated to Cy3 and Cy5 (Table 1)] separated
by 80 nm were clearly visible after 20 s of STORM
acquisition, and a theoretical resolution limit of
24 nm was possible within cells. Further, Zhuang
and coworkers were able to achieve multicolor su-
per-resolution in the fixed cells labeling clathrin-
coated pits,structures implicated in receptor-medi-
ated endocytosis,
simultaneously with micro-
tubules (Fig. 3). This was achieved using combina-
tions of Cy2 and Alexa 647 to label microtubules
and Cy3 and Alexa 647 to label clathrin-coated pits.
Extending these techniques to three dimensions,
they have recently used STORM to study the spa-
tial relationship between the cellular structures of
the mitochondria and microtubules, also in fixed
cells [43]. Their super-resolved 3-D images of the
mitochondria reveal it to have a hollow shape in the
outer membrane and identified two distinct mito-
chondrial morphologies, tubular and globular.
When co-imaged with microtubules, frequent con-
tacts between globular mitochondria and micro-
tubules were observed. The interactions of tubular
mitochondria and microtubules were found to be
more complex with noncontiguous interaction be-
tween the two.
STORM, at present, has the distinct disadvan-
tage of not permitting live-cell imaging due to the
buffer conditions that are necessary for photo-
switching of fluorescent dyes. It has, however, the
advantage of using dyes that have high contrast ra-
tios and are far brighter than fluorescent proteins.
STORM also does not require the construction of
fusion proteins, unlike PALM, although it does re-
quire the availability of antibodies to label the pro-
tein of interest. These advantages make it likely
that the ability to perform live-cell imaging with
STORM imaging will represent a significant ad-
vance in super resolution.
4.4
Stochasticity and the study of gene expression
Another area, which has an intense interest in ad-
vances in the dissolving of the resolution limits of
light microscopy,is the study of nuclear biology.The
spatial organization of the nucleus, the apparently
non-random organization of the genome within the
nucleus and several of the processes surrounding
transcription and gene expression are all questions
ripe for the use of super-resolution techniques
[74]. Current microscopy approaches, in live cells,
have focused on using statistical mapping to un-
derstand genome organization [75], and fluores-
cent microscopy techniques to visualize chromo-
some repositioning within the nucleus [76]. These
approaches have fallen short when it has come to
the study of gene expression and transcription, re-
sorting to fluorescent in situ hybridization (FISH)
to take snap shots of these events in fixed cells, al-
beit at times with single-molecule resolution [77].
Figure 3. STORM imaging of microtubules and clathrin-coated pits. (A–C) BS-C-1 cells: secondary antibodies in microtubule staining are labeled with Cy2
and Alexa 647 (false color green), those for clathrin are labeled with Cy3 and Alexa 647 (false color red). These images demonstrate the level of magnifica-
tion the STORM technique and others such as PALM. (B) Magnification of the boxed section in (A), (C) further magnification of boxed region in (B). Im-
ages modified, with permission from [51] © (2007) American Association for the Advancement of Science.

----!@#$NewPage!@#$----
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
855
Biotechnol. J. 2009, 4, 846–857
www.biotechnology-journal.com
Gene expression is a fundamentally stochastic
process, and in some instances appears purpose-
fully so (reviewed in [78]). This makes the con-
struction and modeling of gene networks especial-
ly difficult. Transcription is thought to occur in
bursts with little known as to their source [79]. One
may be the existence of pre-initiation complexes
formed on the promoter that may permit multiple
rounds of RNA polymerase II transcription [80, 81].
Such complexes could form in so-called “transcrip-
tion factories” where active genes would be recruit-
ed. Some evidence for transcription factories does
exist though at best three differing transcripts have
been imaged in putative factories [82]. Where su-
per-resolution could be most informative in this
field is in the ability to image all the species tran-
scripts within these factories and to understand the
contents of the transcriptional machinery in real
time. This would give key insights into the funda-
mental process of transcription whose “molecular
players” have only been observed as snapshots.
Since there is great heterogeneity in transcription
in even otherwise identical cell types,observing the
minutiae of this process over several cells is per-
haps the only way stochasticity can be understood.
We thank Christophe Zimmer, Mickael Lelek for
valuable comments and help in preparation of
Table 1. We also thank Xiaowei Zhuang for informal
advice and interactions.
The authors have declared no conflict of interest.
References
[1] Abbe, E., Beiträge zur Theorie des Mikroskops und der
mikroskopischen Wahrnehmung. Arch. Mikroskop. Anat.
1873, 9, 413–420.
[2] Pawley, J., Handbook of Biological Confocal Microscopy,
Springer, New York 2006.
[3] Koster,A., Klumperman, J., Electron microscopy in cell biol-
ogy: Integrating structure and function. Nat. Rev. Mol. Cell
Biol. 2003, Supp. SS6–10.
[4] Hell, S., Far-field optical nanoscopy. Science 2007, 316, 1153.
[5] Thompson, R. E., Larson, D. R., Webb, W. W., Precise
nanometer localization analysis for individual fluorescent
probes. Biophys. J. 2002, 82, 2775–2783.
[6] Basché, T., Single-Molecule Optical Detection, Imaging and
Spectroscopy, John Wiley & Sons, New York 1997.
[7] Novotny, L., Hecht, B., Principles of Nano-Optics, Cambridge
University Press, Cambridge 2006.
[8] Cremer, C., Cremer, T., Considerations on a laser-scanning-
microscope with high resolution and depth of field. Microsc.
Acta 1978, 81, 31–44.
[9] Hell, S., Stelzer, E. K., Fundamental improvement of resolu-
tion with a 4Pi-confocal fluorescence microscope using
two-photon excitation. Opt. Commun. 1992, 93, 277–282.
[10] Hell, S., Wichmann, J., Breaking the diffraction resolution
limit by stimulated emission: Stimulated-emission-deple-
tion fluorescence microscopy. Opt. Lett. 1994, 19, 780–782.
[11] Hanninen, P. E., Hell, S.W., Salo, J., Soini, E., Cremer, C.,Two-
photon excitation 4Pi confocal microscope: Enhanced axial
resolution microscope for biological research. Appl. Phys.
Lett. 1995, 66, 1698–1700.
[12] Gustafsson, M. G. L.,Agard, D.A., Sedat, J.W., Sevenfold im-
provement of axial resolution in 3D widefield microscopy
using two objective lenses. Proc. SPIE Int. Soc. Opt. Eng.
1995, 2412, 147–158.
[13] Egner, A., Hell, S., Fluorescence microscopy with super-re-
solved optical sections. Trends Cell Biol. 2005, 15, 207–215.
[14] Bailey, B., Farkas, D., Taylor, D., Lanni, F., Enhancement of
axial resolution in fluorescence microscopy by standing-
wave excitation. Nature 1993, 366, 44–48.
[15] Gustafsson, M., Surpassing the lateral resolution limit by a
factor of two using structured illumination microscopy. J.
Microsc. 2000, 198, 82–87.
Ricardo Henriques received his Physics
Engineering degree in 2005 from the
Faculty of Sciences University of Lis-
bon, Portugal and trained further in mi-
croscopy at the cell imaging facility of
the Instituto Gulbenkian de Ciências,
Oeiras. In 2006 he took a position as
manager of the BioImaging facility in
the Institute of Molecular Medicine,
Lisbon, Portugal, researching new data-
driven adaptive microscopy methods through the use of artificial vi-
sion and robotics. He has been a PhD student in the laboratory of 
Dr. Musa Mhlanga since 2008. His research interest is focused on cre-
ating technological tools for sub-diffraction visualization of the nuclear
architecture and its re-organization driven by gene expression activa-
tion.
Musa Mhlanga received his PhD in
Cell Biology & Molecular Genetics
from New York University, School of
Medicine, in 2003 where he focused on
the visualization of mRNA in living
cells. He went on to do his postdoc at
the Institut Pasteur in Paris, France
where he worked on imaging gene ex-
pression in the nucleus and RNA
transport. Since 2008 he has been a
Research Group Leader at the CSIR in
Pretoria, South Africa and holds an ad-
junct position at the Institute of Molecular Medicine in Lisbon, Portu-
gal. Research in his group spans the disciplines of "super" resolution
imaging and image analysis/signal processing, the study and imaging
of gene expression, and the development of cell-based bioassays for
screening.

----!@#$NewPage!@#$----
Biotechnology
Journal
Biotechnol. J. 2009, 4, 846–857
856
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
[16] Schermelleh, L., Carlton, P., Haase, S., Shao, L. et al., Subdif-
fraction multicolor imaging of the nuclear periphery with
3D structured illumination microscopy. Science 2008, 320,
1332.
[17] Langhorst, M. F., Schaffer, J., Goetze, B., Structure brings
clarity: Structured illumination microscopy in cell biology.
Biotechnol.J. 2009,4,this issue DOI:10.1002/biot.200900025.
[18] Shao, L., Isaac, B., Uzawa, S., Agard, D. et al., I5S:Wide-field
light microscopy with 100-nm-scale resolution in three di-
mensions. Biophys. J. 2008, 94, 4971–4983.
[19] Heintzmann, R., Jovin, T., Cremer, C., Saturated patterned
excitation microscopy—S concept for optical resolution im-
provement. J. Opt. Soc. Am. A 2002, 19, 1599–1609.
[20] Gustafsson, M., Nonlinear structured-illumination mi-
croscopy:Wide-field fluorescence imaging with theoretical-
ly unlimited resolution. Proc. Natl. Acad. Sci. USA 2005, 102,
13081–13086.
[21] Schrader, M., Meinecke, F., Bahlmann, K., Kroug, M. et al.,
Monitoring the excited state of a fluorophore in a micro-
scope by stimulated emission. Bioimaging 1995, 3, 147–153.
[22] Klar, T. A., Jakobs, S., Dyba, M., Egner, A., Hell, S. W., Fluo-
rescence microscopy with diffraction resolution barrier
broken by stimulated emission. Proc. Natl. Acad. Sci. USA
2000, 97, 8206–8210.
[23] Schmidt, R.,Wurm, C., Jakobs, S., Engelhardt, J. et al., Spher-
ical nanosized focal spot unravels the interior of cells. Nat.
Methods 2008, 5, 539–544.
[24] Nan, X., Sims, P., Xie, X., Organelle tracking in a living cell
with microsecond time resolution and nanometer spatial
precision. ChemPhysChem 2008, 9, 707–712.
[25] Burns, D. H., Callis, J. B., Christian, G. D., Davidson, E. R.,
Strategies for attaining superresolution using spectroscop-
ic data as constraints. Appl. Opt. 1985, 24, 154–161.
[26] Bobroff, N., Position measurement with a resolution and
noise-limited instrument. Rev. Sci. Instrum. 1986, 57,
1152–1157.
[27] Hell, S., Soukka, J., Hanninen, P., Two-and multiphoton de-
tection as an imaging mode and means of increasing the
resolution in far-field light microscopy: A study based on
photon-optics. Bioimaging 1995, 3, 64-69.
[28] Gelles, J., Schnapp, B., Sheetz, M., Tracking kinesin-driven
movements with nanometre-scale precision. Nature 1988,
331, 450–453.
[29] Yildiz, A., Forkey, J., McKinney, S., Ha, T. et al., Myosin V
walks hand-over-hand: Single fluorophore imaging with
1.5-nm localization. Science 2003, 300, 2061–2065.
[30] Betzig, E., Proposed method for molecular optical imaging.
Opt. Lett. 1995, 20, 237–239.
[31] Van Oijen, A., Köhler, J., Schmidt, J., Müller, M., Brakenhoff,
G., 3-Dimensional super-resolution by spectrally selective
imaging. Chem. Phys. Lett. 1998, 292, 183–187.
[32] van Oijen, A. M., Kohler, J., Schmidt, J., Muller, M., Braken-
hoff, G. J., Far-field fluorescence microscopy beyond the dif-
fraction limit. J. Opt. Soc. Am. A 1999, 16, 909–915.
[33] Lacoste,T. D., Michalet, X., Pinaud, F., Chemla, D. S. et al., Ul-
trahigh-resolution multicolor colocalization of single fluo-
rescent probes. Proc. Natl. Acad. Sci. USA 2000, 97,
9461–9466.
[34] Gordon, M. P., Ha,T., Selvin, P. R., Single-molecule high-res-
olution imaging with photobleaching. Proc. Natl. Acad. Sci.
USA 2004, 101, 6462–6465.
[35] Qu, X. H., Wu, D., Mets, L., Scherer, N. F., Nanometer-local-
ized multiple single-molecule fluorescence microscopy.
Proc. Natl. Acad. Sci. USA 2004, 101, 11298–11303.
[36] Lidke, K. A., Rieger, B., Jovin, T. M., Heintzmann, R., Super-
resolution by localization of quantum dots using blinking
statistics. Opt. Express. 2005, 13, 7052–7062.
[37] Betzig, E., Patterson, G., Sougrat, R., Lindwasser, O. et al.,
Imaging intracellular fluorescent proteins at nanometer
resolution. Science 2006, 313, 1642–1645.
[38] Hess, S., Girirajan, T., Mason, M., Ultra-high resolution im-
aging by fluorescence photoactivation localization mi-
croscopy. Biophys. J. 2006, 91, 4258.
[39] Rust, M., Bates, M., Zhuang, X., Sub-diffraction-limit imag-
ing by stochastic optical reconstruction microscopy
(STORM). Nat. Methods 2006, 3, 793–796.
[40] Fölling, J., Belov, V., Kunetsky, R., Medda, R. et al., Pho-
tochromic rhodamines provide nanoscopy with optical sec-
tioning. Angew. Chem. Int. Ed 2007, 46, 6266–6270.
[41] Egner, A., Geisler, C., von Middendorff, C., Bock, H. et al.,
Fluorescence nanoscopy in whole cells by asynchronous lo-
calization of photoswitching emitters. Biophys. J. 2007, 93,
3285.
[42] Bock, H., Geisler, C., Wurm, C., von Middendorff, C. et al.,
Two-color far-field fluorescence nanoscopy based on pho-
toswitchable emitters. Appl. Phys. B Lasers Opt. 2007, 88,
161–165.
[43] Huang, B., Wang, W., Bates, M., Zhuang, X., Three-dimen-
sional super-resolution imaging by stochastic optical recon-
struction microscopy. Science 2008, 319, 810–813.
[44] Juette, M., Gould, T., Lessard, M., Mlodzianoski, M. et al.,
Three-dimensional sub–100 nm resolution fluorescence
microscopy of thick samples. Nat. Methods 2008, 5, 527–529.
[45] Patterson, G., Lippincott-Schwartz, J., A photoactivatable
GFP for selective photolabeling of proteins and cells. Sci-
ence 2002, 297, 1873–1877.
[46] Schneider, M., Barozzi, S., Testa, I., Faretta, M., Diaspro, A.,
Two-photon activation and excitation properties of PA-GFP
in the 720–920-nm region. Biophys. J. 2005, 89, 1346–1352.
[47] Hess, S. T., Gould, T. J., Gudheti, M. V., Maas, S. A. et al., Dy-
namic clustered distribution of hemagglutinin resolved at
40 nm in living cell membranes discriminates between raft
theories. Proc. Natl. Acad. Sci. USA 2007, 104, 17370–17375.
[48] Habuchi, S., Tsutsui, H., Kochaniak, A. B., Miyawaki, A., van
Oijen, A. M., mKikGR, a monomeric photoswitchable fluo-
rescent protein. PLoS ONE 2008, 3, e3944.
[49] Shroff, H., Galbraith, C. G., Galbraith, J. A., White, H. et al.,
Dual-color superresolution imaging of genetically ex-
pressed probes within individual adhesion complexes.Proc.
Natl. Acad. Sci. USA 2007, 104, 20308–20313.
[50] Andresen, M., Stiel, A. C., Folling, J.,Wenzel, D. et al., Photo-
switchable fluorescent proteins enable monochromatic
multilabel imaging and dual color fluorescence nanoscopy.
Nat. Biotechnol. 2008, 26, 1035–1040.
[51] Bates, M., Huang, B., Dempsey, G. T., Zhuang, X., Multicolor
super-resolution imaging with photo-switchable fluores-
cent probes. Science 2007, 317, 1749–1753.
[52] Heilemann, M., van de Linde, S., Schuttpelz, M., Kasper, R.
et al., Subdiffraction-resolution fluorescence imaging with
conventional fluorescent probes. Angew. Chem. Int. Ed. Engl.
2008, 47, 6172–6176.
[53] Fernandez-Suarez, M., Ting, A. Y., Fluorescent probes for
super-resolution imaging in living cells. Nat. Rev. Mol. Cell
Biol. 2008, 9, 929–943.
[54] Sinnecker, D., Voigt, P., Hellwig, N., Schaefer, M., Reversible
photobleaching of enhanced green fluorescent proteins.
Biochemistry 2005, 44, 7085–7094.

----!@#$NewPage!@#$----
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
857
Biotechnol. J. 2009, 4, 846–857
www.biotechnology-journal.com
[55] Folling, J., Bossi, M., Bock, H., Medda, R. et al., Fluorescence
nanoscopy by ground-state depletion and single-molecule
return. Nat. Methods 2008, 5, 943–945.
[56] Reymann, J., Baddeley, D., Gunkel, M., Lemmer, P. et al.,
High-precision structural analysis of subnuclear complex-
es in fixed and live cells via spatially modulated illumina-
tion (SMI) microscopy. Chromosome Res. 2008, 16, 367–382.
[57] Baddeley, D., Jayasinghe, I. D., Cremer, C., Cannell, M. B.,
Soeller, C., Light-induced dark states of organic flu-
ochromes enable 30 nm resolution imaging in standard me-
dia. Biophys. J. 2009, 96, L22–24.
[58] Lukyanov, K. A., Chudakov, D. M., Lukyanov, S., Verkhusha,
V.V., Innovation: Photoactivatable fluorescent proteins. Nat.
Rev. Mol. Cell Biol. 2005, 6, 885–891.
[59] Subach, F.V., Patterson, G. H., Manley, S., Gillette, J. M. et al.,
Photoactivatable mCherry for high-resolution two-color
fluorescence microscopy. Nat. Methods 2009, 6, 153–159.
[60] Chudakov, D. M., Lukyanov, S., Lukyanov, K. A.,Tracking in-
tracellular protein movements using photoswitchable fluo-
rescent proteins PS-CFP2 and Dendra2. Nat. Protoc. 2007,
2, 2024–2032.
[61] McKinney, S. A., Murphy, C. S., Hazelwood, K. L., Davidson,
M.W., Looger, L. L.,A bright and photostable photoconvert-
ible fluorescent protein. Nat. Methods 2009, 6, 131–133.
[62] Shroff, H., Galbraith, C. G., Galbraith, J. A., Betzig, E., Live-
cell photoactivated localization microscopy of nanoscale
adhesion dynamics. Nat. Methods 2008, 5, 417–423.
[63] McAnaney, T. B., Zeng, W., Doe, C. F., Bhanji, N. et al., Proto-
nation, photobleaching, and photoactivation of yellow fluo-
rescent protein (YFP 10C): A unifying mechanism. Bio-
chemistry 2005, 44, 5510–5524.
[64] Biteen, J.,Thompson, M.,Tselentis, N., Bowman, G. et al., Su-
per-resolution imaging in live Caulobacter crescentus cells
using photoswitchable EYFP. Nat. Methods 2008.
[65] Högbom, J.,Aperture synthesis with a non-regular distribu-
tion of interferometer baselines. Astron. Astrophys. Suppl.
1974, 15, 417.
[66] Kalaidzidis,Y., Intracellular objects tracking. Eur. J. Cell Biol.
2007, 86, 569–578.
[67] Jaqaman, K., Loerke, D., Mettlen, M., Kuwata, H. et al., Ro-
bust single-particle tracking in live-cell time-lapse se-
quences. Nat. Methods 2008, 5, 695–702.
[68] Serge, A., Bertaux, N., Rigneault, H., Marguet, D., Dynamic
multiple-target tracing to probe spatiotemporal cartogra-
phy of cell membranes. Nat. Methods 2008, 5, 687–694.
[69] Nagorni, M., Hell, S., 4Pi-confocal microscopy provides
three-dimensional images of the microtubule network with
100-to 150-nm resolution. J. Struct. Biol. 1998, 123, 236–247.
[70] Hell, S., Toward fluorescence nanoscopy. Nat. Biotechnol.
2003, 21, 1347–1355.
[71] Lemmer, P., Gunkel, M., Baddeley, D., Kaufmann, R. et al.,
SPDM: Light microscopy with single-molecule resolution at
the nanoscale. Appl. Phys. B Lasers Opt. 2008, 93, 1–12.
[72] Tokunaga, M., Imamoto, N., Sakata-Sogawa, K., Highly in-
clined thin illumination enables clear single-molecule im-
aging in cells. Nat. Methods 2008, 5, 159–161.
[73] Manley, S., Gillette, J. M., Patterson, G. H., Shroff, H. et al.,
High-density mapping of single-molecule trajectories with
photoactivated localization microscopy. Nat. Methods 2008,
5, 155–157.
[74] Gunkel, M., Erdel, F., Rippe, K., Lemmer, P. et al., Dual color
localization microscopy of cellular nanostructures. Biotech-
nol. J. 2009, 4, this issue DOI: 10.1002/biot.200900005.
[75] Berger, A. B., Cabal, G. G., Fabre, E., Duong, T. et al., High-
resolution statistical mapping reveals gene territories in
live yeast. Nat. Methods 2008, 5, 1031–1037.
[76] Cabal, G. G., Genovesio, A., Rodriguez-Navarro, S., Zimmer,
C. et al., SAGA interacting factors confine sub-diffusion of
transcribed genes to the nuclear envelope. Nature 2006, 441,
770–773.
[77] Raj, A., van den Bogaard, P., Rifkin, S. A., van Oudenaarden,
A., Tyagi, S., Imaging individual mRNA molecules using
multiple singly labeled probes. Nat. Methods 2008, 5,
877–879.
[78] Raj, A., van Oudenaarden, A., Nature, nurture, or chance:
Stochastic gene expression and its consequences. Cell 2008,
135, 216–226.
[79] Raj,A.,Peskin,C.S.,Tranchina,D.,Vargas,D.Y.,Tyagi,S.,Sto-
chastic mRNA synthesis in mammalian cells. PLoS Biol.
2006, 4, e309.
[80] Blake, W. J., Balazsi, G., Kohanski, M. A., Isaacs, F. J. et al.,
Phenotypic consequences of promoter-mediated transcrip-
tional noise. Mol. Cell 2006, 24, 853–865.
[81] Blake, W. J., M, K. A., Cantor, C. R., Collins, J. J., Noise in eu-
karyotic gene expression. Nature 2003, 422, 633–637.
[82] Osborne, C. S., Chakalova, L., Brown, K. E., Carter, D. et al.,
Active genes dynamically colocalize to shared sites of ongo-
ing transcription. Nat. Genet. 2004, 36, 1065–1071.