Superresolution imaging with switchable fluorophores based on oxazine auxochromes

Article abstract of DOI:10.1111/php.12100

The spatial resolution of fluorescence microscopes is limited by diffraction to about half of the light wavelength, hampering the observation of many important intracellular processes. Recent emerging techniques have overcome that diffraction barrier usin

Full text of DOI:10.1111/php.12100

Photochemistry and Photobiology, 20**, **: *–*
Superresolution Imaging with Switchable Fluorophores Based on Oxazine
Auxochromes†
Marco Petriella1, Erhan Deniz2, Subramani Swaminathan2, Maria J. Roberti1, Francßisco M. Raymo*2 and
Mariano L. Bossi*1
1
ꢀ
Laboratorio de Nanoscopías Fotonicas, INQUIMAE – DQIAyQF (FCEyN), Universidad de Buenos Aires & Conicet, Buenos
Aires, Argentina
2Department of Chemistry, Laboratory for Molecular Photonics, University of Miami, Miami, FL
Received 15 March 2013, accepted 7 May 2013, DOI: 10.1111/php.12100
ABSTRACT
imaging paths work by focusing light through regular lenses, a
process restricted by diffraction that imposes a limit to the spatial
resolution (DrDL), the minimal distance that two identical objects
can be distinguished. This is the so-called diffraction barrier, and
it is related to the wavelength of the light (k) and the numerical
aperture of the imaging lens (NA) by Eq. (1):
The spatial resolution of fluorescence microscopes is limited
by diffraction to about half of the light wavelength, ham-
pering the observation of many important intracellular pro-
cesses. Recent emerging techniques have overcome that
diffraction barrier using the temporal discrimination of
close objects that are otherwise unresolved or blurred
within the spatial resolution of the microscope. The key of
these techniques is to switch the signal of fluorescence
markers on and off exploiting their distinct molecular
states, and detect and localize these markers at the single-
molecule level. This underlying principle highlights the crit-
ical role of the photophysical properties of the probes, and
the importance of finding adequate switching mechanisms.
Here, we present strategies to achieve fluorescence modula-
tion based on novel molecular assemblies containing a [1,3]
oxazine as the two states, building block responsible for
the transformation. Two different triggering events, based
on the photochromic and halochromic properties of the ox-
azine, induce a large absorption and emission bathochro-
0:61 k
NA
DrDL
ꢀ
ð1Þ
The visible wavelengths involved in the fluorescence process
and the best objective lenses available (NA ~ 1.4) set a value for
DrDL to about half of the wavelength (ca 200–300 nm). Thus,
typical subcellular compartments with sizes below DrDL remain
unresolved in conventional microscopies and important biological
processes are impossible to be observed at the (bio)molecular
level.
Modern and emerging techniques, such as STED (2),
PALM (3), STORM (4), GSDIM (5), and dSTORM (6), have
taken advantage of the molecular states of the fluorescent
probes not only for signal generation but also for overcoming
the limits set by diffraction (7,8). From the first successful
realizations (9) to the present day, all the different techniques
share in common the concept of switching the markers
between an emissive and a dark state (E↔D) (10). The key is
to switch the signal of the fluorescent probes on and off to
achieve a temporal discrimination of subdiffraction-separated
objects, i.e. features within a diffraction spot in the sample.
One of these methods is based upon the detection and locali-
zation of single molecules (SMLM, single-molecule localiza-
tion microscopies) (11). The imaging process consists of a
simple sequence of steps (Scheme 1). Initially all markers
should be in the dark state (D) or switched off with light of
an appropriate wavelength (kOFF). Then a few of them are
switched on (D?E) in random positions, usually by illumina-
tion with light of a defined wavelength kON, to produce a
sparse subset of molecules in the bright state. It is critical at
this stage that each emitter can be resolved with a conven-
tional wide-field (WF) microscope (i.e. average distance
between emitters found simultaneously in the bright state >>
DrDL). An image is recorded under illumination with kEX to
elicit fluorescence, from which the position of each emitter
can be precisely obtained after proper image analysis with an
accuracy Dxy Eq. (2):(12)
mic shift of
a pendant fluorophore, as the ultimate
fluorescence switching event. The implementation of these
approaches to achieve spatial resolution beyond the diffrac-
tion limit is also discussed.
INTRODUCTION
Far-field fluorescence microscopy is a widely applied and power-
ful tool for biological imaging that provides noninvasive 3D
visualization of cells and tissues with a unique combination of
high selectivity and sensitivity, and fast temporal resolution. A
conventional fluorescence microscope is a system of several
lenses that produce a magnified image of the observation object,
usually of micrometer or even submicrometer size (1). It relies
on the excitation of intrinsic or extrinsic fluorescent markers,
introduced through proper labeling protocols, and the collection
of the signal they produce. Both the illumination and the
Corresponding author email: mariano@qi.fcen.uba.ar (Mariano L. Bossi); fraymo@
miami.edu (Francisco M. Raymo)
†This article is part of the Special Issue dedicated to the memory of Elsa Abuin.
© 2013 The American Society of Photobiology
1

2
Marco Petriella et al.
Scheme 1. Superresolution imaging with SMLM. The crosses represent a series of markers labeling an object. Initially (a) all are driven to a dark state,
and a few are switched on to record a camera frame (b). Each emitting molecule appears with a size determined by the standard resolution of the micro-
scope DrDL. Then, those molecules are switched off (c), to repeat the process and acquire the next frame (d). Software analysis is applied to each frame
(f, g) to determine the position of the probes in the bright state, with a localization accuracy improved. The superresolution image is obtained by map-
ping the position of all localized markers (h). Superposition of all frames renders a wide-field image with diffraction-limited resolution (e).
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Fluorescent proteins, both photoswitchable and photoactivat-
4
Dxy¼
2 þ
þ
ꢀ DrDL ꢁðNPHÞꢂ1=2 ð2Þ
ðDrDL
Þ
a2
12NPH
8pðDrDLÞ b2
able, are a common family of markers for SMLM (14). Their
unique advantage is that they are genetically encoded. However,
many of them exhibit low brightness or slow switching kinetics,
and improvement of these key properties may require an arduous
work of molecular cloning and engineering (15). In contrast, the
power of chemical synthesis can be extended with relative simplic-
ity to organic fluorophores to prepare molecular switches with
superior qualities (16). Properly designed organic markers are
brighter, present faster, more reliable switching and negligible
cross-talk (defined as inducing activation with excitation light and
vice versa) (14). Moreover, the size of the probe is becoming an
additional issue to take into account because resolutions on the
order of the size of biomolecules are now at reach. In this regard,
the average size of organic dyes is considerably smaller compared
with that of fluorescent proteins, leading to higher labeling densi-
ties and thus to a better spatial resolution—to reach a certain reso-
lution value, at least two labels should be placed within that
distance, according to the Nyquist criterion (17). Despite several
organic markers have been used in single-molecule localization mi-
croscopies, producing excellent images with high resolution
(<20 nm), most of them rely on a few switching mechanisms
(13,14,17). A frequent mechanism is the reversible interconversion
between the emissive first singlet excited state and the correspond-
ing nonemissive triplet state (5), or radicals formed from the triplet
(18), under illumination. These photoinduced transformations arise
from the transient population of the ground singlet state and are
observed as fluorescence blinking. A clear disadvantage of blink-
ing microscopy is the need of higher light intensities and the use of
chemical additives, usually a cocktail of redox species (e.g. ascor-
bic acid, methyl viologen and thiols such as cysteamine or b-mer-
captoethanol) and oxygen scavengers (e.g. enzymatic systems).
A cumbersome optimization is usually required for each dye, and
they may be incompatible with many samples and measuring
conditions. Another mechanism is the use of caged fluorescent
compounds (19,20,21). These molecules switch irreversibly from a
nonemissive to an emissive state after the photoinduced cleavage
of a protecting group. Such irreversible photoactivation events,
however, prevent the localization of the same marker more than
2
a2ðNPH
Þ
NPH
where NPH is the number of detected photons, a is the pixel size
in nanometers and b is the background noise in photons. If an
adequate pixelation is selected (i.e. ꢀ DrDL), and the background
is low, Dxy can be approximated with the right term in Eq. (2),
depending only on the detected number of photons and the reso-
lution of the microscope. Then, those markers are switched off
again (E?D) so the process is repeated by switching on other
markers. The superresolution image is reconstructed by mapping
the position of a large number of localized events, typically from
several tens of thousands of frames. It must be remarked that the
transition from a diffraction-limited image of resolution DrDL to
a subdiffraction one with a resolution DrSR (equal to Dxy), is
achieved solely by the use of photoswitchable fluorophores
instead of conventional ones.
The modulation of the fluorescence signal can also be per-
formed with a three-state irreversible molecular system. In this
case, makers in the bright state are switched off to another dark
state (i.e. different from the initial one) through a photoinduced
bleaching process (D?E?D*). To distinguish these kinds of
probes, reversible systems are usually referred to as photo-
switchable markers and irreversible ones as photoactivatable
markers.
The imaging scheme outlines the critical role of the molecular
switches used as fluorescent probes (13), in particular the photo-
physical properties of both states and the processes responsible
for the control of signal modulation. Moreover, it also highlights
the importance of finding different and alternative switching
mechanisms at the molecular level, to expand the application
range to varied samples and particular imaging case problems.
The most remarkable qualities required for a suitable molecular
switch are brightness and photostability of the emissive state,
extremely high signal contrast between states, reliable photocon-
trol of the transitions involved and the ability to drive and main-
tain virtually all the ensemble of markers in the field of view to
the dark state during the course of measurement.

Photochemistry and Photobiology
3
and with a PTI Quantamaster fluorescence, respectively. The photochromic
oxazine and the amphiphilic copolymer were prepared according to litera-
ture procedures (26,28).
once. In addition, markers that are activated, but failed to be local-
ized during the data analysis step, do not contribute to the superres-
olution image. Moreover, long time sequences or dynamic imaging
are impossible. Nonetheless, uncaging reactions are independent of
environmental conditions, do not require the presence of additives
(i.e. oxygen scavengers or thiols) and usually present a high con-
trast between the dark and the emissive state. A few examples of
probes used in SML microscopies rely instead on photochromic
transformations as the switching mechanism (22,23). They are in
fact the most promising markers because of their reversible charac-
ter, the possibility to lower the irradiation intensities needed for the
on- and off-switching processes and the opportunity to control and
tune the properties of their interconvertible states independently.
In this study, we present two different strategies to switch flu-
orescence with novel molecular assemblies based on a [1,3]ox-
azine heterocycle (24,25). This particular building block provides
the bistability required at the molecular level for SMLM. In par-
ticular, we have successfully applied two different triggering
events, based on the photochromic (26) and halochromic (27)
properties of this oxazine, to achieve images with a spatial reso-
lution beyond the diffraction limit. In the first one, photoisomer-
ization of the oxazine is induced by direct irradiation of the
chromophore. In the second one, the changes are indirectly trig-
gered by the photoinduced uncaging of a proton from an auxil-
iary group. We demonstrate the potential of our approach by
imaging polymer micelles, incorporating these photoswitchable
compounds, in vitro as well as in cells.
Cell imaging experiments. Baby hamster kidney (BHK) cells were
cultured in 10 cm plastic dishes in complete medium: Dulbecco’s Modi-
fied Eagle Medium supplemented with 10% v/v fetal bovine serum and
penicillin/streptomycin (200 U/mL and 200 ug/mL, respectively), at 37°C
and 5% CO2. For the micelles incorporation experiments, the cells were
splitted and plated onto glass coverslips. After removing the culture med-
ium, they were briefly starved with PBS for 1 h, and then incubated
overnight in complete medium containing a 10% v/v of the micelle dis-
persion prepared as describe elsewhere (28). The samples were finally
washed with PBS to remove unincorporated micelles, fixed with MeOH
(10 min at ꢂ20°C), and mounted in Mowiolâ for subsequent imaging.
Experimental setup. Imaging was performed with a home-built epi-
illumination microscope (Scheme 2). The setup counts with three lasers:
532 nm (SDL-532-LM-200-T; Shanghai Dream Lasers Technology Co.),
355 nm (UVL355-10; LaserLabComponents Inc.) and 405 nm (SDL-
405-LM-20-T; Shanghai Dream Lasers Technology). For photoisomeriza-
tion experiments, we used 532 and 355 nm; for photoacid uncaging
experiments, we used 532 and 405 nm. Lasers were combined with a
dichroic mirror DM1 (ZT405rdc; Chroma Technology Corp.). Each beam
was shaped and expanded to a desired beam waist with a telescope and
spatial filter. Wide-field illumination of the three beams was achieved by
focusing them at the back focal plane of the objective with the WF lens
(f = 500) to ensure nearly constant light intensity in the observation area.
Fluorescence was collected with an infinity-corrected oil immersion
objective lens with a 639 magnification (Plan-Neofluar 63x/1.25; Zeiss)
or 1009 magnification (HCX PL APO 1009/1.4; Leica), separated from
excitation light with a dichroic mirror DM2 (ZT532rdc; Chroma Technol-
ogy Corp.) and filtered with a bandpass emission filter (FF01-640/40;
Semrock Inc.). Excitation residual light was further cleaned with a notch
filter (NF01-532U-25; Semrock Inc.). The image is generated in an EM-
CCD camera (IXON-DU-897; Andor Technology) with
a tubelens
(f = 200 mm) and a telescope to adjust a pixel size of ~100 nm. Image
acquisition was performed with the software of the camera (Andor Solis;
Andor Technology). The excitation source was continuously running
during measurements, and the switching laser was digitally modulated
to irradiate the sample for short periods of time to regulate the desired
MATERIALS AND METHODS
General methods. Steady-state absorption and fluorescence spectra were
recorded with a Shimadzu UV–Visible spectrophotometer model UV-1603
Scheme 2. Schematics of the microscope used for SMLM. The inset shows the pulse sequence used to modulate the switching laser (355 nm or
405 nm), to control the irradiation dose.

4
Marco Petriella et al.
irradiation dose (run at constant intensity and switched on between
frames for 50 ls–5 ms). The latter was achieved with a home-made pulse
generation unit connected to the fire signal of the CCD and the blanking
signal of the lasers.
Image processing software. A typical imaging experiment consisted of
ca 20 000ꢂ50 000 frames, and thus localization of single-molecule
events and the superresolution image reconstruction were performed auto-
matically by a series of home-made routines (written in Matlab Soft-
ware). In summary, each frame is analyzed first by a routine that, after
applying a smoothing Gaussian filter, looks for possible events as a clus-
ter of pixels (typically >5 pixels) over a set threshold and determines the
center of mass. The precise position of each cluster is determined with a
procedure based on a Gaussian mask-fitting algorithm described in detail
elsewhere (12,29). The output is a list of localized events, containing the
position, the intensity in number of photons (based on a proper calibra-
tion of the response of the camera) (30). and the frame where it was
observed. Events appearing in consecutive frames at the same position
(within an error of ca 100 nm) are recalculated as the same molecule.
This treatment also allows an estimation of the average on time of the
markers and, accordingly, the selection of an appropriate frame rate. In
fact, the exposure time of a frame is set equal to the average on time
(Figure S1, see Supplementary Materials). The final image is generated
by plotting a 2D histogram of the list of positions rendered by the locali-
zation routine. The imaged area is reconstructed with a pixel size of
about one half of the theoretical average final resolution (calculated from
the distribution of the number of photons detected per localized event),
and the intensity of each pixel is assigned as the number of SM localized
within its area.
To test the accuracy of the image analysis programs, and the consis-
tency with the expected values predicted by Eq. (2), we run a series of
simulations. In the first one, 200 frames with a simulated image of a mol-
ecule with NPH = 280 photons/frame and a background noise b = 1.7
were localized with our program (Figure S2, see Supplementary Materi-
als); a precision of DrDL 9 0.26 was obtained from the average value of
the distribution of the positions obtained (Eq. (2) predicts DrDL 9 0.22).
The pixelation a was set to 90 nm, a value close to the experimental
pixel size of the camera projected in the image plane. The simulation
was repeated for NPH = 40–10 000 and b = 0–10 (Figure S3, see Supple-
mentary Materials). The trend for different b values was as expected,
with a systematical excess error of 20–25%. This can be reduced with
slower localization algorithms; (12) however, we consider the excess
error acceptable and preferred avoiding extremely long image analysis
times. The second simulation was used to test the ability of the algorithm
to accurately localize two close molecules in a single frame. In this case,
200 frames with two molecules separated by a variable distance d were
analyzed with our localization routine (Fig. 1 and video S1, see Supple-
mentary Materials). The average error in the position decreases when d
Figure 1. (a) Average error in the localization of two adjacent objects as
function of the distance between centers. Simulation parameters:
NPH = 280, b = 0.3, a = (DrDL/0.27). (b–d) The particular cases of
d = 2.1 9 DrDL and d = 1.2 9 DrDL. (b and d) one of the simulated
frames; (c and e) map of localized positions (equivalent to a superresolu-
tion image). The total area in b and d is 6 9 6 pixels, and correspond to
the white square in b and d.
a
increases, as expected, reaching a plateau for values of d > 2 9 DrDL
.
The particular cases d = 2.1 9 DrDL and d = 1.2 9 DrDL are also pre-
sented in the figure; while virtually all events were correctly localized for
the larger distance, a spread of wrongly assigned positions along the axis
connecting the molecules is obtained in the other case. From this simula-
tion, we determined that the average of events per frame, in an area of
(10 9 10) lm2, should be no larger than 10 to minimize the chances of
having two markers on the bright state at the same time at a close dis-
tance. The dose of the laser responsible for switching on the markers was
adjusted accordingly.
matrices (25,31). In addition to an excellent fast response,
these molecular switches show no appreciable fatigue; over hun-
dreds of cycles can be performed in polymer films without
detectable signs of irreversible photobleaching. Unfortunately,
the two chromophores in the closed isomer (an indoline and
a p-nitroanisole), or in the open one (a 3H-indolium and a
p-nitrophenolate) are all nonemissive. However, photoisomeriza-
tion brings the substituents R at the quaternary carbon common
to both fused heterocycles in conjugation with the 3H-indolium
fragment of the open isomer. This fact can be exploited to
induce a strong bathochromic shift in the absorption and emis-
sion of a fluorophore, strategically linked at this position (32).
An example of a compound based on this working principle is
the adduct formed with R = 3-vinyl-7-diethylaminocoumarin
(ox-coum) (26).
RESULTS AND DISCUSSION
[1,3]Oxazines as molecular switches
Photochromic oxazines are a family of compounds with a
2H,3H-indole and a benzo[1,3]oxazine within a single covalent
skeleton, fused by the N–C2 bond of the oxazine. Irradiation
with ultraviolet results in the opening of the oxazine ring
(Fig. 2); the formation of the zwitterionic metastable isomer
occurs on a subnanosecond timescale. This photogenerated open
isomer reverts spontaneously to the stable closed one with a
lifetime ranging from tens of nanoseconds to few microseconds
in acetonitrile and approaching the millisecond domain in polymer
The absorption spectra (Fig. 2) of the closed isomer (D) show
three peaks at 410, 320 and 280 nm, respectively, in agreement
with the three electronically isolated chromophores. The first
band (410 nm) is associated with the coumarin fluorophore,
whereas the spectrum at k < 340 nm is approximately the sum
of the absorptions of the indoline and p-nitroanisole fragments.

Photochemistry and Photobiology
5
Figure 3. Uncaging reaction of the photoacid compound (P-H+) used for
intermolecular proton transfer.
for both open isomers, predicting achievable localization accura-
cies in the order of 20 nm Eq. (2). These values were obtained
under saturating excitation intensities with 532 nm light and a
620–660 nm collection window.
Fluorescence switching strategies
Application of the ox-coum construct as a dual emission switch,
either by direct photoisomerization or by indirect photoinduced
proton transfer, requires readout with blue and green light, and
two-channel detection. However, the complete absence of
absorption of the closed isomer allows its simplified use as an
on/off switch by readout excitation with ca 530–580 nm, and
detection only in a red channel (k > 660 nm). An advantage of
this readout strategy is that a virtually infinite contrast between
states is achieved. In addition, markers that are in the dark state
do not contribute to the background signal, and unnecessary
bleaching of them is avoided.
Modulation in the reversible strategy is achieved with UV
light (≤355 nm) and the thermal natural recovery of the back
isomerization. In the irreversible proton-induced one, photoacti-
vation is achieved by irradiation at the absorption band of the
cage compound (peaking at 355 nm), whereas bleaching with
the same light source used for excitation accounts for the off-
switching process. A key feature of both strategies is that the
three driving processes can be tuned with a high degree of
independence: (1) the rate of the dark to bright transition is reg-
ulated by the intensity (or the dose) of UV or blue light irradia-
tion; (2) photon extraction rate by the intensity of the green light
used for excitation; (3) the rate of the bright to dark transition is
fixed by the thermal reaction rate in the photochromism method,
and by the pseudo–first-order constant of the bleaching process
at the saturation intensity. Although the latter cannot be regu-
lated, typical average times of residence in the bright state are
on the order of a few tens of milliseconds. This imposes frame
exposure times that are perfectly compatible with SMLM mea-
surements (commonly used times range from 10 to 100 ms)
(5,6,19,20).
There are some differences between the two strategies that
must be evaluated for each particular application. The one based
on photochromism has the advantage of being reversible, but
requires the use of ultraviolet light, which may be phototoxic.
Nevertheless, the relatively high isomerization efficiency of our
compound (2%) demands very low intensities. The one based on
the intermolecular proton transfer, is irreversible and requires an
environment that allows proton diffusion (i.e. it does not work in
a rigid matrix). However, it has the advantage that UV light can
be avoided. The photoacid presents a considerable absorption in
the violet region of the visible spectrum (the edge of the band),
and, combined with a high uncaging efficiency, allows the use of
a 405 nm diode laser at low intensities.
Figure 2. (a) Photochromism and halochromism of the ox-coum molecu-
lar switch; (b) Absorption (full lines) and emission (broken lines) spectra
of the thermodynamically stable closed isomer D (black) and the proton-
ated open isomer B-H+ (blue lines) in MeCN solutions. The open isomer
was generated in the cuvette by addition of TFA.
Excitation with blue light (k ~ 405 nm) results in typical couma-
rin emission centered at 490 nm. In contrast, the absorption of
the open isomer (B) is dominated by the polymethinic chromo-
phore, a mixture of a coumarin and a hemicyanine dye, with a
strong blue–violet coloration associated with the absorption band
centered at 580 nm. Excitation of this isomer with green light
elicits red-light emission, centered at 660 nm, with high effi-
ciency (10%).
[1,3]Oxazines also show halochromic properties; (31) acidifi-
cation results in the formation of a cationic open isomer (B-H+)
with the same fluorophore and thus very similar absorption and
emission properties. While this switching-on derived fluorescence
may be used in SML microscopies, for instance, to map with
high-resolution events that generate a proton (e.g. enzymatic
activity), photocontrol of the process is desired and may lead to
broader applications in a light microscope. To achieve phoin-
duced fluorescence activation based on the halochromism of oxa-
zines, we have used our construct in combination with a
photoacid compound (P-H+), based on the common 4,5-dimeth-
oxy-2-nitrobenzyl cage group (Fig. 3) (33).
To implement both mechanisms in SMLM, a necessary condi-
tion is that both emissive isomers must be detected at the single-
molecule level. Polyalkylmethacrylate films embedded with the
dye or with mixtures of the dye and the photoacid proved that
the amount of emitted photons per 10–30 ms frame is over 100

6
Marco Petriella et al.
Scheme 3. Schematic representation of the process used for the preparation of micelles, used as test object in SMLM, and the embedding of the photo-
active markers in their core.
of light to induce switching was determined for each particular
Superresolution imaging
region of a sample. Typically 30 000–90 000 frames were
To implement the two switching strategies for imaging acquisi-
tion with a single-molecule localization approach, objects of
nanoscale dimensions are necessary. Neither the ox-coum probe
nor the photoacid have reactive groups to couple them with anti-
bodies via immunolabeling protocols, or surface staining polymer
or silica nanoparticles. In addition, polar environments affect the
emission of the coumarin, the photochromic behavior of the
oxazine, and thermochromism can also compete with the photo-
induced switching mechanism of the second. Therefore, we
decided to incorporate the markers into an amphiphilic copoly-
mer that forms micellar structures in aqueous solvents
(Scheme 3) (28). The macromolecule used was copolymerized
from two methacrylate monomers, decyl methacrylate and the
other containing hydrophilic poly(ethylene glycol) side chains. In
buffered (pH = 7.4) aqueous solutions, globular constructs are
formed. These micellar structures have an average diameter of
14 nm, and thus are probably an assembly of three to four
chains. Their hydrophobic core is capable of hosting different
hydrophobic guest, in this case compounds ox-coum or a mixture
of ox-coum and P-H+, and is relatively permeable to facilitate
proton diffusion between the two components involved in the
intermolecular transfer.
A drop of buffer suspension of the micelles was spin coated
onto coverslips, previously cleaned for 20–30 min in a commer-
cial UV ozone cleaner and imaged in our home-made micro-
scope. Imaging conditions were slightly varied depending on the
sample. Micelles doped only with the photochrome–fluorophore
marker were excited with green light with an intensity
5 kW cmꢂ2, activated with UV light of 10 W cmꢂ2 in pulses of
100 ls–5 ms, and the exposure time of each frame was 10 ms.
Micelles doped in addition with the caged compound were
excited with an intensity 7 kW cmꢂ2 and activated with violet
light of 100 W cmꢂ2 in pulses of 20 ls–2.5 ms, and the expo-
sure time of each frame was 30 ms. The increment on the dose
acquired; measurements were stopped when very few events
were observed in a frame.
Representative images obtained with the two different strate-
gies are presented in Fig. 4, showing the resolution enhancement
obtained with both of them. Diffraction-limited images have a
measured resolution of ca 280–300 nm, estimated from line pro-
files of the WF image of single emitters. On the contrary, SMLM
images have resolving powers of ca 30 nm, calculated in the
same manner, a value that is in agreement with Eq. (2). The
10-fold resolution increase allows the identification of features,
even single micelles, that are blurred in the conventional image.
The micellar structures prepared here have been used as effi-
cient supramolecular assemblies for intracellular delivery of
insoluble and potentially toxic compounds. In fact, Trypan Blue
staining showed that cell viability after exposure to the structures
remains unaltered, confirming their lack of cytotoxicity (28). To
test the feasibility of our approach in model biological samples,
baby hamster kidney cells (BHK) were incubated in complete
culture medium containing a suspension of the micelles, loaded
only with ox-coum molecular probe; in parallel, control cells
(Figure S4, see Supplementary Materials) were kept in complete
medium in absence of micelles. After overnight incubation and
three cycles of media exchange to remove micelles free in solu-
tion, cells were fixed with MeOH to reduce the background sig-
nal and to prevent cell movement during image acquisition.
Figure 5 shows one image obtained with the reversible switching
strategy.
The wide-field images show a rather homogeneously distrib-
uted signal inside the cells that is attributable to the ox-coum sys-
tem, whereas control cells lack any significant signal (data not
shown). In contrast, the subdiffraction images reveal the presence
of discrete, punctuate structures; most of these features are inside
the cells, although a few are also present outside, possibly
because of unspecific micelle attachment to the glass surface.

Photochemistry and Photobiology
7
Figure 4. SMLM with switchable oxazines. Images obtained with the two proposed strategies are shown in the right panel. The resolution enhancement
can be appreciated by comparing the diffraction-limited (DL) image with that obtained by localization (SR). All image bars are 500 nm.
based on distinct phototriggering, both requiring low light inten-
sities. Spatial resolutions down to 30 nm were achieved in vitro
and inside cells. The principle is straightforward and can be
extended to other fluorophores to increase the emission efficiency
of the bright isomer as a means to improve the spatial resolution
even further. Indeed, this would be mandatory for this type of
markers to work in similar superresolution microscopies in three
dimensions, or in samples presenting larger background signals.
The ox-coum system can be incorporated into live cells with the
aid of the polymer micelles, and provides an effective way to
investigate drug delivery and intracellular trafficking at the sub-
diffraction level. In addition, dye functionalization with reactive
groups will be implemented in the future for alternative labeling
or targeting strategies, in particular for biological applications.
Acknowledgements—M.L.B. is a member of ″Carrera del Investigador
Cientifico″ from CONICET (National Research Council of Argentina), and
part of this work was performed under financial support from the Max
Planck Society (Partner Group Grant), ANPCyT and UBACyT. He is also
grateful to Professor Stefan W. Hell for constant support. F.M.R. thanks the
National Science Foundation (CAREER Award CHE-0237578, CHE-
0749840 and CHE-1049860) for supporting his research program and for
providing funds to purchase a mass spectrometer (CHE-0946858).
Figure 5. Superresolution imaging in cells. Wide-field (a, c) and super-
resolution (b, d) images of BHK cells treated with a suspension of water-
compatible micelles containing the oxazine. The enlarged areas (c, d)
show the resolution enhancement achieved by SMLM, allowing the
detection of discrete structures where the micelles are trapped inside the
cells. Bars are 2 lm (a, b) and 300 nm (c, d).
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online
version of this article:
The punctuate structures are consistent with the uptake of the
micelles in the cellular milieu via the endocytic pathway (34),
and they provide a means to estimate the dimensions of the intra-
cellular vesicles in which they are confined. The size of the fea-
tures is well below the diffraction limit, within the 65–80 nm
range, proving a significant resolution enhancement with the
oxazine system in cell imaging applications.
Figure S1. Frequency histogram of the distribution on-times
of compound ox-coum embedded in block copolymer micelles.
Figure S2. Localization accuracy of the localization software.
Figure S3. Localization accuracy as a function of the number
of detected photons and background noise.
Figure S4. Images of control BHK cells.
Video S1. Localization of two objects at variable distance.
CONCLUSIONS AND OUTLOOK
The signal modulation of a fluorescent coumarin dye, induced by
REFERENCES
electronic changes of
a [1,3]oxazine, was implemented to
achieve far-field images with a resolution well beyond the dif-
fraction limit. Two different switching strategies were applied,
1. Murphy, D. B. (2001) Fundamentals of Light Microscopy and
Electronic Imaging. John Wiley-Liss Inc., New York, NY.

8
Marco Petriella et al.
2. Hell, S. W. and J. Wichmann (1994) Breaking the diffraction resolu-
tion limit by stimulated emission: Stimulated-emission-depletion
fluorescence microscopy. Opt. Lett. 19, 780–782.
3. Betzig, E., G. Patterson, R. Sougrat, O. Wolf Lindwasser, S. Olen-
ych, J. Bonifacino, M. Davidson, J. Lippincott-Schwartz and H. Hess
(2006) Imaging intracellular fluorescent proteins at nanometer resolu-
tion. Science 313, 1642–1645.
for single-molecule imaging in live cells. J. Am. Chem. Soc. 130,
9204–9205.
20. Juette, M. F., T. J. Gould, M. D. Lessard, M. J. Mlodzianoski, B. S.
Nagpure, B. T. Bennett, S. T. Hess and J. Bewersdorf (2008) Three-
dimensional sub-100 nm resolution fluorescence microscopy of thick
samples. Nat. Methods 5, 527–529.
21. Belov, V. N., C. A. Wurm, V. P. Boyarskiy, S. Jakobs and S. W.
Hell (2010) Rhodamines NN: A novel class of caged fluorescent
dyes. Angew. Chem. Int. Ed. 49, 3520–3523.
4. Rust, M. J., M. Bates and X. Zhuang (2006) Sub-diffraction-limit
imaging by stochastic optical reconstruction microscopy (STORM).
Nat. Methods 3, 793–796.
€
€
22. Folling, J., V. Belov, R. Kunetsky, R. Medda, A. Schonle, A. Egner,
C. Eggeling, M. Bossi and S. W. Hell (2007) Photochromic rhodam-
ines provide nanoscopy with optical sectioning. Angew. Chem. Int.
Ed. 46, 6266–6270.
€
5. Folling, J., M. Bossi, H. Bock, R. Medda, C. A. Wurm, B. Hein,
S. Jakobs, C. Eggeling and S. W. Hell (2008) Fluorescence nanos-
copy by ground-state depletion and single-molecule return. Nat.
Methods 5, 943–945.
23. Hu, D., Z. Tian, W. Wu, W. Wan and A. D. Q. Li (2008) Photo-
switchable nanoparticles enable high-resolution cell imaging: PUL-
SAR microscopy. J. Am. Chem. Soc. 130(46), 15279–15281.
24. Tomasulo, M., S. Sortino and F. M. Raymo (2005) A fast and stable
photochromic switch based on the opening and closing of an oxazine
ring. Org. Lett. 7, 1109–1112.
25. Tomasulo, M., S. Sortino, A. J. P. White and F. M. Raymo (2005)
Fast and stable photochromic oxazines. J. Org. Chem. 70, 8180–
8189.
6. Heilemann, M., S. van de Linde, M. Sch€uttpelz, R. Kasper, B. See-
feldt, A. Mukherjee, P. Tinnefeld and M. Sauer (2008) Subdiffrac-
tion-resolution fluorescence imaging with conventional fluorescent
Probes. Angew. Chem. Int. Ed. 47, 6172–6176.
7. Hell, S. W. (2007) Far-field optical nanoscopy. Science 316, 1153–
1158.
8. Hell, S. W. (2009) Microscopy and its focal switch. Nat. Methods 6,
24–32.
9. Klar, T. A., S. Jakobs, M. Dyba, A. Egner and S. W. Hell (2000)
Fluorescence microscopy with diffraction resolution limit broken by
stimulated emission. Proc. Natl. Acad. Sci. U S A 97, 8206–8210.
10. Hell, S. W. (2003) Toward fluorescence nanoscopy. Nat. Biotechnol.
21, 1347–1355.
11. Huang, B., M. Bates and X. Zhuang (2009) Super-resolution fluores-
cence microscopy. Annu. Rev. Biochem. 78, 993–1016.
12. Thompson, R. E., D. R. Larson and W. W. Webb (2002) Precise
nanometer localization analysis for individual fluorescent probes.
Biophys. J. 82, 2775–2783.
13. Heilemann, M., P. Dedecker, J. Hofkens and M. Sauer (2009)
Photoswitches: Key molecules for subdiffraction-resolution fluores-
cence imaging and molecular quantification. Laser & Photon Rev. 3,
180–202.
26. Deniz, E., M. Tomasulo, J. Cusido, I. Yildiz, M. Petriella, M. L.
Bossi, S. Sortino and F. M. Raymo (2012) Photoactivatable fluoro-
phores for super-resolution imaging based on oxazine auxochromes.
J. Phys. Chem. C 116, 6058–6068.
27. Swaminathan, S., M. Petriella, E. Deniz, J. Cusido, J. D. Baker,
M. L. Bossi and F. M. Raymo (2012) Fluorescence photoactiva-
tion by intermolecular proton transfer. J. Phys. Chem. A 116,
9928–9933.
28. Yildiz, I., S. Impellizzeri, E. Deniz, B. McCaughan, J. F. Callan and
F. M. Raymo (2011) Supramolecular strategies to construct biocom-
patible and photoswitchable fluorescent assemblies. J. Am. Chem.
Soc. 133, 871–879.
29. Egner, E., C. Geisler, C. v. Middendorff, H. Bock, D. Wenzel,
R. Medda, M. Andresen, A. C. Stiel, S. Jakobs, C. Eggeling,
ꢀ
ꢀ
€
14. Fernandez-Suarez, M. and A. Y. Ting (2008) Fluorescent probes for
super-resolution imaging in living cells. Nat. Rev. Mol. Cell Biol. 9,
929–943.
A. Schonle and S. W. Hell (2007) Fluorescence nanoscopy in whole
cells by asynchronous localization of photoswitching emitters. Bio-
phys. J. 93, 3285–3290.
€
€
15. Andresen, M., A. C. Stiel, J. Folling, D. Wenzel, A. Schonle, A.
Egner, C. Eggeling, S. W. Hell and S. Jakobs (2008) Photoswitchable
fluorescent proteins enable monochromatic multilabel imaging and
dual color fluorescence nanoscopy. Nature Biotech. 26, 1035–1040.
16. Raymo, F. M. (2012) Photoactivatable synthetic dyes for fluores-
cence imaging at the nanoscale. J. Phys. Chem. Lett. 3, 2379–2385.
17. Vogelsang, J., C. Steinhauer, C. Forthmann, I. H. Stein, B. Person-
Skegro, T. Cordes and P. Tinnefeld (2010) Make them blink: Probes
for super-resolution microscopy. ChemPhysChem 11, 2475–2490.
18. Vogelsang, J., T. Cordes, C. Forthmann, C. Steinhauer and P. Tinne-
feld (2009) controlling the fluorescence of ordinary oxazine dyes for
single-molecule switching and superresolution microscopy. Proc.
Natl. Acad. Sci. U S A 106, 8107–8112.
30. Gould, T., V. Verkhusha and S. Hess (2009) Imaging biological
structures with fluorescence photoactivation localization microscopy.
Nat. Protoc. 4, 291–308.
31. Deniz, E., M. Tomasulo, S. Sortino and F. M. Raymo (2009) Substi-
tuent effects on the photochromism of bichromophoric oxazines.
J. Phys. Chem. C 113, 8491–8497.
32. Deniz, E., M. Tomasulo, J. Cusido, S. Sortino and F. M. Raymo
(2011) Fast and stable photochromic oxazines for fluorescence
switching. Langmuir 27, 11773–11783.
33. Goeldner, M. and R. Givens, Eds. (2005) Dynamic Studies in Biol-
ogy: Phototriggers, Photoswitches and Caged Biomolecules. Wiley-
VCH, Weinheim, Germany.
34. Zhang, L. W. and N. A. Monteiro-Riviere (2009) Mechanisms of
quantum dot nanoparticle cellular uptake. Toxicol. Sci. 110, 138–
155.
19. Lord, S. J., N. R. Conley, H. D. Lee, R. Samuel, N. L., R. J. Twieg
and W. E. Moerner (2008) A photoactivatable pushꢂpull fluorophore