ZurichOpenRepositoryand
Archive
UniversityofZurich
UniversityLibrary
Strickhofstrasse39
CH-8057Zurich
www.zora.uzh.ch
Year:2010
Nanometeraxialresolutionbythree-dimensionalsupercriticalangle
uorescencemicroscopy
Winterood,CM;Ruckstuhl,T;Verdes,D;Seeger,S
Abstract: Wereportanoninvasiveuorescencemicroscopymethodanddemonstratenanometerresolu-
tionalongtheopticalaxis.Thetechniqueisbasedontheinuenceofthemicroscopeslideontheangular
intensitydistributionofuorescence.Axialpositionsaredeterminedbymeasuringtheproportionoflight
emittedbelowthecriticalangleoftotalinternalreection,whichbehavesinaclassicalway,andlight
emittedabovethecriticalangle,whichisexponentiallydependentonthedistanceoftheuorophore
fromthemicroscopeslide.
DOI:https://doi.org/10.1103/PhysRevLett.105.108103
PostedattheZurichOpenRepositoryandArchive,UniversityofZurich
ZORAURL:https://doi.org/10.5167/uzh-43612
JournalArticle
AcceptedVersion
Originallypublishedat:
Winterood,CM;Ruckstuhl,T; Verdes,D;Seeger, S(2010).Nanometeraxialresolutionbythree-
dimensionalsupercriticalangleuorescencemicroscopy.PhysicalReviewLetters,105(10):108103.
DOI:https://doi.org/10.1103/PhysRevLett.105.108103
Nanometer Axial Resolution by Three-Dimensional
Supercritical Angle Fluorescence Microscopy
Christian M. Winterflood, Thomas Ruckstuhl, Dorinel Verdes, and Stefan Seeger
*
Physikalisch-Chemisches Institut, Universita
¨
t Zu
¨
rich, Winterthurerstrasse 190, CH-8057 Zu
¨
rich, Switzerland
(Received 26 February 2010; published 31 August 2010)
We report a noninvasive fluorescence microscopy method and demonstrate nanometer resolution along
the optical axis. The technique is based on the influence of the microscope slide on the angular intensity
distribution of fluorescence. Axial positions are determined by measuring the proportion of light emitted
below the critical angle of total internal reflection, which behaves in a classical way, and light emitted
above the critical angle, which is exponentially dependent on the distance of the fluorophore from the
microscope slide.
DOI:
10.1103/PhysRevLett.105.108103 PACS numbers: 87.64.M!, 68.37.Uv
Fluorescence microscopy is a powerful tool in biology,
providing noninvasive imaging with high biomolecular
specificity. The diffraction-limited resolution of standard
fluorescence microscopy, 200 nm in the lateral and 500 nm
in the axial direction, is insufficient to study the organiza-
tion of the cell at the molecular level. To this end, several
super-resolution microscopy techniques have been intro-
duced that circumvent the diffraction-barrier. Stimulated
emission depletion (STED) microscopy was the first to
achieve super-resolution by far-field optics and has at-
tained 20 nm lateral resolution [
1]. 4Pi microscopy has
shown 40–45 nm in all three dimensions in combination
with STED [
2]. More recently, stochastic optical recon-
struction microscopy (STORM) [
3] and photoactivated
localization microscopy (PALM) [
4] have achieved lateral
resolutions of 20 nm based on single-molecule localization
of photoswitchable fluorescent labels. Extensions of the
latter techniques have attained an axial resolution of 50 nm
using lens astigmatism [
5], 75 nm using double-plane
detection [
6] and sub-20 nm axial resolution using photon
self-interference [
7] or a double-helix point spread func-
tion [
8]. In differential evanescence nanometry (DiNa) [9]
the exponential decay of the evanescent field created by
total internal reflection fluorescence (TIRF) illumination at
the surface is used to measure positions with 10 nm accu-
racy along the optical axis (z axis). The intensity obtained
from a fluorescent particle located within the evanescent
field is not only subject to its distance from the surface but
is also to its intrinsic brightness. Therefore, the intensity of
the TIRF image is calibrated with a consecutive fluores-
cence image obtained with wide-field illumination. The
prerequisite of stable fluorescence intensities hampers the
use of TIRF for z localization of fluctuating emitters, such
as single fluorophores, photoswitchable labels, and quan-
tum dots.
The discontinuity of the refractive index (RI) at the
interface between an aqueous sample and glass has a strong
impact on the angular distribution of radiation of fluoro-
phores nearby. Fluorophores located within one emission
wavelength " from a glass interface, emit a significant
proportion of their fluorescence above the critical angle
#
c
into the glass. For fluorophores located directly above
the glass, supercritical angle fluorescence (SAF) accounts
for 34% of their total emission. The electromagnetic cou-
pling of the dipole emitter’s near-field with the glass de-
cays rapidly with its surface distance z and at a distance of
one emission wavelength SAF is already below 2% of the
total emission [
10,11]. Consequently, exclusive collection
of SAF leads to a high surface-confinement of the detection
volume which is useful for biosensing [
12], surface-
selective cell imaging [
13] and diffusion measurements
in membranes [
14].
Here we introduce three-dimensional SAF microscopy
(3D-SAFM) for z localization with nanoscale resolution. It
is based on simultaneous fluorescence collection in sepa-
rate angular regions, below and above the critical angle for
total internal reflection #
c
¼ sin
!1
ðn
2
=n
1
Þ, where n
1
and
n
2
are the RIs of the glass substrate and the sample me-
dium, respectively. The principle of 3D-SAFM is illus-
trated in Fig.
1. SAF provides an extremely sensitive
measure of the z position due to its rapid decay along the
z axis. The fluorescence emission into surface angles below
#
c
, referred to as undercritical angle fluorescence (UAF)
[
15], is not influenced by an emitter’s z position [16] and
can therefore be used to measure its intrinsic brightness
[Fig.
1(a)]. The parallel detection of SAF and UAF makes
it possible to account for fluorescence intensity fluctuations
of small emitters on any time scale. The z position is
obtained from the measurement of the SAF/UAF ratio
and the theory of its decay along the optical axis.
3D-SAFM was accomplished on a sample-scanning
(Ma
¨
rzha
¨
user, ScanIM) fluorescence microscope with a
custom-made objective [
17,18] consisting of an aspheric
lens (N.A. 0.62 or 24
%
angular aperture, LightPath
Technologies, lens 350340) and a parabolic collector as
shown in Fig.
1(b). The aspheric lens focuses a laser beam
(635 nm, Picoquant, LDM635) with near-diffraction-
limited performance onto the surface of a coverslip. The
PRL 105, 108103 (2010)
P H Y S I C A L R E V I E W L E T T E R S
week ending
3 SEPTEMBER 2010
0031-9007=10=105(10)=108103(4) 108103-1 ! 2010 The American Physical Society
focus has a lateral and axial extent of 433 & 8 nm and
2:5 (m (1=e
2
intensity), respectively. SAF is collected by
the parabolic element and UAF by the aspheric lens. Both
signals are detected with single-photon avalanche diodes
(Perkin Elmer, SPCM-AQR-13) with the small diameter
(180 (m) of the photosensitive area acting as a spatial
filter. The separate optical paths for near- and far-field
emission generate two overlapping fluorescence detection
volumes of completely different axial extent. The decays
of the relative detection efficiencies for SAF and UAF
along the z axis are shown in Fig.
1(c). The z decay of
SAF was calculated in a classical electrodynamic frame-
work, where the fluorescing molecule is considered as a
radiating electric dipole. The flux intensity of the radiation
into angles collected above #
c
was calculated according to
Ref. [
11] assuming a randomized dipole orientation. The z
decay of UAF was computed as the product of the spatial
excitation intensity distribution calculated according to
Refs. [
19–22] and the spatial collection efficiency function
calculated according to Ref. [
23]. The decay of the SAF/
UAF ratio is dependent on the RI of the sample medium
above the coverslip and was calculated according to the
experiment.
We validated 3D-SAFM by measuring the z profile of a
silica sphere of 5 (m diameter (Polysciences) coated with
the fluorescent dye DiIC
18
ð5Þ [DiD, Invitrogen, (650=670)]
[Fig.
2]. The silica bead (n ¼ 1:429) was immobilized on a
glass coverslip (n ¼ 1:523 & 0:002) and immersed in an
index-matched solution of glycerol/water (n ¼ 1:429 &
0:002) to avoid any optical influence of the bead. The
low angle limit for SAF collection was set to #
c
¼ 69:8 &
0:3
%
by means of a circular aperture below the parabolic
collector as shown in Fig.
1(d). A scan of the microsphere
delivers two images with very different intensity distribu-
tions. The depth of the UAF detection volume is about
equal to the radius of the sphere and captures the full lateral
extent of the bead. In the SAF image only the contact
region of the particle with the coverslip appears. The z
profile was calculated pixelwise from the SAF and UAF
intensities and the theory of their z dependencies, which
requires only knowledge of easily accessible experimental
parameters such as RIs, collected angles and emission
wavelength. As shown in Fig.
2(e), the z profile (points)
of the bead is undistorted and follows an ideal sphere (fitted
diameter: 5:1 (m) with excellent agreement demonstrat-
ing the efficacy of 3D-SAFM. The z values deviate (resid-
uals) at most by '60 nm, possibly reflecting the roughness
of the particle.
To further explore the axial resolution of 3D-SAFM,
fluorescent beads of 36 nm diameter (Invitrogen,
Fluospheres (650=670)) were used [Fig.
3]. Because of
their volume far smaller than the volume occupied by the
near-field, the mismatch of their RI with the surrounding
FIG. 2 (color online). 3D-SAFM of a fluorescence coated
microsphere of 5 (m diameter. (a) Schematic of the experiment.
(b) Raw UAF image. (c) Raw SAF image. (d) 3D image of the
bead surface-contact region. Pixels with SAF intensities below a
threshold are shown in black. (e) z profile through the center of
the bead fitted to a sphere of 5:1 ( m diameter. The sample was
scanned with a pixel size of 156 nm and 10 ms integration time
per pixel. '1100–2600 counts per pixel were detected in both
channels together for determining the z positions. Scale bars ¼
1 (m.
FIG. 1 (color online). Principles of 3D-SAFM. (a) Radiant flux
into angles # from the optical axis of a fluorophore with
randomized orientation at different distances z from a water-
glass interface (#
c
¼ 61:1
%
, dashed line) in fractions of ". SAF
emission is significantly reduced already for small distances,
while UAF emission is unaltered. The indicated areas represent
the portions captured by the objective. (b) Optical setup.
(c) Simulated z dependence of the relative detection efficiencies
of SAF and UAF for Cy5 (" ¼ 670 nm) for water-glass. The
moderate decline of UAF follows the laser intensity distribution
near the surface. (d) Enlarged view of the optical paths.
PRL 105, 108103 (2010)
P H Y S I C A L R E V I E W L E T T E R S
week ending
3 SEPTEMBER 2010
108103-2
medium can be neglected. The beads were measured either
in water or agarose gel, both having a RI of 1:333 & 0:002.
The low SAF collection angle limit was set to #
c
¼ 61:1 &
0:1
%
. A narrow z distribution was obtained by nonspecific
adsorption of the beads to a water-glass interface. The SAF
and UAF images of beads directly at the coverslip surface
are indistinguishable when represented with a normalized
color map [Figs.
3(a) and 3(b)], but the SAF count rate is
approximately 5 times higher because of the higher col-
lection efficiency above #
c
. The 3D image was calculated
pixelwise, omitting pixels with a SAF intensity below a
threshold [Fig.
3(c)]. The z positions of the indicated beads
were calculated from the mean value of their image pixels
and the z origin was set to the center of the lowest bead
(highest SAF/UAF ratio). Most of the beads were localized
within a section of a few nanometers and the variation in
the measured z positions is consistent with the size distri-
bution measured with AFM [
24]. The indicated errors of 1–
3 nm were obtained from the standard error of the mean
which represents the localization accuracy of a single scan.
It is noteworthy that the accuracy in the z localization is 2
orders of magnitude higher than the diffraction-limited
lateral resolution. The localization error was governed by
shot noise ( '60 000 counts per bead on average) and can
be reduced by increasing the number of detected photon
counts. The localization error increases with the emitter’s
surface distance as fewer SAF photon counts are detected.
To characterize the localization accuracy deeper inside the
sample, a broad z distribution of beads was prepared by
embedding them in 1% (w/v) agarose gel (n ¼ 1:333 &
0:001) [Figs.
3(d)–3(f)]. A film of agarose was prepared by
spin coating and then covered with water. As expected, the
accuracy decreased along z, but still all beads were local-
ized within an error of 15 nm [Fig.
3(f)].
One can obtain absolute positions with respect to the
coverslip surface from the data by precise calibration of
both detection efficiencies. However, 3D structures can be
resolved accurately even without exact knowledge of ab-
FIG. 3 (color online). 3D-SAFM of 36 nm diameter fluores-
cent nanospheres. (a) Raw UAF image, (b) Raw SAF image and
(c) 3D image of beads adsorbed at the water-coverslip interface.
(d) Raw UAF image, (e) Raw SAF image and (f) 3D image of
beads embedded in agarose gel. The sample was scanned with a
pixel size of 156 nm and 10 ms integration time per pixel. Scale
bars ¼ 2 (m.
(b)
(a)
∆z
1
∆z
2
∆z
3
∆z
0
[nm]
0 25 50
75
100
-8
-6
-4
-2
0
1.0
0.5
0.0
0.03
-0.03
0.00
0 300 600 900
evita
leR
slaudise
R
FIG. 4. Axial localization accuracy of 3D-SAFM. (a) The z
dependency of the relative SAF/UAF ratio at a water-glass
interface for " ¼ 670 nm fitted by a monoexponential function
(solid line). (b) The relative error of the distance measured
between two points separated axially by z
1
¼ 10 nm, z
2
¼
50 nm, and z
3
¼ 200 nm as a function of the error z
0
in
establishing the z origin. Even for a very large z
0
of 100 nm the
relative error of the distance measured between two points is
below 8%.
PRL 105, 108103 (2010)
P H Y S I C A L R E V I E W L E T T E R S
week ending
3 SEPTEMBER 2010
108103-3
solute emitter-coverslip distances due to the quasi mono-
exponential z decay of the SAF/UAF ratio [Fig.
4(a)]. The
proportionality of the SAF/UAF ratios of two axially sepa-
rated point sources is in good approximation constant,
irrespective of their distance from the surface [Fig.
4(b)].
In most cases the highest measured SAF/UAF ratio, e.g.,
from a contact point of the fluorescent sample with the
coverslip, can be used directly to establish the z origin.
Figure
5 shows 3D-SAFM of the microtubule network of
an embryonic NIH 3T3 fibroblast cell by indirect immuno-
fluorescence using a Cy5 labeled (650=670) secondary
antibody. In immunostaining the cell membrane is permea-
bilized allowing for antibodies to enter. As a consequence
the interior of the cell is RI matched with the aqueous
buffer. The antibody-stained microtubules have a diameter
of '60 nm. The bulk of the network is located within the
first wavelength from the coverslip and can be z localized.
Some microtubules further afar cannot, yet they are imaged
by UAF. At the crossings of axially overlapping micro-
tubules the calculated z position represents an average
value. The use of photoswitchable dyes can provide a
way to resolve such axially overlapping objects. By rapid
consecutive single-molecule localization of a large number
of fluorophores an image can be reconstructed as done in
STORM and PALM. For this purpose a SAF microscope
objective with high N.A. optics for UAF is in development.
Furthermore, we note that 3D-SAFM is not restricted to the
use of parabolic collectors and could be performed with
conventional microscope objectives of sufficiently large
(N:A: > 1:45). The presented concept of simultaneous de-
tection of SAF and UAF is very generic and can be
combined with established super-resolution microscopy
techniques including STED.
In conclusion, with 3D-SAF microscopy we have intro-
duced a simple method for noninvasive surface-selective
3D imaging with nanometer axial resolution.
This work was supported by the Swiss National Science
Foundation. The authors thank T. Strassen for providing
the cell strains and assistance with cell culturing.
*s.seeger@pci.uzh.ch
[1] G. Donnert, J. Keller, C. A. Wurm, S. O. Rizzoli, V.
Westphal, A. Scho
¨
nle, R. Jahn, S. Jakobs, C. Eggeling,
and S. W. Hell,
Biophys. J. 92, L67 (2007).
[2] R. Schmidt, C. A. Wurm, S. Jakobs, J. Engelhardt, A.
Egner, and S. W. Hell,
Nat. Methods 5, 539 (2008).
[3] M. J. Rust, M. Bates, and X. Zhuang,
Nat. Methods 3, 793
(2006).
[4] E. Betzig, G. H. Patterson, R. Sougrat, O. W. Lindwasser,
S. Olenych, J. S. Bonifacino, M. W. Davidson, J.
Lippincott-Schwartz, and H. F. Hess,
Science 313, 1642
(2006)
.
[5] B. Huang, W. Wang, M. Bates, and X. Zhuang,
Science
319, 810 (2008).
[6] M. F. Juette, T. J. Gould, M. D. Lessard, M. J.
Mlodzianoski, B. S. Nagpure, B. T. Bennett, S. T. Hess,
and J. Bewersdorf,
Nat. Methods 5, 527 (2008).
[7] G. Shtengel et al.,
Proc. Natl. Acad. Sci. U.S.A. 106, 3125
(2009)
.
[8] S. R. P. Pavani, M. A. Thompson, J. S. Biteen, S. J. Lord,
N. Liu, R. J. Twieg, R. Piestun, and W. E. Moerner,
Proc.
Natl. Acad. Sci. U.S.A. 106, 2995 (2009)
.
[9] S. Saffarian and T. Kirchhausen,
Biophys. J. 94, 2333
(2008)
.
[10] W. Lukosz and R. E. Kunz,
J. Opt. Soc. Am. A 67, 1607
(1977)
.
[11] J. Enderlein, T. Ruckstuhl, and S. Seeger,
Appl. Opt. 38,
724 (1999)
.
[12] T. Ruckstuhl, J. Enderlein, S. Jung, and S. Seeger,
Anal.
Chem. 72, 2117 (2000)
.
[13] D. Axelrod, J Biomed. Opt. 6, 6 (2001).
[14] J. Ries, T. Ruckstuhl, D. Verdes, and P. Schwille,
Biophys.
J. 94, 221 (2008)
.
[15] T. P. Burghardt and K. Ajtai,
J Biomed. Opt. 14, 034036
(2009)
.
[16] L. Novotny,
J. Opt. Soc. Am. A 14
, 91 (1997)
.
[17] T. Ruckstuhl and D. Verdes,
Opt. Express 12, 4246 (2004).
[18] D. Verdes, T. Ruckstuhl, and S. Seeger,
J Biomed. Opt. 12,
034012 (2007)
.
[19] B. Richards and E. Wolf,
Proc. R. Soc. A 253, 358 (1959).
[20] P. To
¨
ro
¨
k, P. Varga, Z. Laczik, and G. R. Booker,
J. Opt.
Soc. Am. A 12, 325 (1995)
.
[21] H. Ling and S.-W. Lee,
J. Opt. Soc. Am. A 1, 965 (1984).
[22] L. Novotny, Lecture Notes on Nano-optics (Rochester
University, New York, 2000).
[23] J. Enderlein,
Opt. Lett. 25, 634 (2000).
[24] S. Seiffert and W. Oppermann,
Polymer 49, 4115 (2008).
FIG. 5 (color online). 3D-SAFM of the microtubule network
of a mouse embryonic fibroblast cell. Microtubules further away
than 640 nm from the coverslip surface (corresponding to <2%
of the normalized SAF/UAF ratio) are shown in gray. (Inset) 3D
representation of the pixels in the region outlined by the box. The
sample was scanned with a pixel size of 156 nm and 3 ms
integration time per pixel. Scale bar ¼ 10 (m.
PRL 105, 108103 (2010)
P H Y S I C A L R E V I E W L E T T E R S
week ending
3 SEPTEMBER 2010
108103-4
