Full text of DOI:10.1002/1438-5171(200006)1:2<129::AID-SIMO129>3.0.CO;2-G

Single 129
RESEARCH PAPER
Molecules
Single Mol. 1 (2000) 2, 129-133
Photonic Force Microscopy:
A New Tool Providing New Methods to Study
Membranes at the Molecular Level
Arnd Pralle, Ernst-Ludwig Florin, Ernst H.K. Stelzer and J. K. Heinrich Hörber
Correspondence to
Introduction
Cell Biology and Biophysics
European Molecular Biology Laboratory (EMBL)
Meyerhoftstrasse 1
D-69117 Heidelberg, Germany
phone +49 06221-387569
fax +49 06221-387306
The Atomic Force Microscope (AFM) [1] allows to study non-
conducting materials under ambient conditions and/or in
solution. It was considered to be the ideal tool for physical
studies of live biological specimens [2], soon after its
invention living cells were imaged [3] and dynamic
processes occurring at the cell membrane were examined
email: lastname@embl-heidelberg.de
[
4]. However, the rough surface of a cell often prevents the
submitted 02 May 2000
published 21 Jul 2000
tip of mechanical cantilever from following fine
a
topographic details. Furthermore, forces of several tens of
piconewtons are necessary for stable imaging which can
have the detrimental effect of deforming soft cellular
structures, such as the plasma membrane. Only stiff
structures, such as the cytoskeleton, are visible when
imaging, whilst imaging inside cells is impossible due to the
mechanical connection to the imaging tip. Therefore, a
Abstract
The PFM is a three-dimensional scanning probe microscope
based on optical tweezers. It evolved from AFM techniques
by replacing the mechanical cantilever of the AFM with
optical tweezers and the cantilever tip by a trapped bead.
Various detection systems measure the three-dimensional
position of the bead with a spatial and temporal resolution
in the nanometer and microsecond range, respectively. Due
to the small trapping force constants, thermal position
fluctuations of the probe are relatively large in comparison
to the thermal motion of an AFM cantilever. However, these
position fluctuations provide information about the local
environment and specific interactions of the probe with
molecules such as membrane proteins.
Within the few last years, the instrument turned out to
be a powerful tool to study properties of the plasma
membrane of intact cells at the nanometer scale. For
instance, the diffusion of membrane components could be
observed over minutes at high spatial and temporal
resolutions. For the first time, the diffusion coefficient
measured locally in the plasma membrane of an intact cell
agreed well with previous measurements for lipid model
membranes, thus providing new ways to characterize
membrane structures with unknown properties, such as lipid
rafts. Furthermore, the technique can be used to determine
the elasticity of the lipid bilayer and the binding properties
of membrane components to the cytoskeleton.
scanning probe microscope without
a mechanical
connection to the tip and working with extremely small
loading forces is desirable. Such an instrument, the
Photonic Force Microscope (PFM), has been developed at
the European Molecular Biology Laboratory (EMBL) in
Heidelberg during the last few years.
Experimental Details
Instrumentation
In the PFM the mechanical cantilever of the AFM is replaced
by the 3-D trapping potential of a laser focus [5]. The ability
to trap small particles with high stability was first described
1986 by Ashkin [6]. In the PFM a submicrometer-sized
particle, e.g. a latex, glass or metal bead is used as a tip.
The difference in the refractive index of the medium and the
bead, the diameter of the bead, the laser intensity and the
intensity profile in the focal volume, determine the strength
of the trapping potential. Depending on the application, the
potential is most easily adjusted by changing the laser
power. In effect, the spring constant of the potential is two

130 Single
Single Mol. 1 (2000) 2
Molecules
RESEARCH PAPER
to three orders of magnitude softer than that of the softest
commercially available cantilevers.
Fig. 1. Scheme of the PFM: The trapping beam is emitted by
a diode-pumped Nd:YVO laser (A) (wavelength: 1064 nm,
4
T20-B10-106Q, Spectra Physics, Germany) with a maximum
power of 4 W in continuous wave mode. To achieve the high-
intensity gradient necessary for a high trapping efficiency,
the back-aperture of the focussing device is overilluminated.
This is accomplished by adapting the effective diameter of
the laser beam with a beam expander (B). For stability pur-
poses the laser runs at high power and the intensity in the
focus is regulated by a polarizing beam splitter (C). The
beam is deflected onto a scanning mirror (E), mounted on a
piezo scanner that can tilt around two axis (PSH2zNV, power
supply: ENV 400/1, Piezosysteme Jena, Germany). The
scan optics allows a lateral translation of the laser focus in
the object plane of 3.9 µm. Inside the microscope, the
beam is directed by a dichroic mirror (G) through the tube
lens (H) onto the high numerical aperture (NA) oil-immersion
objective lens (Neofluar 100x, NA = 1.3, Zeiss) (I). The
objective lens is mounted on a piezoelectric actuator (PI-
Foc, P721.00, Physik Instrumente, Germany) permitting the
positioning of the laser focus along the optical axis over a
range of 100 µm. The PI-Foc and the scanning arrangement
serve as 3-D manipulators for the trapped object. The stan-
dard differential interference contrast (DIC) equipment
includes a Wollaston prism (J) and a second one within the
condenser (K). The DIC illumination (L) used here is
restricted to a wavelength of about 700 nm. Optionally a
dichroic mirror (M) and a Hg-lamp (N) can be placed
together with a filter set into the light path in order to use
fluorescence microscopy simultaneously. The laser light
scattered by a trapped particle and the unscattered light are
collected by the condenser lens (K) and projected onto the
quadrant photodiode (P) (S5981, Hamamatsu, Japan) by a
second dichroic mirror (O) (DELTA Lys. & Optics, Denmark).
Fig. 2. 3-D position fluctuation analysis applying Boltzmann
statistics. (a) Typical 3-D position fluctuation recordings of
an optically trapped bead (430 nm in diameter) at low laser
power in aqueous solution 2 µm from the coverslip surface.
The sampling rate was 25 kHz. The peak-to-peak
fluctuations have a maximum of 200 nm in the lateral
directions (x, y) and 500 nm along the optical axis (z). (b) 3-
D energy isosurface (»5 k T above the global minimum) of
B
the trapping potential calculated from 3-D position
histograms using the data shown in (a).

H. Hörber et al.
Single 131
Molecules
Photonic Force Microscopy: A New Tool Providing New Methods to
Study Membranes at the Molecular Level
RESEARCH PAPER
The PFM contains two position-sensing systems to
determine the position of the trapped sphere relative to that
of the potential minimum. The first sensor records the
fluorescence intensity emitted by fluorophores inside the
trapped sphere, which are excited by the trapping laser via a
two-photon process. The fluorescence intensity provides an
axially sensitive position signal with millisecond resolution
[
7]. The second sensor is based on the interference of the
forward scattered light from the trapped particle with the un-
scattered laser light at a quadrant photodiode [8]. This
provides a fast three-dimensional recording of the particle's
position, and is most sensitive perpendicular to the optical
axis. The position of the bead can be measured with a
spatial resolution of better than one nanometer at a
temporal resolution in the microsecond range. Depending
on the application, either one or both of these detection
systems can be used.
At room temperature the thermal position fluctuations of
the trapped bead in weak trapping potentials reach several
hundred nanometers. At first glance these fluctuations
seem to be disturbing noise that limits the resolution.
However, due to the speed and resolution of the position
sensor based on the forward-scattered light detection, the
fluctuations of the bead can be tracked directly, opening
new ways to analyzing the interaction of the bead with its
environment.
The PFM (Fig.1) development is based on an inverted
optical microscope with a high numerical aperture objective
lens providing a high resolution optical image of the
investigated structures. The laser light is coupled into the
microscope using techniques known from laser scanning
microscopy and is focussed by the objective lens into the
Fig. 3. PFM imaging using a 200nm, fluorophore labelled
latex bead and the two-photon fluorescence as position
sensor. Optical DIC (a, b) and PFM (c) images of a small
neurite (N) growing at a branching point (B) from a major
neurite (M) of a glutaraldehyde fixed rat hippocampal
neuron. The PFM image was acquired in constant height
mode, with a loading force of smaller than 5pN (recording
time about 30s).
specimen plane. A 1064 nm Nd:YVO laser is used since
4
neither water nor biological material have significant
absorption at this wavelength. A dichroic mirror behind the
condenser deflects the laser light onto the quadrant
photodiode. The difference between the left and right half of
the diode provides the x-, the difference between upper and
lower half the y-, and the sum signal the z-position [8]. For
displacements small relative to the focal dimensions, the
signals change linearly with the position of the bead in the
trap.
The trapped particle acts as a Brownian particle in a
potential well and its position distribution is therefore
described by the Boltzmann-distribution. Such distributions
p(r) are readily measured with the high spatial and temporal
resolution available to the PFM. Thus allowing for the
calibration of the three-dimensional trapping potential E(r)
without any further knowledge but the temperature, using
Imaging
Scanning either the laser focus or the sample, the PFM can
be used in a manner similar to the AFM to image surface
topographies. The applied forces range from a few
piconewtons down to fractions of one piconewton. The size
of the smallest distinguishable structure is given by the
diameter of the sphere, which can be some ten nanometers
for metal beads or several hundreds of nanometers for latex
or glass beads. Figure 3 shows an image of a neuronal
dendrite obtained operating the PFM in constant height
mode with a maximal force of about 1 pN using a latex bead
with a diameter of »200 nm. Beside the constant height
mode, other imaging modes known from AFM can be
implemented using a feedback loop. Furthermore, due to
the large thermal fluctuations of the bead a new imaging
mode becomes possible with the PFM by observing the
spatial distribution of the thermal fluctuations of the bead,
which avoids stationary objects. In this way it is possible, to
image for example the 3-D structure of polymer networks.
Additionally, detailed information about the interaction
potential between the bead and the surface of the objects
[
9]
E(r)= -k T*ln p(r) + E
B
0.
A precision of one-tenth the thermal energy k T is achieved
B
with a sufficient number of statistically independent position
readings. Figure 2 shows a typical isoenergy surface of the
trapping potential at »5 k T.
B

132 Single
Single Mol. 1 (2000) 2
Molecules
RESEARCH PAPER
imaged can be obtained. At present, PFM imaging of
biological material is limited by non-specific adhesion
events between the bead and the sample, which can lead to
the loss of the probe due to the weak forces applied.
Although there seems to be no general strategy to prevent
non-specific adhesion to complex biological material, it is
still possible to achieve specific binding to membrane
structures for a limited time as will be shown below (Fig. 5).
described by Saffman and Delbrück [12] using a
hydrodynamic model treating the bilayer as a continuum and
assuming weak coupling to the surrounding liquid medium.
The viscous drag g on a cylindrical particle with radius r in a
m
homogeneous lipid bilayer of thickness h is: g m
=
4ph h/(ln(h h/h r)-e), where h denotes the viscosity of the
m
m
w
w
surrounding fluid, hm the viscosity of the lipid bilayer, and e
the Euler constant. This approximation is valid for proteins
with radii large compared to the size of the lipid molecules
and for h m >> h w, which is true for cellular membranes.
Thus, the membrane viscosity can be determined by binding
a bead to a membrane structure of known diameter and
measuring the total viscous drag on the bead as described
above. Such measurements can be performed with a
temporal resolution of 0.3 s and within areas of 100 nm in
diameter.
Fig. 4. Local viscosity measurement. Decrease of the
diffusion coefficient of a sphere approaching a surface
measured for a polystyrene sphere with r=515nm using the
PFM. The dotted line shows the theoretical prediction by
Happel and Brenner fitted for r=530nm and a free solution
-
13
2
diffusion constant of D=6.9 10 m /s.
Local Viscosity Measurements
The motion of a thermally fluctuating particle in a harmonic
potential like that of the laser trap is to a certain level
characterized by an exponentially decaying position
autocorrelation function with a characteristic autocorrelation
time t=g/k, where g denotes the viscous drag on the sphere
and k the force constant of the optical trap. Thus, the local
viscous drag and the diffusion coefficient D = k T/ g of a
B
sphere in a harmonic potential can be calculated from the
measured autocorrelation time of the motion and stiffness
of the potential [10]. The stiffness of the trapping potential
is determined as described before for the calibration of the
laser trap. To measure diffusion coefficients with a
statistical error smaller than ten percent, the observation
interval must be in the order of 1000 times longer than the
autocorrelation time. Hence, the motion of the bead limits
the temporal resolution of the viscosity measurement, and
not the bandwidth of the detection system. Near surfaces,
e.g. glass surfaces or cell membranes, the diffusion is
reduced because of the spatial confinement (Fig. 4) [11].
Within an obstacle free lipid bilayer, diffusion has been
Fig. 5. The viscous drag g (A) and the lateral spring
constant k (B) during a typical experiment in which an
antibody coated sphere (e.g. polyclonal anti-PLAP, sphere
radius r=100nm) was brought close to the plasma
membrane and held there until the antibody bound to the
membrane protein (“+” x- and “·” y-direction). The traces
show three regions corresponding to the reference
measurement away from the surface (I), the diffusion near
the plasma membrane (II) and after binding to the
membrane protein (III). The lateral confining potential k (B)
remained unchanged during the entire experiment, whereas
the lateral viscous drag g showed a clear chang e upon the
antibody binding.

H. Hörber et al.
Single 133
Molecules
Photonic Force Microscopy: A New Tool Providing New Methods to
Study Membranes at the Molecular Level
RESEARCH PAPER
Local Viscous Drag of Single Membrane
Components
biological samples to a detailed investigation of single
molecule properties.
References
Figure 5 depicts a typical measurement of the lateral
viscous drag and spring constant of the trapping potential
plotted against time for an experiment performed on the
plasma membrane of a PtK2 cell. The measured viscous
[
1] G. Binnig, C.F. Quate and Ch. Gerber, Phys. Rev. Lett.
6, 930 (1986)
5
[2] B. Drake, C.B. Prater, A.L. Weisenhorn, S.A.C. Gould,
T.R. Albrecht, C.F. Quate, D.S. Cannell, H.G. Hansma
and P.K. Hansma, Science 243, 1586 (1989)
[3] W. Häberle, J.K.H.Hörber and G. Binnig, J. Vac. Sci.
Tech., B9, 1210 (1991)
drag g is the sum of the Stokes drag of the sphere g =
s
6
p h r and the viscous drag g of e.g. a single protein in the
w
m
lipid bilayer. The Stokes drag of the sphere near the cell
membrane increases due to the confinement and must be
corrected for the calculation of g . The observation of the
[4] J.K.H. Hörber, W. Häberle, F. Ohnesorge, G. Binnig,
H.G. Liebich, C.P. Czerny, H. Mahnel and A. Mayr,
Scanning Microscopy 6 (4), 919 (1992)
m
strength of the lateral potential ensures that the observed
membrane component does indeed diffuses freely.
[
5] E.-L Florin, A. Pralle, J.K.H. Hörber, and E.H.K. Stelzer,
J. Stru. Biol. 119, 202 (1997)
Using the PFM technique on a single transmembrane
protein of known size it is possible to determine the
membrane viscosity in living cells in areas of about 100 nm
in diameter [13]. Once the membrane viscosity has been
obtained, the PFM technique allows one to determine the
diameter of other membrane structures and to monitor it
continuously for several minutes. We performed such
experiments to determine the size of membrane structures
called "rafts" [13]. Lipid rafts are involved in the polarized
sorting of proteins [14] and cellular signalling [15].
According to our study in the plasma membrane of
fibroblast-like cells, these structures are indeed quite stable
in the time-scale of minutes [15], thus justifying the model
of “rafts” as lipid-protein complexes floating raft-like in the
membrane [16]. Their diameter was determined to be about
[
[
6] A. Ashkin, Opt. Lett. 11, 288 (1986)
7] E.-L Florin, J.K.H. Hörber, and E.H.K. Stelzer, Appl.
Phys. Lett. 69, 446 (1996)
[
[
8] A. Pralle, M. Prummer, E.-L. Florin, E.H.K. Stelzer and
J.K.H. Hörber, Micr. Res. and Tech. 44, 378 (1999)
9] E.-L. Florin, A. Pralle, E.H.K. Stelzer and J.K.H. Hörber,
Appl. Phys. A 66, S75 (1998)
[
[
[
[
10] A. Pralle, E.-L. Florin, E.H.K. Stelzer and J.K.H. Hörber,
Appl. Phys. A 66, S71 (1998)
11] J. Happel and H. Brenner, Low Reynolds number
hydrodynamics, Prentice Hall, New Jersey (1965)
12] P.G. Saffman and M. Delbrück, Proc. Natl. Acad. Sci.
USA 72, 3111 (1975)
13] A.Pralle, P.Keller E.-L. Florin, K. Simons and J.K.H.
Hörber, J. Cell Biol. 148 (5), 997 (2000)
50 nm.
In summary, the PFM has proven to be a powerful tool for
the study of biophysical properties of plasma membranes
as well as the interaction of single molecules or complexes
of molecules within a membrane. Future applications of the
PFM will cover a wide range from 3-D in vivo imaging of
[14] P. Keller and K. Simons, J. Cell Biol. 140, 1357
(1998)
[15] T.P. Stauffer and T. Meyer, J. Cell Biol. 139, 1447
(1997)
[16] K. Simons and E. Ikonen, Nature 387, 569 (1997)