doi:10.1016/S0022-2836(03)00206-7
J. Mol. Biol. (2003) 327, 925–930
COMMUNICATION
Observing Membrane Protein Diffusion at
Subnanometer Resolution
Daniel J. Müller1,2*, Andreas Engel3, Ulrich Matthey4, Thomas Meier4
Peter Dimroth4 and Kitaru Suda4
1
Max-Planck-Institute
of Molecular Cell Biology
and Genetics, Biotechnology
Center, Pfotenhauerstr. 108
01307 Dresden, Germany
2
BIOTEC, Technical University
01062 Dresden, Germany
3
M. E. Müller Institute
Biozentrum, 4056 Basel
Switzerland
4
Institute for Microbiology
ETH, 8092 Zürich
Switzerland
Single sodium-driven rotors from a bacterial ATP synthase were
embedded into a lipid membrane and observed in buffer solution at subnanometer resolution using atomic force microscopy (AFM). Time-lapse
AFM topographs show the movement of single proteins within the
membrane. Subsequent analysis of their individual trajectories, in consideration of the environment surrounding the moving protein, allow
principal modes of the protein motion to be distinguished. Within one
trajectory, individual proteins can undergo movements assigned to free
as well as to obstacled diffusion. The diffusion constants of these two
modes of motion are considerably different. Without the structural information about the membrane environment restricting the moving proteins,
it would not be possible to reveal insight into these mechanisms. The
high-resolution AFM topographs suggest that, in future studies, such
data revealed under various physiological conditions will provide
novel insights into molecular mechanisms that drive membrane protein
assembly and supply excellent boundary conditions to model protein –
protein arrangements.
q 2003 Elsevier Science Ltd. All rights reserved
*Corresponding author
Keywords: atomic-force microscopy; ATP synthase; diffusion; single
molecule dynamics; supported membrane
Molecular interactions regulate life with molecular machineries functioning in a heterogeneous
manner. Single-molecule techniques provide
insight into the individuality of biological macromolecules, their unique function, their reaction
pathways, their trajectories and into their molecular interactions.1,2 Complementary to conventional
ensemble measurements, these techniques will
provide new biological insights and will assist in
understanding molecular principles of modern
biology.
Conventionally, the location, dynamics and function of single biological molecules are detected by
fluorescence microscopy. Fluorescence microscopy,
used frequently in modern biological research,
requires labeling the molecules of particular
interest with fluorophores. These chemically fixed
labels, however, exhibit the potential to influence
the native behavior of the labeled molecule.3,4
Abbreviations used: AFM, atomic-force microscopy;
MSD, mean-square displacement.
E-mail address of the corresponding author:
[email protected]
Chemical reactions limit the lifetime of the fluorophore and, thereby, the visualization process by
optical microscopy. Most importantly, molecules
other than the labeled ones will not be visualized
by optical methods. Multicolor imaging by the use
of various fluorophores allows simultaneous
imaging of different biological structures. Unfortunately, the number of colors that are detectable
simultaneously is presently limited to a few, and
the selection of structures that may be observed
has to be chosen carefully. As a side-effect, such
pre-selection of labeled structures, before optical
imaging, will necessarily exclude most other components of, for example, a biological cell. To detect
single molecules by light microscopy, the density
of the labeling fluorophores should be reasonably
diluted to prevent overlapping of their point
spread functions and, thereby, enabling their detection and localization as single molecules.4 As a
result, optical single-molecule detection allows
observation of only very few different structures,
which are separated by a few tens of nanometers.
Hence, optical single-molecule detection by fluorophores is an excellent method to localize the
0022-2836/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved
926
Diffusion of Single Membrane Proteins
samples is performed in buffer solution and at
ambient temperatures,11 which allows the imaging
of living cells12 and single proteins13 at work.14,15
Such experiments can be performed over a timerange of several hours without destruction of
individual proteins or disturbing their inherent
assembly. We report here, for the first time, that
AFM can be applied to observe the diffusion of
single membrane proteins at a lateral resolution
of , 1 nm and a vertical resolution of 0.1 nm.
Imaging loosely packed membrane proteins at
subnanometer resolution
Figure 1. Sodium-driven rotors from Ilyobacter
tartaricus ATP synthase embedded in a lipid membrane.
Individual rotors were orientated with one end towards
the membrane surface. While one end of the rotor was
plugged by phospholipid headgroups (left inset) the
other end had a central depression (right inset). The
plugged and unplugged ends exhibited an average
outer diameter of 5.4(^0.3) nm and of 5.7(^0.3) nm protruding by 2.5(^0.3) nm and by 1.5(^ 0.3) nm from the
lipid surface, respectively.16 Apparently, some rotors
formed assemblies while others had no contact to surrounding proteins. Vertical brightness range of topograph, 3 nm. Sodium-driven ATP synthase rotors were
isolated from I. tartaricus31 and reconstituted into a
POPC bilayer as described.16 The sample was adsorbed
onto freshly cleaved mica17 in buffer solution (10 mM
Tris – HCl (pH 7.8), 300 mM KCl). To minimize the force
interacting locally between the AFM tip and the protein,
the electrostatic double-layer force was adjusted by
reducing the salt concentration of the buffer solution to
100 mM KCl (10 mM Tris –HCl, pH 7.8).18 The AFM
instrument used was a Nanoscope III (Digital Instruments, Santa Barbara, CA) equipped with a 120 mm
scanner and oxide-sharpened Si3N4 tips on a cantilever
with a nominal spring constant of 0.1 N/m (Olympus,
Tokyo, Japan). Topographs recorded in trace and retrace
scanning directions using contact mode showed no
significant difference. All measurements were carried
out at room temperature.
position of one or a few molecules with a high precision of . 10 nm, but it does not allow one to
image biological structures at such a resolution.
Thus, experiments detecting the movement of
membrane proteins or of lipids by fluorescence
microscopy derive conclusions on the surrounding
membrane structure indirectly from the diffusion
behavior of the proteins.5 – 8
In contrast to optical microscopy, the exceptionally high signal-to-noise ratio of the atomic
force microscope (AFM)9 allows the observation of
single unlabeled macromolecules in a heterogeneous environment.10 AFM imaging of biological
We have reconstituted ion driven rotors of
llyobacter tartaricus ATP synthase into a lipid
(palmitoyl-oleoyl phosphatidylcholine; POPC)
bilayer16 and adsorbed the membranes onto freshly
cleaved mica.17 After this, the sample was imaged
at room temperature in buffer solution by contact
mode AFM. To minimize the interaction between
the AFM stylus and the sample, the force applied
to the stylus was counterbalanced electrostatically
to < 25 pN.18 The AFM topograph revealed single
rotors protruding from the lipid bilayer (Figure 1).
While some of the rotors were assembled into
densely packed regions, other rotors apparently
had no directly adjacent proteins. Imaged at higher
resolution, individual subunits of the ring-shaped
rotors were observed indicating an 11-fold symmetry (Figure 2(A); and see the inset in Figure 1).
Each of the subunits is formed by two transmembrane a-helices connected by a polypeptide loop.19
Observing the motion of single
membrane proteins
Imaging the same membrane area after 90
seconds showed most rotors remained at their
location (Figure 2(B)). A few rotors, however,
changed their position. At the same time as some
rotors disassembled, others assembled. Repeated
imaging of the same area (Figure 2(C)) showed the
continuous movement of individual proteins.
Tracking individual rotors indicated that the
molecules started and stopped their movement at
any given time, independent of whether they had
moved previously. The membrane protein movement observed was independent of the scanning
direction of the AFM stylus (horizontal to topograph) and was not correlated to the location and
movement of other proteins. Therefore, we conclude that the imaging process did not influence
their motion significantly. However, the diffusion
coefficient of < 1.04 £ 1025 mm2/second estimated
from topographs recorded in time-lapse movies
was much lower than expected from peptides
embedded in POPC vesicles (3 £ 1022 –8 £ 1022 mm2/
second).20
The following two reasons may explain the
observed reduced diffusion constant. Firstly,
membrane proteins moving in supported lipid
membranes exhibit diffusion coefficients orders of
927
Diffusion of Single Membrane Proteins
Analyzing the modes of motion
Figure 2. Diffusion of single rotors observed at a
spatial resolution ,1 nm. (A) AFM topographs showing
rotors assembled in the membrane. Subunits of individual rotors were observed directly in the raw data.
(B) Same area imaged after 90 seconds. Individual
rotors changed their location (outlined red) while
others remained at their location (outlined yellow).
(C) Repeated imaging of the same area after an
additional 90 seconds. While some proteins moving in
the previous images interrupted their movement (outlined blue) others continued (outlined red) or began
to move (outlined green). The circles dotted in yellow
outline proteins, which did not change their location.
Vertical brightness range of topograph, 3 nm.
magnitude lower than when measured in freestanding membranes.21,22 When adsorbed to
atomically flat mica, the membranes can be
assumed to be separated by a water-layer of
< 1 nm, which couples frictional forces between
membrane protein and support, but still allows
proteins to move within the membrane.21 The
result of this effect is observed directly by the
reduced diffusion coefficient determined from our
measurements. Thus, from experiments performed
on supported membranes one cannot necessarily
expect diffusion coefficient values similar to that
measured on native cell membranes. Secondly,
because of size constraints, single peptides of the
size of a single transmembrane helix tend to show
a higher diffusion constant compared to the oligomeric ATP synthase rotor containing 22 helices
such as investigated in this work.
Individual rotors can undergo lateral and
rotational movements, as apparent from shifts of
their relative position (Figure 2). The uncertainty
when determining the position of a single rotor
was < 1 nm, on average, which is about 30 times
smaller than the accuracy of optical single
molecule methods to detect molecular diffusion3,4
and about 200– 300 times smaller than the diffraction-limited width of an excellent optic. This
positional accuracy combined with the high spatial
resolution of the AFM topographs allowed us to
visualize the trajectory of individual rotors and to
study their collisions with surrounding molecules.
Analyzing all trajectories, without applying any
further classification procedures, results in a nonlinear diffusion behavior (Figure 3; inset, black
data). From the AFM topographs, however, we
can principally distinguish between the structural
factors that limit the free diffusion of a membrane
protein. It becomes clear that apparent transitions
among modes of diffusion can be observed
directly. Most obviously, assemblies formed by the
tight rotor packing restrict the free diffusion of
moving rotors. Additionally, it was observed that
some of the rotors assembled with other rotors,
bringing their free diffusion through the membrane to an end.
To characterize the diffusion behavior of the free
motion and of the so-called obstacled motion we
analyzed 30 trajectories of rotors, which had no
apparent contact with the surrounding proteins
during their movement (Figure 3; green circles),
and 30 trajectories of rotors, the movements of
which had been restricted by other rotors or by
their assemblies (Figure 3; blue circles). To ensure
that rotors moving large distances had no intermittent contact with surrounding proteins, their
motions were analyzed only from diluted areas.
As expected, the observed mean-square displacements (MSD) and the time lag ðtlag Þ yielded a linear
relationship for the free diffusion (Figure 3; inset,
green data23) with a diffusion constant of 2.04 £
1025 mm2/second. From this almost linear behavior,
it can be assumed that the movement of these
rotors was not affected significantly by non-diffusive motions or by collisions.23 Rotors showing a
restricted motion, however, yielded a non-linear
relationship (Figure 3; inset, red data). Interestingly, the MSD for a given time lag fluctuated,
thereby showing various gradients. However, the
gradient was always smaller compared to that of
the free diffusion and showed an average diffusion
constant of 0.32 £ 1025 mm2/second. As observed
directly from the topographs, it follows that the
obstacled motion is represented by a mixture of
free motion and of the interactions with other
membrane proteins. It may be further assumed
that the density and the distribution of the proteins
influence this effect.
To investigate to what extent the imaging parameters of the AFM showed an influence on the
928
Diffusion of Single Membrane Proteins
individual rotors start and end their movement
independently from each other, we assume the
effect of the scanning process on the rotor movement to be neglectible in this work.
Conclusions and biological significance
Direct observation of membrane protein motion
and membrane structure
Figure 3. Trajectories of the two-dimensional diffusion
of single cation-driven rotors in a lipid membrane.
Analyzing the AFM topographs, it was possible to distinguish between free motion of rotors that had no
apparent contact with surrounding membrane proteins
(green circles) and obstacled motion of rotors that had
limited free space due to interactions with other proteins
(blue circles). Inset, MSD versus time lag calculated for 30
trajectories. The black dots represent all motions, which
were not further classified. The green and blue dots
present the intermediate steps of the free and obstacled
rotor motion, respectively.
rotor movement, we varied the scanning parameters such as scanning speed, feedback and
imaging force. It appeared that the rotor movement
could not be directed by the scanning stylus.
At sufficiently high imaging force, @ 100 pN,
membrane and individual proteins became disrupted by the scanning AFM stylus. Similarly, the
feedback parameters controlling the tip-sample
interaction were crucial to prevent destruction of
the proteins and showed the same effect as
elevated forces. At excessive scanning speeds, the
feedback control is not sufficient and the force
interacting between the AFM stylus and the membrane lacked control. Again, the membrane and
protein surfaces were disrupted by the scanning
stylus. Only if AFM topographs recorded successively from the same membrane surface showed
the individual protein structures unchanged, were
the imaging parameters (force, scanning speed,
feedback control) adjusted optimally to prevent
observable structural or functional influence of the
sample. Nevertheless, no connection between the
diffusion direction of the rotors and the scanning
direction of the AFM stylus was observed. It
cannot be ruled out, however, that the interaction
between the rotors and AFM tip enhances the
probability of a rotor to diffuse or initiates the
rotor diffusion through the membrane. However,
since there was no dependence of the rotor
movement on the scanning direction, and since
Most importantly for molecular cell biologists, it
has been demonstrated that individual biological
molecules can be observed moving through a
heterogeneous assembly of membrane proteins.
As revealed from AFM topographs, the movement
of single rotors was restricted by surrounding single
motors and by rotors forming assemblies. Occasionally, the moving rotors reversibly joined either one
of the obstacles. On the other hand, individual
rotors moved freely through the lipid membrane
without showing apparent interactions with other
proteins. The observed trajectories suggest that the
membrane proteins undergo apparent transitions
between these modes of motion (Figure 3; inset,
black data). Subsequent analysis of the diffusion
behavior revealed the first type of movements to
be comparable to a mixture of corralled and
obstacled diffusion (Figure 3; inset, blue data). The
second type of movement is described best by free
diffusion (Figure 3; inset, green data).
The diffusion constants measured, however,
were orders of magnitude lower than that
observed for a free membrane, a behavior
explained by the ultrathin water layer (< 1 nm)
sandwiched between the membrane and supporting mica surface, which couples the friction
between protein and the support. The example
given in Figure 2 shows only three consecutive
topographs of the same membrane area. Within
these images only a few rotors changed their
location, most of the other rotors remained at
their location. However, observing this and other
areas over a much longer time-range indicated
that almost all proteins moved. Therefore, we
assume that if we could observe the proteins over
an even more extended time-range, the individuality of more and less mobile rotors would be
smeared out over the time. This hypothesis,
however, has to be proven. One other approach to
investigate the individual protein mobility would
be to observe a membrane exhibiting a greater
fluidity (e.g. at enhanced temperatures or different
lipid compositions). To do so, however, AFM topographs must be captured within a much shorter
time-range. Therefore, we are currently establishing high-resolution imaging using fast-speed AFM
in our laboratory (see discussion below).
Heading to address fundamental
biological questions
Driven by their dynamic clustering, lipids, other
amphiphilic molecules and membrane proteins
929
Diffusion of Single Membrane Proteins
have been found to form rafts that move within the
cell membrane.24 Such rafts are proposed to serve
as platforms for the attachment of proteins controlling protein sorting and trafficking through
secretory and endocytic pathways. The assembly
of membrane proteins into rafts appears to be of
significant importance during signal transduction25
and many other biological processes.26 It remains a
challenge to understand the molecular mechanisms driving the assembly of rafts. In this context,
techniques allowing the observation of the de- and
re-assembly of membrane proteins into functional
assemblies27 can be seen as a step towards studying the formation of functional membrane protein
assemblies. Furthermore, the ability to image individual proteins and of their environment will
allow more complex biological systems to be
studied and will provide novel insights into the
biogenesis of various native membranes, into the
interactions between similar or different membrane
proteins, and into their formation of supramolecular complexes.
Towards higher temporal resolution and better
understanding of molecular assemblies
Imaged at a spatial resolution of , 1 nm, the
molecular arrangement of assembled proteins was
resolved in this work. Because of the relatively
low temporal resolution of the AFM used, only
slow motions of the membrane proteins could be
studied in this work. Many biological processes
take part on a much faster time-scale. Fast scanning AFM allowing imaging of the biological
sample at much greater time resolution28 – 30 will
allow observation of the dynamics of biological
processes in greater detail. After establishing these
fast AFM techniques for high-resolution imaging
of single proteins they may be applied to study faster diffusion and assembly processes on supported
membranes exhibiting an enhanced mobility.
Immobilized labels allowing the measurement of
fast protein movements of highly diffusive systems
may be required and provided by adsorbing nanogold particles or filamentous structures onto the
membranes. With these or similar approaches, a
combination of experimentally observed protein
movements, influenced by various biochemical
parameters (e.g. pH, electrolyte, lipid composition,
temperature, trajectories), and of molecular modeling of the complexity of the protein arrangement
will promote the understanding of protein
assembly and will provide unique insights into
mechanisms that drive intermolecular interactions
in cellular membranes.
Acknowledgements
We thank Kurt Anderson, Joe Howard, Kate
Poole, Kai Simons, and Winchil Vaz for critical
reading of the manuscript. This work was
supported by the Maurice E. Müller and Swiss
National Foundation (SNF), by the European Community (EFRE) and the Sächsische Ministerium für
Wissenschaft und Kunst (SMWK), Germany.
References
1. American Association for the Advancement of
Science (1999). Frontiers in chemistry: single molecules (special issue). Science, 283, 1593– 1804.
2. Nature Publishing Group (2000). Monitoring molecular movements (special edition). Nature Struct. Biol,
7, 711 – 743.
3. Schmidt, T., Schütz, G. J., Baumgartner, W., Gruber,
H. J. & Schindler, H. (1996). Imaging of single
molecule diffusion. Proc. Natl Acad. Sci. USA, 93,
2926 –2929.
4. Weiss, S. (1999). Fluorescence spectroscopy of single
biomolecules. Science, 283, 1676 –1683.
5. Fujiwara, T., Ritchie, K., Murakoshi, H., Jacobson, K.
& Kusumi, A. (2002). Phospholipids undergo hop
diffusion in compartmentalized cell membrane.
J. Cell. Biol. 157, 1071– 1081.
6. Schütz, G. J., Kada, G., Pastushenko, V. P. &
Schindler, H. (2000). Properties of lipid microdomains in a muscle cell membrane visualized by
single molecule microscopy. EMBO J. 19, 892– 901.
7. Pralle, A., Keller, P., Florin, E. L., Simons, K. &
Horber, J. K. (2000). Sphingolipid-cholesterol rafts
diffuse as small entities in the plasma membrane of
mammalian cells. J. Cell. Biol. 148, 997– 1008.
8. Tomishige, M. & Kusumi, A. (1999). Compartmentalization of the erythrocyte membrane by the membrane skeleton: intercompartmental hop diffusion of
band 3. Mol. Biol. Cell, 10, 2475– 2479.
9. Binnig, G., Quate, C. F. & Gerber, C. (1986). Atomic
force microscope. Phys. Rev. Letters, 56, 930– 933.
10. Scheuring, S., Fotiadis, D., Möller, C., Müller, S. A.,
Engel, A. & Müller, D. J. (2001). Single proteins
observed by atomic force microscopy. Single Mol. 2,
59 – 67.
11. Drake, B., Prater, C. B., Weisenhorn, A. L., Gould,
S. A. C., Albrecht, T. R., Quate, C. F. et al. (1989).
Imaging crystals, polymers, and processes in water
with the atomic force microscope. Science, 243,
1586– 1588.
12. Domke, J., Parak, W. J., George, M., Gaub, H. E. &
Radmacher, M. (1999). Mapping the mechanical
pulse of single cardiomyocytes with the atomic
force microscope. Eur. Biophys. J. 28, 179– 186.
13. Müller, D. J. & Anderson, K. (2002). Biomolecular
imaging using atomic force microscopy. Trends Biotechnol. 20, S45– S49.
14. Engel, A. & Müller, D. J. (2000). Observing single biomolecules at work with the atomic force microscope.
Nature Struct. Biol. 7, 715– 718.
15. Müller, D. J., Janovjak, H., Lehto, T., Kuerschner, L. &
Anderson, K. (2002). Observing structure, function
and assembly of single proteins by AFM. Prog. Biophys. Mol. Biol. 79, 1 – 43.
16. Stahlberg, H., Müller, D. J., Suda, K., Fotiadis, D.,
Engel, A., Matthey, U. et al. (2001). Bacterial ATP
synthase has an undecameric rotor. EMBO Rep. 21,
1 – 5.
17. Müller, D. J., Amrein, M. & Engel, A. (1997). Adsorption of biological molecules to a solid support for
930
18.
19.
20.
21.
22.
23.
24.
Diffusion of Single Membrane Proteins
scanning probe microscopy. J. Struct. Biol. 119,
172– 188.
Müller, D. J., Fotiadis, D., Scheuring, S., Müller, S. A.
& Engel, A. (1999). Electrostatically balanced subnanometer imaging of biological specimens by
atomic force microscopy. Biophys. J. 76, 1101– 1111.
Stock, D., Leslie, A. G. & Walker, J. E. (1999). Molecular architecture of the rotary motor in ATP synthase.
Science, 286, 1700– 1705.
Frey, S. & Tamm, L. K. (1990). Membrane insertion
and lateral diffusion of fluorescence-labelled
cytochrome c oxidase subunit IV signal peptide in
charged and uncharged phospholipid bilayers. Biochem. J. 272, 713– 719.
Sackmann, E. (1996). Supported membranes: scientific and practical applications. Science, 271, 43 –48.
Sonnleitner, A., Schutz, G. J. & Schmidt, T. (1999).
Free Brownian motion of individual lipid molecules
in biomembranes. Biophys. J. 77, 2638– 2642.
Saxton, M. J. & Jacobson, K. (1997). Single-particle
tracking: applications to membrane dynamics.
Annu. Rev. Biophys. Biomol. Struct. 26, 373–399.
Simons, K. & Ikonen, E. (2000). How cells handle
cholesterol. Science, 290, 1721– 1726.
25. Simons, K. & Toomre, D. (2000). Lipid rafts and
signal transduction. Nature Rev. Mol. Cell Biol. 1,
31 –39.
26. Brown, D. A. & London, E. (1998). Functions of lipid
rafts in biological membranes. Annu. Rev. Cell. Dev.
Biol. 14, 111 – 136.
27. Simons, K. & Ikonen, E. (1997). Functional rafts in
cell membranes. Nature, 387, 569– 572.
28. Viani, M. B., Schäfer, T. E., Chand, A., Rief, M., Gaub,
H. & Hansma, P. K. (1999). Small cantilevers for force
spectroscopy of single molecules. J. Appl. Phys. 86,
2258– 2262.
29. Viani, M. B., Pietrasanta, L. I., Thompson, J. B.,
Chand, A., Gebeshuber, I. C., Kindt, J. H. et al.
(2000). Probing protein – protein interactions in real
time. Nature Struct. Biol. 7, 644– 647.
30. Ando, T., Kodera, N., Takai, E., Maruyama, D., Saito,
K. & Toda, A. (2001). A high-speed atomic force
microscope for studying biological macromolecules.
Proc. Natl Acad. Sci. USA, 98, 12468 –12472.
31. Neumann, S., Matthey, U., Kaim, G. & Dimroth, P.
(1998). Purification and properties of the F1F0 ATPase
of Ilyobacter tartaricus, a sodium ion pump. J. Bacteriol.
180, 3312– 3316.
Edited by W. Baumeister
(Received 14 October 2002; received in revised form 6 January 2003; accepted 6 January 2003)