CLIN.CHEM.37/9, 1497-1501 (1991)
Atomic Force Microscopy: Seeing Molecules of Lipid and Immunoglobulin
Helen G. Hansma,
Albrecht
Allen B. Edmundson,’
L. Weisenhorn,
The atomic force microscope (AFM) can image individual
molecules by raster-scanning a sharp tip over a surface.
In this paper we present molecular-resolution Images of
immunoglobulin M (1gM) and of ultraviolet light-polymerized films of the lipid dimethyl-bis(pentacosadiynoyloxyethyl) ammonium bromide (“BRoNco”). The polar head
groups of individual lipid molecules can be resolved on
the surface of this and other lipid films. These lipid films
also provide a good substrate for AFM imaging of DNA
and of other molecules such as antibodies. Because the
AFM scans surfaces, it is most often successful at imaging either molecules that can form an array on a surface
or molecules that are quite firmly attached to a surface.
The ability of the AFM to operate under water, buffers,
and other liquids makes it possible to study biological
molecules under conditions in which they are physiologically active. Imaging of the actual molecular process of
flbnn polymerization shows the potential of the AFM for
studying biological processes. In the six years since its
invention, the AFM has excited much interest and has
imaged molecules in a wide range of systems.
AddItional Keyphrasss:
raster scanning
polymer films
The atomic force microscope (AFM), invented in 1985
by Binnig, Quate, and Gerber (1,2), images surfaces at
molecular or even atomic resolution,
opening new frontiers in the study of surfaces
and of molecules on
surfaces. One of the most dramatic sets of images
recorded by the AFM shows the process of fibrin polymerization,
molecule
by molecule (3). Images taken
after thrombin
was injected into a buffered solution of
fibrinogen
in the fluid cell of the AFM show that short
polymers of fibrin can join into longer polymers. This
answered
the question of whether fibrin polymers are
lengthened only by the addition of fibrin monomers to
the ends of the polymer or by other means.
The AFM images surfaces by raster-scanning
a sharp
tip gently over the surface at forces as small as the
forces between atoms in molecules (10-#{176}
to 10h1 N) (4).
Raster scanning is the movement of a tip back and forth
across a surface, which produces a topological map of the
surface features. The scanning tip is attached to the end
of a cantilever, which deflects as the tip scans up and
down over features on the surface (Figure 1). The
deflection of the cantilever
is detected by the deflection
Department of Physics, University of California,
Santa Barbara,
CA 93106.
‘Harrington Cancer Center, Amarillo, TX 79106.
2Technischen
Umversitflt
Mflnchen,
Lehrstuhl
E22, 8046 Garching, F.R.G.
Received April 16, 1991; accepted July 10, 1991.
f#{252}r
Biophysik
Hermann
E. Gaub,2 and Paul K. Hansma
LIGHT
A
‘I,
PHOTODIOD
\
B
L .0.
RPG
‘
1-P%jXG.A5
SAUP..(
LICUO
Fig. 1. Diagramof an atomicforce mIcroscope (AFM)
(A) sample (represented as a square) Is scanned with an xyz plezoelectric
translator under a shatp tip at the end of a V-shaped cantilever. Laser light
(downwa,d snows) Is reflected off the cantilever onto a two-segment photodiode, which measuresthe vertical deflection of the cantilever as It scans the
sample. (B) Samplecan be Imagedunder wateror other liquid by use of a fluid
cell
of a laser beam reflected off the cantilever. The reflected
laser beam shines on a two-segment
photodiode, which
measures the magnitude of the deflection. Especially
when imaging soft biological samples, it is important
for
the scanning tip to maintain a constant low force on the
sample surface.
To do this, a feedback loop from the
photodiode to a piezoelectric
translator moves the sample up and down in response to small changes in cantilever deflection. This keeps the cantilever
deflection
nearly constant,
resulting
in a nearly constant force
between
the scanning tip and the sample surface. To
image surfaces
as gently and nondestructively
as possible, one scans the surface at the smallest force that can
be maintained without having the tip lose contact with
the surface. The surface is also raster-scanned
in the x
andy directions by means of the piezoelectric translator.
The scan area can range in size from 5 x 5 nm to 80 x
80 aim. The surface topography
thus revealed
is displayed continuously
on a monitor.
The AFM can image individual atoms on some surfaces (e.g., 5,6). Gold atoms 0.3 nm apart have been
resolved, and the AFM has imaged the electrochemical
CLINICALCHEMISTRY,Vol.37, No.9, 1991 1497
of copper atoms on a gold surface
(7,8).
Biological applications of the AFM are among the most
exciting applications.
The AFM can image not only in
air but also under water or buffer or many other fluids.
This allows us to observe cells and biological molecules
at a resolution not previously possible under physiological conditions (e.g., 9-14). The AFM is at present even
more useful for many industrial applications, including
measurement
of the residual roughness
of polished
optical surfaces and inspection of compact discs and
diffraction
gratings.
deposition
Imaging LipId Molecules
The AFM can image molecules of lipids in LangmuirBlodgett films at high resolution. Langmuir-Blodgett
ifims
are synthetic
lipid layers,
resembling
those
in a
cell membrane, that can form at an air-water interface.
The hydrophobic lipid tails align in the air, and the
hydrophilic lipid headgroups remain in the water or
other aqueous phase. Individual
lipid molecules have
been resolved with the AFM in films of many lipids,
including phospholipids
such as phosphatidylglycerol
and phosphatidylethanolamine,
and also in a mixture of
a positively charged lipid and a negatively charged lipid
(9,13-15).
Lipid films are an especially good biological
system for imaging with the AFM. The AFM images
surfaces,
and lipid films are themselves
a surfaceunlike most biological molecules,
which sit on a foreign
surface in the AFM.
Figure 2 (top) shows what appears to be a molecularresolution image
of the lipid dimethyl-bis(pentacosadiynoyloxyethyl) ammomum
bromide
(“BRONCo”).
BRONCOis a positively charged lipid with two diacetylenic hydrocarbon
chains that can be polymerized by
ultraviolet light. These polymerized lipid films have the
advantage of being stronger than films of unpolymerizable lipids (e.g., those in cell membranes). The BRONCO
sample was prepared
at room temperature
as follows: a
BRONCO solution in chloroform
was spread on the surface of a Langmuir trough by standard LangmuirBlodgett technique, polymerized by exposure
for 30s to
high-intensity ultraviolet light, and transferred
by vertical dipping onto a substrate. The substrate was a
monolayer of cadmium arachidate,
also deposited by the
Langmuir-Blodgett
technique
by vertical dipping onto
freshly cleaved mica. After preparation, the sample was
kept under water, even during imaging
thus, the images in Figure 2 show the hydrophilic surface of the
BRONCOfilm. AFM imaging was done with a Nanoscope
II AFM (Digital Instruments,
Santa Barbara, CA) with
an F-scanner, which has a maximum scan range of 10 x
10 tm. The cantilever was 100 pm long with an integrated tip. Scan speeds ranged from 8 to 26 Hz for scan
lines in the x direction; image resolution was 400 lines
x 400 points/line.
The 9 x 9 nm BRONCOimage of Figure 2 (top) shows
bands of molecular
lattice alternating
with diagonal
ridges. This pattern was common to other high-resolution AFM images of the BRONCOfilm. A two-dimensional Fourier transform of these images gives lattice
1498
CLINICALCHEMISTRY,Vol.37, No. 9, 1991
Fig.2. AFMimagesofa film of ultraviolet light-polymerized
BRONCO
lipid
Because the sample was Imaged under water, the hydrophilicsurface of the
BRoNcofilm is seen. Light areas In the image correspond to high areas in the
lipid film; dark areas correspond to low areas. (Top) 9 x 9 nm image,filtered,
showing light ridges on the lipid film alternating with regions where the
hydrophilicheadgroupsof indMdual lipid moleculescan be seen. (Middle)250
x 250 nm image,unfiltered, showing parallel ridges on surface of BRONCO film
oriented In the same direction as the ridges in the toppanel. (Bottom) 3.5 x 3.5
tan image of BRONCOfilm, unfiltered, showing varied surface structure
spacings of approximately 0.8 nm at 50#{176}
and 0.9 nm at
155#{176},
with a faint periodicity
of -1 nm at 110#{176}.
This is
consistent with the results of Radmacher et al. (16),
who found a spacing of 0.75 nm between
rows of
BRONCOin the AFM imaging. X-ray and electron diffraction data on BRONCOand other diacetylenic lipids
have shown an orthorhombic lattice with a = 0.75 nm
and b = 0.495 nm (17-19); this gives a smaller unit cell
than that seen for BRONCOwith the AFM (Figure 2,
top). Diffraction data are based only on the hydrocarbon tails, whereas AFM data have been obtained only
for the hydrophiuic headgroups. Also, the diffraction
techniques measure only molecular averages for dry
samples, whereas the AFM can image local structure
in fully hydrated samples. It is possible that the
BRONCOheadgroups in the AFM are splayed out relative to the hydrocarbon tails in the lattice-containing
regions of Figure 2 (top) and are squeezed together in
the ridges. This illustrates the power of the AFM to
obtain potentially useful information complementary
to the information
obtained by x-ray and electron
diffraction and other techniques.
Figure 2 (middle) shows a larger area of the same
BRONCOfilm, 250 X 250 nm. Light areas in the AFM
image correspond to high areas on the sample, and dark
areas correspond to low areas. In this image the ridges
are oriented in the same direction as the ridges in
Figure 2 (top) but are much more prominent. The
BRONCOfilm appears to fold into a series of parallel
ridges. Near the upper left-hand corner is an unusual
“Y,” where one ridge splits, probably attributable to a
defect in the packing of the lipid molecules. The dark
“valleys” between the ridges are about 1-2 nm deep.
These images that can be obtained with the AFM are
potentially very useful for quality control of LangmuirBlodgett films and other molecular films.
Figure 2 (bottom) shows a larger image of the same
BRONCOlipid film, 3.5 X 3.5 pm, and demonstrates even
more strikingly the ability of the AFM to show the
quality of lipid films. One can see reasonably flat surfaces of lipid in this image at about four different levels.
AFM images give semiquantitative height information
about samples. The height of each lipid level in Figure 2
(bottom) is -5 nm. One might expect each level to be
one bilayer, or 5-6 nm, higher than the previous level,
and this is consistent with the AFM data.
Magnified images of the BRONCOfilm at each level
showed that all levels had ridges oriented in the same
direction as the ridges in the upper two panels of Figure
2. We have not seen ridges like these in AFM images of
other lipids. This particular BRONCOsample was taken
from a region of the film near the corner between the
moving barrier and the side of the Langmuir trough,
where it appears that the lipid was quite dense. BRONCO
samples taken from other regions on the surface of the
Langmuir trough did not show as many lipid levels in
the AFM.
Because ultraviolet-polymerized
BRONCOmolecules
fluoresce, BRONCOfilms have also been viewed with the
fluorescence microscope (16). The 300-pm images in the
fluorescence microscope also show patches of lipid with a
series of parallel cracks and some bright spots, but
without the height information of the AFM. One impressive feature of the AFM is that it can, on some
samples, give high-resolution images of areas as small
as5 x 5nmandaslargeas8o
x 80pm.
ImagIng Protein Molecules: gM
Figure 3 shows AFM images of 1gM. The 1gM was
obtained by cryoprecipitation at 4#{176}C
overnight from the
serum of a patient with WaldenstrOm macroglobulinemia. The 1gM was purified by redissolving
it at 37#{176}C
in
isotonic saline (0.15 mol/L NaCI) containing
sodium
azide, 1 mmol/L, and cryoprecipitating it five times. 1gM
samples for the AFM were prepared at 37#{176}C
by placing
a freshly cleaved mica surface face down onto a 5-pL
drop of 1gM in isotonic saline on Parafllm.
The 1gM
concentration was sufficient for a maximum surface
coverage of 100 ng/cm2. After a 15-mm incubation, the
samples were removed from the Paraflim and rinsed
with water at 0#{176}C.
The samples were then imaged in
the AFM under ethanol. The AFM images were not
stable long: typically, images with the quality shown in
Figure 3 would last for three or four scans by the AFM
before the molecular shapes deteriorated.
1gM is a large molecule with five major subunits, each
structurally similar to a Y-shaped IgG molecule. These
five subunits are arranged in a wheel shape, with the
arms of the Y pointing outward and the stem of the Y
pointing inward (as determined by electron microscopy).
Because 1gMhas not been crystallized, its structure has
not been determined by x-ray diffraction (20).
Figure 3 (left) shows an unfiltered AFM image of
what appear to be four 1gM molecules in a 200 X 200
nm field. Figure 3 (right) is a filtered image of the same
four molecules, showing some submolecular detail.
Molecules 1 and 3 in Figure 3 (right) have a depression
in the center, appearing as a darker region, and molecule 3 appears to have about five subunits (see arrows).
Electron micrographs of 1gM show some fields of molecules with empty centers (21) and some fields of
molecules with prominent centers (22). One model for
the 1gM molecule shows it as spider-like, with the arms
of the Y’s corresponding to the spider’s legs and the
stems of the Y’s forming the body (20). In this model,
the 1gM molecules with depressed or empty centers
would be lying on their backs, whereas 1gM molecules
with prominent centers would be standing on their
arms
(legs?).
Figure 2 (top) showed lipid molecules with a spacing
of 0.7 nm. Figure 3 (right) shows protein molecules
where the resolution is only about 25 nm. Why is there
so much less resolution on the protein molecules than
on the lipid molecules? Perhaps the isolated protein
molecules show much less detail because of thermal
motion or because the AFM tip pushes the surface of
the molecule around more than in the crystallized lipid
film.
1gM has been imaged with the transmission
electron
CLINICALCHEMISTRY,Vol.37, No.9, 1991 1499
Fig. 3. AFM images of four molecules of 1gM,imaged underethanol
Image areas are 200 x 200 nm. (Left) UnfilteredImageshowspositionsof four 1gM molecules. (Right)FilteredImageshows surface topography of molecules.
Notedepressions In center of molecules 1 and 3. Molecule 3 appears to have about fivesubunits(denoted by snows)
microscope.
Negatively stained 1gM molecules in the
electron microscope appear smaller (20-30 nm in diameter) and thinner than 1gM molecules in the AFM, and
the five subunits or even the 10 “arms” of 1gM can be
more easily identified with the electron microscope (1820). Negative staining tends to decrease the apparent size
of molecules, whereas the AFM broadens molecules by an
amount approximating
the diameter of the AFM tip.
DiscussIon
In conclusion, what are the strengths and weaknesses
of the AFM for biomedical applications? The minimum
limit for sensitivity or detection is a single molecule or a
subunit of that molecule, which makes the AFM one of
the most sensitive instruments known. Current instrument problems include drift, in which features slowly
move through the field of view because of thermal drift,
and horizontal streaks, as in Figure 3. These problems
are being minimized with improvements in instrumentation and image-processing
techniques. The AFM tip
interacts with the sample in ways that are sometimes
damaging and not yet well characterized.
Another important consideration in AFM imaging is
the substrate on which the sample
sits. Mica is a
favorite substrate for many applications, because clean
planar surfaces can be prepared by simply peeling off
the top mica layer with a piece of tape. However, many
biological molecules do not stick well to mica because of
its extreme flatness and negative surface charge. Glass
is also used as a substrate, but in the AFM it shows hills
that can be confused with biological molecules.
Under good working conditions, the AFM can image
samples quickly
and routinely. The time for a single
scan is about lOs, soa sample can be imaged in detail in
less than an hour. Samples can be changed in about 15
miii, allowing several samples to be run in the course of
a morning or an afternoon. It is not usually necessary to
1500 CLINICALCHEMISTRY,Vol.37, No. 9, 1991
have elaborate sample-preparation techniques. Because
the AFM images the surface of the sample itself, the
sample does not need to be embedded or sectioned. There
is much room, however, for new and innovative research
in finding better ways of firmly attaching samples to flat
substrates.
This is a very new field. The AFM was invented only
ago, and its use to investigate biological
samples is even more recent. We have already come a
long way in imaging proteins, lipids, and nucleic acids
in ways that have never been seen before.
What future directions might one envision for the
AFM? The AFM is already being modified for combined
optical and AFM imaging of samples, allowing one to
scan ever smaller samples with the AFM and to scan
selected areas of a sample, e.g., an individual cell or a
specific region on the cell surface. Electrochemistry is a
promising new area of AFM research; with electrodes in
the fluid cell of the AFM, it is possible to observe the
changes in each step of a complete electrochemical cycle
at atomic resolution (8). An AFM may be able to
sequence DNA many orders of magnitude faster than
conventional techniques (21). A dedicated AFM could be
a versatile, yet expensive, biosensor. It could scan a
surface that had an affinity for a particular type of
molecule until it detected an individual molecule of this
type. The AFM may be able to engrave ultra-fine patterns on surfaces (22). The movie of fibrin polymerization (summarized
in reference 3) has aroused much
excitement; with improved data-collection instrumentation, we look forward to making and seeing more such
movies, especially of processes occurring at liquid-solid
interfaces. For an instrument
in its infancy, the AFM
has performed well and shows much promise.
six years
We gratefully
Eloise Martzen
acknowledge the expert technical assistance of
and Gregory Kelderman and the helpful discus-
Scot A. Gould, Srin Manne, and Craig Prater. This work
was supported by an IBM Manufacturing Fellowship (A.L.W.),
the Deutsche Forschungsgemeinschaft
(H.E.G.), NSF Grants
DIR-9018846 and DMR89-17164 (H.G.H., P.K.H.), and the Office
of Naval Research (P.K.H.).
sions of
References
1. Binnig G, Quate C, Gerber
Rev Lett 1986;56:930-3.
2. Rugar D, Hansma
CH. Atomic force microscope. Phys
P. Atomic force microscopy. Phys Today
1990;43:23-30.
3. Drake B, Prater CB, Weisenhorn AL, et al. Imaging crystals,
polymers, and processes in water with the atomic force microscope.
Science 1989243:1586-9.
4. Weisenhorn AL, Hanama PK, AlbrechtTR, Quate CF. Forces in
atomic force microscopy
in air and water. AppI Phys Lett
1989;54:2651-3.
5. Binnig G, Gerber CH, Stoll E, Albrecht TR, Quate CF. Atomic
resolution
with atomic
force microscope. Europhys Lett
1987;3:1281-6.
6. Albrecht
TR, Quate CF.Atomic resolution imaging of a nonconductor by atomic force microscopy. J Appl Phys 1987;62:2599-602.
7. Manse S, Butt H.,J, Gould SAC, }llincImA
PK. Imaging metal
atoms in air and water using the atomic force microscope. Appi
Phys Lett 1990;56:1758-9.
8. Manne 8, linnAmA
PK, Masale J, Elings YB, Gewirth AA.
Atomic-resolution
electrochemistry
with the atomic force microscope: copper deposition on gold. Science 1991251:183-7.
9. Weisenhorn AL, Drake B, Prater CB, et al. Immobilized proteins in buffer imaged at molecular resolution by atomic force
microscopy. Biophys J 1990;58:1251-8.
10. Butt H-J, Downing KH, Hsiniuns% PK. Imaging the membrane
protein bacteriorhodopsin with the atomic force microscope. Biophys J 1990;58:1473-80.
11. Liii JN, Drake B, Lea AS, Hansma PK, AndradeJD. Direct
observation of immunoglobulin adsorption dynamics using the
atomic force microscope. Langmuir 19906:509-11.
12. Edatrom RD, Meinke MH, Yang X, Yang R, Evans DF. Direct
visualization
of phosphorylase-phosphorylase
kinase complexes
by scsnning tunneling microscopy and atomic force microscopy.
Biophys J 1990;58:1-12.
13. Egger M, Ohneeorge F, Weisenhorn AL, et al. Wet lipidprotein membranes imaged at submolecularresolution by atomic
force microscopy. J Struct Biol 1990;103:89-94.
14. Weisenhora AL, Egger M, Ohneaorge F. Molecular-resolution
images of Langmuir-Blodgett
films and DNA by atomic force
microscopy. Langmuir 1990;7:8-12.
15. Albrecht TR, Dovek MM, Lang CA, et al. Imaging and modiby 8csnning tunneling and atomic force
microscopy. J Appi Phys 1988;64:1178-84.
16. Radmacher M, Goettgens BM, Tillxnann RW, Hansma HG,
Hansma PK, GaubHE. In Wickramasinghe K, McDonald FA, eds.
Scanned probe microscopies. New York: Am Institute of Physics,
fication of polymers
1991 (in press).
17. Lieser G, Tieke B, Wegner 0. Structure, phase transitions and
polymerizability of multilayers of some diacetylene monocarboxylic acids. Thin Solid Films 1980;68:77-90.
18. Gobel HI), Gaub HE, MOhwaldH. Shape and microetructure of
crystalline domains in polydiacetylenes
monolayers. Chem Phys
Left 1987;138:441-6.
19. Gobel H, Kjaer K, Ala-Nielsen J. Reorientation
of aliphatic
tails during the photopolymerization
of a diacetylenic lipid. Thin
Solid Films 1989;179:41-62.
20. Feinstein A, Richardson N, Taussig MJ. Immunoglobulin
flexibility in complement activation. Immunol Today 1986;7:16974.
21. Green NM. Electron microscopy of the immunoglobulins [Review]. Adv Immunol 1969;11:1-30.
22. Feinstein A, Munn EA, Richardson NE. The three-dimensional conformation of f and 7A globulin molecules. Ann NY
Acad Sci 1971;190:104-21.
23. Beale D, Feinstein A. Constant regions of immunoglobulins.
Rev Biophys 1976;9:160-3.
24. Hnnkmjl
HG, Weisenhorn AL, Gould SAC, et al. Progress in
sequencing DNA with an atomic force microscope. J Vac Sd Techn
B 1991;9:1282-4.
25. Weisenhorn AL, MacDougall JE, Gould SAC, et al. Imaging
and manipulating
molecules in a zeolite surface with an atomic
force microscope. Science 1990247:1330-3.
CLINICALCHEMISTRY,Vol.37, No.9, 1991 1501