Atomic Force Microscopy Part 2 (Molecular Biology)

1.6. Examples of AFM Imaging

1.6.1. Cells and Cellular Processes

AFM images cellular and subcellular structures in physiological conditions at a resolution far exceeding that of optical microscopes. Living cells have been imaged in aqueous conditions with a resolution as small as 10 nm. By applying a larger imaging force, intracellular organelles and cytoskeleton networks have also been examined (Fig. 2). The ability to view structures beneath the plasma membrane is puzzling. Two possible mechanisms are (1) the tip penetrates the bilayer and images the substructure or (2) the plasma membrane drapes around the cytoskeletal fibers and the tip images the contour of the plasma membrane. If the "drape" model is correct, these observations give wonderful demonstrations of the flexibility of biological membranes and a potential tool for measuring the "drape characteristics" of natural and synthetic membranes.

Figure 2. Images obtained with AFM. (a) Error-mode AFM image of a fixed atrial cell. The cytoskeletal network and ce nucleus are visible. [S. Shroff, D. Saner, and R. Lai (1995) Am. J. Physiol. 269, C286-292.] (b) AFM error mode image a freeze-fracture replica of atrial tissue. Details can be identified in the mitochondrion and in the atrial granules and vesicles. Scale Bar: 1 mm. [L. Kordylewsky, D. Saner, and R. Lai (1994) J. Microsc. 173, 173-181.] (c) Error-mode AFM image of a neurite outgrowth in PC-12 cells treated with nerve growth factor and dibutyryl cyclic AMP. For details, see R. Lai, B. Drake, D. Blumberg, D. Saner, P. K. Hansma, S. Feinstein (1995) Am. J. Physiol. 269, C275-C285. (d) Lamb. DNA under propanol using a regular silicon nitride tip. Note the sharp bends in the strands and a fairly regular series of lumps along the strands, 6 to 8 nm apart. The strand width is 7 to 9 nm, greater than expected, which may result from the relatively large size of the tip. Scan size is 1000 nm x 1000 nm (courtesy H. Hansma, H. G. Hansma, R. L. Sinsheimer, M Q. Li, and P. K. Hansma (1992) Nucleic Acids Res. 20, 3585-3590). (e) AFM height mode image of a single connexon ( gap junction hemichannel) imaged on its cytoplasmic face. The subunit structure, a central pore, and the spacing between connections are apparent [S. A. John, D. Saner, J. Pitts, M. Finbow, and R. Lai (1997) J. Struct. Biol. 120, 22-31).] (f) High-resolution AFM imaging of a single acetylcholine receptor expressed inXenopus oocytes. The channel has a diameter of ~10.5 nm , and the five subunits (~1 to 1.5 nm in diameter) that have a central pore-like structure are shown.

In AFM imaging, the scan area can be varied from the micron to nanometer range, and hence it is possible to image features ranging from whole cells to individual macromolecules, such as ion channels and receptors (Fig. 2). In air-dried, hydrated Xenopus oocytes in which acetylcholine receptor proteins are expressed, the characteristic pentameric subunit structure of the expressed receptor has been observed after removing of the follicle layer. This technique of imaging expressed proteins on or in the surface of the plasma membrane of oocytes will shortly enable the characterization of a myriad of ion channels and receptors expressed in an appropriate expression system.

The major factor limiting resolution in imaging a cell surface is the mobility of the upper plasma membrane, plus the mobility of the macromolecules within the plasma membrane: the lower membrane is anchored to the substrate. Improvements in resolution may be made by (1) increasing the surface rigidity (e.g., suction of cells onto patch pipettes and thus reducing the lateral mobility of proteins); (2) by imaging with low forces (e.g., attractive force mode imaging or magnetic tapping imaging).

1.6.2. Membranes and Membrane Proteins

Imaging membranes, both native and reconstituted, has received wide attention because of their flattened 2-D sheet-like structure and the ease in preparing them. Purified membrane proteins, such as bacteriorhodopsin, gap junctions, acetylcholine receptor, and the hexagonally packed intermediate (HPI) layer from Drosophila radiodurans, have been imaged in aqueous conditions without fixation (Fig. 2). These membrane proteins have been characterized extensively by alternative techniques, such as electron microscopy and electron and X-ray crystallography. The results from AFM studies agree remarkably with those from other techniques. In addition, AFM images provide a direct observation of membrane polarity. For example, extracellular and cytoplasmic surfaces of gap junctions are distinguished unambiguously. The thickness measurement in AFM study is also direct and often very precise.

Synthetic membranes (Langmuir-Blodgett) and reconstituted vesicles have been imaged at molecular resolution. Images of Langmuir-Blodgett films provide direct measurement of lipid membrane thickness, obtained previously by indirect methods and theoretical extrapolations, because the height resolution in AFM is subnanometer. AFM images of Langmuir-Blodgett films show individual polar head groups and their molecular arrangement, including their long-range packing. An advantage of studying these membranes with AFM is that one can change the lipid composition on-line and study lipid-lipid interactions, lipid fluidity, and lipid-protein interactions. AFM images of proteins that are naturally embedded within membranes and form 2-D crystalline arrays and images of LB films provide some of the best evidence that the imaging of biological specimens generally agrees with that by electron microscopy, with the advantage, of course, that AFM imaging occurs in nearly physiological environments. It is also worth noting that such a correlation adds weight to the interpretation of images gathered by electron microscopy.

Proteins immobilized by synthetic membranes have also been imaged. Bacterial porins are one of the best-studied channel-forming membrane proteins. Porins reconstituted as 2-D crystals in lipid vesicles have been imaged in a liquid environment. AFM images at molecular resolution show the trimeric structure of porins illustrated by X-ray crystallography and electron microscopic single-particle reconstruction. In addition, recent studies show that molecular resolution can be obtained on noncrystalline specimens in liquid media. This opens a new avenue for studying the molecular structure of biological macromolecules (e.g., ion channels, receptors) that are easily expressed in an appropriate expression systems, such as the Xenopus oocyte, or simply isolated and anchored properly on suitable substrates. For example, purified cholera toxin molecules incorporated into synthetic phospholipid bilayers by covalent cross-linking, observed at molecular resolution, have the expected pentameric structure. Other membrane proteins imaged by AFM include the hexagonally packed intermediate (HPI) layer of Drosophila radiodurans,, vacuolar proton

As mentioned before, AFM imaging of whole cell membranes has also been achieved. Acetylcholine receptor expressed in Xenopus oocytes has been observed. Calcium channels have been localized on the calyx-type nerve terminal of fixed chick ciliary ganglion in culture by imaging avidin-coated 30-nm gold particles incubated with w-conotoxin GVIA linked to biotin, although the molecular structure of individual calcium channels was not reported. The interchannel spacing of 40 nm was noticed, which may reflect the spatial limitation due to tagging with 30-nm gold particles. Individual calcium channels have much smaller diameters. Other membrane channels, such as sodium channels, potassium channels, and gap junctions, are 6 to 10 nm in diameter. Isolated cellular organelles have been imaged with AFM, including nuclear pore complexes, ~134 nm in outer diameter, with a central pore-like trough.

1.6.3. Isolated Macromolecules, such as DNA, Amino Acids, and Proteins

Imaging isolated macromolecules is challenging because it is difficult to find suitable surfaces to which to anchor the molecules for repeatable and reliable imaging. The recent development of cryo-AFM shows good promise for obtaining high-resolution images of isolated macromolecules. Images at molecular resolution have been obtained of DNA at the plasmid and chromosomal levels, polyamino acids, isolated proteins, and ligand-receptor complexes. Large protein fibers, such as actin and microtubules, have also been imaged at molecular resolution, and it was possible to discern individual actin molecules. Isolated protein molecules show dynamic changes while imaging with the AFM. For example, when glycogen phosphorylase b binds to phosphorylase kinase, the dimensions and shapes of the proteins change noticeably.

Imaging DNA and nucleic acids has been appealing for many reasons. Given their well-known geometry and easy availability, they are readily identified and hence used for calibration and for studies of the interaction between tip and sample. Intriguingly, this may also open a door for structure-based sequencing and mapping of DNA AFM. However, DNA sequencing AFM will require an order-of-magnitude improvement in resolution (to ~2 to 3 A). This increase may come from improvements in hardware and software, but methods to prepare DNA in extended conformations will probably be just as important in improving resolution.

Images of double-stranded DNA at molecular resolution (2 to 3 nm), in which the helical pitch and turns could be deciphered, have been obtained in air and liquid. Occasional images at higher resolution showing individual base pairs have also been obtained. Single-stranded DNA, though, has proven more intractable to image at any molecular resolution.

Images of complexes of DNA and protein A deposited onto mica show single proteins bound to the end of the DNA strands. In addition some single protein molecules bind to up to four DNA strands per protein molecule. When RNA polymerase binds to DNA, AFM images show that the modified DNA is bent at marked angles where the polymerase binds. One appeal of these approaches is in searching for DNA-binding proteins and, intriguingly, perhaps to image the effects of topoisomerase on DNA.

1.6.4. Imaging Dynamic Processes

AFM, unlike other molecular level imaging systems, allows imaging in an aqueous environment. In an elegant set of studies, AFM was used to visualize real-time surface processes on vaccinia virus pox viridae-infected monkey kidney cells. Real-time changes in surface morphology and the exocytosis of enveloped virus and proteins were observed over a period of 19 hours. In contrast, cells not infected with virus showed no appreciable change in surface morphology. The real-time contractile activity of cultured atrial myocytes was also imaged. As the concentration of calcium increased, cells underwent rapid contraction, and a corresponding shortening in cytoplasmic fibers (perhaps cross-bridges) was observed.

Dynamic studies have been conducted on isolated proteins, such as formation of glycogen phosphorylase-phosphorylase kinase complexes, antibody-antigen interactions, dynamics of immunoglobulin adsorption, and binding of streptavidin to a biotinylated lipid bilayer. Real-time polymerization of fibrin, a protein important in blood clotting, shows that the polymer chain grows by the fusion of many short chains, rather than by successive addition of monomers to a few long chains. A change in Langmuir-Blodgett film morphology has been observed as trace amounts of a fluorescent dye are added, suggesting that the perturbation of molecular conformation by the tracer molecules may not be as insignificant as is commonly believed.

AFM can be used for in situ studies of the growth of protein 3-D crystals in their native solution environment and of the role of nucleation centers, lattice defects, and saturation level. These studies may provide clues for growing the 3-D crystals essential for high-resolution X-ray crystallography.

1.7. Structure-Function Studies

AFM can be combined with other techniques, which opens the possibility, as various biochemical, pharmacological, and other perturbations are introduced on-line, of real-time dynamic studies for direct structure-function correlations at the molecular level. For example, "single cell" experiments have been reported where electrical activity and AFM images were obtained from Xenopus oocytes expressing acetylcholine receptor. Electrical recording of acetylcholine-sensitive current and labeling by specific binding of a-bungarotoxin were also conducted in parallel. The receptor density calculated from the AFM studies correlates well with that from electrical measurements and toxin binding, but the clustering of acetylcholine receptor differs from the uniform distribution of the expressed receptors that is commonly assumed in electrophysiological studies. A correlation between patch-clamp electrical recording and AFM of transcription factor IID (TFIID) interactions with the nuclear pore complex has also been reported, showing unplugging of the nuclear pore complex accompanied by prolonged electrical current through the channel, perhaps reflecting the reopening of the channels. A combined AFM and patch-clamp study measured the electrical current in the membrane patches excised from Xenopus oocytes and attached to the pipette tip, while imaging the surface topology with AFM. Although the resolution of such a study is limited, it nevertheless shows the promise of direct structure-function studies of membrane macromolecules. Another study simultaneously measured images of bacteriorhodopsin in purple membranes adsorbed onto a lipid monolayer, ion transport through the membrane, and the electrical properties of the membrane. A recent study simultaneously measured the surface structure of nuclear pore filters and electrical current passing across the filter through pores of different diameters. For such a study, a scanning ion-conductance microscope was developed that records electrical activity, while imaging the 3D-structures of various membranes (Fig. 1).

Imaging force can be varied considerably during experiments. Such a feature has been used to nanomanipulate protein and membrane structures, and two different conformations of membrane proteins have been reported.

As already mentioned, action is at biological surfaces. The potential for rapidly characterizing and cataloguing the structures of synthetic peptides opens vistas for synthesizing drugs that can interact with "similar surfaces" within the body. Because AFM measures the force of an interaction between substrate and cells, interactions of a ligand or agonist should be measurable if well-defined molecules are placed on the tip of the AFM. Then it may be possible to use this ligand as a probe to determine the presence or absence of a receptor in impure, natural preparations. Perhaps more importantly, it may be possible to measure the interactive forces between agonist and receptor. This information will prove useful in designing of drug inhibitors or mimics.

1.8. Analysis of Micro-mechanical Properties

Contrast mechanism and image formation in AFM reflect a sum of many local forces and the micromechanical properties of the specimen. As the choice of specimen is shifted from rigid and hard materials (such as mica, graphite, tungsten) to soft and deformable biological materials, the dominating micro-mechanical properties shift from pure frictional to viscoelastic. Frictional forces on the atomic scale have been measured between two silica surfaces, two thin films, and between tungsten and graphite. By using local frictional force contours, fluorinated and hydrocarbon regions were distinguished in a Langmuir-Blodgett film. The hydrocarbons and fluorocarbons had separate domain structures: hydrocarbons as circular domains, fluorocarbons as the surrounding flat films. The frictional force was higher in the fluorinated region than in the hydrocarbon region. Such compositional studies provide a mechanism to identify individual components in a multicomponent sample like a cell membrane.

The viscoelastic properties of several soft, deformable biological materials, including cells, has also been obtained by a technique called force mapping. Such a study can be undertaken during on-line pharmacological and biochemical perturbations. In each imaging pixel, the tip is brought into proximity with the surface until a preset deflection is reached. The tip then retracts to its original position, and this process is repeated in every pixel. A height image is obtained from the amount of vertical piezo movement necessary at each point to obtain the preset deflection of the cantilever. For each pixel, a deflection versus distance curve is stored, which can be fitted to different models to obtain properties, such as the elastic modulus of the sample surface. Usually, the tip apex is approximated by a semisphere or a cone and the specimen by a spherical or planar model, depending on the shape of the features on the surface.

Thus, AFM is an imaging tool and also a system for analyzing micromechanical properties of cells, subcellular organelles, and macromolecules. It may be possible to study localized viscoelastic properties of molecular motor units, the distribution and propagation of contraction waves in a muscle cell, and the correlation between the calcium concentration wave and electrical propagation. One can also induce local shearing (frictional) force or pressure to assess the effects on the vascular system (mimicking the role of blood flow-induced shearing on vasorelaxation) or to distinguish pressure- or shear-sensitive ion channels in plasma membranes.

1.9. Simultaneous Multimodal Imaging

The simple design of AFM allows integrating it with other techniques, such as light fluorescence microscopy, laser confocal microscopy, and near-field scanning optical microscopies. Such integrated systems permit simultaneous multimodal imaging and provide independent verification with appropriately labeled markers. For example, using appropriately labeled fluorescent signals, one can identify specific areas and then use AFM to obtain high-resolution details.

1.9.1. Combined AFM and Light Fluorescence Microscope

Although conventional AFMs are ideal for high-resolution imaging, they could not be combined with large-aperture optical microscopes. In a few AFMs, however, the cantilever moves and the sample is stationary, permitting the addition of optical microscopes that have high numerical apertures. The most promising of these AFMs has the scanned-cantilever mode in which the cantilever position is accurately tracked by a scanned focused spot (Fig. 1) and is incorporated into an inverted fluorescence microscope. This combined fluorescence and force microscope has been used to image immunolabeled membranes and whole cells (Fig. 2). Fluorescent labels show remarkable correspondence among AFM images and the specificity of the molecules: such correspondence that one can obtain structural information at molecular resolution on biological macromolecules present individually or in small clusters, long as they have detectable fluorescent signals.

1.9.2. Combined Atomic Force Microscope and Confocal Microscope

Early combined AFM and laser-scanning confocal microscopes (LSCM) included features like a stationary sample stage, an AFM with an optical tracking system for the scanned cantilever, and either a scanned-beam or tandem design confocal microscope. The limitations of such systems include a limited scan range of both the AFM and confocal images. The scan range of the independently scanned confocal spot is limited to the size of the field of view of the objective and off-axis optical aberrations, and the scanned-cantilever AFMs with optical level detection require optical tracking of the cantilever for a large scan range. The latter constraint was overcome by the scanned-cantilever (tip) design with an optical tracking feature (the features explained previously in the combined AFM-fluorescence microscope), where the sample is scanned by a piezo system and the AFM tip and objective remain stationary. The AFM registers the topography of the sample surface, and the LSCM laser scans the surface to obtain fluorescent data on the same scan area. Although the AFM scan size is increased in this improved design, the confocal scan size is still limited by the objective. Moreover, although the AFM images are of the sample surface, the confocal image plane may not be the sample surface, but anywhere within the confocal slice, which could be no more than 100 nm thick.

A new combined AFM and LSCM allows simultaneous imaging of the sample surface in both modes, in addition to the conventional confocal imaging through the sample thickness. The salient features of such a combined microscope include a scanned-sample approach wherein the specimen is scanned above an inverted microscope objective with a fixed optical path for fluorescent LSCM imaging. An AFM positioned directly above the sample simultaneously measures the surface topography. Therefore, in this design the confocal spot and AFM cantilever remain stationary. Optical cantilever tracking is not required, and the confocal spot can be centered in the microscope’s objective. The AFM feedback system ensures that the focal point is on or near the surface of the specimen, so that when the cantilever is positioned at the confocal spot, the LSCM and AFM images are acquired in direct registration, allowing image features to be easily correlated.

In this combined multimodal system, the confocal plane can be selected from the topmost region on a specimen surface or anywhere through its depth, so it is quite possible to follow, for example, cytoplasmic signal transduction processes leading to changes in the cellular plasma membrane surface conformations.

1.9.3. Combined Atomic Force Microscopy and Electrophysiological Recording

A combined tapping-mode AFM and a scanning ion-conductance microscope have been developed recently (Fig. 1). One of the salient features of this combined microscope is a bent glass pipette used as both the force sensor and the conductance probe. The force-sensing capability allows measuring of the pipette deflection, which then is used to create surface images in both regular contact mode and tapping mode. The conductance-measuring capability allows recording the electrical current flow across pores in a suitable specimen. Using such a microscope, it is possible to image the structures of channels and receptors and to measure their functional states (conducting vs. nonconducting) (Fig. 3).

Figure 3. Examples of AFM multimodal imaging. (a) and (b) Simultaneous immunofluorescence and atomic force microscopy of amyloid b peptide (AbP) reconstituted into liposomes. All liposomes, with or without AbP, were imaged with AFM (a), a few are shown at higher magnification in the inset. The liposomes were treated with anti-AbP antibody and subsequently identified with fuorescein-conjugated second antibody. The AbP-carrying liposomes showed strong fluorescence signals (b). [S. K. Rhee, A. P. Quist, and R. Lai (1998) J. Biol. Chem. 273, 13379-13382]. (c), (d) and (e) Adhesion sites between a Xenopus retinal glial cell (XR1 cell line) and extracellular matrix material in a cell culture. The fluorescent images show the location of b-integrin (c) and f-actin (d) fibers detected by immunofluorescence, and the tapping mode AFM image (e) reveals the 3-D architecture of the focal point after removing of the cell body [R. Lai and R. Proksch (1997) Int. J. Imaging Syst. Technol. 8, 293-300]. (f) and (g): Simultaneously combined AFM and fluorescence-confocal microscopic images. The sample was a suspension of fuorescently labeled latex beads that were dried into a gel on a plastic diffraction grating. The lines of the grating are visible in the topographic AFM image (f) but not in the confocal fluorescent image (g). The two images allow distinguishing a nonfluorescent particle (left arrow) from a fluorescent particle (right arrow), although both appear as raised bumps in the AFM image. [P. E. Hillner, D. A. Walters, R. Lai, H. G. Hansma, and P. K. Hansma (1995) J. Micro. Soc. Am. 1, 123-126]. (h) and (i): Simultaneously combined AFM and scanning ion-conductance microscope (SICM) electrophysiology. (h) shows a tapping mode AFM image of a nucleopore membrane, and (i) shows the associated ionic conductivity image obtained by tapping mode SICM. Note that there are some differences in the pores detected by the two procedures. For example, the area circled in white shows a groups of pores that appear to be deep in the AFM image and highly conductive in the SICM image. The area circled in grey contains a large pore that appears deep in the AFM image but is nonconductive in the SICM image. The scale bars at the bottom are intensity-coded. Brighter is a greater height in the AFM image and a greater conductance in the SICM image.

Another approach is combining AFM with the patch-clamp technique in the same experiment. Such a combined technique records electrical current in the excised membrane patches from Xenopus oocytes that are attached to the patch pipette tip, while simultaneously observing the surface topology with AFM. Also, The membrane surface is also deformed by applying pressure through the patch pipette and observing the lateral displacement of features. However, the resolution is limited to about 10 nm.