Approaching microtubule structure with the scanning tunneling ...

3127

Journal of Cell Science 107, 3127-3131 (1994) Printed in Great Britain © The Company of Biologists Limited 1994

Approaching microtubule structure with the scanning tunneling microscope (STM) M. Maaloum*, D. Chrétien, E. Karsenti and J. K. H. Hörber† European Molecular Biology Laboratory, Meyerhofstrasse 1, 69112 Heidelberg, Germany *Present address: Institut Charles Sadron, Rue Boussingault 6, 67000 Strasbourg, France †Author for correspondence

SUMMARY We demonstrate that the scanning tunneling microscope can be used to obtain information about arrangement of tubulin subunits in the microtubule wall. Long rows of subunits with a periodicity of 3.8±0.4 nm were clearly visible in the images of microtubules. The separation between the rows of subunits was 4.8±0.4 nm. Close inspection of two images revealed another periodicity of 7.8±0.4 nm in the contour levels of the protofilaments. This indicates that α and β tubulin monomers can be resolved. In these areas the monomers were arranged according to

a ‘B-type’ lattice. Scanning tunneling microscope images confirm that the lateral contacts between tubulin monomers in adjacent protofilaments are compatible with a three-start, left-handed helix model. This study demonstrates that scanning tunneling microscopy can give direct information on the structure and organization of macromolecular assemblies and can complement the classical methods of electron microscopy and X-ray scattering.

INTRODUCTION

and triplets, and it may also have a bearing on the stability and dynamic behavior of microtubules. There are three types of possible lattice organization for the tubulin molecules in microtubules (see Fig. 1). The ‘A-type’ lattice shows ‘alternating’ heterologous monomers (α-β-α...) on the three-start helices. The ‘B-type’ lattice shows alternating homologous monomers (α-α-α..., or β-β-β...,). For a microtubule consisting of 13 protofilaments, the lattice will be interrupted by at least one ‘seam’ corresponding to the ‘A-lattice’. For 14 protofilaments, both types of lattice will be discontinuous. The ‘mixed’ lattice corresponds to a mixture of the two preceding ones. In this paper we show that scanning tunneling microscopy (STM) can give direct information concerning microtubule structure. Our results are consistent with the left-handed threestart helix model originally proposed by Amos and Klug (1974). In some areas scanned the α and β subunits could be resolved and were arranged according to the ‘B-lattice’.

Microtubules are hollow cylinders made of tubulin heterodimers (α, β; Mr ≈ 100×103), which play essential roles in cell organization, division and motility (Hyams and Lloyd, 1994). They interact with a large number of proteins that can be grouped under the epithet microtubule-associated proteins (MAPs); these include the motor proteins kinesin and dynein, and cross-linking proteins such as nexin. A considerable amount of work is in progress with the aim of understanding how these cellular components function. To achieve this aim will ultimately require the collection of a great deal of biochemical and structural information. On the structural side it is necessary to know the organization of tubulin in the microtubule assemblies. Although some recent progress has been made on the structure of the tubulin molecule (Downing and Jontes, 1992), little additional information has been obtained about the structure of microtubules since the pioneering work of Amos and Klug in the 70’s (Amos and Klug, 1974). The exact organization of tubulin molecules in the microtubule wall is unknown and, until recently, the handiness of tubulin subunit organization in helices has been a matter of debate. A common feature of all structural models of microtubules is the head-to-tail alignment of the tubulin heterodimer to give an invariant α/β sequence from dimer to dimer along the protofilaments. The difference between the models lies in the lateral packing of the subunits in adjacent protofilaments. The surface lattice organization can be expected to be an important parameter for the binding of associated proteins and motor proteins, for the construction of microtubule doublets

Key words: tubulin, microtubule, scanning tunneling microscopy

MATERIALS AND METHODS Preparation of pure tubulin Tubulin was isolated from calf brain by two cycles of assembly-disassembly in the presence of glycerol, followed by phosphocellulose chromatography (Mitchison and Kirschner, 1984). Three fresh calf brains were obtained from the slaughterhouse and transferred into PB (polymerization buffer: 100 mM PIPES, 1 mM EGTA, 1 mM MgCl2, pH 6.8 with KOH) at 4°C. Meninges were removed on ice and the cerebellums were discarded. Brains were cut up, washed once in PB at 4°C and weighed. The same weight of PB

3128 M. Maaloum and others

Top

Fig. 1. Schematic representation of three types of possible lattice organization for the tubulin molecules in microtubules. Taken from Wade and Chrétien (1993), adapted from the work of Amos and Klug (1974). containing protease inhibitors (10 µg/ml aprotinin, 10 µg/ml leupeptin, 10 µg/ml pepstatin, 500 µM PMSF, 200 µM benzamidine and 1.0 µg/ml O′-phenanthroline) was added to the preparation. Brain pieces were pounded in a cold room using a Waring blendor running at low speed. Three runs of 20 seconds each were performed with pauses of 20 seconds between each run. The suspension was transferred into type I 19 rotors (Beckman Instruments GmbH, Munich, Germany) and centrifuged at 17,000 rpm, 60 minutes, 4°C. Supernatants were pooled and 0.5 volume of glycerol 12 M in PB, and GTP to a final concentration of 1 mM, were added. The suspension was incubated for 45 minutes at 37°C, transferred into Ti 45 rotors (Beckman) and centrifuged at 30,000 rpm, 30 minutes, 30°C. Supernatants were discarded and the pellets were dissociated in small volumes of PB at 4°C using a 10 ml pipette. The suspension was homogenized using a Dounce homogenizer (type L), incubated for 30 minutes at 4°C and centrifuged in a Ti 45 rotor at 40,000 rpm, 30 minutes, 4°C. Then 0.5 vol. of pre-warmed PB-glycerol was added to the pooled supernatants with 1 mM GTP and incubated for 45 minutes at 37°C. The suspension was transferred in a Ti 70 rotor (Beckman) and centrifuged at 33,000 rpm, 30 minutes, 30°C. Supernatants were discarded and pellets were stored overnight at −80°C. The day after the pellets were thawed at 4°C and resuspended in CB (cycling buffer: 50 mM PIPES, 1 mM EGTA, 0.2 mM MgCl2, pH 6.8, with saturated KOH), containing 1 mM GTP and protease inhibitors (1.0 µg/ml aprotinin, 1.0 µg/ml leupeptin, 1.0 µg/ml pepstatin, 100 µM benzamidine and 0.5 µg/ml O-phenanthroline). The suspension was homogenized using a Dounce homogenizer (type S), incubated for 45 minutes at 4°C and centrifuged in a Ti 70 rotor at 38,000 rpm, 30 minutes, 4°C. A 23 ml volume of supernatant was loaded onto a phosphocellulose column (Whatman P11), and equilibrated in CB and protease inhibitors at 4°C. The column was eluted at 4 ml/min and fractions of 3 ml were collected. The peak flow-through fractions were assayed for protein concentration (Bio-Rad Protein Assay, Bio-Rad Laboratories GmbH, München, Germany), and pooled in three batches, depending on tubulin concentration. The pooled fractions were adjusted to 1 mM GTP, aliquoted, snap frozen and stored in liquid nitrogen. SDS-gel electrophoresis of the three batches of tubulin prepared from the peak flow-through fractions of the phosphocellulose column was carried out on 7.5% polyacrylamide gels (Protean II mini cuves, Bio-Rad). Gels were stained with Coomassie Brilliant Blue (see Fig. 2). Immunofluorescense Microtubules were assembled according to the method of Mitchison and Kirschner (1984). They were diluted to 30 µM in buffer contain-

Bottom

Fig. 2. Analysis by SDS-polyacrylamide gel electrophoresis of the calf on a Macintosh IIcx computer. The upper trace shows the profile of the molecular mass markers and the lower trace shows the profile of the tubulin preparation. The highly contrasted band at the bottom of the gel is due to gel folding. No contaminants were detected in our sample. ing 80 mM PIPES, 1 mM MgCl2, 1 mM EGTA, pH 8.0, with saturated KOH, in the presence of 20 µM taxol. Microtubules were prepared by centrifugation on a small piece of glass coated with a conducting, optically transparent, layer of Indium tin oxide (ITO, Sanyo) or on a highly oriented pyrolitic graphite (HOPG) surface at 20,000 rpm for one hour according to the method of Evans et al. (1985). The samples were post-fixed in methanol at −20°C for 5 minutes, then incubated with a monoclonal antibody and with a fluorescein-labeled anti-mouse antibody and finally mounted in Mowiol. Fig. 3 shows a photograph of fluorescent microtubules on an ITO glass surface taken using a Zeiss Axiophotmicroscope. STM recording The scanning tunneling microscope used was a non-commercial ‘pocket-size’ type (Binnig et al., 1982; Binnig and Smith, 1986; Smith et al., 1987) equipped with tungsten tips etched in KOH by alternating current. There are two ways of recording STM images: one, the constant height mode, records the variations of the tunneling current flowing between sample and tip while the sample surface is scanned

_^

100 µm

Fig. 3. Immunofluorescence microscopy of the microtubule preparation on an ITO-glass surface.

&_

STM observation of microtubules 3129

A

Fig. 5. STM image of a dehydrated, fixed microtubule observed several hours after preparation. This image reveals many distortions in microtubule structure. The bias voltage was 80 mV with a tunneling current of 40 pA.

B

Fig. 4. STM images of unfixed microtubules deposited on ITO surface. These images were obtained in constant current mode. (A) Image of the end of a microtubule observed about one hour after preparation. (B) Image of an intermediate part of a microtubule. The image accentuates the left-handed rotational sense of the microtubule helix. The bias voltage between tip and sample was 50 mV with a tunneling current of 40 pA for both images. by the tip in an x-y plane. The variation of the tunneling current is translated in a gray scale providing the image-contrast information, with the resolution of structural details limited by the tip and the frequency response of the preamplifier. In the other mode, a fixed current is maintained between tip and sample, and the tip is enabled to adjust its distance from the surface automatically with the help of a feedback circuit as it scans the sample (constant current mode). The motion of the tip in the z direction is then plotted as the surface contour. With this mode the height of the structures can be determined directly by the calibrated z-movement of the piezotube. The resolution in this mode is restricted by the tip and by the time constant of the feedback circuit. Microtubule samples were prepared by centrifugation as described for immunofluorescense, but the holder was immediately transferred to the STM, after blotting most of the buffer off. The observations were carried out, with a thin film of water remaining associated with microtubules. Some microtubules could be found in preparations on HOPG, a substrate used for the first attempts to image microtubule structures several years ago (Simic-Krstic et al., 1989; Hameroff et al., 1990). However, this surface did not adsorb microtubules very well. Moreover, the HOPG surface has structural features that make

it difficult to find microtubules in large scan images (a necessary step in finding the microtubules on the substratum in the first place). The ITO glass surfaces had a smooth aspect in STM images on the scale of the tubulin subunits, allowing clear identification of microtubule structures even at low resolution. Therefore we used the ITO support throughout this study. In addition, the same preparation of microtubules could be easily observed, by immunofluorescense microscopy, to determine suitable preparation conditions before the observation by STM (see Fig. 3). In some experiments, microtubules were fixed with glutaraldehyde before centrifugation, although we found it unnecessary in order to obtain good images by STM.

RESULTS AND DISCUSSION Fig. 4 shows STM images of taxol-stabilized, unfixed microtubules observed about one hour after preparation. The height of the structures measured by the z-movement of the tip in the feedback mode was 24 nm and indicates that they were still hydrated due to water molecules tightly bound to the microtubules, whereas most of the surface water covering the substratum had evaporated. The overall appearance, especially the conservation of the microtubule cylinder, suggests that it was still reasonably preserved. Fully dehydrated fixed microtubules, left on the support for several hours to days, revealed many distortions in their structure, and were flattened on the substratum (Fig. 5) giving height measurements of less than 15 nm. The height measurements of 24 nm fit well with the accepted microtubule diameter of about 25 nm. This indicates that microtubules retain their cylindrical structure and do not flatten on the substratum when observed within one hour after sample preparation. Height measurements with the STM do not necessarily correspond to pure topography, the image contrast often being a complex combination of height information, i.e. the shape of the molecules, and the electronic surface structure. This second parameter reflects the energy required to transfer electrons from the tip to the surface of the molecule with the tip polarity used. In our experiments, it seems that the parameter determining the image was mainly the topography, since the height measurements and structural details corresponded well to what is known about microtubules from EM studies. Long rows of subunits with an average periodicity of

3130 M. Maaloum and others Table 1. Analysis of microtubule images Microtubules 1 2 3 4 5 6

Periodicity along the rows (nm)

Separation between the rows (nm)

Angle relative to the longitudinal axis (deg.)

3.8±0.3 4.0±0.3 3.8±0.4 3.7±0.5 3.6±0.4 3.9±0.3

5.0±0.3 4.8±0.4 4.6±0.5 4.7±0.4 5.0±0.2 4.8±0.3

80±2 82±2 78±3 81±3 78±2 80±3

A

Comparison between six microtubule images (mean ± s.d.).

3.8±0.4 nm were clearly visible on STM images of microtubules. The separation between the rows of subunits was on average 4.8±0.4 nm. Each row was slightly shifted with respect to its neighbor by approximately 0.9 nm, giving rise to transverse stripes of subunits with a tilt angle of about 80±3 degrees relative to the longitudinal axis of the microtubule. The variations in the distances of tubulin subunits (see Table 1) are not due to the inaccuracy of the STM, but reflect both imperfections of the microtubule structures and the difficulty of identifying the center of a tubulin subunit by a technique sensing only the electronic surface properties of the structure. Nevertheless, the images are well compatible with the left-handed three-start helix microtubule model originally proposed by Amos and Klug (1974). On ten STM images of microtubules, nine showed a left-handed orientation for the three-start helices, and one was right-handed. The right-handed structure had a thickness of 8 nm, suggesting that it was a sheet of protofilaments. This particular structure probably corresponded to a view of the interior of a microtubule wall. This can happen when microtubules open during adsorption on the substratum. Also, the presence of tubulin sheets at the ends of microtubules have been observed by EM of negatively stained specimens (Simon and Salmon, 1990). Close inspection of STM images made in constant height mode (Fig. 6) revealed an additional periodicity of 7.6±0.4 nm more clearly than the images made in constant current mode. Fig. 7 shows a contour level representation of Fig. 6A with the clearly visible alternation along the protofilaments of darker and brighter structures. This suggests that the α and β tubulin subunits are differentially detected by the STM probe, one appearing consistently darker than the other. It seems unlikely that the difference observed could be attributed to a size difference of the two subunits because their molecular masses are very similar and their sequences are highly homologous. One possible explanation is that one of the two subunits is more oriented towards the interior of the microtubule. Indeed, this would result in a lower signal from the tip passing in constant height mode over the surface of the microtubule. An alternative explanation is that the periodic change in contrast along the protofilament length is produced by conductance variations due to different charged groups exposed at the surface, similar to the observations made on gramicidin, showing that differently charged amino acids modulate the topographic information (Hörber et al., 1991). The first hypothesis, however, fits well with the structural data obtained by Amos and Klug (1974) using electron microscopy. If the difference in contour level indeed represents the two subunits (α and β, not knowing which is which), the images

B

Fig. 6. (A,B) Two examples for the arrangement of the tubulin protofilaments in unfixed microtubules imaged by the STM in constant height mode. The protofilament structure consists of two distinct morphological subunits alternating along a linear chain at approximately 4 nm intervals.

obtained show that the observed microtubule stretches were built according to a three-start helix with a B-lattice, meaning that the nearest neighbor of each subunit is its homolog (α-α, β-β in two adjacent rows, see Fig. 1). This fits with recent results by Song and Mandelkow (1993); however, we still do not have enough images to make sure that other configurations cannot exist (Mandelkow et al., 1986; Wade and Chrétien, 1993) (see Fig. 1). Conclusion This study demonstrates that STM can give direct information on the structure and organization of macromolecular assemblies, and can complement the classical methods of electron

STM observation of microtubules 3131 process and the structure of microtubules growing under more physiological conditions. The interaction of microtubules with motor proteins and microtubule-associated proteins could also be examined at the structural level using a combination of tunneling and force microscopy. REFERENCES

Fig. 7. Contour level representation of the microtubule shown in Fig. 6A. The image shown in Fig. 6A was low pass filtered and digitized in a Macintosh computer using Adobe Photoshop. This procedure reveals more clearly the alternation of subunits along the length of a protofilament.

microscopy and X-ray scattering. The technique provides very local information with high contrast, making averaging unnecessary. Therefore, the method can show all the imperfections in regular structures, reflecting the response to the actual environment. In the case of microtubule structures, this study paves the way for further investigations on the polymerization

Amos, L. A. and Klug, A. (1974). Arrangement of subunits in flagellar microtubules. J. Cell Sci. 14, 523-549. Binnig, G., Rohrer, H., Gerber, C. and Weibel, E. (1982). Tunneling through a controllable vacuum gap. Appl. Phys. Lett. 40, 178. Binnig, G. and Smith, D. P. E. (1986). Single-tube three-dimensional scanner for scanning tunneling microscopy. Rev. Sci. Instrum. 57, 1688. Downing, H. D. and Jontes, J. (1992). Projection map of tubulin in zincinduced sheets at 4 Å resolution. J. Struct. Biol. 109, 152-159. Evans, L., Mitchison, T. J. and Kirschner, M. W. (1985). Influence of the centrosome on the structure of nucleated microtubules. J. Cell Biol. 100, 1185-1191. Hameroff, S., Simic-Krstic, Y., Vernetti, L., Lee, Y.C., Sarid, D., Wiedmann, J., Elings, V., Kjoller, K. and McCuskey, R. (1990). Scanning tunneling microscopy of cytoskeletal proteins: Microtubules and intermediate filaments. J. Vac. Sci. Technol. A8, 687-691. Hörber, J. K. H., Schuler, F. M., Witzemann, V., Schröter, K. H., Müller, H. and Ruppersberg, J. P. (1991). Imaging of cell membrane proteins with a scanning tunneling microscope. J. Vac. Sci. Technol. 9, 1214-1218. Hyams, J. S. and Lloyd, C. W. (1994). Microtubules (ed. J. S. Hyams and C. W. Lloyds). Wiley-Liss, Inc., New-York. Mandelkow, E. M., Schultheiss, R., Rapp, R., Müller, M. and Mandelkow, E. (1986). On the surface lattice of microtubules: helix starts, protofilament number, seam, and handedness. J. Cell Biol. 102, 1067-1073. Mitchison, T. and Kirschner, M. (1984). Microtubule assembly nucleated by isolated centrosomes. Nature 312, 232-237. Simic-Krstic, Y., Kelly, M., Schneiker, C., Krasovich, M., McCuskey, R., Koruga, D. and Hameroff, S. (1989). Direct observation of microtubules with the scanning tunneling microscope. FASEB J. 3, 2184-2188. Simon, J. R. and Salmon, E. D. (1990). The structure of microtubule ends during the elongation and shortening phases of dynamic instability examined by negative-stain electron microscopy. J Cell Sci. 96, 571-582. Smith, D. P. E., Bryant, A., Quate, C. F., Rabe, J. P., Gerber, C. and Swalen, J. D. (1987). Images of lipid bilayer at molecular resolution by scanning tunneling microscopy. Proc. Nat. Acad. Sci. USA 84, 969-972. Song, Y. H. and Mandelkow, E. (1993). Recombinant kinesin motor domain binds to beta-tubulin and decorates microtubules with a B surface lattice. Proc. Nat. Acad. Sci. USA 90, 1671-1675. Wade, R. H. and Chrétien, D. (1993). Cryoelectron microscopy of microtubules. J. Struct. Biol. 110, 1-27. (Received 31 March 1994 - Accepted 18 July 1994)