NANO LETTERS
Microtubule Gliding and Cross-Linked Microtubule Networks on Micropillar Interfaces
2005 Vol. 5, No. 12 2630-2634
Wouter Roos,† Jens Ulmer,† Stefan Gra1 ter,† Thomas Surrey,‡ and Joachim P. Spatz*,† Max-Planck-Institute for Metals Research, Department New Materials & Biosystems, Heisenbergstrasse 3, 70569 Stuttgart, Germany, UniVersity of Heidelberg, Department of Biophysical Chemistry, INF 253, 69120 Heidelberg, Germany, and European Molecular Biology Laboratory, Cell Biology and Biophysics Unit, Meyerhofstrasse 1, 69117 Heidelberg, Germany Received September 18, 2005; Revised Manuscript Received October 21, 2005
ABSTRACT We combined biochemical and topographical patterning to achieve motor-driven microtubule gliding on top of microfabricated pillar arrays with limited and controllable surface interactions of gliding microtubules. Kinesins immobilized on pillar heads pushed microtubules from the top of one micropillar to the next bridging up to 20 µm deep gaps filled with buffer solution. Distances of more than 10 µm were by-passed, and microtubule buckling was occasionally observed. The velocity distributions of microtubules gliding on poly(dimethylsiloxane) (PDMS) pillars, on flat PDMS, and on glass were found to be different, most likely due to topological and/or chemical differences between the substrates. We also used pillar arrays to suspend cross-linked microtubule networks, whose structural characteristics were governed by the topographical characteristics of the pillar pattern. These experiments open new possibilities to study the dynamics and the self-organization of motor/ microtubule networks in defined topologically structured environments.
Microtubules play an essential role in intracellular transport and in cell division. For example, in neuronal axons, microtubules stabilize the axonal structure and serve as tracks for intracellular vesicle transport coming from and going toward the neuronal synapses.1 These transport events are mediated by molecular motors such as conventional kinesin (kinesin-1). During cell division in eukaryotes, microtubules and molecular motors form a complex protein network, called the mitotic spindle, around replicated chromosomes.2,3 The chromosomes are separated by a combination of microtubule polymerization and depolymerization and by active transport mediated by molecular motors. In vitro, the activity of molecular motors can be studied in gliding assays. In these experiments microtubules glide over motor-coated surfaces. A single motor is able to push a microtubule forward in a processive manner4 if it has a high duty ratio, like for example in the case of conventional kinesin.5 Simple dynamic networks of motors and microtubules can also be generated by self-organization in vitro.6 Here, we used microfabricated poly(dimethylsiloxane) (PDMS) pillars to perform microtubule gliding assays on motor protein-functionalized pillar tops (“heads”) and to * To whom correspondence may be addressed. E-mail:
[email protected]. † Max-Planck-Institute for Metals Research and University of Heidelberg. ‡ European Molecular Biology Laboratory. 10.1021/nl051865j CCC: $30.25 Published on Web 11/12/2005
© 2005 American Chemical Society
suspend static microtubule networks on cross-linker-functionalized pillar tops. We demonstrated that in gliding assays moving microtubules can bridge gap distances between pillars of several micrometers. Occasionally, buckling between pillars was observed and was evaluated quantitatively. We show that pillar distances influence the structural characteristics of suspended static microtubule networks. Compared to flat glass substrates the pillar structures are one step further toward mimicking the interior of cells, where regions with obstacles alternate with regions where translocation is undisturbed. Pillar arrays have been used in the past for other applications, for example, to study twodimensional actin networks,7 to measure traction forces in adhering cells,8 and to examine nonwetting phenomena.9 The PDMS pillars were made by photolithographic and replicate molding techniques (the methods are described in detail in the Supporting Information). Key to our experiments is selective immobilization of cross-linkers or motors on the pillar heads and passivation of the pillar sidewalls and of the bottom between pillars. This configuration ensures selective microtubule interactions with pillar heads only. This is different from experiments performed on microstructured, flat surfaces where microtubules glided at the bottom and not on top of the structure.10 We achieved controlled and
Figure 1. (A) Schematic view of flow chamber and microscope. The flow is in the in-plane direction. (B) Microtubules gliding over pillar tops (snapshot). Scale bar 20 µm.
specific positioning of biomolecules to pillar heads using a simple and efficient method of molecule “stamping”. PDMS is a hydrophobic material and PDMS pillar substrates are even more water repellent due to enclosure of water, which is know as the so-called lotus effect.11 This is an important consequence since a water drop deposited on the pillar substrate only wets the pillar heads, but not the region between the pillars if the droplet is in contact with the pillar substrate for a limited time. When the molecule to be stamped is dissolved in such a droplet, only the pillar tops become functionalized with the molecules. In a later step, the sidewalls of the pillars are passivated by nonspecific attachment of bovine serum albumine (BSA). The aqueous BSA solution penetrates the pillar array during the course of long incubation times. Our “stamping” method is a combination of chemical patterning and topographical patterning. Usually these approaches are performed separately.12 It needs to be pointed out that this method of stamping is only possible with pillar substrates having a rather small pillar gap. When pillars are 25 µm apart, the liquid runs between them. In our experiments, kinesin was not directly used in the stamping procedure, but biotin-BSA was stamped first onto the PDMS pillars. Subsequent protein attachments were then performed in a perfusion chamber ensuring that the pillar arrays were continuously surrounded by liquid preventing protein denaturation.13 Finally, the microtubule solution was introduced in the chamber. Figure 1A illustrates schematically our experimental setup. Figure 1B shows a fluorescence image of microtubules gliding on kinesin-functionalized pillar tops. We demonstrate that microtubules having lengths of approximately 15-30 µm glide over the pillar tops and can bridge distances of about 10 µm between pillars without difficulties. In contrast, similar experiments with actin filaments having lengths of approximately 20-30 µm did not show this bridging behavior. This difference can be understood by considering that microtubules have a persistence length of several millimeters,14 whereas filamentous actin has a much smaller Nano Lett., Vol. 5, No. 12, 2005
Figure 2. Microtubule motility assay on top of pillars. At 0 and at 116 s no microtubule was bound. In the intermediate time the microtubule attached to one pillar (at 12 s), glides over it to the next pillar, and finally detached from the second pillar. The microtubule buckled around 58 s and continued after 62 s. The average velocity of the gliding microtubule was 0.4 µm/s. Arrows pointing at the microtubule are added in some images, for emphasis only. Time is indicated in seconds; scale bar is 10 µm. Inset is a schematic image of a buckling microtubule on pillars. On the right pillar the microtubule is stalled by at least two rigid motors (small, dark circles). On the left pillar at least two motors (small, white circles) push the microtubule forward.
persistence length of around 15 µm.15,7 As microtubules possess a rigidity which is high compared to other cytoskeletal protein filaments such as actin and intermediate filaments,14,16 microtubule buckling can readily be observed. Several buckling experiments were performed to measure force/velocity relationships of microtubule polymerization and of directed motor movement.17,18 In our experiments, we observed microtubule buckling on the pillar heads as shown in Figure 2. Such events occurred when their leading end became immobilized and motors situated at the microtubule rear end on another pillar continued to push it forward. In Figure 2, a time series of images of a microtubule gliding from one to another pillar is shown. The microtubule buckled for a short time at 62 s. The occurrence of microtubule buckling could have several reasons. It could result from pillar topography, transient interactions of inactive motors or of an unblocked part of the surface with the front of the microtubule at one pillar. All such events cause increased friction or pinning of the microtubule while the rear part of the microtubule is still being pushed by motors on the other, upper left pillar. During buckling of the microtubule between the two pillars, this rear part of the microtubule did not change orientation at the attachment site at the pillar head, which is an indication of attachment via several motors. This was also estimated from the buckling 2631
force: For the buckling microtubule on top of the pillars, a reference frame with the connecting line between the two pillars denoted as the x-axis and an axis perpendicular to this line as the y-axis was taken (Figure 2 inset). In this frame of reference the following boundary conditions are relevant: y(0) ) y(L) ) 0, and due to the rigidity of the microtubules also y′(0) ) y′(L) ) 0 was supposed. The variable L denotes the pillar gap, and the prime denotes the derivative with respect to x. The critical buckling force Fc of a microtubule fixed at two sides, without the possibility to pivot around the attachment points, is given by19 Fc )
4π2κ L2
(1)
where κ is the microtubule bending elasticity and L is the microtubule length between the attachment points at the onset of buckling. Various values for κ are found in the literature.14,20,17 With the values κ ) 2.2 × 10-23 N m2 14 and L ) 9.6 µm, a critical buckling force of 9.4 pN was calculated for the buckling presented in Figure 2. At least two kinesin motors were required to generate this force as the typical maximum force per motor is around 5 pN.21 During all buckling events on pillar tops, the buckled microtubules remained in focus, similar to buckling events on flat surfaces.18 This implies that both a flat surface and the pillar tops prevent a microtubule from buckling downward or upward. We next measured the velocities of microtubules gliding on PDMS pillar substrates, on flat PDMS, and on glass. We demonstrate that the mean velocity of moving microtubules is of the same order of magnitude for all three substrate types. There was a slight decrease of the average velocity from 470 nm/s on glass similar to earlier measurements22,5 to 440 nm/s on flat PDMS and finally to 430 nm/s on PDMS pillar substrates. More striking, however, were the differences of the velocity distributions on the different substrates (Figure 3) with the velocity distribution on glass being the narrowest and those on either of the PDMS substrates being relatively broad. A first possible explanation for the difference between velocities on glass and on PDMS could be an increased nonspecific interaction between the PDMS and the microtubules or an increased fraction of inactivated motors on PDMS as compared to the glass surface. This latter possibility is likely given the observation that both on the PDMS pillar substrate and on the flat PDMS substrate about half of the microtubules were immobile, while on glass more than 95% of the microtubules moved. This indicates a different passivation quality on glass and PDMS interfaces. The geometry of the pillar substrates could have two additional effects on the gliding velocity of the microtubules. On one side it could be argued that the microtubule is in contact with a smaller area of substrate reducing possible friction of the gliding microtubule with the surface and allowing consequently a higher velocity to be achieved. On the other hand, it could be argued that there was a decrease of the velocity as a result from the three-dimensional geometry. Gliding microtubules fluctuate due to Brownian motion 2632
Figure 3. Histogram of microtubule velocity on several kinesincoated substrates. Experiments on PDMS pillars, on flat PDMS, and on glass were performed. Bin width is 25 nm/s. A Gaussian fit to the histogram data is plotted as a black line. On the pillars 94 microtubules were analyzed, the mean velocity Vj is 430 ( 140 nm/s where the error is the standard deviation. On flat PDMS, 82 microtubules were analyzed and the mean velocity is 439 ( 115 nm/s. On glass 59 microtubules were analyzed, the mean velocity is 469 ( 70 nm/s.
potentially also toward the pillar base at the moment when the microtubule front reaches the next pillar. In this case the microtubule bends and reduces its average velocity. It seems that this latter effect dominates, because the velocity distribution of microtubule gliding on PDMS pillars was broadened toward slower velocities as compared to gliding on flat PDMS. In addition to establishing single microtubule gliding on motor-functionalized microstructured pillar substrates, we also used pillar arrays to generate large suspended microtubule networks connected to pillar tops and interconnected by chemical and biological cross-linkers. Electron microscopy was used as an alternative method to verify that such a cross-linked microtubule network lies on top of the pillar heads (Figure 4A). Here, microtubules were attached electrostatically to pillar heads that were coated with polylysine and static microtubule-microtubule interactions were formed by the chemical cross-linker glutaraldehyde (see Supporting Information). Nano Lett., Vol. 5, No. 12, 2005
Figure 4. Cross-linked microtubules on pillar tops. (A) Microtubules attached to polylysine-coated pillar heads were cross-linked by glutaraldehyde. Scanning electron micrograph, scale bar 5 µm, 30° angle of view. The inset shows a high magnification graph of the microtubules on the pillar tops. Inset full width 10 µm, 45° angle of view. (B and C) Biotin-labeled microtubules cross-linked by inactive Eg5. Fluorescence images, scale bar 20 µm. (B) Pillars with a minimum distance of 5 µm. On this part of the substrate the microtubules formed a network. At some spots pillars were missing, but still network formation was initiated. (C) Pillars with a minimum distance of 11 µm. On this part of the substrate the pillars were too far apart to serve as a scaffold for network formation, instead microtubule selforganization on the prestructured surfaces was observed.
Alternatively, biochemical strategies were employed to attach microtubule networks to pillar heads and to crosslink microtubules within such a network. Biotinylated microtubules were used to specifically bind them directly to biotin-BSA/streptavidin coated pillar tops. The microtubules on the pillar tops were then cross-linked biochemically using the tetrameric kinesin Eg5 (see Supporting Information). This motor is able to cross-link two microtubules23 and did so in our experiment in a static manner because of the absence of ATP. For pillar distances of about 5 µm we observed that twodimensional static networks were formed (Figure 4B). These networks are similar to the networks formed by cross-linking actin filaments with filamin on top of pillar substrates.7 By increase of the pillar distances, a regime was reached where the pillars are too far apart to achieve global cross-linking of the microtubules. The cross-linking became localized, and assembly of microtubule clusters on pillar tops was observed (Figure 4C). Obviously, the prestructured surface determines via the pillar spacing the organization of the microtubule clusters. The results presented here show that microfabricated threedimensional structures can be applied to assemble biofunctional filament systems. We demonstrate this using microtubules and PDMS pillar substrates. Similar assemblies were previously generated with actin filaments on top of silicon and epoxy pillars.7 In the first part of our study we measured microtubule gliding driven by motors specifically immobilized on pillar tops. To immobilize motors or microtubules on the pillar heads, a novel, fast, and easy stamping method was used. Specificity of attachment is achieved by biotin-streptavidin biochemistry. In the second part of our study, we used pillar arrays to assemble static networks of cross-linked microtubules. The pillar structures are ideal to study self-organization of dynamic motor/microtubule systems on prestructured templates in the future. Surface interactions are minimized, and at the same time anchoring points are present to immobilize the structures and to influence self-organization. A further advantage of this PDMS pillar system is that it allows for monitoring forces generated by such self-organising systems by observing the bending of flexible pillars. Nano Lett., Vol. 5, No. 12, 2005
In conclusion, pillar arrays promise to become a versatile tool for studying motor dynamics and microtubule network formation. The combination of micrometer-sized MEMS and protein filaments having nanometer diameters is a useful direction in resolving open questions of intracellular biofilament dynamics such as buckling behavior of microtubules pushed by motors against chromosome or cell membrane, cross-link and network formation behavior of microtubules during mimetic microtubule spindle assembly, or mimetic chromosome positioning and segregation by microtubules. Acknowledgment. We thank Jovana Drinjakovic, Steffi Kandels-Lewis, and Arne Seitz (EMBL) for protein purifications and labeling. VolkswagenStiftung (VW I/80956) and the Max-Planck-Society are acknowledged for financial support. The work has also been initiated in a STREP Program of the European Community (ACTIVE BIOMICS, Contract No. 516989) and the Priority Program SPP 1164: Nano- & Microfluidic of the DFG. Supporting Information Available: Experimental details including proteins used, pillar fabrication, flow cell protocol, light and electron microscopies, and control expeiments. This material is available free of charge via the Internet at http:// pubs.acs.org. References (1) Hirokawa, N. Science 1998, 279, 519-526. (2) Reinsch, S.; Go¨nczy, P. J. Cell Sci. 1998, 111, 2283-2295. (3) Wittmann, T.; Hyman, A.; Desai, A. Nat. Cell Biol. 2001, 3, E28E34. (4) Howard, J.; Hudspeth, A. J.; Vale, R. D. Nature 1989, 342, 154158. (5) Young, E. C.; Mahtani, H. K.; Gelles, J. Biochemistry 1998, 37, 3467-3479. (6) Surrey, T.; Nedelec, F.; Leibler, S.; Karsenti, E. Science 2001, 292, 1167-1171. (7) Roos, W. H.; Roth, A.; Konle, J.; Presting, H.; Sackmann, E.; Spatz, J. P. ChemPhysChem 2003, 4, 872-877. (8) Tan, J. L.; Tien, J.; Pirone, D. M.; Gray, D. S.; Bhadriraju, K.; Chen, C. S. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 1484-1489. (9) Burton, Z.; Bhushan, B. Nano Lett. 2005, 5, 1607-1613. (10) Hess, H.; Clemmens, J.; Howard, J.; Vogel, V. Nano Lett. 2002, 2, 113-116. (11) Patankar, N. A. Langmuir 2004, 20, 8209-8213. 2633
(12) Hess, H.; Clemmens, J.; Matzke, C. M.; Bachand, G. D.; Bunker, B. C.; Vogel, V. Appl. Phys. A: Mater. Sci. Process. 2002, 75, 309313. (13) Janson, M. E.; Dogterom, M. Biophys. J. 2004, 87, 2723-2736. (14) Gittes, F.; Mickey, B.; Nettleton, J.; Howard, J. J. Cell Biol. 1993, 120, 923-934. (15) Le Goff, L.; Hallatschek, O.; Frey, E.; Amblard, F. Phys. ReV. Lett. 2002, 89, 258101_1-4. (16) Mu¨cke, N.; Kreplak, L.; Kirmse, R.; Wedig, T.; Herrmann, H.; Aebi, U.; Langowski, J. J. Mol. Biol. 2004, 335, 1241-1250. (17) Dogterom, M.; Yurke, B. Science 1997, 278, 856-860. (18) Gittes, F.; Meyho¨fer, E.; Baek, S.; Howard, J. Biophys. J. 1996, 70, 418-429.
2634
(19) Landau, L. D.; Lifschitz, E. M. Lehrbuch der Theoretischen Physik, VII Elastizita¨ts- theorie; Akademie Verlag: Berlin, 1991; problem 2, p104. (20) Elbaum, M.; Fygenson, D. K.; Libchaber, A. Phys. ReV. Lett. 1996, 76, 4078-4092. (21) Svoboda K.; Block, S. M. Cell 1994, 77, 773-784. (22) Surrey, T.; Elowitz, M. B.; Wolf, P.; Yang, F.; Ne´de´lec, F.; Shokat, K.; Leibler, S. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 42934298. (23) Kapitein, L. C.; Peterman, E. J. G.; Kwok, B. H.; Kim, J. H.; Kapoor, T. M.; Schmidt, C. F. Nature 2005, 435, 114-118.
NL051865J
Nano Lett., Vol. 5, No. 12, 2005
