NANO LETTERS
Cellular Unbinding Forces of Initial Adhesion Processes on Nanopatterned Surfaces Probed with Magnetic Tweezers
2006 Vol. 6, No. 3 398-402
Nadine Walter,†,‡ Christine Selhuber,†,‡ Horst Kessler,§ and Joachim P. Spatz*,† Max-Planck-Institute for Metals Research, Department New Materials and Biosystems, Heisenbergstrasse 3, D-70569 Stuttgart, Germany, and UniVersity of Heidelberg, Institute for Physical Chemistry, Department Biophysical Chemistry, D-69120 Heidelberg, Germany, and Technical UniVersity of Munich, Department Chemie, Lehrstuhl II fu¨r Organische Chemie, Lichtenbergstrasse 4, D-85747 Garching, Germany Received November 3, 2005; Revised Manuscript Received December 30, 2005
ABSTRACT To study the dependence of unbinding forces on the distance of molecularly defined and integrin specific c(−RGDfK−) ligand patches in initial cellular adhesion processes, we developed a magnetic tweezers setup for applying vertical forces of up to 200 pN to rat embryonic fibroblasts. The ligand patch distance is controlled with a hexagonally close packed pattern of biofunctionalized gold nanoparticles prepared by block− copolymer micelle nanolithography. Each gold nanoparticle potentially activates up to one rvβ3-integrin. The distances between the gold nanoparticles determine the separation of individual integrins and thus the assembly of integrin clusters. The results show an increase in cellular unbinding forces from approximately 6 to more than 200 pN for a decreasing ligand distance of 145 to 58 nm after 5 min of cell adhesion. Furthermore, we observe a strong dependence on adhesion time during the first 10 min of cell surface contact suggesting an active, cooperative cell response that is controlled by the spacing between individually activated integrins.
Integrin-mediated cell adhesion is a dominant process in cell-extracellular matrix (ECM) adhesion. Focal contacts, the tight anchoring points of tissue cells to their environment, are characterized by the clustering of the transmembrane protein integrin and the hierarchical assembly of intracellular adhesion proteins that bind to the actin cytoskeleton.1 Most studies on integrin-mediated cell adhesion are performed on long-term adhesion and mature focal contacts, which develop on a time scale of hours.2,3 The dependence of their formation, size, and load resistance on various specific adhesion ligands and ligand densities, surface structures, and surface rigidity has been the subject of many publications.4-7 In contrast to this, little is known about the physical properties of initial adhesion formation regarding forces and time scales. Although it has been shown that hyaluronanmediated interactions of the pericellular matrix precede early integrin binding,8 the stages of integrin-mediated adhesion remain largely unexplored. Open questions concern the magnitude of initial cellular unbinding forces, the time scale * To whom correspondence may be addressed:
[email protected]. † Max-Planck-Institute for Metals Research and University of Heidelberg. ‡ Both authors contributed equally to the content of this work. § Technical University of Munich. 10.1021/nl052168u CCC: $33.50 Published on Web 02/01/2006
© 2006 American Chemical Society
of integrin activation, ligand density dependencies, and whether cooperative effects occur. In earlier work, cell spreading and integrin activation during long-term adhesion were studied as a function of ligand distance on nanopatterned and biofunctionalized interfaces.9 It was found that at ligand distances of 73 nm or greater, focal contact formation and cell spreading were significantly reduced. From these experiments it is suggested that there exists a critical length scale between 58 and 73 nm for the separation of single integrins which is crucial for integrin clustering and focal contact formation.9 A key aspect is whether distances between integrin ligands already play a role during initial adhesion and how this can be monitored. Force spectroscopy is a powerful tool for quantifying single-molecule interactions also in the case of cell-cell or cell-substrate adhesion. Benoit et al. have quantified deadhesion forces between cells at the resolution of individual cell-adhesion molecules by single-molecule force spectroscopy using the atomic force microscopy technique.10 Garcia et al. used centrifugation forces to measure adhesion strength of cells.11 We have developed a noninvasive approach to pull vertically on cells during their initial adhesion to nanopat-
Figure 1. Sketch of the experiment. A cell adhering onto a substrate is lifted by a magnetic force. The force is applied to the cell via a magnetic bead which is covalently bound to the cell. The detachment of the cell is observed by RICM. Insets: (A-C) Visualization of a detaching cell during the experiment. (A) Cell with bead (arrow) in bright field microscopy. (B) Same cell in RICM with contact area marked (circle). (C) Situation after liftoff in RICM.
Figure 2. Scheme of a cell membrane section on nanostructured and biofunctionalized glass slides. The c(-RGDfK-) ligand arrangement on a glass slide via gold nanoparticles and the integrin receptors of the cell are sketched. SEM images of the hexagonally ordered gold nanostructures with different particle spacing, L, are shown at the bottom: (A) L ) 69 nm, (B) L ) 93 nm, and (C) L ) 145 nm.
terned surfaces and to measure their detachment force. The nanopatterned surfaces used in this study consist of hexagonally arranged gold nanoparticles with a lateral interparticle spacing, L, ranging from 20 to 150 nm ((10%) and a diameter of 6 ((1) nm, allowing the binding of only one integrin per particle.9,12 The gold particles are biofunctionalized with the cyclic RGD peptide c(-RGDfK-), allowing specific binding of cellular RVβ3 integrins.13 To prevent any kind of cell-surface interaction in the area between the functionalized gold particles, the surface is covered with a layer of linear poly(ethylene glycol) (PEG) which is covalently linked to the glass surface and known to be an ideal passivation layer. Furthermore, the nanopattern offers the advantage of a well-defined particle density that is proportional to the inverse square of the distance, L-2. For performing cellular unbinding force measurements, we have designed a magnetic tweezers device to exert a vertical force on cells. A magnetic bead, which is covalently attached to the cell membrane, serves as force transducer (see Figure 1). Reflection interference contrast microscopy (RICM), a height-sensitive microscopy technique, is used to observe the detachment of the cell from the surface. Figure 2 shows a sketch of our investigated cell-substrate system as well as scanning electron microscopy images of the nanopattern in the insets A to C. The experiments were performed with rat embryonic fibroblasts (REF 52, wild type) that mainly express RVβ3 integrins and thus recognize the surface-coupled c(-RGDfK-) peptides.
In the experiments, we measured the forces required to lift cells from biofunctionalized nanostructured surfaces with ligand distances ranging from L ) 58 to 145 nm. A single cell was left to adhere for 5 ( 2 min on the substrate. During pulling on the cell, the force was automatically raised stepwise every 5 s by 3 ((1) pN for forces lower than 20 pN and by 7 ((1) pN above 20 pN, up to a maximum force of 200 pN. Figure 3 shows several cell pulling events from surfaces with (i) differently distanced nanopatterns (L ) 58, 69, 93, 120, and 145 nm) functionalized by the c(-RGDfK-) peptides and PEG, (ii) non-c(-RGDfK-) functionalized nanopatterns but with a PEG passivation layer, and (iii) pure PEG. The data are plotted versus the ligand density, L-2. At least five measurements were performed on each surface type. The accuracy of the force measurements is limited by the single force steps. To verify specific integrin binding to the c(-RGDfK-) ligands and to quantify nonspecific cellsubstrate interactions, experiments on non-nanostructured as well as on non-c(-RGDfK-) functionalized nanopatterns were performed as a control. The upper limit of these interaction forces on pure PEG and on nonfunctionalized surfaces has been determined to be 2.5 ( 1 pN, which is negligible (see Figure 3). Therefore, we attribute the measured cellular unbinding forces on functionalized nanopatterns to the specific interactions between the integrins in the cell membrane and the c(-RGDfK-) peptides on the surface. The data in Figure 3 show a sharp increase of the maximum forces needed to pull cells from the surfaces with
Nano Lett., Vol. 6, No. 3, 2006
399
Figure 3. Initial cellular unbinding forces. The measured overall cellular unbinding forces of cells that were left to adhere for 5 ( 2 min on substrates with ligand distances from L ) 58 nm to L ) 145 nm are shown in dependence of ligand density. Data of control experiments on pure PEG and on nanopatterned but not c(-RGDfK-) functionalized PEG coated surfaces (no RGD) are added. The break in the force axis indicates the maximum tweezer force of 200 pN.
a decreasing ligand distance and increasing ligand density. For each ligand distance, the values of cellular unbinding forces vary from small forces to a characteristic maximum. The scattering of the data increases with smaller ligand distance. However, for larger ligand distances L ) 145 and 120 nm the scattering is small. The interactions of integrins and fibronectin, an ECM protein containing RGD sequences, have already been observed by Garcia et al. within the first 15 min of adhesion where cell adhesion strength increased linearly with ligand density.11 Thoumine et al. measured the adhesion probabilities of cells in the first 16 s by applying forces to cells with optical tweezers.14 Both studies used homogeneously coated fibronectin surfaces and thus obtained neither the accuracy nor the low ligand densities of the c(-RGDfK-) functionalized nanostructures. Regarding smaller ligand distances, the scattering of the data shown in Figure 3 is large and forces increase strongly. We were not able to determine the force for the 58 nm nanostructure because it exceeds the maximum tweezer force of 200 pN. For the 69 nm pattern, the data range from a few piconewtons to more than 200 pN. Besides a statistical variation in cell activity, time dependence of the cellular unbinding force is a possible explanation. In the experiments performed, the adhesion time could only be determined with an uncertainty of (2 min because of the different sedimentation times of the cells in the sample holder and the different success rate of grabbing individual cells with the magnetic tweezers. Subsequently, we performed several cellular unbinding experiments with one single cell by repeating the lift-off where the adhesion time between the pulling attempts was determined exactly in real time. Each cell was lifted from the surface, then the force was switched off. After the cell reached the surface again in almost the same area, it was left to adhere for a defined time before lifting it again. This allows the evaluation of adhesion time dependence on ligand 400
Figure 4. Time dependence of cellular unbinding force. One cell was left to adhere for 1, 5, and 10 min on one nanopatterned surface. A strong time dependence can be observed for nanopattern distances of L ) 58 and L ) 69 nm indicating activation of cellular adhesion processes.
density with the same cell. Figure 4 shows single cell measurement data for cells each adhering on substrates with different ligand distances as well as on PEG for adhesion times of 1, 5, and 10 min. As expected, the PEG surface shows no time dependence of the cellular unbinding force since adhesion is prevented. Cells adhering on substrates where c(-RGDfK-) ligands are separated by 120 and 93 nm show cellular unbinding forces around 40 pN but also no dependence on adhesion time. For 69 nm patterned surfaces initial cell unbinding forces increased substantially with time and after 10 min the force was higher than the maximum tweezer force of 200 pN. On the 58 nm pattern, the cell unbinding force was already too high for the magnetic tweezers after 1 min of adhesion. These observations correspond with the scattering of pull-off forces presented in Figure 3. Cells respond to certain ligand distances of the nanopatterns by increasing cellular unbinding forces with time. To point this out, the force data are plotted versus ligand density in Figure 5 for 1, 5, and 10 min of adhesion time. For patterns of 69 and 58 nm, adhesion force strengthens very fast. This fits very well with our earlier observation where cell spreading was restricted for a ligand separation larger than 73 nm while a separation of 58 nm allowed the formation of focal contacts. The observed development of cellular unbinding forces on differently distanced structures during the first 10 min might display the beginning of cooperative responses of the cells and the onset of adhesion associated protein clustering. It clearly demonstrates the dependence of cellular adhesion activity on the offered ligand densities. In summary, our magnetic tweezers setup turns out to be a powerful technique for characterizing initial interactions between cells and biofunctionalized surfaces. We measured specific integrin-mediated cell unbinding forces during the initial stage of adhesion formation on c(-RGDfK-) functionalized nanopatterned surfaces. Our results show an increase in cellular unbinding forces with decreasing ligand distances and a nonlinear response with respect to adhesion time indicating cell activity and even the beginning of cooperativeness in adhesion formation. Nano Lett., Vol. 6, No. 3, 2006
Figure 5. Cellular unbinding force versus ligand density for different adhesion times. A clear nonlinear behavior is observed for an adhesion time of 1 min, which indicates cooperativeness for c(-RGDfK-) separation distances smaller than L ) 69 nm. Furthermore, the additional two curves for 5 and 10 min of adhesion time show an increase of unbinding forces with decreasing ligand density.
In the following paragraphs, the magnetic tweezers device and the microscopy technique RICM as well as the biofunctionalized nanostructures and cell handling are described in detail. Magnetic Tweezers (MT). The MT exert a vertical force on a paramagnetic bead which is coupled to the cell.15 This is done by adding magnetic beads to the cell suspension. The beads carry surface epoxy groups that bind covalently to free amino groups on the cell membrane. Two identical electromagnets facing each other with opposite magnetic poles create a magnetic gradient field. The electromagnets are made of transformer cores with a size of 100 × 27 × 27 mm and machined pole tips and ca. 700 windings of 0.8 mm copper wire. A crucial parameter for achieving high forces is the distance between the pole tips, which is variable in our setup as the electromagnets are mounted on micrometer screws. Small distances lead to higher fields and field gradients and thus to a higher force. Accordingly, the sample holder and the pole tip shape were designed to meet these requirements and still allow bright field microscopy with a small working distance objective (63×/1.25 Antiflex-Neofluar, oil-immersion, Zeiss, Germany) (see Figure 6A). At the sample position, magnetic fields of 380 mT and gradient fields of 90 T/m can be reached at a minimal tip-to-tip distance of 2.5 mm. No saturation effect of the cores can be observed within the used current range. Paramagnetic Dynabeads (M450 Epoxy, Dynal Biotech, Norway) with a mean diameter of 4.5 µm are used as force transducers. The force calibration was performed using Stokes drag formula, Ffr ) 6πηrV, where Ffr describes the friction force of a bead with radius r moving with a velocity V in a viscous fluid of viscosity η. In the stationary state of constant bead velocity, the friction force equals the force exerted on the bead. Calibration was performed in glycerol (99.5% from GEBRU, Germany) at both 30 and 20 °C to cover different force ranges. By observing the vertical movement of a bead in bright field and comparing the diffraction pattern to reference Nano Lett., Vol. 6, No. 3, 2006
Figure 6. Magnetic tweezers: (A) Calibration. Stokes law is used for the force calibration. To measure the velocity of the bead, its diffraction pattern is recorded during force application. (B) The images show one bead in different focal planes, each separated by 2 µm. (C) Calibration curves. The force-current relation is quadratic for small currents due to the unsaturated paramagnetic beads. For higher currents, there is a linear dependence.
images at different focal planes (see Figure 6B), the beads can be tracked.16 For each calibration measurement, reference images of the bead were taken at different focal planes by changing the objective height manually in 1 µm steps. The RICM focus is used as a reference point in the vertical direction. The analysis is performed with a custom written IDL program (IDL Research Systems). Calibration curves are shown in Figure 6C. At least five measurements per data point are considered. The nonlinearity in the current-force curve for currents lower than 2 A is due to the unsaturated paramagnetic beads; for higher currents there is a linear dependence. To avoid an overheating of the electromagnet coils, the current is restricted and thus the force. The maximum possible tweezers force is 200 pN. The force is constant over a vertical distance of at least 30 µm from the microscope slide and laterally over the observed sample area of about 80 µm × 60 µm. A Lab View program controls the applied force and triggers the recording of images during the lift-off with a high-resolution camera (Digital CCD Camera ORCA-ER, Hamamatsu Photonics, Germany). For evaluating the experiments, the time of cell detachment is correlated with the exerted force. 401
Biofunctionalized Nanostructures. Nanopatterned glass slides are prepared via self-assembly of diblock copolymer micelles in a monomicellar layer during dip coating. A cluster of gold ions is embedded in the core of these micelles. Subsequent hydrogen plasma treatment deposits the gold nanoparticles as elementary gold on the substrate and removes the surrounding polymer matrix. This results in a hexagonally closed packed pattern of gold nanoparticles where the nanoparticle distance is controlled by the size of the polymer micelles.12 The area between the gold particles is passivated with a thin layer of a linear poly(ethylene glycol) by immersing the hydrogen plasma activated glass interfaces (Plasma etcher (100-E) from TePla; 5 min, 0.4 mbar H2, 150 W) in a 1 mmol solution of the linear poly(ethylene glycol) mPEG2000-urea (CH3-O-(CH2-CH2O)43-NH-CdO-NH-CH2-CH2-CH2-Si(OEt)3) and dry toluene (p.A. from Merck) with 0,05% triethylamine at 80 °C under nitrogen atmosphere for at least 16 h. Finally, the substrates are rinsed extensively with ethyl acetate and methanol (all p.a. from Aldrich). The thickness of the PEG layer was approximately 2 nm in a dry state as determined by X-ray photoelecton microscopy measurements.17 Theoretical consideration of brush formation in water estimates the PEG layer thickness to be approximately 5 nm in a hydrated state. The passivated substrates are biofunctionalized by incubation with a 25 µmol c(-RGDfK-)-thiol ()c[RGDfK(AhxMpa)])13,18-water solution for at least 4 h. Subsequently, the samples are rinsed with MilliQ(R > 18 MΩ) and shaken for 24 h with several water exchanges to remove the noncovalently bound c(-RGDfK-)-thiols. Rat Embryonic Fibroblasts. The fibroblasts (REF52, wild type, passage 5 to 25) were cultured in DMEM (Invitrogen, Germany) supplemented with 10% FBS (fetal bovine serum) and 2% L-glutamine (Invitrogen) at 37 °C and 10% CO2. Prior to the experiments, the magnetic beads are covalently attached to the cells. Cells were trypsinized (0.25% trypsin, 1 mM EDTA) and suspended in PBS (phosphate buffered saline). After 2 min the PBS is replaced by DMEM which is supplemented with 1% FBS. This cell suspension is used within 1 h to avoid endocytosis of the magnetic beads.19 MT experiments are performed at 37 °C and 5% CO2. Reflection Interference Contrast Microscopy. RICM is an interferometric technique that visualizes the contour of objects situated within a relatively small range from 2 nm to 1 µm above the microscope slide. Due to its limited imaging depth, it is used as a marker for obtaining a precise
402
time stamp for the detachment of the cell during the experiment (see Figure 1A-C). The principle of the technique is based on the interference of reflected light from the cell and the top interface of the glass slide. The observed interference pattern encodes information on the height difference between the cell contour and the substrate.20 Acknowledgment. We thank Dr. Benjamin Geiger (Weizmann Institute of Science) for his continuous support of this project. The German-Israeli Foundation (G.I.F.) and the Max-Planck-Society are acknowledged for financial support. The work has also been part of a STREP Program of the European Community (Nanocues). C.S. was supported by the Boehringer Ingelheim Fonds. References (1) Zamir, E.; Geiger, B. J. Cell Sci. 2001, 114, 3583-3590. (2) Zaidel-Bar, R.; Cohen, M.; Addadi, L.; Geiger, B. Biochem. Soc. Trans. 2004, 32, 416-420. (3) Cohen, M.; Klein, E.; Geiger, B.; Addadi, L. Biophys. J. 2003, 85, 1996-2005. (4) Kool, L. Y.; Irvine, D. J.; Mayes, A. M.; Lauffenburger, D. A.; Griffith, L. G. J. Cell Sci. 2002, 115, 1423-1433. (5) Maheshwari, G.; Brown, G.; Lauffenburger, D. A.; Wells, A.; Griffith, L. G J. Cell Sci. 2000, 113, 1677-1686. (6) Chen, C. M.; Mkrsich, M.; Huang, S.; Whitesides, G. M.; Ingber, D. E. Science 1997, 276, 1425-1428. (7) Jungbauer, S.; Kemkemer, R.; Gruler, H.; Kaufmann, D.; Spatz, J. ChemPhysChem 2004, 5, 85-92. (8) Zimmerman, E.; Geiger, B.; Addadi, L. Biophys. J. 2002, 82, 18481857. (9) Arnold, M.; Calvacanti-Adam, E. A.; Glass, R.; Blu¨mmel, J.; Eck, W.; Kantlehner, M.; Kessler, H.; Spatz, J. P. ChemPhysChem 2004, 5, 383-388. (10) Benoit, M.; Gabriel, D.; Gerisch, G.; Gaub, H. E. Nat. Cell Biol. 2000, 2, 313-317. (11) Garcia, A. J.; Boettiger, D. Biomaterials 1999, 20, 2427-2433. (12) Glass, R.; Mo¨ller, M.; Spatz, J. P. Nanotechnology 2003, 14, 11531160. (13) Haubner, R.; Finsinger, D.; Kessler, H. Angew. Chem. 1997, 109, 545-556. (14) Thoumine, O.; Kocian, P.; Kottelat, A.; Meister, J.-J. Eur. Biophys. J. 2000, 29, 398-408. (15) Huang, H.; Dong, C. Y.; Kwon, H. S.; Sutin, J.-D., Kamm, R. D.; So, P. T. C. Biophys. J. 2002, 82, 2211-2223. (16) Gosse, C.; Croquette, V. Biophys. J. 2002, 82, 3314-3329. (17) Blu¨mmel, Jacques. Entwicklung biofunktionalisierter Nanostrukturen an Grenzfla¨chen zur Untersuchung der Kinetik des molekularen Motorproteins Eg5. Ph.D. Thesis, University of Heidelberg, 2005. (18) Pfaff, M.; Tangemann, K.; Mu¨ller, B.; Gurrath, M.; Mu¨ller, G.; Kessler, H.; Timpl, R.; Engel. J. J. Biol. Chem. 1994, 269, 2022320238. (19) McKeown, M.; Knowles, G.; McCulloch, C. A. C. Cell Tissue Res. 1990, 262, 523-530. (20) Ra¨dler, J.; Sackmann, E. Langmuir 1992, 8, 848-853.
NL052168U
Nano Lett., Vol. 6, No. 3, 2006
