Surface-induced polymerization of actin

ABSTRACT Living cells contain a very large amount of membrane surface area, which potentially influences the direction, the kinetics, and the localization of biochemical reactions. This paper quantitatively evaluates the possibility that a lipid monolayer can adsorb actin from a nonpolymerizing solution, induce its polymerization, and form a 2D network of individual actin filaments, in conditions that forbid bulk polymerization. G- and F-actin solutions were studied beneath saturated Langmuir monolayers containing phosphatidylcholine (PC, neutral) and stearylamine (SA, a positively charged surfactant) at PC:SA = 3:1 molar ratio. Ellipsometry, tensiometry, shear elastic measurements, electron microscopy, and dark-field light microscopy were used to characterize the adsorption kinetics and the interfacial polymerization of actin. In all cases studied, actin follows a monoexponential reaction-limited adsorption with similar time constants (~10^sup 3^ s). At a longer time scale the shear elasticity of the monomeric actin adsorbate increases only in the presence of lipids, to a 2D shear elastic modulus of mu == 30 mN/m, indicating the formation of a structure coupled to the monolayer. Electron microscopy shows the formation of a 2D network of actin filaments at the PC:SA surface, and several arguments strongly suggest that this network is indeed causing the observed elasticity. Adsorption of F-actin to PC:SA leads more quickly to a silghtly more rigid interface with a modulus of mu == 50 mN/m.

INTRODUCTION

Actin is the major constituent of muscle cells, but it is also expressed in the cytoplasm of all eukaryotic cells, where it plays a central role in numerous cellular functions such as cell motility, cytokinesis, and phagocytosis (Pollard, 1990; Kabsch and Vandekerckhove, 1992). Since actin purification procedures have been available, actin has been studied for its biochemical and physical properties. In vivo, monomeric actin, also called G-actin, can reversibly polymerize into microfilaments called F-actin, under the control of several intricate and distinct regulatory pathways that have been studied extensively (Carlier, 1991; Sheterline and Sparrow, 1994). In vitro, G-actin is classically induced to polymerize by salts such as KCl and MgC12 or polyamines (Oriol-Audit, 1978; Grant and Oriol-Audit, 1983), and the resulting F-actin solutions have been investigated by numerous groups for their macroscopic (Janmey, 1991; Maggs, 1997) and microscopic viscoelastic properties (Ziemann et al., 1994; Amblard et al., 1996; Schnurr et al., 1997).

Motivated by the elucidation of its structure, efforts have been made to crystallize the actin monomer, and various strategies have been used. The crystal structure of the monomer is not yet available, although cocrystals between F-actin and different actin-binding proteins have been made and resolved successfully, such as those with profilin, DNAse I, and gelsolin (Mannherz, 1992, and Pollard et al., 1994, for reviews). On the one hand, in 3D, under well-defined ionic conditions with polycationic molecules or multivalent cations, pure actin assembles into microfilaments that form superstructures such as elongated microcrystals, tubules, or stacks of parallel filament sheets (see Taylor and Taylor, 1992, for a review). On the other hand, solutions of filaments were shown to adsorb to positively charged 2D substrates by electrostatic interaction and assemble into flat paracrystalline filament arrays, from which low-resolution structural information was obtained (Rioux and Gicquaud, 1985; Ward et al., 1990; Taylor and Taylor, 1992, and references therein). Because actin has a low pI (~5.5), its electrostatic adsorption can be obtained at neutral pH on polyamine surfaces, mixtures of neutral and basic lipids (Rioux and Gicquaud, 1985; Laliberte and Gicquaud, 1988; Ward et al., 1990), or with basic surfactants (Taylor and Taylor, 1992). In most studies, the buffer conditions have been such that bulk polymerization in the solution generally precedes the surface adsorption of polymers and paracrystal formation, and not much attention has been paid to the possible scenario of surface-induced polymerization from a nonpolymerizing monomer solution. One report shows surface-induced polymerization by positively charged liposomes, leading to paracrystal formation, but single filaments were not observed (Laliberte and Gicquaud, 1988).

Because the organization of biological membranes is a central question in cell biology, several studies have been devoted to the lateral behavior of membrane proteins and to lipid/lipid and lipid/protein interactions (Jacobson et al., 1995). In this context, the 2D crystallization of soluble proteins by means of a lipid monolayer has been studied by powerful structural methods such as grazing incidence x-ray diffraction (Haas et al., 1995; Lenne, 1998), and electron cryomicroscopy (Henderson et al., 1990). In addition, the measurement of the surface viscoelasticity has been recently combined with ellipsometry (Venien-Bryan et al., 1998). Nevertheless, most proteins studied so far in 2D do not assemble into linear polymers, but rather into 2D arrays, and the possibility of polymerization from a solution of 2D confined monomers has not been investigated.

The aim of the present work is to investigate the process by which a positively charged lipid monolayer deposited at the air/buffer interface could serve as a template for the polymerization of monomeric actin into single filaments. We wish to establish a reproducible set of qualitative and quantitative observations that demonstrates surface-induced polymerization of actin and describes some of its kinetic, mechanical, and ultrastructural aspects. Following initial work (Renault et al., 1997), the adsorption kinetics and the process of surface polymerization are studied by ellipsometry, tensiometry, surface rheology, and dark-field light microscopy. The ultrastructure of single 2D-formed polymers is approached by electron microscopy. The role of the surface lipids and the dimensionality of the polymerization process are evaluated by contrasting experimental conditions: G- or F-actin at a bare air/water interface or under a lipid monolayer.

MATERIALS AND METHODS

Lipid and actin biochemistry

Egg phosphatidylcholine (PC, Sigma catalog no. P-5763) and stearylamine (SA, Sigma catalog no. S-9273) were kept in hexane/ethanol solution (9:1) at -20 deg C. The working solutions were prepared in chloroform or in hexane/ethanol at a final lipid concentration of 6 X 10^sup -4^ M. Positively charged amphiphiles (SA) were mixed with PC as the neutral lipid, at a final molar ratio of PC:SA of 3:1. Following the classical method of Pardee and Spudich (1982), actin acetone powder was prepared from chicken breast muscles and stored at -80 deg C. Actin was then extracted through two or three cycles of polymerization-depolymerization. High-salt washes were performed with 0.65 M KCl for 30 min, and filaments were depolymerized by rapid overnight dialysis in G-buffer (5 mM Tris-HCI, pH 7.4, 0.2 mM Na-ATP, 0.5 mM beta-mercaptoethanol, 0.2 mM CaCI^sub 2^, and 0.01% NaN^sub 3^), followed by ultracentrifugation. The purity was assessed by electrophoresis, using Coomassie staining-overloaded polyacrylamide gels, and by MALDI-TOF mass spectrometry (Perseptive, Framingham, MA). Highly purified G-actin solution was stored at -80 deg C. Samples were prepared by diluting concentrated actin in nonpolymezing buffer (G-buffer) or polymerizing buffer (F-buffer), that is, G-buffer supplemented with 100 mM KCI, 2 mM MgC1^sub 2^, and 0.5 mM Na-ATP.

Ellipsometry, surface tension

The ellipsometric measurements were carried out with a conventional null ellipsometer using a He-Ne laser operating at 632.8 nm (Berge and Renault, 1993). The variation of the ellipsometric angle is a relevant probe for changes occurring at the interface. Ellipsometric angle (delta) and surface pressure were recorded simultaneously. The surface pressure was measured with a Wilhelmy balance. The sample cell is made from Teflon and has a volume of 8 ml. The protein is injected into the subphase, and the buffer is then coated by the lipid monolayer. All the experiments were carried out at room temperature. Initial time points of all graphs (t = 0) correspond to the first possible measurements, once the surface is stable, i.e., a few minutes after mixing.

Shear elastic constant

The rheometer set-up uses the action of a very light float (32 mg), which applies a rotational strain to the monolayer through a magnetic torque (with a pair of Helmholtz coils and a small magnetized pin deposited in the float). This set-up and the procedure for data analysis have been described previously (Venien-Bryan et al., 1998; Zakri-Delplanque, 1997). Briefly, at the center of a 48-mm-diameter Teflon trough, a 10-mm-diameter paraffincoated aluminum disc floats at the air/water interface, surrounded by the monolayer, whose rigidity is measured. The subphase is 5 mm deep. The float carries a small magnet and is kept centered by a permanent field, B^sub o^ = 6 x 10^sup -5^ T, parallel to the Earth's field and created by a little solenoid located just above the float. Sensitive angular detection of the float rotation is achieved by using a mirror fixed on the magnet to reflect a laser beam onto a differential photodiode. A sinusoidal torque excitation is applied to the float in the 0.01-100 Hz frequency range, by an oscillating field perpendicular to the permanent solenoid field. The latter field acts as a restoring torque equivalent to a monolayer with a rigidity of 0.16 mN/m. This number set the sensitivity limit of this rheometer. The device behaves like a simple harmonic oscillator. The angular response is measured in amplitude and phase and is considered to reflect directly the rotational strain of the monolayer (see the Discussion). The data presented here only include the values of the shear elastic constant, mu (mN/m), measured at 5 Hz. Initial time points of all graphs (t = 0) correspond to the first possible measurements, once the magnetic float is centered and stable, i.e., a few minutes after mixing.

Electron microscopy

At the end of the experiments, i.e., after 20 h, plain carbon-coated electron microscope grids were placed on top of the crystallization trough, withdrawn after a few minutes of adsorption, and negatively stained with 2% (w/v) uranyl acetate for 30 s. Negatively stained specimens were examined in a Philips CM200 electron microscope operating at 200 kV. Micrographs were recorded on Kodak SO 163 film under low dose conditions and at a nominal magnification of 27,500X. For some experiments, the subphase was mildly stirred by a slowly rotating magnetic bar lying on the bottom of the trough, with dimensions and an angular velocity such that it does not disrupt the surface.

Dark-field light microscopy

The air/water interface was imaged with an inverted microscope (IX; Olympus, Japan) equipped with a 100-W halogen lamp and commercially available dry optics: a dark-field condenser (U-DCD) and a 60x UPIFL objective. A ULL760 intensified CCD (Lhesa, Cergy Pontoise, France) was used for video-rate imaging of actin filaments at the interface.

RESULTS

G-Actin at the air/buffer interface

Electron microscopy observations

In nonpolymerizing conditions, i.e., without KCl and MgC1^sub 2^, electron micrographs of the bare air/buffer interface after a 20-h incubation show no linear structures that would be indicative of polymerization. In contrast, a dense and regular distribution of dots of similar size appears, which probably correspond to oligomeric aggregates of denatured monomers (not shown).

In the presence of a PC:SA monolayer but in otherwise identical conditions, the surface was sampled in two different sets of conditions: at mechanical rest, or with a mild stirring of the subphase (see Materials and Methods). The motivation of introducing a mild stirring of the subphase was to perturb the conditions of polymerization, by a hydrodynamic flow. This perturbation will be useful in illustrating and analyzing the differences between the present results and those previously obtained in closely related work (Laliberte and Gicquaud, 1988). Under "stirring" conditions, linear structures clearly appear on the surface, consisting of straight filaments (Fig. 5 a), with lengths typically from I to 3.5 mu m. The thickness of each filament appears to be constant at 6-7 nm, in agreement with the known ultrastructure of F-actin. These filaments are arranged in loose parallel patterns (made up of 5-20 filaments), which clearly contrast with the strong order and close spacing of paracrystalline microfilament sheets made from F-actin solutions (Laliberte and Gicquaud, 1988; Ward et al., 1990; Taylor and Taylor, 1992). Here the word "parallel" does not imply the same polarity of the filaments. In these stirring conditions, isolated filaments are barely seen. Fig. 5 b shows the dramatic effect of the lack of stirring of the subphase on actin filament structure, typically observed on samples transferred onto electron microscopic grids after torsion or ellipsometry measurements. The parallel long and straight filaments are no longer present; only single filaments are seen, and these form a continuous 2D network. Another striking feature is the stuctural defects shown by these isolated filaments, such as thickness irregularities and branching points (Fig. 5 b). The lengths of these filaments are also shorter: from 0.5 to 1 jam. Nevertheless, these seemingly "abnormal" structures are essentially ID objects, with an electron density contrast similar to that of straight filaments.

DISCUSSION

The present results show that a nonpolymerizing G-actin solution can be induced to assemble into individual filaments on the surface of a positively charged lipid monolayer. Following initial work (Renault et al., 1997), in this report the combination of ellipsometry, tensiometry, surface rheology, transmission electron microscopy, and dark-field light microscopy is used to investigate this phenomenon and to describe qualitatively and quantitatively some of its kinetic, mechanical, and ultrastructural aspects. Different experimental conditions (G- or F-actin solutions at a bare air/water interface or under a lipid monolayer) are used to evaluate the role of the surface lipids and the dimensionality of the polymerization process, and to kinetically resolve adsorption from rigidification/polymerization.

Purity and reproducibility

The protocols followed for actin purification and manipulation were optimized for an optimal reproducibility of the results: three polymerization/depolymerization cycles with a rapid overnight depolymerization at each step, storage at -80 deg C, reproducible thawing procedure, and only sameday use of thawed solutions. The solutions used typically had a total concentration of contaminants (larger than 5 kDa) on the order of 0.1-0.2%, as determined by polyacrylamide gel electrophoresis (Fig. 6). In addition, the molecular mass of actin determined by mass spectrometry was within ~50 Da of the expected mass with no apparent proteolysis. In our experience, the usual one-cycle purification procedure gave an insufficient purity (~1% contaminant), which gave fluctuating results in surface rheology experiments. The precautions described above are of critical importance and gave us a satisfactory reproducibility, even between different purification batches. Ellipsometry results were obtained four to five times, giving very similar results in terms of the angle Or"maX and the time constant T; indeed, the deviation of the fitted quantities between the different experiments was less than the fitting error (see values in Table 1). Shear elastic measurements were made two or three times for each set of conditions, and the kinetic curves gave similar results (see Table 1).

Adsorption kinetics

The ellipsometric angle A depends on the refractive index profile perpendicular to the air/water interface. It is analyzed here in a semiquantitative way, by considering it as the sum of two terms: a fixed offset representing the lipid monolayer, and the angle contributed by the proteins in molecular contact with the lipid surface, which reflects the average surface density and thickness of the adsorbed actin layer.

Let us first consider G-actin at the bare air/liquid interface or under the PC:SA monolayer. The short-time monoexponential behavior of the ellipsometric angle A (Figs. 1 and 3 a) can be accounted for by an effective first-order adsorption process sketched by

where [A] is the volume concentration of actin, [B] is its surface concentration, and k+ is the on rate, with units of length over time. The equilibrium value [B]/[A] reflects the surface affinity and has units of length. The experimentally observed proportionality between the on rate and the actin concentration (data not shown) is in agreement with this model, as well as the existence of a long time plateau in the ellipsometric angle. One may then ask whether the G-actin adsorption is diffusion limited or not. A simple calculation based on the actin volume concentration, the monomer size (~5 nm), and the diffusion coefficient of G-actin (~50 mu m^sup 2^ s^sup -1^) tells us that the adsorption occurs on a time scale that is within the reaction-limited regime by at least two orders of magnitude. The reaction-limited monoexponential behaviors described for G-actin are also found with F-actin with or without lipids (Figs. 2 a and 4 a), with slightly larger exponential time constants. Reaction limitation of protein adsorption onto lipids, as opposed to diffusion limitation, has also been observed in other experimental models, such as spectrin or protein 4.1 (Kiernan et al., 1997). It is noteworthy that the presence of lipids slightly reduces the adsorption on rate for both G- and F-actin and leads to a denser and/or thicker surface coverage (Table 1).

After the exponential adsorption phase, in the absence of lipids, the long-time behavior of the ellipsometric signal for both G- and F-actin fits very well with a slow adsorption process with a constant rate that does not saturate (Figs. 1 and 2 a). This process is most likely caused by the irreversible denaturation of actin in contact with air. Interestingly, surface denaturation is much slower (~30 times) with Fthan G-actin. In the case of F-actin, the shear rigidity of the interface is not affected in a detectable way by these processes.

The long-time behavior of the G- and F-actin solutions under the PC:SA monolayer are quite different from those in the absence of lipids. Indeed, F-actin under lipids results in an equilibrium situation, with a flat plateau for both the ellipsometric and the rigidity signals (Fig. 4). This suggests that actin filaments are protected from denaturation and build a mechanically stable surface. In contrast to this situation, the lipid interface in contact with a G-actin solution is not stable, but shows a distinctive hyperbolic decrease of the ellipsometric angle. Taken together with the decrease in surface tension, this observation strongly suggests that actin could intercalate into the monolayer, as has been observed in other experimental protein/lipid models (Ellison and Castellino, 1997, and references therein). The hyperbolic variation of the optical signal suggests that the underlying process is a second-order reaction in contrast with the initial exponential adsorption. Moreover, this second-order process occurs on a time scale that is very similar to that of the elasticity increase, thus suggesting a relationship between them.

Surface shear elasticity

Our approach to measuring surface shear elasticity involves a device that applies very small excitation strains (from 10^sup -3^ down to 10^sup -6^). In this range, previous experiments have shown that pure shear elastic response spectra are obtained with 2D protein crystals, and there is a linear stress-strain relationship over the whole range (VenienBryan et al., 1998). This demonstrates that 1) the rotation coupling between the float and the contacting monolayer covered with its underlying structures is satisfactory, and 2) such small strains do not create plastic deformations on fragile surface objects. We currently used strains in the 10-4 range, which is very likely to be inside the linear elastic response domain of surface actin networks. It is noteworthy that surface rheology experiments already reported employed much higher strains in the 0.01-0.05 range (Muller et al., 1991), and that strongly limits the meaning of quantitative comparisons with our work, as explained later.

The following discussion of elasticity focuses on the behavior of G-actin solutions at the PC:SA interface (Fig. 3), and all quantities refer to this experimental situation unless otherwise stated. As previously mentioned, elasticity cannot be measured while the subphase is stirred, so the present discussion only applies to structures obtained in unstirred conditions. Frequency response spectra (data not shown) acquired at different times during the experiments did not show a viscous component, but only an increase in the shear elastic modulus, which was then simply assessed at a fixed rotation frequency of 5 Hz.

Two arguments support the notion that the elasticity increase is entirely due to interfacial structures resulting from the actin-lipid interaction. First, ultrasensitive darkfield video microscopy experiments aimed at visualizing individual filaments over long times revealed filaments "sticking" to the interface without detaching (Fig. 7). In these experiments, no filaments were seen in the bulk solution. Second, the asymptotic value of the elastic modulus is roughly one order of magnitude larger than that for the F-actin solution in the absence of lipids, and the PC:SA monolayer per se has no detectable elasticity-less than 0.1 mN/m. Altogether, these facts suggest that a lipid-actin structure present at the surface is responsible for the observed elasticity.

We must then discuss the nature of these surface structures that increase the surface shear elasticity. Several arguments strongly suggest that actin filaments observed by electron microscopy in unstirred conditions are indeed causing the elasticity increase. In the first place, the analysis of numerous micrographs indicates that filaments form a 2D network that covers most of the surface in a continuous way, with a typical interfilament distance (2D mesh size) of a micron. Is such a network density high enough to account for the elasticity observed at long times? A direct quantitative answer to this question is rather difficult, because the elastic properties of the surface bound polymers seen by electron microscopy are not known. Nevertheless, the 2D mesh size (1 mu m) is close to the 3D mesh size of the 25 jig/ml solution when polymerized (2 /Mm) (Schmidt et al., 1989). In addition, the vertical size of the strained region (5 mm) is four times smaller than its horizontal extension (20 mm). This indicates that the surface elastic modulus of the 2D network (30 mN/m), when corrected for the geometry, is similar to that of the 3D actin solution. This order of magnitude comparison suggests that the surface elasticity of the surface polymerized actin "skin" can indeed be accounted for by the microfilament network seen on the surface. A rigorous comparison of these numbers would require a careful theoretical analysis of the geometry of the strain field, which is simple in 2D but much more complex in the 3D medium. In addition, the actin we used is very pure (0.1-0.2% contaminants in mass), and we did not see any evidence of 2D crystalline arrays of actin by electron microscopy. This is very different from streptavidin crystals formed under a biotinylated lipid monolayer. Indeed, despite the fact that 2D crystals are clearly visible by electron microscopy, they cause an order of magnitude smaller surface shear elasticity, as measured by the same apparatus with an identical geometry (3 mN/m) (Venien-Bryan et al., 1998). This makes it very unlikely that invisible surface structures contribute to the elasticity. Altogether, these results strongly suggest that the 2D filament network formed at the PC:SA interface over a nonpolymerizing actin solution is indeed the essential cause for the observed increase in surface shear elasticity.

The rheology of actin solution has been extensively investigated in bulk (Janmey, 1991; Maller et al., 1991; Wachsstock et al., 1994; Isambert and Maggs, 1996) and at a microscopic scale by the use of microbeads (Ziemann et al., 1994; Amblard et al., 1996; Schnurr et al., 1997; Maggs, 1998). Unfortunately, unlike physical polymer systems, experimental results with actin display a very strong variability. For instance, plateau moduli measured in bulk by different groups are scattered over many orders of magnitude (Maggs, 1997). This comes from some intrinsic features of the actin biochemistry, such as the polydispersity of filament length, from the purity of actin preparations used, and, importantly, from the technical details of rheology experiments. The surface rheology approach described here bears some apparent similarity to an oscillating disk rheometer previously used for measuring the 3D shear elastic modulus of an entangled actin solution (Miller et al., 1991). A typical plateau modulus of 0.1 N/m2 was found for a 0.1 mg/ml actin solution. Can one make quantitative comparisons between that work and ours? Despite apparent similarities between the two instruments, the set-up of Miller et al. has several important features that make it different from ours: 1) The coupling of the rotating float with the solution is made by a neutral lipid monolayer of dimyristoylphosphatidylcholine, which is known not to interact with actin (Bouchard et al., 1998). 2) The strain used (-0.OS) is roughly three orders of magnitude above our range, and this might affect the linearity of the response. 3) The strain geometries are different in the two instruments, and in particular, the strain in the meniscus region is poorly controlled and probably very high in Muller's work and inadequate for measuring 2D shear elasticity. For all of these reasons, it is not meaningful to look for the length scale that is required for comparing surface and bulk elasticity in the two instruments. Nevertheless, these remarks do not invalidate the above comparison made between our measurements of the elasticity of the 2D and 3D networks, because the same instrument was used.

The kinetics of rigidification follow an exponential behavior, from which the nature of the polymerization kinetics cannot be inferred, because the mechanical effect measured here probably has no simple relationship with the amount of polymer formed at the interface. The exponential elasticity increase is preceded by a 5000-s lag phase, which is longer than the 1800-s adsorption time constant (Fig. 3). This delay between the adsorption and the rigidification has also been observed in the process of 2D crystallization of cholera toxin B, where it was related to the percolation of 2D crystals (Venien-Bryan et al., 1998). A similar feature was also observed in kinetic analysis of actin polymerization in a suspension of positively charged liposomes in the absence of Mg2+ and K+ ions (Laliberte and Gicquaud, 1988). Here either the surface-induced polymerization has a lag like that of solution polymerization, and/or the elasticity requires the surface concentration of polymer to rise above a critical threshold. The latter interpretation fits well with the general notion that polymer solutions manifest elastic properties only if their concentration is above a threshold where entanglement appears (De Gennes, 1978).

Assembly and microscopic structure of surfaceinduced actin polymers

CONCLUSION

The present data demonstrate that a nonpolymerizing Gactin solution can be induced to polymerize into single filaments by and at a positively charged lipid monolayer. The mechanism first involves a monoexponential adsorption process, followed by a marked increase in the surface shear elastic modulus. Actin polymerization essentially occurs at the interface and produces a 2D network of actin filaments, which is probably responsible for the observed elasticity. These observations point toward open questions concerning the detailed molecular mechanism of 2D nucleation and polymerization and its relationship with 3D polymerization. The optical, mechanical, and microscopic imaging methods combined in the present work, together with other tools, should be useful for further studies of proteinlipid interactions and 2D molecular assembly.

[Reference]

REFERENCES

[Reference]

Amblard, F., A. C. Maggs, B. Yurke, A. N. Pargellis, and S. Leibler. 1996. Subdiffusion and anomalous viscoelasticity in actin networks. Phys. Rev. Lett. 77:4470-4473.

Berge, B., and A. Renault. 1993. Ellipsometry study of 2D crystallization of 1-alcohol monolayers at the water surface. Europhys. Lett. 21: 773-777.

[Reference]

Bouchard, M., C. Pare, J.-P. Dutasta, J.-P. Chauvet, C. Gicquaud, and M. Auger. 1998. Interaction between G-actin and various types of liposomes: a 19F, 31p, and 2H NMR study. Biochemistry. 37:3149-3155. Bretscher, A., D. Reczek, and M. Berryman. 1997. Ezrin: a protein requiring conformational activation to link microfilaments to the plasma membrane in the assembly of cell surface structures. J. Cell Sci. 110: 3011-3018.

Cao, L.-G., D. J. Fishkind, and Y.-L. Wang. 1993. Localization and dynamics of nonfilamentous actin in cultured cells. J. Cell Biol. 123: 173-181.

[Reference]

Carlier, M.-F. 1991. Actin: protein structure and filament dynamics. J. Biol. Chem. 266:1-4.

De Gennes, P. G. 1978. Scaling Concept in Polymer Physics. Cornell

University Press, Ithaca, NY.

Ellison, E. H., and F. J. Castellino. 1997. Adsorption of bovine prothrombin to spread phospholipid monolayers. Biophys. J. 72:2605-2615. Grant, N. J., and C. Oriol-Audit. 1983. Supramolecular forms of actin induced by polymamines: an electron microscopy study. Eur. J. Cell Biol. 30:67-73.

Haas, H., G. Brezinski, and H. Mohwald. 1995. X-ray diffraction of a protein crystal anchored at the air/water interface. Biophys. J. 68: 312-314.

[Reference]

Han, X., G. Li, G. Li, and K. Lin. 1997. Interactions between smooth muscle a-actinin and lipid bilayers. Biochemistry. 36:10364-10371. Harwell, O. D., M. Sweeney, and F. Kirkpatrick. 1980. Conformation changes of actin during formation of filaments and paracrystals and upon interaction with DNAse I, cytochalasin B and phalloidin. J. Biol. Chem. 255:1210-1220.

Henderson, R., J. Baldwin, T. Ceska, F. Zemlin, E. Beckmann, and K. Downing. 1990. Model for a structure of bacteriorhodopsin based on hig resolution cryo-electron microscopy. J. MoL Biol. 213:899-929. Isambert, H., and A. C. Maggs. 1996. Dynamics and rheology of actin

solutions. Macromolecules. 29:1036-1040.

Jacobson, K., E. D. Sheets, and R. Simson. 1995. Revisiting the fluid mosaic model of membranes. Science. 268:1441-1442. Janmey, P. A. 1991. Physical properties of actin. Curr. Opin. Cell Biol. 2:11-17.

[Reference]

Kabsch, W., and J. Vandekerckhove. 1992. Structure and function of actin. Annu. Rev. Biophys. Biomol. Struct. 21:49-76. Kiernan, A. E., R. I. McDonald, R. C. McDonald, and D. Axelrod. 1997. Cytoskeletal protein binding kinetics at planar phospholipid membranes. Biophys. J. 73:1987-1998.

Lenne, P.-F. 1998. Deux exemples de cristallisation a la surface de l'eau: cristaux 2D de proteines, et monocouches d' alcools en presence d'amphiphiles solubles. Ph.D. thesis. Universite Joseph Fourier, Grenoble, France. 150 pp.

Laliberte, A., and C. Gicquaud. 1988. Polymerization of actin by positively

charged liposomes. J. Cell Biol. 106:1221-1227. Luna, E. J., and A. L. Hitt. 1992. Cytoskeleton-plasma membrane interactions. Science. 258:955-964.

[Reference]

Maggs, A. C. 1997. Two plateau moduli actin gels. Phys. Rev. E. 55: 7396-7400.

[Reference]

Maggs, A. C. 1998. On micro-bead mechanics with actin filaments. Phys. Rev. E. 57:2091-2094.

Mannherz, H. G. 1992. Crystallization of actin in complex with actin binding proteins. J. BioL Chem. 267:11661-1164. Mller, O., H. E. Gaub, M. Barmann, and E. Sackmann. 1991. Viscoelastic moduli of sterically and chemically cross-linked actin networks in the dilute to semi-dilute regime: measurement by an oscillating disk rheometer. Macromolecules. 24:3111-3120.

Naumann, C., C. Dietrich, A. Behrisch, T. Bayerl, M. Schleicher, D. Bucknall, and E. Sackmann. 1996. Hisactophilin-mediated binding of actin to lipid lamellae: a neutron reflectivity study of protein membrane coupling. Biophys. J. 71:811-823.

Oriol-Audit, C.1978. Polyamine induced actin polymerization. Eur. J. Biochem. 87:371-376.

[Reference]

Pardee, J. D., and J. A. Spudich. 1982. Purification of muscle actin. Methods Enzymol. 85:164-181.

Pollard, T. D. 1990. Actin. Curr. Opin. Cell Biol. 2:33-40. Pollard, T. D., S. Almo, S. Quirk, V. Vinson, and E. E. Lattman. 1994. Structure of actin-binding protein: insight about function at atomic resolution. Annu. Rev. Cell Biol. 10:207-249. Renault, A., F. Amblard, C. Gicquaud, and B. Berge. 1997. Actin adsorbed under phospholipid layers. Eur. Biophys. Conf Eur. Biophys. J. 26:119. Rioux, L., and C. Gicquaud. 1985. Actin paracrystalline sheets formed at the surface of positively charged liposomes. J. Ultrastruct. Res. 93: 42-49.

[Reference]

Schmidt, C. F., M. Barmann, G. Isenberg, and E. Sackmann. 1989. Chain dynamics, mesh size, and diffusive transport in networks of polymerized actin. A quasielastic light scattering and microfluorescence light study. Macromolecules. 22:3638-3649.

[Reference]

Schnurr, B., F. Gittes, F. C. Macintosh, and C. F. Schmidt. 1997. Determining microscopic viscoelasticity in flexible and semiflexible polymer networks from thermal fluctuations. Macromolecules. 30:7781-7792. Sheterline, P., and J. C. Sparrow. 1994. Biological characteristics of actin. Protein Profile. 1:1-12.

Steinmetz, M. O., K. N. Goldie, and U. Aebi. 1997. A correlative analysis of actin filament assembly, structure, and dynamics. J. Cell BioL 138: 559-574.

[Reference]

Taylor, K. E., and D. W. Taylor. 1992. Formation of 2-D paracrystals of F-actin on phospholipid layers mixed with quaternary ammonium surfactants. J. Struct. Biol. 108:140-147.

Venien-Bryan, C., P.-F. Lenne, C. Zakri, A. Renault, A. Brisson, J.-F. Legrand, and B. Berge. 1998. Characterization of the growth of 2D protein crystal on a lipid monolayer by ellipsometry and rigidity measurements coupled to electron microscopy. Biophys. J. 74:2649-2657. Wachsstock, D. H., W. H. Schwarz, and T. D. Pollard. 1994. Cross-linker dynamics determine the mechanical properties of actin gels. Biophys. J. 66:801-809.

[Reference]

Ward, R. J., J. F. Menetret, F. Pattus, and K. Leonard. 1990. Method for forming two-dimensional paracrystals of biological filaments on lipid monolayers. J. Electron Microsc. Tech. 14:335-341. Yasuda, R., H. Miyata, and K. Kinosita. 1996. Direct measurement of torsional rigidity of single actin filaments. J. Mol. Biol. 263:227-236. Zakri-Delplanque, C. 1997. Etude de la fusion-crystallisation de monocouches de 1-alcool 1 la surface de l'eau: mesures d'elasticite laterale par diffraction de rayons X et par une methode mecanique. Ph.D. thesis University Joseph Fourier, Grenoble.

Ziemann, F., J. Radler, and E. Sackmann. 1994. Local measurements of viscoelastic moduli of entangled actin networks using an oscillating magnetic bead microrheometer. Biophys. J. 66:1-7.

[Author Affiliation]

Anne Renault,* Pierre-Fran4;ois Lenne,* Cecile Zakri,* Achod Aradian, Catherine Vdnien-Bryan,# and Fran ois Amblard

[Author Affiliation]

*Laboratoire de Spectrometrie Physique, Centre National de la Recherche Scientifique, UMR-5588, BP87, 38402 St Martin d'Heres; #Institut de Biologie Structurale Jean-Pierre Ebel, Commissariat a l'Energie Atomique et Centre National de la Recherche Scientifique, 38027 Grenoble Cedex 1; and Laboratoire de Physiologie, Ecole Superieure de Physique et Chimie Industrielle et Centre National de la Recherche Scientifique, UMR-7637-Neurobiologie, 75231 Paris Cedex 05, France

[Author Affiliation]

Received for publication 2 April 1998 and in final form 25 November 1998. Address reprint requests to Dr. Franqois Amblard, Laboratoire de PhysicoChimie, Institut Curie, 11 rue Pierre et Marie Curie, 75005 Paris, France. Tel.: 33-1-42346795; Fax: 33-1-42346795; E-mail: francois.amblard@ curie.fr.

[Author Affiliation]

We are indebted to Claude Gicquaud for initiating our interest in actin/lipid interactions and for collaborating on the first experiments. We thank Bruno Berge for his constant scientific support. S. Charpak and J. Rossier are acknowlegded for their warm support, E. Beaurepaire for his assistance, and J. P. Le Caer for purity tests using mass spectrometry. This work was made possible partly by a grant from the Defense Ministry to FA (contract 961177).