Micromechanics and ultrastructure of actin filament networks crosslinked by human fascin: A comparison with α-actinin1

doi:10.1006/jmbi.2001.4716 available online at http://www.idealibrary.com on

J. Mol. Biol. (2001) 310, 351±366

Micromechanics and Ultrastructure of Actin Filament Networks Crosslinked by Human Fascin: A Comparison with a -Actinin Yiider Tseng1, Elena Fedorov2, J. Michael McCaffery3, Steven C. Almo2 and Denis Wirtz1,4* 1

Department of Chemical Engineering, The Johns Hopkins University, 3400 N. Charles St., Baltimore MD 21218, USA 2

Department of Biochemistry Albert Einstein College of Medicine, Yeshiva University 1300 Morris Park Ave., Bronx NY 10461, USA 3

Department of Biology, Integrated Imaging Center, The Johns Hopkins University, 3400 N. Charles St., Baltimore MD 21218, USA 4

Department of Materials Science and Engineering The Johns Hopkins University, 3400 N. Charles St., Baltimore MD 21218, USA

Fascin is an actin crosslinking protein that organizes actin ®laments into tightly packed bundles believed to mediate the formation of cellular protrusions and to provide mechanical support to stress ®bers. Using quantitative rheological methods, we studied the evolution of the mechanical behavior of ®lamentous actin (F-actin) networks assembled in the presence of human fascin. The mechanical properties of F-actin/fascin networks were directly compared with those formed by a-actinin, a prototypical actin ®lament crosslinking/bundling protein. Gelation of F-actin networks in the presence of fascin (fascin to actin molar ratio >1:50) exhibits a non-monotonic behavior characterized by a burst of elasticity followed by a slow decline over time. Moreover, the rate of gelation shows a non-monotonic dependence on fascin concentration. In contrast, a-actinin increased the F-actin network elasticity and the rate of gelation monotonically. Time-resolved multiple-angle light scattering and confocal and electron microscopies suggest that this unique behavior is due to competition between fascin-mediated crosslinking and side-branching of actin ®laments and bundles, on the one hand, and delayed actin assembly and enhanced network micro-heterogeneity, on the other hand. The behavior of F-actin/fascin solutions under oscillatory shear of different frequencies, which mimics the cell's response to forces applied at different rates, supports a key role for fascin-mediated F-actin side-branching. F-actin side-branching promotes the formation of interconnected networks, which completely inhibits the motion of actin ®laments and bundles. Our results therefore show that despite sharing seemingly similar F-actin crosslinking/bundling activity, a-actinin and fascin display completely different mechanical behavior. When viewed in the context of recent microrheological measurements in living cells, these results provide the basis for understanding the synergy between multiple crosslinking proteins, and in particular the complementary mechanical roles of fascin and a-actinin in vivo. # 2001 Academic Press

*Corresponding author

Keywords: side-branching; time-dependent confocal microscopy; rheology; time-dependent multiple-angle static light scattering

Introduction

Abbreviations used: F-actin, ®lamentous actin; G-actin, globular (unpolymerized) actin; Arp, actin related protein; MSD, mean squared displacement; EM, electron microscopy. E-mail address of the corresponding author: [email protected] 0022-2836/01/020351±16 $35.00/0

Fascin is a ubiquitous actin ®lament crosslinking protein (Mr 55-58 kDa) expressed in a variety of cell types.1 ± 3 Fascin was ®rst identi®ed in cytoplasmic extracts of sea urchin eggs4 and homologs have been detected in a wide range of invertebrates and in vertebrates, including Drosophila,5 star®sh,6 Xenopus,7 rodents,8 and humans.9 Sequence alignment indicates that the members of the fascin family form a distinct class of proteins as # 2001 Academic Press

352 they share a high degree of sequence identity with one another, but do not have signi®cant sequence similarities with other known proteins.3,10 Unlike the homodimeric and ¯exible protein a-actinin, fascin appears to be a monomeric and compact molecule11 that crosslinks actin ®laments via two distinct actin-binding sites, one of which has been localized to the carboxyl-terminal half of the molecule.12. Presumably because of its compact structure (S.C.A. et al., unpublished results), fascin organizes actin ®laments into tight bundles.13 In living cells, fascin localizes to a number of highly dynamic cellular structures that require strong mechanical support, including microvilli,3,5,10,18,19 microstress ®bers,14 ± 17, 3,14,16,17,20 ± 24 spikes, and lamellipodia.16,25 For instance, fascin helps organize actin ®laments into well-ordered bundles inside microvilli that rapidly elongate from the surface of a sea-urchin egg after fertilization at the points of sperm contact.26 Fascin is also located in the core actin bundles of ®lopodia formed during the activation of coelomoytes (phagocytic cells from the coelomic cavity).22,23 B lymphocyte activation is accompanied by extensive membrane ruf¯ing and pseudopod formation, which appears to be mediated by fascin.27 Moreover, fascin is thought to be responsible for the generation of dynamic membrane protrusions, including microvilli on the apical surface and lamellipodia at the basolateral surface, of motile epithelial cells.16 These observations together with the actin-bundling activity of fascin and its intracellular localization suggest that fascin plays a key role in coordinating the organization of the cytoskeleton, directing the extension of plasma membrane, and providing mechanical support to cellular protrusions. In vitro, networks of pure F-actin offer relatively poor resistance to applied mechanical deformations: actin ®laments start breaking at shear deformations as small as 1.5-3 %28 and exhibit a very low elasticity even at high (physiological) concentrations.29 Moreover, microrheological studies in living cells have revealed that F-actin is necessary, but not suf®cient, to maintain cell morphology and associated elasticity.30 It is unclear, however, whether fascin alone can suf®ciently enhance the resistance of F-actin to mechanical stresses where actin and fascin are predominant proteins, i.e. in ®lopodia and in stress ®bers. Can fascin provide supplemental stiffness to F-actin? On one hand, the ¯exural rigidity of a fascin-mediated bundle is expected to increase with the number of ®laments in the bundle, which is dependent on fascin concentration. Increased bundle rigidity would increase the elasticity of an F-actin network.31 On the other hand, extensive F-actin bundling by fascin will decrease the effective concentration of actin polymers (®laments ‡ bundles), which would decrease the elasticity of the network.32 It is also unclear whether the fascin:actin stochiometry, which controls the growth

Properties of F-actin/Fascin Networks

rate of F-actin/fascin bundles,33 regulates the rate of gelation of F-actin/fascin networks. To investigate the mechanical role of fascin, we studied the formation, organization, and mechanical behavior of reconstituted F-actin/fascin networks over a wide range of fascin concentrations using rheological methods, time-resolved electron and confocal microscopies, and time-resolved multiple-angle light scattering. These results were directly compared to those obtained with the prototypical F-actin crosslinking protein a-actinin, which co-localizes with fascin to stress ®bers and lamellipodia of adherent cells.34 Like a-actinin, fascin crosslinks and bundles actin ®laments, yet the mechanical behavior of fascin is completely different from that of a-actinin.

Results Gelation kinetics of F-actin/fascin and F-actin/ a -actinin solutions To characterize the mechanical properties of actin solutions in the presence of fascin, we measured the rate and extent of network gelation upon addition of actin-polymerizing salt to Gactin/fascin solutions, as well as the frequencydependent viscoelastic moduli at steady state. The gelation of F-actin/fascin networks was quantitatively monitored using time-resolved rheometry over a wide range of fascin concentrations. In the absence of fascin and in actin/fascin solutions for molar ratios of fascin to actin of up to 1:50, the elastic modulus increased monotonically with time (Figure 1(a)). The associated phase angle, which compares viscous and elastic moduli (see Materials and Methods), decreased steadily during gelation, i.e. F-actin/fascin solutions progressively adopted a solid-like character (Figure 1(b)). Notably, the increase in elasticity and corresponding decrease in the phase angle of F-actin were evident even for very low fascin:actin stochiometries ([fascin]:[actin] ˆ 1:500; Figure 1(a) and (b)), for which no bundling was apparent as detected by electron microscopy.33 This observation suggests that fascin can promote the interactions between actin ®laments (i.e. crosslinking of ®laments) without the formation of bundles. This crosslinking activity, as detected by enhanced elasticity, was also displayed by F-actin/a-actinin solutions, albeit at much higher a-actinin concentrations; we only observed an enhancement of F-actin elasticity for [a-actinin]: [actin] >1:50 (data not shown). Interestingly, whereas the extent of actin solution gelation increased with fascin concentration, the rate of gelation decreased with fascin concentration for molar ratios of fascin to actin of up to 1:50 (Figure 2(a)). This trend was reversed at high fascin concentrations; past [fascin]: [actin] ˆ 1:50, the rate of gelation of F-actin/fascin networks started increasing with fascin concentration and became much higher than the gelation rate of pure F-actin networks (Figure 2(a)). More-

Properties of F-actin/Fascin Networks

353

Figure 1. Non-monotonic gelation of actin in the presence of fascin. (a) Monotonic enhancement of elasticity of F-actin/fascin solutions for fascin concentration between 0 and 0.48 mM during gelation. Error bars are as large or smaller than the symbols. The arrowhead indicates the onset of reduction of elasticity at 0.48 mM. (b) Monotonic decrease of the phase angle during gelation. Same conditions as in (a). The phase angle measures whether the material is solid-like (d ˆ 0  ), liquid-like (d ˆ 90  ), or viscoelastic (90  > d > 0  ). (c) Non-monotonic enhancement of elasticity of F-actin/fascin solutions during gelation at early times for fascin concentration between 0.8 and 2.4 mM. Inset: evolution of the elasticity at long times. (d) Non-monotonic decrease of the phase angle during gelation at early times. Same conditions as in (c). Inset: evolution of the phase angle at long times. (e) Monotonic enhancement of elasticity of F-actin/a-actinin solutions for a-actinin concentrations between 0 and 2.4 mM. (f) Corresponding monotonic decrease of the phase angle of F-actin/a-actinin solutions. Elasticity and phase angle were monitored by imposing a shear deformation of 1 % amplitude at a frequency of 1 rad/second every 30 seconds (see Materials and Methods). Actin concentration was 24 mM ( 1 mg/ml); fascin concentrations are indicated in the Figures.

354

Properties of F-actin/Fascin Networks

Figure 2. Characterization of the non-monotonic behavior of F-actin/fascin solutions during gelation. (a) Rate of gelation de®ned as the inverse of the time it takes to reach 90 % of the steady state value of elasticity (see Materials and Methods). Inset: rate of gelation of F-actin/a-actinin solutions as a function of a-actinin concentration. (b) Maximum elasticity (squares) and elasticity at long times (six hours) (circles) as a function of fascin concentration (obtained from gelation curves in Figure 1(a) and (c)). Inset: long-time elasticity of F-actin/a-actinin solutions as a function of a-actinin concentration. (c) Minimum phase angle (squares) and phase angle at long times (six hours) (circles) as a function of fascin concentration (obtained from Figure 1(b) and (d)). Inset: steady-state phase angle of F-actin/ a-actinin solutions as a function of a-actinin concentration. (d) Time at which the elasticity passed through a maximum during the gelation of F-actin/fascin solutions as a function of fascin concentration (obtained from Figure 1(a) and (c)). Actin concentration in all experiments was 24 mM.

over, the gelation of F-actin/fascin became nonmonotonic with the elastic modulus passing through a distinct maximum (Figure 1(c)), i.e. F-actin ®rst stiffened and then softened with time. The difference between the elasticity measured at long times and this maximum elasticity, which was approximately zero up to [fascin]:[actin] ˆ 1:50, became dramatically dependent on fascin concentration past 1:50 (Figure 2(b) and (c)). Simultaneously, the phase angle passed through a minimum, describing a tendency for F-actin/fascin networks to ®rst rigidify and then liquefy during the gelation process (Figure 1(d)). The time point at which G0 and d reversed their course during gelation was a very strong function of fascin concentration, decreasing 290-fold for [fascin]:[actin] between 1:100 and 1:10 (Figure 2(d)). This peculiar, non-monotonic F-actin network gelation behavior was remarkably different from

that exhibited by F-actin in the presence of the prototypical crosslinking protein a-actinin. The gelation of F-actin/a-actinin solutions was accompanied by a monotonic increase (decrease) of the elasticity (phase angle) over a wide range of a-actinin concentrations (Figure 1(e) and (f)). This means that, unlike fascin/actin solutions, the stiffness and solid-like character of F-actin/a-actinin solutions were enhanced monotonically during the gelation process. Moreover, the rate of gelation decrease monotonically with a-actinin concentration (inset in Figure 2(a)); steady-state elasticity and phase angle, respectively, increased and decreased monotonically with a-actinin concentration (insets in Figure 2(b) and (c)). Therefore, despite apparently similar F-actin crosslinking and bundling activities, a-actinin and fascin enhanced F-actin stiffness in a dramatically different fashion.

Properties of F-actin/Fascin Networks

Evolution of the overall organization of actin/ fascin solutions as probed by time-resolved confocal microscopy The effect of fascin on the evolution of the organization of actin/fascin solutions during gelation was examined by scanning-laser confocal microscopy. We emphasize that confocal microscopy was conducted in conditions identical to those used for rheometry. Since we used small

355 amounts of TRITC-phalloidin to prevent (or at least delay) signal saturation during F-actin network formation, confocal micrographs of actin ®laments displayed little apparent ¯uorescence in the absence of fascin (data not shown); however, when co-polymerized with fascin, F-actin solutions became dramatically ¯uorescent (Figure 3(a)-(d)). During the early stages of the gelation of F-actin/ fascin solutions, ®laments rapidly interacted to

Figure 3. Actin ®lament organization of F-actin/fascin solutions as probed by time-resolved confocal microscopy. (a)-(d) Evolution of the F-actin organization during the gelation of a 24 mM F-actin solution in the presence of 2.4 mM fascin, (a) ten minutes, (b) 30 minutes, (c) 90 minutes, and (d) 300 minutes after addition of polymerizing salt to the actin/fascin solution. Exposure time for image acquisition was constant during the course of network formation. (e)-(g) Steady state organization of actin ®laments in F-actin/fascin solutions for various concentrations of fascin: (e) 24 mM F-actin ‡ 0.48 mM fascin, (f) 24 mM F-actin ‡ 1.2 mM fascin, (g) 24 mM F-actin ‡ 2.4 mM fascin. Images (e)-(g) were collected 16 hours after addition of polymerizing salt to actin. All images are maximal projections of Factin gels, which were obtained by laser-scanning confocal microscopy (see Materials and Methods). Note that the exposure time in (e)-(g) was much smaller than in (a)-(d) to prevent saturation of the signal. Calibration bar is 10 mm.

356 form rigid bundles, which grew in size over time, and subsequently formed highly interconnected networks (Figure 3(a)-(d)). At later times, the thickness of the bundles continued to grow, but their number declined as extremely large bundles formed (Figure 3(e)-(g)). By scanning the sample chamber, it was apparent that fascin induced the F-actin network to heterogeneously ``contract'', which resulted in actin-depleted zones in the chamber (data not shown). This fascin-induced micro-heterogeneity was quantitatively measured using multiple particle tracking in a separate paper.37 At low fascin concentrations, network contraction was less pronounced and the number and the size of bundles was lower (Figure 3(e)-(f)).

Actin assembly and morphology in the presence of fascin as probed by time-resolved electron microscopy Electron microscopy (EM) was used to examine the structure of individual actin ®laments and the detailed time-dependent morphology of F-actin bundles. These EM measurements complement earlier studies of fascin-containing sea urchin egg extracts33 by using puri®ed fascin and probing F-actin bundle formation at early times. EM experiments were conducted at molar ratios of fascin-to-actin equal to those used for confocal microscopy and rheometry. Solutions of unpolymerized (gel-®ltrated) actin in the absence of fascin displayed small disordered actin aggregates (see also44) (Figure 4(a)), which progressively vanished after addition of polymerized salt (Figure 4(f) and (g)). Upon addition of salt, we found that polymerizing actin ®laments exhibited a rather ragged and lightly branched appearance at early times (Figure 4(f)), which mostly vanished with time (Figure 4(g)). Similar observations have been reported by Aebi and coworkers.45,46 At steady state (1 hour), actin ®laments showed a smooth morphology (Figure 4(g)). In the presence of fascin and for short times after initiation of actin polymerization, actin ®laments also appeared ragged but displayed a more highly branched morphology (Figure 4(d) and (i)). Moreover, some actin did not incorporate into the ®laments (Figure 4(b)) and careful examination of EM micrographs showed that, unlike G-actin, these structures contained short ®laments (compare Figure 4(b) and (c)). At later times (one to six hours), well-developed actin ®laments and bundles were formed (Figure 4(e) and (h)), but these actin structures mostly preserved their side-branches (Figures 4(e) and 8) (see more below). At long times (1.5 days), micrographs showed both relatively disordered bundles and ordered F-actin bundles, as well as extended side-branching (Figure 8; see more below).

Properties of F-actin/Fascin Networks

Actin polymerization in the presence of fascin as probed by time-resolved multiple-angle light scattering To further study the early steps of actin assembly and morphology in the presence of fascin, we used time-resolved multiple-angle static light scattering. This method has the salient advantage that it does not require dyes or ®xatives and provides insight into actin ®lament morphology in solution.42 We measured the time-dependent scattering intensity, I, simultaneously at 18 different angles y, i.e. as a function of the scattering-vector amplitude q (see Materials and Methods). Light scattering from solutions of G-actin, a-actinin, or fascin showed no dependence on q (data not shown), which is expected since the length scales probed by the light-scattering instrument (ˆ2p/q) are in the range 0.25-1.22 mm, which is much larger than the size of individual G-actin, a-actinin, and fascin molecules. Polymerized actin exhibited a q-dependent spectrum, a result consistent with the fact that length scales 2p/q are larger than the diameter of F-actin (7-9 nm), but shorter than the average length of an actin ®lament in vitro (7-15 mm). Light scattering spectra of F-actin solutions without auxiliary molecules scaled as an effective power law I(q)  q ÿ a (Figure 5(a)). The exponent a was initially equal to 1.41(0.05), and decreased within ten minutes after addition of polymerizing salt to a value of 1.13(0.04), and within 20 minutes to a steady state value of 1.07(0.03) (Figure 5(a) and (c)). This steady state exponent was very close to that predicted for a rigid rod, for which I(q)  qÿ1.43 This result is consistent with the fact that F-actin has a persistence length of 17 mm47,48 and is therefore rigid-like at length scales of the order of a micron. In striking contrast to this control, F-actin/fascin solutions displayed scattering spectra that scaled with an exponent a which increased from 1.41(0.05) to 1.59(0.02) within ten minutes (Figure 5(a)). The time required for the exponent a to reach a steady-state value slightly decreased for increasing fascin concentrations (Figure 5(c)). In contrast to F-actin/fascin solutions, the timedependent shape of the scattering spectra of F-actin/a-actinin solutions was indistinguishable from that of F-actin and relatively independent of a-actinin concentration (Figure 5(b) and (c)). Similarly to F-actin, the exponent a of the lightscattering spectra displayed by F-actin/a-actinin solutions steadily decreased towards unity with time (Figure 5(c)). Therefore, in view of the differences in scattering spectra (q-dependence of the light intensity) (Figure 5), the morphology of actin ®laments in solutions of F-actin/fascin is fundamentally different from that of actin ®laments in solutions of F-actin or F-actin/a-actinin. Time-resolved light scattering measurements also complemented the EM measurements and allowed the kinetics of actin assembly and morphology to be monitored in solution. We measured

Properties of F-actin/Fascin Networks

357

Figure 4. Actin and actin/fascin morphology as probed by time-resolved electron microscopy. (a) G-actin. (b) Typical, early-time F-actin/fascin aggregate, which does not incorporate into ®lamentous structures such as those shown in (d). (b) and (c) were observed a short time after addition of polymerizing salt (ten minutes) (see Materials and Methods). (c) Detail of the G-actin aggregate taken from (a). (d) Typical F-actin/fascin structure observed at short times after the onset of polymerization (ten minutes) (e) Typical ordered F-actin/fascin structure observed one hour after the onset of polymerization. (f) and (g) Typical F-actin structure in the absence of fascin observed 20 minutes (f) and one hour (g) after the onset of polymerization. (h) Detail of an F-actin/fascin bundle and a (long) side branch (30 minutes). (i) Detail of a (short) side-branch observed at ten minutes. Electron micrographs were obtained by negative staining of F-actin (see Materials and Methods). The concentrations of actin and fascin were 1 mM and 0.1 mM, respectively. Calibration bar is 0.1 mm.

the scattering light intensity at 90  from the direction of incident light, which corresponds to a ®xed wavevector q ˆ 18.7  10ÿ2 nmÿ1 or a length scale of 0.34 mm. These light scattering experiments showed that fascin greatly decreased the rate of

change of scattered light intensity from polymerizing actin in solution (Figure 6(a) and (b)); in contrast, a-actinin did not signi®cantly alter the rate of intensity increase of polymerizing actin (Figure 6(a) and (b)). Therefore, the kinetics of actin assembly

358

Properties of F-actin/Fascin Networks

and morphological change in the presence of fascin are dramatically different from that in the presence of a-actinin.

Dynamic properties of fascin at steady state

Figure 5. Actin ®lament structure in solution as probed by time-resolved static light scattering. (a) Scattering-vector amplitude-dependent light scattering intensity, I(q), for solutions of F-actin and F-actin/fascin measured ten minutes after addition of polymerizing salt. Exponents a were obtained from power-law ®ts of the type I(q)  q ÿ a and are indicated by the corresponding curves. (b) I(q) and associated exponent a for solutions of F-actin/a-actinin also measured after ten minutes. (c) Time-dependent exponent a for solutions of F-actin, F-actin/fascin, and F-actin/a-actinin. Symbols in Figures correspond to solutions of F-actin (®lled triangles), F-actin/fascin 20:1 (®lled circles), F-actin/fascin 50:1 (®lled squares), F-actin/a-actinin 20:1 (open circles), and F-actin/a-actinin 50:1 (open squares). Actin concentration was 24 mM.

Rheometric measurements with the crosslinking protein a-actinin suggest that the lifetime of binding of an auxiliary protein to F-actin controls the dynamic response of a crosslinked actin network at steady state.49 ± 51 The dynamic viscoelastic moduli of F-actin/fascin networks were probed after gelation had reached a steady state by applying oscillatory deformations (see Materials and Methods). Such oscillatory deformations mimic the response of the cell to deformations of different rates. The elastic modulus, G0 (o), of F-actin/fascin networks rapidly became independent of the frequency o with increasing fascin concentrations, indicating inhibited movements of actin ®laments and bundles presumably because of the slow kinetics of dissociation of fascin from F-actin (Figure 7(a)).28,51 Moreover, F-actin/fascin networks were relatively more viscous (i.e. large phase angle) at low and high frequencies and more elastic (i.e. small phase angle) at intermediate frequencies (Figure 7(b)). In contrast to F-actin/fascin solutions, F-actin/ a-actinin solutions displayed a very dynamic behavior (inset in Figure 7(a)) and a phase angle that depended weakly on frequency even at high a-actinin concentrations for which a-actinin-mediated ®lament bundling is induced (inset in Figure 7(b)) (see also49,51 ± 53). Therefore, the dynamic properties of the two crosslinking proteins fascin and a-actinin are diametrically opposite from each other. To examine the dependence of the steady state mechanical properties of F-actin on fascin concentration, elasticity and phase angle of the networks were measured at a ®xed frequency over a wide range of fascin concentrations. That frequency (o ˆ 1 rad/second) was chosen to correspond with the plateau modulus of F-actin/fascin solutions (see Figure 7(a)). The stiffness (i.e. the value of the plateau modulus) of a 24 mM F-actin network quickly increased tenfold for fascin concentrations up to 0.48 mM, but leveled off thereafter (Figure 2(b)). The viscoelastic character (determined by the phase angle) of F-actin/fascin networks was solid-like at low fascin concentrations, but became progressively more liquid-like for fascin concentrations higher than 0.48 mM (Figure 2(c) and 7(b)). In contrast, the elasticity of F-actin/ a-actinin steadily increased with a-actinin concentration and showed no saturation up to 1:10 stochiometries (inset in Figure 2(b)). Moreover, in contrast to fascin, the phase angle of F-actin networks decreased only slightly with a-actinin concentration (inset in Figure 2(c)). Therefore, over a wide range of concentrations of auxiliary proteins, the mechanical properties of F-actin/fascin and F-actin/a-actinin networks at steady state are fundamentally different.

359

Properties of F-actin/Fascin Networks

Figure 6. Actin ®lament assembly in the presence of fascin. (a) Time-dependent light scattering intensity, measured at 90  from the direction of incident light, for solutions of F-actin, F-actin/fascin, and F-actin/a-actinin. Symbols correspond to solutions of F-actin (®lled triangles), F-actin/fascin 20:1 (®lled circles), F-actin/fascin 50:1 (®lled squares), F-actin/a-actinin 20:1 (open circles), and F-actin/a-actinin 50:1 (open squares). (b) Rate of change of scattering light intensity measured at 90  . This rate is the inverse of the time required to reach 90 % of the apparent steady state light intensity. Actin concentration was 24 mM.

Fascin-mediated side-branching of F-actin EM was used to relate the mechanical behavior of F-actin/fascin networks to actin ®lament organization at steady state. At steady state, the organization and type of structures present in F-actin/ fascin solutions was fundamentally different from those in the absence of fascin. In addition to the well-documented bundling activity of fascin (Figure 8; see also33), we found that the sidebranching was maintained not only at short times (Figure 4), but also at long times (6-48 hours) after the onset of polymerization (Figure 8(a)-(d)). Sidebranching promoted the interconnection of bundles, which formed highly extended structures (Figure 8(a) and (b)). These side-branches often consisted of thin bundles originating isotropically(i.e. no preferred angle) from a (slightly) thicker actin bundle (Figure 8(b)). Ultimately, highly ordered bundles of actin ®laments appeared (Figure 8(e)-(h)).

Discussion In vivo experiments suggest that one of the physiological functions of fascin-containing bundles is to provide mechanical support to cellular protrusions such as microvilli, lamellipodia, and ®lopodia. This paper presents a systematic study of the mechanical properties and ultrastructure of F-actin/fascin solutions in vitro. We found that: (1) The gelation of actin ®lament networks in the presence of fascin is accompanied by a major reorganization of ®laments in solution, which, at relatively high fascin concentrations, causes the network to undergo a transient burst followed by a slow decline in elasticity. Such an ``elastic collapse'' does not occur during the gelation of F-actin and F-actin/a-actinin solutions. (2) Fascin delays the

gelation of F-actin networks and induces a decrease in the rate of change of scattering light intensity. (3) In contrast to a-actinin, crosslinking of F-actin by fascin almost completely inhibits the motion of ®laments in solution even at long time scales. (4) Fascin enhances actin-®lament sidebranching, which results in reticulated networks. Non-monotonic gelation of F-actin Time-resolved EM and light microscopy suggest an explanation for the non-monotonic course of gelation of F-actin/fascin solutions (Figure 1), which is not exhibited by a-actinin (Figure 1(e) and (f)), nor by EF-1a, another well-known actin bundling protein (Y.T. and D.W., unpublished results).54 At early times, fascin-mediated side-branching and ®lament bundling creates a highly interconnected network of both ®laments and bundles (Figures 3, 4 and 8), which quickly increases the overall elasticity of the gel. At later times, however, more extensive fascin-induced bundling and associated contraction of the actin ®lament network creates large heterogeneities. At steady state, the degree of heterogeneity in F-actin/fascin networks is indeed much higher than found in F-actin networks as recently shown by multiple-particle tracking measurements.37 To quantitatively assess F-actin network heterogeneity, microspheres were dispersed in F-actin networks in the presence and absence of fascin and video-tracked with 5 nm resolution. The width of the mean square displacement (MSD) distribution normalized by the ensemble-average MSD was much larger in the presence of fascin and correlated with a high degree of heterogeneity compared to F-actin alone.37 This local microheterogeneity can globally weaken the F-actin/fascin network, which may ultimately

360

Properties of F-actin/Fascin Networks

rate of light intensity increase was decreased in the presence of fascin, but not with a-actinin (Figure 6). Using the pyrene ¯uorescence assay, we recently found that the rate of actin assembly was monotonically reduced for increasing concentration of fascin; by contrast, time-dependent ¯uorescence of polymerizing actin was unmodi®ed by a-actinin (Y.T. and D.W., unpublished data). This delayed actin assembly increases the time necessary for ®laments/bundles to form entangled, elastic networks. This in turn reduces the relative rate of increase of elasticity during the early phase of gelation for increasing fascin concentrations (Figure 2(a)). However, at high fascin concentrations, despite the delay in actin assembly, enhanced side-branching promotes rapid ®lament crosslinking and faster interconnectivity of actin ®laments, which greatly increases the rate of F-actin network gelation (Figure 2(a)). These competing functions, F-actin side-branching/ crosslinking and delayed actin assembly, provide fascin with multiple regulatory functions. Enhancement of F-actin elasticity by fascin

Figure 7. Dynamic crosslinking of F-actin by fascin at steady state. (a) Frequency-dependent elasticity of F-actin/fascin solutions. Inset: frequency-dependent elasticity of a F-actin/a-actinin solution. (b) Frequencydependent phase angle. Inset frequency-dependent phase angle of a F-actin/a-actinin solution. Frequencydependent viscoelastic moduli were measured by sequentially applying a 1 %-amplitude shear deformation of frequencies between 0.01 and 100 rad/second (see Materials and Methods). All measurements were conducted six hours after addition of polymerizing salt to actin. Actin concentration in all experiments was 24 mM; fascin concentrations are indicated in the Figures; a-actinin concentration was 0.48 mM.

induce the observed elastic collapse. Our rheological data suggest that increasing the concentration of fascin simply expedites the formation of interconnected networks, which is partly due to the enhanced rate of bundle formation;33 this initially increases the network's rate of elasticity increase. But increasing fascin concentration also accelerates the formation of heterogeneities, which reduces the time to induce an elastic collapse during gelation. Our multiple-angle light scattering and EM observations suggest that fascin prevents normal addition of actin monomers to growing ®laments (Figures 4-6). In particular, we observed that the

Increasing the concentration of fascin in solution could a priori have contrasting effects on the steady-state stiffness of an F-actin network. On one hand, the number of ®laments per bundle increases with fascin concentration,33 which would enhance the ¯exural (i.e. bending) rigidity of F-actin bundles.47 On the other hand, the elasticity of a solution of semi¯exible polymers is predicted to depend very weakly on the intrinsic rigidity of the polymer (G0  k2/5, where k is the ¯exural rigidity of the polymer) and instead should depend strongly on polymer concentration (G0  c7/5).29,32,55 An increasing fascin concentration increases the number of ®laments per bundle,33 which decreases the total number of polymers (®laments ‡ bundles) in solution and should lead to global weakening of an F-actin network. Here we ®nd that, despite this decreased polymer concentration, an increasing fascin concentration enhances the stiffness of F-actin solutions (Figure 2(b)), at least at low fascin concentrations, which may be partially due to the fact that an increasing fascin concentration also lengthens F-actin/fascin bundles.33 High fascin-toactin stochiometries (>1/50) have no effect on the steady-state elasticity of F-actin solutions (Figure 2(b)). We speculate that the reduction of polymer concentration, which would contribute to lower network stiffness, is compensated by fascinmediated lengthening of the bundles and F-actin side-branching, which effectively interconnects actin ®laments and F-actin bundles (Figure 7), and in turn increases the steady state elasticity of F-actin networks. F-actin side-branching has not previously been reported to be enhanced by fascin, which may be due to the fact that previous EM studies made use of sea urchin egg extracts,33,56 which feature a multitude of actin-binding proteins (see more below).

Properties of F-actin/Fascin Networks

361

Figure 8. Ultrastructure of F-actin/fascin networks at long times. (a) Low-magni®cation micrograph of an F-actin bundle network. Note the extensive actin-bundle side-branching, indicated by arrowheads. (b) This micrograph shows typical side-branching of F-actin bundles (left arrow) and overlapping bundles (i.e. bundles are on top of each other; right arrow) observed six hours after onset of polymerization. (c) Typical side branching of ®lamentous bundles. (d) Ordered bundles that overlap each other. (e)-(h)) Highly ordered bundles generated 48 hours after onset of polymerization. Electron micrographs were obtained by negative staining of F-actin (see Materials and Methods). Actin concentration was 1 mM, fascin concentration was 0.1 mM. Calibration bars: 0.1 mm in (a), (d), (e) and (f), and 0.05 mm in (b), (c), (g) and (h).

362 The observed frequency pro®le of the elastic modulus (Figure 7) is consistent with a signi®cant contribution of F-actin side-branching to the mechanical behavior of F-actin/fascin solutions. In the absence of fascin, F-actin networks exhibit a relatively low-amplitude plateau modulus at low frequencies53 because ®laments can move and ultimately relax the stress at long time scales (note that this ®nal relaxation is not probed here).57 At high frequencies (probed for example by diffusing wave spectroscopy or laser-de¯ection particle tracking), the frequency dependence of the elastic modulus is steep because the network can relax the stress via bending ¯uctuations between points of ®lament entanglement.29,30,53,58 In the presence of a-actinin, this increase in elasticity occurs at much lower frequencies because steric constraints are reduced, which allows rapid stress relaxation (see39 for a review of how to interpret rheological data).50 In contrast to F-actin/a-actinin, F-actin/ fascin bundles feature extensive side-branching, which would presumably hinder the lateral (i.e. sliding) movement of the ®laments/bundles57 and is likely to inhibit relaxation of the stress even at low frequencies (Figure 7). This in turn would ¯atten the frequency-dependent elasticity spectra of F-actin/fascin solutions. Our EM studies suggest that F-actin side-branching as mediated by fascin involves a mechanism different from that promoted by the protein complex Arp2/3. The Arp2/3 complex induces the formation of dendritic structures in reconstituted F-actin solutions and at the leading edge of actinrich lamellipodia of moving cells.59 ± 61 Electron microscopy and kinetics experiments suggest that the Arp2/3 complex links the pointed end of the growing ®laments to the side of other ®laments at an angle of 70  from the ®lament axis.59 In contrast, our EM studies indicate that fascin promotes side-branching of mostly F-actin bundles by splitting thick bundles into thinner ones with no preferred angular correlation. This is an organization different from side-branching observed in actin ®laments during the early phase of polymerization, which rapidly disappears.45,46,62 Moreover, the rate of actin assembly is reduced by fascin; hence, unlike Arp2/3, fascin does not appear to nucleate actin assembly under the present buffer conditions. Implications for the cell Changing patterns of fascin expression seem to be a common feature of cell transformation.1 ± 3 For instance, latent Epstein-Barr virus (EBV) infection of B lymphocytes (B cell) is accompanied by a 200fold increase in fascin mRNA.63 Assuming that mRNA levels are directly related to the expression of fascin, our results obtained in vitro suggest that the fascin-rich perinuclear cytoplasm and ®lopodia-like protrusions in infected B cells will undergo a major cytoskeletal reorganization, promoting the formation of reticulated networks of side-branched ®laments and bundles. EBV transformation of B

Properties of F-actin/Fascin Networks

cells indeed induces extensive fascin and actin-containing microspikes (Y.T. and D.W., unpublished data). Our results also predict that the cytoskeleton of EBV-transformed B cells should become at least tenfold stiffer than in uninfected B cells, which may play a key role in phenotypes associated with EBV infection of B cells, including impaired chemotaxis through connective tissue and endothelial cells. Using cryo-EM and microrheological methods, we are currently testing this model of actin cytoskeleton reorganization and stiffening in EBV-transformed B cells. Our rheological data show that reconstituted F-actin/fascin networks exhibit very low dissipation of energy under stress. This observation together with the fact that fascin localizes at the extreme tip of ®broblast ®lopodia16 supports the idea that by crosslinking actin ®laments proximal to the plasma membrane, fascin can help limit frictional dissipation of internal bending modes of ®laments when the ®laments recoil and push the membrane forward after barbed-end monomer addition.64,65 Fascin also localizes in the lamellipodia of adherent cells; using laser-de¯ection particle-tracking microrheology, we recently found that the cytoskeleton of the lamellipodia of COS7 cells exhibited an elasticity of G0 ˆ 820(520) dyne/cm2 (at 1 rad/ second).30 That study also showed that polymerized actin is necessary, but not suf®cient, to generate these levels of elasticity as cells treated with the actin-depolymerizing drug latrunculin A66 and reconstituted F-actin networks29 both exhibited much lower elasticity. Therefore, the formation and stabilization of lamellipodia requires the action of structural proteins that organize F-actin into assemblies compatible with the large stresses associated with these cellular structures.29,58 Our rheological results indicate that the crosslinking/ bundling proteins fascin and a-actinin, which both localize along the stress ®bers of the lamellipodia of motile epithelial cells,16 cannot individually account for the reinforcement of the F-actin network observed in those cells. Moreover, our microrheology measurements established that the cytoskeleton in the lamellipodia of living cells is extremely dynamic since it displayed an elastic modulus which increased from 850 to 1700 dyne/cm2 over a 1-100 rad/second frequency range.30 The results presented here show that neither fascin nor a-actinin produce networks that are suf®ciently dynamic to account for the crosslinking dynamics of F-actin in the lamellipodia of living cells. The different mechanical behavior of fascin and a-actinin suggests mechanical complementarity and synergy between these two proteins in subcellular regions where these proteins co-localize. Interestingly, we found that equimolar amounts (0.6 mM each) of a-actinin and fascin are indeed able to synergistically generate levels of elasticity of G0  990 dyne/cm2 (at 1 rad/second) (Y.T. and D.W., unpublished results), a feat achieved neither

363

Properties of F-actin/Fascin Networks

by a-actinin nor by fascin individually (this work). We also found that a-actinin regulated the dynamics of F-actin/fascin networks: together, fascin and a-actinin provided enhanced stiffness to F-actin gels for fast deformations (1930 dyne/cm2 at 100 rad/second) and less resistance to slow deformations (990 dyne/cm2 at 1 rad/second) as observed in living cells.30 Again, such highly dynamic behavior cannot be induced by a-actinin or fascin individually (Figure 7). Finally, a-actinin eliminated the ``elastic collapse'' observed during F-actin/fascin gelation. This suggests that fascin and a-actinin (or a crosslinking protein functionally similar to a-actinin) may be required to sustain the force involved in the formation and function of lamellipodium and ®lopodia. The involvement of multiple crosslinking/bundling proteins is in line with the co-requirement of the forked and singed proteins in the development of Drosophilia bristles.67,68 We believe that the possibility of functional synergism between the two F-actin crosslinking proteins fascin and a-actinin deserves further consideration and are pursuing a systematic study of the mechanical complementarity between a-actinin and fascin in vitro and in vivo.

Materials and Methods Preparation of proteins Actin was obtained from chicken skeletal acetone powder by the method of Spudich and Watt35 and gel ®ltered on Sephacryl S-300 HR36 as described.29 The puri®ed actin was stored as Ca2‡-actin in continuous dialysis at 4  C against buffer G (0.2 mM ATP, 0.5 mM DTT, 0.2 mM CaCl2, 1 mM Na azide, 2 mM Tris-HCl, pH 8.0). Mg2‡-actin ®laments were formed by adding one volume of 10  KMEI (500 mM KCl, 10 mM MgCl2, 10 mM EGTA, 100 mM imidazole, pH 7.0) polymerizing salt buffer to nine volumes of G-actin in buffer G. Human fascin was obtained as described.37 Brie¯y, fascin was expressed as a GST fusion using pGEX2T in the E. coli strain JR600. Cultures were grown to an A600 of 0.6 at 37  C and cooled to 25  C. Protein expression was induced with the addition of 1 mM IPTG and growth continued overnight at 25  C. Cells were harvested by centrifugation, washed with PBS, resuspended in PBS containing 1 mM DTT, 0.2 mM PMSF and disrupted by sonication. The lysate was clari®ed by centrifugation in a SS-34 rotor at 15,000 rpm for 20 minutes and applied to a glutathione-Sepharose column equilibrated in PBS containing 1 mM DTT (PBS/DTT). The fusion protein was eluted with 10 mM glutathione, 50 mM Tris (pH 8.0), 1 mM DTT, 0.1 M NaCl, and dialyzed against PBS/DTT. Fascin was liberated from GST by cleavage with thrombin, followed by glutathione-Sepharose chromatography. The fascin was dialyzed against 10 mM Tris (pH 8.0), 1 mM DTT, and applied to a DE52 column that was developed with a linear gradient of 0 to 70 mM NaCl in 10 mM Tris (pH 8.0), 1 mM DTT. The puri®ed fascin was dialyzed against 10 mM Tris (pH 8.0), 10 mM NaCl, 30 mM KCl, 0.1 mM EDTA, 1 mM DTT, and stored at ÿ70  C. Chicken gizzard a-actinin was puri®ed as described.28

Mechanical rheometry The mechanical properties of F-actin/fascin networks were probed using a strain-controlled, cone-and-plate rheometer (ARES-100 Rheometrics, Piscataway, NJ) as described.29,38 The temperature of the solution was maintained at 25  C to within 0.1 deg. C. The G-actin solution (24 mM) was mixed with fascin (0.048-2.4 mM) in a test tube, one tenth volume 10  KMEI was added to the solution (total volume 1.4 ml), and immediately loaded onto the lower plate of the rheometer using a micropipette with a cut-off tip. The dead time between protein mixing and the beginning of data collection is ®xed and equal to 30 seconds for all rheological experiments. The plate is coupled to a computer-controlled motor, which can apply steady or oscillatory shear deformations of controlled frequency and amplitude. The cone is connected to a torque transducer, which measures the stress induced by the shear deformation within the F-actin specimen. When an oscillatory deformation of frequency o is applied, the instrument computes the in-phase and out-of-phase components of the stress divided by the amplitude of the oscillatory deformation, which are the storage (or elastic) modulus, G0 (o), and the loss (or viscous) modulus, G00 (o), respectively. Given the large torques created by the stiff actin gels probed here (which are well within the range of sensitivity of the transducer), we estimated that viscoelastic measurements were accurate to within 10 %. Here, we report G0 and the phase angle d ˆ tanÿ1 (G00 /G0 ), which determines whether the material behaves like a Hookean solid (d ˆ 0  ), a viscous liquid (d ˆ 90  ), or a viscoelastic element (0  < d < 90  ). A glossary of rheological de®nitions is given in a recent review.39 For each specimen, F-actin solution gelation was measured by recording G0 and d every 30 seconds, by application of two cycles of shear deformation of 1 % amplitude and 1 rad/second frequency (see Figure 1). The gelation kinetics did not follow a simple exponential behavior of the type G0 (t) ˆ G[1 ÿ exp(ÿkt)], where G would be the long-time elasticity and k the rate of gelation. Therefore, the rate of gelation was conveniently de®ned as the inverse of the time necessary to reach 90 % of G. After steady state was reached, frequencydependent elasticity G0 (o) and phase angle d(o) were measured by setting the amplitude of the oscillatory deformation to g ˆ 1 % and sequentially varying the frequency from o ˆ 0.01 to 100 rad/second (see Figure 6). Light and electron microscopies Electron microscopy was used to examine the type of F-actin structures present in solution during actin gelation in the presence/absence of fascin and to characterize F-actin morphology at steady state. 1 mM actin samples, containing different actin-to-fascin molar ratios, were polymerized by the addition of 10  KMEI and negatively stained with 2 % (w/v) uranyl acetate (pH 4.6), placed onto Para®lm-coated, glow-discharged, 300meshed nickel or copper grids. Specimens were examined with a Phillips 420 electron microscope (Netherlands). To visualize F-actin by ¯uorescence microscopy, 10 ml of 6.6 mM rhodamine phalloidin (Molecular Probes, Eugene, OR) in methanol was deposited in the microscopy chamber (PC20 CoverWell, Grace Biolab, Eugene, OR), and allowed to dry for ten minutes as described.37 Protein mixtures were suspended in a ¯uorescence buffer containing 1 mM MgCl2, 50 mM KCl,

364 10 mM imidazole (pH 7.0), 100 mM DTT, 10 mM EGTA, as well as 0.018 mg/ml catalase, 0.1 mg/ml glucose oxidase, and 3 mg/ml glucose to reduce photobleaching. The sample was placed immediately into the chamber well, mounted on a glass slide as described40 and allowed to polymerize overnight at 4  C. Samples were examined with a Nikon PCM 2000 laser-scanning confocal system attached to a Nikon TE300 inverted microscope, equipped with a 100  magni®cation, oilimmersion objective (numerical aperture 1.3) (Nikon, Melville, NY). A total of 20 optical sections separated by 0.5 mm were captured in series and combined to generate maximum projections of the actin gels using the SimplePCI software (Compix, Cranberry Township, PA). Time-resolved multiple-angle steady light scattering Time-resolved multiple-angle steady light scattering (Wyatt Instruments, Santa Barbara, CA) was used to monitor actin assembly and the time-dependent morphology of actin ®laments in solution.41,42 The beam of a 5 mW He-Ne was incident upon an actin solution or mixtures of actin and fascin or actin and a-actinin contained in a glass vial (sample volume 4 ml). The scattered light intensity was measured simultaneously by 18 transimpedance photodiodes, which are arranged on a circle around the scattering volume. The background light intensity from the buffer was independently measured and subtracted from the measured light intensities as described.42 This paper reports both the timedependent light intensity measured at 90  from the axis of incident light and the light intensity, I(q), as a function of the scattering wavevector amplitude, q ˆ (4pn/l0) sin(y/2). Here, l0 ˆ 632.8 nm is the wavelength of the incident laser light in vacuum, n ˆ 1.332 is the refractive index of the buffer, and y ˆ 22-147  is the scattering angle between the direction of the incident light and those of the detectors. The ®nal concentration of actin used in the light scattering experiments was 5 mM; at that concentration, actin ®laments overlap.32 The concentration of the crosslinking protein (fascin or a-actinin) was either 0.10 or 0.25 mM to yield molar ratios identical to those used in the rheological and confocal microscopy measurements. Following classical analysis of scattering spectra,43 qdependent scattering-light intensities were ®tted with power laws of the type I(q)  qÿa. The exponent a re¯ects the structure of the ``scatterers'' in solution at length scales 2p/q. I(q) is independent of q when the scatterers are much smaller that the length scales probed by the instrument. These length scales are much smaller than the average length of actin ®laments and much larger than the diameter of the ®laments (see more in the text); hence spectra from solutions of polymerized actin, with and without auxiliary proteins, were q-dependent. Note that the range of q values accessible to our instrument spanned less than a decade; we therefore used the exponent a as a semi-quantitative marker of local structure of F-actin. Moreover, since extracting morphological information from scattering spectra is an inverse problem,43 there are typically multiple possible morphologies consistent with a given q-dependent spectrum. Hence, we used multiple-angle light scattering to complement our time-resolved EM and light-microscopy measurements. The rate of light intensity increase was extracted from time-dependent light intensity measured at 90  from the incident light direction and corresponds to the inverse of the time required to reach 90 % of the intensity measured at one hour (see Figure 6(b)).

Properties of F-actin/Fascin Networks

Acknowledgments S.C.A. acknowledges ®nancial support from NIH (AR44417); D.W. acknowledges ®nancial support from the NSF (CTS9812624 and DB19729358).

References 1. Otto, J. J. (1994). Actin-bundling proteins. Curr. Opin. Cell Biol. 6, 105-109. 2. Matsudaira, P. (1994). Actin crosslinking proteins at the leading edge. Sem. Cell Biol. 5, 165-174. 3. Edwards, R. A. & Bryan, J. (1995). Fascins, a family of actin bundling proteins. Cell Motil. Cytoskel. 32, 1-9. 4. Kane, R. E. (1975). Preparation and puri®cation of polymerized actin from sea-urchin egg extracts. J. Cell Biol. 66, 305-315. 5. Cant, K., Knowles, B. A., Mooseker, M. S., MahajanMiklos, S., Heintzelman, M. & Cooley, L. (1994). Drosophila singed, a fascin homolog, is required for actin bundle formation during oogenesis and bristle extension. J. Cell Biol. 125, 369-380. 6. Maekawa, S., Endo, S. & Sakai, H. (1982). A protein in star®sh sperm head which bundles actin ®laments in vitro: puri®cation and characterization. J. Biochem. (Tokyo), 92, 1959-1972. 7. Holthuis, J. C., Schoonderwoert, V. T. & Martens, C. J. (1994). A vertebrate homolog of the actin-bundling protein fascin. Biochim. Biophys. Acta, 1219, 184188. 8. Edwards, R. A., Herrera-Sosa, H., Otto, J. & Bryan, J. (1995). Cloning and expression of a murine fascin homolog from mouse brain. J. Biol. Chem. 270, 10764-10770. 9. Duh, F. M., Latif, M., Weng, Y., Geil, L., Modi, W. & Stackhouse, T. et al. (1994). cDNA cloning and expression of the human homolog of the sea urchin fascin and Drosophila singed genes which encodes an actin-bundling protein. DNA Cell Biol. 13, 821827. 10. Bryan, J., Edwards, R., Matsudaira, P., Otto, J. & Wulfkuhle, J. (1993). Fascin, an echinoid actin-bundling protein, is a homolog of the Drosophila singed gene product. Proc. Natl Acad. Sci. USA, 90, 91159119. 11. Yamashiro-Matsumura, S. & Matsumura, F. (1985). Puri®cation and characterization of an F-actin-bundling 55-kilodalton protein from HeLa cells. J. Biol. Chem. 260, 5087-5097. 12. Ono, S., Yamakita, Y., Yamashiro, S., Matsudaira, P. T., Gnarra, J. R. & Obinata, T. et al. (1997). Identi®cation of an actin binding region and a protein kinase C phosphorylation site on human fascin. J. Biol. Chem. 272, 2527-2533. 13. DeRosier, D. J. & Censullo, R. (1981). Structure of F-actin needles from extracts of sea-urchin oocytes. J. Mol. Biol. 146, 77-99. 14. Yamakita, Y., Ono, S., Matsumura, F. & Yamashiro, S. (1996). Phosphorylation of human fascin inhibits its actin binding and bundling activities. J. Biol. Chem. 271, 12632-12638. 15. Lin, X. H., Grako, K. A., Burg, M. A. & Stallcup, W. B. (1996). NG2 proteoglycan and the actin-binding protein fascin de®ne separate populations of actin-containing ®lopodia and lamellipodia during

Properties of F-actin/Fascin Networks

16.

17.

18.

19.

20.

21. 22.

23. 24.

25.

26. 27.

28.

29. 30. 31.

32.

cell spreading and migration. Mol. Biol. Cell, 7, 19771993. Yamashiro, S., Yamakita, Y., Ono, S. & Matsumura, F. (1998). Fascin, an actin-bundling protein, induces membrane protrusions and increases cell motility of epithelial cells. Mol. Biol. Cell, 9, 993-1006. Ishikawa, R., Yamashiro, S., Kohama, K. & Matsumura, F. (1998). Regulation of actin binding and actin bundling activities of fascin by caldesmon coupled with tropomyosin. J. Biol. Chem. 273, 2699126997. Walker, G. R., Kane, R. & Burgess, D. R. (1994). Isolation and characterization of a sea urchin zygote cortex that supports in vitro contraction and reactivation of furrowing. J. Cell Sci. 107, 2239-2248. Cant, K., Knowles, B. A., Mahajan-Miklos, S., Heintzelman, M. & Cooley, L. (1998). Drosophila fascin mutants are rescued by overexpression of the villin- like protein, quail. J. Cell Sci. 111, 213-221. Fischer, D., Tucker, R. P., Chiquet-Ehrismann, R. & Adams, J. C. (1997). Cell-adhesive responses to tenascin-C splice variants involve formation of fascin microspikes. Mol. Biol. Cell, 8, 2055-2075. Adams, J. C. (1997). Characterization of cell-matrix adhesion requirements for the formation of fascin microspikes. Mol. Biol. Cell, 8, 2345-2363. Otto, J. J., Kane, R. E. & Bryan, J. (1979). Formation of ®lopodia in coelomocytes: localization of fascin, a 58,000 dalton actin cross-linking protein. Cell, 17, 285-293. DeRosier, D. J. & Edds, K. T. (1980). Evidence for fascin cross-links between the actin ®laments in coelomocyte ®lopodia. Exp. Cell Res. 126, 490-494. Sasaki, Y., Hayashi, K., Shirao, T., Ishikawa, R. & Kohama, K. (1996). Inhibition by drebrin of the actin-bundling activity of brain fascin, a protein localized in ®lopodia of growth cones. J. Neurochem. 66, 980-988. Fang, X., Burg, M. A., Barritt, D., Dahlin-Huppe, K., Nishiyama, A. & Stallcup, W. B. (1999). Cytoskeletal reorganization induced by engagement of the NG2 proteoglycan leads to cell spreading and migration. Mol. Biol. Cell, 10, 3373-3387. Otto, J. J., Kane, R. E. & Bryan, J. (1980). Redistribution of actin and fascin in sea urchin eggs after fertilization. Cell Motil. 1, 31-40. Nakamura, S., Nagahama, M., Kagami, Y., Yatabe, Y., Takeuchi, T. & Kojima, M. et al. (1999). Hodgkin's disease expressing follicular dendritic cell marker CD21 without any other B-cell marker: a clinicopathologic study of nine cases. Am. J. Surg. Pathol. 23, 363-376. Xu, J., Tseng, Y. & Wirtz, D. (2000). Strain-hardening of actin ®lament networks: regulation by the dynamic crosslinking protein a-actinin. J. Biol. Chem. 275, 35886-35892. Palmer, A., Xu, J., Kuo, S. C. & Wirtz, D. (1999). Diffusing wave spectroscopy microrheology of actin ®lament networks. Biophys. J. 76, 1063-1071. Yamada, S., Wirtz, D. & Kuo, S. C. (2000). Mechanics of living cells measured by laser tracking microrheology. Biophys. J. 78, 1736-1747. Morse, D. C. (1998). Viscoelasticity of concentrated isotropic solutions of semi¯exible polymers. 1. Model and stress tensor. Macromolecules, 31, 70307043. Morse, D. C. (1998). Viscoelasticity of concentrated isotropic solutions of semi¯exible polymers. 2. Linear response. Macromolecules, 31, 7044-7067.

365 33. Stokes, D. L. & DeRosier, D. J. (1991). Growth conditions control the size and order of actin bundles in vitro. Biophys. J. 59, 456-465. 34. Yamashiro-Matsumura, S. & Matsumura, F. (1986). Intracellular localization of the 55-kD actin-bundling protein in cultured cells: spatial relationships with actin, alpha-actinin, tropomyosin, and ®mbrin. J. Cell Biol. 103, 631-640. 35. Spudich, J. A. & Watt, S. (1971). The regulation of rabbit skeletal muscle contraction. J. Biol. Chem. 246, 4866-4871. 36. MacLean-Fletcher, S. D. & Pollard, T. D. (1980). Viscometric analysis of the gelation of acanthamoeba extracts and puri®cation of two gelation factors. J. Cell Biol. 85, 414-428. 37. Apgar, J., Tseng, Y., Federov, E., Herwig, M. B., Almo, S. C. & Wirtz, D. (2000). Multiple particle tracking measurements of the heterogeneities of solutions of actin ®laments and actin bundles. Biophys. J. 79, 1095-1106. 38. Ma, L., Xu, J., Coulombe, P. A. & Wirtz, D. (1999). Epidermal keratin suspensions have unique micromechanical properties. J. Biol. Chem. 274, 1914519151. 39. Coulombe, P. A., Bousquet, O., Ma, L., Yamada, S. & Wirtz, D. (2000). The ``ins'' and ``outs'' of intermediate ®lament organization. Trends Cell Biol. 10, 420-428. 40. Leduc, P., Haber, C., Bao, G. & Wirtz, D. (1999). Dynamics of individual ¯exible polymers in a shear ¯ow. Nature, 399, 564-566. 41. Wirtz, D., Berend, K. & Fuller, G. G. (1992). Electric®eld induced structure in polymer solutions near the critical point. Macromolecules, 25, 7234-7246. 42. Palmer, A. (1998). Dynamics and kinetics of actin networks in the presence of auxiliary proteins. PhD dissertation, The Johns Hopkins University, Baltimore. 43. Higgins, J. S. & Benoit, H. C. (1994). Polymers and Neutron Scattering, Oxford University Press, Oxford. 44. Sato, M., Leimbach, G., Schwarz, W. H. & Pollard, T. D. (1985). Mechanical properties of actin. J. Biol. Chem. 260, 8585-8592. 45. Shoenenberger, C.-A., Steinmetz, M. O., Stof¯er, D., Mandinova, A. & Aebi, U. (1999). Structure, assembly, and dynamics of actin ®laments in situ and in vitro. Microsc. Res. Tech. 47, 38-50. 46. Steinmetz, M. O., Goldie, K. N. & Aebi, U. (1997). A correlative analysis of actin ®lament assembly, structure, and dynamics. J. Cell Biol. 138, 559-574. 47. Gittes, F., Mickey, B., Nettleton, J. & Howard, J. (1993). Flexural rigidity of microtubules and actin ®laments measured from thermal ¯uctuations in shape. J. Cell Biol. 120, 923-934. 48. Ott, A., Magnasco, M., Simon, A. & Libchaber, A. (1993). Measurement of the persistence length of polymerized actin using ¯uorescence microscopy. Phys. Rev. E. Stat. Phys., Plasmas, Fluids, Rel. Interdisc. Top. 48, R1642-R1645. 49. Wachsstock, D., Schwarz, W. H. & Pollard, T. D. (1994). Crosslinker dynamics determine the mechanical properties of actin gels. Biophys. J. 66, 801-809. 50. Tempel, M., Isenberg, G. & Sackmann, E. (1996). Temperature-induced sol-gel transition and microgel formation in a-actinin cross-linked actin networks: a rheological study. Phys. Rev. E, 54, 1802-1808. 51. Xu, J., Wirtz, D. & Pollard, T. D. (1998). Dynamic cross-linking by a-actinin determines the mechanical

366

52.

53.

54.

55.

56. 57. 58.

59.

60.

Properties of F-actin/Fascin Networks

properties of actin ®lament networks. J. Biol. Chem. 273, 9570-9576. Sato, M., Schwarz, W. H. & Pollard, T. D. (1987). Dependence of the mechanical properties of actin/ a-actinin gels on deformation rate. Nature, 325, 828-830. Xu, J., Palmer, A. & Wirtz, D. (1998). Rheology and microrheology of semi¯exible polymer solutions: actin ®lament networks. Macromolecules, 31, 64866492. Murray, J. W., Edmonds, B. T., Liu, G. & Condeelis, J. (1996). Bundling of actin ®laments by elongation factor 1 alpha inhibits polymerization at ®lament ends. J. Cell Biol. 135, 1309-1321. Hinner, B., Tempel, M., Sackmann, E., Kroy, K. & Frey, E. (1998). Entanglement, elasticity, and viscous relaxation of actin solutions. Phys. Rev. Letters, 81, 2614-2617. Bryan, J. & Kane, R. E. (1978). Separation and activation of major components of sea-urchin actin gel. J. Mol. Biol. 125, 207-224. Kas, J., Strey, H. & Sackmann, E. (1994). Direct imaging of reptation for semi¯exible actin ®laments. Nature, 368, 226-229. Palmer, A., Xu, J. & Wirtz, D. (1998). Highfrequency rheology of crosslinked actin networks measured by diffusing wave spectroscopy. Rheologica Acta, 37, 97-108. Mullins, R. D., Heuser, J. A. & Pollard, T. D. (1998). The interaction of Arp2/3 complex with actin: nucleation, high af®nity pointed end capping, and formation of branching networks of ®laments. Proc. Natl Acad. Sci. USA, 95, 6181-6186. Svitkina, T. M. & Borisy, G. G. (1999). Arp2/3 complex and actin depolymerizing factor co®lin in den-

61.

62. 63.

64. 65. 66.

67.

68.

dritic organization and treadmilling of actin ®lament array in lamellipodia. J. Cell Biol. 145, 1009-1026. Blanchoin, L., Amann, K. J., Higgs, H. N., Marchand, J. B., Kaiser, D. A. & Pollard, T. D. (2000). Direct observation of dendritic actin ®lament networks nucleated by Arp2/3 complex and WASP/Scar proteins. Nature, 404, 1007-1011. Shao, Z., Shi, D. & Somlyo, A. V. (2000). Cryoatomic force microscopy of ®lamentous actin. Biophys. J. 78, 950-958. Mosialos, G., Yamashiro, S., Baughman, R. W., Matsudaira, P., Vara, L. F. & Matsumura, F. et al. (1994). Epstein-Barr virus infection induces expression in B lymphocytes of a novel gene encoding an evolutionarily conserved 55-kilodalton actinbundling protein. J. Virol. 68, 7320-7328. Borisy, G. G. & Svitkina, T. M. (2000). Actin machinery: pushing the envelope. Cur. Opin. Cell Biol. 12, 104-112. Mogilner, A. & Oster, G. (1996). Cell motility driven by actin polymerization. Biophys. J. 71, 3030-3045. Palmer, A., Cha, B. & Wirtz, D. (1998). Structure and dynamics of actin ®lament solutions in the presence of latrunculin A J. Polym. Sci. Physics. Ed. 36, 3007-3015. Tilney, L. G., Tilney, M. S. & Guild, G. M. (1995). Factin bundles in Drosophila bristles. I. Two ®lament cross-links are involved in bundling. J. Cell Biol. 130, 629-638. Tilney, L. G., Connelly, P. S., Vranich, K. A., Shaw, M. K. & Guild, G. M. (1998). Why are two different cross-linkers necessary for actin bundle formation in vivo and what does each cross-link contribute? J. Cell Biol. 143, 121-133.

Edited by W. Baumeister (Received 10 November 2000; received in revised form 6 April 2001; accepted 11 April 2001)