Vol. 258, No. 18, Issue of September 25, pp. 11004-11009,1983 Printed in U.5. A.

THE JOURNAL

OF BIOLOGICAL CHEMISTRY

Physical Basis of the Rheologic Properties of F-actin*

(Received for publication, December 14,1982)

Ken Scott ZanerS and Thomas P. Stossel

From the Hematology-Oncology Unit, MassachusettsGeneral Hospital, Department of Medicine, Harvard Medical School, Boston, Massachusetts 14 021

actin filaments result inan easily disrupted network (7, 13, 17). However, thixotropicmaterialsare usually composed of elongated molecules of concentration sufficient that their rotational domains overlap, topologically constraining their diffusion to motion along their length. Although these materials have the gross appearance of a gel, the polymeric strands are notnecessarily bound together by cross-linkages (5). The observed properties of these materials depend on the stress history and the time following the application of the stress. Although most accumuIated rheologic data are for flexible polymers, similar properties are exhibited by concentrated solutions of rod-like molecules (11,2 2 ) . A simple means of quantitating the properties of thixotropic polymer systems is to perform a creep experiment wherein a time-dependent strain is observed following the application of a constant stress. The time-dependent strain is proportionalto the compliance, conventionally designated J ( t ) , which transiently plateaus as thepolymers collide in response to stress, but at times longer than the molecular diffusion time will demonstrate Newtonian flow as the filaments diffuse past each other. If the stress is applied sinusoidally at frequency w, the compliance can be resolved into an elastic component, the dynamic storage modulus, G ( w ) , and a viscous component, the dynamic loss modulus, G ( w ) . The compliance under constantstress at aparticulartime will correspond to a dynamic modulus measurement at a particularfrequency. The The cortical cytoplasm of many cells is thought to contain dynamic moduli at a given frequency can be related mathethe machinery responsible for a varietyof cell functions. This matically to the compliance by a suitable Fourier transforarea of the cell is replete with actin filaments (F-actin) as mation. The dynamic moduli of polymer solutions sufficiently well as several proteins which can alter actin filament length dilute to permit free chain rotation generally change rapidly or induce a branched structure. The rigorous rheologic study with frequency, the loss modulus being higher than the storage of solutions of purified cytoskeletal proteins is a direct probe modulus at low frequencies. In contrast,when the concentraof their mechanical properties, and should contribute to the tion is high enough for topologic constraints, there is a freunderstanding of the role that theseproteins play in the quency range over which the dynamic moduli decrease slowly functions of the cell. In addition to its empiric value, these as the frequency falls, and thestorage modulus is greater than data are necessary to elucidate further the molecular mecha- the loss modulus. This frequency range, designated the planisms responsible for the mechanical properties of these sys- teau zone, reflects temporary network behavior. At much tems. lower frequencies, the material flows, as evidenced by the fact It has been observed that F-actin is thixotropic i e . a that the dynamic moduli decrease rapidly with a falling fresufficiently concentrated solution of F-actin appears solid, quency, and the loss modulus is greater than the storage but liquefies under the influence of applied stress and solidi- modulus. An additional rheologic parameter, the absolute fies again after removal of the stress. Moreover, the viscosity value of the complex dynamic viscosity, I q * ( w ) l , can be calcan be inversely proportional to the rate shear, from which culatedfrom the dynamic moduli and has been shown to of it has been inferred that the viscosity of F-actin would be closely approximate the shear rate-dependent viscosity, q ( + ) , infinite at a zero shear rate and that weak bonds between for a wide range of polymeric systems. Measurement of the absolute value of the complex dynamic viscosity is often * This work was supported by United States Public Health Service facilitated by the fact that the frequency is easier to vary as Grants HL-19429 and HL-00912, and a grant from the Council for an independent parameter than is the rate of shear. Tobacco Research. The costs of publication of this article were The development of theories of solutions of entangled flexdefrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 ible polymers has greatly facilitated theinterpretation of rheologic parameters. Recently, Jain and Cohen (12) have U.S.C. Section 1734 solely to indicate this fact. presented a theory of the rheologic properties of stiff rods in $ To whom correspondence should be sent.

The viscoelastic properties of purified rabbit skeletal muscle actinfilaments(F-actin)were measured at physiologic ionic strength and pH over a range of concentrations and filament lengths. Although F-actin demonstrated transitory elastic behavior, viscous flow was observed at longer times consistent with a high degree of filament overlap. The compliance was independent of stress over a 4-fold range, implying that the measurement did not disrupt any interfilament bonds. The dynamic storage modulus increased monotonically with frequency over the range measured, whereas the dynamic loss modulus had a relative minimum and was always less than the dynamic storage modulus. These observations are typical of topologically constrained behavior. The absolute value of the complex dynamic viscosity of F-actin, varied as the -0.8 power of the frequency and at a frequency of 0.1 radiansls was proportional to the product ofthe weight average filament length raised to the 0.7 power and the concentration. The experimental data agreed well with the predictions of a theory of the rheologic behavior of stiff rods in semidilute solutions. We conclude that the mechanical behavior of pure F-action solutions can be explained on the basis of the mutual topologic constraints to diffusion of long stiff rods which do not otherwise interact.

11004

Rheology of F-actin
concentrated solution, which should be applicable to the data obtained from solutions of the long stiff F-actin molecules. This theory is based on earlier work dealing with the physics of polymer solutions (l), and derives its predictions from the fact that long stiff molecules will interfere with each others free rotary diffusion (2, 3). In the present report, we have measured several rheologic parameters for F-actin at a number of concentrations and average lengths. We have compared the experimental data to values of the rheologic parameters calculated from the theory of Jain and Cohen, under the same experimental conditions. The experimental data, and their agreement with the theoretical predictions, strongly support the contention that the rheologic properties of F-actin solutions can ascribed solely be to the fact that the F-actin filaments will topologically constrain each others free diffusion because of their extreme length and extension.
MATERIALS ANDMETHODS

11005

Proteins-Actin purified by the method of Spudich and Watt (19) was used for most of the studies. In some experiments, the actin was subjected to anadditional chromatographic purification as previously described (18).All measurements were made no longer than 3 days after preparation of the actin. ABP and gelsolin were purified from rabbit lung macrophages by the method of Hartwig and Stossel (10) and of Yin and Stossel(23), respectively. Protein concentrationswere measured by the method of Lowry et al. (16). The lengths of the filaments were modulated by the addition of Ca.gelsolin complex, which has been shown to shorten F-actin in astoichiometric fashion (24). Actin filament lengths were estimated as previously described by determining the amount of ABP necessary to cause an abrupt increase in apparent viscosity, assumed to be gelation. The weight average length, L,, was calculated from (10): (actin)
Lw

sured at three different stresses,0.020,0.039, and 0.076 dynes/ cm2,were superimposable, demonstrating that all of the measurements were within the linear viscoelastic region. This result implies that the stress applied to the sample is not significantly perturbing the relative orientations of the filaments insolution. The curve demonstrates three regions, resembling thoseoften seen inconcentratedsolutions of flexible polymers. The first region arising at short times is a rapidly increasing compliance usually ascribed to be due to short range configurational changes. The second is a relative plateau at longer times, representing the temporary appearance of solidity. Although there is a relative flattening of the compliance, the slope is not zero, probably due to flow of the shorter filaments even at short times, a result which is distinctly different from that expected for a true solid which would have a slope of zero. The third is an upward deviation of the compliance at longer times due to long range diffusive movements of the molecules. Newtonian flow, which would be demonstrated by a slope of unity on a log-log plot, would only be observable at a time greaterthan therotary diffusion time which can be calculated from the expression stated in Appendix B to be about lo7 s for a filament length of 2.5 pm.

Lo#

Jlt)

...e+.*....
1 .

= (2 X 370 X ABP)
-06
I

where the concentration of actin and ABP are in moles/liter and the numerical constant of370 is the number of actin monomers per micrometer of filament length (9). Dynamic Rheologic Measurements-G, G, and * were measured I using the viscoelastometer previously described (25, 26). The instrument measures the rheologic properties of materials by observing the motion of a mica strip through a sample. The mica strip is suspended from a rod attached to a tilt table on which the sample is placed. Motion is induced by tilting the table to a fixed angle or in an oscillatory fashion to obtain respectively time-dependent or frequency-dependent data. After all components were combined, they were immediately placed in the viscoelastometer, and monitored by the measurement of G and G at 20-30-min intervals. Frequency-dependent measurements were begun when these values had reached a constant value, usually a t 3-3.5 h. Steady State Compliance-The method by which the time-dependent measurements were made is discussed in Appendix A (see miniprint). The compliance was calculated by means of Equation A4. The strain asfunction of time was entered intoan Apple I1 computer a with 64K memory and thecalculation was performed.
RESULTS

1.0
Log I

20

SD

FIG. 1. Time dependence of the compliance of F-actin at Log different initial stresses. of the compliance, J ( t ) ,in cm2/dyne is plotted againstthe log of the time for 1mg/ml F-actin in a solution containing 10 m imidazole-HC1,0.2 m CaC12, 0.2 m ATP, 0.1 M M M M KCI, 2 m M&lz, 0.2 m dithiothreitol, pH 7.5 at 25 C, at three M M different initial stresses: 0, 0.020 dyne/cm2; A, 0.039 dyne/cm2; 0.076 dyne/cm2.

.,

Compliance of F-actin-The log of the compliance of Factin as a function of the log of the time at a concentration of 1 mg/ml is shown in Fig. 1. The compliance curves, meaPortions of this paper (including Appendices A and B with Figs. A 1 and A2) are presented in miniprint a t the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No. 82-3346, cite authors, and include a check or money order for $3.60 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that available from Waverly is Press.

The abbreviation used is: ABP, actin-binding protein.

-I. 0

-I 5

-I 0

-0. 5

C 3

Log

FIG.2. Frequency dependence of the dynamic moduli for 1 mg/ml F-actin. The log of the dynamic storage ( G ) and loss (G) modulus in dynes/cm for 1 mg/ml F-actin, is plotted against the log of the frequency (radians/s). The weight average filament length as determined by the gel point method was 2.5 pm. Buffer conditions were the same as in Fig. 1.

11006

Rheology of F-actin
the gel-point method as described underMaterials and Methods. G shows a gradual monotonic decrease, with decreasing frequency whereas G shows a relative minimum. Dynamic Viscosity-The log of the absolute value of the dynamic viscosity, 1 v* I, plotted as a functionof the frequency for several concentrations and filament lengths is shown in Figs. 3 and 4. The curves are linear for all the samples measured witha slope of -0.8. The log of the dynamic viscosity at 0.1 radian/s is plotted againstthe log of the weight average length and is shown in Fig. 5. All of the data points are for 1 mg/ml actin, and include several actin preparations in addition to the data in Fig. 3. The plot is approximately linear with a slope of 0.7, indicating that 1q*( a L,.? at a

1.2

-1.5

- 1 .o

-0.5

Log w
FIG. 3. Frequency dependence of the absolute value of the complex dynamic viscosity for 1 mg/ml F-actin at a number of filament lengths. The log of the absolute value of the dynamic viscosity, lq*l (poise) for 1 mg/ml F-actin is plotted against the log of the frequency (radians/s). Buffer conditions were as in Fig. 1. The lengths were modulated by means of the addition of Ca.gelsolin complex and were measured by the gel point method as described under Materials and Methods. The lengths in micrometers are shown at the right

0.8

F 0)

0
4

0.4

Log L, (microns) FIG. 5. Weight average length dependence (A,) of the ab. solute value ofthe complex dynamic viscosity at 0 1 radian/ s. Log IT*[ a t 0.1 radianls is plotted against log for 1 mg/ml Factin obtained from several preparations: 0, data obtained from Fig. 3; data obtained from Fig. 4; A,column-purified and A, nonpurified actin (Ref. 26).

.,
1.6

! -

\ \

-J

1.0

1.2

j
I)

c \

0)

0
J

0.8

-1.5

- 1 .o

-0.5

Log

FIG. 4. Frequency dependence of the absolute value of the complex dynamic viscosity for F-actin a number of concenat trations. The log ofthe absolute value of the dynamic viscosity, Iq* I (poise) for F-actin is plotted againstthe log ofthe frequency (radians/ s) at 25 C .Buffer conditions were as in Fig. 1. The concentration in milligrams/ml is shown at theright.

0.4 -C

0.0

0.4

Dynamic Moduli-The log of the dynamic storage modulus, G, and dynamic loss modulus, G, as a function of the log of the frequency, for 1 mg/ml F-actin, are shown in Fig. 2. The weight average filament length of 2.5 km was determined by

Log c (mglml) FIG. 6. Concentration dependence of the absolute value of the complex dynamic viscosity. Log I?*I/LY0. F-actin at 0.1 for radian/s is plotted against log C. L, was determined for each preparation by the gel point method as described under Materials and Methods.

Rheology of F-actin

11007

in Appendix B. The theoretical value of the absolutevalue of In the complex dynamic viscosity is given by: I t ) * I a C.~L,.~. addition, a numerical calculation of the dynamic storage and loss modulus, shown in Fig. 7, closely approximates the experimental data, although the theoretical dynamicloss modulus is somewhat less than the experimental dynamic loss modulus. Fig. 8 shows a comparison between theoretical and experimental compliancevalues, with good agreement between them.
DISCUSSION

The rheologic properties of F-actin solutions presented are consistent with the conclusion that the relatively stiff F-actin filaments will impede each others free rotation a topologic by mechanism. These collisions leadto transient elastic behavior the degree of which depends on filament concentration and length. In the absencepermanent cross-linkages, increasing of rates of shear progressively align the molecules, thereby decreasing the viscosity. There is no need to postulate breaking a of interfilament bonds. This physical pictureisfurther supported by the fact that much of the rheologic data presented are consistent with the theoryof stiff chains in solution. -I 5 -I 0 -05 The compliance of F-actin continues to rise over the 30 Log w min of applied stress, rather than attaining constant value a FIG. 7. Theoretical values of the dynamic moduli. Log G which would be characteristic of a solid. Although there is a and G were calculated from Equations B6 and B7 as described under tendency for the strain to plateau at short times, this plateau Appendix B and are plotted against log w. It was assumed that the is followed byaconcave upward deviation.Areasonable concentration was 1 mg/ml and L, was 2.5 pm. The doshed c u m is explanation of these data is that, at times shorter than the the data from Fig. 2, presented for comparison. rotary diffusion time, the molecules impede each others motion and the material acts instantaneously as a solid. The elastic properties of the material are generated by the fact that a shearstress slightly rotatesthefilaments,storing energy whichis recoverable as the orientation the filaments of ,:andomize by their Brownian motion. At longer times, there is long range movement of the filaments and an approach toward flow, although Newtonian flow does not occur. This interpretation is further substantiated by the fact that the theoretical compliance curve shown in Fig. 8 for an F-actin concentration and length similar to that determined experimentally reasonably approximates theempirical values. As discussed in the Introduction, the fact that the storage loss modulus G over the modulus G is greater than the entire frequency range measured provides rheologic evidence THEOR. / THEOR. for the existence of filament constraints. In addition, the EXPTL. finding that the values of G decrease with decreasing frequency, and that G has a relative minimum suggests that the filaments are not permanently cross-linked. The theoret-I0 0 I.o 2.0 3 .o 40 ical curves reasonably approximate the experimental curves as shown in Fig. 7, although the theoretical G curve has a Log t lower minimum than that observed. This discrepancy is probFIG. 8. Theoretical values of the compliance. Log J ( t ) wae ably in part due to the partial flexibility (21) of the actin calculated from Equation B8 as described under Results, and is filament which contributes additional relaxation modes, thus plotted against log t, for the same concentration and L, as in Fig. 1. decreasing the theoretical minimums. The dashed curue is the compliance data from Fig. 1 replotted for The dynamic viscosity data presented in Figs. 3 and 4 shows comparison. that the lines are virtually parallel for all lengths and concentrations measured with a slope of about -0.8. This slope is fixed concentration. For this reason, the ratioof Iq*l to L,0.7 typical of the power law exponent which has been demonis plotted against concentration in Fig. 6 and an approxi- strated for numerous synthetic flexible polymers in concenmately linear curveis obtained witha slope of 1.0. These two tratedsolutions of uniform length ( 6 ) andhas alsobeen curves suggest that 1q*1 a cL,.~ F-actin for this range of previously demonstrated for F-actin (13, 17). The theoretical for frequency. treatments of concentrated solutions of stiff rods have preTheoretical Calculations-A theory of the rheologic prop- dicted a power law slope of about -1 (12). The somewhat erties of stiff rods in solution has been presented recently by lower slope obtained with the present data may result from (12). We have calculatedrheologic values for the slight flexibility and polydisperse filament lengths of FJain and Cohen conditions identical with those present in the experimental actin. The data presented in Figs. 5 and 6 imply the relationmeasurements. The details of this calculation are presented ship I v* I 0: cL,O. which is close to the theoretical relationship

11008

Rheology of F-actin

I q*( c1.4L,0.4 may be useful and for predicting i vivo agents shown to shorten actin n filaments. At longer lengths or properties of F-actin in cytoplasm. higher concentrations, F-actin can possibly exist in a situation Previous studiesof the rheologic properties of F-actin have similar to a glassy state (4), wherein the filaments are topobeen interpreted in terms of the classic theory of rubber logically constrained to such an extent that motion is virtually elasticity. This approach hasbeen motivated by the fact that impossible, and a veryhighrigidity is exhibited.A small F-actin solutionsat sufficient concentration appear to gels. quantity of actin-binding protein which causes right angle be Although it might be appropriate to apply the term gel in a branching of F-actin (10) could drastically reduce the free phenomenologic sense to describe an apparent elasticity for diffusion of actin filaments, thereby modulating its rheologic F-actin,as shown by the previousdiscussion, it does not properties. necessarily arise from a rubber-like state with interfilament Acknowledgments-We wish to acknowledge the useful discussions bonds. If such bonds exist, it is unclear by what mechanism factors which bind actin filaments together at their crossover with Dr. Jonathan Bennett and Dr. Paul Janmey. points might alter the viscoelastic properties of actin solutions REFERENCES under situations of low shear, since the number of points at 1. de Gennes, P. G . (1971) J . Chem. Phys. 5 5 , 572-579 which actin filaments can cross in solution is a function of 2. Doi, M., and Edwards, S. F. (1978) J. Chem. SOC. Trans. Faraday the amount protein present.If it is assumed that the spatial of Soc. ZZ 74,560-570 positions of the filaments canbe approximated by a tetrahe3. Doi, M., and Edwards, S. F. (1978) J. Chem. SOC.Trans. Faraday dral array, it can be easily calculated that there should be SOC. 74,918-932 ZZ 4. Edwards, S. F., and Evans, K. E. (1982) J. Chem. SOC. Faraday about 2.8 x 1013 crossover points/ml for a 1 mg/ml F-actin Trans. 78,113-121 solution. However, an exponential distribution of filaments Viscoehtic Properties of Polymers, John Wiley with an average length of 2.5 pm would have about 2.8 X 1 1 0 3 5. Ferry, J. D. (1980)York and Sons, New filamentends/ml.Thisresult implies that the number of 6. Graessley, W. (1974) Adu. Polymer Sci. 9 , 1-178 effective overlappoints, equal to the difference between the 7. Griffth, L. M., and Pollard, T. D. (1982) J. Biol. Chem. 2 5 7 , 9135-9142 total number of overlap points and the number of ends, is 8. Hansen, J., and Lowey, J. (1963) J. Mol. Bioi. 6,46-60 very small. Therefore, a small change in the average filament 9. Hartwig, J. H., and Stossel, T. P. (1979) J. Mol. Biol. 1 3 4 , 539length, increasing the number of free ends present should 553 cause a radical change in rheologic behavior. However, as is 10. Hartwig, J. H., and Stossel, T. P. (1981) J. Mol. Biol. 1 4 5 , 563case shown inFig. 3, this is not the whena quantity of gelsolin 581 sufficient to causea 5-fold decrease in length is present. 11. Helders, F. E., Ferry, J. D., Markovitz, H., and Zapas, L.S. (1956) J. Phys. Chem. 60,1575-1578 It could be argued that the applied stress as exertedFig. in 1 is breaking a system of weak interfilament bonds. If this 12. Jain, S., and Cohen, C. (1981) Macromolecules 1 4 , 759-765 were the case, the shape of the compliance curves shown in 13. Kasai, M., Kawashima, H., and Oosawa, F. (1960) J. Polymer Sei. 44,51-69 Fig. 1 should vary with increasing stress as more bonds are 14. Kawamura, M., and Maruyama, K. (1970) J. Biochem. (Tokyo) broken. However, identical compliance curves are obtaineda t 3,437-457 stresses which range over a factor of 4, suggesting that no 15. Kirkwood, J. G., and Auer, P. L. (1951) J. Chem. Phys. 19,281283 interfilament bonds havebeen disrupted. However, it may be H., possible that all of the bonds havebeen broken at thelowest 16. Lowry, 0. Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J. Biol. Chem. 1 9 3 , 265-275 applied stress, which is 0.058 dyne. The distance the probe 17. Maruyama, K., Kaibara, M., and Fukada, E. (1974) Biochim. cm so that the work done is moves at this stress is 7.8 X Biophys. Acta 371, 20-29 about 4.5 X ergs. Since there are of the order of 1013 18. Rees, M. K., and Young, M. (1967) J. Biol. Chem. 2 4 2 , 4449bonds in the sample and assuming all mechanical work was 4458 converted to disrupting chemical bonds, the energy of the 19. Spudich, J. A., and Watt, S. (1971) J. Biol. Chem. 2 4 6 , 48664871 interaction would be less than 0.01 kcal/mol. The chemical nature of a bond this weak and the role which it would play 20. Schwarzl, F. R., and Sturik, L. C. E. (1967-1968) Adu. Mol.Relax. Processes 1 , 201-255 in the rheologic properties of F-actin is unclear. 21. Takebayashi, T., Morita, Y., and Oosawa, F. (1977) Biochim. There areseveral implications of a model for F-actinbased Biophys. Acta 492,357-363 on topologic constraints to diffusion of stiff chains. An F- 22. Tschoegl, N. W., and Ferry, J. D. (1964) J. Phys. Chem. 68,867871 actinsolution will have a range of predictableproperties 23. Yin, H., and Stossel, T. P.(1979) Nature (Lord.)281,583-586 depending on the filament length, as has beenpreviously suggested (9,26), the concentration, and the time It will 24. Yin, H. L., Zaner, K. S., and Stossel, T. P. (1980) J. Bwl. Chem. scale. 255,9494-9500 be able to resistelastically a brief mechanical insult, thereby 25. Zaner, K. S., Fotland, R., and Stossel, T. P. (1981) Reu. Sci. Znstr. protecting its integrity, but slowly deform after sufficient will 52,85-87 time to allow for a shape change. These properties can be 26. Zaner, K. S., and Stossel, T. P. (1982) J. Cell Bid. 93,987-991 easily modulated by decreases in length caused by one of the 27. Zero, K. M., and Pecora, R. (1982) Macromolecules 15,87-93

Rheology of F-actin
Appendix A-Method of Ccmpliance W s u r e m e n t , Supplementary mtecial to .Physical Baais of the Rheologic Properties Of ?-actin., XS ZMer and TP stoLllle1.

11009

Appendix 8- Theoretical Calculations Of Rheologic Parameters, Supplementary material to 'Phyeical Basis of the Rheologlc Properties Of F-actin'. KS Z m e r and TP Stossel.
W have calculated e

The compliance, Jltl, Of a material is by definition:

J(t1

= T(tl/

on

(A1 1

where I(t1 is the strain exhibited by a mteri.1 i n response t o a fired 0. A stepped atrees Can be easily achieved stress 0 applied at t uaing theoviscoelastoleter by tilting the tabla to a m n s t a n t angle at tit 0. Emever ae the probe moves through the m t e r i a l the angle from vertical rill decrease, thus d e c r e a d n g the stress. This change in stress can be corrected for by a aequential application of the Boltzmann superposition principle which state. that the results of r c h a n i c a l hietory are simply addltive ( 5 ) . T o derive M a l g o r i t b for this mrrection we ILBSUIC that the total t i n following the eteppea (Itrose is divided into n equal s-nts of length At. POI simplicity re introduce the notation that for any function ? P Flnlt). The stcan. 0 ia applied at time that in the first of ti- the stress t 0 and it is s&dd changes by a negligible amount, so that:

the values of the absolute value Of the complex dynamic viscosity In in the power law zone, the frequency dependence of G' and G . , and tie t i k dependence of the compliance uaing as a starting point a theory recently developed for the solution properties of stiff rods (12). The principal q u a t i m a of this theory are:

where c is the mncentration of filaments per ml., w IS the frequency, I is the ofilaDent length k is the Boltamsnn constant, T IS the absolute temperature, d has the of 2 in the Original formulation. B is a con(lta"t, and A = ~ n ~ c : L ~ / 6 k ~ ln(L/d,,) where q s is the solvent viscosity T

kite

and d is the filament diameter. has the physical oignifmance of beingothe inverse of the rotary diffusion constant D Because of the experimental similarity in power law behavior of &tn to that of flexible polymers it i* assumed that the behavior of 1. 11 in the power law zone can be adequately described by its reduced form:

[MJ=
"0
"0

!R31

where TI ia the zero shear viscosity Y the c r i t x a l frequency for zero shear g h a v i o r , and k is the experimLnt8lly determined slope of 0.8. At frequencies below the critical frequency h u ( < 1 , and since
I ! l * I = m ) . + ( 6/W 7 it c a n be calculated from equations 81 and 82 that . 2 n-6,kT o.c,8 where the second term in the brackets of equation 81 ha8 been neglected because it ie much smaller than the first. Svbstltutinq into equation 8 3 :

I'ixl

= co

kBT A"'u-~

(Bi)

Substituting the expression for h and the fact that c=c L where c is the concentration in mg/ml the foliowznq expression is obeained at any particular frequency in the power law zone: IRSI

' I 1
"""---

......--

The denominator in equatlon 8 5 can be taken as constant to a good approximation , since it varies far less than the numerator over the ranqe of filament lengths consldered. It is of interest to Calculate the dynamic storage and loss moduli predicted by this d e l , which can be accomplished by Deans Of equations 81 and 8 2 taking into account the polydispersity of the filaments. The filament lengths were a B I u e d t o be exponentially distributed. and noreover it was a s a u n d thst fil-nte of less than about 0.5 microns ln length rould diffuse llthin the 'pores' of the a c t i n solution and behave as I f in dilute solution. Therefore a dilute solution contribution to the d u l l was included for filaments of less than 150 mnomers. In addition because of the fact that the experimentally determined power law slope was - 0 . 8 rather than -1.0 as predicted by the theory, the v a l u e Of d in equations 81 and 82 was chosen to be l . E rather than 2.0. The cause Of th19 deviation 11 unclear, but m y in w r t be due to the fact that F-actin fzlaments are 1 not completely stiff. The value8 far the number average aodull C' and G . are shown in Pig. 7 and yere calculated from equations 86 Lnd"i7. Thf. calculation consists of a number average integration of equations 81 and 8 2 with an additional dilute solution term:

' h 6
LOO
1

Figure 11. Denendence of the l q strain on loq t i e for 1009 qlycerolo The loq of the m v e m e n t of the probe of the via-lastmeter in arbitrary unite for 1008 91 cero1 is plotted against the log of the time at a t e m m r a t u r e of 25%. The dotted linea reoresente the maximum strain obthinable which corresponde to the probe'retuming to the midline.

where G ' I l . w l and G'll,wl are given by equations 81 and 82 G' ( 1 L ) and G ll.wl . are given by the well known dilute s o l u t m n expre;sio$s i151, N ' 1 s the number fraction Of filaments Of length 1 and a 19 150. For stmplicity a number average was used although a hrgher order average m y be m r e appropriate for a concentrated polydisperse solution. Equations 86 and 8 7 were integrated numerically by means Of Slmpsonr' rule. The constant E was choeen to be 64 BO that the theoretical cvrves nost closely approximated the experimental data. This value represents a somewhat greater .effective length' than has been prev~ously reported for synthetic stiff rods 112.271. . The compliance, J(t1, can be calculated f r m the dynamic noduli, if they are known over the complete range of frequencies. If they are not known over the entire range of frequencies the following formula is an excellent approximation (201;
~ ~~~~

J(t1

J'(AI

0 . 3 3 7 J":.iZia)

: .R)

where t=l/w and the expressions J'=G'/lG'2+G"21 and J"=G"/lG'2+G"21 are used. The calculated values are s h o w in Fig. 8 and a r e quite s i m i l a r both qvalitstloely and qUaotitatiYely to t h e data ~n Fig. 1. It is noted that Newtonian flow is not demonstrated by the Curve shown in flg. & 7 This polnt would be reached at a tme t=h which in thls case i s .boot seconds.

Flgyre A2. Compliance far 1008 glycerol- The log of the conplsance ~n m /dynes for 1 O O E glycerol 1s plotted against the log of the time in Seconds. The curve was Calculated from the data in Plg. 1 as oatlined in the Dethods section.