THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 255, No. 4, I sue of February 25, pp. 1670-1676.1980 Printed in U.S.A.

Spectrin/Actin Complex Isolated from Sheep Erythrocytes Accelerates Actin Polymerization by Simple Nucleation EVIDENCE FOR OLIGOMERIC ACTIN IN THE ERYTHROCYTE CYTOSKELETON*

(Received for publication, August 29, 1979)

Stephen L. BrennerS and Edward D. Korns From the +Physical Sciences Laboratory, Division of Computer Research and Technology, and the 9Laboratory of Cell Biology, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland 20205

A complex of spectrin and actin, isolated from sheep erythrocyte ghosts, accelerates the polymerization of actin in buffer containing 0.3 m~ MgC12. The rate of actin polymerization is similarly increased by soni- cated F-actin nuclei. At steady state, the critical con- centration of actin is lower when polymerization occurs in the presence of spectrin/actin complex than in its absence. Polymerization of actin in the presence of spectrin/actin complex is inhibited by substoichiome- tric concentrations of cytochalasin D, which is thought to block the net polymerizing ends of growing actin filaments (Brenner, S . L., and Korn, E. D. (1979) J. Biol. Chem 254, 9982-9985), in the same way that cytocha- lasin D inhibits actin polymerization in the absence of complex. However, actin filaments formed in the pres- ence of complex are more stable to cytochalasin D added after polymerization has occurred than are fila- ments formed in the absence of complex. We conclude from these data that spectrin/actin complex acceler- ates actin polymerization by simple nucleation and, therefore, that the complex consists of short actin oli- gomers cross-linked by spectrin tetramers (Brenner, S. L., and Korn, E. D. (1979) J. Biol. Chem 254,8620-8627) which stabilize the net depolymerizing ends of the actin oligomers. The isolated spectrin/actin complex is a fragment of the erythrocyte cytoskeleton. From our proposal for the structure of the complex and estimates of others for the relative amounts of actin, spectrin, and spectrin/actin complex-related high affinity cyto- chalasin binding sites in human erythrocyte ghosts, we suggest that the cytoskeletal network contains short oligomers of actin consisting on average of about ten subunits each, with one of every two actin subunits cross-linked by a spectrin tetramer to a subunit of another actin oligomer.

The isolated erythrocyte ghost consists of the plasma mem- brane and an underlying cytoskeleton that is thought to be the primary determinant of erythrocyte shape and deforma- bility in situ (1,2). The major constituents of the cytoskeleton are spectrin (bands 1 and 2, according to the nomenclature of Fairbanks et al. (3)), band 4.1, and actin (band 5) in the approximate molar ratio of 1:1:2 (4). Spectrin is a heterodimer of a 240,000-dalton polypeptide and a 220,000-dalton phospho- rylated polypeptide (3, 5) that further associates to a tetra- meric species (6) believed to be the native form of the protein in the erythrocyte (6, 7). Band 4.1 has a molecular weight of

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

about 78,000 and actin of about 42,000. The cytoskeleton is linked to the plasma membrane apparently through band 2.1, a protein of molecular weight about 210,000 that possesses binding sites both for spectrin (8,9), and an integral membrane protein that migrates electrophoretically with band 3 (10). Until very recently, however, there has been almost no evi- dence for the nature of the interactions among the three major cytoskeletal proteins, spectrin, band 4.1, and actin.

The first important step in this regard was the suggestion by Pinder et al. (11, 12), derived from experiments with partially purified low ionic strength extracts of human eryth- rocyte ghosts, that spectrin could induce the polymerization of G-actin to F-actin. After Ungewickell and Gratzer (6) and Ralston (13) showed that the spectrin in crude extracts of human erythrocyte ghosts can be separated by gel chroma- tography into three fractions, a spectrin/actin/band 4.1 com- plex, spectrin tetramer, and spectrin dimer, we found that only the spectridactin complex isolated from sheep erythro- cyte ghosts (this complex does not contain detectable levels of band 4.1) has the ability to accelerate the polymerization of actin (14). We found no evidence for the interaction of spectrin dimer or spectrin tetramer with monomeric G-actin. Spectrin dimer does bind to F-actin, however, and spectrin tetramer, irrespective of its state of phosphorylation, cross-links F-actin into highly viscous gels (14).

From these data, from the relative amounts of spectrin and actin in erythrocyte ghosts, and from the apparent absence of detectable actin filaments in ghosts (15), we proposed (14) that the erythrocyte cytoskeleton consists, in part, of short actin oligomers cross-linked by spectrin tetramers. On this assumption, the spectrin/actin complex isolated from the low ionic strength extract of sheep erythrocyte ghosts could be viewed as a partial fragment, missing band 4.1, of the eryth- rocyte cytoskeleton. Before considering how such a complex might affect the polymerization of actin in uztro, and what biological significance to attribute to this phenomenon, it is necessary to summarize briefly the properties of actin polym- erization as presently understood (16-18).

The rate-determining step in the polymerization of G-actin to F-actin is the condensation of several actin monomers to form an actin nucleus. Under polymerizing conditions, actin monomers probably associate and dissociate from both ends of the nucleus but with different rate constants at the two ends (18). Normally, under optimal polymerizing conditions, 1 molecule of ATP is hydrolyzed for each molecule of actin monomer added to the growing filament. Polymerization will cease, and the ratio of polymer to monomer will reach a steady state, when the net rate of addition of monomers equals the net rate of loss of monomers. When ATP is present, Wegner has shown (18) that each end of the filament will have its own critical concentration, i.e. the concentration of monomer in

1670

by guest on October 31, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Spectrin/Actin Complex and the Erythrocyte Cytoskeleton 1671

equilibrium with polymer. Therefore, neither end will be in equilibrium when the system as a whole is a t steady state and each filament will have a net polymerizing end and a net depolymerizing end. Similarly, polymerization will not occur, or pre-existing actin filaments will depolymerize, under ionic conditions in which the net rate of loss of actin monomers exceeds the net rate of addition of actin monomers. Further- more, during polymerization or depolymerization, conditions could exist in which net addition or net loss of monomers occurs at both ends of the actin filament simultaneously, i.e. the filament could have two net polymerizing ends or two net depolymerizing ends.

If the spectrin/actin complex were, in fact, a fragment of the erythrocyte cytoskeleton consisting of oligomeric actin cross-linked by spectrin tetramer, it would be expected to have the effect on actin polymerization fmst described by Pinder et al. (11, 12) by providing stable actin nuclei with accessible net polymerizing ends. The experiments described in this paper were designed to test this hypothesis. While they were in progress, a paper was published by Lin and Lin (19) which had an important influence on some of our experiments.

MATERIALS AND METHODS

Muscle actin was prepared from rabbit back and leg muscles by a modification (20) of the method of Spudich and Watt (21). Acantha- moeba profilin was prepared according to Reichstein and Korn (22). ATP and phenylmethanesulfonyl fluoride were purchased from Sigma, Tris from Bethesda Research Laboratories, reagents for poly- acrylamide gel electrophoresis from Bio-Rad, and cytochalasins B, D, and E from Aldrich. Stock solutions of 10 mM cytochalasins in dimethyl sulfoxide were stored at 4°C and diluted with dimethyl sulfoxide as required to maintain 1% dimethyl sulfoxide in the exper- imental solutions while varying the concentrations of the cytochala- sins. All other chemicals were at least reagent grade and deionized water was used.

Purified actin was quantified spectrophotometrically using extinc- tion coefficients a t 290 nm of 617 cm'/g and 638 cm'/g for G-actin and F-actin, respectively (23). Spectrin/actin complex and profiiin were quantified by the method of Lowry et al. (24) with bovine serum albumin (Armour) as standard. Dodecyl sulfate-polyacrylamide gel electrophoresis was carried out as described by Brenner and Korn (14). Samples for viscometry were mixed on ice and 0.5 ml was transferred to Cannon-Manning semi-microviscometers (size 100) a t 30°C and the flow times of the samples were measured. Specific viscosities were calculated as follows: (sample flow time - buffer flow time)/(buffer flow time). The buffer flow time was about 60 to 70 s.

Spectrin dimer and spectrin/actin complex were prepared by chro- matography of low ionic strength extracts of sheep erythrocyte ghosts on Sepharose 4B (Pharmacia) (14).' Spectrin tetramer was prepared from spectrin dimer (14). The pooled peak fractions, containing spectrin/actin complex a t a concentration of 0.05 to 0.1 mg/ml, were dialyzed a t 4°C against a buffer containing 5 mM Tris-chloride, pH 8.0, 0.2 mM ATP, 0.2 mM dithiothreitol, 0.1 mM CaCI2, and 0.01% NaN,, and stored on ice. The most active spectrin/actin complex (used in Figs. 1 and 6) was prepared by a modification of this procedure; 50 p~ ATP was added to the extract which was then dialyzed against a modified column buffer (14) containing 10% sucrose and 50 p~ ATP before application to the Sepharose 4B column, The isolated complex was dialyzed as before except for the presence of 10% sucrose in the dialysis buffer. Spectrin/actin complex prepared in buffers containing sucrose remained active for several weeks when stored on ice. In addition to the spectrin/actin complex, the void fraction from the Sepharose 4B column might have contained oligo- meric spectrin and oligomeric actin not associated with each other. However, native oligomeric actin capable of nucleation should have depolymerized in these buffers were it not stabilized in some way and native spectrin does not associate beyond the tetrameric state (6). Moreover, the molar ratio of spectrin dimer to actin in the void

' Due to a printer's error, the relevant figure (Fig. lA) was omitted from Ref. 14. The figure can be found in the "Additions and Correc- tions" section of J. Biol. Chem. (1979) 254, 12738 and corrected reprints are available from the authors.

fraction was 1:2, as it is in erythrocyte ghosts (4), suggesting that they are linked in a specific complex.

RESULTS

Effect of Spectrin/Actin Complex on Actin Polymeriza- tion-As measured by the increase in viscosity, polymeriza- tion of actin at 30°C is complete within 5 to 10 min when 2 mM MgCl, is added to G-actin in 5 mM Tris-chloride, pH 8.0, 0.2 mM ATP, 0.2 mM dithiothreitol, and 0.1 mM CaC12 (17). It requires 25 min to attain 50% polymerization in 0.5 mM MgCl, and 50 min for 50% polymerization in 0.4 mM MgC12, while in 0.3 mM MgC12, polymerization is usually less than 7% complete after 60 min (17). A few of our actin preparations polymerize less well and behave in 0.4 mM MgCl2 as most preparations do in 0.3 mM MgC12. Since all the solutions contain 0.2 mM ATP, the free M P concentration is about 0.2 mM less than the total Mg2+ concentration.

The data in Fig. 1 demonstrate that polymerization of actin in the presence of 0.3 mM MgC12 is greatly accelerated by addition of small amounts of spectrin/actin complex from sheep erythrocytes. In the presence of 72 pg/ml of complex, the rate and extent of polymerization of 20 p~ actin in 0.3 mM MgCL is essentially the same as is obtained in 2 mM MgC12 (Fig. 1). This is the same observation previously reported for the polymerization of actin in 10 m~ phosphate (11, 12, 14) or in 0.4 mM MgCI2 (19) and interpreted as induction of polym- erization in nonpolymerizing buffers. In fact, however, actin alone does polymerize, although slowly, in 0.3 mM MgC1, and, if polymerization is allowed to proceed to completion, similar steady state viscosities are obtained in the absence and in the presence of spectrin/actin complex (Table I).

The rate of actin polymerization depends on the concentra- tion of actin nuclei. Since the only compositional difference between spectrin/actin complex and spectrin dimer or tetra- mer, neither of which has any effect on the rate of polymeri-

1 0 , , I I I I

0 10 M 30 TIME, mm

FIG. 1. Effect of spectrin/actin complex on the rate of actin polymerization. Spectrin/actin complex was added to 20 ~ L M G-actin at 0°C in 5 mM Tris-chloride, 0.2 mM ATP, 0.2 mM dithiothreitol, 1% dimethyl sulfoxide, 0.1 mM CaCI2, 8.5% sucrose, pH 8.0. Polymeriza- tion was initiated by the addition of 0.3 mM MgCI2 and 0.5-1111 samples were transferred to viscometers a t 30°C. A single sample was used to measure the time course of polymerization for each concentration of complex. The spectrin/actin complex concentrations were: 72 pg/ml (0). 36 pg/ml (0),18 p g / d ( 4 , 9 pg/ml (m), 4.5 pg/ml (O), and 0 pg/ ml (A). For comparison, actin was also polymerized by the addition of 2 mM MgCL (0) in the absence of complex.

by guest on October 31, 2020

http://ww

w.jbc.org/

Dow

nloaded from

1672 Spectrin/Actin Complex and the Erythrocyte Cytoskeleton

zation of actin (14), is the presence of actin in the complex (Fig. 2), it seemed likely that spectrin/actin complex acceler- ates actin polymerization by providing a source of net polym- erizing ends to which actin monomers can add. As would be expected if this were the case, the effect of spectrin/actin complex can be duplicated by addition of sonicated F-actin nuclei (Fig. 3); 0.04 mg/ml(l PM) of sonicated F-actin produces essentially the same rate of polymerization of 20 PM G-actin as is obtained by 0.05 mg/ml of spectrin/actin complex and about the same rate of polymerization as is obtained by adding 2 mM MgCI2. The steady state viscosity in 0.4 m~ MgC12, however, is lower in the absence of spectrin/actin complex for reasons that are discussed later. Although they are qualita-

TABLE I Steady state viscosities of actin at low ionic strength in the

presence and absence of spectrin/actin complex Either spectrin/actin complex (79 pg/ml), 0.3 mM MgCI?, or both,

were added to 20 PM C-actin a t 0°C in 5 mM Tris-chloride, 0.2 mM ATP, 0.2 mM dithiothreitol, 0.1 mM CaC12, 0.01% NaN:,, 1% dimethyl sulfoxide, 7.8% sucrose, pH 8.0. Samples were incubated at 30°C for 19 h and transferred to viscometers for measurement of their steady state viscosities. Each number is the average of at least three meas- urements on the same sample.

Additions Specific ViscOsiKy

Actin + complex Complex + MgCl? Actin + MgCI? Actin + comdex + MeCL

0.02 0.02 0.77 0.80

1 2

5

COMPLEX S PECTR IN FIG. 2. Dodecyl sulfate-polyacrylamide gel electrophoresis

of spectrin/actin complex and spectrin tetramer. Spectrin/actin complex (15 pg) and spectrin tetramer (10 pg) were applied to 7.5% polyacrylamide slab gels. Spectrin (bands 1 and 2) and actin (band 5) are labeled. Complex was prepared as in Ref. 14. The molar ratio of spectrin dimer to actin monomer in the complex is 1:2 by scans of the negative.

1

0 10 20

TIME,rnin

FIG. 3. Effect of sonicated F-actin nuclei on actin polymeri- zation. Sonicated F-actin nuclei were added to 20 p~ G-actin at 0°C immediately after addition of 0.4 mM MgCL to buffer containing 5 mM Tris-chloride, 0.2 mM ATP, 0.2 mM dithiothreitol. 0.1 mM CaClp, pH 8.0. Samples of 0.5 ml were rapidly transferred to viscometers a t 30°C and their flow times were measured. The concentrations of added nuclei were: 1 PM (5% of total actin) (0) and 0.2 p~ (1% of total actin) (0). Controls are shown in which sonicated buffer (A), 54 pg/ ml of spectrin/actin complex in 0.4 mM MgCl? (O), or 2 mM MgCl? (A) were added to the G-actin. F-actin was prepared by incubating 20 pM G-actin overnight at 30°C in the above buffer containing 0.5 mM MgCln and 0.01% NaN,. Aliquots of 0.5 ml were sonicated immediately before use for 20 s at setting 4 on a Heat Systems Ultrasonic model W185 sonifier. Samples of sonicated F-actin nuclei (a maximum of 30 pl) were rapidly added to the C-actin (570 pl) on ice.

tively similar, spectrin/actin complex is much more effective than sonicated F-actin nuclei/mol of actin, since the complex is only about 10% actin by mass. This probably means that the average length of actin oligomers in the spectrin/actin complex is sufficiently less than the average length of soni- cated F-actin nuclei that approximately 10 times more of the latter must be added to obtain an equivalent concentration of net polymerizing filament ends. A still greater concentration of sonicated F-actin is required when the F-actin is polymer- ized and sonicated in 2 m~ MgCI2 (data not shown), rather than in 0.5 mM MgCL, presumably either because F-actin is more stable or because fragmented F-actin reanneals more rapidly, thus providing fewer filament ends, in the higher concentration of MgCI2.

Effect of Spectrin/Actin Complex on the Critical Concen- tration for Actin Polymerization-In the context of the po- lymerization model discussed above, the critical concentration at the net polymerizing end of the actin filament must, by definition, be lower than the critical concentration at the net depolymerizing end. The experimentally determined critical concentration will be between the two (18). If spectrin/actin complex does consist of short actin oligomers cross-linked and stabilized by spectrin tetramers, the critical actin concentra- tion should be different for actin polymerized in the presence and absence of complex. Spectrin/actin complex would not be expected to affect the steady state rate constants for the on- off reactions at the net polymerizing ends of actin filaments, because they will be distant from the complex, but the spectrin cross-links should greatly affect the rate constants, and effec- tively block the release of monomers, a t the spectrin-blocked

by guest on October 31, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Spectrin/Actin Complex and the Erythrocyte Cytoskeleton 1673

polymerization of actin in the absence of spectrin/actin com- plex just as in its presence. These data led to the hypothesis (17) that cytochalasins act primarily by blocking the net polymerizing ends of actin nuclei and filaments. One demon- strable consequence of that interaction is that addition of cytochalasin D to actin filaments, in the absence of spectrin/

ACTIN,/tM

FIG. 4. Effect of spectrin/actin complex on the critical con- centration for actin polymerization. Actin was polymerized for 19 h at 30°C in 5 mM Tris-chloride, 0.2 mM ATP, 0.1 mM CaC12, 0.2 mM dithiothreitol, 0.01% NaNa 1% dimethyl sulfoxide, 0.4 m~ MgClz, and 7.8% sucrose, pH 8.0, in the presence (0) or absence (0) of 78 pg/ml of spectrin/actin complex. Samples of 0.5 ml were transferred to viscometers at 30°C and their viscosities were measured. The inter- cept at zero specific viscosity is the critical actin concentration for polymerization.

ends of filaments at which net depolymerization would oth- erwise have occurred. Complex should reduce the critical concentration at the net depolymerizing ends and, therefore, the measured critical concentration for actin polymerization should be lower in the presence of spectrin/actin complex than in its absence. In the experiment described in Fig. 4, the presence of spectrin/actin complex did reduce the critical concentration (the actin concentration at zero specific viscos- ity) from the control value of 4.5 pM to about 0.2 pM. If d of the filaments at steady state in the presence of complex were associated with complex, the measured critical concentration of 0.2 ~ L M would be the critical concentration, i.e. the reciprocal of the equilibrium constant for association, for the net poly- merizing ends of actin filaments.

Effect of Cytochalasins on the Polymerization of Actin in the Presence and Absence of Spectrin/Actin Complex- While the experiments described in this paper were in prog- ress, Lin and Lin (19) discovered that the high affinity cyto- chalasin binding sites of human erythrocytes that are not involved in sugar transport are associated with the spectrin/ actin/band 4.1 complex and that cytochalasins inhibit the ability of that complex to accelerate the polymerization of actin. In complete agreement with Lin and Lin (19), we find (Fig. 5) that substoichiometric concentrations of cytochalasin D (0.2 p ~ ) inhibit the polymerization of actin (20 p ~ ) in the presence of sheep erythrocyte spectrin/actin complex (55 pg/ ml). As reported by Lin and Lin (19), cytochalasin B is only about one-tenth as effective an inhibitor of actin polymeriza- tion as is cytochalasin D (Fig. 5).

From the facts that the cytochalasin-inhibited complex from sheep erythrocytes contains only spectrin and actin (Fig. 2), that pure spectrin does not bind cytochalasins (19) nor detectably affects actin polymerization (14), and that the effects of complex on actin polymerization can be duplicated by sonicated F-actin nuclei (Fig. 3), we deduced that the cytochalasins probably interact specifically with stabilized actin oligomers in the spectrin/actin complex. Cytochalasins, therefore, should also effectively inhibit actin polymerization in the absence of complex. We were able to show (17) that substoichiometric concentrations of cytochalasin D inhibit the

0.8 I I I

0

TIME.min

FIG. 5. Effects of cytochalasins on the rate of actin polym- erization in the presence of spectrin/actin complex. Cytocha- lasin was added to solutions of 20 PM G-actin and 55 pg/ml of spectrin/ actin complex in 5 mM Tris-chloride, 0.2 m~ ATP, 0.1 mM CaCI2, 0.3 mM MgC12, 0.2 mM dithiothreitol, pH 8.0, at 0°C. All solutions contained 1% dimethyl sulfoxide, the solvent for the cytochalasins. Samples of 0.5 ml were transferred to viscometers at 30°C and the time course of polymerization was followed. Samples contained cy- tochalasin D at 0.2 p~ (=), 0.02 PM (A), or 0.002 PM (0); cytochalasin B at 0.2 PM (A) or just dimethyl sulfoxide but no cytochalasin (0).

0.8

0.6

v) 0 V E ' 0.4 u LL

a

c

s v) ' --C-"+-"""w,-r

0.2

0 0 10 20 30

FIG. 6. Effect of cytochalasin D on actin polymerized in the presence of spectrin/actin complex. Seven separate solutions of 20 p~ G-actin and 36 pg/ml of spectrin/actin complex in 5 m~ Tris- chloride, 0.2 m~ ATP, 0.2 mM dithiothreitol, 0.1 mM CaCIZ, 8.5% sucrose, pH 8.0, were stored on ice. Cytochalasin D was added at 0.2 p~ to one sample (0) and all the samples were made 0.3 m~ in MgCle. The samples containing cytochalasin D (0) and a control sample without cytochalasin but with dimethyl sulfoxide (0) were transferred to viscometers a t 30°C and their viscosities were measured with time. The five other samples were incubated at 30°C and cytochalasin D at 0.2 PM was added at 2 min (A), 5 min (A), 10 min (0). and 15 min (W). Dimethyl sulfoxide was added to another control at 5 min (0). Immediately after addition of cytochalasin D or dimethyl sulfoxide, these samples were transferred to viscometers and their viscosities were followed with time.

TIME, rnm

by guest on October 31, 2020

http://ww

w.jbc.org/

Dow

nloaded from

1674 Spectrin/Actin Complex and the Erythrocyte Cytoskeleton

actin complex, caused a rapid decrease in viscosity to the same low viscosity that was attained when cytochalasin D has been present from the beginning of the polymerization (17). This would result from cytochalasin blocking the net polymerizing ends of the growing actin filaments while allowing the net depolymerizing ends to reach their independent equilibrium with the pool of monomeric actin.

If spectrin/actin complex does, as we propose, block the net depolymerizing ends of growing actin fiaments, we would expect a different result when cytochalasin D is added to actin filaments polymerizing in the presence of complex than was obtained in the absence of complex. Cytochalasin should, as before, inhibit further filament growth but filaments attached to complex should not depolymerize.

Experimentally, addition of 0.2 p~ cytochalasin D, a con- centration that strongly inhibits actin polymerization, a t var- ious times after initiation of actin polymerization in the pres- ence of spectrin/actin complex, immediately stops further polymerization (Fig. 6 ) as predicted. Morever, the viscosity does not fall to the very low level observed when cytochalasin D is present from the beginning of the incubation, as it does when the same experiment is carried out in the absence of complex (17). The small decrease in viscosity that occurs, especially when cytochalasin is added near the end of polym- erization, is probably due to the presence of actin filaments that are not attached to complex. Such unattached actin fiaments could arise either from slow self-nucleation of po- lymerization, as occurs under these conditions in the absence of complex (Figs. 1, 3, 4, 7, and Table I), or from dissociation of filaments from complex subsequent to their formation.

Effect of Profilin on the Polymerization of Actin in the Presence of Spectrin/Actin Complex-Profiiin is a low molec- ular weight protein that inhibits actin polymerization (22,25) by forming a 1:l complex with monomeric actin. Under the conditions of polymerization used in this paper, spectrin/actin

I I f I f

TIME, min

FIG. 7. Effect of profilin on the polymerization of actin in the presence of spectrin/actin complex. Acanthamoeba profiiin (400 VM in 5 mM potassium phosphate, 0.5 mM dithiothreitol, pH 7.5) was mixed with an equal volume of 100 p~ G-actin in 5 mM Tris- chloride, 0.2 mM ATP, 0.2 mM dithiothreitol, 0.1 m~ CaCL, pH 8.0, and incubated at 25°C for 10 min. A control mixture without profiiin was also prepared. The two samples were then chilled on ice and spectrin/actin complex and MgCL were added to give the following concentrations; 80 pM profiiin, 20 pM G-actin, 54 pg/ml of spectrin/ actin complex in 4 mM Tris-chloride, 4.6% sucrose, 0.2 mM ATP, 0.1 mM CaCl?, 0.3 mM dithiothreitol, 0.4 m~ MgCI2, 1% dimethyl sulfox- ide, pH 8.0. Samples were rapidly transferred to viscometers at 3OoC and their viscosities were followed with time. Results are shown for four mixtures: actin (e), actin + complex (=I, actin + profiiin (A), actin + complex + profiin (0).

complex largely counteracts the inhibition by profiin (Fig. 7) just as actin nuclei were previously shown to do (22). This is additional evidence for the qualitative similarity of the polym- erization processes in the presence and absence of complex.

DISCUSSION

We have demonstrated in this paper that spectrin/actin complex isolated from sheep erythrocyte ghosts accelerates the rate of actin polymerization and lowers the steady state critical actin concentration under conditions in which actin alone will also polymerize, albeit more slowly than in the presence of complex. The kinetic effect of spectrin/actin com- plex is mimicked by sonicated F-actin nuclei. Cytochalasins inhibit complex-accelerated actin polymerization and interact with actin fiiaments formed in the presence of complex in a manner generally predictable from their effects on actin po- lymerization in the absence of complex (17). The results reported in this paper were all obtained in 5 mM Tris-chloride containing 0.3 or 0.4 mM MgC12, a buffer similar to that used by Lin and Lin (19), but identical results have also been obtained in 10 mM phosphate, the buffer used by Pinder et al. (11, 12).

We believe, on the basis of these observations, that there is no need to attribute any unique properties to the spectrin/ actin complex. The data are consistent with the simple pro- posal that the spectrin/actin complex consists of short actin oligomers cross-linked by spectrin tetramers (14). Complex would thus provide stabilized actin nuclei with accessible net polymerizing ends and blocked net depolymerizing ends. In fact, under the conditions of the isolation of complex, both ends of the actin oligomers might be net depolymerizing ends and the oligomers may survive only because depolymerization is blocked at both ends by cross-linking spectrin tetramers. Cytochalasin would inhibit complex-accelerated actin polym- erization, as it does actin polymerization in the absence of complex, by blocking net polymerizing ends under otherwise polymerizing conditions. Little depolymerization of actin fiia- ments occurs when cytochalasin is added to fiiaments formed in the presence of complex because, even at steady state, the net depolymerizing ends of many of the actin filaments are blocked by their association with spectrin/actin complex. This model for the structure of the spectrin/actin complex, and the effect of complex and the cytochalasins on actin polymeriza- tion, is schematically illustrated in Fig. 8.

The isolated spectrin/actin complex is most simply consid- ered as a partial fragment of the erythrocyte cytoskeleton. The possible size of actin oligomers in the erythrocyte can then be estimated from the above model and the further assumption that each actin oligomer has one free end acces- sible to cytochalasins. Steck (4) has estimated that each human erythrocyte ghost contains about 3.5 X lo5 actin mon- omers and Lin and Lin (19) have estimated that the spectrin/ actin/band 4.1 complex accounts for about 3 to 4 X lo4 high affinity cytochalasin binding sites in each erythrocyte ghost. On this basis, the average actin oligomer would contain about 10 actin monomers and be only about 26 nm long (26) , if the actin oligomer is a typical double helical actin structure. The actin oligomer, therefore, would be considerably shorter than the maximum length of just one spectrin tetramer, which is estimated to be about 200 nm long (27). From the reported molar ratio of spectrin to actin in human erythrocyte ghosts ( 4 , approximately one of every two monomers in each actin oligomer would be cross-linked by a spectrin tetramer to a subunit of another actin oligomer in a continuous cytoskeletal network (Fig. 9, top). This model is compatible with the continuous, web-like reticulum of filaments 5 to 40 nm in diameter and 20 to 100 nm in length revealed by scanning

by guest on October 31, 2020

http://ww

w.jbc.org/

Dow

nloaded from

Spectrin/Actin Complex and the Erythrocyte Cytoskeleton 1675

Spectrin Tetramer Actin Oligomer

+ Cytochalasin

I + G - a c t i n n

FIG. 8. Schematic representation of the proposed structure of spectrin/actin complex and the mechanism of its interaction with actin monomers and cytochalasin. In this model, the com- plex consists of short actin oligomers cross-linked by spectrin tetra- mers. Spectrin tetramers stabilize the actin oligomers by blocking depolymerizing ends. Under appropriate conditions, the actin oligo- mers provide net polymerizing ends for filament growth. It is proposed that cytochalasins block filament growth by blocking net polymerizing ends of actin oligomers or filaments. If the net depolymerizing ends were not blocked by spectrin/actin complex, the filaments would depolymerize completely when monomer addition was blocked by cytochalasin.

actin oligornel

band 3.

spectrin tetramer actin oligomer

FIG. 9. Schematic representation of the proposed structure of the spectrin/actin cytoskeleton and its association with the erythrocyte plasma membrane. Top, an en face view of the cytoskeleton parallel to the plane of the membrane. Actin oligomers of about 10 subunits are interconnected by spectrin tetramers. Ap- proximately 50% of the actin subunits are involved in such linkages. The arrangement of the linkages is arbitrary with some spectrin tetramers crossing over others. The actin oligomers will be about 26 nm long and the spectrin tetramer maximally about 200 nm long. The location of band 4.1 is not known and it is, therefore, not included in the model. Possibly it stabilizes the structure by interconnecting spectrin and actin. Bottom, cross-sectional view of the cytoskeleton showing its possible linkages to the plasma membrane. Each spectrin tetramer is linked through band 2.1 to the integral membrane protein that is one of the components of band 3. Bands 2.1 and 3 may be present as dimers (10). Band 4.2 may also be involved in this inter- action but it is not included in the model. The evidence for these models is developed more fully in the text.

electron microscopy of the cytoplasmic surface of lysed eryth- rocyte membranes (28); the narrower filaments could be actin oligomers and the wider filaments might be spectrin tetramers.

In addition to spectrin and actin, the complexes isolated from human erythrocytes by Lin and Lin (19) and Cohen and Branton (29) also contain band 4.1. Although band 4.1 is apparently not necessary for the acceleration of actin polym- erization by the complex, nor for the inhibition of that effect by cytochalasins, band 4.1 is probably an important structural element of the erythrocyte cytoskeleton. Possibly, band 4.1, which is present in approximately molar equivalence to spec-

trin (4), provides stability to the cytoskeleton by cross-linking spectrin to actin near the sites of their interaction. Another protein, band 4.2, that is believed to interact with this system (30), may be associated (8, 10) with the linkage of the cyto- skeleton to the plasma membrane through band 2.1 (Fig. 9, bottom).

Finally, we attribute no particular biological significance to the ability of isolated spectrin/actin complex to accelerate the polymerization of actin. Rather, it seems to be just a fortuitous consequence of the structure of the erythrocyte cytoskeleton from which the complex is derived. Cohen and Branton (29) have reached a similar conclusion from their observation that the orientation of the newly polymerized actin filaments as- sociated with complex in vitro is the opposite of that generally found for membrane-associated filaments in uiuo. However, it should be remembered that, in practice as well as in theory, conditions can be found in which either or both (31) ends of an actin filament can be net polymerizing ends. Nonetheless we would expect that any complex of actin filaments cross- linked and stabilized by any actin-binding protein, irrespective of its possible membrane association or function in situ would, when isolated from cells, have properties similar to those of the spectrin/actin complex isolated from erythrocytes, i.e. the ability to bind cytochalasins and to nucleate cytochalasin- inhibitable actin polymerization.

Addendum-After this paper was accepted for publication, a paper appeared (32) with evidence that band 4.1 may be required to form stable complexes between spectrin tetramer and F-actin. That is not incompatible with the data in this paper and Ref. 14 for direct interaction between spectrin and actin and agrees with our specula- tion that spectrin/actin interactions in situ might be stabilized by cross-links of band 4.1.

Acknowledgments-We thank Dr. Lois Greene for critically read- ing the manuscript and for suggesting experiments that have materi- ally improved the paper and Dr. Stephen C. Mockrin for assistance in the experiments with profilin.

REFERENCES

5.

6.

7.

8.

9.

10. 11.

12.

13. 14.

15. 16.

17.

1. Pinder, J. C., Ungewickell, E., Bray, D., and Gratzer, W. B. (1978)

2. Ralston, G. B. (1978) Trends Biochem. Sei. 3, 195-198 3. Fairbanks, G., Steck, T. L., and Wallach, D. F. H. (1971) Bio-

4. Steck, T. L. (1974) J. Cell Biol. 62, 1-19

J. Supramol. Struct. 8,439-445

chemistry 10,2606-2617

Clarke, M. (1971) Biochem. Biophys. Res. Commun. 45, 1063- 1070

Ungewickell, E., and Gratzer, W. B. (1978) Eur. J. Biochem. 88,

Ralston, G., Dunbar. J., and White, M. (1977) Biochim. Biophys.

Luna, E. J., Kidd, G. H., and Branton, D. (1979) J. Biol. Chem.

Bennett, V., and Stenbuck, P. J. (19791 J. B i d Chem. 254,2533-

Bennett, V., and Stenbuck, P. J. (1979) Nature 280, 468-473 Pinder, J. C., Bray, D., and Gratzer, W. B. (1975) Nature 258,

Pinder, J. C., Bray, D., and Gratzer, W. B. (1977) Nature 270,

Ralston, G. B. (1978) J. Supramol. Struct. 8, 361-373 Brenner, S. L., and Korn, E. D. (1979) J. Biol. Chem. 254,8620-

Tilney, L. G., and Detmers, P. (1975) J. Cell BioL 66, 508-520 Oosawa, F., and Asakura, S. (1975) Thermodynamics of the

Brenner, S. L., and Korn, E. D. (1979) J. Biol. Chem. 254,9982-

379-385

Acta 491, 345-348

254,2526-2532

2541

765-766

752-754

8627

Polymerization of Proteins, Academic Press, New York

9985 18. Wegner, A. (1976) J. Mol. Biol. 108, 139-150 19. Lin, D. C., and Lin, S. (1979) Proc. Natl. Acad. Sei. U. S. A. 76,

2345-2349

by guest on October 31, 2020

http://ww

w.jbc.org/

Dow

nloaded from

1676 Spectrin/Actin Complex and the Erythrocyte Cytoskeleton

20. Eisenberg, E., and Kielley, W. W. (1974) J. Biol. Chem. 249,

21. Spudich, J. A,, and Watt, S. (1971) J. Biol. Chem. 246,4866-4871 22. Reichstein, E., and Korn, E. D. (1979) J. Biol. Chem. 254, 6174-

23. Gordon, D. J., Boyer, J. L., and Korn, E. D. (1977) J. Biol. Chem.

24. Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J.

25. Markey, F., Lindberg, U., and Eriksson, L. (1978) FEBS Lett. 88,

26. Huxley, H. E., and Brown, W. (1967) J. Mol. Bwl. 30, 383-434

4742-4748

61 79

252,8300-8309

(1951) J. Biol. Chem. 193,265-275

75-79

27. Shotten, D., Burke, B., and Branton, D. (1978) Biochim. Biophys.

28. Hainfeld, J. F., and Steck, T. L. (1977) J. Supramol. Struct. 6,

29. Cohen, C. M., and Branton, D. (1979) Nature 279, 163-165 30. Yu, J., Fischman, D. A., and Steck, T. L. (1973) J. Supramol.

31. Woodrum, D. T., Rich, S. A,, and Pollard, T. D. (1975) J. Cell

32. Ungewickell, E., Bennett, P. M., Calvert, R., Ohanian, V., and

Acta 536,313-317

301-311

Struct. 1, 233-248

Biol. 67,231-237

Gratzer, W. B. (1979) Nature 280,811-814

by guest on October 31, 2020

http://ww

w.jbc.org/

Dow

nloaded from

S L Brenner and E D Kornerythrocyte cytoskeleton.

polymerization by simple nucleation. Evidence for oligomeric actin in the Spectrin/actin complex isolated from sheep erythrocytes accelerates actin

1980, 255:1670-1676.J. Biol. Chem. 

  http://www.jbc.org/content/255/4/1670.citation

Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/255/4/1670.citation.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on October 31, 2020

http://ww

w.jbc.org/

Dow

nloaded from