Val. 267, NO.
THEJOURNALOF RIOLOCICALCHEMISTRY
62 1992 by The American Society for Biochemistry and Molecular Biology, Inc.
Issue of October 5, PP. 20339-20345,1992
Printed in U.S A .
Nucleotide Exchange, Structure, and Mechanical Properties
of
Filaments Assembled from ATP-actin and ADP-actin*
(Received for publication, April 6, 1992)
Thomas D. Pollard$, Ilya Goldberg, and William H.
Schwarz$
From the Departmentof Cell Biology and Anatomy, The Johns HopkinsMedical School, Baltimore, Maryland 21205 and the
21218
§Department of Chemical Engineering, The Johns Hopkins University, Baltimore, Maryland
The adenine nucleotide bound in the central cleft of the
actin molecule stabilizes the protein, but in spite of extensive
research (for review see Korn et al., 1987; Pollard, 1990; and
Carlier, 1991) the underlying function of the nucleotide and
its hydrolysis in the assembly of actin filaments in live cells
has remained an enigma for years (for review see Cooper,
1991). It is well established that when actin with bound ATP
(referred to here as ATP-actin)polymerizes, the ATP on each
molecule is hydrolyzed (Strauband Feuer, 1950) afterits
incorporation into a filament (Pollard and Weeds, 1984) in
an irreversible reaction (Carlier et al., 1988) to ADP and
inorganic phosphate (Pi). The hydrolysis reaction may occur
randomly on the polymerized subunits (Pollard and Weeds,
1984) or may be favored at the interface between a central
region consisting of ADP-actin and newly added ATP-actin
subunits near the ends of the filament (Carlier et al., 1987).
The Pi remains bound to the filament for some time (Carlier
and Pantaloni, 1986),making rapidly growing filamentsa
heterogeneous mixture of subunits with ATP, ADP-P,, or
ADP. Eventually the Pi dissociates leaving filaments composed largely of ADP-actin with a few ATP- and ADP-Piactin subunits near the ends.
* This work was supported by National Institutes of Health Research Grant GM-26338. 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.
$ T o whom correspondenceshould be addressed Dept. of Cell
Biology and Anatomy, The Johns Hopkins Medical School, 725 N.
Wolfe St., Baltimore, MD 21205. Tel.: 410-955-5664; Fax: 410.9554129.
The nucleotide composition of monomers has an easily
measured impact on the elongation of actin filaments (for
review see Korn et al., 1987). At the barbed end, we find by
direct electron microscopic measurements that ATP-actin
binds with a diffusion-limited rate constantof lo7M" s-l and
dissociates slowly, about 1 s" (Pollard, 1986; Drenckhahn
and Pollard, 1986). (Similar values of these and other rate
constants have been measured by other methods; for review
see Korn et al., 1987.) In contrast, ADP-actin binds about
20% as fast and dissociates about 5 times faster than ATPactin. ADP-P-actin has a critical concentration similar to
ATP-actin (Rickard and Sheterline, 1988; Wanger and Wegner, 1987; Carlier and Pantaloni, 1988), but there arenot yet
any direct measurements of the elongation rate constants.
The effect of nucleotide on growth at thepointed end is less
clear. Both ATP- andADP-actin associate at rates lower than
expected for a diffusion-limited reaction (Pollard, 1986;
Drenckhahn and Pollard, 1986). Both dissociate slowly, especially ADP-actin. The ratios of the association and dissociation rate constantsgive the critical actin monomer concentrations for elongation. In ATP with physiological concentrations of divalent cations, the critical concentration at the
pointed end is substantially higher than at the barbed end.
This should lead to a slow flux of subunitsthrough the
filaments from the barbed end to the pointed end driven by
ATP hydrolysis.
These properties suggest that nucleotide hydrolysis might
be used by cells to power subunit flux through actin filaments
and possibly also to prepare the filaments for rapid disassembly at a time different from the time of assembly. In live cells
actin subunits flux rapidly through filaments at the leading
edge (Wang, 1985; Forscher and Smith,1988), and many actin
filaments turn over rapidly (Amato and Taylor, 1986; Theriot
and Mitchison, 1991). Neither the rapid flux nor turnover
appears to be compatible with the relatively slow dynamics of
purified actin filaments i n vitro, so we do not yet understand
how the polymer dynamics in vivo are related to nucleotide
hydrolysis (for reviewsee Cooper, 1991). Excluded volume
effects (Drenckhahn and Pollard, 1986) or accessory proteins
could speed up subunit exchange in the cell.
Recent observations of Janmey et al. (1990), showing a
possible effect of nucleotide hydrolysis on actin filament
structure, mechanical properties, and nucleotide exchange,
suggested a particularlyexciting new hypothesis regarding the
function of the nucleotide. Janmey et al. presented evidence
that filaments assembled from ADP-actin differed from filaments assembled from ATP-actin in several ways eventhough
both consist of ADP-actin subunits. Their ADP-actin filaments were irregular in contourand slightly larger in diameter
thanATP-actin filaments. Nevertheless, solutions of the
ADP-filaments had a much higher dynamic elastic modulus
than ATP-actin filaments, a difference that the authors at-
20339
Downloaded from www.jbc.org at Yale University on April 7, 2009
Actin monomers with bound ATP, ADP, or fluorescent analogues of these nucleotides exchange the nucleotide on a second time scale, whereas filaments assembled from each of these species exchange their nucleotide with the solution at least 1,000 times slower
than monomers. Filaments assembled from either
ATP-actin or ADP-actin are indistinguishable by electron microscopy after negative staining. The dynamic
elasticity and viscosity of filaments assembled from
ATP-actin or ADP-actin or mixtures of these two species are the same over a wide range of frequencies.
These observations do not support a recent suggestion
(Janmey, P. A., Hvidt, S., Oster, G. F., Lamb, J., Stossel, T. P., and Hartwig, J. H. (1990) Nature 347, 9599) that ATP hydrolysis within actinfilaments stiffens
the polymer and alters both their structure and affinity
for nucleotides. The difference in observations between these two studies may be related to time-dependent changes in ADP-actin prepared in slightlydifferent
ways.
20340
Nucleotides
Actin
1983) and a two-step mechanism
EXPERIMENTALPROCEDURES
Protein Preparation-Actin was purified from rabbit skeletal muscle and gel filtered in Buffer G on Sephadex G-150 (MacLean-Fletcher
and Pollard, 1980). Buffer G consists of 2 mM Tris-C1, pH 8.0, at
25 "c,0.2 mM ATP, 0.5 mM dithiothreitol, 0.1 mM CaC12, 1 mM
sodium azide. Mg-ADP-actin was made by the method of Pollard
(1986) using soluble hexokinase or, for the nucleotide exchange experiments, by a modification using hexokinase-agarose beads as follows. One milliliter of actin at a concentration of 30-50 yM was mixed
gently with 50 pl of Dowex 1 (Bio-Rad AG-X2 washed in 2 mM Tris,
p H 8.0) for 5 min at 4 "C. The Dowex binds free nucleotides. The
Dowex was removed by centrifugation at 14,000 X g for 30 s, and 852
p1 of nucleotide-free actin monomers from the supernatant was added
t o a microcentrifuge containing 100 pl of packed hexokinase-agarose
beads (Sigma H-2005; washed with 2 mM Tris-C1, pH 8.0, 0.5 mM
dithiothreitol; final concentration 2 units/ml), 5 or 10 pl of 100 mM
ADP (final concentration 0.5 or 1 mM), 8 p1 of 10 mM MgCl, (final
concentration 80 PM), 20 p1 of 10 mM EGTA' (final concentration
200 yM), and 10 pl of 100 mM glucose (final concentration 1 mM).
The sample was mixed gently for 4 h at 0-4 "C. After pelleting the
hexokinase-agarose beads, the Mg-ADP-actin in the supernatantwas
stored on ice and used the same day. Etheno-ADP (Molecular Probes,
Inc.) was exchanged onto the actin in the same way using a final
concentration of etheno-ADP of 0.5 mM. Etheno-ADP-actin is not
stable for more than a few hours at 0 "C based on a progressive loss
in theamplitude of the signal when ATP is exchanged for the ethenoADP. Etheno-ATP was exchanged onto the actin in the same way
with the omission of the hexokinase beads using a finalconcentration
of 0.5 mM etheno-ATP.
Actin-Nucleotide Exchange Experiments-The binding or dissociation of etheno-ATP or etheno-ADP was measured at 25 "C by fluorescence (Waechterand Engel, 1975) with a Perkin-Elmer-Cetus
Instruments 640-A spectrofluorometerusing an excitation wavelength of 365 nm and emission wavelength of 410 nm. The fluorescence of the etheno-nucleotides is higher when bound to actin than
when free. The relationship of the fluorescence intensitytothe
absolute extent of a reaction was established as follows. First, we
measured the fluorescence change at the end of each reaction as a
function of the actin monomer concentration (as in Fig. 1). The
fluorescence intensity was always directly proportional to the actin
monomerconcentration, Second, from the initial fluorescence in
experiments such as Fig. 2 0 with mixtures of filaments and monomers, we established that thefluorescence of etheno-ADP is the same
bound to monomeric and polymeric actin. (A similar comparison is
not possible for etheno-ATP, since it is hydrolyzed during polymeri-
eN
+ A + eN*A
N+AeNA
where e N is etheno-ATP or etheno-ADP, A is actin, eN* is the high
fluorescence form of the nucleotide analog bound to actin, and N is
ATP or ADP. The simulations were initiated by adding the second
nucleotide to anequilibrium mixture of actin and thefirst nucleotide.
The association rate constantswere assumed to be 10 PM-' s" (Nowak
et al., 1988). We searched by trial and error for the single set of
dissociation rate constants, one for each of the four nucleotides, that
allowed the kinetic simulations to fit simultaneously the full-time
courses of the fluorescence changes in all four types experiments and
all actin concentrations in Fig. 1. Initially the ratios of these dissociation rate constants were constrained by the ratios of the corresponding Kd values as determined above but had to be adjusted
slightly during the simulations to achieve a good fit to all of the data.
The optimal rate constants are given in the legend of Fig. 1, and
simulations are illustrated in Fig. 1 as continuous curves.
Actin Polymerization-Some of the actin was modified with pyrenyl-iodoacetamide(MolecularProbes, Inc.) and used to measure
actin polymerization (Pollard, 1984). The critical concentration for
polymerization was determined from the dependence of the steadystate fluorescence or the initial rate of nucleated actin polymerization
on the concentration of actin (Pollard, 1984). Having establishedthat
actin filaments do not exchange bound nucleotide at an appreciable
rate, we also used the extent of fluorescent nucleotide exchange as a
function of the concentration of actin as a new method to measure
the critical concentration of monomers in the presence of filaments.
Electron Microscopy-Actin filaments on glow-discharged carbon
films were negatively stained with 1% uranyl acetate (Cooper and
Pollard, 1982). Four grids
each of ATP-actin andADP-actin filaments
were coded and randomized before one author took micrographs at
46,000 X of one area of eight grid squares with well spread and evenly
stained filaments. Coded prints were scored for straight filaments by
two blinded observers.
Rheological Measurements-The dynamic viscosity and elasticity
were measured in a cone and plate rheometer with small amplitude
oscillations as described by Sato et al. (1985). ADP-actin was made
by the method of Pollard (1986). Samples were degassed to avoid
bubbles during the long incubations. Actin stocks were warmed to
25 "C and diluted with ADP or ATP buffer and concentrated salts to
give final concentrations of 34 p M actin, 50 mM KCl, 1 mMMgC12, 1
mM EGTA, 0.2 mM nucleotide, 10 mM imidazole, pH 7.0. Samples
were immediately transferred to the
rheometer and incubated at 25 "C
The abbreviation used is: EGTA, [ethylenebis(oxyethyleneni- for 8-12 h to attain an equilibrium configuration before measurements were made of the dynamic viscosity and elasticity.
tri1o)jtetraacetic acid.
Downloaded from www.jbc.org at Yale University on April 7, 2009
tributed to the greaterflexibility of the ADP-actin filaments, zation.) Third, we titrated 4 PM etheno-ATP-actin or etheno-ADPwhich caused them to be more entangled than the stiff rod- actin with ATP. From the titration data we calculated the fluoreslikeATP-actin
filaments. TheADP-actinfilaments
ex- cence of the etheno-nucleotides when free and when bound to actin.
changed their bound ADP faster than the ADPof filaments For etheno-ATP binding to actin increased the fluorescence 16 fold;
for etheno-ADP binding increased the fluorescence 18-fold. From the
assembled from ATP-actin. The addition of ATP stiffened dependence
of the equilibrium fluorescence on the concentration of
the ADP-actin filaments andgave them properties more like ATP at each point in the titration, we calculated the relative values
ATP-actinfilaments. It was postulated that hydrolysis of of the dissociation constants for the etheno-nucleotides and ATP
ATP within an actin filament "traps . . . the monomers . . . (similar to themethod of Waechter and Engel, 1975). The ratios were
conformationally and stores elastic energy" that might be Kd ATP/Kd etheno-ATP = 0.3 and Kd ATP/Kd etheno-ADP = 0.015.
"available for releaseby actin-binding proteins that transduceFrom the nucleotide concentrations and thesteady-state fluorescence
at the end of the experiments in (Fig. 1C) we calculated the ratio of
force or sever actin filaments."
Kd ADP/Kd etheno-ATP to be 1.5. These relative equilibrium conSince this is the most interesting suggestion in years restants and thenucleotide concentrations were used to scale all of the
garding the function of the bound nucleotide of actin, we fluorescence data.
thought that it is important to confirm the
key observations
For exchange experiments with polymerized actin, we first deterof Janmey et al. (1990). In this paper we report new data on mined the critical concentration for the actin species being tested
nucleotideexchange,
filamentstructure,andmechanical
(see below) and then prepared the samples for exchange experiments
properties of actin assembled from ADP-actin or ATP-actin. by mixing concentrated monomer and polymer stocks to give the
We prepared the ADP-actinquickly in the cold by a modifi- criticalconcentration of monomer and variableconcentrations of
cation of current methods. Using this ADP-actin and ATP- polymer. This avoided prolonged incubations that otherwise would
been required to allow either monomers or polymers to attain
actin, we found noevidence for rapidexchange of any nucleo- have
steady state after dilution into the polymerizing buffer.
tide between polymerized actin and the solution and
were
Kinetic Simulation Methods-Rate constants for the exchange of
unable to detect
significant differences in theoverall structure nucleotides on actin monomers were determined by kinetic simulation
or mechanical properties of filaments assembled from ATP- using an Apple Macintosh version (HOPKINSIM, Daniel Wachsactin or ADP-actin.
stock, Johns Hopkins Medical School) of KINSIM (Barshop et al.,
Nucleotides
Actin
20341
Although the exchange of actin-bound nucleotide is relatively
slow in calcium-containing solutions (Gershman etal.,
Preparation of Actin-Nucleotide Complexes-Actin with
1989),
ATP, ADP, etheno-ATP,and etheno-ADPall exbound Mg2+ and ADP or etheno-ADP prepared by exchange
in a low ionic strength buffer containing M e , EGTA, glu- change on a second time scale in buffers with M e and EGTA
cose, and hexokinase had a critical concentration10-20 times (Fig. 1).Under these conditions, M$+ is bound to the high
et al., 1986) in the
higher than the corresponding ATP-actins (Table I). The affinity divalent cation site (Gershman
cleft
of
the
actin
molecule
(Kabsch
et
al.,
1990). The time
hexokinase acts asa scavenger for any ATP dissociated from
course
is
essentially
the
same
when
the
Mg-EGTA
buffer has
the actin or synthesized from ADP by contaminating adenylate kinase (Pollard, 1986; Gershman et al., 1989). Soluble either a low ionic strength favoring depolymerization of the
hexokinase was left in the ADP-actin preparations used for actin or high enough concentrations ofKC1 and M e to
rheological studies to remove any ATP generated during the strongly favor polymerization of the actin molecules.
Simulation of the time course of these exchange reactions
equilibration and measurements.
allowed
us toselect dissociation rate constantsfor each of the
To measure the exchange of ADP bound to the actin for
ATP from the buffer, it was necessary to remove the hexoki- four nucleotides (Fig. 1 legend) that together account the
nase from the actin. This was accomplished by a new method kinetic data from four different types of nucleotide exchange
using hexokinase immobilized on agarose beads. The hexoki- experiments (Fig. 1).We used the two-step mechanism exnase activity that could be achieved in the exchange reaction plained under“Experimental Procedures.” The simulation
us to model the reactions without simpliusing the commercial preparation of hexokinase-agarose was approach allowed
fying assumptions about irreversibility used in earlier analytabout 10 times lower than theconcentration of soluble hexoical solutions (Waechter and Engel, 1975). Although the nukinase known to be effective in producing ATP-free actin in
cleotide exchange rate constants are very sensitive to experiprevious studies (Pollard, 1986). To reduce the need for high
mental conditions (Frieden and Patane, 1988), our complete
hexokinase activity, free ATP was removed from the starting
set of values for the rate constants is in general agreement
sample of actin before adding ADP andthe immobilized
with literature values obtained for subsets of these constants
enzyme. In steady-state criticalconcentrationexperiments
under somewhat differentconditions(Kuehl
and Gergely,
the resulting Mg-ADP-actin has the same critical concentra1969; Waechter and Engel, 1975; Nowak et al., 1988).
tion that made with soluble hexokinase (Table I). Thus the
Nucleotide Exchange by PolymerizedActin-Samples
of
two preparations of Mg-ADP-actin have the same ratio of
polymerized actin assembled from monomers with bound
dissociation and association rate constants for polymer elonATP, ADP, etheno-ATP, or etheno-ADP exchanged a small
gation. It is formally possible that theabsolute values of these fraction of their nucleotide on a second time scale, but the
rate constants may differ for the two preparations. Since the
rate of exchange by most of the actin in these samples was
critical concentration is very sensitive to low concentrations too slow to detect in
20 min (Fig. 2). This confirms for all
of ATP-actin in ADP-actin (Ohm and Wegner, 1991), these
nucleotides previous observations onfilaments assembled
results are strong evidence that the hexokinase-agarose effi- from ATP-actin (Martonosi et al., 1960).
ciently removes all of the ATP from the ADP-actin preparaThe time course for the exchangeable fraction of nucleotide
tion like soluble hexokinase (Pollard, 1984).
in each of these polymerized actin samples was similar to
Nucleotide Exchange on Actin Monomers-We first carried samples of monomer without filaments, and theamplitude of
out a comprehensive set of nucleotide exchange experiments the signal was equal to or slightly less than that predicted by
with actin monomers (Fig. 1) to establisha quantitative the critical concentration of monomers measured independmechanism that would allowus to interpretexperiments with ently by steady-state or kinetic methods. Beyond this signal
more complex mixtures containing both monomers and fila- arising from exchange of nucleotide by the monomers, there
ments. Others have investigated nucleotide exchange by actin was no additional reproducible signal attributable to nucleomonomers (see for example Kuehl and Gergely, 1969; tide exchange by the polymeric actin over the time course of
Waechter and Engel, 1975; Frieden and Patane, 1988; Nowak our experiments. In the experiments with ATP-actin plus
et al., 1988), but not for all of the relevant nucleotides and etheno-ATP, we detected no exchange of nucleotide by polynot in a buffer than approximates physiological ionic condi- mer concentrations up to 50 times the critical concentration
tions.
of monomers, where exchange, even at a low rate, would have
been obvious (Fig. 2 A ) . The signal to noise ratio is even better
TABLE
I
with etheno-ATP-actin reacted with ATP. In these experiCritical concentrationsfor actin polymerization
ments, polymer concentrations 60 times higher than the critConditions: 50 mM KCI, 1 mM MgCl,, 1 mM EGTA, 0.2 mM
ical concentration did not exchange over a period of 20 min
nucleotide, 0.5 mM dithiothreitol, 10 mM imidazole (pH 7.0). The
underlined values were obtained with samples of actin prepared with (Fig. 2B). In the experiments with ADP-actin filaments the
hexokinase-agarose beads. The other ADP-actin samples were pre- critical concentration is much higher, so the highest polymer
pared with soluble hexokinase. The samples used for elongation rate concentration tested was six times the monomer concentraand steady-state polymer concentration measurements contained 5% tion. Nevertheless, no nucleotide exchange of the polymerized
pyrene-labeled actin.
actin was detected over times greater than 10 half-times for
exchange of nucleotide on the monomers in the same sample
Method
ATP-actin
ADP-actin
(Fig. 2C). Similarresults were obtained with filaments of
PM
PM
PM
etheno-ADP-actin where the signal to noise ratio was even
Elongation rate
0.04
1.35
better (Fig. 2 0 ) .
0.90
0.05
Given the sensitivity of these assays, the half-time for the
0.05
exchange
of nucleotide on the subunits of actin filaments is
Steady-state polymer
0.10
1.20
concentration
0.05
more than 1,000 times greater than the half-times for the
1.15
1.35
exchange of the same nucleotide bound to actin monomers.
Nucleotide exchange
1.0
This conclusion holds true for filaments assembled from ATP0.9
actin, ADP-actin, etheno-ATP-actin, or etheno-ADP actin.
RESULTS
Downloaded from www.jbc.org at Yale University on April 7, 2009
Ethe::iy-
20342
Actin Nucleotides
1 .o
4
TA+eT==eTA+T
A
o'8
0.6
t1
J
1000
500
0
1500
2000
3
a 4
I-
C
DA + eT==eTA + D ,
a
200
D
400
600
800
eDA + T == TA +eD
2
1
1
~
~
100
"
"
200
"
"
300
"
'
400
0
500
600
0
50
100
150
200
Seconds
FIG. 1. Time course of nucleotide exchange by actin monomers. Experimental data are given as open symbols and dashed line.
Theoretical kinetic curves shown as continuous lines were calculated using KINSIM and the following rate constants: k , = 10 pL"' s-' for all
nucleotide binding reactions; k- = 0.0045 s" for ATP, 0.015 s-' for etheno-ATP, 0.023 s-l for ADP, and 0.09 s" for etheno-ADP. Conditions:
2 mM Tris-C1, pH 7.5; 500 pM dithiothreitol; 200 pM EGTA, 80 p~ MgCI,. All samples contained nucleotides and etheno-adeninenucleotides
as indicated and low concentrations of CaCl, that varied with the actin concentrations in the range of 3-10 p ~ Panel
.
A , exchange of ATP
bound to actin for etheno-ATP. ATP-actin concentrations were 0.2 p M monomers plus 1 F M filaments (0)
and 0.5 p~ monomers (0).The
final ATP concentrations were 0.8 and 2 p ~ The
.
etheno-ATP concentration was 200 p ~ Panel
.
B , exchange of etheno-ATP bound to actin
for ATP. The etheno-ATP-actinmonomer concentrations were 1 p~ (0),
2 /IM( O ) ,and 3 p~ (A). The final etheno-ATP concentrations were
17, 34, and 51 p ~ The
.
ATP concentration was 1,000 PM. Panel C , exchange of ADP hound to actin for etheno-ATP. ADP-actin monomer
concentrations were 1 F M (O),2 pM (El),and 3 FM (A). The final ADP concentrations were 31,62, and 93 p ~ The
.
etheno-ATPconcentration
was 200 p ~ Under
.
these conditions there was a significant concentration of ADP-actin in equilibrium with etheno-ATP-actin and the free
nucleotides at the end of the reaction. Punel D, exchange of etheno-ADP bound to actin for ATP. The etheno-ADP-actin monomer
concentrations were 1 p~ (O),2 p~ (El),
and 3 p~ (A). The final etheno-ADP concentrations were 17,34, and51 p M . The ATP concentration
was 1,000 pM.
Electron Microscopy of Actin Filaments Assembled from
DISCUSSION
ATP- or ADP-Monomers-Although there is alwayssome
In a physiological buffer containing M e , KCl, and EGTA,
heterogeneity in the appearance of actin filaments prepared
a
single
set of rate constants can account
for the timecourses
for electron microscopy by negative staining, we could not
detect any systematic differences in the gross appearance of of a wide range of nucleotide exchange experiments (Fig. I),
values are reliable. All are
actin filaments assembled from ATP-actin (Fig. 3, A-C) or so we are confident that these
ADP-actin (Fig. 3, D-F). In particular, all of the grids con- based on a reasonable assumption that the association rate
s-', a value citedby Nowak et al. (1988)
tained filaments with smooth, gently curving contours. Most constants are 10 p"'
from unpublishedwork.
ofthe grids alsocontained someirregular filaments,but
Four adeninenucleotides exchanged on a second time scale
blinded observers could not distinguish two populations of
of the from actin monomers, but we detected very little exchange on
grids made with ATP-actin or ADP-actin. The quality
stain varied from one grid square to the next but was not actin filaments over much longerperiods of time even in
related to the nucleotide. One of four grids with ADP-actin experiments with high concentrations of filaments (Fig. 2).
had considerable lateral aggregation of the filaments, but this Conservatively, we estimate that the
nucleotide bound to actin
was not characteristic of the other ADP grids. These obser- filaments polymerized from either ATP- or ADP-actin exvations rule out gross differences in filaments of ATP-actin changes 1,000 times slower than ATP or ADP on actin monand ADP-actin but do not preclude subtle differences in the omers. This exchange rate for subunits within filamentsis an
packing or shape of the subunits that might be detected by upper limit, since some nucleotide exchanges into polymers
high resolution reconstructions.
Mechanical Properties of Actin Filaments Assembled from indirectly because of nucleotide exchange on monomers that
off the endsof filaments. Theslow rate
ATP- or ADP-Monomers-We detected no reproducible dif- then exchange on and
of
nucleotide
exchange
by polymerized actinisconsistent
ferences in the viscous or elastic modulii of actin filaments
assembled from ATP-actin or ADP-actin over a wide range with the atomic model of the actin filament (Holmes et al.,
long pitch helix appear
of frequencies (Fig. 4). Filaments assembled from mixturesof 1990) wheresubunit contacts along the
well
ATP- and ADP-actin (made by converting most but not all to block the exitof the nucleotide from its binding site as
as constraining the type of interdomain motions that open
of the buffer ATP to ADP) had similar mechanical properties.
Downloaded from www.jbc.org at Yale University on April 7, 2009
2
0
0
3
3 t
0
eTA+T==TA+eT
B
20343
Actin Nucleotides
TA+eT==eTA+T
0.6
"*-
-
0.2 uM monomer
+1 uM filaments
0.1 uM monomer
+1 uM filaments
"0"+2 uM filaments
+3 uM filaments
-
+3 uM filaments
0.4
0.2
E
o
loo0
500
1500
2000
0.0
0
200
. ,
1.6
D
DA+eT==eTA+D
400
.
,
600
,
,
1000
800
,
.
.
eDA+T==TA+eD
0.8
0.4
1 uM monomer
o . o ~ " ' " ' ~ " r
0.0
0
100
200
300
400
0
500
100
200
300
400
500
.
these conditions there was a significant concentration of ADP-actin in equilibrium with etheno-ATPwere 31, 62, 93, and 124 p ~ Under
actin and thefree nucleotides at the endof the reaction, accounting for the incomplete exchange of ADP. Panel D,exchange of etheno-ADP
bound to actin for ATP. All samples contained 1.0 p~ etheno-ADP-actin monomers and 1,000 p~ ATP. The etheno-ADP-actin filament
2 p M (O), and 3 pM (A). The final etheno-ADP concentrationswere 17, 34, 51, and 68 p ~ .
concentrations were 0 (bold line), 1 p~ (0),
the nucleotide binding site in hexokinase.
Our conclusions are infull agreement with the long accepted
belief that ADP is tightly bound to the subunits in actin
filaments (Martonosiet al., 1960) but differ from two previous
studies of nucleotide exchange by actin filaments using etheno-ATP fluorescence. Wang and Taylor (1981) followed the
exchange of etheno-ATP for ATP and the reverse reaction
with 5 p~ actin in a polymerization buffer with 2 mM MgC12
& 100 mMKC1.
They observed two phases of fluorescence
change. Initially, fluorescence changed rapidly with an amplitude of about 20% (-1 p M actin) and half-time of 5-10 min
( k 0.001 s-l). They interpreted thisfirst phase as therapid
exchange of nucleotide on monomers and addition of some of
these monomers to theends of filaments. They did not measure the critical concentration, but the amplitude of their fast
phase was about 10 times larger than expected from later
determinations under similar conditions
(Table I).We cannot
explain this difference from our experiments. The rate of the
second phase of the fluorescence change was variable. In 100
mMKC1 with 50 pM MgCI2 the rate was near zero. In 2 mM
MgC12 the maximum rate was about 0.1 nM s" and was
attributed to the flux of subunits throughfilaments. Our
calculation of their subunit exchange rate (0.1 nM subunits/
s/2 nM filamentends = 0.05 subunits/filament end/s) is
consistent with the rate of subunit flux (about 0.1 s-l) calculated from the elongation rate constants (Pollard, 1986), so it
is not necessary to postulate rapid exchange of nucleotide by
subunits within the polymers. These conclusions are consistent with ours.
Janmey et al. (1990) reported that solutions of ADP-actin
filaments exchange bound nucleotide more rapidly than ATPactin filaments. They capped the barbed end of the filaments
-
to eliminate the slow flux of subunits observed by Wang and
Taylor (1981). From the the partial
time courses of nucleotide
exchange in polymeric actin reported by Janmey et al. we
estimate first-order rate constantsof 0.005 s-l for ADP-actin
and 0.001 s" for ATP-actin.Their signal came from the
binding of etheno-ATP, butthese values should approximate
the dissociation rate constantsfor ADP and ATP. Both values
are about &fold lower than our dissociation rate constants
for ADP and ATP from monomeric actin, a difference that is
expected because of the Ca2+in their buffer which slows the
rate of nucleotide dissociation (Frieden and Patane, 1988).
We suggest that their signal arose in large part from the
critical concentration of monomers in their filament samples,
as in our experiments in Fig. 2. The reason is that Janmeyet
al. calibrated their fluorescence signal by assuming that 100%
of the nucleotide in samples of filaments exchanged in 20 h.
They calculated the fluorescence increment per molecule of
actin-bound nucleotide from the totalactin concentrationand
the fluorescence change at that time. Twenty hours is probably inadequate for complete equilibration of nucleotide with
filaments capped on their rapidly exchanging ends by gelsolin.
Consequently, they may have underestimated the magnitude
of the fluorescence change. This explains why their experiments appear to show extensive exchange of nucleotide on
polymerized actin on a minute time scale. The presence of
gelsolin may also account for the relatively small difference
in the amplitudes of exchange that they observed for ATPactin andADP-actin. Without gelsolin the fluorescence signal
arising from the monomers present ina filament sample is 10
times higher for ADP-actin than ATP-actin (Fig. 2) because
of the large difference in their critical concentrations at the
barbed end. However, with the barbed ends capped by gelsolin,
Downloaded from www.jbc.org at Yale University on April 7, 2009
Seconds
FIG. 2. Time course of nucleotide exchange by mixtures containing a critical concentration of actin monomers and a range
of concentrations of actin filaments. Note that therewas no exchange by the filaments above that expected from the critical concentrations
of actin monomers. Panel A , exchange of ATP bound to actin for etheno-ATP. All samples contained 0.2 p~ ATP-actin monomers and 200
f i etheno-ATP.
~
ATP-actin filament concentrations were 1p M (0)
and 3 pM (A).The final ATP concentrations were 7.2 and 20.4 p ~ Panel
.
B , exchange of etheno-ATP bound to actin for ATP. All samples contained 0.1 @M etheno-ATP-actin monomers and 1,000 PM ATP. The
etheno-ATP-actin filament concentrations were 0 (bold line), 1 pM (O),2 pM (U), and 3 pM (A). The final etheno-ATP concentrations were
17, 19, 36, and 53 p M . Panel C , exchange of ADP bound to actin for etheno-ATP. All samples contained 1.0 p~ ADP-actin monomers and
200 p~ etheno-ATP. ADP-actin filament concentrations were 0 (bold line), 1 p~ (0),
2 p M (O), and 3 p M (A). The final ADP concentrations
20344
Nucleotides
Actin
0
ATP. q'
ATP. G
0
ADP-q'
ADP.G
8
E
100
0 1
FIG.4. Comparison of the rheological properties of ATPand ADP-actin filaments. Actin filaments at were analyzedby
oscillation in a cone-plate rheometer at 25 "C over a wide range of
frequencies. Conditions:34 PM ATP-actin (circles);34 p~ ADP-actin
(squares). The filled symbols represent the elastic modulus ( G ' ) is
dynes/cm2, and the open symbob represent the dynamic viscosity in
poise.
the critical concentrationsare determined by the pointed ends
where the ATP-actin and ADP-actin differ only by a factor
of 2-3 (Coue and Korn, 1985; Pollard, 1986).
We used nucleotide exchange by monomers and competitive
nucleotide bindingexperiments (see "Experimental Procedures") to calibrate the absolute fluorescence increment for
each nucleotide exchange pair. In our experiments with mixtures of filaments and thecritical concentrationof monomer,
only the critical concentration of actin monomers exchanged
rapidly. Even without capping the barbed ends, no nucleotide
exchange pair gave a substantial fluorescence signal from
filaments during the 20-min duration of our experiments.
Because the nucleotide exchange rate by polymerized actin is
so slow, a new experimental design will be required to detect
any subtle differences in nucleotide exchange rates of filaments assembled from ADP- and ATP-actin.
We found no difference in the electron microscopic appearance (Fig. 3) or mechanical properties (Fig. 4) of actin filaments assembled from ADP- or ATP-actin. Our observations
do not exclude small differences in the structureof ADP- and
ATP-actin filaments that may exist at the atomic level, but
both the microscopy and rheology differ from the report of
Janmey et al. (1990). First, our ADP-actin filaments have
smooth, not irregular, profiles just like ATP-actin filaments.
Downloaded from www.jbc.org at Yale University on April 7, 2009
FIG. 3. Electron micrographs of negatively stained actin filaments. A-C, filaments assembled from ATP-actin. D-F, filaments
assembled from ADP-actin. Magnifications:A, R, D,and E = 113,000 X; C and F = 173,000 X. Bars = 0.1 um.
20345
Actin Nucleotides
through the filaments and to preparefilaments for rapid
disassembly at a time subsequentto theirassembly.
Acknowledgments-WethankDr.
Pascal Goldschmidt-Clermont
for suggesting the use of the hexokinase-agarose beads for making
ADP-actin andDr.PaulJanmeyfor
exchanging ideas, data, and
proteins in our ongoing effort to reconcile our experimental results.
REFERENCES
Amato, P. A., and Taylor, D. L. (1986) J. Cell Bid. 102,1074-1084
Barshop, B. A,, Wrenn, R. F., and Frieden,C. (1983) Anal. Biochem. 1 3 0 , 134145
Carlier, M.-F. (1991) J. Bid. Chem. 2 6 6 , 1-4
Carlier, M.-F., and Pantaloni, D. (1986) Biochemistry 26,7789-7792
Carlier, M.-F., and Pantaloni, D.(1988) J. Biol. Chem. 263,817-825
Carlier, M.-F., Pantaloni, D., and Korn, E.D. (1987) J. Bid. Chem. 262,30523059
Carlier, M.-F., Pantaloni, D., Evans, J. A., Lambooy, P. K., Korn, E. D., and
Webb, M. R. (1988) FEBS Lett. 236,211-214
Cooper, J. A. (1991) Annu. Reu. Physiol. 5 3 , 585-605
Cooper, J. A., and Pollard, T. D. (1982) Methods Enzymol. 85,182-210
Coue, M., and Korn, E. D. (1985) J. Biol. Chem. 260,15033-15041
Drenckhahn, D., and Pollard, T. D. (1986) J. Bid. Chem. 261,12754-12758
Drewes, G., and Faulstich, H. (1991) J. Bid. Chem. 2 6 6 , 5508-5513
Forscher, P., and Smith,S. J. (1988) J. Cell Biol. 107,1505-1516
Frieden, C., and Patane, K. (1988) Biochemistry 2 7 , 3812-3820
Gershman, L., Selden, L., and Estes,J. (1986) Biochem. Biophys. Res. Commun.
135,607-614
Gershman. L.. Selden., L... Kinosian. H. J.. and Estes., J . (1989)
Biochim. Bioohvs.
.
. ”
Acta 996,109-115
Holmes, K., Popp, D., Gebhard, W., and Kabsch, W. (1990) Nature 3 4 7 , 4449
Janmey, P.A., Hvidt, S., Peetermans, J., Lamb, J., Ferry, J. D., and Stossel, T.
P. (1988) Biochemistry 27,8218-8227
Janmey, P. A,, Hvidt, S., Oster, G. F., Lamb, J., Stossel, T. P., andHartwig, J.
H. (1990) Nature 347,95-99
Kabsch, W., Mannherz, H. G., Suck, D., Pai, E. F., and Holmes, K. C. (1990)
Nature 347,37-44
Korn, E. D., Carlier, M.-F., and Pantaloni, J. (1987) Science 238,638-644
Kuehl, W. M., and Gergely, J. (1969) J. Biol. Chem. 244,4720-4729
MacLean-Fletcher, S., and Pollard, T. D. (1980) Biochem. Biophys. Res. Commun. 9 6 , 18-27
Martonosi, A,, Gouvea, M. A,, and Gergely, J. (1960)J. Bid. Chem. 236,17001711
Nowak, E., Strezelecka-Golaszwska,H., and Goody, R. S. (1988) Biochemistry
2 7 , 1785-1792
O’Donoghue, S. I., Miki, M., and dosRemedios, C.G. (1992) Arch. Biochem.
Biophys. 2 9 3 , 110-116
Ohm, T., and Wegner, A. (1991) Biochemistry 3 0 , 1193-1197
Pollard, T. D. (1984) J. Cell Biol. 9 9 , 769-777
Pollard, T. D. (1986) J. Cell Biol. 103,2747-2754
Pollard, T. D. (1990) Curr. Opin. Cell Biol. 2 , 33-40
Pollard, T.D., and Weeds, A. G. (1984) FEBS Lett. 170,94-98
Rickard, J. E., and Sheterline, P. (1988) J. Mol. Bid. 201,675-681
Sato. M..Leimbach. G.. Schwarz. W. H.. and Pollard. T. D. (1985)
J. Bid.
,
,
Chem. 260,8585-8592
Straub, F. B., and Feuer, G. (1950) Biochim. Biophys. Acta 4,455-470
Theriot, J. A., and Mitchison, T. J. (1991) Nature 3 5 2 , 126-131
Waechter, F., and Engel, J. (1975) Eur. J. Biochem. 57,453-459
Wang, Y:L. (1985) J. CellBiol. 1 0 1 , 597-602
Wang, Y.-L., and Taylor, D. L. (1981) Proc. Natl. Acad. Sci. U. S. A. 78,55035507
Wanger, M., and Wegner, A. (1987) Biochim. Biophys. Acta 9 1 4 , 105-113
,
I
,
Downloaded from www.jbc.org at Yale University on April 7, 2009
Second, our ADP-actin filaments and ATP-actin filaments
had similar mechanical properties, whereas the ADP-actin
filaments of Janmey et al. were about 10 times more rigid
than their ATP-actin filaments. Furthermore, the absolute
values of the elastic modulii of filaments prepared from fresh
actin are much lower in our laboratory than Janmey’s, an
unresolved, long-standing difference (see Sato et al., 1985;
Janmey et al., 1988). The experimental design differed in two
ways: ( a ) methods to prepare the actin, and ( b ) the inclusion
of gelsolin in the rheological experiments of Janmey et al. to
control the length of the filaments. Differences in the actin
rather than the gelsolin seem more likely to account for the
difference in rigidity.
The differences in the observations between the two laboratories have not yet been resolved but may lie in the preparation of the ADP-actin. Our current methods take advantage
of rapid exchange of nucleotide by Mg-actin (Gershman et al.,
1989) to make ADP-actin quickly in the cold in an effort to
avoid an inevitable, time-dependent (k = 2 X loT5s-l at 0 “C)
conformational change that can be detected in ADP-actinby
sulfhydryl titration of Cys-10 (Drewes and Faulstich, 1991).
We never freeze purified actin preparations. Janmey et al.
used a slower, early method from our laboratory (Pollard,
1984) to prepareADP-actin from frozen stocks. Ca-ATPactin is incubated with hexokinase and glucose at room temperature for several hours. These ADP-actin preparationsare
not completely stable (Pollard, 1984), perhaps because of the
conformational change. It was necessary to incubate their
ADP-actin for 2 h at room temperature before the differences
with ATP-actinbecame apparent. TheCys-10 conformational
change should have occurred on part of the ADP-actin prepared by this method and may account for the substantial
differences in the properties of the ADP-actin in the two
laboratories. It is particularly interesting that the structural
differences in ADP-actin filamentsobserved by Janmey et al.
(1990) were reversed rapidly by ATP just like the Cys-10
conformational change (Drewes and Faulstich, 1991). O’Donoghue et al. (1992) also observed that ADP-actin prepared by
our 1984 method has a higher viscosity than ATP-actin.
Given that ADP- and ATP-actinfilaments are the same by
our criteria, new evidence will be necessary to confirm the
hypothesis that nucleotide hydrolysis within actin filaments
is used to modulate their mechanical properties. Rather, we
expect that nucleotide hydrolysis is used to drive subunit flux
~