Detailed Tuning of Structure and Intramolecular Communication
Are Dispensable for Processive Motion of Myosin VI
Mary Williard Elting,†‡Zev Bryant,§Jung-Chi Liao,§and James A. Spudich†*
†Department of Biochemistry,‡Department of Applied Physics, and§Department of Bioengineering, Stanford University, Stanford, California
mechanism of this processive motion by measuring the impact of structural and chemical perturbations on single-molecule
processivity. Processivity is maintained despite major alterations in lever arm structure, including replacement of light chain
binding regions and elimination of the medial tail. We present kinetic models that can explain the ATP concentration-dependent
processivities of myosin VI constructs containing either native or artificial lever arms. We conclude that detailed tuning of struc-
ture and intramolecular communication are dispensable for processive motion, and further show theoretically that one proposed
type of nucleotide gating can be detrimental rather than beneficial for myosin processivity.
Dimeric myosin VI moves processively hand-over-hand along actin filaments. We have characterized the
Myosin VI has cellular roles ranging from transporter to
anchor in vivo (1,2). In vitro, it is processive (defined as
taking multiple steps along an actin filament before detach-
ing) and shares features with processive (þ)-end directed
myosins, including a hand-over-hand mechanism (3), strides
that approximately span the pseudorepeat of the actin
helix (4), and intramolecular gating (5). However, the
detailed mechanisms of myosin VI processivity might be
expected to diverge from other myosins due to a number
of unique structural and functional characteristics: it
achieves (?)-end-directed motion due to a unique insert
following the catalytic domain that redirects its lever
arm in the opposite direction from other known myosins
(6–8), and it has an unusual lever arm structure with two
light-chain-binding domains (9) followed by a sequence
that forms an extended single a-helix as an isolated frag-
Myosin VI stepping is gated (5): communication between
the two heads keeps their kinetic cycles out of phase. This
communication is thought to occur via tension transmitted
through the lever arms that connect the heads, altering the
kinetics in such a way that front head release is prevented
(11,12). Previous work suggests two possible mechanisms
producing gating in myosin VI. The rate of ADP release
may be reduced in the front head when both heads are bound
to actin (5), as is known to occur in myosin V (13). Reported
reductions in this rate range from two- (14) to at least
10-fold (15) for myosin VI.
Alternatively, there is evidence of a reduced rate of ATP
binding to the front head (16). Either mechanism is capable
of preventing front head release, because the myosin VI
head remains strongly attached to the actin filament in
both the ADP-bound and the nucleotide-free state. Although
it has been suggested that myosin VI must have a different
gating mechanism than myosin V because of its reverse
directionality (17), ADP release could be affected by
multiple structural distortions, and either gating mechanism
seems physically plausible. The properties of the lever arm
and tail structure of myosin VI, and of other molecular
motors, have been proposed to be important for successful
intramolecular tension sensing, and therefore for gating
and processivity (15,18,19).
In this work, we probe the structures and mechanisms
underlying myosin VI processivity. Mechanistic models
of myosins (15,20–25) and of kinesin (26–29) have been
tested by engineering changes into these molecular motors
and examining whether their behaviors respond predict-
ably. We previously incorporated an artificial lever arm
into a myosin VI dimer and showed that the chimera
remains processive (21). Here, we examine the same
chimera in detail to probe the effects of the structural
changes on processivity. We use kinetic modeling, fit to
our processivity data for constructs with native or artificial
lever arms, to examine the relevance of gating to
We present models that explain our observed processiv-
ities over a wide range of conditions, and show theoretically
that gating of ADP release yields only modest improve-
ments in processivity, while gating of ATP binding actually
decreases processivity at physiological ATP concentrations.
Processivity seems to be a robust characteristic of dimeric
molecular motors with high duty ratios even in the absence
of gating. This idea is underscored by the processive
activity of additional chimeric constructs in which we
have incorporated increasingly dramatic changes in lever
Submitted June 22, 2010, and accepted for publication November 16, 2010.
Jung-Chi Liao’s present address is Department of Mechanical Engineering,
Columbia University, New York, NY.
Editor: Claudia Veigel.
? 2011 by the Biophysical Society
430Biophysical Journal Volume 100 January 2011 430–439
MATERIALS AND METHODS
Protein expression and purification
Proteins were expressed as described (7,21). Briefly, myosin VI or chimeras
were cloned into pBiEx-1 (see also the Supporting Material) and trans-
fected into SF9 cells, Flag affinity purified, and labeled with TMR-HaloTag
ligand (Promega, Madison, WI).
Single-molecule TIRF microscopy
Single-molecule TIRF microscopy was performed as described (21) (see
also the Supporting Material). Briefly, actin was attached to coverslip
surfaces using heavy meromyosin (HMM) prepared from rabbit skeletal
muscle and treated with n-ethyl maleimide (NEM) to inactivate it so that
it would bind actin but not detach. After rinsing out unattached actin fila-
ments, motor was flowed into the chamber in assay buffer including ATP
and reagents for ATP regeneration and oxygen scavenging. Additionally,
MgCl2was added at a concentration that varied with ATP concentration
and was chosen to result in 2 mM free Mg2þ(or other Mg2þconcentration
as noted). Expected free Mg2þconcentrations were calculated using
MAXCHELATOR (http://maxchelator.stanford.edu). A 532-nm laser was
used for excitation, and its power was lowered to prevent misinterpreting
photobleaching as run termination when measuring processivity. Images
were analyzed by single-molecule tracking (30).
Stride size and dwell-time measurement
One-dimensional distance versus time traces created from two-dimensional
tracking data and steps were chosen using a step-picking algorithm based
on Kerssemakers et al. (31) (see also the Supporting Material). We used
this algorithm to generate fits with varying numbers of steps, but picked
the number of steps in each run by eye.
Run-length and velocity measurements
For details, see the Supporting Material. In brief, to ensure unbiased run
selection, custom software was used to create a kymograph for each actin
filament (Fig. S1). Runs were selected from these kymographs. The ex-
pected run-length, corrected for premature terminations of runs that ended
at the end of a filament, was calculated using the Kaplan-Meier survivor
function, a maximum likelihood estimator of the mean (32,33). The error
bars are the error on the mean estimate from the Kaplan-Meier survivor
function (32). The average velocity was calculated using a bootstrap fit
of the slope of a plot of the average distance traveled for every d-time for
all runs (Fig. S2). The mean and standard deviation of 50 bootstraps
were used as the velocity and error estimate for each data set.
A simplified model of the kinetic cycle of myosin VI was created with only
three states for each head: bound to actin and ADP, bound to actin but not
ADP, and bound to ATP and released from actin. Some states of a more
complete kinetic cycle (see later in Fig. 2 a) were omitted (and assumed
to be rarely visited) or were collapsed into one of the three remaining states.
Transfers between states were approximated as irreversible (see later in
Fig. 2 b). A branched kinetic model was created that included all combina-
tions of these three states in each of the two heads (see later in Fig. 2 c).
This model included a total of five parameters: two ADP release rates
corresponding to the front and rear heads, two effective ATP binding
rates corresponding to the front and rear heads, and the rebinding rate of
the free head.
The relative rates of ADP release from the front and rear heads or of ATP
binding to the front and rear heads were not allowed to vary independently,
but were set at a fixed ratio depending on the gating mechanism used for the
fit, as described in the text. An additional parameter was added to compen-
sate for premature run termination due to actin attachment or defects in the
actin filament, since we found that actin attachment mechanisms affect
processivity (see below). Run-lengths and durations as a function of ATP
concentration were analytically calculated by summing over all possible
paths to run termination (see derivation in the Note in the Supporting Mate-
rial). This modeling was verified by Monte Carlo simulations of the same
process, which agreed well with the analytic version (Fig. S3).
To fit parameters to the model, the apparent ATP binding rate and ADP
release rates were fit in a least-squares manner to the velocity data. (Note
that although these velocity plots resemble Michaelis-Menten plots, they
are not fit to simple Michaelis-Menten kinetics. Instead, the data are fit
directly to velocities calculated using the comprehensive kinetic model.)
These rates were fixed and the rebinding rate and actin defect parameters
were fit using the run-length data. Alternating fits were then performed of
thevelocity and run-length data, as described, until all four rates converged.
Errorestimatesfor thefits weregeneratedbyusingthe standarddeviationof
the parameters from a parametric bootstrap comprising 1000 randomly
generated data sets based on the measured mean and error for each run-
length or velocity data point. (See Methods in the Supporting Material
for more details.)
The behavior of dimeric myosin VI and a chimeric
construct are similar
Gating is thought to arise from tension transmitted through
the lever arm. To assess the impact on processivity of
altering the lever am structure, we conducted a detailed
characterization of an engineered processive myosin we
previously constructed (21). In M6DI816-2R~MT-GCN4,
the IQ and proximal tail domains were removed and
replaced with an artificial lever arm (two spectrin repeats
from a-actinin, abbreviated as 2R (20)) fused after the distal
portion of the unique insert, followed by a (GSG)3flexible-
linker swivel (~), and the medial tail (MT), with GCN4 used
to ensure dimerization. We compared this to our myosin VI
control dimer (4), M6-GCN4, which contains native
sequence through the medial tail, followed by GCN4 for
dimerization (Fig. 1 and Movie S1 and Movie S2 in the
Supporting Material). Using single molecule total internal
reflection fluorescence (TIRF) microscopy, we found that
the stride size distributions for M6DI816-2R~MT-GCN4
and M6-GCN4 are similar (Fig. 1, c and d), as we showed
previously under different actin conditions (21). Here, we
show that the two constructs also have similar dwell-time
distributions (Fig. 1 e). We then compared them to test for
effects of the artificial lever arm on gating and processivity.
Modeling effects of gating on processivity
Kinetic models of processive motion can be tested by
measuring run-length as a function of nucleotide conditions
(34). To generate predictions of the effects of gating mech-
anisms, we constructed models of myosin VI processivity.
Biophysical Journal 100(2) 430–439
Requirements for Myosin VI Processivity 431
We first describe a two-state, conceptual model, followed by
the three-state model we used to perform fits.
The simplest model includes only two states for each
head: bound to or detached from the actin filament. There
are two different effects that can break symmetry in
a two-headed walker (35):
1. It can be more likely for the rear head to detach than for
the front head.
2. A released head can preferentially bind in front of, rather
than behind, the remaining attached head.
Either of these effects is sufficient on its own to ensure
directional motion. We assume that the second effect always
occurs for myosin VI. It is the first effect that we describe
here as ‘‘gating’’; thus ‘‘perfect gating’’ means that the front
head never detaches.
In this simple scheme, it is clear that a high duty ratio can
result in significant processivity, as has been suggested (11).
The probability of the dimer detaching during any cycle
(meaning the second head detaches during the period of
the cycle that the first head is already detached) is 1?r,
where r corresponds to the duty ratio of a monomeric
motor, and is given by the rebinding rate over the sum of
the rebinding and detachment rates. The average number
of ATP cycles (equivalent to head detachment/reattachment
cycles) carried out by the motor before completely dissoci-
ating is r/(1?r). (See Note in the Supporting Material.) This
also represents the average number of forward strides in the
presence of perfect gating. However, statistically half of
these cycles would be unproductive in the absence of
gating, yielding an observed run-length of r/(2*(1?r).)
Thus, perfect gating only increases the run-length twofold
in this model, while at high duty ratios run-length is
inversely proportional to the percentage of the cycle spent
Though this two-state model is instructive, and we expect
its trends to carry through to more complicated models, its
simplicity prevents it from differentiating among types of
gating mechanisms. We therefore used a model with three
states: bound to both actin filament and ADP; bound to
the actin filament while not bound to nucleotide; and
detached from the actin filament while bound to ATP (see
Fig. 2, a and b, and Materials and Methods). Transfers
between states are irreversible (as we assume the reverse
rates are slow enough to be ignored) and we assume the
free head always rebinds in front of the attached head,
guaranteeing unidirectional travel. Three rate constants are
needed to describe this model: the actomyosin ADP release
rate, the apparent actomyosin ATP binding rate, and the
actin-rebinding rate of the detached head.
M6-GCN4 and M6DI816-2R~MT-GCN4. (a) Sche-
matics of constructs. (Gray, head of native myosin
VI; green (C), converter domain; purple (UI),
calmodulin binding unique insert; cyan (IQ), IQ
domain; orange (PT), proximal tail domain; red
(MT), medial tail domain; brown (G), GCN4;
yellow, HaloTag; blue (2R), two spectrin repeats
from a-actinin; and wavy line, (GSG)3 flexible
linkers.) (b) Cartoon of M6-GCN4 structure,
color-coded to match part a. (c) Example stepping
traces for M6-GCN4 (red) and M6DI816-2R~MT-
GCN4 (blue). Steps fit to these traces (black).
(d) Distributions of stride sizes for M6-GCN4
and M6DI816-2R~MT-GCN4. Stride size histo-
grams (markers) are shown with a fit to the distri-
bution of positive stride sizes (solid). Error bars are
calculated as the square-root of the number of
strides in each bin, scaled to proportion of strides.
Peak positive stride sizes are 33.5 5 0.7 nm
(N ¼ 158) for M6-GCN4 (red) and 30.1 5
0.6 nm (N ¼ 252) for M6DI816-2R-MT-GCN4
(blue). (e) Dwell-time distributions for M6-
GCN4 (red), mean dwell-time of 8.7 5 0.7 s and
M6DI816-2R~MT-GCN4 (blue), mean dwell-time
of 7.4 5 0.4 s. Short dwells below a cutoff (dashed
lines) were ignored to avoid undersampling near
our time resolution. Curves (solid black) are expo-
nential distributions with decay constants equal to
the mean dwell-time, shifted by the undersampling
Single molecule stepping results for
Biophysical Journal 100(2) 430–439
432 Elting et al.
This three-state model for each head can be used as the
foundation for a dimeric model by allowing each head to
cycle through these three states independently, creating
a branched kinetic pathway with a total of nine states
(Fig. 2 c). Which pathways of this model are predominantly
visited depends on the mechanism of gating. If ADP release
is gated, myosin VI predominantly visits a pathway through
states A, B, and C1(shaded pink) because transitions from
A/D and from B/E are inhibited by the slowed ADP
release rate from the front head. If binding of ATP is gated,
but release of ADP is not, myosin VI will predominantly
follow the green-shaded pathways (which include the
pink-shaded area) because D/C2 and E/F2 transitions
are inhibited by the slowed front head ATP binding rate.
In the absence of gating, all pathways (shaded blue) are
A total of five parameters are possible in the model
described thus far because the ADP release rates and
apparent ATP binding rates may be different between the
front and rear heads. However, to assess the effects of gating
mechanisms, for each fit we set the ratio of ADP release
rates and the ratio of ATP binding rates to be fixed constants
(which varied depending on the gating mechanism),
reducing the number of parameters to three. Additionally,
a fourth parameter was added to account for runs that termi-
nate prematurely due to actin filament effects, such as
defects in the actin filaments or their interactions with the
coverslip (as described below).
Assuming that a productive step occurs when the rear
head detaches while the front head remains attached and
strokes, and that a run terminates when both heads detach
simultaneously, we used this model to derive analytic
kinetic cycle. (a) Schematic of kinetic pathway
are grouped together or ignored in the simplified
kinetic model. (Green) States in which myosin is
to the actin filament. (Red) States in which myosin
is bound to actin and ADP. The remaining states
(b) Cartoon of the modeled kinetic cycleof a single
myosin head. States and transitions between states
are colored to match panel a. The prestroke state
(yellow box) is bound to ADP.Pi(green circle) and
unbound from actin. At rate krebind, myosin rebinds
to a poststroke state (red box) where it is bound to
actin and ADP (red circle). Note that krebindmay
actually be limited by phosphate release or the
weak to strong transition instead of rebinding.
to actin butnotnucleotide (green box). The cycle is
completed when myosin binds ATP at rate kATPon
and releases from the actin filament. (c) Diagram
of transitions in a dimeric motor, highlighting the
paths that predominate with different mechanisms
are directed toward the left (black arrow). Myosin
heads are marked as containing ATP or ADP.Pi
(green dot), ADP (red dot), or no nucleotide (no
dot) at the nucleotide binding site. Transitions that
are in competition with rebinding (thinner arrows)
have low probabilities because the rebinding rate
is the fastest rate in the cycle. Shading indicates
the predominant pathways in the case of gating
of ADP release (pink); in the case of gating of
ATP binding (green shading, which includes the
pink shaded region); and in the case of no gating
(blue shading, which encompasses all states). Inhi-
of ADP release (red) or gating of ATP binding
Three-state model for myosin VI
Biophysical Journal 100(2) 430–439
Requirements for Myosin VI Processivity433
expressionsfor averagerun-length andvelocityasafunction
of ATP concentration, effective ATP binding rates to the
front and rear heads, ADP release rates from the front and
rear heads, actin rebinding rate of the detached head, and
actin defect parameter (see Note in the Supporting Mate-
rial). These expressions were derived by summing over all
possible pathways toward a productive step or dissociation
(see Fig. 2 c, Fig. S4 and Fig. S5, and Note in the Supporting
Material). These summations were calculated in a manner
similar to that discussed in Ninio (36) by considering the
relative probabilities of transitions at each branch point,
were verified by Monte Carlo simulation (Fig. S3), and
are derived in detail in the Supporting Material. Gating
mechanisms were incorporated into the model by varying
the ratio of front/rear head ATP binding rates or front/rear
head ADP release rates.
Comparison of modeled and measured
We measured average velocities and run-lengths, correcting
for the finite length of the actin filaments, across a range of
ATP concentrations (Fig. 3 and Materials and Methods).
Both M6DI816-2R~MT-GCN4 and M6-GCN4 exhibited
similar velocities as a function of ATP concentration. Both
constructs showed increased run-lengths at lower concentra-
tions of ATP, as expected because the duty ratio is increased
when ATP-binding is rate-limiting.
We first fit our processivity and velocity data for M6-
GCN4 to the model assuming 10-fold gating of ADP release
(that is, slowing the release of ADP from the front head by
a factor of 10 when both heads are attached). This has
recently been proposed as a lower bound on ADP release
gating, based on high time-resolution, single-molecule
myosin VI measurements (15). This model fits quantita-
tively to our data (Fig. 3, c and d, and Table 1), and yields
parameters in good agreement with previously published
values based on bulk kinetic measurements (5,16).
Although the chimera M6DI816-2R~MT-GCN4 is highly
processive, it is less processive than M6-GCN4 (Fig. 3 d).
One explanation is that gating is damaged in this construct
because the forces transmitted between the heads are altered
by the artificial lever arm. This is likely because the flexible
M6DI816-2R~MT-GCN4. (a) M6-GCN4 (red) and (b) M6DI816-2R~MT-
GCN4 (blue) at 2 mM ATP. Kaplan-Meiersurvivorfunctions,with compen-
sation for runs terminating at filament ends (32,33), are shown (color).
Exponential distributions based on the Kaplan-Meier estimate of the
mean (0.556 5 0.050 mm for M6-GCN4; 0.226 5 0.016 mm for
M6DI816-2R~MT-GCN4) are shown (solid black). Runs shorter than
0.216 mm (two pixels) are truncated due to undersampling of runs close
to our time resolution, and the mean estimator has been shifted accordingly
(see MaterialsandMethods).(c) Velocityand (d) run-lengthasa functionof
ATP concentration for M6-GCN4 (red) and M6DI816-2R~MT-GCN4
(blue). M6DI816-2R~MT-GCN4 is slightlyslower than M6-GCN4.If veloc-
ities for both constructs are scaled to an expected stepping rate (based on
measured stepsize), the velocity difference is reduced, because the chimera
has a slightly shorter stepsize (Fig. S7). M6DI816-2R~MT-GCN4 shows
reduced processivity compared to M6-GCN4, possibly as a result of
damaged gating. (Error bars) Standard deviation of bootstrapped mean
(for velocity) and Kaplan Meier estimate for standard deviation of the
mean (run-length). (Solid lines) Fit to processivity model (see parameters
in Table 1).
Velocity and run-length measurements for M6-GCN4 and
TABLE 1 Modeling parameters for fits/models shown in Figs. 3 and 4
(mM?1s?1)Steps per actin defect
ADP-release gating model
ATP-binding gating model
No gating model
Relevant previous measurements
5.1 5 0.2*
3.8 5 0.1*
0.016 5 0.001*
57 5 12*
48 5 11*
Parameters marked in bold were allowed to vary independently in the fits. The front head ADP release and ATP binding rates were varied with fixed ratios to
the rear head rates; all other parameters not in bold were fixed. Error estimates were generated using parametric bootstrapping (see Materials and Methods in
this article, and Methods in the Supporting Material).
*Parameters allowed to vary independently in fits.
yFrom de la Cruz et al. (5).
zFrom Sweeney et al. (16).
Biophysical Journal 100(2) 430–439
434 Elting et al.
linker between the artificial lever arm and medial tail should
serve to reduce tension that might otherwise be transmitted
between the heads. The velocity data for the chimera are
slower at high concentrations of ATP, when ADP release
is rate-limiting, but not at low concentrations of ATP,
when ATP binding is rate-limiting, suggesting that the rate
of ADP release is somewhat slowed in the chimera whereas
the rate of ATP binding remains the same (Fig. 3 c).
To test consistency with our model, we fit a new ADP
release rate to the M6DI816-2R~MT-GCN4 velocity data
and otherwise used the same parameters from the
the front and rear head ADP release rates equal. The theory
agrees with the processivity and velocity data for this
construct, using only a one-parameter fit to the velocity
and a zero-parameter fit to the processivity (Fig. 3, c and d,
(10-fold slowing of ADP release to the front head in
M6-GCN4 and no gating in M6DI816-2R~MT-GCN4) are
consistent with our processivity data, other interpretations
are also consistent with this type of modeling. For instance,
it would be possible to fit the chimera processivity data by
altering all four modeling parameters and including some
gating, or to fit the M6-GCN4 data without gating. However,
these fits require more free parameters than the fit we have
shown, yield parameters with less consistency between
constructs, and give rebinding rates that are less consistent
with previously measured values (Fig. S6 and Table S1).
Surface attachment effects on processivity
Processivity in both constructs was affected by the mecha-
nism of attachment of actin filaments to the coverslip.
When we used a higher concentration of NEM-treated
HMM (see Materials and Methods) to attach actin to the
surface, we observed shorter runs, suggesting that myosin
bumping into an attachment point more frequently increases
the frequency of run termination (Table S2). We also found
that biotinylating actin to attach it to the surface, likely
a less flexible attachment mechanism than NEM-HMM,
shortened the runs (data not shown). These effects led us to
add an additional actin defect parameter to our modeling,
or NEM-HMM (see Table 1 and Table S3). Because these
results showed that the state of the actin binding to the
surface affects the run-length, we collected all directly-
compared processivity data (e.g., all data in Fig. 3 and, sepa-
minimum amount of NEM-HMM necessary to attach actin.
Effect of free Mg2þon processivity
As a further demonstration of the effectiveness of the model,
we used it to make predictions of the effect on processivity
of altering free Mg2þconcentration. Free Mg2þconcentra-
tion can alter thevelocity (37,38) and processivity (25,39) of
a molecular motor. Therefore, we compensated for changes
in the ATP concentration to keep the amount of free Mg2þ
constant in the experiments described above (see Materials
and Methods). It has been observed for myosin V and
myosin I that the ADP release rate is dependent on the
Mg2þconcentration, with the suggested mechanism that
Mg2þmust release from the nucleotide binding pocket
before ADP (37–39). We observe a similar effect in myosin
VI: at high ATP concentrations, when ADP release is rate-
limiting, higher Mg2þconcentrations result in slower veloc-
ities (Fig. S8 a).
Assuming free Mg2þaffects only ADP-release, we fit that
parameter to each point in the velocity data, and used our
model to predict the corresponding expected run-length
(with the additional actin binding parameter fit as well
across all Mg2þconcentrations) (Fig. S8 and Table S3).
This predicts that higher Mg2þconcentrations and slower
velocities should cause a corresponding increase in proces-
sivity, as we observe.
Effects of gating mechanisms on processivity
After creating and testing a kinetic model (see Fig. 2 c and
Note in the Supporting Material) for myosin VI stepping, we
used the model to better understand how gating should be
expected to affect processivity. We fixed the modeling
parameters to the values described in the earlier fit to
M6-GCN4, and calculated processivities resulting from
different gating mechanisms. We compared three possibili-
ties: no gating (each head cycles independently); 10-fold
slowing of ADP release from the front head; and 10-fold
slowing of ATP binding to the front head (Fig. 4). Clearly,
effective processivity is possible in the absence of intramo-
lecular gating. In fact, if we look at the most physiologically
nisms of gating: no gating (blue), 10-fold gating of ADP release from front
head (red), and 10-fold gating of ATP binding to front head (green). Param-
eters input into this model are the same as the M6-GCN4 fit shown in Fig. 3.
All parameters are kept constant when comparing the models; only the
mechanism of gating is changed (see Table 1).
Comparison of processivity model using different mecha-
Biophysical Journal 100(2) 430–439
Requirements for Myosin VI Processivity435
relevant region of this curve, at the highest ATP concentra-
tions, we find that gating of ADP release only improves
processivity by approximately a factor of 2.5, and that,
perhaps surprisingly, gating of ATP binding slightly
The potential for ATP gating to disrupt processivity can
be understood by considering what happens when the front
head releases ADP before the rear head and the motor enters
state D (Fig. 2 c). This event occurs frequently unless ADP
release is gated. At high ATP concentration in the absence of
any gating, the front head will usually rebind ATP and
regenerate state A via a pathway (D/C2/A) that wastes
an ATP hydrolysis but is relatively safe from dissociation
because the one-head bound state (C2) has ADP bound.
However, ATP gating specifically inhibits this pathway,
and thus the motor proceeds through the only available alter-
native pathway, through state E to the very vulnerable state
F (Fig. 2 c, yellow box).
At low ATP concentrations, D/C2 becomes unlikely
and the motor always proceeds through states E and F.
Under these conditions, ATP gating is preferable to no
gating because the motor proceeds through the productive
state F1 rather than the unproductive state F2. There must
thus be a crossover ATP concentration at which the protec-
tive effect of ATP rebinding to the front head fully compen-
sates for the detrimental effect of unproductive cycling, as
shown in Fig. 4.
Structural requirements for myosin VI
We created two additional chimeras to further test the
structural robustness of myosin VI processivity. These
constructs are: M6PI790-2R~MT-GCN4, identical to the
previous chimera except that the distal, calmodulin-binding
portion of the unique insert has been removed, and the
artificial lever arm is fused after the proximal portion of
the unique insert, which redirects the lever arm; and
M6DI816-2R~GCN4IL, identical to the first chimera except
that the medial tail has been removed and a hyperstable
mutant of GCN4 (40) immediately follows the swivel
(Fig. 5 a). (We were not able to demonstrate processivity
in a M6DI816-2R~GCN4 construct with GCN4 alone.)
Results with these constructs demonstrate that myosin
VI processivity is possible without the medial tail and in
the absence of calmodulin (see Fig. 5, and Movie S3 and
Movie S4). Removal of the unique insert calmodulin is
a dramatic structural modification, considering the close
interaction of this calmodulin with the converter domain
(6,41). All three chimeric structures show remarkable simi-
larity in their basic behavior to our M6-GCN4 control: they
are able to take multiple steps along the actin filament
without detaching, with similar stride sizes and kinetics
(Fig. 5, b–d). All three chimeras do have slightly shorter
stride sizes than M6-GCN4, showing that processivity is
2R~GCN4IL at 5 mM ATP. (a) Schematic of
constructs. (Gray, head of native myosin VI; green
(C), converter domain; purple (UI), calmodulin
binding unique insert or unique insert truncated
at residue 790 (before calmodulin binding site);
red (MT), medial tail domain; brown (G/IL),
GCN4 or GCN4IL; yellow, HaloTag; blue (2R),
two spectrin repeats from a-actinin; and wavy
line, (GSG)3 flexible linkers.) (b) Dwell-time
distributions at 5 mM ATP for M6PI790-2R~MT-
GCN4 (green), mean dwell-time of 11.7 5 0.7 s
dwell-time of 6.3 5 0.3 s. (Dotted lines) Cutoff
of short dwells from undersampling near our time
resolution. Curves (solid black) are exponential
distributions with time constants of the mean
dwells, shifted by the undersampling cutoff. (c)
Example stepping traces of M6PI790-2R~MT-
(magenta). Modeled steps are shown (black). (d)
Stride size distributions of chimeras with M6-
GCN4 for comparison. Histograms are shown
with fits as in Fig. 1. Peak positive stride sizes
are 33.5 5 0.7 nm (N ¼ 158) for M6-GCN4
(red), 27.9 5 0.6 nm (N ¼ 238) for M6PI790-
2R~MT-GCN4 (green), and 26.2 5 0.5 nm (N ¼
335) for M6DI816-2R~GCN4IL(magenta).
Single molecule stepping results for
Biophysical Journal 100(2) 430–439
436Elting et al.
2. Spudich, J. A., and S. Sivaramakrishnan. 2010. Myosin VI: an innova-
tive motor that challenged the swinging lever arm hypothesis. Nat. Rev.
Mol. Cell Biol. 11:128–137.
3. Okten, Z., L. S. Churchman, ., J. A. Spudich. 2004. Myosin VI walks
hand-over-hand along actin. Nat. Struct. Mol. Biol. 11:884–887.
4. Rock, R. S., S. E. Rice, ., H. L. Sweeney. 2001. Myosin VI is a proc-
essive motor with a large step size. Proc. Natl. Acad. Sci. USA.
5. de la Cruz, E. M., E. M. Ostap, and H. L. Sweeney. 2001. Kinetic
mechanism and regulation of myosin VI. J. Biol. Chem. 276:32373–
6. Me ´ne ´trey, J., A. Bahloul, ., A. Houdusse. 2005. The structure of the
myosin VI motor reveals the mechanism of directionality reversal.
7. Bryant, Z., D. Altman, and J. A. Spudich. 2007. The power stroke of
myosin VI and the basis of reverse directionality. Proc. Natl. Acad.
Sci. USA. 104:772–777.
8. Park, H.,A.Li, .,H.L. Sweeney. 2007.Theuniqueinsertat the endof
the myosin VI motor is the sole determinant of directionality. Proc.
Natl. Acad. Sci. USA. 104:778–783.
9. Bahloul, A., G. Chevreux, ., H. L. Sweeney. 2004. The unique insert
in myosin VI is a structural calcium-calmodulin binding site. Proc.
Natl. Acad. Sci. USA. 101:4787–4792.
10. Spink, B. J., S. Sivaramakrishnan, ., J. A. Spudich. 2008. Long single
a-helical tail domains bridge the gap between structure and function of
myosin VI. Nat. Struct. Mol. Biol. 15:591–597.
11. Veigel, C., F. Wang, ., J. E. Molloy. 2002. The gated gait of the proc-
essive molecular motor, myosin V. Nat. Cell Biol. 4:59–65.
12. Robblee, J. P., A. O. Olivares, and E. M. de la Cruz. 2004. Mechanism
of nucleotide binding to actomyosin VI: evidence for allosteric head-
head communication. J. Biol. Chem. 279:38608–38617.
13. Rief, M., R. S. Rock, ., J. A. Spudich. 2000. Myosin-V stepping
kinetics: a molecular model for processivity. Proc. Natl. Acad. Sci.
14. Oguchi, Y., S. V. Mikhailenko, ., S. Ishiwata. 2008. Load-depen-
dent ADP binding to myosins V and VI: implications for subunit
coordination and function. Proc. Natl. Acad. Sci. USA. 105:7714–
15. Dunn, A. R., P. Chuan, ., J. A. Spudich. 2010. Contribution of the
myosin VI tail domain to processive stepping and intramolecular
tension sensing. Proc. Natl. Acad. Sci. USA. 107:7746–7750.
16. Sweeney, H. L., H. Park, ., S. S. Rosenfeld. 2007. How myosin VI
coordinates its heads during processive movement. EMBO J.
17. Sweeney, H. L., and A. Houdusse. 2010. Myosin VI rewrites the rules
for myosin motors. Cell. 141:573–582.
18. Lan, G., and S. X. Sun. 2005. Dynamics of myosin-V processivity.
Biophys. J. 88:999–1008.
19. Vilfan, A. 2005. Elastic lever-arm model for myosin V. Biophys. J.
20. Anson, M., M. A. Geeves, ., D. J. Manstein. 1996. Myosin motors
with artificial lever arms. EMBO J. 15:6069–6074.
21. Liao, J. C., M. W. Elting, ., Z. Bryant. 2009. Engineered myosin VI
motors reveal minimal structural determinants of directionality and
processivity. J. Mol. Biol. 392:862–867.
22. Manstein, D. J. 2004. Molecular engineering of myosin. Philos. Trans.
R. Soc. Lond. B Biol. Sci. 359:1907–1912.
23. Homma, K., M. Yoshimura, ., M. Ikebe. 2001. The core of the motor
domain determines the direction of myosin movement. Nature.
24. Baboolal, T. G., T. Sakamoto,., M. Peckham. 2009.The SAH domain
extends the functional length of the myosin lever. Proc. Natl. Acad. Sci.
25. Amrute-Nayak, M., R. P. Diensthuber, ., G. Tsiavaliaris. 2010. Tar-
geted optimization of a protein nanomachine for operation in biohybrid
devices. Angew. Chem. Int. Ed. Engl. 49:312–316.
26. Thorn, K. S., J. A. Ubersax, and R. D. Vale. 2000. Engineering the
processive run length of the kinesin motor. J. Cell Biol. 151:1093–
27. Yildiz, A., M. Tomishige, ., R. D. Vale. 2008. Intramolecular strain
coordinates kinesin stepping behavior along microtubules. Cell.
28. Adio, S., J. Jaud, ., G. Woehlke. 2009. Dissection of kinesin’s proc-
essivity. PLoS ONE. 4:e4612.
29. Miyazono, Y., M. Hayashi, ., H. Tadakuma. 2010. Strain through the
neck linker ensures processive runs: a DNA-kinesin hybrid nanoma-
chine study. EMBO J. 29:93–106.
30. Yildiz, A., J. N. Forkey, ., P. R. Selvin. 2003. Myosin V walks hand-
over-hand: single fluorophore imaging with 1.5-nm localization.
31. Kerssemakers, J. W., E. L. Munteanu, ., M. Dogterom. 2006.
Assembly dynamics of microtubules at molecular resolution. Nature.
32. Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from
incomplete observations. J. Am. Stat. Assoc. 53:457–481.
33. Nagy, S., B. L. Ricca, ., R. S. Rock. 2008. A myosin motor that
selects bundled actin for motility. Proc. Natl. Acad. Sci. USA.
34. Baker, J. E., E. B. Krementsova, ., D. M. Warshaw. 2004. Myosin V
processivity: multiple kinetic pathways for head-to-head coordination.
Proc. Natl. Acad. Sci. USA. 101:5542–5546.
35. Astumian, R. D. 2010. Thermodynamics and kinetics of molecular
motors. Biophys. J. 98:2401–2409.
36. Ninio, J. 1987. Alternative to the steady-state method: derivation of
reaction rates from first-passage times and pathway probabilities.
Proc. Natl. Acad. Sci. USA. 84:663–667.
37. Fujita-Becker, S., U. Du ¨rrwang, ., D. J. Manstein. 2005. Changes
in Mg2þion concentration and heavy chain phosphorylation regulate
the motor activity of a class I myosin. J. Biol. Chem. 280:6064–
38. Rosenfeld, S. S., A. Houdusse, and H. L. Sweeney. 2005. Magnesium
regulates ADP dissociation from myosin V. J. Biol. Chem. 280:6072–
39. Taft, M. H., F. K. Hartmann, ., G. Tsiavaliaris. 2008. Dictyostelium
myosin-5b is a conditional processive motor. J. Biol. Chem.
40. Harbury, P. B., T. Zhang, ., T. Alber. 1993. A switch between two-,
three-, and four-stranded coiled coils in GCN4 leucine zipper mutants.
41. Me ´ne ´trey, J., P. Llinas, ., A. Houdusse. 2008. The post-rigor structure
of myosin VI and implications for the recovery stroke. EMBO J.
42. Purcell, T. J., C. Morris, ., H. L. Sweeney. 2002. Role of the lever arm
in the processive stepping of myosin V. Proc. Natl. Acad. Sci. USA.
43. Sakamoto,T., A. Yildez, ., J. R. Sellers. 2005. Step-sizeis determined
by neck length in myosin V. Biochemistry. 44:16203–16210.
44. Sakamoto, T., M. R. Webb, ., J. R. Sellers. 2008. Direct observation
of the mechanochemical coupling in myosin Va during processive
movement. Nature. 455:128–132.
45. Block, S. M. 2007. Kinesin motor mechanics: binding, stepping,
tracking, gating, and limping. Biophys. J. 92:2986–2995.
46. Uemura, S., H. Higuchi, ., S. Ishiwata. 2004. Mechanochemical
coupling of two substeps in a single myosin V motor. Nat. Struct.
Mol. Biol. 11:877–883.
47. Gebhardt, J. C., A. E. Clemen, ., M. Rief. 2006. Myosin-V is
a mechanical ratchet. Proc. Natl. Acad. Sci. USA. 103:8680–8685.
Biophysical Journal 100(2) 430–439
438Elting et al.
48. Nishikawa, S., K. Homma, ., M. Ikebe. 2002. Class VI myosin moves
processively along actin filaments backward with large steps. Biochem.
Biophys. Res. Commun. 290:311–317.
49. Sun, Y., H. W. Schroeder, 3rd, ., Y. E. Goldman. 2007. Myosin VI
walks ‘‘wiggly’’ on actin with large and variable tilting. Mol. Cell.
50. Park, H., B. Ramamurthy, ., H. L. Sweeney. 2006. Full-length myosin
VI dimerizes and moves processively alongactin filaments uponmono-
mer clustering. Mol. Cell. 21:331–336.
51. Mukherjea, M., P. Llinas, ., H. L. Sweeney. 2009. Myosin VI dimer-
ization triggers an unfolding of a three-helix bundle in order to extend
its reach. Mol. Cell. 35:305–315.
52. Lister, I., S. Schmitz, ., J. Kendrick-Jones. 2004. A monomeric
myosin VI with a large working stroke. EMBO J. 23:1729–1738.
53. Phichith, D., M. Travaglia, ., H. L. Sweeney. 2009. Cargo binding
induces dimerization of myosin VI. Proc. Natl. Acad. Sci. USA.
Biophysical Journal 100(2) 430–439
Requirements for Myosin VI Processivity439