Single molecule high-resolution colocalization of Cy3 and Cy5 attached to macromolecules measures intramolecular distances through time.

Abstract

Here we present a technique called single-molecule high-resolution colocalization (SHREC) of fluorescent dyes that allows the measurement of interfluorophore distances in macromolecules and macromolecular complexes with better than 10-nm resolution. By using two chromatically differing fluorescent molecules as probes, we are able to circumvent the Rayleigh criterion and measure distances much smaller than 250 nm. The probes are imaged separately and localized individually with high precision. The registration between the two imaging channels is measured by using fiduciary markers, and the centers of the two probes are mapped onto the same space. Multiple measurements can be made before the fluorophores photobleach, allowing intramolecular and intermolecular distances to be tracked through time. This technique's lower resolution limit lies at the upper resolution limit of single molecule FRET (smFRET) microscopy. The instrumentation and fluorophores used for SHREC can also be used for smFRET, allowing the two types of measurements to be made interchangeably, covering a wide range of interfluorophore distances. A dual-labeled duplex DNA molecule (30 bp) was used as a 10-nm molecular ruler to confirm the validity of the method. We also used SHREC to study the motion of myosin V. We directly observed myosin V's alternating heads while it walked hand-over-hand along an actin filament.

Full text

Page 1

Single molecule high-resolution colocalization of Cy3
and Cy5 attached to macromolecules measures
intramolecular distances through time
L. Stirling Churchman*†, Zeynep O¨kten*‡, Ronald S. Rock*§, John F. Dawson*¶, and James A. Spudich*?
*Department of Biochemistry, Stanford University School of Medicine, Stanford, CA 94305;†Department of Physics, Stanford University, Stanford, CA 94305;
‡Department of Biology, Chemistry, and Pharmacy, Freie Universitaet Berlin, 14195 Berlin, Germany;§Department of Biochemistry and Molecular Biology,
University of Chicago, Chicago, IL 60637; and¶Department of Microbiology and Cellular Biology, University of Guelph, Guelph, ON, Canada N1G 2W1
Contributed by James A. Spudich, December 19, 2004
Here we present a technique called single-molecule high-resolution
colocalization (SHREC) of fluorescent dyes that allows the measure-
ment of interfluorophore distances in macromolecules and macro-
molecularcomplexeswithbetterthan10-nmresolution.Byusingtwo
chromatically differing fluorescent molecules as probes, we are able
to circumvent the Rayleigh criterion and measure distances much
smaller than 250 nm. The probes are imaged separately and localized
individually with high precision. The registration between the two
imaging channels is measured by using fiduciary markers, and the
centers of the two probes are mapped onto the same space. Multiple
measurements can be made before the fluorophores photobleach,
allowing intramolecular and intermolecular distances to be tracked
throughtime.Thistechnique’slowerresolutionlimitliesattheupper
resolution limit of single molecule FRET (smFRET) microscopy. The
instrumentation and fluorophores used for SHREC can also be used
for smFRET, allowing the two types of measurements to be made
interchangeably, covering a wide range of interfluorophore dis-
tances. A dual-labeled duplex DNA molecule (30 bp) was used as a
10-nm molecular ruler to confirm the validity of the method. We also
used SHREC to study the motion of myosin V. We directly observed
myosin V’s alternating heads while it walked hand-over-hand along
an actin filament.
centroid tracking ? dynamic conformational changes ? fluorescence ?
molecular motors ? total internal reflection microscopy
S
structuralconstraintsforanenzymeduringitsbiochemicalcycle.
To probe close to this interface between structure and function,
distancesmustberesolvedontheorderofthesizeoftheenzyme.
However, far field fluorescence microscopy is limited in its
resolution by the Rayleigh criterion at ?250 nm. On the other
end of the size spectrum, single molecule FRET (smFRET)
provides a way to estimate intramolecular distances ?10 nm and
has yielded insights on a range of biological molecules from the
Tetrahymena ribozyme to the ribosome (2–4).
Recent methods, single-molecule high-resolution imaging
with photobleaching (SHRImP) and nanometer-localized mul-
tiple single-molecule (NALMS), have been developed in an
attempt to bridge this ‘‘gap’’ in distance resolution. These
methods are capable of measuring an intramolecular distance to
high precision, but only once, because of the intrinsic destruction
of one of the probes (5, 6). A time series of the intramolecular
distance is not possible with these methods, removing one of the
greatest potential benefits of single molecule imaging.
Addressing the gap in imaging techniques suitable for studying
macromolecules, we introduce a technique, single-molecule
high-resolution colocalization (SHREC) of fluorescent probes,
that can precisely measure intramolecular and intermolecular
distances through time. By separately imaging two chromatically
different fluorophores (Cy3 and Cy5) conjugated to a single
molecule, we are able to localize both probes simultaneously. A
ingle molecule fluorescence techniques are bridging the
areas of structural biology and biochemistry (1). They yield
transformation mapping for the two imaging channels is applied
to the localization results, and the relative positions of the two
probes in the plane of the microscope stage can then be
accurately determined to better than 10 nm, as was confirmed by
measuring the end-to-end distance of dual-labeled duplex DNA.
SHREC uses small probes (fluorescent molecules) that allow
a direct intramolecular distance measurement because they can
be located specifically on the domains being studied. The
dramatic decrease in photons emitted by single fluorescent dyes
compared with large fluorescent beads and nanocrystals used in
previous multicolor colocalization studies (7) presents chal-
lenges to achieving high resolution. The development of fluo-
rescence imaging with one nanometer accuracy (FIONA) has
addressed the challenges involved in precisely localizing a single
fluorescent dye (8–11). FIONA measures the location of a single
molecular domain through time with high precision but cannot
measure how the distance between two points within a molecule
are changing with time.
The power of SHREC is that it can probe distances on the
scale of biological macromolecules with relevant time resolution.
Distance measurements in the range of 10–200 nm, difficult to
probe by other techniques, are now accessible. This is precisely
the range of distances that applies to the wide variety of
macromolecules and macromolecular complexes that are the
players in establishing the processes fundamental to all of cell
and developmental biology. These processes are dynamic, and,
therefore, the changes in the distance measurements need to be
determined with good time resolution. To illustrate SHREC’s
capability in measuring intramolecular distances through time,
we use it here to analyze the myosin V walking mechanism.
Myosin V is a member of the myosin family of molecular motors
and is responsible for cargo transport along actin filaments in a
wide variety of cell types. Myosin V is an ideal enzyme to study
with the SHREC technique because it undergoes multiple
catalytic cycles per encounter with actin, each of which is
coupled to a large conformational change in the protein. The
myosin V heavy chain consists of three major domains: the
catalytic head domain; the light chain binding domain, which
binds six light chains and serves as a lever arm; and the
C-terminal tail domain that forms a dimer with another myosin
V heavy chain by means of coiled-coil interactions. Recent
studies on the myosin V walking mechanism strongly implied
that myosin V uses its lever arms to walk hand-over-hand along
actin (9, 12–15, 17–19). The hand-over-hand walking mechanism
predicts an alternation of the catalytic heads while the protein
walks along the actin filament. Here we directly show the
Freely available online through the PNAS open access option.
Abbreviations: smFRET, single molecule FRET; SHREC, single molecule high-resolution
colocalization.
?To whom correspondence should be addressed at: Stanford University School of Medicine,
BeckmanCenter,279CampusDrive,Stanford,CA94305-5307.E-mail:jspudich@stanford.edu.
© 2005 by The National Academy of Sciences of the USA
www.pnas.org?cgi?doi?10.1073?pnas.0409487102PNAS ?
February 1, 2005 ?
vol. 102 ?
no. 5 ?
1419–1423
BIOPHYSICS

Page 2

alternation of the heads by differentially labeling the two lever
arms and imaging them by using the SHREC technique.
Materials and Methods
Duplex DNA Preparation. To make the 30-bp duplex DNA mole-
cules, two oligonucleotides with the following sequence and
modifications were hybridized (Integrated DNA Technologies,
Coralville, IA). The oligonucleotide 5?-GGGTATGGAGATT-
TTTAGCGGAGTGACAGC-3? was labeled with Cy5 at its 5?
end and with Cy3 at its 3? end. The complementary oligonucle-
otide was labeled with biotin at both ends.
Protein Expression and Purification. The yellow fluorescent protein
(YFP) gene in the p2Bac?pFastBac-YFP-M5-CaM plasmid (a
gift from H. Lee Sweeney, University of Pennsylvania, Phila-
delphia) was deleted by PCR. The resulting p2Bac?pFastBac-
M5-CaM plasmid codes for chicken myosin V that is truncated
at Glu-1099. A leucine zipper followed the native coiled coil to
ensure dimerization. To facilitate purification, the myosin V
protein was N-terminally tagged with a FLAG-tag (DYKD-
DDDK). Two recombinant baculoviruses were generated for
protein expression in Sf9 cells. One encoded the truncated
myosin V and the Drosophila melanogaster calmodulin derived
from the p2Bac?pFastBac-M5-CaM plasmid. The second virus
encoded the human essential light chain derived from the
p2Bac?pFastBac-ELC plasmid. Both viruses were used for coin-
fection of Sf9 cells. The protein was expressed and purified as
described by Sweeney et al. (20).
Calmodulin Labeling and Exchange. Calmodulin was expressed and
labeled with Cy3 or Cy5 as follows: A single cysteine was
introduced in sea urchin calmodulin by means of the mutation
Q143C. This calmodulin was expressed in Escherichia coli and
purified as described in ref. 21. The calmodulin [800 ?M in 50
mM Tris?HCl, pH 7.4/1 mM Tris-(2-carboxyethyl)phosphine]
was labeled with either Cy3-maleimide or Cy5-maleimide (Am-
ersham Biosciences) stock solutions (in DMSO) at a 1.4-fold
molar ratio for 20 min. Excess dye was removed by gel filtration
intoexchangebuffer(below),followedbyovernighthydrophobic
absorption onto BioBeads (Bio-Rad). Label incorporation was
102%forCy3-calmodulinand75%forCy5-calmodulin.Aliquots
were stored at ?80°C.
The exchange of labeled calmodulin onto myosin V was per-
formed as described but with some modifications (22). Myosin V
(150 nM) was incubated with 0.7 nM Cy3-labeled calmodulin and
0.8nMCy5-labeledcalmodulininexchangebuffer(25mMKCl?20
mM imidazole HCl, pH 7.4?2 mM MgCl2?1 mM EGTA) at 22°C
for 2 min. To exchange calmodulins, 0.9 mM CaCl2was added and
the mixture was incubated at 22°C for 5 min. The reaction was
quenched with 7 mM EGTA. The reaction mixture was applied
onto a 100K MWCO Nanocep ultrafiltration device (Pall) and
washed three times with equal volumes of AB (below) to purify
myosin V from excess calmodulin.
Flow Cell Preparation. The flow cell was constructed with a glass
microscope slide and highly refracting coverslips made of
NLAF21 (VA Optical Labs, San Anselmo, CA) held together by
pieces of double-sided Scotch tape.
For the myosin V experiments, 15 ?l of biotin-BSA (1
mg?ml?1) was incubated in the flow cell for 2 min, after which 30
?l of AB buffer (25 mM KCl?25 mM imidazole HCl, pH 7.4?1
mM EGTA?4 mM MgCl2?10 mM DTT?1 mg?ml?1BSA) was
passed through the flow cell to wash it out. Fifteen microliters
of 0.5 mg?ml NeutrAvidin (Molecular Probes) was added to the
cell and was allowed to incubate for 2 min, after which another
wash was done with AB buffer. The flow cell was incubated with
15 ?l of biotinylated phalloidin actin filaments (250 nM) for 5
min, followed by a 100-?l wash with AB buffer. Finally, the cell
was loaded with 20 ?l of imaging buffer [10 mM Tris, pH 8.0?50
mM NaCl?10 mM MgCl2?an oxygen scavenger system consisting
of 17 units of glucose oxidase (Sigma) and 260 units of catalase
(Roche) from a stock that had been filtered with a 0.2-?m
syringe filter and centrifuged for 5 min at 13,000 ? g?0.4%
(vol?vol) glucose?1% (vol?vol) ?-mercaptoethanol] including
calmodulin-exchanged myosin V, 5 ?M calmodulin, 300 nM
ATP, an ATP regeneration system (0.1 mg?ml?1creatine phos-
phokinase?1 mM creatine phosphate), and 0.5% (vol?vol) Tri-
ton-X100toreducenonspecificbinding.Thecellwassealedwith
vacuum grease and imaged immediately. The final motor con-
centration resulted in sparsely decorated actin and thus allowed
an analysis of single myosin V molecules.
For the DNA experiments, the biotin-BSA and NeutrAvidin
were loaded in the same manner as for the myosin V experiments,
but the washes were done with T50 buffer (10 mM Tris, pH 8.0?50
mM NaCl). Once the flow cell had a NeutrAvidin coating, 15 ?l of
dsDNA (30 nM) in T50 buffer with 1 mg?ml?1BSA were flowed in
and incubated for 2 min, after which 100 ?l of imaging buffer was
flowed in. The flow cell was then sealed with vacuum grease and
promptly imaged. The resulting decoration of DNA molecules on
the surface was sufficiently sparse that fluorescent spots from
different molecules rarely overlapped.
Microscope Setup. The total internal reflection fluorescence
(TIRF) microscope was set up as described in ref. 8, with some
modifications (Fig. 5, which is published as supporting informa-
tion on the PNAS web site). Excitation source beams at 532 nm
(Coherent, Santa Clara, CA) and 633 nm (JDS Uniphase, San
Jose, CA) were combined by a dichroic mirror and expanded to
a 7-mm diameter. These sources were focused (focal length ?
500 mm) on the back focal plane of an Olympus 1.65 NA ?100
TIRF objective by means of a laser line dichroic on a linear
translation stage that allows microscope operation in either
epifluorescence or TIRF modes. The objective was positioned
under a closed-loop, two-axis, piezo nanotranslation stage
equipped with capacitive sensors for position measurement
(Physik Instrumente, Auburn, MA). The reflected light exiting
the back aperture of the objective was directed on a quadrant
photodiode to provide a signal for a focus feedback loop that
clamped the distance between the objective and sample with an
electrostrictive actuator (Newport, Fountain Valley, CA). Flu-
orescence emission was collected by the objective, passed
through two StopLine thin film notch filters (Semrock, Roch-
ester, NY), one for each excitation source, and transmitted
through a dual-view apparatus that allowed simultaneous imag-
ing of the Cy3 and Cy5 channels on a single iXon DV 887
EMCCD camera (Andor Technology, Belfast, Ireland). All data
were imaged with a 0.5-sec integration time. Crosstalk between
the two channels (10%) presumably biases the distance mea-
surements toward smaller values. Our experimental error, how-
ever, renders it undetectable, as shown by Monte Carlo simu-
lations in which 100 pairs of modeled images of single
fluorophores separated by 10 nm were created and analyzed
identically as with the DNA data (described below). Simulated
images of fluorescent point spread functions were calculated by
generating an integrated 2D Gaussian intensity distribution with
a constant background to which shot noise was added. The
simulated peaks had the same dimensions and signal-to-noise
ratio as the real peaks. Simulated data with 10% crosstalk and
simulated data without crosstalk are indistinguishable by a
Kolmogorov–Smirnov test (N1? 100, N2? 100, p ? 0.05).
Data Analysis. A registration mapping of the Cy3 and Cy5
channels was performed with fiducials consisting of 100-nm
TransFluoSphere beads (Molecular Probes). They were excited
at 532 nm, and they emit with a broad emission spectrum. The
bead was detectable in both of our channels. We located a single
1420 ?
www.pnas.org?cgi?doi?10.1073?pnas.0409487102 Churchman et al.

Page 3

bead associated to the coverslip, and with our piezo stage we
stepped it with nanometer precision in a grid pattern (with a
0.5-?m spacing), taking an image at every stop. The end result
was a stack of 312 images that shows the bead in both channels
at different positions (Fig. 1a).
The centers of the beads were found by a least squares fit to
a 2D elliptical Gaussian function with a background,
f?x, y? ? z0? A exp?
1
2??
x ? ?x
?x?
2
??
y ? ?y
?y?
2??,
[1]
to determine its location in its respective space (Fig. 1b). In
practice, ?xand ?ywere nearly identical. These locations allowed
us to calculate a local weighted mean mapping (23) from the Cy5
channel to the Cy3 channel (Fig. 1c). This mapping is a weighted
sum of second-order polynomials determined locally around
each fiducial (within a six fiducial radius). It corrects registration
errors that arise locally without allowing their influence to
extend to the rest of the space. The target registration error
(TRE) that accompanies this mapping is calculated in the
following way. Each fiducial is set aside one at a time, and a
mapping is calculated by using the others. The left-out fiducial’s
position in the Cy5 channel is then mapped into the Cy3
channel’s space, and the deviation of the mapped location to the
fiducial’s location in the Cy3 channel is recorded. The mean
value of the deviations’ magnitudes calculated for each fiducial
is the TRE. This mapping can then be applied to any dye
localized in the Cy5 channel to determine where it resides
relative to positions found in the Cy3 channel. The dyes’
locations are found in units of pixels and must be converted by
means of the pixel size. Because we know the differences in the
locations of the bead in real space as we move it around with the
piezo stage, we are also able to get an accurate measurement of
the pixel size (110 nm) with this same calibration. We found that,
in general, this calibration yields a transformation mapping that
is valid over a few weeks. All data reported here, however, were
collected on the same day that a calibration was performed.
The DNA data were analyzed by homemade software using
MATLAB (Mathworks, Natick, MA). The routine automatically
locates peaks in the image by determining the pixels of high
intensity and ensures that the surrounding pixels have intensities
as expected for a single fluorophore. Before fitting the peaks to
a 2D Gaussian function to get its center locations, we search for
pairs of peaks. The pixels found in the Cy5 channel are mapped
onto the Cy3 channel, and a search for peaks in the surrounding
Cy3 pixels is performed. If a peak is found, then the Cy5 peak
and the Cy3 peak are considered a pair for further analysis. Each
peak is fit to a 2D Gaussian function in its respective space as
described for the beads above (Eq. 1). The error on the fit mean
location is calculated with the number of photons (N?) collected,
??? ???x
2? ?y
2???N?? 1.
[2]
The Cy5’s location is mapped to the Cy3 channel, and the distance
betweenthemiscalculated.Afterthecomputationalanalysisofthe
DNA data, the identified fluorophore pairs were screened manu-
allytoensurethatthepairswerenotclosetoanyotherfluorophores
that would have interfered in the pairs’ analysis.
The myosin V data were analyzed by first identifying a moving
motor by eye. Homemade software written in MATLAB then fit
the starting pair of fluorescent peaks to 2D Gaussian functions
as described above and continued to track the peaks for as long
as the user specified. Motors whose conjugated dyes each
photobleached in a single step were analyzed, as was the case for
most of the motors observed, which ensured that we indeed were
studying motors with only one Cy3 label and one Cy5 label. The
resulting trajectories were registered by mapping the Cy5 tra-
jectory onto the Cy3 channel. A least squares linear fit to the
points from both trajectories gave the orientation of the actin
filament to which the trajectories were projected. Distances
along the linear fit (the actin filament) were plotted against time
to see the heads’ relative distances (Fig. 4).
Results
Image Registration Achieved to Within 3.3 nm. Here we describe
imaging and colocalization of two different dyes with distinct
emission spectra, Cy3 and Cy5, with high resolution. In our dual
view imaging setup, the images of the spectrally separated
fluorophores are transmitted through two separate light paths
containing separate sets of optics that distort the images in
different ways (Fig. 5). We mapped this distortion by using
fluorescent beads that have a broad emission spectrum (see
Materials and Methods). These beads can be imaged in both of
our detection channels and thus are considered fiducial markers.
By finding the center locations of the beads at multiple positions
in the two channels, we were able to calculate a mapping from
the Cy5 channel to the Cy3 channel with a target registration
error of 3.3 nm (Fig. 1).
End-to-End Length Measurements of dsDNA 30-mers. We created a
molecular ruler consisting of 30 bp of dsDNA with Cy5 at one
end and Cy3 at the other. We then imaged the Cy5 and the Cy3
dyes on each end of the DNA molecule simultaneously and on
separate halves of the CCD camera. The locations of each dye
were determined by fitting to a 2D Gaussian function. The Cy5’s
location was mapped onto the Cy3 channel, and then the
distance between the dyes was calculated by using the center
Fig. 1.
Fiducial grids with 0.5-?m spacing were made by translating a fluorescent
bead detectable in both channels by a piezo stage. At each point in the grid,
an image of the bead was taken. (a) The resulting stack of images was
projectedthroughztoobservethepositionsthatthecalibrationseriesvisited.
(b) The center locations of the beads were found by means of a fit to a 2D
Gaussianfunction.(c)Thesubsequentpairsoflocationswereusedtocalculate
a transformation mapping of the Cy5 channel onto the Cy3 channel.
Precise alignment of the two imaging channels to within 3.3 nm.
Churchman et al.
PNAS ?
February 1, 2005 ?
vol. 102 ?
no. 5 ?
1421
BIOPHYSICS

Page 4

locations found from the registered 2D Gaussian fits, (xCy3, yCy3)
and (xCy5, yCy5):
d ? ?pixel size???yCy3? yCy5?2? ?xCy3? xCy5?2.
[3]
We solved the expected distance distribution numerically by
means of Monte Carlo simulations. We modeled a Gaussian
distribution of points around two positions separated by a fixed
distance. Pairs of those points were picked at random, and the
distance between them was calculated and graphed as histograms
(Fig. 2a). The reason for the skewed shape of the simulated
histogram is evident from geometric considerations. For a dsDNA
moleculeoflengthslyingalongthexaxiswithoneendattheorigin
and the other at the position x ? s (Fig. 2b), the experimental error
gives a probability of a position measurement of these ends that is
different from their true values. This probability decays from the
true location in a Gaussian manner (see gray cloud in Fig. 2b). If
we measure the first end of the molecule to be at the origin, to
measure the correct end-to-end separation s, the second point will
need to lie on the circumference of the semicircle originating at (0,
0) and cutting through the Gaussian gray cloud at a radius s. The
points in the Gaussian gray cloud lying inside this semicircle will
givedistances?s,andthepointslyingoutsidewillgivedistances?s.
MoreoftheGaussiangraycloudliesoutsidethecircle,soitismore
likely to measure a distance ?s; hence, the distribution of possible
distance measurements from the origin will have a skew toward
large values.
A useful formula can be derived relating the end-to-end
distance (or separation s) to measured quantities in SHREC. By
mathematically defining the mean distance squared between a
point from a 2D Gaussian probability distribution with a vari-
ance ?1
distribution with a variance ?2
?d2? ??
?? ???? ??
2and a point from a second 2D Gaussian probability
2a distance s away as
???
???
???
??
??x1? x2?2? ?y1? y2?2?
?exp(??x1
2? y1
2??2?1
2)exp(???x2? s?2? y2
4?2?1
2??2?2
2)
2?2
2
? dx1dy1dx2dy2? s2? 2??1
2? ?2
2?,
one can derive an expression for the end-to-end separation, s,
based on the mean of the distribution and the variance of the
distribution along with the localization errors:
s ? separation ??d?2? variance?d? ? 2??1
2? ?2
2?.
[4]
The above formula was tested with Monte Carlo simulations of
various absolute separations, and it correctly yields the true
separation (Fig. 2c).
We measured the end-to-end lengths of 482 DNA molecules
(Fig. 3). The data are skewed toward large values, which is
expected from the above description. Applying Eq. 4 to our data,
we measure an end-to-end distance of 10 ? 1 nm (95% confi-
dence limits were determined from 1,000 bootstrap distribu-
tions) using the following mean errors on our localizations:
??Cy3? ? ????Cy3? 5 nm, ??Cy5? ? ??????Cy5
? ??52? 3.32? ? 6 nm,
2
?TRE2?
Fig. 2.
distribution of distance measurements. (a) The distance probability distribu-
tion was calculated by means of Monte Carlo simulations. (b) The skew in the
histograms toward large values can be understood from a geometric argu-
ment.IfoneendoftheDNAmoleculeismeasuredtoresideattheorigin,then
only points lying on the circumference of a circle with radius s and origin (0,
0) will yield the true end-to-end distance. It is more likely that a point will lie
outside of the dashed-line semicircle than inside it, which gives the distance
distribution a long tail. (c) Despite the non-Gaussian nature of the distance
distribution,theend-to-endseparationcanbecalculatedusingthegeometric
mean, variance, and localization errors.
Determining the accurate end-to-end distance from the skewed
Fig. 3.
measured by SHREC.
End-to-end distances of 482 duplex DNA molecules (30 bp) as
1422 ?
www.pnas.org?cgi?doi?10.1073?pnas.0409487102 Churchman et al.

Page 5

where??istheerroronthefitmeanlocationandTREisthetarget
registrationerror.Withapersistencelengthof50nm(16),the30-bp
duplexDNAmoleculecanbeconsideredarigidrodwithacontour
length of 10 nm, assuming a 0.34-nm rise per base. All data were
imaged with a 0.5-sec integration time, which is long enough such
that all thermal motion of the DNA molecules, their conjugated
probes, and linkers are averaged out. The expected measurement
of the end-to-end length of the DNA molecules is thus the contour
length, 10 nm, which is in excellent agreement with our result.
Watching both Heads of Myosin V Walk Hand-over-Hand. Myosin V
is an excellent model enzyme with which to test SHREC’s ability
to track distances through time, because its hand-over-hand
walking is confirmed and well understood (9, 12–15, 17–19). It
takes 36-nm steps, which is well above our experimental error.
By exchanging the native calmodulins with a mixture of Cy3- and
Cy5-labeled calmodulins, we can observe two different legs of
the same myosin molecule. The Cy3 and the Cy5 data are
obtained by analyzing each fluorescent channel’s ‘‘movie’’ sep-
arately, mapping the results on top of one another, and then
projecting each of the trajectories onto the inferred position of
the actin filament. A time trace is made by plotting the location
of each probe along the actin filament against time. The sample
trace shows the trajectory of a single myosin V molecule as an
alternating staircase, which is expected from the hand-over-hand
model of processivity (Fig. 4). Each leg’s step size is ?72 nm, as
expected for calmodulins exchanged close to the motor domain.
Additionally, the motor domains are separated from each other
by ?36 nm along the axis of the actin filament, as expected.
Discussion
The distance that SHREC measures is a 2D projection of the true
intramolecular distance. In our studies, we have arranged the
geometries such that the intramolecular distances lie in the x–y
plane, a common geometry for molecular motor studies and other
single molecule measurements. In other cases, an absolute distance
might not be measurable due to limitations in the possible exper-
imental geometries. Nevertheless, a change in the 2D projection of
the intramolecular distance would also yield valuable information.
The limited number of emitted photons per fluorescent dye
determines the possible temporal resolution and the total ob-
servation time. We show here that temporal resolutions of
interest in biomolecular processes are achievable. Thus, it is
possible to track the colocalization of two single fluorophores
through time with high resolution.
Knowledge of an enzyme’s structural constraints while it
progresses through its biological function is an excellent com-
plement to a crystal structure, which gives only a snapshot of the
structure in one of its conformations. SHREC measurements
can be performed on the same microscopes used in smFRET
studies and with the same fluorophores, providing a smooth
transition from smFRET analysis to SHREC analysis. When
smFRET begins to lose effectiveness (?10 nm), SHREC can
resolve the fluorophores’ locations. Combining smFRET and
SHREC, structural information can be tracked for long periods
of time and over broad distance scales.
We thank Lee Sweeney for the pFastBac-YFP-M5-CaM plasmid con-
struct. We are grateful to Michael Levene (Yale University, New Haven,
CT) for his thoughtful comments on the manuscript. We thank Zev
Bryant, David Altman, and other members of J.A.S.’s laboratory for
their critical comments on the work. This work was supported by
Boehringer Ingelheim Fonds (Z.O ¨.), a Burroughs Wellcome Career
Award at the Scientific Interface (to R.S.R.), the Natural Sciences and
Engineering Research Council of Canada (J.F.D.), and National Insti-
tutes of Health Grant GM33289 (to J.A.S.).
1. Weiss, S. (1999) Science 283, 1676–1683.
2. Blanchard, S. C., Kim, H. D., Gonzalez, R. L., Jr., Puglisi, J. D. & Chu, S. (2004)
Proc. Natl. Acad. Sci. USA 101, 12893–12898.
3. Blanchard, S. C., Gonzalez, R. L., Kim, H. D., Chu, S. & Puglisi, J. D. (2004)
Nat. Struct. Mol. Biol. 11, 1008–1014.
4. Zhuang, X., Bartley, L. E., Babcock, H. P., Russell, R., Ha, T., Hershlag, D.,
Chu, S. (2000) Science 288, 2048–2051.
5. Qu, X., Wu, D., Mets, L. & Scherer, N. F. (2004) Proc. Natl. Acad. Sci. USA
101, 11298–11303.
6. Gordon, M. P., Ha, T. & Selvin, P. R. (2004) Proc. Natl. Acad. Sci. USA 101,
6462–6465.
7. Lacoste, T. D., Michalet, X., Pinaud, F., Chemla, D. S., Alivisatos, A. P. &
Weiss, S. (2000) Proc. Natl. Acad. Sci. USA 97, 9461–9466.
8. Okten, Z., Churchman, L. S., Rock, R. S. & Spudich, J. A. (2004) Nat. Struct.
Mol. Biol. 11, 884–887.
9. Yildiz, A., Forkey, J. N., McKinney, S. A., Ha, T., Goldman, Y. E. & Selvin,
P. R. (2003) Science 300, 2061–2065.
10. Yildiz, A., Tomishige, M., Vale, R. D. & Selvin, P. R. (2004) Science 303,
676–678.
11. Yildiz, A., Park, H., Safer, D., Yang, Z., Chen, L. Q., Selvin, P. R. & Sweeney,
H. L. (2004) J. Biol. Chem. 279, 37223–37226.
12. DeLaCruz,E.M.,Wells,A.L.,Rosenfeld,S.S.,Ostap,E.M.&Sweeney,H.L.
(1999) Proc. Natl. Acad. Sci. USA 96, 13726–13731.
13. Mehta, A. D., Rock, R. S., Rief, M., Spudich, J. A., Mooseker, M. S. & Cheney,
R. E. (1999) Nature 400, 590–593.
14. Rief, M., Rock, R. S., Mehta, A. D., Mooseker, M. S., Cheney, R. E. & Spudich,
J. A. (2000) Proc. Natl. Acad. Sci. USA 97, 9482–9486.
15. Moore, J. R., Krementsova, E. B., Trybus, K. M. & Warshaw, D. M. (2001)
J. Cell Biol. 155, 625–635.
16. Hagerman, P. J. (1988) Annu. Rev. Biophys. Biophys. Chem. 17, 265–286.
17. Purcell, T. J., Morris, C., Spudich, J. A. & Sweeney, H. L. (2002) Proc. Natl.
Acad. Sci. USA 99, 14159–14164.
18. Veigel, C., Wang, F., Bartoo, M. L., Sellers, J. R. & Molloy, J. E. (2002) Nat.
Cell Biol. 4, 59–65.
19. Forkey, J. N., Quinlan, M. E., Alexander Shaw, M., Corrie, J. E. & Goldman,
Y. E. (2003) Nature 422, 399–404.
20. Sweeney, H. L., Rosenfeld, S. S., Brown, F., Faust, L., Smith, J., Xing, J., Stein,
L. A. & Sellers, J. R. (1998) J. Biol. Chem. 273, 6262–6270.
21. Gopalakrishna, R. & Anderson, W. B. (1982) Biochem. Biophys. Res. Commun.
104, 830–836.
22. Sakamoto, T., Amitani, I., Yokota, E. & Ando, T. (2000) Biochem. Biophys.
Res. Commun. 272, 586–590.
23. Goshtasby, A. (1988) Image and Vision Computing 6, 255–261.
Fig. 4.
along an actin filament. The labels (Cy3 and Cy5) are covalently attached to
calmodulins that were exchanged onto the myosin V molecule. In this trace,
both of the fluorescent probe’s locations are taking 72-nm steps, indicating
that the calmodulins were exchanged close to the motor domain. The alter-
nating positions of the probes provide a direct observation of myosin V’s
hand-over-hand walking mechanism.
Time trace of a differentially labeled myosin V molecule walking
Churchman et al.
PNAS ?
February 1, 2005 ?
vol. 102 ?
no. 5 ?
1423
BIOPHYSICS