An Integrated Laser Trap/Flow Control Video Microscope for the Study of Single Biomolecules

Department of Physics, University of California, Berkeley, California 94720 USA.
Biophysical Journal (Impact Factor: 3.97). 09/2000; 79(2):1155-67. DOI: 10.1016/S0006-3495(00)76369-7
Source: PubMed
ABSTRACT
We have developed an integrated laser trap/flow control video microscope for mechanical manipulation of single biopolymers. The instrument is automated to maximize experimental throughput. A single-beam optical trap capable of trapping micron-scale polystyrene beads in the middle of a 200-microm-deep microchamber is used, making it possible to insert a micropipette inside this chamber to hold a second bead by suction. Together, these beads function as easily exchangeable surfaces between which macromolecules of interest can be attached. A computer-controlled flow system is used to exchange the liquid in the chamber and to establish a flow rate with high precision. The flow and the optical trap can be used to exert forces on the beads, the displacements of which can be measured either by video microscopy or by laser deflection. To test the performance of this instrument, individual biotinylated DNA molecules were assembled between two streptavidin beads, and the DNA elasticity was characterized using both laser trap and flow forces. DNA extension under varying forces was measured by video microscopy. The combination of the flow system and video microscopy is a versatile design that is particularly useful for the study of systems susceptible to laser-induced damage. This capability was demonstrated by following the translocation of transcribing RNA polymerase up to 650 s.

Full-text

Available from: Gijs Jan Lodewijk Wuite
An Integrated Laser Trap/Flow Control Video Microscope for the Study of
Single Biomolecules
Gijs J. L. Wuite,* R. John Davenport,
Aaron Rappaport,
and Carlos Bustamante*
Institute of Molecular Biology, University of Oregon, Eugene, Oregon 97403, and
Department of Molecular and Cell Biology and
*Department of Physics, University of California, Berkeley, California 94720 USA
ABSTRACT We have developed an integrated laser trap/flow control video microscope for mechanical manipulation of
single biopolymers. The instrument is automated to maximize experimental throughput. A single-beam optical trap capable
of trapping micron-scale polystyrene beads in the middle of a 200-
m-deep microchamber is used, making it possible to
insert a micropipette inside this chamber to hold a second bead by suction. Together, these beads function as easily
exchangeable surfaces between which macromolecules of interest can be attached. A computer-controlled flow system is
used to exchange the liquid in the chamber and to establish a flow rate with high precision. The flow and the optical trap can
be used to exert forces on the beads, the displacements of which can be measured either by video microscopy or by laser
deflection. To test the performance of this instrument, individual biotinylated DNA molecules were assembled between two
streptavidin beads, and the DNA elasticity was characterized using both laser trap and flow forces. DNA extension under
varying forces was measured by video microscopy. The combination of the flow system and video microscopy is a versatile
design that is particularly useful for the study of systems susceptible to laser-induced damage. This capability was
demonstrated by following the translocation of transcribing RNA polymerase up to 650 s.
INTRODUCTION
Since its demonstration by Arthur Ashkin in 1986 (Ashkin
et al., 1986), the optical trap has become an increasingly
important tool in biophysics and cell biology. Experimental
laser trapping techniques have been used to directly manip-
ulate and trap cells, organelles inside cells, and viruses (Kuo
and Sheetz, 1992; Ashkin and Dziedzic, 1987). In addition,
optical trapping, combined with microsphere handles linked
to molecules of interest, has been used to measure the
force-producing properties of various biological systems,
including kinesin moving along microtubules (Svoboda et
al., 1993), actomyosin complexes (Finer et al., 1994), RNA
polymerase (Davenport et al., 2000; Wang et al., 1998; Yin
et al., 1995), and DNA polymerase (Wuite et al., 2000), as
well as to measure the elastic properties of DNA (Cluzel et
al., 1996; Smith et al., 1996) and the giant muscle protein
titin (Kellermayer et al., 1997; Tskhovrebova et al., 1997).
Optical traps can stably trap particles of a wide range of
sizes, from diameters much smaller than the wavelength of
light (Rayleigh-size regime) to diameters much larger than
the wavelength of light (ray optics regime). The size of
particles manipulated in many biophysical applications of
optical trapping microscopy are on the order of the wave-
length of light and therefore fall between these two regimes.
A theoretical description of trapping in this intermediate
regime is difficult. Ray optics can nonetheless be used to
obtain a qualitative model of the two main forces acting on
intermediate-size beads (Ashkin, 1992). One force is largely
due to backscattering of photons by the sphere and tends to
push it down the optical axis; this force is often referred to
as the scattering force, F
scat
. A second force, the intensity
gradient force, F
grad
, tends to pull the object in the direction
of increasing intensity of the beam. For a laser beam fo-
cused by a high-numerical-aperture lens, a substantial in-
tensity gradient develops perpendicular to the beam axis, as
well as along the beam axis, pulling the bead toward the focus
and counteracting the backscattering force. When these forces
balance each other, a dielectric object can be held at a stable
position just downstream of the objective focus.
The forces involved in many biological processes fall
within the range that can be exerted by an optical trap. For
instance, the reported stalling forces for kinesin, myosin,
RNA polymerase, and DNA polymerase are 5–6 pN (Svo-
boda et al., 1993), 3.5 pN (Finer et al., 1994), 14–25 pN
(Davenport et al., 2000; Wang et al., 1998; Yin et al., 1995),
and 34 pN (Wuite et al., 2000), respectively. Force-induced
structural transition in nucleic acids (Cluzel et al., 1996;
Smith et al., 1996) or proteins (Kellermayer et al., 1997;
Tskhovrebova et al., 1997) require forces up to 80 pN or so.
These forces can be measured by optical trapping techniques,
because for a small external force a trapped object is displaced
a distance proportional to that force. This displacement (and
hence the external force) can be measured by detecting corre-
sponding changes in the polarization of the laser beam (Denk
and Webb, 1990; Svoboda et al., 1993) or deflection of the
laser beam by the use of photodiodes (Davenport et al., 2000;
Ghislain et al., 1994; Smith et al., 1996).
Hydrodynamic flow has also been successfully used to
apply forces to single molecules. For instance, the dynamic
response of single DNA molecules to a hydrodynamic drag
force has been studied in laminar flows (Perkins et al., 1997;
Smith and Chu, 1998). The force-extension behavior of
Received for publication 19 March 1999 and in final form 12 May 2000.
Address reprint requests to Dr. Carlos Bustamente, Department of Physics,
University of California, 173 Birge Hall, Berkeley, CA 94720. Tel.: 510-
643-9706; Fax: 510-642-5943; E-mail: carlos@alice.berkeley.edu.
© 2000 by the Biophysical Society
0006-3495/00/08/1155/13 $2.00
1155Biophysical Journal Volume 79 August 2000 1155–1167
Page 1
DNA has been characterized with a combination of mag-
netic and drag forces on beads tethered to DNA (Smith et
al., 1992). Because laminar flows are used to generate
forces in these single-molecule studies, drag forces can be
calculated using Stokes’ law. For any size sphere in water,
Stokes’ law remains valid for forces up to 10 nN, at which
point the Reynolds’ number exceeds unity. Single-molecule
experiments are almost always carried out in the confined
space of a sample chamber. Therefore, force determination
should take into account the modification of Stokes’ law
due to the hydrodynamic coupling between the bead and the
boundaries of the microchamber (Happel and Brenner,
1991; Lorentz, 1907; Smith et al., 1992).
Both optical trapping and flow control systems represent
important technical developments in biophysics that are
readily applicable to the study of single-molecule mechan-
ics and force-generating systems. Moreover, these methods
have complementary features: laser traps afford excellent
spatial resolution and force control, but can induce radiation
damage in the samples. Problems also arise from the use of
oil immersion objective lenses, which form optical traps
near the surface of the sample chamber. Calibration of such
traps is difficult because small changes in distance from the
coverslip result in large differences in hydrodynamic cou-
pling of the beads to the glass surface and in trap stiffness
due to spherical aberration. Hydrodynamic flows, on the
other hand, greatly extend the range of forces that can be
applied with laser traps, without inducing radiation damage.
However, video detection combined with flow systems has
lower spatial resolution and reduced force control as com-
pared to laser trapping detection systems. In addition, there
are experimental drawbacks common to the two techniques.
In both setups, experiments in which molecules are tethered
to the glass surface of the fluid chamber require setting up
a new chamber each time the experimental sample has
become inactive or unusable, increasing experimental times
and reducing overall throughput. Furthermore, if a polymer
molecule is attached to a surface and aligned at an angle to
the vector of force application, a further correction for this
angle must be made. One solution of this problem was
offered by Ishijima et al. (1998), who used a surface with a
microfabricated pedestal.
It is therefore advantageous to design and build an in-
strument that combines the complementary capabilities of
both optical trapping and hydrodynamic flow control while
being easy to calibrate, having improved experimental ge-
ometry, faster data throughput, and controllable laser expo-
sure. We communicate here the design of an integrated laser
trap/flow control video microscope that is automated to
maximize its throughput in single-biomolecule experiments.
This instrument was previously used in the study of RNA
polymerase by Davenport et al. (2000). This instrument can
be used in one of three operation modes: 1) laminar flow
mode, where forces are exerted through the hydrodynamic
drag experienced by a bead and the bead positions are
determined by video microscopy; 2) laser trap mode, where
the laser power is used both to exert force on the bead and
to determine the bead position; and 3) combined laminar/
optical mode, in which the force is exerted both hydrody-
namically and optically, but the bead position is tracked by
the laser beam, at high or low power. The use of exchange-
able beads in this design greatly decreases the experimental
setup times. Surface effects are minimized by trapping these
beads deep inside the microchamber, through the use of a
water immersion, long-working-distance objective. Finally,
through the use of either the laminar flow mode or the com-
bined mode, radiation damage can be minimized. Two exper-
iments illustrate the versatility of this system. The elasticity of
DNA was characterized by using both the combined mode and
the laminar flow mode. Furthermore, transcriptional transloca-
tion by RNA polymerase was followed over long times with-
out radiation damage, using the laminar flow mode.
INSTRUMENTAL DESIGN
Optics
The optics of the laser trap/flow control video microscope
are depicted in Fig. 1. The optical trap in this instrument can
stably trap refractive objects far from the cover slide (up to
150
m) with a single laser beam. This capability results
from the use of a high numerical aperture water immersion
objective lens, which has much less spherical aberration
than an oil immersion objective (Brenner, 1994). This type
of objective is commonly used for imaging through thick
water layers or thick samples (Perkins et al., 1997; Smith
and Chu, 1998). High NA oil immersion objectives are
commonly used to generate high-angle peripheral rays,
which are necessary to create a refractive force capable of
counteracting the backscattering force on a dielectric object.
An oil immersion lens forms a stiffer trap close to a surface
than is possible with a water immersion objective. Deeper
inside the sample chamber, however, rays are not directed
toward a single focus, and an optical trap formed by an oil
immersion lens rapidly decreases in stiffness. Particles can
no longer be stably trapped at distances beyond several
microns from the surface (e.g., a 1-
m polystyrene sphere
cannot be trapped at a depth greater than 21
m from the
surface; Ghislain et al., 1994). A single beam optical trap
that uses a water immersion, high-NA objective (typically
1.2) can be used to focus the axial and peripheral rays to the
same focus both close to and far away from the coverglass
(Davenport et al., 2000). As a result, the trap stiffness is
independent of the relative distance between the trap and the
coverslide surface. Dual beam traps, which use two coun-
terpropagating, confocal laser beams, suffer from spherical
aberration as well, but the gradient forces dominate because
the scattering forces of the two beams cancel. Therefore
they can trap objects stably far from the glass surface, using
low numerical aperture lenses (Smith et al., manuscript in
1156 Wuite et al.
Biophysical Journal 79(2) 1155–1167
Page 2
preparation). Dual beam traps, however, are somewhat dif-
ficult to align and are more sensitive to optical drift. The
single-lens design is easy to set up and align.
Micropipette and exchangeable surfaces
Trapping objects deep inside the sample chamber makes the
introduction of a micropipette possible (Fig. 2). A bead held
by suction on the tip of the pipette can be used as an easily
exchangeable surface of well-defined chemical properties to
which molecules can be attached. With the use of beads as
replaceable attachment surfaces, many single-molecule exper-
iments can be performed without replacing the entire fluid
chamber. Moreover, this design can easily be adapted to a large
variety of molecular systems by simply changing the chemistry
of the beads or appropriately labeling macromolecules.
The assembly of macromolecules between beads can be
performed in a very controlled fashion by using a two-bead
FIGURE 1 Schematic of integrated optical trapping/flow control video microscope. The optical trapping microscope is constructed horizontally on an
optical breadboard (Newport Research Corporation, Irvine, CA) and suspended by elastic shock cords. Commercial microscope configurations have been
avoided to minimize undesirable mechanical vibration. The laser beam (LAS) (from a single-mode, circular corrected, linearly polarized diode,
835
nm, 178 mW maximum power; Melles Griot, Boulder, CO) passes through a computer-controlled shutter (SH) (Melles Griot), is reflected by a broadband
polarizing beamsplitter cube (PBS), and is focused (LS) 160 mm upstream from the objective lens. The objective, a water immersion, high numerical
aperture (NA) lens (OL) (PlanApo 60, NA 1.2, Nikon; Meridian Instruments, Okemos, MI) focuses the laser beam in the middle of a 200-
m-deep
sample chamber (SC). The working distance of the objective is 220
m from the coverslide surface. The exiting laser light is collected by a large-aperture
oil immersion condenser (CL). The beam is then reflected by a short wave pass filter (SWF) (SPF 650; CVI Laser Corp., Albuquerque, NM) onto a 2D
position-sensitive photodetector (PD) (DL10; UDT Sensors, Hawthorne, CA). A fiber optic visible light source (FO) (Tri-Lite; World Precision Instruments,
Sarasota, FL) propagating in the direction opposite that the laser beam is used to illuminate the sample. The image formed in transmission by the objective
is then projected through a beamsplitting cube onto a CCD camera (CCD) (Watec monochrome CCD; Edmund Scientific, Barrington, NJ). Polyethylene
tubing (ID 0.28 mm, OD 0.61 mm; Becton Dickinson, Sparks, MD) and flangeless fittings (Upchurch Scientific, Oakharbor, WA) are used to connect
all components of the flow system. The total dead volume between the pressure bottles (PFLOW) and the middle of the sample chamber is 55
l. The
pressure in the bottles is regulated by three computer-controlled solenoid valves (PVL) (Clippard, Cincinnati, OH): one is connected to a 0.5-atm pressure
source (PRS), another to a 0.5-atm vacuum (VAC), and the third to atmospheric pressure (ATM). The pressure or vacuum solenoid valve is opened for
7 ms to incrementally increase or decrease, respectively, the pressure in the line; this in turn increases or decreases the rate of buffer flow through the
chamber. Buffer flow is stopped by opening the atmosphere solenoid valve. The pressure exerted on the solutions in the bottles is monitored by the computer
via a pressure transducer (accuracy 0.1%). The instrument functions and data collection are controlled by the personal computer (COM) (Intel 486, 66
MHz; Comtrade, City of Industry, CA) and custom-written software. The computer controls the position of the sample chamber via an internal I/O card
(CIO-DIO96; Computer Boards, Mansfield, MA) connected to custom-built digital-to-analog converters. The signal from these converters modulates the
voltage output of 1000-V power operational amplifiers (BOP-1000M; Kepco, Flushing, NY), which control the x and y piezo actuators (40
m piezos;
Burleigh, Fisher, NY) on the flexure stage (STxy). To allow for a larger degree of automation in this instrument, the I/O card also interfaces between a
variety of other components of the instrument, including a piezoelectric controller (MDT691; Thorlabs, Newton, NJ), which allows computer control of
the objective focus via a piezoelectric actuator (PA) (PE4; Thorlabs), the shutter, and the pressure solenoid valves. The CCD camera is connected through
a video monitor (VID) to a video board (DT55–60; Data Translation, Marlboro, MA) in the computer. The camera and the video board are used to determine
the position and size of objects in the video image (bandwidth 60 Hz; maximum sample frequency 30 Hz). An analog-to-digital converter card (DAS-802;
Keithley, Taunton, MA) is used for data collection from the photodetector and from the pressure meter. The selector valves (VL 1, 2, and 3) (MVP modular
positioner; Hamilton, Reno, NV) and the automatic syringe (MOT SYR) (PSD/2; Hamilton) are connected in a daisy chain to the computer parallel port.
Liquids from the bottles controlled by selector valve 3 are introduced into the fluid chamber with the automatic syringe, which functions as a pump.
Integrated Microscopy of Single Biomolecules 1157
Biophysical Journal 79(2) 1155–1167
Page 3
design. We have attached single biomolecules between a
bead in the trap and a bead on the pipette as depicted in Fig.
3. At the beginning of an experiment, streptavidin-coated
polystyrene microspheres are injected into the fluid cham-
ber, and when the laser trap is turned on, one such bead is
captured by the trap. Next, the pipette is brought close to the
trap and the bead in the trap is transferred to the tip of the
micropipette. A second bead is then captured in the laser
trap. With a bead on the micropipette and one in the laser
trap, appropriately labeled molecules (either linear DNA
with biotin tags at either end or biotinylated transcription
complexes for the applications demonstrated here; Appen-
dix A) are introduced into the flow chamber. When one end
of a molecule attaches to the bead, the molecule extends by
viscous drag in the direction of the flow (Fig. 4). The pipette
bead can then be moved near the trapped bead to promote
attachment of the free end of a molecule to its surface
(Appendix A). This procedure can be used for a wide range
of biopolymer lengths; DNA molecules as small as 100 nm
have been tethered between beads.
During the tethering process of force-sensitive biological
systems such as RNA polymerase between two beads, care
must be taken to avoid irreversible damage to the molecule
by minimizing the force applied to it. To this end, the
tethering procedure described above has been completely
automated and is controlled by the computer. During the
procedure, the computer controls the back and forth move-
ment of the pipette bead relative to the trapped bead via the
fine piezo control of the stage position, while the force on
the trapped bead is monitored by the computer. When a
tether is formed the bead is displaced from the center of the
trap and experiences a sudden increase in force. Because the
FIGURE 2 The sample chamber is assembled by placing two layers of
parafilm cut in the chamber shape between a microscope slide and a
coverglass. The chamber is sealed to withstand the fluid pressure by
heating the two parafilm layers that are pressed between the slide and the
coverglass. A 100-
m-inner diameter quartz catheter tube is placed be-
tween the two layers of parafilm to allow the introduction of a micropipette
into the chamber. Two holes drilled through the coverglass form the
entrance and exit ports for the flow chamber. The chamber is clamped on
a holder by two aluminum brackets that have threaded holes to connect
tubing to the chamber by means of flangeless fittings. Micropipettes are
pulled from glass capillaries (ID 40 6.4
m, OD 80 6.4
m,
KG-33 glass; Garner Glass Co., Claremont, CA) in a hand-built resistance
element/gravity pipette puller. The final pipette internal diameter varies
from 0.5 to 1
m tapered over 1.5–2 mm. The pipettes are glued into
polyethylene tubing (ID 0.28 mm), using UV-curing epoxy (Norland
Products, New Brunswick, NJ). Polystyrene beads of 1–10
m diameter
can be held on the tip of the pipette by suction. The pipettes are introduced
into the sample chamber through the catheter tube and clamped on the
sample chamber holder with a Plexiglas bracket. The catheter tube is sealed
with agarose to prevent leakage of fluid from the chamber. A syringe
attached to the other end of the tubing is used to apply suction and pressure
in the pipette. At the end of the usable lifetime of a pipette, a new chamber
can be constructed or the micropipette in the existing chamber can be
replaced. The sample chamber itself is mounted on a custom-built x-y
flexure stage driven by 40-
m piezos. The position of the stage is under
computer control and moves the pipette relative to the fixed position of the
laser trap in the plane perpendicular to the optical axis. The relative z
positions of the optical trap and the pipette are controlled by the z position
of the focus of the objective.
FIGURE 3 Assembly of DNA and DNA-protein complexes. (A) To start
an experiment, polystyrene beads modified for a specific biological exper-
iment are allowed to flow into the fluid chamber. One bead is held on the
tip of the micropipette by suction and one is held in the laser trap. (B)
Macromolecules (i.e., DNA or stalled transcription complexes; Appendix
A) flow in, and their appropriately modified ends can attach to the beads.
Upon attachment the DNA is stretched out in the flow because of the drag
force experienced by the beads. (C) The bead on the tip of the pipette is
moved close to the bead in the trap to attach the free end of the macro-
molecule to the bead in the trap. (D) An attachment can be confirmed when
the bead in the trap experiences a fast increase of force over a short
distance. Once an attachment is made, measurements can be taken.
1158 Wuite et al.
Biophysical Journal 79(2) 1155–1167
Page 4
computer can monitor this force change with higher preci-
sion than a manual operator can, the automated procedure
limits the maximum force experienced by the molecule to
10 pN (manual tethering 20 pN). In addition to mini-
mizing the applied force, the computer-controlled procedure
is less error prone than it would be if it were carried out
manually, and thus it increases molecule assembly throughput.
There are some drawbacks to the use of a micropipette.
The tip of the pipette is very fragile and can easily be
damaged if it is handled outside the chamber. Inside the
chamber, however, it can be used over many months by
rinsing the chamber and pipette with distilled, deionized
water after use and storing them dry to prevent bacterial
contamination. Another problem is that a bead held by
suction does not always seal the pipette tip completely. This
introduces a residual suction that can disrupt an experiment
(i.e., a DNA molecule tethered between two beads can be
sucked into the pipette). This problem can easily be over-
come by decreasing the vacuum in the pipette to only the
amount necessary to keep the bead on the pipette. This
greatly diminishes the residual suction of the pipette when it
is not sealed entirely. While a second optical trap could be
substituted for the pipette, a pipette is much stiffer (4
nN/
m) than an optical trap. In addition, forces of 150 pN
or more do not dislodge a bead held on the pipette by
suction. Thus the pipette functions as a stationary wall.
Flow system
Flow and buffer exchange in the sample chamber are con-
trolled via an automated syringe, pressure bottles, and au-
tomated selector valves (Fig. 1). The automated syringe is
used to inject volumes of suspended beads, to clean the
chamber, and to remove air bubbles from the flow system.
Buffer flow from the pressure bottles is achieved by hydro-
static pressure on the buffer. A desired buffer is selected by
connecting that pressure bottle to the buffer line entering the
chamber via the automated selector valves. The fluid waste
exits the chamber and is disposed in a container. The three
selector valves can be used to program automatic cycles of
buffer exchange through the chamber.
The flow rate is linear with applied pressure and vacuum
(Fig. 5 A) and is capable of producing hydrodynamic drag
forces up to 100 pN on a 1.1-
m-radius bead in both the
forward and backward flow directions. Flow speeds used in
this instrument are less than 4000
m/s, using 1.1-
m beads
(2.4
l/s). The total volume of fluids in the pressure bottles
(4 ml) is small compared with the volume of air in the
pressure system (80 cc), and the average flow rates used
in this system are small (1
l/s). Therefore, the change in
FIGURE 4 A micron-sized bead is kept in a laser trap while 10-kb-
long DNA molecules pass through the fluid chamber. Upon connection of
a biotinylated DNA molecule to the streptavidin on the beads, the measured
drag force of the bead increases (1). This distinct increase in drag allows
precise knowledge of the number of molecules attached to the bead.
FIGURE 5 (A) At each pressure, the flow speed is measured as described
for the flow speed calibration. The flow speed is linear with applied
pressure and suction. (B) A 2.2-
m bead is trapped in a 178-mW laser
beam while a flow is applied through the fluid chamber. Analysis of
short-term flow stability: data were taken at a 1-kHz sample rate. The drag
force experienced by the bead is 10.6 pN. The Brownian motion of the
trapped bead causes a 0.6-pN fluctuation (RMS value) or 6nm(X
rms
,
DNA
110 pN/
m). Inset: Long-term flow stability. The data points are
averaged over a 0.2-s period. The fluctuation in drag force is on the order
of 0.04 pN.
Integrated Microscopy of Single Biomolecules 1159
Biophysical Journal 79(2) 1155–1167
Page 5
gas volume due to displaced fluid is small, and the flow is
stable over many minutes (Fig. 5 B). Moreover, the flow
rate can be ramped by increasing or decreasing the pressure.
Sets of buffer bottles for different experiments are contained
on easily exchangeable fluid cartridges, making the transfer
between new experimental conditions rapid and efficient.
Position and force detector
For the high-resolution detection of bead displacements and
forces we use the laser deflection method of Ghislain et al.
(Ghislain et al., 1994; Ghislain and Webb, 1993), except
that a 2D lateral effect position detector is used instead of a
single photodiode. This detector determines the weighted
average center of mass of the laser spot on the detector in
two dimensions. Unlike the method of Ghislain et al., the
active area of the detector is larger than the intercepted laser
beam and thus is insensitive to bead movements along the
optical axis. A microscope condenser collects the exiting
light after it is refracted by the trapped bead. The condenser
lens casts the laser beam directly on the 2D position detec-
tor, which is placed on the optical axis far behind the
intermediate image plane, so that the defocused laser beam
is 0.8 cm
2
, somewhat smaller than the size of the detector.
Bead displacements relative to the fixed position of the
optical trap deflect the laser beam on the detector. Placing
the detector behind the image plane instead of in the back-
focal plane makes the detector sensitive to translations of
the x-y position of the optical axis, which can be considered
a drawback of this method (Gittes and Schmidt, 1998;
Visscher et al., 1996). The position-sensitive detector em-
ployed here has instrumental noise comparable to that of
interferometry-based systems (0.015 nm/Hz
1/2
) while
having sensitivity in two dimensions. The position-sensing
photodetector also has advantages over a quadrant detector
because it has a larger linear range than the latter and is less
sensitive to changes in the shape and size of the projected
beam. However, it is slower (B ⫽⬃100 kHz) and some-
what less sensitive than a quadrant detector.
The instrument described here uses a diode laser; these
lasers suffer from “mode hopping” (frequency shifting),
which includes beam pointing instabilities. Mode hopping
arises from temperature fluctuations at the junction in the
diode and can create an artificial shift in the bead position
measurement. However, diode lasers are convenient be-
cause diode temperature can be well controlled with a
thermoelectric cooler (resolution ⫾⬃0.01°C). Moreover,
antireflection coatings that reduce back-reflections can be
used to further minimize mode hopping.
The output from the position-sensitive detector is con-
verted into x position, y position, and total intensity signals
by a custom-built analog electronics board. We used a
feedback resistor of 15 kin this current-to-voltage circuit,
which gave a good compromise between resistor noise and
gain, because the direct laser light on the detector induces a
strong signal (mA) (the detector responsivity is 0.55
mA/mW at 835 nm). For flexibility, the deflection signal is
normalized to the total intensity with software if desired.
The signal is low-pass filtered by analog electronics before
sampling. This low-pass frequency is equal to the Nyquist
frequency (f
nyq
) for the sample rate used (f
s
2f
nyq
)
(Horowitz and Hill, 1989). The low-pass frequency can
easily be altered in our electronics by using different capac-
itors, creating flexibility in the sampling rate (f
s
0.1–2
kHz).
Automation
The throughput of this instrument has been maximized, in
part by extensive automation of many instrument functions.
Automation reduces operator error and ensures that proce-
dures are repeated exactly from experiment to experiment.
In particular, functions that are required for multiple buffers
flowing through the sample chamber can be prepro-
grammed. These functions include injection of beads,
changing buffers during experiments, washing out the
chamber at the end of an experiment to initiate a new one,
and eliminating air bubbles from the flow system. Other
instrumental functions such as calibration of flow rates,
position, and force calibration of the photodetector and the
determination of the stiffness of the trap are also fully
automated. Many operations required to assemble mole-
cules between a bead in the trap and a bead on the micropi-
pette are also automated and are under computer control.
These operations include catching a bead in the trap, deter-
mination of bead positions, allowing the molecules to flow
in, attaching them between the beads (limiting the force
applied to them), and releasing beads from the trap. Finally,
the force applied to a molecule during an experiment can be
adjusted through computer control of the flow speed and/or
position of the bead in the laser trap.
Throughput of single-molecule experiments
Quantifying throughput for single-molecule experiments is
difficult, because throughput is dependent on both the in-
strument and the experiment. Besides increasing the number
of experiments that can be performed in a given time period,
automation of an instrument allows the user to monitor the
performed experiment without the need to control every
detail. Moreover, automation increases the repeatability of
the experiments, facilitates the use of the instrument, and
affords a finer control and gentler manipulation of the
biological system. To give an approximation of the through-
put in our system, we considered the assembly of a DNA
molecule between a trapped bead and a stationary surface.
The combination of flow control, exchangeable beads, and
automation allows us to tether a DNA between a trapped
bead and a bead fixed on a pipette in 4 min, starting from
1160 Wuite et al.
Biophysical Journal 79(2) 1155–1167
Page 6
an empty flow chamber. More significantly, the experi-
menter only has to fix a bead on the pipette (a procedure that
takes a few seconds), while the rest of the tethering proce-
dure is fully automated. Because the flow chamber is emp-
tied by pushing an air bubble through it, the chamber does
not have to be replaced between experiments. Therefore, the
tethering can be repeated many times in an hour, and
because the operator only participates during brief periods
of time during the experiment, it is possible to run the
instrument nearly continuously. Finally, starting the instru-
ment, filling the pressure bottles with the desired solution,
flushing the system, and calibrating it takes less than 30
min. The combination and consistency of these procedures
maximize the throughput of this instrument.
CALIBRATION AND PERFORMANCE
Position calibration
To calibrate detector output to the position of a bead in the
laser trap, an increasing flow was used to displace a bead
held in the trap. At each flow rate, the bead displacements
were measured by video microscopy and were used to
calibrate the resulting deflection measured by the photode-
tector (Fig. 6). This procedure is fully automated in our
instrument. However, this method allows only calibration of
one of the detector axes. To confirm that the photodetector
has the same sensitivity in both directions we employed a
second calibration method as well. A fixed bead, either
adhering to the coverglass surface or attached to the tip of
the micropipette, was moved in a raster pattern through the
beam, using the translation stage. This method allows cal-
ibration of both detector axes but needs a calibrated stage
(Svoboda and Block, 1994b).
Trap stiffness calibration
Hydrodynamic drag force on a trapped particle was used to
obtain a force displacement calibration for objects in the
laser trap. The slope of such a calibration gives the trap
stiffness. Hydrodynamic drag produced by periodic move-
ment of the microscope stage (Ashkin, 1992; Kuo and
Sheetz, 1993; Simmons et al., 1996; Wang et al., 1997)
requires a well-calibrated translation stage. Here continuous
flows were used instead. The drag force on a particle of
known size at a known flow speed was calculated from
Stokes’ law. Bead displacements and the corresponding
laser deflection at various flow rates were used to obtain
plots of force and bead displacement versus detector output.
Conversion factors from the detector output to displacement
and to force were thus obtained. The stiffness of the trap
was calculated from the ratio of these conversion factors.
This calibration method does not require the shape of the
trap to be a perfect parabolic but does rely on a position-
calibrated video microscope as well as on known flow rates.
Stiffness calibration is automated in our instrument, and the
error in stiffness determination with this method is ulti-
mately dictated by the accuracy with which the flow speed
through the cell can be determined (see below).
The force acting on a trapped bead, its displacement in
the trap, and the corresponding deflection of the laser were
found to be related linearly within a region of 400 nm
from the center of the trap for a 2.2-
m bead at the maxi-
mum laser power (Fig. 6). Displacements greater than 400
nm from the center of the single-beam trap make the bead
move down the optical axis as well, causing it to escape.
The trap stiffness is directly proportional to the laser power.
At the maximum laser power (maximum output of the diode
laser 178 mW), the laser trap has an average escape force of
45 pN and an average stiffness of 110 pN/
m (Fig. 6, inset).
The same calibration values were obtained when the corner
frequency of the Brownian motion of a trapped bead was
used to calibrate the stiffness (Fig. 7; Svoboda and Block,
1994a).
Flow speed calibration
Two different methods were used to determine the flow
speed in the middle of the fluid chamber. In the first
method, beads are passed through the chamber; the concen-
tration is low enough (2.5 10
3
%, w/v) so that only a
few beads are in the camera view at one time. The computer
measures the time it takes for the edge of a bead to travel
from one side of the field of view to the other. Because the
distance of the camera view is calibrated, the speed of the
bead can be calculated. This method works at relatively low
FIGURE 6 The calibration curves are taken with a 2.2-
m polystyrene
bead at 178-mW laser power. The flow, which imparts a drag force on the
bead, is reversed back and forth and is incrementally increased until the
bead escapes the trap. The system is linear over 400 nm. Inset: The
displacement measurements of the bead are taken directly from the video
image of the bead in the trap. The escape force of this system at 178-mW
laser power is 45 pN for a 2.2-
m-diameter bead; this value is similar to
the escape forces of other single-beam optical traps.
Integrated Microscopy of Single Biomolecules 1161
Biophysical Journal 79(2) 1155–1167
Page 7
flow rates (up to 400
m/s) and is fully automated. The
error of the flow speed determination of this method is
10%. The limitation of the flow rate arises from the fact
that the CCD camera integrates the signal for 1/60 of a
second. The contrast of the bead edge compared with the
background is reduced when the bead moves faster, because
the image of the bead is spread out over a larger area of the
CCD camera. The lower contrast at higher speeds makes the
computer-controlled recognition of the bead edge more
difficult.
The second method of flow speed determination uses a
trapped bead that is released from the trap at different flow
rates by closing the shutter. The next video frame immedi-
ately after the bead is released is automatically captured by
the computer. The escaped bead appears as a long streak on
the video frame because of the integration time of the CCD
camera (16.6 ms). The length of the streak is measured
manually. The flow rate is calculated from the length of the
streak and the integration time of the CCD camera. This
method works well up to 1600
m/s and has an error of
5%. At higher flow rates the blurred bead image is not
completely contained within the width of one video frame,
and the flow rate cannot be determined by this method.
Instrument performance
Power spectra were compared under four different condi-
tions: dark noise, bright noise, “fixed-bead” noise, and the
Brownian motion of a trapped bead (Fig. 7). The bright
noise was measured by projecting the laser light that has
passed through the instrument on the detector, and the
fixed-bead noise is the detector signal created by the laser,
which interacts with a bead adsorbed to the surface of the
microscope chamber in the laser path.
The lower noise limit in the optical trap, represented by
the dark noise, is 0.007 nm/Hz
1/2
(X
rms
0.15 nm, B 0.5
kHz); this noise is predominately electronic in origin. The
noise increase between the bright noise and dark noise
power spectra in Fig. 7 reflects the magnitude of the optical
noise sources (largely shot noise) of the instrument. This
noise increase is 0.008 nm/Hz
1/2
, resulting in a total r.m.s.
value of 0.35 nm (B 0.5 kHz). The “fixed bead” power
spectrum shows a slight increase in the noise at lower
frequencies due to stage drift. The noise level of 0.015
nm/Hz
1/2
is comparable to other optical trap systems (Svo-
boda and Block, 1994a; Veigel et al., 1998) and is 10-fold
smaller than the Brownian motion of a trapped bead (see
below). The small peak at 60 Hz in curves a, b, and c
(Fig. 7) is caused by the A/D converter, which digitizes the
detector output.
Noise due to Brownian motion of a trapped bead deter-
mines the maximum spatial resolution attainable with the
instrument, given a certain bandwidth. In an actual experi-
ment a tethered bead experiences Brownian motion similar
to that of an untethered trapped bead. Between 7 Hz and 600
Hz, the Brownian motion of the trapped bead is (0.25
nm/Hz
1/2
), resulting in a r.m.s. value of 6.1 nm (B 600
Hz). This r.m.s. value is similar to the expected value
calculated from the equipartition theorem (
trap
110
pN/
m, X
rms
6.2 nm).
The stiffness of a pipette is 4 nN/
m, as determined
from the equipartition theorem:
1
2
x
2
典⫽
1
2
k
B
T, where k
B
is Boltzmann’s constant and T is the absolute temperature.
A bead was held on the pipette tip and positioned in the
center of the laser beam. The mean square displacement of
the bead x
2
obtained with the calibrated photodetector was
used to calculated the pipette stiffness. Because the pipette
is mechanically attached to the chamber, systematic errors
like electronic noise and drift, which are present in the
power spectrum of a fixed bead (Fig. 7), will be present in
the noise of a bead on a pipette. Systematic sources of noise
will artificially increase x
2
and decrease the apparent stiff-
ness. Therefore, 4 nN/
m is a lower estimate. However,
because the stiffness of the pipette is nearly two orders of
magnitude larger than that of the laser trap, the pipette
functions in effect as a stationary wall.
In the laminar flow mode, as for a typical experiment
described by Davenport et al. (2000), when a 2.2-
m bead is
held at the end of a 3.5-
m DNA molecule and extended by
5.5 pN of drag force, the experimental uncertainty in the length
determination of the tether is 11 nm (X
rms
). This error is due
to Brownian motion and the uncertainty of the centroid deter-
mination of the beads (Appendix B). Error due to drift in the
microscope stage is eliminated from the measurements in this
mode because the relative distance between the two beads can
FIGURE 7 Noise power spectral density of the detector signal. The data
in this plot are not normalized. (a) Dark noise: detector signal with no light
hitting detector; (b) bright noise: noise of the laser beam without an object
in the focal region; (c) fixed-bead noise: laser noise with a 2.2-
m
polystyrene bead fixed to the microchamber coverglass and sitting in the
focal region. The bead was fixed by drying the bead on a coverglass
surface, after which water was again added to the coverslide. (d) Brownian
motion: the power spectral density of a trapped bead. This data set is taken
with a 2.2-
m bead in full laser power (178 mW). Corner frequency f
c
800 Hz; trap stiffness 105 pN/
m.
1162 Wuite et al.
Biophysical Journal 79(2) 1155–1167
Page 8
be obtained with the video microscope and the two beads move
simultaneously with the pipette.
TEST APPLICATIONS
Experimental setup
Two experiments, stretching DNA and monitoring the ac-
tivity of single RNA polymerase molecules, are described
here to illustrate the capabilities of this instrument. The
majority of tasks involved in these experiments are auto-
mated and can be performed without operator intervention.
An operator initiates an experiment by fixing a bead on the
tip of the pipette by suction and starting the automated
procedures. A dilute solution of beads is allowed to flow
through the chamber, and the computer can detect when a
bead has been trapped, because the drag force on a trapped
bead will deflect the laser light. The instrument can auto-
matically detect the position of beads by analyzing captured
video frames. It can also differentiate between a trapped
bead and a bead on the pipette by comparing the bead
positions in two video frames, before and after the pipette
has been moved to a different position. Once the position of
the trap is known, the size of the trapped object can be
determined by measuring the size of the black ring (the edge
of the bead). If the laser traps multiple beads, which create
an image larger than a single bead, a shutter blocks the laser
light briefly to release them. The computer then continues
the search for a single bead. Once a single bead is in the trap
and a bead is on the pipette, the experiment can be contin-
ued. The instrument moves the pipette in front of the laser
trap to shield the trapped bead and applies high flow to
remove the remaining beads from the chamber. The auto-
mated positioning of the pipette uses either the pipette bead
or the edge of the pipette itself as a reference to determine
the position of the pipette in relation to the trap. After a set
time, the flow stops and the pipette moves away from the
trap. At this point, the buffer switches are set to allow the
flow of the experimental materials, for example, biotinyl-
ated DNA molecules. The flow speed is increased while the
force on the trapped bead is monitored, until the drag force
on the bead attains a value of 30 pN. The fluid exchange
at this flow speed is fast without risking the escape of the
bead from the trap. After buffer with macromolecules is
allowed to flow in for a fixed time, the computer switches
back to buffer alone, and flow continues while DNA at-
taches to the beads. After this, the automated assembly
procedure (described earlier) starts.
When a biomolecule is assembled between two beads, the
instrument can be used in several modes. In the laminar
flow mode, the bead positions are obtained directly from the
video image. The laminar flow mode is used here in the
study of RNA polymerase to avoid laser damage to the
enzyme. Once the transcription complex assembly has been
made, the laser beam is blocked and the free bead is ex-
tended in flow. Buffer containing nucleoside triphosphates
(NTPs) is automatically allowed to flow into the chamber to
initiate transcription. The load force on the molecules can
then be varied by varying the flow speed. For experiments
that are insensitive to laser exposure, such as stretching
DNA, the bead can be retained in the laser trap, and the
instrument is operated in the laser trap mode or the com-
bined laminar/optical mode, using high laser power. When
operating in these modes, we measure the bead positions by
tracking the laser deflection.
Upon manual termination of the experiment by the oper-
ator, an air bubble is pushed through the chamber and fresh
beads are injected. After a successful experiment the auto-
mated calibration procedure can be started (see Calibration
and Performance). As a result of these preprogrammed
functions, the instrument operator needs only to correct
errors that cannot be solved by the computer. The few
manual tasks that remain include preparing solutions of
biomolecules, filling buffer bottles, and placing a bead on
the tip of the pipette at the beginning of each experiment.
DNA elasticity
The elastic properties of
DNA molecules were used to test
the application of force by either the laser trap or the
flow-control system The elastic properties of DNA have
been well characterized (Cluzel et al., 1996; Smith et al.,
1996). Linear DNA was assembled between two streptavi-
din-coated beads (Appendix A) by the automated assembly
procedure. In the first experiment, the force on the DNA
was produced using the combined laminar/optical mode.
The combination of the laser trap and flow can produce
forces in excess of 80 pN (Fig. 8). The DNA was stretched
by moving the micropipette away from the laser trap oppo-
site the direction of the flow, increasing the force exerted on
the trapped bead. Fig. 9 shows a force-extension curve of
full-length
DNA (Fig. 9A) and a 9866-bp fragment of
DNA (Fig. 9 B) in 0.5 M NaCl. As expected, both lengths
of DNA undergo an overstretching transition at approxi-
mately 65 pN (Cluzel et al., 1996; Smith et al., 1996).
As an alternative to force generation with the laser trap,
the laminar flow mode was also used to stretch DNA. In this
experiment, the trapped bead was released from the trap
after the DNA was assembled between the beads. The DNA
and the previously trapped bead were then extended in the
flow. This flow can provide a drag force of less than 1 pN
and up to 70 pN. The flow speed was slowly ramped up and
down to vary the drag force on the bead. As before, the
extension of the DNA was monitored by measuring the bead
positions through video microscopy. Fig. 10 shows a force-
extension curve obtained by this method for a 14,771-bp DNA
fragment in 0.5 M NaCl. The force on the DNA molecule was
calculated from the flow speed using Stokes’ law and was
corrected for the hydrodynamic shielding generated by the
micropipette. The DNA undergoes an overstretching transition
Integrated Microscopy of Single Biomolecules 1163
Biophysical Journal 79(2) 1155–1167
Page 9
at 68 pN, indicating that forces exerted by either the laser
trap or the flow can be determined accurately.
NTP-dependent tether-shortening by
RNA polymerase
Stalled, biotinylated transcription elongation complexes
were prepared (Appendix A) and assembled between
streptavidin beads held on the micropipette and in the laser
trap (Davenport et al., 2000); the automated assembly pro-
cedure was used to prevent irreversible stalling of the stalled
complexes due to excessive force. After a stalled complex
was assembled between the two beads, the trapped bead was
released from the laser trap to avoid laser damage (laminar
flow mode). The bead was then extended by maintaining
flow, which provided a drag force of less than 15 pN. To
initiate transcription, all four NTPs were allowed to flow
into the fluid chamber. Upon transcription of the DNA
tether, the distance separating the two beads was expected to
shorten; to assay for this shortening, the positions of the two
beads were measured directly from the video image. Be-
cause the flow direction is perpendicular to the optical axis
and hence is in the plane of the video image, changes in
tether length should directly reflect movement of RNA
polymerase along the DNA (Appendix C). Fig. 11 shows an
example of NTP-dependent shortening of the distance be-
tween beads tethered by a stalled transcription elongation
complex. The contour length (L) is calculated from the
end-to-end distance of the DNA (x) as described in Appen-
dix C. Because the DNA was extended in constant flow, the
experiment was performed under constant force conditions.
The average peak rate measured by this method was 7.3
3 bp/s (for 0.2 mM NTPs; n 53), comparable to rates
measured by other in vitro microscopy methods (5–20 bp/s;
Schafer et al., 1991; Wang et al., 1998; Yin et al., 1994, 1995).
The average time for the RNA polymerase to be transcribed in
the laminar flow mode was 400 250s(n 53), which is
significantly longer than 82 58 s when the polymerase is
exposed to 90-mW laser power, as was reported by Yin et al.
(1995). Possibly free radicals are formed in the focus of the
laser light. These radicals might react with and disable the
RNA polymerase. Longer observation times allow transcrip-
tion to be followed over thousands of base pairs, making it
possible to study multiple pausing events by one molecule
and/or to reveal long-lived substates of the enzyme.
CONCLUSION
As a complement to bulk biochemistry, optical trapping
detection permits the investigation of the behavior of single
FIGURE 8 In the combined laminar/optical mode the useful range of
force application by the optical trap can be doubled. (A) Without flow the
equilibrium position of the bead in the trap is in the center of the beam
(position 1). (C) The maximum tension that can be applied to the DNA
with the optical trap is F (position 3). (B) The laminar flow creates a drag
force F on the bead and shifts accordingly the equilibrium position of the
bead in the trap to position 2. (D) From position 2, the laser trap can create
a maximum tension in the DNA of 2F (position 3), twice the amount
possible without the flow present.
FIGURE 9 Overstretching DNA with laser trap force. DNA was assem-
bled between two beads, using the automated assembly procedure, and
stretched by stepping the micropipette away from the laser trap. Experi-
ments were performed in 10 mM Tris-Cl (pH 8.0), 0.5 M NaCl. (A)
Bacteriophage
DNA (48,502 bp). (B) HindIII-BamHI fragment of bac-
teriophage
DNA (9866 bp).
1164 Wuite et al.
Biophysical Journal 79(2) 1155–1167
Page 10
biomolecules and the study of the effect of force on such
molecules. These studies are necessarily limited by the require-
ment to measure one molecule at a time. Moreover, high-
intensity laser light can damage some biological systems. We
have developed an integrated laser trap/flow control video
microscope that is automated to maximize throughput of sin-
gle-molecule experiments while minimizing laser light expo-
sure when necessary. Its various modes of operation make it
possible to choose between high spatial resolution measure-
ments or low laser exposure in single-molecule experiments.
We have studied the elasticity of DNA, using either the com-
bined laminar/optical mode or the flow control system alone
(laminar flow mode) to exert the force. The stretching of DNA
shows not only that the optical trap can be used as a force
transducer, but that it can also be combined with the flow
system and video detection to create a flexible system that can
be adjusted to best suit new biological systems for study.
Moreover, the extensive automation of flow control and sam-
ple manipulation allows exact repetition of experimental pro-
cedures on separate molecules, minimizes operator error, and
increases experimental throughput in the instrument. Using
chemically modified beads as surfaces to which molecules can
be attached greatly increases the number and variety of bio-
logical systems accessible to study. This system is especially
well suited to monitoring the activity of biological enzymes
that translocate along or change the structure of nucleic acids in
a time-dependent manner. We have shown that the laminar
flow mode allows the observation of translocation by Esche-
richia coli RNA polymerase for long times and over long
distances. In addition, the control of flow and buffer conditions
afforded by the instrument is ideal for studying the effect of the
addition or removal of buffer constituents or accessory factors
that modify the system under study.
APPENDICES
A: Experimental materials
Carboxylated polystyrene microspheres (2.2-
m diameter or 3.5-
m di-
ameter, 5% w/v; Sphero, Libertyville, IL) covalently coated with strepta-
vidin were used in all experiments. Beads (2.5% w/v) were coated by
incubation with 5 mg/ml EDAC (1-ethyl-3-(dimethylaminopropyl)carbo-
diimide, hydrochloride; Molecular Probes, Eugene, OR) and 0.5 mg/ml
streptavidin (Gibco BRL, Gaithersburg, MD) in 5 mM sodium acetate (pH
5.1) (2 ml total volume) for 2 h with gentle shaking. Beads were washed
by centrifugation at 3000 g for 15 min, followed by resuspension in 2 ml
phosphate-buffered saline. After a final centrifugation at 3000 g for 15
min, the bead pellet was resuspended in 1 ml phosphate-buffered saline
0.1% sodium azide (final concentration 2.5% w/v). For experiments the
bead stock solution is diluted by 1:1000, and an automated syringe is used
to inject exact and repeatable volumes of suspended beads (80
l).
For development of the assembly procedure and for measurement of the
elasticity of DNA, three DNA molecules were used: linearized monomers
of bacteriophage
DNA (48,502 bp); linearized pLT1 plasmid (a deriva-
tive of pBW8; Feiss et al., 1983) (14,771 bp); or a HindIII-BamHI fragment
of bacteriophage
DNA (9866 bp). DNA fragments were labeled with
biotinylated nucleotides by filling in 5 overhangs from restriction endo-
nuclease digestion. DNA (0.1 pmol) was labeled by incubation with
biotin-14-dATP and biotin-14-dCTP (25
M each) (Gibco BRL) and
dGTP and dTTP (100
M each) (Sigma) and either Exo
klenow fragment
of DNA polymerase I (New England Biolabs, Beverly, MA) or Exo
T4
DNA polymerase (Worthington, Freehold, NJ). Labeling reactions were
performed in a total volume of 60
l.
Labeled DNA was diluted to a working concentration of 10 pM in
buffer containing 10 mM Tris-Cl (pH 8.0), 500 mM NaCl, and 0.1%
sodium azide. To assemble DNA between beads in the microscope, the
DNA solution was allowed to flow into the chamber for 100 s at a rate that
exerted a force of 30 pN on a bead held in the optical trap. Attachment
of DNA molecules to the bead can be visualized as a sudden increase in
drag on the bead (Fig. 4). Typically, three to five molecules of DNA attach
to the bead during the time in which the DNA flows through the chamber.
The likelihood of attaching two molecules simultaneously during the
fishing procedure is small; moreover, once a tether is made between the
two beads and flow is lowered or halted, DNA molecules that have not
FIGURE 11 NTP-dependent tether-shortening by transcription of RNA
polymerase. Stalled transcription elongation complexes were assembled
between two beads by the automated assembly procedure. After successful
assembly the trapped bead was released from the laser to avoid laser
damage, and the DNA was extended by maintaining a flow. NTPs were
allowed to flow into the chamber while the distance between the beads (x
m
)
was measured directly from the video image.
FIGURE 10 Overstretching 14,771-bp DNA with flow force. Linearized
pBW8 plasmid DNA was assembled between two beads, using the auto-
mated assembly procedure. The laser-trapped bead was released from the
trap and extended by flow. The DNA was stretched by ramping the flow to
increase the hydrodynamic drag force.
Integrated Microscopy of Single Biomolecules 1165
Biophysical Journal 79(2) 1155–1167
Page 11
attached to both beads are not extended far enough to make an attachment.
The presence of only one double-stranded molecule of DNA is verified by
examining its characteristic stretching curve: a single molecule undergoes
an overstretching plateau at 65–70 pN; multiple molecules will produce an
overstretching transition at multiples of the single-molecule value.
For biotinylation of the transcription template, pPIA6 was linearized
with BamHI restriction endonuclease (New England Biolabs). Linearized
plasmid (5 pmol) was incubated for 20 min at room temperature in 60
l
total volume 1BamHI buffer (New England Biolabs) containing bio-14-
dATP and bio-14-dCTP (25
M each) (GibcoBRL), dTTP and dGTP (100
M each) (Sigma), and 0.5
l Exo
T4 DNA polymerase (Worthington).
The labeling reaction was quenched with 3
l of 0.5 M EDTA (pH 8.0) and
spun in a Microcon 100 (Amicon, Beverly, MA). The Microcon 100 was
washed twice with 400
l 10 mM Tris (pH 7.9), and the sample was
recovered. The recovered sample was digested with SacI restriction endo-
nuclease (New England Biolabs) to produce a linear DNA fragment bio-
tinylated at only one end. This DNA fragment was then gel purified on a
0.7% agarose-TAE gel and electroeluted into low TE in a dialysis bag
(Spectrapor 2, 12,000–14,000 molecular weight cutoff). The eluent was
extracted once with phenol/chloroform/isoamyl alcohol and once with
chloroform; the DNA in the aqueous phase was precipitated with ethanol,
resuspended in low TE, and stored at 4°C. (The biotinylation seems to
degrade when the sample is stored at 20°C.)
Transcription complexes were stalled at position
70 by incubating 10
nM linearized DNA template with 40 nM biotinylated RNA polymerase
(gift of Robert Landick) in 1 transcription buffer (50 mM Tris-HCl (pH
8.0), 200 mM KCl, 10 mM MgCl
2
, 1 mM dithiothreitol) for 10 min at
37°C; then adding 200 mg/ml Heparin, 100
M ATP, 100
M GTP, and
20
M UTP; and incubating for 10 min at 25°C. Stalled complexes were
diluted to a working concentration of 40 pM in 1 transcription buffer
containing 1% dithiothreitol and 0.1% NaAzide.
B: Limits to the spatial resolution of video
microscopy and laminar flow mode
The image of the CCD camera is calibrated with a calibration grid grad-
uated in 10-
m increments (Nikon). Beads imaged by the CCD camera
appear as light, circular objects surrounded by a black edge. The bead
position is determined by averaging the positions of all of the pixels inside
the circular black edge of the bead image. An image of a bead 2.2
min
diameter typically contains 1000 pixels (1 pixel ⫽⬃60 nm). The
centroid of the bead can be calculated by averaging the x and y positions
of all of the pixels:
X
centroid
n
x
n
N
and Y
centroid
n
y
n
N
(1)
where x
n
and y
n
are the coordinates of the individual pixels and N is the
total number of pixels in the image. The position uncertainty in one pixel
is 30 nm, but the standard deviation of the mean decreases with the
number of pixels in the image:
centroid
pixel
N
(2)
The centroid of a 2.2-
m bead image can thus be theoretically determined
with an accuracy of 1 nm (0.03 pixel). The resolution is essentially
linearly related to the radius of the bead image (or the pixels on the edge
of the image). It might be possible to obtain higher resolution measure-
ments by using the exact gray values of the whole bead image. The
empirically measured error in the centroid position is 3 nm. Although the
bead position can be determined, in principle, at a maximum sampling rate
of 30 Hz (bandwidth of the CCD camera 60 Hz), in practice this rate is
limited to 20 Hz because of the time required to extract the distance data
from the video frame. To extract data in real time, only the portion of the
video frame that contains the bead images is used.
After the bead is released from the laser trap and extended by hydro-
dynamic drag, the Brownian motion perpendicular to the flow is greater
than that along the axis of the flow, i.e., along the tether between the two
beads. The theoretical error in position determination due to thermal noise
can be approximated from the fluctuation-dissipation theorem (Landau and
Lifshitz, 1980):
X
rms
1
DNA
trap
2k
B
T
B
with B
/
c
and
6
␲␩
r (3)
where
DNA
and
trap
are the spring constants of DNA and the laser trap
respectively, B is the bandwidth of the measurements in radians/s,
c
is the
corner frequency of the bead in the potential well,
is the viscosity of the
medium, r is the radius of the bead, k
B
is the Boltzmann constant, and T is
the absolute temperature. In the laminar flow mode
trap
is 0 pN. The
spring constant of DNA can be calculated from the expression (Bustamante
et al., 1994; Marko and Siggia, 1995)
F
k
B
T
P
1
41 x/L兲兲
2
x
L
1
4
(4)
where F is the force applied to a bead tethered to a DNA molecule attached
at its other end to a fixed support, P is the persistence length of DNA
(P 53 nm in the ionic strength conditions used in these experiments), x
is the end-to-end distance at that force; and L is the contour length. The
spring constant can be determined as
DNA
F
x
xL
4F
3/2
L
P
k
B
T
(5)
The position of a 2.2-
m bead held in the laser trap (
trap
90 pN/
m)
in the absence of flow can be empirically determined with an accuracy of
4.6 nm (X
rms
), using video microscopy. As expected, this error is equal to
the error due to Brownian motion of the bead (Eq. 3; 1.6 nm; B 60 Hz)
plus the error of the centroid determination (3 nm). When a 2.2-
m bead
is held at the end of a 3.5-
m DNA molecule and extended by 5.5 pN of
drag force, the experimental uncertainty in position determination is 6.8
nm. The error due to Brownian motion is higher in this case, namely 3.1 nm
(Eq. 5;
DNA
50 pN/
m at 5.5 pN).
C: Calculation of the contour length from the
bead-bead distance
The end-to-end distance of the DNA (x) is obtained by subtracting the radii
of the beads (B
corr
) from the center-to-center bead distance (x
m
). The
applied force (F), created by hydrodynamic drag by the buffer flow on the
bead, was calibrated to the flow speed as described previously. The actual
force applied to the molecule is the force determined by the calibration
minus the drag reduction caused by the presence of the pipette. The value
of such drag reduction is obtained by holding an untethered trapped bead
behind the pipette at various distances and at various flow speeds. The
force detection capability of the laser allows direct measurement of the
hydrodynamic shielding of the pipette.
A DNA molecule displays elastic behavior when pulled by its ends. The
separation between the beads gives the end-to-end distance of the DNA,
which