28.06.2013 Views

Steve Smith - Bustamante Group - University of California, Berkeley

Steve Smith - Bustamante Group - University of California, Berkeley

Steve Smith - Bustamante Group - University of California, Berkeley

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Introduction to Optical Tweezers<br />

<strong>Steve</strong> <strong>Smith</strong><br />

<strong>Bustamante</strong> <strong>Group</strong>, Physics Dept.<br />

Howard Hughes Medical Institute<br />

<strong>University</strong> <strong>of</strong> <strong>California</strong>, <strong>Berkeley</strong>


Comet tail<br />

Light transfers momentum to ma ter<br />

Light exerts force on matter<br />

James Clerk Maxwell<br />

1831-1879


Electromagnetic waves interact<br />

electric<br />

with electrons in matter<br />

DP = DU/c<br />

V<br />

e-<br />

E<br />

B<br />

F=eBV<br />

P = h k


Arthur Ashkin builds first optical trap<br />

Single-beam trap<br />

Dual-beam trap<br />

1970<br />

Axial<br />

escape


Photon meets refracting object<br />

Photon<br />

momentum<br />

P = h/l<br />

Pin<br />

F = dP/dt<br />

q<br />

Pout<br />

DP<br />

For every action there exists an equal but opposite reaction<br />

Sir Isaac Newton<br />

ashkin1.EXE


a stable single-beam trap<br />

Anti-scattering force:<br />

Forward momentum is<br />

increased by lens -focusing<br />

effect.<br />

ASHKIN2.EXE


Trap live<br />

bacteria<br />

Infrared trap supports life !<br />

Sort living<br />

cells<br />

Manipulate<br />

organelles<br />

inside cells<br />

Nature, 1987


Estimating Forces<br />

trap center<br />

Dx<br />

1. Assume a linear-spring restoring force<br />

2. Determine trap stiffness k<br />

3. Measure Dx relative to trap center<br />

F = k Dx


Calibrating trap stiffness<br />

(1) Stokes’ law<br />

(2) Corner frequency<br />

(3) Equipartition


Glass chamber<br />

Fluid drag test force<br />

Stokes’ law F = 6prhV but corrected for proximity to walls<br />

Distance<br />

detector<br />

Data<br />

acquisition<br />

Motorized<br />

Stage with<br />

encoder<br />

Glass<br />

Glass<br />

water


Brownian noise as test force<br />

Drag force<br />

= 6phr<br />

for a sphere<br />

Langevin equation:<br />

x<br />

F(<br />

t)<br />

kDx<br />

Dx<br />

2<br />

(<br />

f<br />

Fluctuating force<br />

= 0<br />

< F(t) F(t’) > = 2k BTd(t-t’)<br />

)<br />

<br />

<br />

4k<br />

B<br />

T<br />

2 2<br />

f f <br />

Lorentzian power spectrum<br />

c<br />

Trap force<br />

Corner<br />

frequency<br />

f c = 2p k /


Power (nm 2 /Hz)<br />

Power spectra<br />

f c=k/2p<br />

Frequency (Hz)<br />

1/f 2<br />

4k BT/k 2


Does not use drag coefficient, but rather ..<br />

Equipartition method<br />

integrate area under power curve to get <br />

k = k BT / <br />

However, you must have<br />

an accurate measure <strong>of</strong><br />

Dx at high bandwidth.<br />

This value is more<br />

easily taken in AFMs<br />

than optical traps.<br />

½ k = ½ k BT


Position clamp avoids problems with trap linearity


Measuring forces by analyzing<br />

momentum <strong>of</strong> the trap beam<br />

Light ray with power W<br />

dP/dt = nW/c<br />

input<br />

(nW/c) sin q<br />

F = -D dP/dt<br />

(nW/c) (1-cos q)<br />

q<br />

dP/dt<br />

output<br />

DdP/dt


Counter-propagating beams for<br />

LASER<br />

position<br />

detector<br />

narrow-angle trap<br />

pbs<br />

OBJ<br />

DNA<br />

pipette<br />

OBJ<br />

qwp<br />

liquid<br />

chamber<br />

position<br />

detector<br />

LASER


Narrow beams stay within NA <strong>of</strong> lenses<br />

liquid<br />

external<br />

force<br />

air<br />

detector


Light leaving trap obeys Abbe sine condition<br />

Front focus<br />

Focal length L<br />

Back focal plane<br />

X = n L sinq<br />

Objective lens<br />

BFP : where angle q is best<br />

represented by <strong>of</strong>fset x


How to measure light <strong>of</strong>fset?<br />

quadrant photodiode


versus PSD photodiode


+<br />

_ _<br />

+<br />

_ _<br />

_<br />

_<br />

QPD<br />

PSD<br />

+


PSD<br />

(position sensitive detector)<br />

Plate resistors<br />

separated by<br />

reverse-biased<br />

PIN photodiode<br />

In 1<br />

N<br />

Out 1<br />

P<br />

P<br />

Out 2<br />

N<br />

In 2<br />

opposite electrodes held at same potential<br />

no conduction unless there is light


PSD<br />

(position sensitive detector)<br />

Plate resistors<br />

separated by<br />

reverse-biased<br />

PIN photodiode<br />

In 1<br />

N<br />

Out 1<br />

P<br />

P<br />

Out 2<br />

Suppose we shine ray <strong>of</strong> light with intensity W i in exact center <strong>of</strong> detector:<br />

In 1 + In 2 = Out 1 + Out 2 = W i<br />

by charge conservation<br />

Out 1 = Out 2 = ½ W<br />

In 1 = In 2 = ½W<br />

by symmetry<br />

N<br />

In 2<br />

(sensitivity = 1)


PSD<br />

(position sensitive detector)<br />

In 1<br />

Now suppose the ray <strong>of</strong> is <strong>of</strong>f center.<br />

N<br />

Out 1<br />

P<br />

P<br />

Out 2<br />

N<br />

Out 1 + Out 2 = W = In 1 + In 2 still holds<br />

In 1 > In 2 and Out 1 > Out 2 due to resistance asymmetry<br />

Opposite electrodes held at equal potential so currents to those<br />

electrodes divide inversely to the distance <strong>of</strong> the spot from electrode.<br />

In 2


PSD<br />

(position sensitive detector)<br />

In 1<br />

N<br />

Out 1<br />

P<br />

P<br />

Out 2<br />

Multiple rays add their currents linearly to the electrodes,<br />

where each ray’s power adds W i current to the total sum.<br />

N<br />

In 2


PSD<br />

(position sensitive detector)<br />

In 1<br />

N<br />

Define x-y coordinates centered on detector<br />

it can be shown<br />

Out 1<br />

Out 2<br />

y<br />

P<br />

P<br />

N<br />

In 1 – In 2 = S W i x i / R D<br />

Out 1 – Out 2 = S W i y i / R D<br />

In 2<br />

where R D is the half-width (or “radius”) <strong>of</strong> the detector<br />

x


PSD<br />

(position sensitive detector)<br />

In 1<br />

N<br />

Out 1<br />

P<br />

P<br />

Out 2<br />

For arbitrary light distribution, centroid position given by difference <strong>of</strong> electrode currents<br />

where sum = In 1 + In 2 = Out 1 + Out 2 = S W i<br />

N<br />

X center= R D (In 1 –In 2) / sum<br />

Y center = R D (Out 1 – Out 2 ) / sum<br />

Sensitivity does not depend on spot size or shape<br />

In 2


PSD<br />

force<br />

sensor<br />

samples<br />

unfocused<br />

beam<br />

In 1<br />

N<br />

Out 1<br />

P<br />

P<br />

Out 2<br />

N<br />

S X = In 1 – In 2 = S W i x i / R D<br />

S Y = Out 1 – Out 2 = S W i y i / R D<br />

In 2


Detecting<br />

external<br />

force<br />

from<br />

changes<br />

in<br />

light<br />

momentum<br />

flux<br />

liquid<br />

n L<br />

air<br />

detector<br />

external<br />

force<br />

External force = light force = effect from all rays:<br />

Collector lens transforms exit angles into ray <strong>of</strong>fsets<br />

by Abbe Sine Condition: x i = L n L sin q i<br />

PSD sums over rays to give signal S X R D= SW i x i<br />

2L<br />

F x = dP/dt = (n L/c) SW i sin q i<br />

Then external transverse force is given by<br />

F X = S XR D /cL<br />

X


Momentum sensor calibration<br />

Calibrate signal to power ratio for PSDs / objectives with<br />

power meter and ruler.<br />

No test force is used.<br />

Calibration does not change with particle size, particle<br />

shape or laser power. Particle and trap are not being<br />

calibrated (don’t matter).<br />

Methods in Enzymology v.361 (2003)


Measuring axial forces<br />

Light ray with power W<br />

dP/dt = n LW/c<br />

input<br />

(n LW/c) sin q<br />

F = -D dP/dt<br />

(n LW/c) (1-cos q)<br />

q<br />

dP/dt<br />

output<br />

DdP/dt


Size <strong>of</strong> exit beam indicates axial force on<br />

trapped object<br />

Laser beam


Correct weighting function to extract axial<br />

momentum flux is semi-circle<br />

transmission<br />

radius = n L * L<br />

bulls-eye<br />

optical<br />

attenuator


Plain<br />

Photo<br />

diode<br />

Placement <strong>of</strong> axial force sensors<br />

Bulls-eye<br />

attenuator


Path <strong>of</strong> bacterium<br />

Flagellum wobble<br />

10 nm<br />

1000 samples/sec


Force (pN)<br />

Force - Extension Behavior <strong>of</strong> dsDNA<br />

and ssDNA<br />

Fractional Extension


Bockelmann, Heslot, 2002<br />

S. Koch, M. Wang, 2003<br />

Felix Ritort et al., in preparation<br />

Unzipping dsDNA<br />

ssDNA<br />

Protein<br />

ssDNA


17 pN<br />

16 pN<br />

15 pN<br />

60 nm


Motor step size:<br />

how small can we detect?<br />

• Effects <strong>of</strong> thermal noise and tether elasticity


Springs in series for motor<br />

Trap center<br />

Light<br />

spring<br />

k 2<br />

Bead moves<br />

Dx sig = Dx s k 1<br />

k 1+k 2<br />

Tether<br />

spring<br />

k 1<br />

Motor steps<br />

Dx s


Springs in parallel for thermal noise<br />

Trap center<br />

Light<br />

spring<br />

k 2<br />

Combined<br />

potential<br />

k 1 k 2<br />

Tether<br />

spring<br />

k 1


(pN 2 /Hz)<br />

Force-noise spectral density is<br />

10 -3<br />

10 -4<br />

10 -5<br />

10 -6<br />

10 -7<br />

10 -8<br />

proportional to bead size<br />

0.27<br />

0.54<br />

0.82<br />

2.03<br />

5.10<br />

10 100 1000 10 4<br />

Frequency (Hz)<br />

10 5<br />

= 4k BT <br />

at low frequencies


Thermal noise<br />

Dx therm = DF therm / (k 1 + k 2)<br />

= 2 (k BT B) 1/2 / (k 1 + k 2)<br />

where B is bandwidth in Hz<br />

Signal to noise ratio<br />

SNR >1 when Dx sig > Dx therm<br />

Dx s k 1 / (k 1 + k 2) > 2(k BT B) 1/2 / (k 1 + k 2)<br />

Dx step > 2(k BT B) 1/2 / k 1<br />

Tether<br />

stiffness


Thermal limit to step detection<br />

Dx step > 2(k BT B) 1/2 / k 1<br />

Resolution depends only on tether stiffness, not trap stiffness.<br />

Resolution degrades as (drag) 1/2<br />

Comparing AFM to laser tweezers, the force noise scales as<br />

sqrt(cantilever length / bead diameter). Therefore a 100um cantilever has<br />

10x more force noise than a 1 um bead, and 10x bigger distance noise<br />

for fixed k 1.<br />

A stiff linkage (large k 1) gives an AFM very good resolution when it<br />

pushes against a hard sample. To make a DNA tether stiff requires some<br />

tension in the tether.


Signal<br />

100<br />

50<br />

0<br />

-50<br />

Averaging reduces bandwidth, suppresses noise<br />

Noise plus Steps<br />

-100<br />

0 5000 10000 15000 20000 25000 30000<br />

filtered<br />

time<br />

100<br />

50<br />

0<br />

-50<br />

Running Window 500<br />

-100<br />

0 5000 10000 15000 20000 25000 30000<br />

time<br />

filtered<br />

100<br />

50<br />

0<br />

-50<br />

Running Window 10<br />

-100<br />

0 5000 10000 15000 20000 25000 30000<br />

time<br />

filtered<br />

100<br />

50<br />

0<br />

-50<br />

filtered<br />

100<br />

50<br />

0<br />

-50<br />

Running Window 100<br />

-100<br />

0 5000 10000 15000 20000 25000 30000 35000<br />

Running Window 1000<br />

-100<br />

0 5000 10000 15000 20000 25000 30000<br />

time<br />

time


For example:<br />

Bead is 2 um diameter, immersed in water.<br />

Tether is 10 kbp <strong>of</strong> dsDNA and tether tension is either 2 pN or 20 pN.<br />

Signal <strong>of</strong> interest is at 1 Hz, so that much bandwidth is required.<br />

Tether stiffness k 1 = dF/dx for WLC at either tension (assume P~50nm).<br />

at 2 pN tension, k 1 = 12 pN/um<br />

at 20 pN tension, k 1 = 170 pN/um<br />

Then smallest resolvable step Dx s= 2(k BT B) 1/2 / k 1<br />

Dx s= 1.5 nm @ 2 pN tension<br />

Dx s = 0.1 nm @ 20 pN tension (in 1 Hz bandwidth)<br />

Averaging for infinite time will reduce B to zero and resolve infinitely small steps<br />

[ but completely lose temporal resolution]<br />

Slow down the process?<br />

[ now limited by position drift <strong>of</strong> instrument ]


Work-Horse Optical Trap<br />

Measures force by light momentum change<br />

10 years gave<br />

over 30 papers


Movable microchamber, fixed trap position<br />

X-Y-Z piezo-flexure stage (Martoc)


Typical configuration to pull a molecule<br />

Piezo stage<br />

Glass<br />

pipette


1.3nm<br />

pipette<br />

Methods in<br />

Enzymology<br />

volume 361 (2003)<br />

Special configuration tests “drift” noise<br />

Force (pN)<br />

0 1 2 3<br />

Stage position (nm)<br />

8 0 -8 -16


Characterizing “drift” with velocity histograms<br />

Low-pass filter position signal<br />

(here 1 Hz)<br />

Score average velocities<br />

in 2 sec intervals<br />

Fit distribution to Gaussian<br />

Drift <strong>of</strong>fset = 0.3 bp/s<br />

Velocity SD = 2 bp/s @ 1 Hz<br />

count<br />

see<br />

Neuman and Block<br />

Cell, 2003<br />

basepair / sec


Technical problems<br />

Optics sensitive to:<br />

Operator’s touch, breath, voice<br />

Changes in room temperature, air-flow, vibration<br />

Atmospheric fluctuations (star twinkle)<br />

become management problems<br />

Only works in special room in basement<br />

Needs operator training for good data<br />

Competition for machine time<br />

Expensive to build extra machines


Our solution: “Mini-Tweezers”


Table-top<br />

instrument<br />

Optics head<br />

hangs from<br />

bungee cord<br />

Works OK on<br />

upper floors


Mini uses telecom “pump” lasers<br />

300 mW typical<br />

single-mode fiber<br />

output<br />

975 nm or 845 nm


Fixed chamber, movable traps<br />

gives increased stability


Fiber wiggler moves trap


position<br />

Moving the fiber is faster<br />

than moving the chamber<br />

Martock flexure stage Fiber wiggler<br />

1.32 1.34 1.36 1.38 1.4 1.42 0.06 0.08 0.1 0.12 0.14 0.16<br />

time (s) time (s)


Compact optical path avoids “twinkle” effect<br />

10 cm from fiber to trap (3 cm air)


eam<br />

Motorized<br />

stage<br />

remains<br />

fixed<br />

Beams move up and down<br />

Pipette bead remains fixed<br />

0.5 nm steps at 1 Hz


Wooden box


count<br />

Less velocity-noise with<br />

mini-Tweezers<br />

basepair / sec<br />

standard<br />

mini<br />

Velocity noise = 0.4 bp/s @ 1Hz BW


Analytical<br />

optical<br />

traps<br />

can do:<br />

RNA hairpins assay helicase activity<br />

RNA secondary structure, folding and refolding<br />

Phage packaging motors<br />

Polymer entropic elasticity<br />

DNA mechanics (torsional rigidity, phase transitions)<br />

DNA condensation phase transitions<br />

DNA thermodynamics, base-pair energies<br />

Force-melting DNA shows sequence (unzipping)<br />

Molecular motors in muscle (myosin, actin)<br />

Cell transport: kinesin on tubulin, dynein on tubulin<br />

Cell import: endosome degradation<br />

Protein folding and refolding (RnaseH, T4 Lysozyme)<br />

Protein folding multimers (Titin)<br />

Enzyme movements, kinetics: topoisomerase, gyrase<br />

Polymerases (DNA, RNA)<br />

Affinity studies: antibody, ligand<br />

DNA/protein binding, e.g. recA,<br />

Chromatin structure and remodeling<br />

Combinatorial chemistry, bead sorting<br />

Cell sorting by drag coefficients<br />

Rheology <strong>of</strong> polymers<br />

Reptation studies<br />

Electrophoresis forces<br />

Cell wall deformability<br />

Statistical mechanics (Jarzynski, Crooks theorems)<br />

Bacterial motility (swimming force) in 3 dimensions<br />

Education / training in biophysics


Thanks:<br />

• Carlos <strong>Bustamante</strong> and lab members<br />

• Howard Hughes Medical Institute<br />

• Claudio Rivetti, <strong>University</strong> <strong>of</strong> Parma<br />

http:// tweezerslab.unipr.it<br />

• Agilent Technologies Foundation

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!