Sub-diffraction-limit imaging by Stochastic Optical Reconstruction Microscopy (STORM) Michael J. Rust, Mark Bates, Xiaowei Zhuang Harvard University Published.

Sub-diffraction-limit imaging by Stochastic Optical Reconstruction

Microscopy (STORM)

Michael J. Rust, Mark Bates, Xiaowei ZhuangHarvard University

Published Online August 9, 2006Nature Methods Vol.3 No.10

Presented by Artie Wu

STORM

• High-resolution fluorescence microscopy method based on high-accuracy localization of photoswitchable fluorophores

• Imaging resolution of 20nm

Outline

• Background– fluorescence microscopy– diffraction

• Motivation• Fluorescence Microscopy Alternatives• STORM• Results• Conclusions

Background

Background

Motivation I

• Resolution limit set by diffraction of light

• Fluorescence microscopy widely used in molecular and cell biology

Fluorescence Microscopy Alternatives

• Lateral resolution of 10s of nanometers– Near-field scanning optical microscopy (NSOM)– Multiphoton fluorscence– Stimulated emission depletion (STED)– Saturated structured-illumination microscopy (SSIM)

Near-field scanning optical microscopy (NSOM)

• Image at interface due to evanescent field

• Study what goes on near membrane– Exocytosis & endocytosis

• Build up point by point• Drawback: low imaging

depth

Motivation II

• Single-molecule detection leads to sub-diffraction-limit spatial resolution

• Stochastic optical reconstruction microscopy (STORM)– Fluorescence image constructed from high-accuracy

localization of individual fluorescent molecules– Imaging resolution: ~20nm using TIRF and

photoswitchable cyanine dye, Cy5

STORM

•Cy5: fluorescent and dark state using different λ•Cy3: secondary dye•Series of imaging cycle•In each cycle

•Only 1-3 switches in FOV are switched ON•Stochastically different subset of fluorophores are ON

•Red: 633nm, 30W/cm2, 2s•Green: 532nm, 1W/cm2, 0.5s•Photobleaching: 230s

Resolution

• Limited by accuracy of localization of switches

• 2d Gaussian fit to PSF used to find centroid position of switch

Centroid position

• Fit to pixelated Gaussian function

A: background fluorescence level

Io: amplitude of peak

a,b: widths of Gaussian distribution

xo,yo: center coordinates of peak

δ: fixed half-width of pixel in object plane

b

yyerf

b

yyerf

a

xxerf

a

xxerf

abIA

eIdYdXAyxI

ooooo

b

yY

a

xX

o

y

y

x

x

o

4

,2/0

Results I

• Linear, dsDNA with 2 switches separated by 135 bps (46nm)

• Theoretical dist = 40nm• Experimental dist = 41nm

Results II

• Longer DNA with 4 switches spaced 46 nm apart

• Localize large number of switches within diffraction-limited spot by cycling switches on/off

Conclusions

• STORM capable of imaging biological structures with sub-diffraction-limit resolution

• Resolution limited by # photons emitted per switch cycle– Cyanine switch ~3000 photons/cycle

• Theoretical localization accuracy of 4nm

• Corresponds to imaging resolution of ~20nm

– Imaging speed improved by increasing switching rate• Stronger excitation or fluorophores with faster switching kinetics

• Valuable tool for high-resolution in situ hybridization and immunofluorescence imaging