Super-Resolution Optical Microscopy Bo Huang Light Microscopy May 10, 2010

Naked eye: ~ μm 1595, Zaccharias and Hans Janssen First microscope, 9x magnification Antony Van Leeuwenhoek ( ), 200x Ernst Abbe ( ) The “physical” diffraction limit Compound microscope >1000x 2000 d d  2 NA

The diffraction barrier Atomic Cellular Sub-cellular Molecular Fotin et al., Nature 2004; μm Diffraction limit: ~ 250 nm lateral ~ 600 nm axial

50 years to extend the resolution Confocal microscopy (1957) Near-field scanning optical microscopy (1972/1984) Multiphoton microscopy (1990) 4-Pi microscopy / I 5 M ( ) Structured illumination microscopy (2000) Negative refractive index (2006)

Near-field scanning optical microscopy Excitation light Optical fiber Aperture SampleIanoul et al., 2005 β 2 adrenergic receptor clusters on the plasma membrane ~ 50 nm

4-Pi / I 5 M d d  2 NA  NA = n sin  Major advantage: Similar z resolution as x-y resolution Major advantage: Similar z resolution as x-y resolution

Patterned illumination =x x ExcitationDetection Detector

Structured Illumination Microscopy (SIM) 9 images Reconstruction WFSIM Gustafsson, J Microscopy =

The diffraction limit still exists NA d 22 1   Confocal 4Pi / I 5 M SIM

Breaking the diffraction barrier

Confocal 4Pi / I 5 M SIM Breaking the diffraction barrier

Stimulated Emission Stimulated Emission Depletion (STED) Send to a dark state Detector h 2h Excitation Fluorescence FL 0 0 IsIs

STED microscopy Excitation Fluorescence Stimulated Emission ÷= ExcitationSTED pattern Effective PSF Hell 1994, Hell 2000 Detector Excitation Depletion Light modulator ?

Saturated depletion I STED = I S I STED = 2 I S I STED = 10 I S I STED = 100 I S STED pattern Saturated Depletion zero point NA II d s 2 /1 1   

STED images of microtubules Wildanger et al., 2009

The “patterned illumination” approach ÷ ExcitationDepletion pattern Ground state Triplet state Isomerization etc. = Multiple cycles

Saturated SIM I ex FL Fluorescence saturation Saturation level Saturated illumination pattern Sharp zero lines Gustaffson, PNAS 2005 WFDeconvolution SIMSSIM 50 nm resolution Suffers from fast photobleaching under saturated excitation condition

The single-molecule switching approach

Single-Molecule Localization Yildiz et al., Science, 2003 Image of one fluorescent molecule

Single-molecule localization precision 1 photon 10 photons100 photons1000 photons d d  2 NA NA N d 2 1  

Super-resolution imaging by localization Raw imagesConventional fluorescence STORM Image 2x real time ActivationLocalization Deactivation Also named as PALM (Betzig et al., Science, 2006) and FPALM (Hess et al., Biophys. J. 2006) Huang et al., Annu Rev Biochem, 2009 Stochastic Optical Reconstruction Microscopy = STORM

Photoswitching of red cyanine dyes photoactivation Deactivation 650 nm 360 nm 650 nm Fluorescent Dark Cy5 / Alexa 647 N N + + thiol Bates eta l., PRL 2005, Bates et al., Science 2007, Dempsey et al., JACS 2009

B-SC-1 cell, anti-β tubulin Commercial secondary antibody 5 μm 500 nm Alexa ,000 frames, 1,502,569 localization points FWHM = 24 nm stdev = 10 nm

The “single-molecule switching” approach Multiple photons Photoswitching Blinking Diffusion Binding etc. + Stochastic Switching =

STORM probes commercially available or already in your lab nm Cyanine dye + thiol system Cy5 Cy5.5 Cy7 Rhodamine dye + redox system Atto590 Alexa568 Atto655 Atto700 Alexa488 Heilemann et al., 2009 Atto565 Bates et al., 2005, Bates et al., 2007, Huang et al., 2008 Photoactivatible fluorescent proteins PA-GFP PS-CFP2 Dronpa mEosFP2 Dendra2 EYFP Reviews: Lukyanov et al., Nat. Rev. Cell Biol., 2005 Lippincott-Schwartz et al., Trends Cell Biol., 2009 Alexa532 Atto520 PAmCherry Alexa647

In a 2D world… Satellite image of ??? Google maps

3D STED Harke et al., Nano Lett, 2008

3D STORM/PALM EMCCD (x, y, z ) z (nm) (x, y) Astigmatic imaging Huang et al., Science 2008 Bi-plane imaging Double-helical PSF Juette et al., Science 2008 EMCCD SLM z (nm) Pavani et al., PNAS 2009

5 μm Huang, Wang, Bates and Zhuang, Science, 2008 Scale bar: 200 nm 3D Imaging of the Microtubule Network z (nm)

5 μm Huang, Wang, Bates and Zhuang, Science, 2008 Small, isolated clusters 22 nm28 nm55 nm 3D Imaging of the Microtubule Network z (nm) FWHM

I5SI5S isoSTED iPALM 4Pi scheme The use of two opposing objectives Near isotropic 3D resolution Shal et al., Biophys J 2008 Schmidt et al., Nano Lett 2009 Shtengel et al., PNAS 2009

3D resolution of super-resolution methods x-y (nm) z (nm) Opposing objectives (nm) Two-photon Conventional Pi: 120 SIM I 5 S: 120 xyz STED ~30~100isoSTED: 30 xyz100 µm deep STORM/PALM iPALM: 20 xy, 10 z

Multi-color Imaging

Excitation 2 STED 2 Muticolor STED Excitation STED 2 color isoSTED resolving the inner and outer membrane of mitochondria Schmidt et al., Nat Methods µm

Multicolor STORM/PALM: Emission n1n1 n2n2 n 1 = n 2  50% SRA % SRA617?  100% SRA577? Single-molecule detection! 3-color imaging with one excitation wavelength and two detection channels Bossi et al., Nano Lett 2008

Multicolor STORM/PALM: activation photoactivation Deactivation 650 nm 360 nm 650 nm Fluorescent Dark Cy5 Cy3 532 nm

1 μm Bates, Huang, Dempsey and Zhuang, Science, 2007 █ Cy3 / Alexa 647: Clathrin █ Cy2 / Alexa 647: Microtubule Crosstalk subtracted Laser sequence …… Cy3A647Cy2A647

Multicolor imaging Multicolor capability Conventional SIM 4 colors in the visible range STED2 colors so far STORM/PALM3 activation x 3 emission

Live Cell Imaging

SIM STORM/PALM STED 2 µm Kner, Chhun et al., Nat Methods, 2009 Nagerl et al., PNAS, 2008 Schroff et al., Nat Methods, 2008

The limit of “Super-Resolution”

Unbound theoretical resolution STORM/PALM – 6,000 photons  5 nm – 100,000 photos during Cy5 life time  < 1 nm STED – 1:100 contrast of the donut  20 nm – Diamond defects: 8 nm NA d 2 S 1   N S = 1+ I/I s S =

Effective resolution: Probe size matters 100 nm Antibodies: ~ 10 nm Localization precision:22 nm Measured width by STORM: 56 nm Actual microtubule diameter:25 nm Localization precision:22 nm Measured width by STORM: 56 nm Actual microtubule diameter:25 nm Small fluorophores: ~ 1 nm 100 nm Fluorescent Proteins: ~ 3 nm Bacillus subtilis spore 500 nm < 1000 photons ~ 6000 photons

…… Conventional image STORM: a “time-for-space” strategy STORM image Time

Effective resolution: Density matters Frames for image reconstruction: ,0005,00040,000 Point to point distance ≈ Feature size Point to point distance < ½ Feature size This labeling density limit of resolution applies to all fluorescence microscopy methods Nyquist criteria

Effective resolution: Contrast matters photoactivation Deactivation 650 nm 360 nm 650 nm Fluorescent Dark e.g. 1% e.g. 99% 1% means… Sparsely labeled sample Densely labeled sample

Effective resolution: Contrast matters photoactivation Deactivation 650 nm 360 nm 650 nm Fluorescent Dark e.g. 1% e.g. 99% Homogeneous sampleMicrotubule Average point-to-point distance: 40 nm 14 nm 1% means… Common blinking dyes: >3% Cy5 + mercaptoethylamine : % mEosFP: 0.001% Common blinking dyes: >3% Cy5 + mercaptoethylamine : % mEosFP: 0.001%

Live cell STORM: Density matters 100x real time 1 μm Plasma membrane staining of a BS-C-1 cell Assuming: 1 molecule occupies 500 × 500 nm ↓ On average 0.1 point / 0.25 µm 2 ·frame ↓ 70 nm resolution ≡ 2000 frames ↓ 100 fps = 20 sec time resolution

Stochastic switching + particle tracking 1 μm Effective D = 0.66 μm 2 /s 1000 frames, 10 sec total time Number of localizations Displacement / frame (nm) DiD stained plasma membrane 1 μm

Comparison of time resolution 2D Spatial resolution Time resolution SIMWide-field120 nm9 frames (0.09 sec) STEDScanning60 nm 1 x 2 µm: 0.03 sec 10 x 20 µm: 3 sec STORM/PALMWide-field60 nm3000 frames (3 sec) 3D Spatial resolution Time resolution SIMWide-field120 nm15 frames x 10 (1.5 sec) STEDScanning60 nm 1 x 2 x 0.6 µm: 0.6 sec 10 x 20 x 0.6 µm: 60 sec STORM/PALMWide-field60 nm3000 frames (3 sec) – no scan!