Review of Super-Resolution Fluorescence Microscopy for Biology · PDF fileSTORM, their applications in biology, and technical considerations for imple-menting these methods. STRUCTURED

focal point reviewBONNIE O LEUNG AND KENG C CHOU

DEPARTMENT OF CHEMISTRY

UNIVERSITY OF BRITISH COLUMBIA

VANCOUVER BC V6T 1Z1 CANADA

Review of Super-Resolution FluorescenceMicroscopy for Biology

Several methodologies have been developed

over the past several years for super-resolution

fluorescence microscopy including saturated

structured-illumination microscopy (SSIM)

stimulated emission depletion microscopy

(STED) photoactivated localization microsco-

py (PALM) fluorescence photoactivation local-

ization microscopy (FPALM) and stochastic

optical reconstruction microscopy (STORM)

While they have shown great promise for

biological research these techniques all have

individual strengths and weaknesses This

review will describe the basic principles for

achieving super resolution demonstrate some

applications in biology and provide an over-

view of technical considerations for implement-

ing these methods

Index Headings Super-resolution Fluorescence

imaging Saturated structured-illumination mi-

croscopy Stimulated emission depletion mi-

croscopy Photoactivated localization

microscopy Fluorescence photoactivation lo-

calization microscopy Stochastic optical recon-

struction microscopy

INTRODUCTION

Fluorescence microscopy is cur-rently one of the more powerfuland versatile techniques available

for biological studies1 The techniqueuses fluorophores that have large ab-sorption cross-sections at a specificwavelength and emit light at a longerwavelength Because of the combinationof high absorption cross-section andhigh quantum efficiency fluorophore-labeled molecules are very bright andreadily distinguishable from other back-ground signals This optical propertymakes it fairly straightforward to obtainimages of the labeled molecules withhigh contrast With the development ofgenetically encoded fluorescent proteins(FPs) it has become possible to imageprotein expression localization andactivity in living cells2 Additionallythe combination of fluorescence andconfocal microscopies has allowed re-searchers to render the images in threedimensions (3D)3ndash5 However opticalmicroscopes have an inherent limitationin spatial resolution because of the wavenature of light The effect of lightdiffraction limits the resolution of anoptical microscope to approximatelyhalf of the wavelength of light used6ndash8

With the best optics the resolution offluorescence microscopy is limited to ~200 nm which cannot resolve many finecellular structures

To visualize cellular structures small-er than 200 nm researchers have reliedon electron microscopy (EM) As the de

Broglie wavelength of an electron ismuch shorter than visible light EM hasa much higher resolution than opticalmicroscopes Although atomic resolu-tion has been achieved9 EM has manypractical issues that limit its utility forbiological studies The main issue is thatelectrons carry an electric charge andthey interact with the medium throughwhich they travel Since electrons inter-act with molecules in air samples haveto be viewed in vacuum or under lowpressure The interaction between theelectrons and the sample prohibits theelectrons from penetrating deep into thesample Therefore fixation dehydrationand thin sectioning are required toprepare a sample for EM With all thesesample preparation procedures one ma-jor uncertainty in interpreting imagesacquired by EM is that these proceduresmay result in artifacts such as shrinkageof the specimen and alteration of tissuestructures10 The low pressure and thinsectioning requirements also make EMproblematic for live-cell imaging There-fore a microscopic technique that com-bines both the nondestructive nature ofoptical microscopy and the nanometerresolution of EM is highly desirable forbiological research

For decades researchers have beenseeking optical techniques that break thediffraction limit of an optical micro-

Received 29 June 2011 accepted 29 June2011

Author to whom correspondence should besent E-mail kcchouchemubccaDOI 10136611-06398

APPLIED SPECTROSCOPY 967

scope One of the most intuitive ways tobreak the diffraction limit was proposedby Synge in 1928 using a sub-wave-length aperture to image a surface11 Theconcept was realized by Ash andNichols in 1972 using microwaves witha wavelength of 3 cm and the diffrac-tion limit was overcome for the firsttime12 In 1984 the technique wasdemonstrated in the visible region1314

and it evolved into a rapidly expandingfield now known as near-field scanningoptical microscopy (NSOM)15 Onelimitation of NSOM is that the aperturehas to be placed within one wavelengthof the light to obtain sub-diffraction-limit resolution before the light spatiallydiverges The resolution decreases withincreasing distance between the apertureand the sample Therefore NSOM islimited to the study of surfaces and itsapplicability in biology is limited

On the other hand efforts to developfar-field optical microscopy with sub-diffraction-limit resolution have contin-ued Interference is probably the easiestmethod by which to obtain a sub-wavelength pattern for far-field micros-copy In 1978 C Cremer and T Cremerproposed that a laser-scanning micro-scope using a point-hologram wouldproduce a spot with a diameter consid-erably smaller than the wavelengthused16 In 1993 sub-diffraction-limitedaxial resolution was demonstrated byusing standing-wave excitation17 Instanding-wave fluorescence microscopy(SWFM) two counter-propagating non-focused laser beams are used to form astanding wave which creates an excita-tion field with closely spaced nodes andantinodes to improve the axial resolu-tion In 1995 a technique called imageinterference microscopy (I2M) was de-veloped In I2M the fluorescence fromboth sides of a sample is collected usingtwo opposed objective lenses These twoimages are combined by a beam-splitterand superposed on the camera tointerfere then images with a higherresolution can be obtained from theinterference pattern18 I2M can be com-bined with the incoherent interferenceillumination (I3) used in SWFM thencalled I5M microscopy to achieve ahigher resolution I5M has achieved anaxial resolution seven-fold better than atypical wide-field microscope19

An interferometric technique was alsoproposed for laser scanning microscopyIn 1992 a detailed theory of the 4Piconfocal fluorescence microscope waspublished by Hell and Stelzer20 In a 4Piconfocal microscope the specimen isilluminated by two coherent laser beamsthrough two opposed objective lenses21

The interference of the counter-propa-gating beams can greatly reduce the sizeof the focal spot in the axial direction Aseven-fold increase in axial resolutionwas achieved22 However these axialinterferometric techniques offered littleor no improvement in the lateral resolu-tion and have not been widely adoptedamong biologists To improve the lateralresolution for wide-field microscopy aninterferometric technique called struc-tured illumination microscopy (SIM)was proposed in the late 1990s2324

SIM superposes a known excitationpattern over the unknown sample pat-tern then a higher resolution image canbe deconvolved from the resultingfringes25 Using saturated excitationpatterns to introduce higher-order termssaturated structured-illumination micros-copy (SSIM) has shown a lateralresolution of 50 nm on sufficientlybright and photostable samples26

In 1994 a new type of scanningfluorescence microscope was proposedby Hell and Wichmann to improve thelateral resolution21 The basic principleof the new laser scanning fluorescencemicroscope is that the size of thefluorescence spot the point spreadfunction (PSF) can be reduced byemploying a second laser beam tostimulate the fluorescent emission inthe outer regions of the excitation spotThe second laser beam has a wavelengthlonger than the detected fluorescencetherefore the stimulated emission willalso emit at a longer wavelength notseen by the detector27 In this newapproach the shape of the effectivefluorescent spot which determines thespatial resolution is sharpened by thesecond laser beam via the stimulatedemission Creating a desired depletionbeam profile at the focal point is one ofthe major challenges to improve theresolution Between 1994 and 2006Hell and his co-workers published aseries of different designs to manipulatethe profile of the depletion beam212829

The final experimental setup that attract-ed a great deal of attention was carriedout in 2006 using the spatial phasemodulation pattern described by Torokand Munro3031 At this point thepotential of stimulated emission deple-tion (STED) microscopy in biologicalresearch became convincing and com-mercial STED microscopes have recent-ly become available

On the other hand new developmentsin single-molecule spectroscopy in theearly 1990s have enabled a differentapproach to achieving nanometer-scaleoptical microscopy3233 When imaginga single isolated molecule the size of themolecule that appears in the image iethe PSF is much larger than the actualsize of the molecule because of thediffraction limit However the locationof the molecule (or the center of thePSF) can be determined with a muchhigher accuracy A technique that de-pends on this effect was proposed byBetzig in 199535 In 2006 three re-search groups independently demon-strated super-resolution microscopy us-ing high-precision localization of singlefluorophores named as photoactivatedlocalization microscopy (PALM)35 sto-chastic optical reconstruction micros-copy (STORM)36 and fluorescencephotoactivation localization microscopy(FPALM)37

In this review we will briefly sum-marize far-field super-resolution opticalmicroscopy As SIM STED (F)PALMand STORM microscopy are now be-coming commercially available for biol-ogists their broad impact on biologicalresearch can be expected Howeverthese super-resolution microscopeswhich were mostly developed anddesigned by physicists and chemistsare not as user-friendly as conventionaloptical microscopes Currently there isstill no ideal system that offers high-speed 3D nanometer spatial resolutionwith multicolor capabilities for live-cellimaging Each technique has itsstrengths and weaknesses Thereforeresearchers need to be informed thatthere are trade-offs between resolutionsensitivity and depth of view Experi-mental designs are often also limited bythe available fluorescence probes Thisreview will describe the basic principleof SSIM STED (F)PALM and

968 Volume 65 Number 9 2011

focal point review

STORM their applications in biologyand technical considerations for imple-menting these methods

STRUCTURED ILLUMINATIONMICROSCOPY

The basic concept of structured illu-mination microscopy (SIM) is to illumi-nate a sample with patterned light andmeasure the moire pattern A moirepattern is an interference pattern createdby overlaying two grids with differentangles or mesh sizes If one of thepatterns is an unknown structure suchas Fig 1a and the other pattern is aknown pattern shown in Fig 1b themoire fringes (Fig 1c) obtained byoverlapping Figs 1a and 1b will containmore information about the unknownstructure than the original pattern in Fig1a If the known patterns have higherspatial frequencies the technique willoffer better spatial resolution Unfortu-nately the spatial frequencies that onecan create optically are also limited bydiffraction Therefore SIM can onlyimprove the resolution by a factor ofapproximately 225 Figure 2 shows theactin cytoskeleton of a HeLa cell imagedby conventional microscopy and SIMThe full widths at half-maximum(FWHM) of the fibers are 280 nm fora conventional microscope and 110 nmfor SIM

In 2005 Gustafsson further improvedthe resolution using nonlinear structured

FIG 1 (a) A sample containing unknown structures (b) A known structured patterned (c) The moire pattern generated by overlaying (a)and (b)

FIG 2 The actin cytoskeleton of a HeLa cell as imaged by (a c) conventional and (b d)structured illumination microscopy (c d) Enlargements of the boxed areas in (a) and (b)respectively The widths (FWHM) of the finest protruding fibers [small arrows in (a b)] are110 6 120 nm in (b) compared to 280 6 300 nm in (a) (Reproduced with permission fromRef 25)


illumination26 Under intense illumina-tion the emission intensity can dependnonlinearly on the illumination intensityAs shown in Fig 3a the fluorescenceemission intensity (rate) will saturate athigh illumination intensity This nonlin-

earity would cause the illuminationpattern to saturate at the high-intensityillumination region as indicated in Fig3b With high illumination intensity theeffective illumination pattern in Fig 3bcontains harmonics with spatial frequen-

cies that are higher than the originalillumination spatial frequency Thesehigher spatial frequency componentshave provided a resolution of 50 nmon bright and photostable samples26

Figure 4 shows 50 nm fluorescent beadsimaged by conventional microscopy(Fig 4a) conventional microscopy plusfiltering (Fig 4b) linear SIM (Fig 4c)and saturated structured-illumination mi-croscopy (SSIM) (Fig 4d) taking intoaccount three harmonic orders in theimage processing

Multicolored 3D-SIM has been re-ported 3D-SIM was achieved by illu-minating the sample with three beams ofinterfering light and observing theinterference pattern along the x y andz axes As shown in Fig 5 C2C12 cellnuclei were stained with antibodiesagainst lamin and nuclear pore complex(NPC) and chromatin was stained with4 06-diaminidino-2-phenylindole (DA-PI) The study could image single NPCsthat were co-localized with the laminnetwork and peripheral heterochromatinFurthermore double-layered invagina-tions of the nuclear envelope in pro-phase was obtained which were

FIG 3 (a) The nonlinear dependence of the fluorescent emission rate on the illuminationintensity in the saturation regime (b) The effective emission pattern resulting fromsinusoidally patterned illumination with peak pulse energy densities of (from bottom to top)025 1 4 16 and 64 times the saturation threshold (Reproduced with permission from Ref26)

FIG 4 A field of 50 nm fluorescent beads imaged by (a) conventional microscopy (b) conventional microscopy plus filtering (c) linearstructured illumination and (d) saturated structured illumination taking into account three harmonic orders in the processing (Reproducedwith permission from Ref 26)


focal point review

previously only acquired via electronmicroscopy38 3D-SIM is currently the3D super-resolution imaging techniquethat can easily be expanded to threecolors or more which makes it the mostwidely accepted super-resolution mi-croscopy technique among biologists

STIMULATED EMISSIONDEPLETION MICROSCOPY

Stimulated emission depletion(STED) microscopy is based on laserscanning confocal microscopy In a laserscanning confocal microscope a laserbeam is focused by an objective lensinto a small focal spot within a speci-men then an image is acquired point-by-point by scanning either the speci-men or the laser beam The point-by-point approach allows an easy 3Dreconstruction of a complex object Forbiological studies the specimens areusually fluorophore-labeled Thereforethe wavelength of the excitation beamshould match either the single-photon(1P) or two-photon (2P) absorption(excitation) bands of the fluorophoreBecause of the important applications offluorescent proteins in biological sci-ence the following discussion will usegreen fluorescent protein (GFP) as anexample Figure 6 shows the absorptionand emission spectra of enhanced GFP(EGFP) For STED microscopy both 1Pand 2P excitations have been demon-strated for GFP using excitation beamsat 490 nm and 840 nm respectively3940

In both cases a second laser beam near

580 nm was used as the depletion beam

to reduce the effective PSF

Figure 7a shows the energy diagram

for a typical STED process Excitation

from the ground state S0 via 1P or 2P

processes quickly relaxes to lower

vibrational levels in the S1 state With-

out an external field the population in

S1 decays to the vibrational ground-state

in S0 via spontaneous emission which

peaks at 509 nm as shown in Fig 6

(green curve) For GFP a 580 nm beam

can effectively stimulate the emission to

the higher vibrational states in S0213940

Since 580 nm is not within the detection

band the stimulated emission does not

enter the detection system and the

spontaneous emission is depleted Fig-

FIG 5 Simultaneous imaging of DNA nuclear lamina and NPC epitopes in C2C12 cells by confocal laser scanning microscopy (CLSM)CLSMthorn deconvolution and 3D-SIM C2C12 cells are immunostained with antibodies against lamin B (green) and antibodies that recognizedifferent NPC epitopes (red) DNA is counterstained with DAPI (blue) With 3D-SIM the spatial separation of NPC lamina and chromatin andchromatin-free channels underneath nuclear pores are clearly visible (Reproduced with permission from Ref 38)

FIG 6 The absorption and emission spectra of EGFP The arrows indicate the wavelengthfor 1P excitation at 490 nm (blue) 2P excitation at 840 nm (red) and depletion beam at 580nm (dark yellow) for STED microscopy The detection band is indicated by the detection bar(green)


ure 7b shows the detected fluorescence

of GFP as a function of the 580 nm

beam intensity when the excitation and

depletion beams are spatially over-

lapped Generally 100 depletion of

the spontaneous emission is not achiev-

able but a 90ndash95 depletion will

produce a STED image with an accept-

able contrast ratio In principle the

depletion efficiency will increase if the

depletion beam is blue shifted Howev-er the depletion beam is also capable of

exciting the fluorophore and should not

be too close to the absorption band

The layout of a typical STED micro-

scope is shown in Fig 8 Two laser

beams at different wavelengths arerequired for STED microscopy As

mass-produced laser diodes are becom-

ing more powerful 1P excitation STED

likely can be built with lower cost41

However 2P excitation STED has adeeper penetration depth which allowsimaging of thick tissues4243 The exci-tation and depletion beams are com-bined using a dichroic mirror prior toentering the objective lens The dough-nut-shaped focal intensity profile of thedepletion beam (Fig 8b) can be ob-tained using a spiral spatial phasemodulation pattern proposed by Toroket al31 Both pulsed and continuouswave (CW) lasers have been used forSTED microscopy If pulsed lasers areused these two beams have to besynchronized temporally to maximizethe spontaneous fluorescence depletionWith the doughnut-shaped depletionbeam the spontaneous fluorescenceemission from the periphery of theexcitation spot is inhibited by thestimulated emission resulting in afluorescent spot size below the diffrac-tion limit Because the intensity of thedepletion beam is near zero at the centerof the doughnut the spontaneous fluo-rescence emission at the central regionremains unchanged

Figures 9a and 9b show EGFP-taggedcaveolin 1 (Cav1-GFP) in Chinesehamster ovary (CHO) cells imaged bya 2P fluorescence microscope withoutand with the doughnut-shaped depletionbeam respectively40 Fluorescence spotsof various sizes can be seen in Fig 9because caveolins exist in differentcellular structures in the cytoplasmThe small Cav1-GFP domains are pre-sumed to be caveolar vesicles or pro-teinndashlipid complexes which havediameters of 50 to 100 nm under anelectron microscope4445 Figures 9c and9d are magnified views of the markedareas in Figs 9a and 9b respectivelyThe regular 2P fluorescence microscopehas a diffraction-limited excitation spotwith a FWHM of ~ 250 nm With thedoughnut-shaped depletion beam twocaveolar vesicles located within a dis-tance smaller than the 250 nm diffrac-tion limit can be distinguished

The resolution of the STED micro-scope can be described by

Dr~k

2NAffiffiffiffiffiffiffiffiffiffiffi1thorn I

Is

q

FIG 7 (a) Energy diagram for the STED microscopy and (b) measured 2P fluorescencedepletion efficiency for GFP The excitation beam was a 130 fs Ti sapphire laser with awavelength of 840 nm The depletion beam was 200 ps with a wavelength of 580 nm

FIG 8 (a) The layout of a typical STED microscope (b c) The intensity profiles of thedepletion and excitation beams respectively which were recorded by the scattering lightfrom a 100 nm gold particle (Reproduced with permission from Ref 40)


focal point review

where NA is the numerical aperture ofthe objective lens k is the wavelength oflight I is the intensity of the STEDdoughnut maximum and Is is the STEDintensity used to reduce the fluorescenceintensity by one-half4647 The spatialresolution of STED microscopy dependson the intensity of the depletion beamAs the intensity of the depletion beamincreases the resolution improves Al-though there is no theoretical resolutionlimit in practice photodamage of thesample usually sets the intensity limit onthe depletion beam particularly forbiological samples

The doughnut-shaped depletion beamworks well to deplete the spontaneousfluorescence emission in the lateraldirection but it offers no resolutionimprovement in the axial direction (z-axis) Figure 10 shows the intensityprofiles of the excitation beam thedoughnut-shaped depletion beam andthe resulting effective PSF As shown inFigs 10b and 10e the intensity of thedoughnut-shaped depletion beam haslittle intensity at the center along the z-axis Therefore it does not reduce thelength of the effective PSF as shown inFig 10f To improve the axial resolu-tion a third beam has been used todeplete the spontaneous fluorescence inthe axial direction48

Two-color STED microscopy wasdemonstrated by Donnert et al in 2007using green and red fluorescent beads49

Multicolor imaging significantly in-creases the complexity of the experi-mental setup as two laser beams withdifferent colors are needed for eachfluorophore one for excitation and onefor depletion Simultaneous imaging oftwo colors is not straightforward forSTED microscopy because four laserbeams with four different wavelengthscan produce undesired interference Toavoid the interference Donnert et alcarried out the two-color STED imagingby scanning one color at a timeRecently Buckers et al demonstrated aquasi-simultaneous recording of two-color STED by a 40 ns time shiftbetween the pairs of excitation anddepletion beams50 Figure 11 shows athree-color STED image by using twodyes with similar spectral properties butdifferent lifetimes (KK 114 ATTO647N) and a third dye with different

absorption and fluorescence spectra

(ATTO 590) to label the lamin tubulin

and clathrin in human glioblastoma

cells respectively50

One of the first studies addressed by

STED was the outcome of synaptic

vesicles employed after synaptic trans-

mission The main limitation prior to

STED was that the 50 nm vesicles were

too small and closely packed to be

resolved by conventional fluorescence

microscopy Vesicles that were under-

going endocytosis were labeled by

attaching a green emitting dye to

antibodies capable of binding to the

synaptotagmin protein present on the

vesicle membranes30 Comparison be-

tween the membrane-associated and

internalized fluorescent dots within cul-

tured hippocampal neurons revealed that

FIG 9 GFP-tagged caveolin 1 in CHO cells imaged by a 2P STED microscope (a) withoutand (b) with the doughnut-shaped depletion beam (c) Magnified view of the marked area in(a) (d) Magnified view of the same marked area in (b) The scale bars are 1 lm in (a) and (b)and 200 nm in (c) and (d) The pixel sizes are 40 nm for (c) and 20 nm for (d) (Reproducedwith permission from Ref 40)

FIG 10 The intensity profiles of the excitation beam the doughnut-shaped depletionbeam and the resulting effective PSF in (andashc) the xndashy plane and (dndashf) the xndashz plane (d-f)


the synaptotagmin protein remains inclusters during recycling and is notdiffuse throughout the membrane

Acetylcholine receptor (AChR) su-pramolecular aggregates involved infast synaptic transmission were exam-ined with STED microscopy51 Thestudy found that the removal of choles-terol from the cell surface changes theorganization and distribution of theAChR aggregates Compared to con-trols the decreased cholesterol resultedin fewer AChR aggregates of larger sizeThis cholesterol-dependent clustering ofAChR aggregates was well below thediffraction limit and could not bedetected by a conventional opticalmicroscope

Dendritic spines have been shown tohave structural diversity related to po-tential disease such as depression42

These spines can vary from 02 to 2lm in size and hence are easilyassessable by conventional fluorescencemicroscopy However changes in thesize and shape of the spines are belowthe diffraction limit STED was able toimage the dendritic spines of YFP-transgenic mice as time lapse movies at20 sframe With STED changes in theneck width and curvature of the spinescould be captured

STED was also used to investigate thespatial distribution of Human Immuno-deficiency Virus (HIV) between infecteddendritic cells and uninfected CD4thorn Tcells52 STED was able to resolvefluorescent virus clustering at the syn-apses with single-virion resolution

In the last ten years studies haveemerged showing that the cytoskeletonthat was previously believed to existonly in eukaryotic cells has bacterialhomologs found in almost all prokary-otes53 The tubulin homolog cell divi-sion protein FtsZ helix in fixed cells wasimaged with STED and its structure wasfound to be patchy and discontinuouswhich contrasts sharply with the diffrac-tion-limited conventional fluorescencemicroscopy which showed FtsZ to bea smooth structure Furthermore STEDmicroscopy showed a much narrowerperiodic structure of FtsZ (140ndash200 nm)compared to conventional fluorescencemicroscopy (550 nm) this tight helixwas a new sub-cellular structure imagedfor the first time with STED microscopy

STED has also been used to probe thelipids of cell membranes Lipid lsquolsquoraftsrsquorsquohave been difficult to detect in livingcells due to their small size (5ndash200 nm)and rapid diffusion54 Sphingomyelinwhich is believed to form cholesterol-

mediated lipid nanodomains and a non-integrating lipid phosphoethanolaminewere fluorescently labeled and thediffusion of these single molecules wasmonitored by both confocal and STEDmicroscopy While the confocal resultsshowed comparable results for bothlipids the STED measurements revealedphosphoethanolamine with one type ofdiffusion with fluorescent bursts 1 mswhile sphingomyelin showed heteroge-neous bursts ranging from 1 to 50 msDepleting the cholesterol in the cellmembrane by cholesterol oxidase re-vealed that the longer transit time ofsphingomyelin was found to result fromcholesterol-mediated molecular trap-ping

Syntaxin 1 and syntaxin 4 clustersfound on the plasma membrane areinvolved with dockingfusion of secre-tory vesicles and caveolae respective-ly55 Although it was previouslybelieved that cholesterol was involvedin cluster formation STED microscopystudies revealed that cholesterol alonedid not result in correct clustering butthat the SNARE motif (a homologoussequence of 60 to 70 amino acids)involving cytoplasmic proteinndashproteininteractions is required for cluster for-mation Using a construct without theSNARE motif of syntax 1A significantlyreduced its co-clustering ability withunmodified syntax 1A

Proteins such as flotillin-1 and flotti-lin-2 which are associated with theformation of the neurotoxic plaquesfound in Alzheimerrsquos disease have beenprobed with STED56 It was found thatdepletion of flotillin-2 resulted in asignificant reduction in the amyloidprecursor protein (APP) cluster sizewhile flotillin-1 did not result in variedcluster size or distribution The resultsfrom this study suggest that the presenceof flotillin-2 can stimulate APP cluster-ing potentially promoting endocytosiswhich can result in the release of theneurotoxic amyloid-beta peptide

STED microscopy is also capable ofimaging within a living cell Usingfluorescently labeled proteins the endo-plasmic reticulum (ER) of mammalian

FIG 11 Lamin (blue) Clathrin (red) and Tubulin (green) imaged with confocal and STEDmicroscopy (Reproduced with permission from Ref 50)

SNARE is an acronym for soluble N-ethyl-maleimide-sensitive factor attachment proteinreceptors


focal point review

cells was imaged using STED48 Sincethe ER can be folded in a complexstructure STED was able to profilesingle stacked ER strands using 150nm optical sectioning Furthermoreusing a lateral spatial resolution of 50nm time-resolved movies (10s framerate) of ER movement were obtainedrevealing ER closure formations and re-openings

SINGLE-FLUOROPHORESUPER-RESOLUTIONMICROSCOPY

The basic principle of single-fluoro-phore-based super-resolution microsco-py is that the position of a spatiallyisolated fluorophore can be determinedwith an accuracy higher than the widthof the PSF Single-molecule detection isthe first step in locating the moleculebeyond the diffraction limit Single-molecule spectroscopy was first demon-strated in 1989 at liquid-helium temper-atures32 However with a modernelectron multiplying charge-coupled de-vice (CCD) single-molecule imaging isnow easily achievable at room temper-ature Given an isolated single moleculeas shown in Fig 12a the image size ofthe molecule on a CCD is diffractionlimited as shown in Fig 12b If it isknown that there is only one moleculethe position of the molecule can bedetermined by fitting the image in Fig12b with a Gaussian distribution tolocate the center of the PSF Theuncertainty in the position of the mole-cule is determined by the uncertainty infitting the center of the PSF Theoreti-cally the precision scales inversely withthe square of the number of detectedphotons which is described by

Dlocalization~DPSFffiffiffiffi

Np

where Dlocalization is the precision of thelocalization DPSF is the width of thePSF and N is the number of photonsdetected from the molecule OnceDlocalization is obtained the image of themolecule can be reconstructed as shownin Fig 12c

For single-fluorophore-based super-resolution microscopy each moleculemust be well separated or a false imagereconstruction will arise as shown in

Figs 12d through 12f Figure 12d showstwo molecules located within the dif-fraction limit These two molecules arenot distinguishable on the CCD asshown in Fig 12e In this case onlyone molecule will be reconstructed andit will be located near the center of thesetwo molecules Therefore a sample withwell-separated molecules which areindividually distinguishable on theCCD is required to avoid a false imagereconstruction

The requirement for spatially distin-guishable molecules generally cannot befulfilled in biological samples becauseproteins within a cell have very highlocal density and are not spatiallydistinguishable on a CCD This difficul-ty was overcome in a very clever wayusing photoactivatable fluorescent la-bels5758 Photoactivatable fluorescentlabels allow a controlled activation of asubset of the fluorescent moleculesBecause the number of activated mole-cules is small the chance of having twoor more of them located within adiffraction-limited volume is lowTherefore it is possible to image asubset of well-separated molecules oneat a time with high resolution then thecomplete image can be obtained byadding all images of the subsets togeth-

er This approach was independentlyreported by three research groups in200635ndash37 The methodology and theirapplications are described as follows

PHOTOACTIVATEDLOCALIZATIONMICROSCOPY

Photoactivated localization microsco-py (PALM) was first carried out using aphotoconvertible FP named EosFP35

The emission wavelength of photocon-vertible FPs can be optically convertedfrom one wavelength to another Undernormal conditions EosFP emits greenfluorescence at 516 nm Upon irradiationnear 400 nm its fluorescence changes to581 nm because of a photo-inducedmodification involving a break in thepeptide backbone next to the chromo-phore59 This optical property makes itpossible to convert a subset of EosFPs toemit in the yellow region (581 nm)When the number of converted proteinsis small the proteins will be wellseparated and can be imaged in theyellow region with high resolutionWhen this particular subset of EosFPsis photobleached another subset ofEosFPs can be converted and imagedThis process may be cycled 104 to 105

FIG 12 (a) Location of a single molecule (b) Image observed by a CCD for the molecule in(a) (c) Reconstructed images of the molecule in (a) (d) Two closely located molecules (e)Molecules in (d) observed by a CCD (f) A false image reconstruction that places a moleculenear the center of these two molecules


times until the population of EosFPs isdepleted FPALM utilizes a similarapproach and was first demonstratedusing photoactivable GFP37

Since 2002 many other photoconver-tible photoactivatable and photoswitch-able FPs have been developed Photo-activatable FPs can be activated from adark state to a bright state using UVlight and photoswitchable FPs canalternatively be switched on or off withspecific illumination A summary ofthese proteins can be found in Ref 60although not all of them have beenproven to be effective for super-resolu-tion microscopy

Three-dimensional FPALM wasachieved in 2008 using biplane (BP)detection61 A beam-splitter splits thefluorescence light into a shorter andlonger path to form two detection planesfor z-position determination A resolu-tion of ~30 nm laterally and ~75 nmaxially was achieved by BP FPALM61

In 2009 interferometric PALM(iPALM) was also introduced and pro-vided sub-20-nm 3D protein localiza-tion62

Two-color PALM was implementedin 2009 COS-7 cells were tagged withtransferring receptor (TfR)-PAmCherry1and PA-green fluorescent protein (GFP)-clathrin light chain (CLC) and thenalternately imaged at 561 nm and 468nm to excite the red (PAmCherry1) andgreen (PAGFP) dye as shown in Fig1363 The results showed 200 nmclusters of transferring receptors and

clathrin light chains at a lateral resolu-tion of 20 nm

PALM was used to study the E colichemotaxis receptor cluster aggregationat localized poles of the bacteria64

Three proteins (the Tar receptor CheYand CheW) involved in chemotaxis wereimaged with a resolution of 15 nm Nocharacteristic cluster sizes were foundThe results showed that the receptorswere not location specific but ratherdiffusion (or stochastic self-assembly) ofthe receptors can explain the imagesobtained with PALM

Dendritic spines protrude from neu-ronal dendrites and hold postsynapticmolecules65 Changes in the shape of thespine arise from the actin cytoskeletonwithin the dendritic spines PALM wasused to track a single molecule of F-actin in cultured hippocampal neuronswith a resolution of 30 nm This studyfound the dynamics of F-actin to behighly heterogeneous The results pro-pose short actin filaments in the dendrit-ic spine

The FtsZ protein is involved inprokaryotic cell division and is shapedlike a ring called the Z-ring66 It hasbeen difficult to image the Z-ring withelectron microscopy due to the densecytoplasm PALM microscopy was ableto image the Z-ring in E coli with aspatial resolution of 35 nm and showedthat in addition to a ring structure thering exists as a helical conformationThe thickness of the ring is approxi-mately 110 nm and is likely formed by a

loose bundle of randomly overlappedFtsZ protofilaments

STOCHASTIC OPTICALRECONSTRUCTIONMICROSCOPY

Stochastic optical reconstruction mi-croscopy (STORM) was first introducedusing Cy3ndashCy5 dye pairs as the opticalswitch to activate a subset of fluoro-phores36 The Cy5 works as the primaryfluorophore which can be reversiblyswitched between a fluorescent and adark state in a controlled fashion by 532nm and 633 nm light respectively andthe Cy3 is used to facilitate the switch-ing of Cy558 The optical switch can becycled on and off hundreds or thousandsof times before becoming permanentlyphotobleached A resolution of 20 nmwas demonstrated for RecA-coated cir-cular plasmid DNA36

Figure 14 shows the comparisonbetween imaging microtubules in a BS-C-1 cell using conventional confocalfluorescence versus STORM Multi-color STORM was demonstrated formicrotubules Furthermore the tubuleswere imaged with clathrin-coated pitswhich are involved in receptor-mediatedendocytosis67 Cy2-Alexa647 and Cy3-Alexa647 were used to label the micro-tubules and clathrin respectively Laserswith wavelengths of 457 and 532 nmwere used to selectively excite theswitch pairs STORM was able to imagethe microtubules separately from thecircular clathrin pits with ~ 30 nmspatial resolution

Three-dimensional STORM wasachieved by introducing astigmatism tothe image with a cylindrical lens Thusthe position of the fluorophore wasmonitored by examining the ellipticityof the circular fluorophore Imagesabove and below the focal plane wereellipsoidal and by fitting the image witha 2D elliptical Gaussian function thepeak widths of the xy coordinates wereobtained which allowed for the calcu-lation of the z coordinate68 Recently3D spatial resolution of ~30 nm in thelateral direction and ~50 nm in the axialdirection with time resolution as fast as1 to 2 simage have been reported69 3DSTORM has been used to image micro-tubule networks in green monkey kidneyepithelial cells labeled with a Cy3-

FIG 13 Distribution of TfR and CLC using two-color PALM (right) compared toconventional total internal reflection (TIRF) microscopy (left) TfR clusters are red CLCclusters are green Scale bar is 2 lm (Reproduced with permission from Ref 63)


focal point review

Alexa647 switch 3D STORM was ableto capture multiple layers of microtubulefilaments in cell cross-sections and twomicrotubules 100 nm apart in the zdirection were easily separated

Two-color imaging combined with3D STORM of tubular mitochondriaand microtubules in BS-C-1 cells la-beled with A405-Cy5 and A555-Cy5 (orA488-Cy5) respectively has been re-ported70 Laser pulses at 405 nm wereused to activate A405-Cy5 while laserpulses at 542 nm (or 460 nm) were usedto activate the A555A488-Cy5 dyesThe probes were imaged during theactivated state by a 657 nm laser todetermine the 3D position of the fluo-rophores Compared to conventionalfluorescence microscopy STORM wasable to minimize false contacts betweenthe mitochondria and microtubulesshowing a 150 nm separation between

microtubule and mitochondria that ap-peared to be touching in the conven-tional fluorescence microscopy images

Recently Heilemann et al introduceddirect STORM (dSTORM) which usesconventional photoswitchable fluores-cent dyes that can be reversibly cycledbetween a fluorescent and a dark state byirradiation with light of different wave-lengths71 This approach does not re-quire special fluorophore pairs used inSTORM dSTORM was able to visual-ize cellular structures with a resolutionof approximately 20 nm without theneed of an activator molecule

TECHNICALCONSIDERATIONS

Photobleaching Photobleaching isthe light-induced destruction of fluoro-phores The phenomenon is often an

important limiting factor in imagingbiological samples especially for time-lapse studies When a fluorophore ab-sorbs a photon an electron is excitedfrom the ground state to an excited stateWhen fluorophores are in the excitedstate they are more likely to react withother molecules The mechanism ofphotobleaching is not fully understoodfor most molecules but it is generallyassumed to be related to excitations totriplet states as triplet states have longerlifetimes and are more reactive72 Thereare anti-photobleaching agents that re-duce the amount of oxygen in thesample to prevent reactions with oxy-gen but many of them are toxic forliving cells Some fluorophores are morephotostable than the others A typicalorganic dye can emit 105 to 106 photonsbefore it is photobleached and a fluo-rescent protein typically can emit 104 to

FIG 14 (B D F) STORM and (A C E) conventional fluorescence imaging of microtubules in a mammalian cell (Reproduced withpermission from Ref 67)


105 photons Because the directions ofthe emissions are random only afraction of them can enter the detectionsystem The number of detected photonsis a limiting factor for super-resolutionmicroscopy

In general photobleaching is more aproblem for SSIM and STED than forthe other super-resolution imaging ap-proaches as these two techniques requiresaturated fluorescence excitation anddepletion respectively For SSIM pho-tobleaching will cause dimmer subse-quent images causing problems for thereconstruction For STED microscopybecause the resolution is improved bydecreasing the effective PSF the detect-ed photon counts decrease dramaticallyas the size of the PSF decreasesAlthough the photon count per pixelcan be increased by increasing the dwelltime per pixel a longer dwell time willincrease the photobleaching For biolog-ical studies the number of fluorophoresis predetermined by the nature of thespecimens In these cases the repetitionrate of the excitation can be reduced toreduce photobleaching It has beenreported that a 5- to 25-fold increase intotal fluorescence yield for GFP andAtto532 can be obtained when therepetition rate of pulsed illumination islowered to 1 MHz73 This fluores-cence signal gain was observed both for1P and 2P excitation However it willsignificantly increase the image acquisi-tion time Generally 50 photonspixelwill produce a good quality image forSTED Because of the low photoncount images obtained by STED areoften smoothed or restored by variousfilters to remove noise30474974

For single-fluorophore-based super-resolution microscopy such as (F)PALMand STORM the spatial resolution islimited by the number of photons asindicated in Eq 2 Unlike STED whichdepletes the fluorescence to improveresolution single-fluorophore-based su-per-resolution microscopy collects themaximum number of photons emittedfrom each molecule and the resolutionis determined by the number of photonsdetected The number of photons amolecule can emit before it is photo-bleached (or switched off) varies Forexample EosFP one of the brighter PA-FPS for PALM can emit ~750 photons

per molecule75 and the switchable fluo-rophore pair Cy3ndashCy5 used for STORMcan provide ~6000 photons per moleculeper switching cycle and lasts ~200switching cycles5867 A review of fluo-rescent probes for super-resolution imag-ing can be found in Ref 75 Overallphotoswitchable dyes have larger numbersof collected photons per molecule com-pared with fluorescence proteins

Quantum dots (QDs) offer excellentphotostability for long-term imagingSeveral applications of QDs for super-resolution microscopy have been dem-onstrated76ndash79 Compared to traditionalfluorophores QDs have a broader ab-sorption spectrum and a narrower emis-sion spectrum which is beneficial forsimultaneous multicolor imaging withthe same excitation light8081 QDs havebeen estimated to be up to 20 timesbrighter and 100 times more stable thanorganic dyes such as rhodamine82

However for specific labeling QDsmust be conjugated to biomolecules thatprovide binding specificity8384 Whilecommercial toolboxes and establishedlabeling protocols are widely availablefor conjugating dyes to biomoleculesconjugating QDs to biomolecules stillremains challenging85

Spatial Resolution For STED mi-croscopy the achievable resolution isstrongly dependent on the photostabilityof a sample A resolution of 58 nm hasbeen achieved on color centers of nano-crystals86 For biological samples aresolution of 50ndash100 nm can be ob-tained without causing significant dam-age Beside photobleaching a commonissue that reduces the resolution ofSTED microscopy is a non-ideal deple-tion beam profile From a practicalstandpoint it is difficult to generate anideal doughnut-shaped depletion beamwith zero intensity at the center Whenthe wavefront (spatial phase) of theincident beam is distorted or the spatialphase modulation is imperfect theresulting doughnut will be asymmetricand there is a finite intensity at thecenter The asymmetric depletion profileproduces an asymmetric effective PSFbut it is not a critical issue when the sizeof the effective PSF is smaller orcomparable to the scale of interestOne major problem with having a finiteintensity at the center of the doughnut

beam is that it also depletes thefluorescence at the center and reducesthe number of photons detected Anoth-er problem comes from the aforemen-tioned excitation by the depletion beamIn an ideal case the fluorescence excitedby the doughnut-shaped depletion beamcan be mostly excluded from enteringthe detection system by a spatial filtersuch as a pinhole in front of thedetector However when the profile ofthe depletion beam is imperfect thefluorescence excited by the depletionbeam will increase Typically the wave-front distortion can be minimized byputting the laser beam through a single-mode fiber or a pinhole and thedepletion beam profile can be checkedby measuring the scattering light from agold nanoparticle to ensure a reasonablygood doughnut profile

For single-fluorophore-based micros-copy such as (F)PALM and STORMthe resolution limit is given by thenumber of detected photons as indicatedin Eq 2 Theoretically a 10 nmresolution is possible with ~1000 pho-tons but practically a resolution of ~ 20nm in the lateral direction is commoneven with 6000 photons3667 Manyother factors can potentially increasethe uncertainty in determining the posi-tion of a molecule based on the imagerecorded on a CCD The mechanicaldrift of the imaging system is typicallyin the range of tens of nanometers Forslow imaging luminescent beads addedsparsely to the sample have been used tocalibrate the drift during the post-acquisition analysis35 The location ofa fluorophore is usually determined byGaussian fitting However the PSF of afluorophore may not be a perfectGaussian function8788 It has beenreported that the fluorescence intensitydistribution of a molecule also dependson its three-dimensional orientation89

Additionally the camera pixels maydiffer in quantum efficiency and gain90

CONCLUSIONS ANDOUTLOOK

Super-resolution fluorescence micros-copy developed in the past five years hasshown a number of important applica-tions and shows great promise for abetter understanding of biology Al-though several methodologies have been


focal point review

developed for sub-diffraction-limit im-aging these techniques have theirstrengths and weaknesses The advan-tage of STED microscopy is that it canbe carried out using traditional dyes andfluorescent proteins The disadvantage isthat STED is technically more compli-cated and the resolution is typicallylimited to 50 to 100 nm for biologicalsamples On the other hand (F)PALMand STORM are technically easier toimplement and a higher resolution (~20nm) is easier to achieve However(F)PALM and STORM require specialphotoactivatable (or photoswitchable)fluorophores There is still no idealsystem that offers user-friendly high-speed multicolor 3D and multicolorimaging with a nanometer spatial reso-lution However super-resolution mi-croscopy is still in its infancy and newdevelopments in the next 5 to 10 yearscan be expected

Super-resolution fluorescence micros-copy for biological research has been acollaborative work among physicistschemists and biologists Further devel-opment will continue to rely on newoptical methods new fluorophores anda better understanding of biology at thenanometer scale The research willbenefit from a more sensitive CCDfaster data processing and cheaperlasers Certainly the development ofphotobleach-resistant fluorophores willlead to greater advancement in super-resolution imaging Quantum dots canpotentially become excellent labels forsuper-resolution microscopy as theyhave great photostability and brightnessIn the future new developments willenable more user-friendly multicolored3D nanoscale imaging to unlock morebiological secrets that still remain hid-den

1 J W Lichtman and J A Conchello NatMethods 2 910 (2005)

2 B N G Giepmans S R Adams M HEllisman and R Y Tsien Science 312 217(2006)

3 M Minsky Scanning 10 128 (1988)4 J B Pawley Zool Stud 34 117 (1995)5 T Wilson J Microsc 242 111 (2011)6 F M Grimaldi Physico-Mathesis de Lumine

Coloribus Et Iride Aliisque Adnexis LibriDuo (Bologna Italy 1665 reprinted by Kes-singer Publishing Whitefish MT 2010)

7 H Helmholtz The Monthly MicroscopicalJournal 16 15 (1876)

8 E Abbe Journal of the Royal MicroscopicalSociety ser 2 1 388 (1881)

9 R Erni M D Rossell C Kisielowski and UDahmen Phys Rev Lett 102 4 (2009)

10 R F E Crang and K L Klomparens EdsArtifacts in Biological Electron Microscopy(Plenum Press New York 1988)

11 E H Synge Philos Mag 6 356 (1928)12 E A Ash and G Nicholls Nature 237 510

(1972)13 D W Pohl W Denk and M Lanz Appl

Phys Lett 44 651 (1984)14 A Lewis M Isaacson A Harootunian and

A Muray Ultramicroscopy 13 227 (1984)15 E Betzig and J K Trautman Science 257

189 (1992)16 C Cremer and T Cremer Microsc Acta 81

31 (1978)17 B Bailey D L Farkas D L Taylor and F

Lanni Nature 366 44 (1993)18 M G L Gustafsson D A Agard and J W

Sedat in Three-Dimensional MicroscopyImage Acquisition and Processing II TWilson and C J Cogswell Eds (SPIE-IntSoc Opt Engineering Bellingham1995)vol 2412 p 147

19 M G L Gustafsson D A Agard and J WSedat J Microsc-Oxford 195 10 (1999)

20 S Hell and E H K Stelzer Opt Commun93 277 (1992)

21 S W Hell and J Wichmann Opt Exp 19780 (1994)

22 S W Hell S Lindek C Cremer and E H KStelzer Appl Phys Lett 64 1335 (1994)

23 M G L Gustafsson D A Agard and J WSedat United States Patent Patent number5671085 (1997)

24 R Heintzmann and C Cremer Proc SPIE-IntSoc Opt Eng 3568 185 (1998)

25 M G L Gustafsson J Microsc-Oxford 19882 (2000)

26 M G L Gustafsson Proc Natl Acad SciUSA 102 13081 (2005)

27 S W Hell and M Kroug Appl Phys B-Lasers Opt 60 495 (1995)

28 T A Klar E Engel and S W Hell PhysRev E 64 06613 (2001)

29 M Dyba and S W Hell Phys Rev Lett 884 (2002)

30 K I Willig S O Rizzoli V Westphal RJahn and S W Hell Nature 440 935 (2006)

31 P Torok and P R T Munro Opt Exp 123605 (2004)

32 W E Moerner and L Kador Phys Rev Lett62 2535 (1989)

33 M Orrit and J Bernard Phys Rev Lett 652716 (1990)

34 E Betzig Opt Exp 20 237 (1995)35 E Betzig G H Patterson R Sougrat O W

Lindwasser S Olenych J S Bonifacino MW Davidson J Lippincott-Schwartz and HF Hess Science 313 1642 (2006)

36 M J Rust M Bates and X W Zhuang NatMethods 3 793 (2006)

37 S T Hess T P K Girirajan and M DMason Biophys J 91 4258 (2006)

38 L Schermelleh P M Carlton S Haase LShao L Winoto P Kner B Burke M CCardoso D A Agard M G L GustafssonH Leonhardt and J W Sedat Science 3201332 (2008)

39 K I Willig R R Kellner R Medda B Hein

S Jakobs and S W Hell Nat Methods 3721 (2006)

40 Q F Li S S H Wu and K C ChouBiophys J 97 3224 (2009)

41 S Schrof T Staudt E Rittweger NWittenmayer T Dresbach J Engelhardtand S W Hell Opt Exp 19 8066 (2011)

42 U V Nagerl K I Willig B Hein S WHell and T Bonhoeffer Proc Natl Acad SciUSA 105 18982 (2008)

43 J B Ding K T Takasaki and B L SabatiniNeuron 63 429 (2009)

44 R G Parton and K Simons Nat Rev MolCell Biol 8 185 (2007)

45 L Campbell A J Hollins A Al-Eid G RNewman C von Ruhland and M Gumble-ton Biochem Biophys Res Co 262 744(1999)

46 U V Naegerl and T Bonhoeffer J Neuro-science 30 9341 (2010)

47 B Harke J Keller C K Ullal V WestphalA Schoenle and S W Hell Opt Exp 164154 (2008)

48 B Hein K I Willig and S W Hell ProcNatl Acad Sci USA 105 14271 (2008)

49 G Donnert J Keller C A Wurm S ORizzoli V Westphal A Schonle R Jahn SJakobs C Eggeling and S W Hell BiophysJ 92 L67 (2007)

50 J Buckers D Wildanger G Vicidomini LKastrup and S W Hell Opt Exp 19 3130(2011)

51 R R Kellner C J Baier K I Willig S WHell and F J Barrantes Neuroscience 144135 (2007)

52 R L Felts K Narayan J D Estes D Shi CM Trubey J Fu L M Hartnell G T RuthelD K Schneider K Nagashima J W Bess SBavari B C Lowekamp D Bliss J DLifson and S Subramaniam Proc NatlAcad Sci USA 107 13336 (2010)

53 P C Jennings G C Cox L G Monahanand E J Harry Micron 42 336 (2011)

54 C Eggeling C Ringemann R Medda GSchwarzmann K Sandhoff S Polyakova VN Belov B Hein C von Middendorff ASchonle and S W Hell Nature 457 1159(2009)

55 J J Sieber K I Willig R Heintzmann S WHell and T Lang Biophys J 90 2843(2006)

56 A Schneider L Rajendran M Honsho MGralle G Donnert F Wouters S W Helland M Simons J Neuroscience 28 2874(2008)

57 G H Patterson and J Lippincott-SchwartzScience 297 1873 (2002)

58 M Bates T R Blosser and X W ZhuangPhys Rev Lett 94 108101 (2005)

59 J Wiedenmann S Ivanchenko F Oswald FSchmitt C Rocker A Salih K D Spindlerand G U Nienhaus Proc Natl Acad SciUSA 101 15905 (2004)

60 G Patterson M Davidson S Manley and JLippincott-Schwartz Annu Rev Phys Chem61 345 (2010)

61 M F Juette T J Gould M D Lessard M JMlodzianoski B S Nagpure B T BennettS T Hess and J Bewersdorf Nat Methods 5527 (2008)

62 G Shtengel J A Galbraith C G GalbraithJ Lippincott-Schwartz J M Gillette S


Manley R Sougrat C M Waterman PKanchanawong M W Davidson R DFetter and H F Hess Proc Natl Acad SciUSA 106 3125 (2009)

63 F V Subach G H Patterson S Manley JM Gillette J Lippincott-Schwartz and V VVerkhusha Nature Methods 6 153 (2009)

64 D Greenfield A L McEvoy H Shroff G ECrooks N S Wingreen E Betzig and JLiphardt Plos Biology 7 e1000137 (2009)

65 V Tatavarty E J Kim V Rodionov and JYu Plos One 4 e7724 (2009)

66 G Fu T Huang J Buss C Coltharp ZHensel and J Xiao Plos One 5 e12682(2010)

67 M Bates B Huang G T Dempsey and XW Zhuang Science 317 1749 (2007)

68 B Huang W Q Wang M Bates and X WZhuang Science 319 810 (2008)

69 S A Jones S Shim J He and X ZhuangNat Methods 8 499 (2011)

70 B Huang S A Jones B Brandenburg andX W Zhuang Nat Methods 5 1047 (2008)

71 M Heilemann S van de Linde M Schutt-

pelz R Kasper B Seefeldt A Mukherjee PTinnefeld and M Sauer Angew Chem IntEd 47 6172 (2008)

72 L L Song E J Hennink I T Young and HJ Tanke Biophys J 68 2588 (1995)

73 G Donnert C Eggeling and S W Hell NatMethods 4 81 (2007)

74 K I Willig J Keller M Bossi and S WHell New J Phys 8 106 (2006)

75 H Shroff C G Galbraith J A Galbraith andE Betzig Nat Methods 5 417 (2008)

76 S E Irvine T Staudt E Rittweger JEngelhardt and S W Hell Angew Chem

Int Ed 47 2685 (2008)

77 K A Lidke B Rieger T M Jovin and RHeintzmann Opt Exp 13 7052 (2005)

78 K A Lidke B Rieger T M Jovin and RHeintzmann Biophys J 88 346a (2005)

79 P Hoyer T Staudt J Engelhardt and S WHell Nano Lett 11 245 (2011)

80 J K Jaiswal H Mattoussi J M Mauro andS M Simon Nat Biotechnol 21 47 (2003)

81 J B Delehanty H Mattoussi and I L

Medintz Anal Bioanal Chem 393 1091(2009)

82 W C W Chan and S M Nie Science 2812016 (1998)

83 J K Jaiswal E R Goldman H Mattoussiand S M Simon Nat Methods 1 73 (2004)

84 G A Silva in Nanoneuroscience andNanoneuropharmacology (Elsevier ScienceBv Amsterdam 2009) vol 180 p 19

85 U Resch-Genger M Grabolle S Cavaliere-Jaricot R Nitschke and T Nann NatMethods 5 763 (2008)

86 E Rittweger K Y Han S E Irvine CEggeling and S W Hell Nature Photon 3144 (2009)

87 L S Churchman H Flyvbjerg and J ASpudich Biophys J 90 668 (2006)

88 K I Mortensen L S Churchman J ASpudich and H Flyvbjerg Nat Methods 7377 (2010)

89 J Enderlein E Toprak and P R Selvin OptExp 14 8111 (2006)

90 B Huang M Bates and X W ZhuangAnnu Rev Biochem 78 993 (2009)


focal point review

scope One of the most intuitive ways tobreak the diffraction limit was proposedby Synge in 1928 using a sub-wave-length aperture to image a surface11 Theconcept was realized by Ash andNichols in 1972 using microwaves witha wavelength of 3 cm and the diffrac-tion limit was overcome for the firsttime12 In 1984 the technique wasdemonstrated in the visible region1314

and it evolved into a rapidly expandingfield now known as near-field scanningoptical microscopy (NSOM)15 Onelimitation of NSOM is that the aperturehas to be placed within one wavelengthof the light to obtain sub-diffraction-limit resolution before the light spatiallydiverges The resolution decreases withincreasing distance between the apertureand the sample Therefore NSOM islimited to the study of surfaces and itsapplicability in biology is limited

On the other hand efforts to developfar-field optical microscopy with sub-diffraction-limit resolution have contin-ued Interference is probably the easiestmethod by which to obtain a sub-wavelength pattern for far-field micros-copy In 1978 C Cremer and T Cremerproposed that a laser-scanning micro-scope using a point-hologram wouldproduce a spot with a diameter consid-erably smaller than the wavelengthused16 In 1993 sub-diffraction-limitedaxial resolution was demonstrated byusing standing-wave excitation17 Instanding-wave fluorescence microscopy(SWFM) two counter-propagating non-focused laser beams are used to form astanding wave which creates an excita-tion field with closely spaced nodes andantinodes to improve the axial resolu-tion In 1995 a technique called imageinterference microscopy (I2M) was de-veloped In I2M the fluorescence fromboth sides of a sample is collected usingtwo opposed objective lenses These twoimages are combined by a beam-splitterand superposed on the camera tointerfere then images with a higherresolution can be obtained from theinterference pattern18 I2M can be com-bined with the incoherent interferenceillumination (I3) used in SWFM thencalled I5M microscopy to achieve ahigher resolution I5M has achieved anaxial resolution seven-fold better than atypical wide-field microscope19

An interferometric technique was alsoproposed for laser scanning microscopyIn 1992 a detailed theory of the 4Piconfocal fluorescence microscope waspublished by Hell and Stelzer20 In a 4Piconfocal microscope the specimen isilluminated by two coherent laser beamsthrough two opposed objective lenses21

The interference of the counter-propa-gating beams can greatly reduce the sizeof the focal spot in the axial direction Aseven-fold increase in axial resolutionwas achieved22 However these axialinterferometric techniques offered littleor no improvement in the lateral resolu-tion and have not been widely adoptedamong biologists To improve the lateralresolution for wide-field microscopy aninterferometric technique called struc-tured illumination microscopy (SIM)was proposed in the late 1990s2324

SIM superposes a known excitationpattern over the unknown sample pat-tern then a higher resolution image canbe deconvolved from the resultingfringes25 Using saturated excitationpatterns to introduce higher-order termssaturated structured-illumination micros-copy (SSIM) has shown a lateralresolution of 50 nm on sufficientlybright and photostable samples26

In 1994 a new type of scanningfluorescence microscope was proposedby Hell and Wichmann to improve thelateral resolution21 The basic principleof the new laser scanning fluorescencemicroscope is that the size of thefluorescence spot the point spreadfunction (PSF) can be reduced byemploying a second laser beam tostimulate the fluorescent emission inthe outer regions of the excitation spotThe second laser beam has a wavelengthlonger than the detected fluorescencetherefore the stimulated emission willalso emit at a longer wavelength notseen by the detector27 In this newapproach the shape of the effectivefluorescent spot which determines thespatial resolution is sharpened by thesecond laser beam via the stimulatedemission Creating a desired depletionbeam profile at the focal point is one ofthe major challenges to improve theresolution Between 1994 and 2006Hell and his co-workers published aseries of different designs to manipulatethe profile of the depletion beam212829

The final experimental setup that attract-ed a great deal of attention was carriedout in 2006 using the spatial phasemodulation pattern described by Torokand Munro3031 At this point thepotential of stimulated emission deple-tion (STED) microscopy in biologicalresearch became convincing and com-mercial STED microscopes have recent-ly become available

On the other hand new developmentsin single-molecule spectroscopy in theearly 1990s have enabled a differentapproach to achieving nanometer-scaleoptical microscopy3233 When imaginga single isolated molecule the size of themolecule that appears in the image iethe PSF is much larger than the actualsize of the molecule because of thediffraction limit However the locationof the molecule (or the center of thePSF) can be determined with a muchhigher accuracy A technique that de-pends on this effect was proposed byBetzig in 199535 In 2006 three re-search groups independently demon-strated super-resolution microscopy us-ing high-precision localization of singlefluorophores named as photoactivatedlocalization microscopy (PALM)35 sto-chastic optical reconstruction micros-copy (STORM)36 and fluorescencephotoactivation localization microscopy(FPALM)37

In this review we will briefly sum-marize far-field super-resolution opticalmicroscopy As SIM STED (F)PALMand STORM microscopy are now be-coming commercially available for biol-ogists their broad impact on biologicalresearch can be expected Howeverthese super-resolution microscopeswhich were mostly developed anddesigned by physicists and chemistsare not as user-friendly as conventionaloptical microscopes Currently there isstill no ideal system that offers high-speed 3D nanometer spatial resolutionwith multicolor capabilities for live-cellimaging Each technique has itsstrengths and weaknesses Thereforeresearchers need to be informed thatthere are trade-offs between resolutionsensitivity and depth of view Experi-mental designs are often also limited bythe available fluorescence probes Thisreview will describe the basic principleof SSIM STED (F)PALM and


focal point review
















focal point review
















































Dr~k


Is

q




focal point review










































focal point review





Np




























focal point review























focal point review

















































































Int Ed 47 2685 (2008)

















focal point review
















focal point review
















































Dr~k


Is

q




focal point review










































focal point review





Np




























focal point review























focal point review

















































































Int Ed 47 2685 (2008)

















focal point review









focal point review
















































Dr~k


Is

q




focal point review










































focal point review





Np




























focal point review























focal point review

















































































Int Ed 47 2685 (2008)

















focal point review
















































Dr~k


Is

q




focal point review










































focal point review





Np




























focal point review























focal point review

















































































Int Ed 47 2685 (2008)

















focal point review























Dr~k


Is

q




focal point review










































focal point review





Np




























focal point review























focal point review

















































































Int Ed 47 2685 (2008)

















focal point review










































focal point review





Np




























focal point review























focal point review

















































































Int Ed 47 2685 (2008)

















focal point review















focal point review





Np




























focal point review























focal point review

















































































Int Ed 47 2685 (2008)

















focal point review





Np




























focal point review























focal point review

















































































Int Ed 47 2685 (2008)

















focal point review


















focal point review























focal point review

















































































Int Ed 47 2685 (2008)

















focal point review























focal point review

















































































Int Ed 47 2685 (2008)

















focal point review














focal point review

















































































Int Ed 47 2685 (2008)

















focal point review

















































































Int Ed 47 2685 (2008)

















focal point review

















Int Ed 47 2685 (2008)

















focal point review

Documents

Review of Super-Resolution Fluorescence Microscopy for Biology · PDF fileSTORM, their applications in biology, and technical considerations for imple-menting these methods. STRUCTURED