Parallelized ground state depletion [Elektronische Ressource] / by Miriam A. Schwentker

-

English
106 Pages
Read an excerpt
Gain access to the library to view online
Learn more

Description

DissertationsubmittedtotheCombinedFacultyfor theNaturalSciencesandforMathmaticsoftheRuperto CarolaUniversityofHeidelberg,GermanyforthedegreeofDoctorofNaturalSciencesbyDiplom PhysikerinMiriamA.Schwentkerborn13.07.1977inAugsburgoralexam: 06.06.2007GelängeunsdochderGriffausdemDunkelnachdenwirklichenDingen!(HeinzCzechowski)ParallelizedGroundStateDepletionGutachter: Prof. Dr. StefanW.HellProf. Dr. JosefBilleAbstractRESOLFT was introduced to break the limited resolution in fluorescence microscopygiven by thephysical limit of diffraction. RESOLFT needs saturable deple tion and illumination with spatial regions of zero intensity. Improvements in RESOLFTare still essential since (i) photobleaching restricts the use of large intensities obligate forhigh resolution, (ii) long acquisition times emerge from single point scanning, and (iii) RE SOLFTputsaheadspecialrequirementsonthedye´sphotophysicalproperties. Tocounteractthese problems, this work presents a RESOLFT type microscope implementing parallelizedground state depletion (GSD). Parallelization allows rather fast image acquisition apply ing structured illumination combined with widefield detection on a camera. The depletionof the fluophore´s ground state, and thus the fluorescence, is realized using a pump probeilluminationscheme. Toincreasephotostability,thesamplewasevacuatedandcooledtoap proximately80K.

Subjects

Informations

Published by
Published 01 January 2007
Reads 8
Language English
Document size 22 MB
Report a problem

Dissertation
submittedtothe
CombinedFacultyfortheNaturalSciencesandforMathmatics
oftheRuperto CarolaUniversityofHeidelberg,Germany
forthedegreeof
DoctorofNaturalSciences
by
Diplom PhysikerinMiriamA.Schwentker
born13.07.1977inAugsburg
oralexam: 06.06.2007Gelängeunsdoch
derGriffausdemDunkel
nachdenwirklichenDingen!
(HeinzCzechowski)ParallelizedGroundStateDepletion
Gutachter: Prof. Dr. StefanW.Hell
Prof. Dr. JosefBilleAbstract
RESOLFT was introduced to break the limited resolution in fluorescence microscopy
given by thephysical limit of diffraction. RESOLFT needs saturable deple
tion and illumination with spatial regions of zero intensity. Improvements in RESOLFT
are still essential since (i) photobleaching restricts the use of large intensities obligate for
high resolution, (ii) long acquisition times emerge from single point scanning, and (iii) RE
SOLFTputsaheadspecialrequirementsonthedye´sphotophysicalproperties. Tocounteract
these problems, this work presents a RESOLFT type microscope implementing parallelized
ground state depletion (GSD). Parallelization allows rather fast image acquisition apply
ing structured illumination combined with widefield detection on a camera. The depletion
of the fluophore´s ground state, and thus the fluorescence, is realized using a pump probe
illuminationscheme. Toincreasephotostability,thesamplewasevacuatedandcooledtoap
proximately80K. Since the population of the triplet state is an intrinsic property of almost
every dye and is enhanced by evacuation and low temperatures, parallelized GSD is gener-
ally applicable to a wide range of fluorescent markers. Applying RESOLFT an
enhancementinresolutionwasprovenforabeadsampleaswellasinsideacell. Parallelized
GSD microscopy further prevents photostress on both fluorescent label and sample, since it
2encountersmoderateilluminationintensities<kW/cm .
Zusammenfassung
DasRESOLFT Konzeptwurdeentwickelt,umdieAuflösungsgrenzeinderFluoreszenz
mikroskopie zu überwinden. RESOLFT benötigt einen sättigbare Fluoreszenzauslöschung
mechanismusundeineräumlicheBeleuchtungmitdunklenBereichen. Verbesserungensind
notwendig, da (i) starkes Photobleichen den Einsatz hoher Laser Intensitäten verbietet, die
jedoch für eine hohe Auflösung notwendig sind, (ii) die bisherigen RESOLFT Verfahren
LaserrasterMethodeneinsetzenundlangeAufnahmezeitenbedingen,und(iii)spezielleAn
forderungen an die eingesetzten Farbstoffe gesetzt sind. Um diese Problemen zu beheben,
wurde in der vorliegenden Arbeit ein neues RESOLFT Mikroskop entwickelt, welches den
sättigbarenFluoreszenunterdrückungsprozeßdurchEntleerungdesGrundzustandesüberdie
Bevölkerung des Tripletzustandes realisiert. Die Bildaufnahme ist durch die parallelisierte
Aufnahme mit Hilfe von strukturierter Beleuchtung und großflächiger Fluoreszenzdetek
tion über eine Kamera wesentlich beschleunigt. Die Entleerung des Grundzustandes eines
Farbstoffmoleküles und Löschung der Fluoreszenz geschieht durch ein Zwei Puls Beleuch
tungsschema. Um die Photostabilität der Farbstoffe zu erhöhen, wurde die Probe evakuiert
und auf 80K gekühlt. Da Tripletbevölkerung eine Eigenschaft fast jeden Farbstoffs ist und
durch Kühlung verstärkt wird, ist das vorgestellte RESOLFT Konzept weitreichend einset
zbar. Die vorgestellte Mikroskopiemethode erzielte Auflösungen unterhalb der Beugungs
grenze sowohl für punktförmige Farbstoffproben, als auch innerhalb von Zellen. Die dabei
2verwendetenIntensitätenvon<kW/cm übenkaumPhotostreßaufdieProbeaus.Contents
1 Introduction 1
2 TheRESOLFTconcept 3
2.1 ResolutionEnhancementwithRESOLFT . . . . . . . . . . . . . . . . . . 3
2.2 GroundStateDepletion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 ProblemsinherenttoRESOLFTmicroscopy . . . . . . . . . . . . . . . . . 13
3 ExperimentalRealisation 17
3.1 Time: Parallelisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1.1 Theoreticalexplanation . . . . . . . . . . . . . . . . . . . . . . . . 17
3.1.2 ExperimentalImplementation . . . . . . . . . . . . . . . . . . . . 32
3.2 Bleaching: Workingatlowtemperatures . . . . . . . . . . . . . . . . . . . 33
3.3 Photophysics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4 ParallelizedGroundStateDepletion 52
4.1 ExperimentalSetup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.2 Imagereconstruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.3 SingleBeads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.3.1 ConstrictionoftheMaxima . . . . . . . . . . . . . . . . . . . . . 59
4.3.2 Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.4 BiologicalImaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.5 Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5 Discussion 75
A Appendix 77
A.1 MaterialsandMethods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
A.1.1 SamplePreparation . . . . . . . . . . . . . . . . . . . . . . . . . . 77
A.1.2 Cellcultureandimmunocytochemistry . . . . . . . . . . . . . . . 78
A.2 Derivations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
A.2.1 Pointspreadfunction . . . . . . . . . . . . . . . . . . . . . . . . . 79
A.2.2 RESOLFTrateequations . . . . . . . . . . . . . . . . . . . . . . . 80
A.2.3resolutionequation . . . . . . . . . . . . . . . . . . . . 81
A.2.4 Threestaterateequationsandsteadystatelevels . . . . . . . . . . 82
A.2.5 Derivationoffouriercoefficientsforsinesquare . . . . . . . . . . . 85
A.2.6 Derivationofthepump probecharacteristics . . . . . . . . . . . . 88
iA.2.7 DerivationoftheplateaufluorescencewithreverseISC . . . . . . . 89
A.2.8 Characterizationoftheemissionprobabilitypattern . . 91
A.3 Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
Bibliography 96
ii1 Introduction
Duringthelastcenturiesmicroscopyhasincreasinglybecomeanimportanttoolinmanydif
ferentscientifcareas,especiallyinthefieldofbiology. Themostcommonlyusedmicroscopy
technique in biology is optical microscopy, using fluorescent markers to identify certain
structures or proteins. One demand on microscopy is high resolution - that is, the smallest
distanceatwhichtwoobjectscanstillberesolved. In1873ErnstAbbeuncoveredthephys
ical limit of resolution (the so called diffraction limit), which is approximately half of the
wavelength of the light used for imaging [1]. With the use of visible light (400 −800nm),
afar fieldmicroscopecannotdiscernalikeobjectslessthan ≈ 200nmapart. Asmostofthe
interestingcellularfeaturesareintherangeofafewnmthereisaneedforhigherresolution.
Because of phototoxicity it is not feasible to reduce the applied wavelength further to the
ultraviolet(wavelength< 350nm,[2])forresolutionenhancement.
The diffraction limit derived by Abbe was perceived as being fixed for a long time, but
throughout the last century different approaches in high resolution microscopy have been
pursued, e.g. electron microscopy (EM) [3], atomic force microscopy (AFM) [4] or scan
ning near field opticaly (SNOM) [5]. But all these methods have features, which
reduce their applicability to life sciences. EM samples have to be prepared laboriously, that
is dehydrated, cut in very thin slices and even coated with metal, because of the high ab
sorption of electrons. Three dimensional images from cellular compounds are possible, but
need extensive resonstruction using a PC due to the poor labelling possibilities in compari
sontofluorescencemicroscopy. AFMisrestrictedtosurfacesandverymuchdependsonthe
interactions between scanning tip and sample. SNOM takes advantage of the broad range
of fluorescent markers, but is restricted to surfaces as well. All these techniques feature
veryhighresolutionintherangeoffewnm,buttheirpreparationisveryelaborateanddoes
not allow live cell imaging, and three dimensional imaging is hardly possible without huge
effort.
Eventually,theinterestinbiologicalmicroscopyistheimagingoflivingcellswithhigh
resolutionandgoodcontrast. Therefore,fluorecencemicropscopyisstillthebestcandidate,
partly because of the range of fluorescent labels, from organic dyes to fluorescent proteins.
Confocal microscopy even allows very fast acquisition of three dimensional images of a
cellular structures ([6]). The issue of overcoming the diffraction barrier is addressed by
a new concept using Reversible Saturable Optical (Fluorescence) Transitions, RESOLFT
([7],[8],[9],[10]).
The RESOLFT technique exploits a saturable nonlinear relationship between illumina
tion light and resulting fluorescence, using a transition between two states one of which
should be fluorescing. This light driven saturable is used to restrict fluorescence
emission to spots or lines of minimized size. So far, RESOLFT has been implemented2
in STimulated Emission Depletion (STED) microscopy ([11],[12], [13]), photoswitchable
chromophores([14])andphotoswitchableproteins([15],[16]). Anothersaturableprocessis
GroundStateDepletion (GSD) as theortically suggested by Hell et.al. ([17]) which will be
usedinthisproject. Contrarytopreviouslyreportedsub diffractionresolutiontechniques,it
relies both on modest continuous wave (CW) illumination intensities and on standard fluo
rescent markers. However, it is challenged by photobleaching, which accounts for the fact,
that, although GSD microscopy has been suggested over ten years ago ([17]), experimental
realizationhasnotbeenpursued. ThepresentworkdiscoversoptimizedconditionsforGSD
microscopy realizing its experimental implementation. Large illumination intensities and
thushighsaturationoffluorescencedepletionismadepossiblebyevacuatingandcoolingof
the sample, which is known to reduce bleaching ([18]). A setup will be adapted allowing
subdiffractionfluorescenceimagingonevacuatedandfrozensamples.
Usually RESOLFT is applied to decrease the size of the confocal scanning spot in one,
two or even three dimenstions. To cover the same image area as in regular confocal micr-
socopyRESOLFT microscopyhastoperformcomparativeymorescanningsteps,eachwith
a longer dwell time, since the reduced spot covers a smaller region, eliciting less fluores
cence signal. To counteract the longer image acquisition time, parallelization by structured
illuminationandcamera baseddetectionwillbeintroduced.
The presented work depicts the experimental implementation of parallelized GSD at
low temperatures. The following chapter will explain the concept of resolution enhance
ment using RESOLFT and outline the principle of GSD. The third chapter introduces the
experimental realization and challenges of parallelization and imaging at low temperatures.
Chapterfourdescribestheexperimentalsetupandandoutlinesaresolutionenhancementof
afactorofupto3withthepresentsetupofparallelizedGSDatlowtemperatures.2 TheRESOLFTconcept
InthischapterthenatureofthediffractionlimitwillbeexplainedandhowRESOLFTallows
imagingwithgreaterdetailthanimposedbythisdiffractionlimit. Groundstatedepletionwill
be presented as one method for RESOLFT imaging, which was used in this project, and the
generalproblemsofthetechniquewillbeadressed.
2.1 ResolutionEnhancementwithRESOLFT
Every image taken of an object using light misses small details of that object, either tele
scopicimagesofthestars,photographsofoursurroundingsormicrsopsopicimagesofcells
– even with perfect lenses. This is due to the wave nature of light. Imaging light emitted
fromonepointdoesnotresultinonesinglepointbutablurredspot. Anobjectcanbetaken
asaseriesofpoint likelightemitters,resultinginaseriesofblurredspots,andthereforethe
imageofanobjectwillbeblurredandlackdetails.
imaging
O(x) B(x)
Figure2.1:BlurringduetoImagingImageofAbbe(left)isblurredduetoimaging(right)andthereforelacks
detail in comparison to the original object. In mathmatical terms the objectO(x) is convolved with the point
h(x)spreadfuntionh(x)toyieldtheimageB(x).
The resolution describes the ability of an optical system - telescope, photographic cam B(x) = O(x) h(x)
eraormicroscope-toimagedetailsoftheoriginalobject. Thehighertheresolutionthemore
detailscanbeseenintheimage. Attheendofthe19thcenturyErnstAbbefoundarelation
shipbetweenthesizeΔxoftheblurredspotemittedbyapointsourceandthewavelengthof
theemittedlight[1]. Thefullwidthathalfmaximum(FWHM)ofthatblurredspotwaslater
quantified by Lord Rayleigh and is still known as the Abbe diffraction limit (see appendix
A.2.1forderivation):2.1ResolutionEnhancementwithRESOLFT 4
λ
Δx = 0.61 (2.1)
n·sinα
λ is the wavelength of the emitted light, n the refractive index of the material between
imaging lens and sample (immersion medium) and α half the angle under which the light
emittedbythesinglespotiscollected. Theproductn·sinαiscalledthenumericalapterure
NAoftheimagingoptics. Thespatialdistributionofthelightemittedbyapoint likesource
andimagedbyanopticalsystemiscalledthePointSpreadFunction(PSF)andischaracter-
istictotheopticalsystem.
As described before, the object can be seen as a series of point like emitters, each of
which is imaged as a blurred spot (the PSF) and the final images results of an assembly
of these blurred spots. In mathmatical terms this mechanism is described as a convolution
between the spatial (x) distribution of the object O(x) and the PSF h(x). The final image
B(x)is:
B(x) =O(x)⊗h(x) (2.2)
Inbiology,far fieldfluorescencemicroscopyismostcommonlyusedforimaging. Struc
turesormoleculestobeimagedarespecificallystainedwithafluorescencemarker,forexam
pleanorganicdyeorafluorescentprotein,whosefluorescenceemissionisexcited(e.g. bya
laser),andthefluorescenceisdetectedbytheopticalsystemandaphotondetector. Far field
fluorescence microscopy not only allows imaging of surfaces (as in electron microscopy or
near fieldopticaly)butwithinathree dimensionalobjectwiththree dimensional
imagereconstruction.
The object that is imaged is the distribution of dye molecules within the sample, that is
thedensityofdyemoleculesρ (x)dependingonthespatialpositionx. Atmoderateexcita dye
tionintensitiesI(x)thefluorescenceF(x)emittedfromonedyemoleculeisproportionalto
I(x) and the resultingS(x) from the sample is the product of the fluorescence
signalF(x),detectedfromonedyemolecule,andthedyedistributionρ (x):dye
S(x) =ρ (x)·F(x)dye
∝ρ (x)·I(x) (2.3)dye
with F(x)∝I(x)
ThewavelengthdeterminingthesizeofthePSFinequation2.1isthewavelengthofthe
fluorescencelightλ ,whichduetotheStokesshiftislargerthantheexcitationwavelengthem
λ . Toavoidphototoxicitytheexcitationwavelengthλ shouldnotbe< 400nm. Thebestex ex
objective lenses used in modern light microscopy allow a semi aperture angle of α ≈ 73°.
Thisleadstoahighestresolution(theFWHMofthePSF)of≈ 170nm(andNA = 1.5).
The resolution in confocal microscopy is slightly higher than in standard widefield mi
croscopy. Inthelatterthesampleisilluminatedhomogeneouslyandtheresolutionisdirectly
linked to the Abbe diffraction limit with λ . In confocal microscopy (see figure 2.2) theem
sampleisscannedwithahighlyfocusedbeamandthefinalimageisreconstructedfromthe